{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 1937,
  "global_step": 19365,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 5.1639555899819264e-05,
      "grad_norm": 151.14651489257812,
      "learning_rate": 5.1546391752577325e-08,
      "loss": 0.5101,
      "step": 1
    },
    {
      "epoch": 0.00010327911179963853,
      "grad_norm": 208.51341247558594,
      "learning_rate": 1.0309278350515465e-07,
      "loss": 0.7166,
      "step": 2
    },
    {
      "epoch": 0.00015491866769945777,
      "grad_norm": 302.1886291503906,
      "learning_rate": 1.5463917525773197e-07,
      "loss": 1.8659,
      "step": 3
    },
    {
      "epoch": 0.00020655822359927706,
      "grad_norm": 384.33966064453125,
      "learning_rate": 2.061855670103093e-07,
      "loss": 0.6037,
      "step": 4
    },
    {
      "epoch": 0.0002581977794990963,
      "grad_norm": 161.31434631347656,
      "learning_rate": 2.577319587628866e-07,
      "loss": 1.2018,
      "step": 5
    },
    {
      "epoch": 0.00030983733539891554,
      "grad_norm": 306.5719909667969,
      "learning_rate": 3.0927835051546394e-07,
      "loss": 1.201,
      "step": 6
    },
    {
      "epoch": 0.00036147689129873486,
      "grad_norm": 387.6163024902344,
      "learning_rate": 3.608247422680412e-07,
      "loss": 1.247,
      "step": 7
    },
    {
      "epoch": 0.0004131164471985541,
      "grad_norm": 179.30392456054688,
      "learning_rate": 4.123711340206186e-07,
      "loss": 1.4575,
      "step": 8
    },
    {
      "epoch": 0.00046475600309837337,
      "grad_norm": 169.96343994140625,
      "learning_rate": 4.6391752577319593e-07,
      "loss": 1.067,
      "step": 9
    },
    {
      "epoch": 0.0005163955589981926,
      "grad_norm": 131.09303283691406,
      "learning_rate": 5.154639175257732e-07,
      "loss": 0.581,
      "step": 10
    },
    {
      "epoch": 0.0005680351148980118,
      "grad_norm": 38.55739974975586,
      "learning_rate": 5.670103092783505e-07,
      "loss": 0.346,
      "step": 11
    },
    {
      "epoch": 0.0006196746707978311,
      "grad_norm": 271.78271484375,
      "learning_rate": 6.185567010309279e-07,
      "loss": 2.3254,
      "step": 12
    },
    {
      "epoch": 0.0006713142266976503,
      "grad_norm": 384.5425720214844,
      "learning_rate": 6.701030927835052e-07,
      "loss": 1.2953,
      "step": 13
    },
    {
      "epoch": 0.0007229537825974697,
      "grad_norm": 190.00439453125,
      "learning_rate": 7.216494845360824e-07,
      "loss": 0.6619,
      "step": 14
    },
    {
      "epoch": 0.000774593338497289,
      "grad_norm": 73.83168029785156,
      "learning_rate": 7.731958762886599e-07,
      "loss": 0.3189,
      "step": 15
    },
    {
      "epoch": 0.0008262328943971082,
      "grad_norm": 191.72682189941406,
      "learning_rate": 8.247422680412372e-07,
      "loss": 0.6895,
      "step": 16
    },
    {
      "epoch": 0.0008778724502969275,
      "grad_norm": 98.48928833007812,
      "learning_rate": 8.762886597938144e-07,
      "loss": 0.4238,
      "step": 17
    },
    {
      "epoch": 0.0009295120061967467,
      "grad_norm": 288.5750732421875,
      "learning_rate": 9.278350515463919e-07,
      "loss": 1.4426,
      "step": 18
    },
    {
      "epoch": 0.000981151562096566,
      "grad_norm": 142.048583984375,
      "learning_rate": 9.793814432989692e-07,
      "loss": 0.7616,
      "step": 19
    },
    {
      "epoch": 0.0010327911179963851,
      "grad_norm": 140.44268798828125,
      "learning_rate": 1.0309278350515464e-06,
      "loss": 0.3471,
      "step": 20
    },
    {
      "epoch": 0.0010844306738962044,
      "grad_norm": 52.17066955566406,
      "learning_rate": 1.0824742268041239e-06,
      "loss": 0.2587,
      "step": 21
    },
    {
      "epoch": 0.0011360702297960237,
      "grad_norm": 49.1740608215332,
      "learning_rate": 1.134020618556701e-06,
      "loss": 0.3009,
      "step": 22
    },
    {
      "epoch": 0.001187709785695843,
      "grad_norm": 262.36798095703125,
      "learning_rate": 1.1855670103092783e-06,
      "loss": 0.425,
      "step": 23
    },
    {
      "epoch": 0.0012393493415956622,
      "grad_norm": 23.182348251342773,
      "learning_rate": 1.2371134020618557e-06,
      "loss": 0.302,
      "step": 24
    },
    {
      "epoch": 0.0012909888974954814,
      "grad_norm": 19.936756134033203,
      "learning_rate": 1.288659793814433e-06,
      "loss": 0.2938,
      "step": 25
    },
    {
      "epoch": 0.0013426284533953007,
      "grad_norm": 23.371959686279297,
      "learning_rate": 1.3402061855670104e-06,
      "loss": 0.3026,
      "step": 26
    },
    {
      "epoch": 0.0013942680092951202,
      "grad_norm": 31.33051300048828,
      "learning_rate": 1.3917525773195878e-06,
      "loss": 0.2442,
      "step": 27
    },
    {
      "epoch": 0.0014459075651949394,
      "grad_norm": 25.045204162597656,
      "learning_rate": 1.4432989690721649e-06,
      "loss": 0.2894,
      "step": 28
    },
    {
      "epoch": 0.0014975471210947587,
      "grad_norm": 25.97345733642578,
      "learning_rate": 1.4948453608247423e-06,
      "loss": 0.2349,
      "step": 29
    },
    {
      "epoch": 0.001549186676994578,
      "grad_norm": 42.69337463378906,
      "learning_rate": 1.5463917525773197e-06,
      "loss": 0.2964,
      "step": 30
    },
    {
      "epoch": 0.0016008262328943972,
      "grad_norm": 41.879661560058594,
      "learning_rate": 1.597938144329897e-06,
      "loss": 0.3117,
      "step": 31
    },
    {
      "epoch": 0.0016524657887942165,
      "grad_norm": 54.57902908325195,
      "learning_rate": 1.6494845360824744e-06,
      "loss": 0.3115,
      "step": 32
    },
    {
      "epoch": 0.0017041053446940357,
      "grad_norm": 19.439001083374023,
      "learning_rate": 1.7010309278350518e-06,
      "loss": 0.2794,
      "step": 33
    },
    {
      "epoch": 0.001755744900593855,
      "grad_norm": 18.906618118286133,
      "learning_rate": 1.7525773195876288e-06,
      "loss": 0.1786,
      "step": 34
    },
    {
      "epoch": 0.0018073844564936742,
      "grad_norm": 30.162254333496094,
      "learning_rate": 1.8041237113402063e-06,
      "loss": 0.2463,
      "step": 35
    },
    {
      "epoch": 0.0018590240123934935,
      "grad_norm": 33.67458724975586,
      "learning_rate": 1.8556701030927837e-06,
      "loss": 0.1903,
      "step": 36
    },
    {
      "epoch": 0.0019106635682933127,
      "grad_norm": 21.45730972290039,
      "learning_rate": 1.907216494845361e-06,
      "loss": 0.2232,
      "step": 37
    },
    {
      "epoch": 0.001962303124193132,
      "grad_norm": 30.974803924560547,
      "learning_rate": 1.9587628865979384e-06,
      "loss": 0.2194,
      "step": 38
    },
    {
      "epoch": 0.002013942680092951,
      "grad_norm": 28.242895126342773,
      "learning_rate": 2.010309278350516e-06,
      "loss": 0.2709,
      "step": 39
    },
    {
      "epoch": 0.0020655822359927703,
      "grad_norm": 23.270423889160156,
      "learning_rate": 2.061855670103093e-06,
      "loss": 0.205,
      "step": 40
    },
    {
      "epoch": 0.0021172217918925896,
      "grad_norm": 14.165643692016602,
      "learning_rate": 2.1134020618556703e-06,
      "loss": 0.1617,
      "step": 41
    },
    {
      "epoch": 0.002168861347792409,
      "grad_norm": 23.96873664855957,
      "learning_rate": 2.1649484536082477e-06,
      "loss": 0.1773,
      "step": 42
    },
    {
      "epoch": 0.002220500903692228,
      "grad_norm": 47.468841552734375,
      "learning_rate": 2.2164948453608247e-06,
      "loss": 0.2227,
      "step": 43
    },
    {
      "epoch": 0.0022721404595920473,
      "grad_norm": 16.021501541137695,
      "learning_rate": 2.268041237113402e-06,
      "loss": 0.1959,
      "step": 44
    },
    {
      "epoch": 0.0023237800154918666,
      "grad_norm": 34.86495590209961,
      "learning_rate": 2.3195876288659796e-06,
      "loss": 0.2823,
      "step": 45
    },
    {
      "epoch": 0.002375419571391686,
      "grad_norm": 33.38130187988281,
      "learning_rate": 2.3711340206185566e-06,
      "loss": 0.1484,
      "step": 46
    },
    {
      "epoch": 0.002427059127291505,
      "grad_norm": 21.61211395263672,
      "learning_rate": 2.422680412371134e-06,
      "loss": 0.1708,
      "step": 47
    },
    {
      "epoch": 0.0024786986831913244,
      "grad_norm": 13.025924682617188,
      "learning_rate": 2.4742268041237115e-06,
      "loss": 0.1836,
      "step": 48
    },
    {
      "epoch": 0.0025303382390911436,
      "grad_norm": 14.986786842346191,
      "learning_rate": 2.525773195876289e-06,
      "loss": 0.1343,
      "step": 49
    },
    {
      "epoch": 0.002581977794990963,
      "grad_norm": 11.67739486694336,
      "learning_rate": 2.577319587628866e-06,
      "loss": 0.1471,
      "step": 50
    },
    {
      "epoch": 0.002633617350890782,
      "grad_norm": 28.498422622680664,
      "learning_rate": 2.628865979381444e-06,
      "loss": 0.1409,
      "step": 51
    },
    {
      "epoch": 0.0026852569067906014,
      "grad_norm": 36.6308479309082,
      "learning_rate": 2.680412371134021e-06,
      "loss": 0.1627,
      "step": 52
    },
    {
      "epoch": 0.0027368964626904206,
      "grad_norm": 15.6350679397583,
      "learning_rate": 2.731958762886598e-06,
      "loss": 0.1531,
      "step": 53
    },
    {
      "epoch": 0.0027885360185902403,
      "grad_norm": 18.23200035095215,
      "learning_rate": 2.7835051546391757e-06,
      "loss": 0.1558,
      "step": 54
    },
    {
      "epoch": 0.0028401755744900596,
      "grad_norm": 51.352928161621094,
      "learning_rate": 2.8350515463917527e-06,
      "loss": 0.3178,
      "step": 55
    },
    {
      "epoch": 0.002891815130389879,
      "grad_norm": 52.27288055419922,
      "learning_rate": 2.8865979381443297e-06,
      "loss": 0.227,
      "step": 56
    },
    {
      "epoch": 0.002943454686289698,
      "grad_norm": 62.46421432495117,
      "learning_rate": 2.9381443298969076e-06,
      "loss": 0.3296,
      "step": 57
    },
    {
      "epoch": 0.0029950942421895174,
      "grad_norm": 43.33767318725586,
      "learning_rate": 2.9896907216494846e-06,
      "loss": 0.2508,
      "step": 58
    },
    {
      "epoch": 0.0030467337980893366,
      "grad_norm": 20.027721405029297,
      "learning_rate": 3.041237113402062e-06,
      "loss": 0.2312,
      "step": 59
    },
    {
      "epoch": 0.003098373353989156,
      "grad_norm": 12.170441627502441,
      "learning_rate": 3.0927835051546395e-06,
      "loss": 0.1206,
      "step": 60
    },
    {
      "epoch": 0.003150012909888975,
      "grad_norm": 22.435781478881836,
      "learning_rate": 3.1443298969072165e-06,
      "loss": 0.1571,
      "step": 61
    },
    {
      "epoch": 0.0032016524657887944,
      "grad_norm": 19.189538955688477,
      "learning_rate": 3.195876288659794e-06,
      "loss": 0.1332,
      "step": 62
    },
    {
      "epoch": 0.0032532920216886137,
      "grad_norm": 16.46035385131836,
      "learning_rate": 3.2474226804123714e-06,
      "loss": 0.1308,
      "step": 63
    },
    {
      "epoch": 0.003304931577588433,
      "grad_norm": 12.474228858947754,
      "learning_rate": 3.298969072164949e-06,
      "loss": 0.1013,
      "step": 64
    },
    {
      "epoch": 0.003356571133488252,
      "grad_norm": 10.034826278686523,
      "learning_rate": 3.350515463917526e-06,
      "loss": 0.1112,
      "step": 65
    },
    {
      "epoch": 0.0034082106893880714,
      "grad_norm": 15.092479705810547,
      "learning_rate": 3.4020618556701037e-06,
      "loss": 0.1264,
      "step": 66
    },
    {
      "epoch": 0.0034598502452878907,
      "grad_norm": 11.673678398132324,
      "learning_rate": 3.4536082474226807e-06,
      "loss": 0.1463,
      "step": 67
    },
    {
      "epoch": 0.00351148980118771,
      "grad_norm": 9.947355270385742,
      "learning_rate": 3.5051546391752577e-06,
      "loss": 0.0369,
      "step": 68
    },
    {
      "epoch": 0.003563129357087529,
      "grad_norm": 43.462608337402344,
      "learning_rate": 3.5567010309278356e-06,
      "loss": 0.1686,
      "step": 69
    },
    {
      "epoch": 0.0036147689129873485,
      "grad_norm": 30.30590057373047,
      "learning_rate": 3.6082474226804126e-06,
      "loss": 0.2268,
      "step": 70
    },
    {
      "epoch": 0.0036664084688871677,
      "grad_norm": 63.40016174316406,
      "learning_rate": 3.6597938144329896e-06,
      "loss": 0.2511,
      "step": 71
    },
    {
      "epoch": 0.003718048024786987,
      "grad_norm": 44.29582214355469,
      "learning_rate": 3.7113402061855674e-06,
      "loss": 0.1343,
      "step": 72
    },
    {
      "epoch": 0.0037696875806868062,
      "grad_norm": 23.782470703125,
      "learning_rate": 3.7628865979381445e-06,
      "loss": 0.1308,
      "step": 73
    },
    {
      "epoch": 0.0038213271365866255,
      "grad_norm": 44.18037033081055,
      "learning_rate": 3.814432989690722e-06,
      "loss": 0.236,
      "step": 74
    },
    {
      "epoch": 0.0038729666924864447,
      "grad_norm": 57.37205123901367,
      "learning_rate": 3.865979381443299e-06,
      "loss": 0.4287,
      "step": 75
    },
    {
      "epoch": 0.003924606248386264,
      "grad_norm": 14.491768836975098,
      "learning_rate": 3.917525773195877e-06,
      "loss": 0.1471,
      "step": 76
    },
    {
      "epoch": 0.003976245804286083,
      "grad_norm": 21.034934997558594,
      "learning_rate": 3.969072164948453e-06,
      "loss": 0.2352,
      "step": 77
    },
    {
      "epoch": 0.004027885360185902,
      "grad_norm": 20.42064666748047,
      "learning_rate": 4.020618556701032e-06,
      "loss": 0.1475,
      "step": 78
    },
    {
      "epoch": 0.004079524916085721,
      "grad_norm": 15.45172119140625,
      "learning_rate": 4.072164948453608e-06,
      "loss": 0.1875,
      "step": 79
    },
    {
      "epoch": 0.004131164471985541,
      "grad_norm": 52.90830993652344,
      "learning_rate": 4.123711340206186e-06,
      "loss": 0.2503,
      "step": 80
    },
    {
      "epoch": 0.00418280402788536,
      "grad_norm": 35.40111541748047,
      "learning_rate": 4.175257731958763e-06,
      "loss": 0.1944,
      "step": 81
    },
    {
      "epoch": 0.004234443583785179,
      "grad_norm": 45.74510192871094,
      "learning_rate": 4.2268041237113405e-06,
      "loss": 0.2096,
      "step": 82
    },
    {
      "epoch": 0.004286083139684998,
      "grad_norm": 34.63525390625,
      "learning_rate": 4.278350515463918e-06,
      "loss": 0.1588,
      "step": 83
    },
    {
      "epoch": 0.004337722695584818,
      "grad_norm": 13.451525688171387,
      "learning_rate": 4.329896907216495e-06,
      "loss": 0.2038,
      "step": 84
    },
    {
      "epoch": 0.004389362251484637,
      "grad_norm": 20.81909942626953,
      "learning_rate": 4.381443298969073e-06,
      "loss": 0.1713,
      "step": 85
    },
    {
      "epoch": 0.004441001807384456,
      "grad_norm": 18.89186668395996,
      "learning_rate": 4.4329896907216494e-06,
      "loss": 0.1282,
      "step": 86
    },
    {
      "epoch": 0.004492641363284275,
      "grad_norm": 15.41633129119873,
      "learning_rate": 4.484536082474228e-06,
      "loss": 0.1314,
      "step": 87
    },
    {
      "epoch": 0.004544280919184095,
      "grad_norm": 28.283287048339844,
      "learning_rate": 4.536082474226804e-06,
      "loss": 0.2327,
      "step": 88
    },
    {
      "epoch": 0.004595920475083914,
      "grad_norm": 9.663429260253906,
      "learning_rate": 4.587628865979382e-06,
      "loss": 0.1547,
      "step": 89
    },
    {
      "epoch": 0.004647560030983733,
      "grad_norm": 10.074034690856934,
      "learning_rate": 4.639175257731959e-06,
      "loss": 0.1164,
      "step": 90
    },
    {
      "epoch": 0.004699199586883552,
      "grad_norm": 15.28135871887207,
      "learning_rate": 4.690721649484537e-06,
      "loss": 0.1669,
      "step": 91
    },
    {
      "epoch": 0.004750839142783372,
      "grad_norm": 34.13657760620117,
      "learning_rate": 4.742268041237113e-06,
      "loss": 0.1485,
      "step": 92
    },
    {
      "epoch": 0.004802478698683191,
      "grad_norm": 18.038320541381836,
      "learning_rate": 4.7938144329896915e-06,
      "loss": 0.1457,
      "step": 93
    },
    {
      "epoch": 0.00485411825458301,
      "grad_norm": 8.730405807495117,
      "learning_rate": 4.845360824742268e-06,
      "loss": 0.1062,
      "step": 94
    },
    {
      "epoch": 0.0049057578104828295,
      "grad_norm": 8.752382278442383,
      "learning_rate": 4.8969072164948455e-06,
      "loss": 0.1001,
      "step": 95
    },
    {
      "epoch": 0.004957397366382649,
      "grad_norm": 33.873817443847656,
      "learning_rate": 4.948453608247423e-06,
      "loss": 0.1753,
      "step": 96
    },
    {
      "epoch": 0.005009036922282468,
      "grad_norm": 10.068822860717773,
      "learning_rate": 5e-06,
      "loss": 0.1686,
      "step": 97
    },
    {
      "epoch": 0.005060676478182287,
      "grad_norm": 7.206031322479248,
      "learning_rate": 5.051546391752578e-06,
      "loss": 0.0889,
      "step": 98
    },
    {
      "epoch": 0.0051123160340821065,
      "grad_norm": 7.78395938873291,
      "learning_rate": 5.1030927835051544e-06,
      "loss": 0.1309,
      "step": 99
    },
    {
      "epoch": 0.005163955589981926,
      "grad_norm": 19.48714828491211,
      "learning_rate": 5.154639175257732e-06,
      "loss": 0.2159,
      "step": 100
    },
    {
      "epoch": 0.005215595145881745,
      "grad_norm": 15.681782722473145,
      "learning_rate": 5.20618556701031e-06,
      "loss": 0.1489,
      "step": 101
    },
    {
      "epoch": 0.005267234701781564,
      "grad_norm": 12.479681968688965,
      "learning_rate": 5.257731958762888e-06,
      "loss": 0.1782,
      "step": 102
    },
    {
      "epoch": 0.0053188742576813835,
      "grad_norm": 10.832316398620605,
      "learning_rate": 5.309278350515464e-06,
      "loss": 0.0954,
      "step": 103
    },
    {
      "epoch": 0.005370513813581203,
      "grad_norm": 59.741554260253906,
      "learning_rate": 5.360824742268042e-06,
      "loss": 0.1718,
      "step": 104
    },
    {
      "epoch": 0.005422153369481022,
      "grad_norm": 20.77020835876465,
      "learning_rate": 5.412371134020619e-06,
      "loss": 0.1081,
      "step": 105
    },
    {
      "epoch": 0.005473792925380841,
      "grad_norm": 18.219430923461914,
      "learning_rate": 5.463917525773196e-06,
      "loss": 0.1711,
      "step": 106
    },
    {
      "epoch": 0.005525432481280661,
      "grad_norm": 16.551544189453125,
      "learning_rate": 5.515463917525774e-06,
      "loss": 0.1062,
      "step": 107
    },
    {
      "epoch": 0.005577072037180481,
      "grad_norm": 9.044017791748047,
      "learning_rate": 5.567010309278351e-06,
      "loss": 0.107,
      "step": 108
    },
    {
      "epoch": 0.0056287115930803,
      "grad_norm": 20.787067413330078,
      "learning_rate": 5.618556701030928e-06,
      "loss": 0.1434,
      "step": 109
    },
    {
      "epoch": 0.005680351148980119,
      "grad_norm": 15.563191413879395,
      "learning_rate": 5.670103092783505e-06,
      "loss": 0.107,
      "step": 110
    },
    {
      "epoch": 0.0057319907048799384,
      "grad_norm": 20.330432891845703,
      "learning_rate": 5.721649484536083e-06,
      "loss": 0.2336,
      "step": 111
    },
    {
      "epoch": 0.005783630260779758,
      "grad_norm": 48.23778533935547,
      "learning_rate": 5.7731958762886594e-06,
      "loss": 0.2776,
      "step": 112
    },
    {
      "epoch": 0.005835269816679577,
      "grad_norm": 5.85194730758667,
      "learning_rate": 5.824742268041238e-06,
      "loss": 0.1114,
      "step": 113
    },
    {
      "epoch": 0.005886909372579396,
      "grad_norm": 21.70347023010254,
      "learning_rate": 5.876288659793815e-06,
      "loss": 0.1197,
      "step": 114
    },
    {
      "epoch": 0.0059385489284792155,
      "grad_norm": 9.90454387664795,
      "learning_rate": 5.927835051546393e-06,
      "loss": 0.1333,
      "step": 115
    },
    {
      "epoch": 0.005990188484379035,
      "grad_norm": 11.491694450378418,
      "learning_rate": 5.979381443298969e-06,
      "loss": 0.0798,
      "step": 116
    },
    {
      "epoch": 0.006041828040278854,
      "grad_norm": 8.78729248046875,
      "learning_rate": 6.030927835051547e-06,
      "loss": 0.1279,
      "step": 117
    },
    {
      "epoch": 0.006093467596178673,
      "grad_norm": 11.919055938720703,
      "learning_rate": 6.082474226804124e-06,
      "loss": 0.0878,
      "step": 118
    },
    {
      "epoch": 0.0061451071520784925,
      "grad_norm": 24.196142196655273,
      "learning_rate": 6.134020618556702e-06,
      "loss": 0.1503,
      "step": 119
    },
    {
      "epoch": 0.006196746707978312,
      "grad_norm": 15.570281982421875,
      "learning_rate": 6.185567010309279e-06,
      "loss": 0.1353,
      "step": 120
    },
    {
      "epoch": 0.006248386263878131,
      "grad_norm": 9.711990356445312,
      "learning_rate": 6.237113402061856e-06,
      "loss": 0.0932,
      "step": 121
    },
    {
      "epoch": 0.00630002581977795,
      "grad_norm": 27.815322875976562,
      "learning_rate": 6.288659793814433e-06,
      "loss": 0.2513,
      "step": 122
    },
    {
      "epoch": 0.0063516653756777695,
      "grad_norm": 11.150771141052246,
      "learning_rate": 6.34020618556701e-06,
      "loss": 0.1087,
      "step": 123
    },
    {
      "epoch": 0.006403304931577589,
      "grad_norm": 41.11570358276367,
      "learning_rate": 6.391752577319588e-06,
      "loss": 0.1745,
      "step": 124
    },
    {
      "epoch": 0.006454944487477408,
      "grad_norm": 31.128684997558594,
      "learning_rate": 6.443298969072166e-06,
      "loss": 0.1589,
      "step": 125
    },
    {
      "epoch": 0.006506584043377227,
      "grad_norm": 16.766277313232422,
      "learning_rate": 6.494845360824743e-06,
      "loss": 0.0833,
      "step": 126
    },
    {
      "epoch": 0.006558223599277047,
      "grad_norm": 22.7692813873291,
      "learning_rate": 6.54639175257732e-06,
      "loss": 0.1153,
      "step": 127
    },
    {
      "epoch": 0.006609863155176866,
      "grad_norm": 12.015410423278809,
      "learning_rate": 6.597938144329898e-06,
      "loss": 0.1505,
      "step": 128
    },
    {
      "epoch": 0.006661502711076685,
      "grad_norm": 4.090978145599365,
      "learning_rate": 6.649484536082474e-06,
      "loss": 0.111,
      "step": 129
    },
    {
      "epoch": 0.006713142266976504,
      "grad_norm": 7.551740646362305,
      "learning_rate": 6.701030927835052e-06,
      "loss": 0.1255,
      "step": 130
    },
    {
      "epoch": 0.006764781822876324,
      "grad_norm": 18.28023910522461,
      "learning_rate": 6.75257731958763e-06,
      "loss": 0.1589,
      "step": 131
    },
    {
      "epoch": 0.006816421378776143,
      "grad_norm": 33.776939392089844,
      "learning_rate": 6.804123711340207e-06,
      "loss": 0.2154,
      "step": 132
    },
    {
      "epoch": 0.006868060934675962,
      "grad_norm": 15.398185729980469,
      "learning_rate": 6.855670103092784e-06,
      "loss": 0.1222,
      "step": 133
    },
    {
      "epoch": 0.006919700490575781,
      "grad_norm": 56.12717819213867,
      "learning_rate": 6.907216494845361e-06,
      "loss": 0.2544,
      "step": 134
    },
    {
      "epoch": 0.006971340046475601,
      "grad_norm": 13.180216789245605,
      "learning_rate": 6.958762886597939e-06,
      "loss": 0.1757,
      "step": 135
    },
    {
      "epoch": 0.00702297960237542,
      "grad_norm": 21.102319717407227,
      "learning_rate": 7.010309278350515e-06,
      "loss": 0.1206,
      "step": 136
    },
    {
      "epoch": 0.007074619158275239,
      "grad_norm": 15.09914779663086,
      "learning_rate": 7.061855670103094e-06,
      "loss": 0.1213,
      "step": 137
    },
    {
      "epoch": 0.007126258714175058,
      "grad_norm": 7.538786888122559,
      "learning_rate": 7.113402061855671e-06,
      "loss": 0.0536,
      "step": 138
    },
    {
      "epoch": 0.007177898270074878,
      "grad_norm": 18.547950744628906,
      "learning_rate": 7.164948453608248e-06,
      "loss": 0.1318,
      "step": 139
    },
    {
      "epoch": 0.007229537825974697,
      "grad_norm": 24.347200393676758,
      "learning_rate": 7.216494845360825e-06,
      "loss": 0.2564,
      "step": 140
    },
    {
      "epoch": 0.007281177381874516,
      "grad_norm": 21.751266479492188,
      "learning_rate": 7.2680412371134026e-06,
      "loss": 0.1377,
      "step": 141
    },
    {
      "epoch": 0.007332816937774335,
      "grad_norm": 17.687992095947266,
      "learning_rate": 7.319587628865979e-06,
      "loss": 0.1254,
      "step": 142
    },
    {
      "epoch": 0.007384456493674155,
      "grad_norm": 23.3682918548584,
      "learning_rate": 7.3711340206185574e-06,
      "loss": 0.1474,
      "step": 143
    },
    {
      "epoch": 0.007436096049573974,
      "grad_norm": 34.79859924316406,
      "learning_rate": 7.422680412371135e-06,
      "loss": 0.2216,
      "step": 144
    },
    {
      "epoch": 0.007487735605473793,
      "grad_norm": 8.752110481262207,
      "learning_rate": 7.474226804123712e-06,
      "loss": 0.1174,
      "step": 145
    },
    {
      "epoch": 0.0075393751613736125,
      "grad_norm": 5.8171772956848145,
      "learning_rate": 7.525773195876289e-06,
      "loss": 0.1141,
      "step": 146
    },
    {
      "epoch": 0.007591014717273432,
      "grad_norm": 22.51607894897461,
      "learning_rate": 7.577319587628866e-06,
      "loss": 0.1855,
      "step": 147
    },
    {
      "epoch": 0.007642654273173251,
      "grad_norm": 8.971314430236816,
      "learning_rate": 7.628865979381444e-06,
      "loss": 0.1973,
      "step": 148
    },
    {
      "epoch": 0.00769429382907307,
      "grad_norm": 6.435577392578125,
      "learning_rate": 7.680412371134021e-06,
      "loss": 0.1105,
      "step": 149
    },
    {
      "epoch": 0.0077459333849728895,
      "grad_norm": 10.445040702819824,
      "learning_rate": 7.731958762886599e-06,
      "loss": 0.1081,
      "step": 150
    },
    {
      "epoch": 0.007797572940872709,
      "grad_norm": 6.77109432220459,
      "learning_rate": 7.783505154639176e-06,
      "loss": 0.091,
      "step": 151
    },
    {
      "epoch": 0.007849212496772527,
      "grad_norm": 19.688772201538086,
      "learning_rate": 7.835051546391754e-06,
      "loss": 0.1583,
      "step": 152
    },
    {
      "epoch": 0.007900852052672347,
      "grad_norm": 10.278484344482422,
      "learning_rate": 7.886597938144331e-06,
      "loss": 0.0917,
      "step": 153
    },
    {
      "epoch": 0.007952491608572166,
      "grad_norm": 15.34327220916748,
      "learning_rate": 7.938144329896907e-06,
      "loss": 0.2181,
      "step": 154
    },
    {
      "epoch": 0.008004131164471986,
      "grad_norm": 26.09906768798828,
      "learning_rate": 7.989690721649486e-06,
      "loss": 0.1245,
      "step": 155
    },
    {
      "epoch": 0.008055770720371804,
      "grad_norm": 16.510021209716797,
      "learning_rate": 8.041237113402063e-06,
      "loss": 0.0587,
      "step": 156
    },
    {
      "epoch": 0.008107410276271624,
      "grad_norm": 12.442309379577637,
      "learning_rate": 8.09278350515464e-06,
      "loss": 0.0862,
      "step": 157
    },
    {
      "epoch": 0.008159049832171443,
      "grad_norm": 15.944640159606934,
      "learning_rate": 8.144329896907216e-06,
      "loss": 0.1026,
      "step": 158
    },
    {
      "epoch": 0.008210689388071263,
      "grad_norm": 28.59109115600586,
      "learning_rate": 8.195876288659794e-06,
      "loss": 0.2038,
      "step": 159
    },
    {
      "epoch": 0.008262328943971081,
      "grad_norm": 33.630611419677734,
      "learning_rate": 8.247422680412371e-06,
      "loss": 0.1736,
      "step": 160
    },
    {
      "epoch": 0.008313968499870901,
      "grad_norm": 24.19276237487793,
      "learning_rate": 8.29896907216495e-06,
      "loss": 0.141,
      "step": 161
    },
    {
      "epoch": 0.00836560805577072,
      "grad_norm": 14.12147045135498,
      "learning_rate": 8.350515463917526e-06,
      "loss": 0.156,
      "step": 162
    },
    {
      "epoch": 0.00841724761167054,
      "grad_norm": 23.958498001098633,
      "learning_rate": 8.402061855670104e-06,
      "loss": 0.1575,
      "step": 163
    },
    {
      "epoch": 0.008468887167570358,
      "grad_norm": 26.457298278808594,
      "learning_rate": 8.453608247422681e-06,
      "loss": 0.2319,
      "step": 164
    },
    {
      "epoch": 0.008520526723470178,
      "grad_norm": 13.230968475341797,
      "learning_rate": 8.505154639175259e-06,
      "loss": 0.1793,
      "step": 165
    },
    {
      "epoch": 0.008572166279369997,
      "grad_norm": 9.68136215209961,
      "learning_rate": 8.556701030927836e-06,
      "loss": 0.1641,
      "step": 166
    },
    {
      "epoch": 0.008623805835269817,
      "grad_norm": 8.471491813659668,
      "learning_rate": 8.608247422680413e-06,
      "loss": 0.0868,
      "step": 167
    },
    {
      "epoch": 0.008675445391169635,
      "grad_norm": 12.195115089416504,
      "learning_rate": 8.65979381443299e-06,
      "loss": 0.1429,
      "step": 168
    },
    {
      "epoch": 0.008727084947069455,
      "grad_norm": 14.088054656982422,
      "learning_rate": 8.711340206185568e-06,
      "loss": 0.1665,
      "step": 169
    },
    {
      "epoch": 0.008778724502969274,
      "grad_norm": 29.61151123046875,
      "learning_rate": 8.762886597938146e-06,
      "loss": 0.2205,
      "step": 170
    },
    {
      "epoch": 0.008830364058869094,
      "grad_norm": 9.755768775939941,
      "learning_rate": 8.814432989690721e-06,
      "loss": 0.1262,
      "step": 171
    },
    {
      "epoch": 0.008882003614768912,
      "grad_norm": 38.25715255737305,
      "learning_rate": 8.865979381443299e-06,
      "loss": 0.2824,
      "step": 172
    },
    {
      "epoch": 0.008933643170668732,
      "grad_norm": 9.558396339416504,
      "learning_rate": 8.917525773195878e-06,
      "loss": 0.1857,
      "step": 173
    },
    {
      "epoch": 0.00898528272656855,
      "grad_norm": 19.081363677978516,
      "learning_rate": 8.969072164948455e-06,
      "loss": 0.1027,
      "step": 174
    },
    {
      "epoch": 0.009036922282468371,
      "grad_norm": 6.29206657409668,
      "learning_rate": 9.020618556701031e-06,
      "loss": 0.1279,
      "step": 175
    },
    {
      "epoch": 0.00908856183836819,
      "grad_norm": 5.9786505699157715,
      "learning_rate": 9.072164948453609e-06,
      "loss": 0.1323,
      "step": 176
    },
    {
      "epoch": 0.00914020139426801,
      "grad_norm": 15.601408958435059,
      "learning_rate": 9.123711340206186e-06,
      "loss": 0.1481,
      "step": 177
    },
    {
      "epoch": 0.009191840950167828,
      "grad_norm": 6.908166408538818,
      "learning_rate": 9.175257731958764e-06,
      "loss": 0.0673,
      "step": 178
    },
    {
      "epoch": 0.009243480506067648,
      "grad_norm": 26.444934844970703,
      "learning_rate": 9.226804123711341e-06,
      "loss": 0.202,
      "step": 179
    },
    {
      "epoch": 0.009295120061967466,
      "grad_norm": 4.882022380828857,
      "learning_rate": 9.278350515463918e-06,
      "loss": 0.0948,
      "step": 180
    },
    {
      "epoch": 0.009346759617867286,
      "grad_norm": 7.198930263519287,
      "learning_rate": 9.329896907216496e-06,
      "loss": 0.1181,
      "step": 181
    },
    {
      "epoch": 0.009398399173767105,
      "grad_norm": 22.072280883789062,
      "learning_rate": 9.381443298969073e-06,
      "loss": 0.0946,
      "step": 182
    },
    {
      "epoch": 0.009450038729666925,
      "grad_norm": 12.008312225341797,
      "learning_rate": 9.43298969072165e-06,
      "loss": 0.097,
      "step": 183
    },
    {
      "epoch": 0.009501678285566743,
      "grad_norm": 11.344054222106934,
      "learning_rate": 9.484536082474226e-06,
      "loss": 0.096,
      "step": 184
    },
    {
      "epoch": 0.009553317841466564,
      "grad_norm": 11.103641510009766,
      "learning_rate": 9.536082474226806e-06,
      "loss": 0.0971,
      "step": 185
    },
    {
      "epoch": 0.009604957397366382,
      "grad_norm": 19.186378479003906,
      "learning_rate": 9.587628865979383e-06,
      "loss": 0.1276,
      "step": 186
    },
    {
      "epoch": 0.009656596953266202,
      "grad_norm": 7.595647811889648,
      "learning_rate": 9.63917525773196e-06,
      "loss": 0.1109,
      "step": 187
    },
    {
      "epoch": 0.00970823650916602,
      "grad_norm": 15.829177856445312,
      "learning_rate": 9.690721649484536e-06,
      "loss": 0.1229,
      "step": 188
    },
    {
      "epoch": 0.00975987606506584,
      "grad_norm": 40.41967010498047,
      "learning_rate": 9.742268041237114e-06,
      "loss": 0.2046,
      "step": 189
    },
    {
      "epoch": 0.009811515620965659,
      "grad_norm": 41.628543853759766,
      "learning_rate": 9.793814432989691e-06,
      "loss": 0.2241,
      "step": 190
    },
    {
      "epoch": 0.009863155176865479,
      "grad_norm": 17.800569534301758,
      "learning_rate": 9.84536082474227e-06,
      "loss": 0.1603,
      "step": 191
    },
    {
      "epoch": 0.009914794732765297,
      "grad_norm": 7.987938404083252,
      "learning_rate": 9.896907216494846e-06,
      "loss": 0.127,
      "step": 192
    },
    {
      "epoch": 0.009966434288665118,
      "grad_norm": 35.01975631713867,
      "learning_rate": 9.948453608247423e-06,
      "loss": 0.2215,
      "step": 193
    },
    {
      "epoch": 0.010018073844564936,
      "grad_norm": 37.709381103515625,
      "learning_rate": 1e-05,
      "loss": 0.2059,
      "step": 194
    },
    {
      "epoch": 0.010069713400464756,
      "grad_norm": 14.37966251373291,
      "learning_rate": 9.999999932864811e-06,
      "loss": 0.1296,
      "step": 195
    },
    {
      "epoch": 0.010121352956364574,
      "grad_norm": 29.046964645385742,
      "learning_rate": 9.999999731459245e-06,
      "loss": 0.1992,
      "step": 196
    },
    {
      "epoch": 0.010172992512264395,
      "grad_norm": 11.284933090209961,
      "learning_rate": 9.999999395783307e-06,
      "loss": 0.1107,
      "step": 197
    },
    {
      "epoch": 0.010224632068164213,
      "grad_norm": 10.117998123168945,
      "learning_rate": 9.999998925837007e-06,
      "loss": 0.1401,
      "step": 198
    },
    {
      "epoch": 0.010276271624064033,
      "grad_norm": 22.498088836669922,
      "learning_rate": 9.999998321620355e-06,
      "loss": 0.0869,
      "step": 199
    },
    {
      "epoch": 0.010327911179963851,
      "grad_norm": 18.761680603027344,
      "learning_rate": 9.99999758313337e-06,
      "loss": 0.0951,
      "step": 200
    },
    {
      "epoch": 0.010379550735863672,
      "grad_norm": 9.443952560424805,
      "learning_rate": 9.999996710376072e-06,
      "loss": 0.0732,
      "step": 201
    },
    {
      "epoch": 0.01043119029176349,
      "grad_norm": 9.90606689453125,
      "learning_rate": 9.999995703348482e-06,
      "loss": 0.0613,
      "step": 202
    },
    {
      "epoch": 0.01048282984766331,
      "grad_norm": 49.51936721801758,
      "learning_rate": 9.99999456205063e-06,
      "loss": 0.1636,
      "step": 203
    },
    {
      "epoch": 0.010534469403563129,
      "grad_norm": 36.0008430480957,
      "learning_rate": 9.999993286482545e-06,
      "loss": 0.1934,
      "step": 204
    },
    {
      "epoch": 0.010586108959462949,
      "grad_norm": 41.58720016479492,
      "learning_rate": 9.99999187664426e-06,
      "loss": 0.1789,
      "step": 205
    },
    {
      "epoch": 0.010637748515362767,
      "grad_norm": 11.938361167907715,
      "learning_rate": 9.999990332535816e-06,
      "loss": 0.1267,
      "step": 206
    },
    {
      "epoch": 0.010689388071262587,
      "grad_norm": 33.009700775146484,
      "learning_rate": 9.999988654157251e-06,
      "loss": 0.129,
      "step": 207
    },
    {
      "epoch": 0.010741027627162406,
      "grad_norm": 8.432607650756836,
      "learning_rate": 9.999986841508613e-06,
      "loss": 0.1569,
      "step": 208
    },
    {
      "epoch": 0.010792667183062226,
      "grad_norm": 10.464057922363281,
      "learning_rate": 9.99998489458995e-06,
      "loss": 0.0922,
      "step": 209
    },
    {
      "epoch": 0.010844306738962044,
      "grad_norm": 7.249148368835449,
      "learning_rate": 9.999982813401313e-06,
      "loss": 0.1195,
      "step": 210
    },
    {
      "epoch": 0.010895946294861864,
      "grad_norm": 15.076309204101562,
      "learning_rate": 9.999980597942758e-06,
      "loss": 0.1387,
      "step": 211
    },
    {
      "epoch": 0.010947585850761683,
      "grad_norm": 13.332361221313477,
      "learning_rate": 9.999978248214346e-06,
      "loss": 0.1524,
      "step": 212
    },
    {
      "epoch": 0.010999225406661503,
      "grad_norm": 5.2526631355285645,
      "learning_rate": 9.999975764216138e-06,
      "loss": 0.1151,
      "step": 213
    },
    {
      "epoch": 0.011050864962561323,
      "grad_norm": 9.555336952209473,
      "learning_rate": 9.999973145948204e-06,
      "loss": 0.1419,
      "step": 214
    },
    {
      "epoch": 0.011102504518461141,
      "grad_norm": 10.909151077270508,
      "learning_rate": 9.999970393410612e-06,
      "loss": 0.1221,
      "step": 215
    },
    {
      "epoch": 0.011154144074360961,
      "grad_norm": 32.13146209716797,
      "learning_rate": 9.999967506603435e-06,
      "loss": 0.1903,
      "step": 216
    },
    {
      "epoch": 0.01120578363026078,
      "grad_norm": 10.533029556274414,
      "learning_rate": 9.999964485526752e-06,
      "loss": 0.1322,
      "step": 217
    },
    {
      "epoch": 0.0112574231861606,
      "grad_norm": 48.03980255126953,
      "learning_rate": 9.999961330180643e-06,
      "loss": 0.1386,
      "step": 218
    },
    {
      "epoch": 0.011309062742060418,
      "grad_norm": 11.67361068725586,
      "learning_rate": 9.999958040565197e-06,
      "loss": 0.1272,
      "step": 219
    },
    {
      "epoch": 0.011360702297960238,
      "grad_norm": 7.972373008728027,
      "learning_rate": 9.999954616680496e-06,
      "loss": 0.0783,
      "step": 220
    },
    {
      "epoch": 0.011412341853860057,
      "grad_norm": 8.311938285827637,
      "learning_rate": 9.999951058526634e-06,
      "loss": 0.0601,
      "step": 221
    },
    {
      "epoch": 0.011463981409759877,
      "grad_norm": 7.425189971923828,
      "learning_rate": 9.999947366103709e-06,
      "loss": 0.072,
      "step": 222
    },
    {
      "epoch": 0.011515620965659695,
      "grad_norm": 9.216312408447266,
      "learning_rate": 9.999943539411818e-06,
      "loss": 0.0878,
      "step": 223
    },
    {
      "epoch": 0.011567260521559515,
      "grad_norm": 21.65372657775879,
      "learning_rate": 9.999939578451065e-06,
      "loss": 0.1473,
      "step": 224
    },
    {
      "epoch": 0.011618900077459334,
      "grad_norm": 25.643774032592773,
      "learning_rate": 9.999935483221556e-06,
      "loss": 0.2874,
      "step": 225
    },
    {
      "epoch": 0.011670539633359154,
      "grad_norm": 13.649435997009277,
      "learning_rate": 9.9999312537234e-06,
      "loss": 0.0716,
      "step": 226
    },
    {
      "epoch": 0.011722179189258972,
      "grad_norm": 39.2706184387207,
      "learning_rate": 9.99992688995671e-06,
      "loss": 0.2717,
      "step": 227
    },
    {
      "epoch": 0.011773818745158792,
      "grad_norm": 37.21033477783203,
      "learning_rate": 9.999922391921607e-06,
      "loss": 0.1626,
      "step": 228
    },
    {
      "epoch": 0.01182545830105861,
      "grad_norm": 18.96639633178711,
      "learning_rate": 9.999917759618207e-06,
      "loss": 0.1375,
      "step": 229
    },
    {
      "epoch": 0.011877097856958431,
      "grad_norm": 9.030610084533691,
      "learning_rate": 9.999912993046638e-06,
      "loss": 0.1065,
      "step": 230
    },
    {
      "epoch": 0.01192873741285825,
      "grad_norm": 11.053308486938477,
      "learning_rate": 9.999908092207027e-06,
      "loss": 0.1162,
      "step": 231
    },
    {
      "epoch": 0.01198037696875807,
      "grad_norm": 18.441356658935547,
      "learning_rate": 9.999903057099504e-06,
      "loss": 0.1053,
      "step": 232
    },
    {
      "epoch": 0.012032016524657888,
      "grad_norm": 9.130353927612305,
      "learning_rate": 9.999897887724206e-06,
      "loss": 0.0954,
      "step": 233
    },
    {
      "epoch": 0.012083656080557708,
      "grad_norm": 11.05713176727295,
      "learning_rate": 9.99989258408127e-06,
      "loss": 0.1176,
      "step": 234
    },
    {
      "epoch": 0.012135295636457526,
      "grad_norm": 17.10800552368164,
      "learning_rate": 9.999887146170841e-06,
      "loss": 0.2014,
      "step": 235
    },
    {
      "epoch": 0.012186935192357346,
      "grad_norm": 8.205573081970215,
      "learning_rate": 9.999881573993065e-06,
      "loss": 0.0661,
      "step": 236
    },
    {
      "epoch": 0.012238574748257165,
      "grad_norm": 9.793214797973633,
      "learning_rate": 9.999875867548087e-06,
      "loss": 0.106,
      "step": 237
    },
    {
      "epoch": 0.012290214304156985,
      "grad_norm": 5.333937168121338,
      "learning_rate": 9.999870026836066e-06,
      "loss": 0.0744,
      "step": 238
    },
    {
      "epoch": 0.012341853860056803,
      "grad_norm": 16.152050018310547,
      "learning_rate": 9.999864051857155e-06,
      "loss": 0.2105,
      "step": 239
    },
    {
      "epoch": 0.012393493415956624,
      "grad_norm": 4.425943851470947,
      "learning_rate": 9.999857942611518e-06,
      "loss": 0.0948,
      "step": 240
    },
    {
      "epoch": 0.012445132971856442,
      "grad_norm": 17.699235916137695,
      "learning_rate": 9.999851699099315e-06,
      "loss": 0.1342,
      "step": 241
    },
    {
      "epoch": 0.012496772527756262,
      "grad_norm": 8.45376968383789,
      "learning_rate": 9.999845321320715e-06,
      "loss": 0.0826,
      "step": 242
    },
    {
      "epoch": 0.01254841208365608,
      "grad_norm": 6.530562400817871,
      "learning_rate": 9.999838809275892e-06,
      "loss": 0.1157,
      "step": 243
    },
    {
      "epoch": 0.0126000516395559,
      "grad_norm": 8.514091491699219,
      "learning_rate": 9.999832162965019e-06,
      "loss": 0.0826,
      "step": 244
    },
    {
      "epoch": 0.012651691195455719,
      "grad_norm": 7.234795570373535,
      "learning_rate": 9.999825382388273e-06,
      "loss": 0.0871,
      "step": 245
    },
    {
      "epoch": 0.012703330751355539,
      "grad_norm": 12.943209648132324,
      "learning_rate": 9.999818467545837e-06,
      "loss": 0.0859,
      "step": 246
    },
    {
      "epoch": 0.012754970307255357,
      "grad_norm": 13.129032135009766,
      "learning_rate": 9.999811418437898e-06,
      "loss": 0.0922,
      "step": 247
    },
    {
      "epoch": 0.012806609863155178,
      "grad_norm": 9.882725715637207,
      "learning_rate": 9.999804235064643e-06,
      "loss": 0.1159,
      "step": 248
    },
    {
      "epoch": 0.012858249419054996,
      "grad_norm": 38.55314254760742,
      "learning_rate": 9.999796917426267e-06,
      "loss": 0.1524,
      "step": 249
    },
    {
      "epoch": 0.012909888974954816,
      "grad_norm": 19.44582176208496,
      "learning_rate": 9.999789465522967e-06,
      "loss": 0.1005,
      "step": 250
    },
    {
      "epoch": 0.012961528530854634,
      "grad_norm": 24.89863395690918,
      "learning_rate": 9.99978187935494e-06,
      "loss": 0.2146,
      "step": 251
    },
    {
      "epoch": 0.013013168086754455,
      "grad_norm": 15.435675621032715,
      "learning_rate": 9.999774158922393e-06,
      "loss": 0.1158,
      "step": 252
    },
    {
      "epoch": 0.013064807642654273,
      "grad_norm": 4.622774600982666,
      "learning_rate": 9.99976630422553e-06,
      "loss": 0.0798,
      "step": 253
    },
    {
      "epoch": 0.013116447198554093,
      "grad_norm": 4.962852478027344,
      "learning_rate": 9.999758315264566e-06,
      "loss": 0.1094,
      "step": 254
    },
    {
      "epoch": 0.013168086754453912,
      "grad_norm": 6.094913005828857,
      "learning_rate": 9.999750192039712e-06,
      "loss": 0.0763,
      "step": 255
    },
    {
      "epoch": 0.013219726310353732,
      "grad_norm": 37.34068298339844,
      "learning_rate": 9.999741934551187e-06,
      "loss": 0.2643,
      "step": 256
    },
    {
      "epoch": 0.01327136586625355,
      "grad_norm": 21.78446388244629,
      "learning_rate": 9.999733542799213e-06,
      "loss": 0.1205,
      "step": 257
    },
    {
      "epoch": 0.01332300542215337,
      "grad_norm": 5.846582889556885,
      "learning_rate": 9.999725016784017e-06,
      "loss": 0.1045,
      "step": 258
    },
    {
      "epoch": 0.013374644978053189,
      "grad_norm": 12.54537296295166,
      "learning_rate": 9.999716356505826e-06,
      "loss": 0.1142,
      "step": 259
    },
    {
      "epoch": 0.013426284533953009,
      "grad_norm": 11.519002914428711,
      "learning_rate": 9.999707561964874e-06,
      "loss": 0.1133,
      "step": 260
    },
    {
      "epoch": 0.013477924089852827,
      "grad_norm": 9.588472366333008,
      "learning_rate": 9.999698633161395e-06,
      "loss": 0.1307,
      "step": 261
    },
    {
      "epoch": 0.013529563645752647,
      "grad_norm": 18.402437210083008,
      "learning_rate": 9.99968957009563e-06,
      "loss": 0.1367,
      "step": 262
    },
    {
      "epoch": 0.013581203201652466,
      "grad_norm": 4.345107078552246,
      "learning_rate": 9.999680372767823e-06,
      "loss": 0.1111,
      "step": 263
    },
    {
      "epoch": 0.013632842757552286,
      "grad_norm": 13.820584297180176,
      "learning_rate": 9.999671041178221e-06,
      "loss": 0.1492,
      "step": 264
    },
    {
      "epoch": 0.013684482313452104,
      "grad_norm": 24.28551483154297,
      "learning_rate": 9.999661575327074e-06,
      "loss": 0.1586,
      "step": 265
    },
    {
      "epoch": 0.013736121869351924,
      "grad_norm": 20.930953979492188,
      "learning_rate": 9.999651975214634e-06,
      "loss": 0.1365,
      "step": 266
    },
    {
      "epoch": 0.013787761425251743,
      "grad_norm": 14.212982177734375,
      "learning_rate": 9.999642240841164e-06,
      "loss": 0.1095,
      "step": 267
    },
    {
      "epoch": 0.013839400981151563,
      "grad_norm": 3.4185128211975098,
      "learning_rate": 9.99963237220692e-06,
      "loss": 0.0836,
      "step": 268
    },
    {
      "epoch": 0.013891040537051381,
      "grad_norm": 17.86728858947754,
      "learning_rate": 9.99962236931217e-06,
      "loss": 0.1616,
      "step": 269
    },
    {
      "epoch": 0.013942680092951201,
      "grad_norm": 26.125591278076172,
      "learning_rate": 9.999612232157183e-06,
      "loss": 0.1999,
      "step": 270
    },
    {
      "epoch": 0.01399431964885102,
      "grad_norm": 8.479475975036621,
      "learning_rate": 9.99960196074223e-06,
      "loss": 0.0763,
      "step": 271
    },
    {
      "epoch": 0.01404595920475084,
      "grad_norm": 24.021392822265625,
      "learning_rate": 9.999591555067587e-06,
      "loss": 0.1563,
      "step": 272
    },
    {
      "epoch": 0.014097598760650658,
      "grad_norm": 17.387971878051758,
      "learning_rate": 9.999581015133532e-06,
      "loss": 0.174,
      "step": 273
    },
    {
      "epoch": 0.014149238316550478,
      "grad_norm": 8.603208541870117,
      "learning_rate": 9.999570340940353e-06,
      "loss": 0.0486,
      "step": 274
    },
    {
      "epoch": 0.014200877872450297,
      "grad_norm": 7.017289161682129,
      "learning_rate": 9.99955953248833e-06,
      "loss": 0.1194,
      "step": 275
    },
    {
      "epoch": 0.014252517428350117,
      "grad_norm": 5.310054779052734,
      "learning_rate": 9.999548589777758e-06,
      "loss": 0.0821,
      "step": 276
    },
    {
      "epoch": 0.014304156984249935,
      "grad_norm": 6.546644687652588,
      "learning_rate": 9.99953751280893e-06,
      "loss": 0.1004,
      "step": 277
    },
    {
      "epoch": 0.014355796540149755,
      "grad_norm": 4.293127536773682,
      "learning_rate": 9.999526301582141e-06,
      "loss": 0.0315,
      "step": 278
    },
    {
      "epoch": 0.014407436096049574,
      "grad_norm": 24.88428497314453,
      "learning_rate": 9.999514956097695e-06,
      "loss": 0.2357,
      "step": 279
    },
    {
      "epoch": 0.014459075651949394,
      "grad_norm": 3.610429286956787,
      "learning_rate": 9.999503476355894e-06,
      "loss": 0.034,
      "step": 280
    },
    {
      "epoch": 0.014510715207849212,
      "grad_norm": 4.161850929260254,
      "learning_rate": 9.999491862357049e-06,
      "loss": 0.0648,
      "step": 281
    },
    {
      "epoch": 0.014562354763749032,
      "grad_norm": 13.997310638427734,
      "learning_rate": 9.999480114101472e-06,
      "loss": 0.1021,
      "step": 282
    },
    {
      "epoch": 0.01461399431964885,
      "grad_norm": 24.589012145996094,
      "learning_rate": 9.999468231589476e-06,
      "loss": 0.3044,
      "step": 283
    },
    {
      "epoch": 0.01466563387554867,
      "grad_norm": 30.076644897460938,
      "learning_rate": 9.99945621482138e-06,
      "loss": 0.3047,
      "step": 284
    },
    {
      "epoch": 0.01471727343144849,
      "grad_norm": 6.001841068267822,
      "learning_rate": 9.99944406379751e-06,
      "loss": 0.068,
      "step": 285
    },
    {
      "epoch": 0.01476891298734831,
      "grad_norm": 21.53598403930664,
      "learning_rate": 9.99943177851819e-06,
      "loss": 0.1067,
      "step": 286
    },
    {
      "epoch": 0.014820552543248128,
      "grad_norm": 26.438024520874023,
      "learning_rate": 9.999419358983749e-06,
      "loss": 0.2406,
      "step": 287
    },
    {
      "epoch": 0.014872192099147948,
      "grad_norm": 11.010616302490234,
      "learning_rate": 9.999406805194523e-06,
      "loss": 0.1314,
      "step": 288
    },
    {
      "epoch": 0.014923831655047766,
      "grad_norm": 18.78911590576172,
      "learning_rate": 9.999394117150846e-06,
      "loss": 0.157,
      "step": 289
    },
    {
      "epoch": 0.014975471210947586,
      "grad_norm": 6.76024055480957,
      "learning_rate": 9.999381294853061e-06,
      "loss": 0.1112,
      "step": 290
    },
    {
      "epoch": 0.015027110766847405,
      "grad_norm": 8.964905738830566,
      "learning_rate": 9.999368338301514e-06,
      "loss": 0.128,
      "step": 291
    },
    {
      "epoch": 0.015078750322747225,
      "grad_norm": 8.138101577758789,
      "learning_rate": 9.999355247496549e-06,
      "loss": 0.0978,
      "step": 292
    },
    {
      "epoch": 0.015130389878647043,
      "grad_norm": 8.906871795654297,
      "learning_rate": 9.99934202243852e-06,
      "loss": 0.1888,
      "step": 293
    },
    {
      "epoch": 0.015182029434546863,
      "grad_norm": 6.523357391357422,
      "learning_rate": 9.99932866312778e-06,
      "loss": 0.1088,
      "step": 294
    },
    {
      "epoch": 0.015233668990446682,
      "grad_norm": 5.614871025085449,
      "learning_rate": 9.999315169564692e-06,
      "loss": 0.1334,
      "step": 295
    },
    {
      "epoch": 0.015285308546346502,
      "grad_norm": 12.402887344360352,
      "learning_rate": 9.999301541749614e-06,
      "loss": 0.156,
      "step": 296
    },
    {
      "epoch": 0.01533694810224632,
      "grad_norm": 7.2405900955200195,
      "learning_rate": 9.999287779682913e-06,
      "loss": 0.1022,
      "step": 297
    },
    {
      "epoch": 0.01538858765814614,
      "grad_norm": 7.025831699371338,
      "learning_rate": 9.999273883364959e-06,
      "loss": 0.1078,
      "step": 298
    },
    {
      "epoch": 0.015440227214045959,
      "grad_norm": 8.307042121887207,
      "learning_rate": 9.999259852796128e-06,
      "loss": 0.0924,
      "step": 299
    },
    {
      "epoch": 0.015491866769945779,
      "grad_norm": 19.204036712646484,
      "learning_rate": 9.999245687976791e-06,
      "loss": 0.2631,
      "step": 300
    },
    {
      "epoch": 0.015543506325845597,
      "grad_norm": 13.534343719482422,
      "learning_rate": 9.999231388907333e-06,
      "loss": 0.1673,
      "step": 301
    },
    {
      "epoch": 0.015595145881745417,
      "grad_norm": 13.22003173828125,
      "learning_rate": 9.999216955588135e-06,
      "loss": 0.1524,
      "step": 302
    },
    {
      "epoch": 0.015646785437645236,
      "grad_norm": 5.594106197357178,
      "learning_rate": 9.999202388019589e-06,
      "loss": 0.0659,
      "step": 303
    },
    {
      "epoch": 0.015698424993545054,
      "grad_norm": 8.673233985900879,
      "learning_rate": 9.99918768620208e-06,
      "loss": 0.0962,
      "step": 304
    },
    {
      "epoch": 0.015750064549444876,
      "grad_norm": 12.684235572814941,
      "learning_rate": 9.99917285013601e-06,
      "loss": 0.1541,
      "step": 305
    },
    {
      "epoch": 0.015801704105344695,
      "grad_norm": 8.134642601013184,
      "learning_rate": 9.99915787982177e-06,
      "loss": 0.1168,
      "step": 306
    },
    {
      "epoch": 0.015853343661244513,
      "grad_norm": 17.34668731689453,
      "learning_rate": 9.999142775259767e-06,
      "loss": 0.1797,
      "step": 307
    },
    {
      "epoch": 0.01590498321714433,
      "grad_norm": 4.937404632568359,
      "learning_rate": 9.999127536450403e-06,
      "loss": 0.072,
      "step": 308
    },
    {
      "epoch": 0.015956622773044153,
      "grad_norm": 10.710586547851562,
      "learning_rate": 9.999112163394092e-06,
      "loss": 0.1334,
      "step": 309
    },
    {
      "epoch": 0.01600826232894397,
      "grad_norm": 6.9057135581970215,
      "learning_rate": 9.999096656091243e-06,
      "loss": 0.0771,
      "step": 310
    },
    {
      "epoch": 0.01605990188484379,
      "grad_norm": 11.050850868225098,
      "learning_rate": 9.999081014542274e-06,
      "loss": 0.0898,
      "step": 311
    },
    {
      "epoch": 0.01611154144074361,
      "grad_norm": 17.627029418945312,
      "learning_rate": 9.999065238747605e-06,
      "loss": 0.1286,
      "step": 312
    },
    {
      "epoch": 0.01616318099664343,
      "grad_norm": 14.302743911743164,
      "learning_rate": 9.999049328707657e-06,
      "loss": 0.1148,
      "step": 313
    },
    {
      "epoch": 0.01621482055254325,
      "grad_norm": 17.83171272277832,
      "learning_rate": 9.999033284422863e-06,
      "loss": 0.1423,
      "step": 314
    },
    {
      "epoch": 0.016266460108443067,
      "grad_norm": 6.618483543395996,
      "learning_rate": 9.999017105893649e-06,
      "loss": 0.0759,
      "step": 315
    },
    {
      "epoch": 0.016318099664342885,
      "grad_norm": 18.665203094482422,
      "learning_rate": 9.999000793120449e-06,
      "loss": 0.144,
      "step": 316
    },
    {
      "epoch": 0.016369739220242707,
      "grad_norm": 38.835533142089844,
      "learning_rate": 9.998984346103703e-06,
      "loss": 0.1981,
      "step": 317
    },
    {
      "epoch": 0.016421378776142526,
      "grad_norm": 31.776832580566406,
      "learning_rate": 9.998967764843855e-06,
      "loss": 0.2409,
      "step": 318
    },
    {
      "epoch": 0.016473018332042344,
      "grad_norm": 14.646016120910645,
      "learning_rate": 9.998951049341348e-06,
      "loss": 0.1043,
      "step": 319
    },
    {
      "epoch": 0.016524657887942162,
      "grad_norm": 36.23216247558594,
      "learning_rate": 9.998934199596629e-06,
      "loss": 0.2675,
      "step": 320
    },
    {
      "epoch": 0.016576297443841984,
      "grad_norm": 21.602018356323242,
      "learning_rate": 9.998917215610153e-06,
      "loss": 0.1503,
      "step": 321
    },
    {
      "epoch": 0.016627936999741803,
      "grad_norm": 7.502382755279541,
      "learning_rate": 9.998900097382373e-06,
      "loss": 0.0948,
      "step": 322
    },
    {
      "epoch": 0.01667957655564162,
      "grad_norm": 25.088001251220703,
      "learning_rate": 9.998882844913755e-06,
      "loss": 0.1413,
      "step": 323
    },
    {
      "epoch": 0.01673121611154144,
      "grad_norm": 32.51912307739258,
      "learning_rate": 9.998865458204756e-06,
      "loss": 0.2611,
      "step": 324
    },
    {
      "epoch": 0.01678285566744126,
      "grad_norm": 21.996585845947266,
      "learning_rate": 9.998847937255844e-06,
      "loss": 0.1768,
      "step": 325
    },
    {
      "epoch": 0.01683449522334108,
      "grad_norm": 41.177242279052734,
      "learning_rate": 9.998830282067493e-06,
      "loss": 0.2007,
      "step": 326
    },
    {
      "epoch": 0.016886134779240898,
      "grad_norm": 32.52981948852539,
      "learning_rate": 9.998812492640174e-06,
      "loss": 0.2907,
      "step": 327
    },
    {
      "epoch": 0.016937774335140716,
      "grad_norm": 9.30726146697998,
      "learning_rate": 9.998794568974365e-06,
      "loss": 0.1342,
      "step": 328
    },
    {
      "epoch": 0.01698941389104054,
      "grad_norm": 4.2823686599731445,
      "learning_rate": 9.998776511070548e-06,
      "loss": 0.1069,
      "step": 329
    },
    {
      "epoch": 0.017041053446940357,
      "grad_norm": 7.465283393859863,
      "learning_rate": 9.99875831892921e-06,
      "loss": 0.0918,
      "step": 330
    },
    {
      "epoch": 0.017092693002840175,
      "grad_norm": 5.722858905792236,
      "learning_rate": 9.998739992550834e-06,
      "loss": 0.0694,
      "step": 331
    },
    {
      "epoch": 0.017144332558739993,
      "grad_norm": 6.753835201263428,
      "learning_rate": 9.998721531935917e-06,
      "loss": 0.0851,
      "step": 332
    },
    {
      "epoch": 0.017195972114639815,
      "grad_norm": 27.466909408569336,
      "learning_rate": 9.998702937084954e-06,
      "loss": 0.163,
      "step": 333
    },
    {
      "epoch": 0.017247611670539634,
      "grad_norm": 20.13273048400879,
      "learning_rate": 9.998684207998443e-06,
      "loss": 0.1759,
      "step": 334
    },
    {
      "epoch": 0.017299251226439452,
      "grad_norm": 9.254204750061035,
      "learning_rate": 9.998665344676887e-06,
      "loss": 0.1343,
      "step": 335
    },
    {
      "epoch": 0.01735089078233927,
      "grad_norm": 24.64960289001465,
      "learning_rate": 9.998646347120794e-06,
      "loss": 0.1911,
      "step": 336
    },
    {
      "epoch": 0.017402530338239092,
      "grad_norm": 17.566349029541016,
      "learning_rate": 9.998627215330671e-06,
      "loss": 0.1622,
      "step": 337
    },
    {
      "epoch": 0.01745416989413891,
      "grad_norm": 7.801143169403076,
      "learning_rate": 9.998607949307036e-06,
      "loss": 0.1336,
      "step": 338
    },
    {
      "epoch": 0.01750580945003873,
      "grad_norm": 14.900544166564941,
      "learning_rate": 9.998588549050404e-06,
      "loss": 0.1136,
      "step": 339
    },
    {
      "epoch": 0.017557449005938548,
      "grad_norm": 9.415428161621094,
      "learning_rate": 9.998569014561297e-06,
      "loss": 0.0721,
      "step": 340
    },
    {
      "epoch": 0.01760908856183837,
      "grad_norm": 10.380762100219727,
      "learning_rate": 9.998549345840239e-06,
      "loss": 0.1425,
      "step": 341
    },
    {
      "epoch": 0.017660728117738188,
      "grad_norm": 7.175077438354492,
      "learning_rate": 9.998529542887756e-06,
      "loss": 0.1036,
      "step": 342
    },
    {
      "epoch": 0.017712367673638006,
      "grad_norm": 13.045048713684082,
      "learning_rate": 9.998509605704383e-06,
      "loss": 0.0862,
      "step": 343
    },
    {
      "epoch": 0.017764007229537825,
      "grad_norm": 19.894134521484375,
      "learning_rate": 9.998489534290654e-06,
      "loss": 0.1169,
      "step": 344
    },
    {
      "epoch": 0.017815646785437646,
      "grad_norm": 28.85492706298828,
      "learning_rate": 9.99846932864711e-06,
      "loss": 0.2348,
      "step": 345
    },
    {
      "epoch": 0.017867286341337465,
      "grad_norm": 11.613121032714844,
      "learning_rate": 9.998448988774289e-06,
      "loss": 0.1341,
      "step": 346
    },
    {
      "epoch": 0.017918925897237283,
      "grad_norm": 10.074936866760254,
      "learning_rate": 9.998428514672742e-06,
      "loss": 0.1175,
      "step": 347
    },
    {
      "epoch": 0.0179705654531371,
      "grad_norm": 15.10318660736084,
      "learning_rate": 9.998407906343017e-06,
      "loss": 0.1027,
      "step": 348
    },
    {
      "epoch": 0.018022205009036923,
      "grad_norm": 6.825178623199463,
      "learning_rate": 9.998387163785667e-06,
      "loss": 0.1243,
      "step": 349
    },
    {
      "epoch": 0.018073844564936742,
      "grad_norm": 9.804069519042969,
      "learning_rate": 9.998366287001249e-06,
      "loss": 0.0965,
      "step": 350
    },
    {
      "epoch": 0.01812548412083656,
      "grad_norm": 7.033784866333008,
      "learning_rate": 9.998345275990324e-06,
      "loss": 0.1408,
      "step": 351
    },
    {
      "epoch": 0.01817712367673638,
      "grad_norm": 4.864927768707275,
      "learning_rate": 9.998324130753457e-06,
      "loss": 0.0946,
      "step": 352
    },
    {
      "epoch": 0.0182287632326362,
      "grad_norm": 13.85106086730957,
      "learning_rate": 9.998302851291215e-06,
      "loss": 0.1172,
      "step": 353
    },
    {
      "epoch": 0.01828040278853602,
      "grad_norm": 12.242929458618164,
      "learning_rate": 9.99828143760417e-06,
      "loss": 0.125,
      "step": 354
    },
    {
      "epoch": 0.018332042344435837,
      "grad_norm": 15.313193321228027,
      "learning_rate": 9.998259889692894e-06,
      "loss": 0.1016,
      "step": 355
    },
    {
      "epoch": 0.018383681900335656,
      "grad_norm": 15.326311111450195,
      "learning_rate": 9.99823820755797e-06,
      "loss": 0.1697,
      "step": 356
    },
    {
      "epoch": 0.018435321456235478,
      "grad_norm": 9.838823318481445,
      "learning_rate": 9.998216391199979e-06,
      "loss": 0.0732,
      "step": 357
    },
    {
      "epoch": 0.018486961012135296,
      "grad_norm": 67.01812744140625,
      "learning_rate": 9.998194440619504e-06,
      "loss": 0.1916,
      "step": 358
    },
    {
      "epoch": 0.018538600568035114,
      "grad_norm": 16.33999252319336,
      "learning_rate": 9.99817235581714e-06,
      "loss": 0.2974,
      "step": 359
    },
    {
      "epoch": 0.018590240123934933,
      "grad_norm": 42.344905853271484,
      "learning_rate": 9.998150136793475e-06,
      "loss": 0.3304,
      "step": 360
    },
    {
      "epoch": 0.018641879679834755,
      "grad_norm": 14.757149696350098,
      "learning_rate": 9.998127783549107e-06,
      "loss": 0.1042,
      "step": 361
    },
    {
      "epoch": 0.018693519235734573,
      "grad_norm": 5.670040130615234,
      "learning_rate": 9.998105296084636e-06,
      "loss": 0.0731,
      "step": 362
    },
    {
      "epoch": 0.01874515879163439,
      "grad_norm": 10.958182334899902,
      "learning_rate": 9.998082674400669e-06,
      "loss": 0.1372,
      "step": 363
    },
    {
      "epoch": 0.01879679834753421,
      "grad_norm": 7.005664348602295,
      "learning_rate": 9.99805991849781e-06,
      "loss": 0.1166,
      "step": 364
    },
    {
      "epoch": 0.01884843790343403,
      "grad_norm": 14.996322631835938,
      "learning_rate": 9.998037028376672e-06,
      "loss": 0.1887,
      "step": 365
    },
    {
      "epoch": 0.01890007745933385,
      "grad_norm": 24.673765182495117,
      "learning_rate": 9.998014004037869e-06,
      "loss": 0.0843,
      "step": 366
    },
    {
      "epoch": 0.01895171701523367,
      "grad_norm": 9.441974639892578,
      "learning_rate": 9.997990845482017e-06,
      "loss": 0.1232,
      "step": 367
    },
    {
      "epoch": 0.019003356571133487,
      "grad_norm": 6.611137866973877,
      "learning_rate": 9.997967552709741e-06,
      "loss": 0.0686,
      "step": 368
    },
    {
      "epoch": 0.01905499612703331,
      "grad_norm": 8.048462867736816,
      "learning_rate": 9.997944125721666e-06,
      "loss": 0.1321,
      "step": 369
    },
    {
      "epoch": 0.019106635682933127,
      "grad_norm": 3.0205495357513428,
      "learning_rate": 9.997920564518422e-06,
      "loss": 0.0803,
      "step": 370
    },
    {
      "epoch": 0.019158275238832945,
      "grad_norm": 10.812738418579102,
      "learning_rate": 9.99789686910064e-06,
      "loss": 0.1354,
      "step": 371
    },
    {
      "epoch": 0.019209914794732764,
      "grad_norm": 7.724452495574951,
      "learning_rate": 9.997873039468954e-06,
      "loss": 0.09,
      "step": 372
    },
    {
      "epoch": 0.019261554350632586,
      "grad_norm": 4.808595180511475,
      "learning_rate": 9.997849075624011e-06,
      "loss": 0.0774,
      "step": 373
    },
    {
      "epoch": 0.019313193906532404,
      "grad_norm": 6.970510482788086,
      "learning_rate": 9.997824977566447e-06,
      "loss": 0.0855,
      "step": 374
    },
    {
      "epoch": 0.019364833462432222,
      "grad_norm": 12.346540451049805,
      "learning_rate": 9.997800745296914e-06,
      "loss": 0.0681,
      "step": 375
    },
    {
      "epoch": 0.01941647301833204,
      "grad_norm": 11.915414810180664,
      "learning_rate": 9.99777637881606e-06,
      "loss": 0.0867,
      "step": 376
    },
    {
      "epoch": 0.019468112574231863,
      "grad_norm": 6.757781505584717,
      "learning_rate": 9.99775187812454e-06,
      "loss": 0.1056,
      "step": 377
    },
    {
      "epoch": 0.01951975213013168,
      "grad_norm": 25.082059860229492,
      "learning_rate": 9.997727243223015e-06,
      "loss": 0.1062,
      "step": 378
    },
    {
      "epoch": 0.0195713916860315,
      "grad_norm": 18.497692108154297,
      "learning_rate": 9.997702474112142e-06,
      "loss": 0.13,
      "step": 379
    },
    {
      "epoch": 0.019623031241931318,
      "grad_norm": 18.001497268676758,
      "learning_rate": 9.997677570792587e-06,
      "loss": 0.195,
      "step": 380
    },
    {
      "epoch": 0.01967467079783114,
      "grad_norm": 14.321142196655273,
      "learning_rate": 9.997652533265021e-06,
      "loss": 0.1187,
      "step": 381
    },
    {
      "epoch": 0.019726310353730958,
      "grad_norm": 9.309700965881348,
      "learning_rate": 9.997627361530116e-06,
      "loss": 0.1118,
      "step": 382
    },
    {
      "epoch": 0.019777949909630776,
      "grad_norm": 27.14641761779785,
      "learning_rate": 9.997602055588547e-06,
      "loss": 0.1394,
      "step": 383
    },
    {
      "epoch": 0.019829589465530595,
      "grad_norm": 13.872747421264648,
      "learning_rate": 9.997576615440992e-06,
      "loss": 0.0881,
      "step": 384
    },
    {
      "epoch": 0.019881229021430417,
      "grad_norm": 7.571065425872803,
      "learning_rate": 9.997551041088137e-06,
      "loss": 0.1071,
      "step": 385
    },
    {
      "epoch": 0.019932868577330235,
      "grad_norm": 9.406332969665527,
      "learning_rate": 9.997525332530668e-06,
      "loss": 0.1096,
      "step": 386
    },
    {
      "epoch": 0.019984508133230054,
      "grad_norm": 10.985230445861816,
      "learning_rate": 9.997499489769274e-06,
      "loss": 0.0891,
      "step": 387
    },
    {
      "epoch": 0.020036147689129872,
      "grad_norm": 29.810497283935547,
      "learning_rate": 9.997473512804653e-06,
      "loss": 0.2531,
      "step": 388
    },
    {
      "epoch": 0.020087787245029694,
      "grad_norm": 22.303865432739258,
      "learning_rate": 9.997447401637496e-06,
      "loss": 0.1651,
      "step": 389
    },
    {
      "epoch": 0.020139426800929512,
      "grad_norm": 21.7547664642334,
      "learning_rate": 9.997421156268508e-06,
      "loss": 0.1708,
      "step": 390
    },
    {
      "epoch": 0.02019106635682933,
      "grad_norm": 4.276151657104492,
      "learning_rate": 9.997394776698395e-06,
      "loss": 0.0783,
      "step": 391
    },
    {
      "epoch": 0.02024270591272915,
      "grad_norm": 10.791491508483887,
      "learning_rate": 9.997368262927864e-06,
      "loss": 0.1193,
      "step": 392
    },
    {
      "epoch": 0.02029434546862897,
      "grad_norm": 24.49895668029785,
      "learning_rate": 9.997341614957626e-06,
      "loss": 0.1391,
      "step": 393
    },
    {
      "epoch": 0.02034598502452879,
      "grad_norm": 20.48058319091797,
      "learning_rate": 9.9973148327884e-06,
      "loss": 0.1485,
      "step": 394
    },
    {
      "epoch": 0.020397624580428608,
      "grad_norm": 7.04456090927124,
      "learning_rate": 9.9972879164209e-06,
      "loss": 0.0943,
      "step": 395
    },
    {
      "epoch": 0.020449264136328426,
      "grad_norm": 3.294736385345459,
      "learning_rate": 9.997260865855853e-06,
      "loss": 0.0676,
      "step": 396
    },
    {
      "epoch": 0.020500903692228248,
      "grad_norm": 3.8832812309265137,
      "learning_rate": 9.997233681093984e-06,
      "loss": 0.075,
      "step": 397
    },
    {
      "epoch": 0.020552543248128066,
      "grad_norm": 14.853604316711426,
      "learning_rate": 9.997206362136022e-06,
      "loss": 0.1301,
      "step": 398
    },
    {
      "epoch": 0.020604182804027885,
      "grad_norm": 19.36187171936035,
      "learning_rate": 9.997178908982704e-06,
      "loss": 0.157,
      "step": 399
    },
    {
      "epoch": 0.020655822359927703,
      "grad_norm": 5.767361164093018,
      "learning_rate": 9.997151321634762e-06,
      "loss": 0.1402,
      "step": 400
    },
    {
      "epoch": 0.020707461915827525,
      "grad_norm": 9.692453384399414,
      "learning_rate": 9.997123600092942e-06,
      "loss": 0.1302,
      "step": 401
    },
    {
      "epoch": 0.020759101471727343,
      "grad_norm": 10.823790550231934,
      "learning_rate": 9.997095744357984e-06,
      "loss": 0.0673,
      "step": 402
    },
    {
      "epoch": 0.02081074102762716,
      "grad_norm": 9.872780799865723,
      "learning_rate": 9.997067754430641e-06,
      "loss": 0.106,
      "step": 403
    },
    {
      "epoch": 0.02086238058352698,
      "grad_norm": 8.685038566589355,
      "learning_rate": 9.997039630311659e-06,
      "loss": 0.0781,
      "step": 404
    },
    {
      "epoch": 0.020914020139426802,
      "grad_norm": 6.771312713623047,
      "learning_rate": 9.997011372001798e-06,
      "loss": 0.0252,
      "step": 405
    },
    {
      "epoch": 0.02096565969532662,
      "grad_norm": 14.544816970825195,
      "learning_rate": 9.996982979501814e-06,
      "loss": 0.0758,
      "step": 406
    },
    {
      "epoch": 0.02101729925122644,
      "grad_norm": 14.603408813476562,
      "learning_rate": 9.99695445281247e-06,
      "loss": 0.123,
      "step": 407
    },
    {
      "epoch": 0.021068938807126257,
      "grad_norm": 7.56645393371582,
      "learning_rate": 9.996925791934536e-06,
      "loss": 0.0881,
      "step": 408
    },
    {
      "epoch": 0.02112057836302608,
      "grad_norm": 13.42009162902832,
      "learning_rate": 9.996896996868774e-06,
      "loss": 0.1351,
      "step": 409
    },
    {
      "epoch": 0.021172217918925897,
      "grad_norm": 7.904331684112549,
      "learning_rate": 9.996868067615963e-06,
      "loss": 0.0627,
      "step": 410
    },
    {
      "epoch": 0.021223857474825716,
      "grad_norm": 10.172428131103516,
      "learning_rate": 9.99683900417688e-06,
      "loss": 0.0971,
      "step": 411
    },
    {
      "epoch": 0.021275497030725534,
      "grad_norm": 18.051517486572266,
      "learning_rate": 9.996809806552301e-06,
      "loss": 0.1706,
      "step": 412
    },
    {
      "epoch": 0.021327136586625356,
      "grad_norm": 33.455482482910156,
      "learning_rate": 9.996780474743013e-06,
      "loss": 0.1743,
      "step": 413
    },
    {
      "epoch": 0.021378776142525174,
      "grad_norm": 5.754262924194336,
      "learning_rate": 9.996751008749806e-06,
      "loss": 0.0699,
      "step": 414
    },
    {
      "epoch": 0.021430415698424993,
      "grad_norm": 12.862781524658203,
      "learning_rate": 9.996721408573467e-06,
      "loss": 0.0944,
      "step": 415
    },
    {
      "epoch": 0.02148205525432481,
      "grad_norm": 8.74370288848877,
      "learning_rate": 9.996691674214791e-06,
      "loss": 0.1206,
      "step": 416
    },
    {
      "epoch": 0.021533694810224633,
      "grad_norm": 16.886606216430664,
      "learning_rate": 9.996661805674582e-06,
      "loss": 0.1328,
      "step": 417
    },
    {
      "epoch": 0.02158533436612445,
      "grad_norm": 10.108582496643066,
      "learning_rate": 9.996631802953634e-06,
      "loss": 0.0804,
      "step": 418
    },
    {
      "epoch": 0.02163697392202427,
      "grad_norm": 18.77193260192871,
      "learning_rate": 9.99660166605276e-06,
      "loss": 0.1254,
      "step": 419
    },
    {
      "epoch": 0.021688613477924088,
      "grad_norm": 6.81450891494751,
      "learning_rate": 9.996571394972765e-06,
      "loss": 0.084,
      "step": 420
    },
    {
      "epoch": 0.02174025303382391,
      "grad_norm": 5.898200035095215,
      "learning_rate": 9.996540989714464e-06,
      "loss": 0.1256,
      "step": 421
    },
    {
      "epoch": 0.02179189258972373,
      "grad_norm": 7.039751052856445,
      "learning_rate": 9.996510450278672e-06,
      "loss": 0.087,
      "step": 422
    },
    {
      "epoch": 0.021843532145623547,
      "grad_norm": 10.412111282348633,
      "learning_rate": 9.99647977666621e-06,
      "loss": 0.1043,
      "step": 423
    },
    {
      "epoch": 0.021895171701523365,
      "grad_norm": 8.930888175964355,
      "learning_rate": 9.9964489688779e-06,
      "loss": 0.0867,
      "step": 424
    },
    {
      "epoch": 0.021946811257423187,
      "grad_norm": 15.340171813964844,
      "learning_rate": 9.996418026914574e-06,
      "loss": 0.1289,
      "step": 425
    },
    {
      "epoch": 0.021998450813323005,
      "grad_norm": 7.914361476898193,
      "learning_rate": 9.996386950777058e-06,
      "loss": 0.0632,
      "step": 426
    },
    {
      "epoch": 0.022050090369222824,
      "grad_norm": 7.893770217895508,
      "learning_rate": 9.996355740466189e-06,
      "loss": 0.1612,
      "step": 427
    },
    {
      "epoch": 0.022101729925122646,
      "grad_norm": 12.19672966003418,
      "learning_rate": 9.996324395982802e-06,
      "loss": 0.1061,
      "step": 428
    },
    {
      "epoch": 0.022153369481022464,
      "grad_norm": 6.550032138824463,
      "learning_rate": 9.996292917327743e-06,
      "loss": 0.1065,
      "step": 429
    },
    {
      "epoch": 0.022205009036922282,
      "grad_norm": 7.044384479522705,
      "learning_rate": 9.996261304501856e-06,
      "loss": 0.1349,
      "step": 430
    },
    {
      "epoch": 0.0222566485928221,
      "grad_norm": 12.376392364501953,
      "learning_rate": 9.996229557505988e-06,
      "loss": 0.1291,
      "step": 431
    },
    {
      "epoch": 0.022308288148721923,
      "grad_norm": 5.430670261383057,
      "learning_rate": 9.996197676340993e-06,
      "loss": 0.0755,
      "step": 432
    },
    {
      "epoch": 0.02235992770462174,
      "grad_norm": 7.82625150680542,
      "learning_rate": 9.996165661007726e-06,
      "loss": 0.061,
      "step": 433
    },
    {
      "epoch": 0.02241156726052156,
      "grad_norm": 6.756915092468262,
      "learning_rate": 9.996133511507048e-06,
      "loss": 0.054,
      "step": 434
    },
    {
      "epoch": 0.022463206816421378,
      "grad_norm": 8.417162895202637,
      "learning_rate": 9.996101227839825e-06,
      "loss": 0.0781,
      "step": 435
    },
    {
      "epoch": 0.0225148463723212,
      "grad_norm": 10.190946578979492,
      "learning_rate": 9.996068810006918e-06,
      "loss": 0.128,
      "step": 436
    },
    {
      "epoch": 0.022566485928221018,
      "grad_norm": 12.927014350891113,
      "learning_rate": 9.996036258009203e-06,
      "loss": 0.1029,
      "step": 437
    },
    {
      "epoch": 0.022618125484120837,
      "grad_norm": 6.577816963195801,
      "learning_rate": 9.99600357184755e-06,
      "loss": 0.0861,
      "step": 438
    },
    {
      "epoch": 0.022669765040020655,
      "grad_norm": 16.644426345825195,
      "learning_rate": 9.995970751522839e-06,
      "loss": 0.1366,
      "step": 439
    },
    {
      "epoch": 0.022721404595920477,
      "grad_norm": 11.476631164550781,
      "learning_rate": 9.99593779703595e-06,
      "loss": 0.1016,
      "step": 440
    },
    {
      "epoch": 0.022773044151820295,
      "grad_norm": 5.854076862335205,
      "learning_rate": 9.99590470838777e-06,
      "loss": 0.1055,
      "step": 441
    },
    {
      "epoch": 0.022824683707720114,
      "grad_norm": 13.150372505187988,
      "learning_rate": 9.995871485579187e-06,
      "loss": 0.111,
      "step": 442
    },
    {
      "epoch": 0.022876323263619932,
      "grad_norm": 26.723979949951172,
      "learning_rate": 9.995838128611091e-06,
      "loss": 0.1477,
      "step": 443
    },
    {
      "epoch": 0.022927962819519754,
      "grad_norm": 6.101738452911377,
      "learning_rate": 9.99580463748438e-06,
      "loss": 0.1003,
      "step": 444
    },
    {
      "epoch": 0.022979602375419572,
      "grad_norm": 6.648772239685059,
      "learning_rate": 9.995771012199954e-06,
      "loss": 0.0558,
      "step": 445
    },
    {
      "epoch": 0.02303124193131939,
      "grad_norm": 9.533580780029297,
      "learning_rate": 9.995737252758712e-06,
      "loss": 0.0577,
      "step": 446
    },
    {
      "epoch": 0.02308288148721921,
      "grad_norm": 6.946924209594727,
      "learning_rate": 9.995703359161565e-06,
      "loss": 0.0945,
      "step": 447
    },
    {
      "epoch": 0.02313452104311903,
      "grad_norm": 11.819528579711914,
      "learning_rate": 9.995669331409422e-06,
      "loss": 0.087,
      "step": 448
    },
    {
      "epoch": 0.02318616059901885,
      "grad_norm": 16.106975555419922,
      "learning_rate": 9.995635169503194e-06,
      "loss": 0.1374,
      "step": 449
    },
    {
      "epoch": 0.023237800154918668,
      "grad_norm": 12.490364074707031,
      "learning_rate": 9.995600873443802e-06,
      "loss": 0.1199,
      "step": 450
    },
    {
      "epoch": 0.023289439710818486,
      "grad_norm": 11.135107040405273,
      "learning_rate": 9.995566443232165e-06,
      "loss": 0.0672,
      "step": 451
    },
    {
      "epoch": 0.023341079266718308,
      "grad_norm": 9.718745231628418,
      "learning_rate": 9.995531878869208e-06,
      "loss": 0.1532,
      "step": 452
    },
    {
      "epoch": 0.023392718822618126,
      "grad_norm": 4.589501857757568,
      "learning_rate": 9.995497180355862e-06,
      "loss": 0.0978,
      "step": 453
    },
    {
      "epoch": 0.023444358378517945,
      "grad_norm": 12.413463592529297,
      "learning_rate": 9.995462347693054e-06,
      "loss": 0.1206,
      "step": 454
    },
    {
      "epoch": 0.023495997934417763,
      "grad_norm": 27.07497787475586,
      "learning_rate": 9.995427380881722e-06,
      "loss": 0.2681,
      "step": 455
    },
    {
      "epoch": 0.023547637490317585,
      "grad_norm": 29.669431686401367,
      "learning_rate": 9.995392279922804e-06,
      "loss": 0.2076,
      "step": 456
    },
    {
      "epoch": 0.023599277046217403,
      "grad_norm": 24.655376434326172,
      "learning_rate": 9.995357044817245e-06,
      "loss": 0.2173,
      "step": 457
    },
    {
      "epoch": 0.02365091660211722,
      "grad_norm": 21.96120834350586,
      "learning_rate": 9.995321675565987e-06,
      "loss": 0.1429,
      "step": 458
    },
    {
      "epoch": 0.02370255615801704,
      "grad_norm": 15.15005111694336,
      "learning_rate": 9.995286172169983e-06,
      "loss": 0.1607,
      "step": 459
    },
    {
      "epoch": 0.023754195713916862,
      "grad_norm": 22.052322387695312,
      "learning_rate": 9.995250534630187e-06,
      "loss": 0.1954,
      "step": 460
    },
    {
      "epoch": 0.02380583526981668,
      "grad_norm": 4.656030178070068,
      "learning_rate": 9.995214762947553e-06,
      "loss": 0.1146,
      "step": 461
    },
    {
      "epoch": 0.0238574748257165,
      "grad_norm": 7.910061836242676,
      "learning_rate": 9.995178857123043e-06,
      "loss": 0.1405,
      "step": 462
    },
    {
      "epoch": 0.023909114381616317,
      "grad_norm": 18.872121810913086,
      "learning_rate": 9.995142817157623e-06,
      "loss": 0.2157,
      "step": 463
    },
    {
      "epoch": 0.02396075393751614,
      "grad_norm": 6.695632457733154,
      "learning_rate": 9.995106643052258e-06,
      "loss": 0.1515,
      "step": 464
    },
    {
      "epoch": 0.024012393493415957,
      "grad_norm": 8.220040321350098,
      "learning_rate": 9.99507033480792e-06,
      "loss": 0.1056,
      "step": 465
    },
    {
      "epoch": 0.024064033049315776,
      "grad_norm": 9.099470138549805,
      "learning_rate": 9.995033892425587e-06,
      "loss": 0.0854,
      "step": 466
    },
    {
      "epoch": 0.024115672605215594,
      "grad_norm": 8.987198829650879,
      "learning_rate": 9.994997315906233e-06,
      "loss": 0.1293,
      "step": 467
    },
    {
      "epoch": 0.024167312161115416,
      "grad_norm": 5.712615013122559,
      "learning_rate": 9.994960605250846e-06,
      "loss": 0.1123,
      "step": 468
    },
    {
      "epoch": 0.024218951717015234,
      "grad_norm": 19.32622718811035,
      "learning_rate": 9.994923760460406e-06,
      "loss": 0.1639,
      "step": 469
    },
    {
      "epoch": 0.024270591272915053,
      "grad_norm": 3.5177435874938965,
      "learning_rate": 9.994886781535906e-06,
      "loss": 0.0731,
      "step": 470
    },
    {
      "epoch": 0.02432223082881487,
      "grad_norm": 9.9025297164917,
      "learning_rate": 9.994849668478336e-06,
      "loss": 0.1172,
      "step": 471
    },
    {
      "epoch": 0.024373870384714693,
      "grad_norm": 3.6962082386016846,
      "learning_rate": 9.994812421288695e-06,
      "loss": 0.0819,
      "step": 472
    },
    {
      "epoch": 0.02442550994061451,
      "grad_norm": 4.436151504516602,
      "learning_rate": 9.994775039967984e-06,
      "loss": 0.083,
      "step": 473
    },
    {
      "epoch": 0.02447714949651433,
      "grad_norm": 3.100613832473755,
      "learning_rate": 9.994737524517205e-06,
      "loss": 0.0586,
      "step": 474
    },
    {
      "epoch": 0.024528789052414148,
      "grad_norm": 4.939920902252197,
      "learning_rate": 9.994699874937365e-06,
      "loss": 0.0826,
      "step": 475
    },
    {
      "epoch": 0.02458042860831397,
      "grad_norm": 10.579439163208008,
      "learning_rate": 9.994662091229479e-06,
      "loss": 0.063,
      "step": 476
    },
    {
      "epoch": 0.02463206816421379,
      "grad_norm": 28.35911750793457,
      "learning_rate": 9.994624173394556e-06,
      "loss": 0.2501,
      "step": 477
    },
    {
      "epoch": 0.024683707720113607,
      "grad_norm": 12.926342010498047,
      "learning_rate": 9.994586121433619e-06,
      "loss": 0.1745,
      "step": 478
    },
    {
      "epoch": 0.024735347276013425,
      "grad_norm": 18.308156967163086,
      "learning_rate": 9.994547935347685e-06,
      "loss": 0.1357,
      "step": 479
    },
    {
      "epoch": 0.024786986831913247,
      "grad_norm": 12.205229759216309,
      "learning_rate": 9.994509615137785e-06,
      "loss": 0.0556,
      "step": 480
    },
    {
      "epoch": 0.024838626387813065,
      "grad_norm": 19.95629119873047,
      "learning_rate": 9.994471160804944e-06,
      "loss": 0.1919,
      "step": 481
    },
    {
      "epoch": 0.024890265943712884,
      "grad_norm": 13.892300605773926,
      "learning_rate": 9.994432572350196e-06,
      "loss": 0.1331,
      "step": 482
    },
    {
      "epoch": 0.024941905499612702,
      "grad_norm": 7.795734405517578,
      "learning_rate": 9.994393849774577e-06,
      "loss": 0.0909,
      "step": 483
    },
    {
      "epoch": 0.024993545055512524,
      "grad_norm": 8.759286880493164,
      "learning_rate": 9.994354993079127e-06,
      "loss": 0.1314,
      "step": 484
    },
    {
      "epoch": 0.025045184611412342,
      "grad_norm": 4.752614498138428,
      "learning_rate": 9.99431600226489e-06,
      "loss": 0.0452,
      "step": 485
    },
    {
      "epoch": 0.02509682416731216,
      "grad_norm": 6.893033504486084,
      "learning_rate": 9.994276877332913e-06,
      "loss": 0.1052,
      "step": 486
    },
    {
      "epoch": 0.02514846372321198,
      "grad_norm": 10.91378402709961,
      "learning_rate": 9.994237618284247e-06,
      "loss": 0.1335,
      "step": 487
    },
    {
      "epoch": 0.0252001032791118,
      "grad_norm": 13.984848976135254,
      "learning_rate": 9.994198225119944e-06,
      "loss": 0.0888,
      "step": 488
    },
    {
      "epoch": 0.02525174283501162,
      "grad_norm": 4.099421977996826,
      "learning_rate": 9.994158697841065e-06,
      "loss": 0.0608,
      "step": 489
    },
    {
      "epoch": 0.025303382390911438,
      "grad_norm": 3.017310857772827,
      "learning_rate": 9.994119036448668e-06,
      "loss": 0.0552,
      "step": 490
    },
    {
      "epoch": 0.025355021946811256,
      "grad_norm": 5.476449012756348,
      "learning_rate": 9.994079240943822e-06,
      "loss": 0.0696,
      "step": 491
    },
    {
      "epoch": 0.025406661502711078,
      "grad_norm": 6.260684490203857,
      "learning_rate": 9.994039311327593e-06,
      "loss": 0.0978,
      "step": 492
    },
    {
      "epoch": 0.025458301058610897,
      "grad_norm": 22.360870361328125,
      "learning_rate": 9.993999247601053e-06,
      "loss": 0.163,
      "step": 493
    },
    {
      "epoch": 0.025509940614510715,
      "grad_norm": 10.362800598144531,
      "learning_rate": 9.99395904976528e-06,
      "loss": 0.1727,
      "step": 494
    },
    {
      "epoch": 0.025561580170410533,
      "grad_norm": 5.921308517456055,
      "learning_rate": 9.993918717821351e-06,
      "loss": 0.0954,
      "step": 495
    },
    {
      "epoch": 0.025613219726310355,
      "grad_norm": 5.541549205780029,
      "learning_rate": 9.99387825177035e-06,
      "loss": 0.0739,
      "step": 496
    },
    {
      "epoch": 0.025664859282210174,
      "grad_norm": 9.813212394714355,
      "learning_rate": 9.993837651613365e-06,
      "loss": 0.0797,
      "step": 497
    },
    {
      "epoch": 0.025716498838109992,
      "grad_norm": 29.81727409362793,
      "learning_rate": 9.993796917351486e-06,
      "loss": 0.3028,
      "step": 498
    },
    {
      "epoch": 0.02576813839400981,
      "grad_norm": 17.940351486206055,
      "learning_rate": 9.993756048985805e-06,
      "loss": 0.1515,
      "step": 499
    },
    {
      "epoch": 0.025819777949909632,
      "grad_norm": 15.54222297668457,
      "learning_rate": 9.993715046517421e-06,
      "loss": 0.0581,
      "step": 500
    },
    {
      "epoch": 0.02587141750580945,
      "grad_norm": 5.618869304656982,
      "learning_rate": 9.993673909947434e-06,
      "loss": 0.0736,
      "step": 501
    },
    {
      "epoch": 0.02592305706170927,
      "grad_norm": 8.664447784423828,
      "learning_rate": 9.99363263927695e-06,
      "loss": 0.1033,
      "step": 502
    },
    {
      "epoch": 0.025974696617609087,
      "grad_norm": 8.658576011657715,
      "learning_rate": 9.993591234507078e-06,
      "loss": 0.0954,
      "step": 503
    },
    {
      "epoch": 0.02602633617350891,
      "grad_norm": 17.012659072875977,
      "learning_rate": 9.993549695638928e-06,
      "loss": 0.1713,
      "step": 504
    },
    {
      "epoch": 0.026077975729408728,
      "grad_norm": 38.5374641418457,
      "learning_rate": 9.993508022673615e-06,
      "loss": 0.157,
      "step": 505
    },
    {
      "epoch": 0.026129615285308546,
      "grad_norm": 16.602373123168945,
      "learning_rate": 9.99346621561226e-06,
      "loss": 0.2732,
      "step": 506
    },
    {
      "epoch": 0.026181254841208364,
      "grad_norm": 10.644806861877441,
      "learning_rate": 9.993424274455985e-06,
      "loss": 0.13,
      "step": 507
    },
    {
      "epoch": 0.026232894397108186,
      "grad_norm": 12.803956031799316,
      "learning_rate": 9.993382199205917e-06,
      "loss": 0.1378,
      "step": 508
    },
    {
      "epoch": 0.026284533953008005,
      "grad_norm": 12.235467910766602,
      "learning_rate": 9.993339989863181e-06,
      "loss": 0.2416,
      "step": 509
    },
    {
      "epoch": 0.026336173508907823,
      "grad_norm": 15.189654350280762,
      "learning_rate": 9.99329764642892e-06,
      "loss": 0.2236,
      "step": 510
    },
    {
      "epoch": 0.02638781306480764,
      "grad_norm": 8.829648971557617,
      "learning_rate": 9.993255168904262e-06,
      "loss": 0.1102,
      "step": 511
    },
    {
      "epoch": 0.026439452620707463,
      "grad_norm": 14.719709396362305,
      "learning_rate": 9.993212557290353e-06,
      "loss": 0.1286,
      "step": 512
    },
    {
      "epoch": 0.02649109217660728,
      "grad_norm": 14.020724296569824,
      "learning_rate": 9.993169811588335e-06,
      "loss": 0.1525,
      "step": 513
    },
    {
      "epoch": 0.0265427317325071,
      "grad_norm": 8.926763534545898,
      "learning_rate": 9.993126931799356e-06,
      "loss": 0.1091,
      "step": 514
    },
    {
      "epoch": 0.02659437128840692,
      "grad_norm": 7.222027778625488,
      "learning_rate": 9.993083917924568e-06,
      "loss": 0.0935,
      "step": 515
    },
    {
      "epoch": 0.02664601084430674,
      "grad_norm": 5.25782585144043,
      "learning_rate": 9.993040769965128e-06,
      "loss": 0.1076,
      "step": 516
    },
    {
      "epoch": 0.02669765040020656,
      "grad_norm": 10.245491981506348,
      "learning_rate": 9.992997487922192e-06,
      "loss": 0.1861,
      "step": 517
    },
    {
      "epoch": 0.026749289956106377,
      "grad_norm": 3.566049098968506,
      "learning_rate": 9.992954071796921e-06,
      "loss": 0.0795,
      "step": 518
    },
    {
      "epoch": 0.026800929512006195,
      "grad_norm": 8.995718955993652,
      "learning_rate": 9.992910521590486e-06,
      "loss": 0.0863,
      "step": 519
    },
    {
      "epoch": 0.026852569067906017,
      "grad_norm": 7.871212482452393,
      "learning_rate": 9.992866837304051e-06,
      "loss": 0.0767,
      "step": 520
    },
    {
      "epoch": 0.026904208623805836,
      "grad_norm": 9.703657150268555,
      "learning_rate": 9.992823018938793e-06,
      "loss": 0.1041,
      "step": 521
    },
    {
      "epoch": 0.026955848179705654,
      "grad_norm": 12.919182777404785,
      "learning_rate": 9.992779066495887e-06,
      "loss": 0.1599,
      "step": 522
    },
    {
      "epoch": 0.027007487735605473,
      "grad_norm": 12.050824165344238,
      "learning_rate": 9.992734979976515e-06,
      "loss": 0.0945,
      "step": 523
    },
    {
      "epoch": 0.027059127291505294,
      "grad_norm": 24.438077926635742,
      "learning_rate": 9.992690759381859e-06,
      "loss": 0.1986,
      "step": 524
    },
    {
      "epoch": 0.027110766847405113,
      "grad_norm": 3.721461772918701,
      "learning_rate": 9.992646404713105e-06,
      "loss": 0.039,
      "step": 525
    },
    {
      "epoch": 0.02716240640330493,
      "grad_norm": 3.684659719467163,
      "learning_rate": 9.992601915971447e-06,
      "loss": 0.0493,
      "step": 526
    },
    {
      "epoch": 0.02721404595920475,
      "grad_norm": 28.474292755126953,
      "learning_rate": 9.99255729315808e-06,
      "loss": 0.1503,
      "step": 527
    },
    {
      "epoch": 0.02726568551510457,
      "grad_norm": 6.253152370452881,
      "learning_rate": 9.9925125362742e-06,
      "loss": 0.0438,
      "step": 528
    },
    {
      "epoch": 0.02731732507100439,
      "grad_norm": 6.457488059997559,
      "learning_rate": 9.992467645321012e-06,
      "loss": 0.0908,
      "step": 529
    },
    {
      "epoch": 0.027368964626904208,
      "grad_norm": 8.703802108764648,
      "learning_rate": 9.992422620299716e-06,
      "loss": 0.1347,
      "step": 530
    },
    {
      "epoch": 0.027420604182804027,
      "grad_norm": 5.973473072052002,
      "learning_rate": 9.992377461211527e-06,
      "loss": 0.0525,
      "step": 531
    },
    {
      "epoch": 0.02747224373870385,
      "grad_norm": 23.384191513061523,
      "learning_rate": 9.992332168057654e-06,
      "loss": 0.1441,
      "step": 532
    },
    {
      "epoch": 0.027523883294603667,
      "grad_norm": 6.915075302124023,
      "learning_rate": 9.992286740839316e-06,
      "loss": 0.0802,
      "step": 533
    },
    {
      "epoch": 0.027575522850503485,
      "grad_norm": 7.862879753112793,
      "learning_rate": 9.992241179557732e-06,
      "loss": 0.0821,
      "step": 534
    },
    {
      "epoch": 0.027627162406403304,
      "grad_norm": 11.977999687194824,
      "learning_rate": 9.992195484214125e-06,
      "loss": 0.1677,
      "step": 535
    },
    {
      "epoch": 0.027678801962303125,
      "grad_norm": 6.649957656860352,
      "learning_rate": 9.99214965480972e-06,
      "loss": 0.134,
      "step": 536
    },
    {
      "epoch": 0.027730441518202944,
      "grad_norm": 13.904911994934082,
      "learning_rate": 9.992103691345752e-06,
      "loss": 0.1306,
      "step": 537
    },
    {
      "epoch": 0.027782081074102762,
      "grad_norm": 5.140995025634766,
      "learning_rate": 9.992057593823453e-06,
      "loss": 0.0913,
      "step": 538
    },
    {
      "epoch": 0.02783372063000258,
      "grad_norm": 16.583709716796875,
      "learning_rate": 9.99201136224406e-06,
      "loss": 0.1495,
      "step": 539
    },
    {
      "epoch": 0.027885360185902403,
      "grad_norm": 11.721223831176758,
      "learning_rate": 9.991964996608817e-06,
      "loss": 0.1091,
      "step": 540
    },
    {
      "epoch": 0.02793699974180222,
      "grad_norm": 29.797531127929688,
      "learning_rate": 9.991918496918966e-06,
      "loss": 0.1518,
      "step": 541
    },
    {
      "epoch": 0.02798863929770204,
      "grad_norm": 6.41659688949585,
      "learning_rate": 9.99187186317576e-06,
      "loss": 0.0743,
      "step": 542
    },
    {
      "epoch": 0.028040278853601858,
      "grad_norm": 8.4193696975708,
      "learning_rate": 9.991825095380448e-06,
      "loss": 0.0604,
      "step": 543
    },
    {
      "epoch": 0.02809191840950168,
      "grad_norm": 6.7128005027771,
      "learning_rate": 9.991778193534285e-06,
      "loss": 0.0792,
      "step": 544
    },
    {
      "epoch": 0.028143557965401498,
      "grad_norm": 9.621326446533203,
      "learning_rate": 9.991731157638533e-06,
      "loss": 0.105,
      "step": 545
    },
    {
      "epoch": 0.028195197521301316,
      "grad_norm": 4.856945991516113,
      "learning_rate": 9.991683987694454e-06,
      "loss": 0.0877,
      "step": 546
    },
    {
      "epoch": 0.028246837077201135,
      "grad_norm": 14.089775085449219,
      "learning_rate": 9.991636683703316e-06,
      "loss": 0.2163,
      "step": 547
    },
    {
      "epoch": 0.028298476633100957,
      "grad_norm": 12.682099342346191,
      "learning_rate": 9.991589245666386e-06,
      "loss": 0.1128,
      "step": 548
    },
    {
      "epoch": 0.028350116189000775,
      "grad_norm": 4.204016208648682,
      "learning_rate": 9.991541673584943e-06,
      "loss": 0.0427,
      "step": 549
    },
    {
      "epoch": 0.028401755744900593,
      "grad_norm": 3.4143574237823486,
      "learning_rate": 9.99149396746026e-06,
      "loss": 0.093,
      "step": 550
    },
    {
      "epoch": 0.028453395300800412,
      "grad_norm": 11.550357818603516,
      "learning_rate": 9.99144612729362e-06,
      "loss": 0.0536,
      "step": 551
    },
    {
      "epoch": 0.028505034856700234,
      "grad_norm": 5.962544918060303,
      "learning_rate": 9.991398153086308e-06,
      "loss": 0.1005,
      "step": 552
    },
    {
      "epoch": 0.028556674412600052,
      "grad_norm": 13.265331268310547,
      "learning_rate": 9.99135004483961e-06,
      "loss": 0.2093,
      "step": 553
    },
    {
      "epoch": 0.02860831396849987,
      "grad_norm": 5.466770172119141,
      "learning_rate": 9.991301802554822e-06,
      "loss": 0.138,
      "step": 554
    },
    {
      "epoch": 0.02865995352439969,
      "grad_norm": 6.0533928871154785,
      "learning_rate": 9.991253426233235e-06,
      "loss": 0.0921,
      "step": 555
    },
    {
      "epoch": 0.02871159308029951,
      "grad_norm": 8.82242488861084,
      "learning_rate": 9.991204915876151e-06,
      "loss": 0.1661,
      "step": 556
    },
    {
      "epoch": 0.02876323263619933,
      "grad_norm": 13.641070365905762,
      "learning_rate": 9.991156271484873e-06,
      "loss": 0.1383,
      "step": 557
    },
    {
      "epoch": 0.028814872192099147,
      "grad_norm": 5.661549091339111,
      "learning_rate": 9.991107493060705e-06,
      "loss": 0.1066,
      "step": 558
    },
    {
      "epoch": 0.028866511747998966,
      "grad_norm": 10.416232109069824,
      "learning_rate": 9.99105858060496e-06,
      "loss": 0.1114,
      "step": 559
    },
    {
      "epoch": 0.028918151303898788,
      "grad_norm": 8.741594314575195,
      "learning_rate": 9.991009534118949e-06,
      "loss": 0.0866,
      "step": 560
    },
    {
      "epoch": 0.028969790859798606,
      "grad_norm": 4.669722557067871,
      "learning_rate": 9.990960353603989e-06,
      "loss": 0.0856,
      "step": 561
    },
    {
      "epoch": 0.029021430415698424,
      "grad_norm": 8.832385063171387,
      "learning_rate": 9.990911039061402e-06,
      "loss": 0.132,
      "step": 562
    },
    {
      "epoch": 0.029073069971598243,
      "grad_norm": 11.469136238098145,
      "learning_rate": 9.990861590492512e-06,
      "loss": 0.1135,
      "step": 563
    },
    {
      "epoch": 0.029124709527498065,
      "grad_norm": 4.8284993171691895,
      "learning_rate": 9.990812007898645e-06,
      "loss": 0.0879,
      "step": 564
    },
    {
      "epoch": 0.029176349083397883,
      "grad_norm": 5.787387371063232,
      "learning_rate": 9.990762291281136e-06,
      "loss": 0.0756,
      "step": 565
    },
    {
      "epoch": 0.0292279886392977,
      "grad_norm": 7.247158527374268,
      "learning_rate": 9.990712440641317e-06,
      "loss": 0.0589,
      "step": 566
    },
    {
      "epoch": 0.02927962819519752,
      "grad_norm": 5.577150344848633,
      "learning_rate": 9.99066245598053e-06,
      "loss": 0.0471,
      "step": 567
    },
    {
      "epoch": 0.02933126775109734,
      "grad_norm": 11.37957763671875,
      "learning_rate": 9.990612337300112e-06,
      "loss": 0.0847,
      "step": 568
    },
    {
      "epoch": 0.02938290730699716,
      "grad_norm": 22.263378143310547,
      "learning_rate": 9.990562084601415e-06,
      "loss": 0.1469,
      "step": 569
    },
    {
      "epoch": 0.02943454686289698,
      "grad_norm": 11.28761100769043,
      "learning_rate": 9.990511697885785e-06,
      "loss": 0.1726,
      "step": 570
    },
    {
      "epoch": 0.029486186418796797,
      "grad_norm": 12.662792205810547,
      "learning_rate": 9.990461177154576e-06,
      "loss": 0.1086,
      "step": 571
    },
    {
      "epoch": 0.02953782597469662,
      "grad_norm": 6.088356018066406,
      "learning_rate": 9.990410522409143e-06,
      "loss": 0.069,
      "step": 572
    },
    {
      "epoch": 0.029589465530596437,
      "grad_norm": 12.128657341003418,
      "learning_rate": 9.990359733650848e-06,
      "loss": 0.0975,
      "step": 573
    },
    {
      "epoch": 0.029641105086496256,
      "grad_norm": 15.973663330078125,
      "learning_rate": 9.990308810881056e-06,
      "loss": 0.0882,
      "step": 574
    },
    {
      "epoch": 0.029692744642396074,
      "grad_norm": 6.499279499053955,
      "learning_rate": 9.990257754101131e-06,
      "loss": 0.0697,
      "step": 575
    },
    {
      "epoch": 0.029744384198295896,
      "grad_norm": 26.538454055786133,
      "learning_rate": 9.990206563312447e-06,
      "loss": 0.0779,
      "step": 576
    },
    {
      "epoch": 0.029796023754195714,
      "grad_norm": 12.38139533996582,
      "learning_rate": 9.990155238516379e-06,
      "loss": 0.171,
      "step": 577
    },
    {
      "epoch": 0.029847663310095533,
      "grad_norm": 13.101884841918945,
      "learning_rate": 9.990103779714302e-06,
      "loss": 0.1094,
      "step": 578
    },
    {
      "epoch": 0.02989930286599535,
      "grad_norm": 7.582159519195557,
      "learning_rate": 9.990052186907601e-06,
      "loss": 0.0861,
      "step": 579
    },
    {
      "epoch": 0.029950942421895173,
      "grad_norm": 15.18481159210205,
      "learning_rate": 9.990000460097663e-06,
      "loss": 0.2477,
      "step": 580
    },
    {
      "epoch": 0.03000258197779499,
      "grad_norm": 12.850144386291504,
      "learning_rate": 9.98994859928587e-06,
      "loss": 0.0877,
      "step": 581
    },
    {
      "epoch": 0.03005422153369481,
      "grad_norm": 8.706390380859375,
      "learning_rate": 9.98989660447362e-06,
      "loss": 0.2024,
      "step": 582
    },
    {
      "epoch": 0.030105861089594628,
      "grad_norm": 7.6663713455200195,
      "learning_rate": 9.989844475662311e-06,
      "loss": 0.1373,
      "step": 583
    },
    {
      "epoch": 0.03015750064549445,
      "grad_norm": 11.306324005126953,
      "learning_rate": 9.98979221285334e-06,
      "loss": 0.0878,
      "step": 584
    },
    {
      "epoch": 0.030209140201394268,
      "grad_norm": 6.720804691314697,
      "learning_rate": 9.98973981604811e-06,
      "loss": 0.1267,
      "step": 585
    },
    {
      "epoch": 0.030260779757294087,
      "grad_norm": 9.121199607849121,
      "learning_rate": 9.989687285248028e-06,
      "loss": 0.1651,
      "step": 586
    },
    {
      "epoch": 0.030312419313193905,
      "grad_norm": 11.007227897644043,
      "learning_rate": 9.989634620454507e-06,
      "loss": 0.1035,
      "step": 587
    },
    {
      "epoch": 0.030364058869093727,
      "grad_norm": 12.392702102661133,
      "learning_rate": 9.989581821668959e-06,
      "loss": 0.077,
      "step": 588
    },
    {
      "epoch": 0.030415698424993545,
      "grad_norm": 13.25352954864502,
      "learning_rate": 9.989528888892802e-06,
      "loss": 0.1576,
      "step": 589
    },
    {
      "epoch": 0.030467337980893364,
      "grad_norm": 6.1588335037231445,
      "learning_rate": 9.98947582212746e-06,
      "loss": 0.0841,
      "step": 590
    },
    {
      "epoch": 0.030518977536793182,
      "grad_norm": 3.499821186065674,
      "learning_rate": 9.989422621374354e-06,
      "loss": 0.1154,
      "step": 591
    },
    {
      "epoch": 0.030570617092693004,
      "grad_norm": 4.502922534942627,
      "learning_rate": 9.989369286634917e-06,
      "loss": 0.1023,
      "step": 592
    },
    {
      "epoch": 0.030622256648592822,
      "grad_norm": 27.592018127441406,
      "learning_rate": 9.989315817910578e-06,
      "loss": 0.1006,
      "step": 593
    },
    {
      "epoch": 0.03067389620449264,
      "grad_norm": 5.338855266571045,
      "learning_rate": 9.989262215202775e-06,
      "loss": 0.0817,
      "step": 594
    },
    {
      "epoch": 0.03072553576039246,
      "grad_norm": 10.35525894165039,
      "learning_rate": 9.989208478512943e-06,
      "loss": 0.1485,
      "step": 595
    },
    {
      "epoch": 0.03077717531629228,
      "grad_norm": 18.726520538330078,
      "learning_rate": 9.989154607842532e-06,
      "loss": 0.095,
      "step": 596
    },
    {
      "epoch": 0.0308288148721921,
      "grad_norm": 17.615339279174805,
      "learning_rate": 9.989100603192984e-06,
      "loss": 0.1235,
      "step": 597
    },
    {
      "epoch": 0.030880454428091918,
      "grad_norm": 12.92137336730957,
      "learning_rate": 9.98904646456575e-06,
      "loss": 0.1381,
      "step": 598
    },
    {
      "epoch": 0.030932093983991736,
      "grad_norm": 3.539713144302368,
      "learning_rate": 9.988992191962286e-06,
      "loss": 0.0673,
      "step": 599
    },
    {
      "epoch": 0.030983733539891558,
      "grad_norm": 20.13332176208496,
      "learning_rate": 9.988937785384046e-06,
      "loss": 0.1954,
      "step": 600
    },
    {
      "epoch": 0.031035373095791376,
      "grad_norm": 14.007573127746582,
      "learning_rate": 9.988883244832492e-06,
      "loss": 0.1174,
      "step": 601
    },
    {
      "epoch": 0.031087012651691195,
      "grad_norm": 4.6299543380737305,
      "learning_rate": 9.988828570309091e-06,
      "loss": 0.072,
      "step": 602
    },
    {
      "epoch": 0.031138652207591013,
      "grad_norm": 12.229937553405762,
      "learning_rate": 9.98877376181531e-06,
      "loss": 0.1119,
      "step": 603
    },
    {
      "epoch": 0.031190291763490835,
      "grad_norm": 4.655936241149902,
      "learning_rate": 9.98871881935262e-06,
      "loss": 0.0791,
      "step": 604
    },
    {
      "epoch": 0.031241931319390653,
      "grad_norm": 5.407815933227539,
      "learning_rate": 9.988663742922495e-06,
      "loss": 0.0785,
      "step": 605
    },
    {
      "epoch": 0.03129357087529047,
      "grad_norm": 6.09136962890625,
      "learning_rate": 9.988608532526415e-06,
      "loss": 0.0899,
      "step": 606
    },
    {
      "epoch": 0.03134521043119029,
      "grad_norm": 18.561464309692383,
      "learning_rate": 9.988553188165866e-06,
      "loss": 0.1227,
      "step": 607
    },
    {
      "epoch": 0.03139684998709011,
      "grad_norm": 14.250049591064453,
      "learning_rate": 9.98849770984233e-06,
      "loss": 0.1155,
      "step": 608
    },
    {
      "epoch": 0.03144848954298993,
      "grad_norm": 6.5807719230651855,
      "learning_rate": 9.9884420975573e-06,
      "loss": 0.1104,
      "step": 609
    },
    {
      "epoch": 0.03150012909888975,
      "grad_norm": 3.569276809692383,
      "learning_rate": 9.988386351312267e-06,
      "loss": 0.0713,
      "step": 610
    },
    {
      "epoch": 0.03155176865478957,
      "grad_norm": 5.338319301605225,
      "learning_rate": 9.98833047110873e-06,
      "loss": 0.0722,
      "step": 611
    },
    {
      "epoch": 0.03160340821068939,
      "grad_norm": 5.639990329742432,
      "learning_rate": 9.988274456948188e-06,
      "loss": 0.0775,
      "step": 612
    },
    {
      "epoch": 0.03165504776658921,
      "grad_norm": 10.824724197387695,
      "learning_rate": 9.988218308832146e-06,
      "loss": 0.1524,
      "step": 613
    },
    {
      "epoch": 0.031706687322489026,
      "grad_norm": 11.50710678100586,
      "learning_rate": 9.98816202676211e-06,
      "loss": 0.1284,
      "step": 614
    },
    {
      "epoch": 0.031758326878388844,
      "grad_norm": 6.021254539489746,
      "learning_rate": 9.988105610739595e-06,
      "loss": 0.1461,
      "step": 615
    },
    {
      "epoch": 0.03180996643428866,
      "grad_norm": 10.806495666503906,
      "learning_rate": 9.988049060766114e-06,
      "loss": 0.1206,
      "step": 616
    },
    {
      "epoch": 0.03186160599018848,
      "grad_norm": 5.310125827789307,
      "learning_rate": 9.987992376843184e-06,
      "loss": 0.1372,
      "step": 617
    },
    {
      "epoch": 0.031913245546088306,
      "grad_norm": 5.37982177734375,
      "learning_rate": 9.98793555897233e-06,
      "loss": 0.0552,
      "step": 618
    },
    {
      "epoch": 0.031964885101988125,
      "grad_norm": 15.467684745788574,
      "learning_rate": 9.987878607155074e-06,
      "loss": 0.1311,
      "step": 619
    },
    {
      "epoch": 0.03201652465788794,
      "grad_norm": 7.843317985534668,
      "learning_rate": 9.98782152139295e-06,
      "loss": 0.0946,
      "step": 620
    },
    {
      "epoch": 0.03206816421378776,
      "grad_norm": 3.7167348861694336,
      "learning_rate": 9.98776430168749e-06,
      "loss": 0.0781,
      "step": 621
    },
    {
      "epoch": 0.03211980376968758,
      "grad_norm": 7.386531829833984,
      "learning_rate": 9.987706948040228e-06,
      "loss": 0.1314,
      "step": 622
    },
    {
      "epoch": 0.0321714433255874,
      "grad_norm": 13.234283447265625,
      "learning_rate": 9.987649460452707e-06,
      "loss": 0.0923,
      "step": 623
    },
    {
      "epoch": 0.03222308288148722,
      "grad_norm": 11.289145469665527,
      "learning_rate": 9.987591838926467e-06,
      "loss": 0.0976,
      "step": 624
    },
    {
      "epoch": 0.032274722437387035,
      "grad_norm": 6.71480131149292,
      "learning_rate": 9.987534083463059e-06,
      "loss": 0.0707,
      "step": 625
    },
    {
      "epoch": 0.03232636199328686,
      "grad_norm": 4.602721691131592,
      "learning_rate": 9.987476194064033e-06,
      "loss": 0.0516,
      "step": 626
    },
    {
      "epoch": 0.03237800154918668,
      "grad_norm": 10.202340126037598,
      "learning_rate": 9.987418170730945e-06,
      "loss": 0.1419,
      "step": 627
    },
    {
      "epoch": 0.0324296411050865,
      "grad_norm": 18.311201095581055,
      "learning_rate": 9.98736001346535e-06,
      "loss": 0.2683,
      "step": 628
    },
    {
      "epoch": 0.032481280660986316,
      "grad_norm": 6.571619510650635,
      "learning_rate": 9.98730172226881e-06,
      "loss": 0.0625,
      "step": 629
    },
    {
      "epoch": 0.032532920216886134,
      "grad_norm": 9.705106735229492,
      "learning_rate": 9.987243297142892e-06,
      "loss": 0.0837,
      "step": 630
    },
    {
      "epoch": 0.03258455977278595,
      "grad_norm": 4.648707866668701,
      "learning_rate": 9.987184738089166e-06,
      "loss": 0.0629,
      "step": 631
    },
    {
      "epoch": 0.03263619932868577,
      "grad_norm": 20.985868453979492,
      "learning_rate": 9.987126045109203e-06,
      "loss": 0.1415,
      "step": 632
    },
    {
      "epoch": 0.03268783888458559,
      "grad_norm": 11.308107376098633,
      "learning_rate": 9.987067218204579e-06,
      "loss": 0.1104,
      "step": 633
    },
    {
      "epoch": 0.032739478440485414,
      "grad_norm": 2.884181261062622,
      "learning_rate": 9.987008257376873e-06,
      "loss": 0.0484,
      "step": 634
    },
    {
      "epoch": 0.03279111799638523,
      "grad_norm": 8.976666450500488,
      "learning_rate": 9.986949162627671e-06,
      "loss": 0.1458,
      "step": 635
    },
    {
      "epoch": 0.03284275755228505,
      "grad_norm": 4.451235294342041,
      "learning_rate": 9.986889933958557e-06,
      "loss": 0.0503,
      "step": 636
    },
    {
      "epoch": 0.03289439710818487,
      "grad_norm": 22.054903030395508,
      "learning_rate": 9.986830571371124e-06,
      "loss": 0.1452,
      "step": 637
    },
    {
      "epoch": 0.03294603666408469,
      "grad_norm": 7.556265354156494,
      "learning_rate": 9.986771074866965e-06,
      "loss": 0.0656,
      "step": 638
    },
    {
      "epoch": 0.032997676219984506,
      "grad_norm": 9.578719139099121,
      "learning_rate": 9.986711444447676e-06,
      "loss": 0.0874,
      "step": 639
    },
    {
      "epoch": 0.033049315775884325,
      "grad_norm": 17.911502838134766,
      "learning_rate": 9.986651680114862e-06,
      "loss": 0.1213,
      "step": 640
    },
    {
      "epoch": 0.03310095533178415,
      "grad_norm": 15.02785587310791,
      "learning_rate": 9.986591781870124e-06,
      "loss": 0.0917,
      "step": 641
    },
    {
      "epoch": 0.03315259488768397,
      "grad_norm": 18.11434555053711,
      "learning_rate": 9.986531749715074e-06,
      "loss": 0.1248,
      "step": 642
    },
    {
      "epoch": 0.03320423444358379,
      "grad_norm": 6.803393363952637,
      "learning_rate": 9.986471583651321e-06,
      "loss": 0.0804,
      "step": 643
    },
    {
      "epoch": 0.033255873999483605,
      "grad_norm": 18.66649627685547,
      "learning_rate": 9.986411283680484e-06,
      "loss": 0.1727,
      "step": 644
    },
    {
      "epoch": 0.033307513555383424,
      "grad_norm": 10.807612419128418,
      "learning_rate": 9.98635084980418e-06,
      "loss": 0.1045,
      "step": 645
    },
    {
      "epoch": 0.03335915311128324,
      "grad_norm": 4.080796241760254,
      "learning_rate": 9.986290282024032e-06,
      "loss": 0.0415,
      "step": 646
    },
    {
      "epoch": 0.03341079266718306,
      "grad_norm": 21.921825408935547,
      "learning_rate": 9.986229580341666e-06,
      "loss": 0.1528,
      "step": 647
    },
    {
      "epoch": 0.03346243222308288,
      "grad_norm": 8.579906463623047,
      "learning_rate": 9.986168744758715e-06,
      "loss": 0.1094,
      "step": 648
    },
    {
      "epoch": 0.033514071778982704,
      "grad_norm": 9.186421394348145,
      "learning_rate": 9.98610777527681e-06,
      "loss": 0.1111,
      "step": 649
    },
    {
      "epoch": 0.03356571133488252,
      "grad_norm": 5.281872272491455,
      "learning_rate": 9.986046671897589e-06,
      "loss": 0.0796,
      "step": 650
    },
    {
      "epoch": 0.03361735089078234,
      "grad_norm": 5.2483649253845215,
      "learning_rate": 9.985985434622693e-06,
      "loss": 0.1059,
      "step": 651
    },
    {
      "epoch": 0.03366899044668216,
      "grad_norm": 3.484017848968506,
      "learning_rate": 9.985924063453765e-06,
      "loss": 0.0427,
      "step": 652
    },
    {
      "epoch": 0.03372063000258198,
      "grad_norm": 10.210417747497559,
      "learning_rate": 9.985862558392457e-06,
      "loss": 0.1025,
      "step": 653
    },
    {
      "epoch": 0.033772269558481796,
      "grad_norm": 20.06682014465332,
      "learning_rate": 9.985800919440415e-06,
      "loss": 0.1601,
      "step": 654
    },
    {
      "epoch": 0.033823909114381615,
      "grad_norm": 22.510507583618164,
      "learning_rate": 9.985739146599297e-06,
      "loss": 0.1409,
      "step": 655
    },
    {
      "epoch": 0.03387554867028143,
      "grad_norm": 17.763166427612305,
      "learning_rate": 9.985677239870766e-06,
      "loss": 0.1204,
      "step": 656
    },
    {
      "epoch": 0.03392718822618126,
      "grad_norm": 21.931194305419922,
      "learning_rate": 9.985615199256477e-06,
      "loss": 0.1249,
      "step": 657
    },
    {
      "epoch": 0.03397882778208108,
      "grad_norm": 5.058078289031982,
      "learning_rate": 9.985553024758103e-06,
      "loss": 0.0797,
      "step": 658
    },
    {
      "epoch": 0.034030467337980895,
      "grad_norm": 5.3657050132751465,
      "learning_rate": 9.98549071637731e-06,
      "loss": 0.0539,
      "step": 659
    },
    {
      "epoch": 0.03408210689388071,
      "grad_norm": 33.50774383544922,
      "learning_rate": 9.985428274115768e-06,
      "loss": 0.1597,
      "step": 660
    },
    {
      "epoch": 0.03413374644978053,
      "grad_norm": 29.93160057067871,
      "learning_rate": 9.98536569797516e-06,
      "loss": 0.2443,
      "step": 661
    },
    {
      "epoch": 0.03418538600568035,
      "grad_norm": 13.168280601501465,
      "learning_rate": 9.985302987957162e-06,
      "loss": 0.081,
      "step": 662
    },
    {
      "epoch": 0.03423702556158017,
      "grad_norm": 4.077238082885742,
      "learning_rate": 9.985240144063462e-06,
      "loss": 0.0188,
      "step": 663
    },
    {
      "epoch": 0.03428866511747999,
      "grad_norm": 5.480165004730225,
      "learning_rate": 9.985177166295743e-06,
      "loss": 0.072,
      "step": 664
    },
    {
      "epoch": 0.03434030467337981,
      "grad_norm": 6.923101425170898,
      "learning_rate": 9.9851140546557e-06,
      "loss": 0.0944,
      "step": 665
    },
    {
      "epoch": 0.03439194422927963,
      "grad_norm": 6.852527618408203,
      "learning_rate": 9.985050809145027e-06,
      "loss": 0.1326,
      "step": 666
    },
    {
      "epoch": 0.03444358378517945,
      "grad_norm": 12.148775100708008,
      "learning_rate": 9.984987429765421e-06,
      "loss": 0.0983,
      "step": 667
    },
    {
      "epoch": 0.03449522334107927,
      "grad_norm": 16.896757125854492,
      "learning_rate": 9.984923916518585e-06,
      "loss": 0.1325,
      "step": 668
    },
    {
      "epoch": 0.034546862896979086,
      "grad_norm": 21.185226440429688,
      "learning_rate": 9.984860269406224e-06,
      "loss": 0.1505,
      "step": 669
    },
    {
      "epoch": 0.034598502452878904,
      "grad_norm": 16.581623077392578,
      "learning_rate": 9.98479648843005e-06,
      "loss": 0.1288,
      "step": 670
    },
    {
      "epoch": 0.03465014200877872,
      "grad_norm": 8.019001007080078,
      "learning_rate": 9.984732573591771e-06,
      "loss": 0.0833,
      "step": 671
    },
    {
      "epoch": 0.03470178156467854,
      "grad_norm": 26.858346939086914,
      "learning_rate": 9.984668524893106e-06,
      "loss": 0.1804,
      "step": 672
    },
    {
      "epoch": 0.034753421120578366,
      "grad_norm": 4.312213897705078,
      "learning_rate": 9.984604342335776e-06,
      "loss": 0.0785,
      "step": 673
    },
    {
      "epoch": 0.034805060676478185,
      "grad_norm": 25.54507827758789,
      "learning_rate": 9.9845400259215e-06,
      "loss": 0.3169,
      "step": 674
    },
    {
      "epoch": 0.034856700232378,
      "grad_norm": 8.73337459564209,
      "learning_rate": 9.984475575652013e-06,
      "loss": 0.1015,
      "step": 675
    },
    {
      "epoch": 0.03490833978827782,
      "grad_norm": 10.717947006225586,
      "learning_rate": 9.984410991529038e-06,
      "loss": 0.0978,
      "step": 676
    },
    {
      "epoch": 0.03495997934417764,
      "grad_norm": 9.459535598754883,
      "learning_rate": 9.984346273554313e-06,
      "loss": 0.1011,
      "step": 677
    },
    {
      "epoch": 0.03501161890007746,
      "grad_norm": 12.877370834350586,
      "learning_rate": 9.984281421729576e-06,
      "loss": 0.1242,
      "step": 678
    },
    {
      "epoch": 0.03506325845597728,
      "grad_norm": 10.192405700683594,
      "learning_rate": 9.984216436056568e-06,
      "loss": 0.0968,
      "step": 679
    },
    {
      "epoch": 0.035114898011877095,
      "grad_norm": 7.938048839569092,
      "learning_rate": 9.984151316537033e-06,
      "loss": 0.1217,
      "step": 680
    },
    {
      "epoch": 0.03516653756777692,
      "grad_norm": 5.764493465423584,
      "learning_rate": 9.984086063172723e-06,
      "loss": 0.0882,
      "step": 681
    },
    {
      "epoch": 0.03521817712367674,
      "grad_norm": 8.38943099975586,
      "learning_rate": 9.984020675965387e-06,
      "loss": 0.1301,
      "step": 682
    },
    {
      "epoch": 0.03526981667957656,
      "grad_norm": 6.866271495819092,
      "learning_rate": 9.983955154916783e-06,
      "loss": 0.126,
      "step": 683
    },
    {
      "epoch": 0.035321456235476376,
      "grad_norm": 7.476076602935791,
      "learning_rate": 9.983889500028668e-06,
      "loss": 0.1361,
      "step": 684
    },
    {
      "epoch": 0.035373095791376194,
      "grad_norm": 17.56871795654297,
      "learning_rate": 9.983823711302808e-06,
      "loss": 0.1226,
      "step": 685
    },
    {
      "epoch": 0.03542473534727601,
      "grad_norm": 13.629590034484863,
      "learning_rate": 9.983757788740968e-06,
      "loss": 0.0981,
      "step": 686
    },
    {
      "epoch": 0.03547637490317583,
      "grad_norm": 3.5428617000579834,
      "learning_rate": 9.98369173234492e-06,
      "loss": 0.079,
      "step": 687
    },
    {
      "epoch": 0.03552801445907565,
      "grad_norm": 5.559525489807129,
      "learning_rate": 9.983625542116433e-06,
      "loss": 0.0876,
      "step": 688
    },
    {
      "epoch": 0.035579654014975474,
      "grad_norm": 10.205683708190918,
      "learning_rate": 9.983559218057292e-06,
      "loss": 0.0939,
      "step": 689
    },
    {
      "epoch": 0.03563129357087529,
      "grad_norm": 29.17209815979004,
      "learning_rate": 9.983492760169273e-06,
      "loss": 0.1515,
      "step": 690
    },
    {
      "epoch": 0.03568293312677511,
      "grad_norm": 7.536749362945557,
      "learning_rate": 9.98342616845416e-06,
      "loss": 0.056,
      "step": 691
    },
    {
      "epoch": 0.03573457268267493,
      "grad_norm": 18.971330642700195,
      "learning_rate": 9.983359442913744e-06,
      "loss": 0.1797,
      "step": 692
    },
    {
      "epoch": 0.03578621223857475,
      "grad_norm": 20.879150390625,
      "learning_rate": 9.983292583549818e-06,
      "loss": 0.1426,
      "step": 693
    },
    {
      "epoch": 0.035837851794474566,
      "grad_norm": 13.804621696472168,
      "learning_rate": 9.983225590364171e-06,
      "loss": 0.0907,
      "step": 694
    },
    {
      "epoch": 0.035889491350374385,
      "grad_norm": 10.660599708557129,
      "learning_rate": 9.983158463358611e-06,
      "loss": 0.1227,
      "step": 695
    },
    {
      "epoch": 0.0359411309062742,
      "grad_norm": 4.2470479011535645,
      "learning_rate": 9.983091202534934e-06,
      "loss": 0.0538,
      "step": 696
    },
    {
      "epoch": 0.03599277046217403,
      "grad_norm": 11.279118537902832,
      "learning_rate": 9.983023807894947e-06,
      "loss": 0.1277,
      "step": 697
    },
    {
      "epoch": 0.03604441001807385,
      "grad_norm": 14.318135261535645,
      "learning_rate": 9.982956279440463e-06,
      "loss": 0.0948,
      "step": 698
    },
    {
      "epoch": 0.036096049573973665,
      "grad_norm": 55.45172882080078,
      "learning_rate": 9.982888617173292e-06,
      "loss": 0.4257,
      "step": 699
    },
    {
      "epoch": 0.036147689129873484,
      "grad_norm": 10.424166679382324,
      "learning_rate": 9.982820821095253e-06,
      "loss": 0.0654,
      "step": 700
    },
    {
      "epoch": 0.0361993286857733,
      "grad_norm": 14.550618171691895,
      "learning_rate": 9.982752891208168e-06,
      "loss": 0.1161,
      "step": 701
    },
    {
      "epoch": 0.03625096824167312,
      "grad_norm": 20.458398818969727,
      "learning_rate": 9.982684827513857e-06,
      "loss": 0.1394,
      "step": 702
    },
    {
      "epoch": 0.03630260779757294,
      "grad_norm": 6.772975444793701,
      "learning_rate": 9.98261663001415e-06,
      "loss": 0.0906,
      "step": 703
    },
    {
      "epoch": 0.03635424735347276,
      "grad_norm": 11.587423324584961,
      "learning_rate": 9.98254829871088e-06,
      "loss": 0.0859,
      "step": 704
    },
    {
      "epoch": 0.03640588690937258,
      "grad_norm": 7.036760330200195,
      "learning_rate": 9.98247983360588e-06,
      "loss": 0.064,
      "step": 705
    },
    {
      "epoch": 0.0364575264652724,
      "grad_norm": 13.29023551940918,
      "learning_rate": 9.982411234700988e-06,
      "loss": 0.0655,
      "step": 706
    },
    {
      "epoch": 0.03650916602117222,
      "grad_norm": 7.405342102050781,
      "learning_rate": 9.982342501998048e-06,
      "loss": 0.0931,
      "step": 707
    },
    {
      "epoch": 0.03656080557707204,
      "grad_norm": 16.9232177734375,
      "learning_rate": 9.982273635498905e-06,
      "loss": 0.1545,
      "step": 708
    },
    {
      "epoch": 0.036612445132971856,
      "grad_norm": 11.770524978637695,
      "learning_rate": 9.982204635205408e-06,
      "loss": 0.1135,
      "step": 709
    },
    {
      "epoch": 0.036664084688871675,
      "grad_norm": 20.671833038330078,
      "learning_rate": 9.982135501119411e-06,
      "loss": 0.1029,
      "step": 710
    },
    {
      "epoch": 0.03671572424477149,
      "grad_norm": 15.9071683883667,
      "learning_rate": 9.98206623324277e-06,
      "loss": 0.0749,
      "step": 711
    },
    {
      "epoch": 0.03676736380067131,
      "grad_norm": 13.798171043395996,
      "learning_rate": 9.981996831577343e-06,
      "loss": 0.1195,
      "step": 712
    },
    {
      "epoch": 0.03681900335657114,
      "grad_norm": 10.923251152038574,
      "learning_rate": 9.981927296124996e-06,
      "loss": 0.1939,
      "step": 713
    },
    {
      "epoch": 0.036870642912470955,
      "grad_norm": 8.162737846374512,
      "learning_rate": 9.981857626887596e-06,
      "loss": 0.2072,
      "step": 714
    },
    {
      "epoch": 0.03692228246837077,
      "grad_norm": 10.172099113464355,
      "learning_rate": 9.981787823867016e-06,
      "loss": 0.0839,
      "step": 715
    },
    {
      "epoch": 0.03697392202427059,
      "grad_norm": 16.27180290222168,
      "learning_rate": 9.981717887065125e-06,
      "loss": 0.1232,
      "step": 716
    },
    {
      "epoch": 0.03702556158017041,
      "grad_norm": 8.764130592346191,
      "learning_rate": 9.981647816483806e-06,
      "loss": 0.1492,
      "step": 717
    },
    {
      "epoch": 0.03707720113607023,
      "grad_norm": 9.86923599243164,
      "learning_rate": 9.981577612124939e-06,
      "loss": 0.1553,
      "step": 718
    },
    {
      "epoch": 0.03712884069197005,
      "grad_norm": 10.040117263793945,
      "learning_rate": 9.981507273990408e-06,
      "loss": 0.1477,
      "step": 719
    },
    {
      "epoch": 0.037180480247869865,
      "grad_norm": 8.76535415649414,
      "learning_rate": 9.981436802082106e-06,
      "loss": 0.146,
      "step": 720
    },
    {
      "epoch": 0.03723211980376969,
      "grad_norm": 5.88051700592041,
      "learning_rate": 9.98136619640192e-06,
      "loss": 0.1001,
      "step": 721
    },
    {
      "epoch": 0.03728375935966951,
      "grad_norm": 7.897507667541504,
      "learning_rate": 9.98129545695175e-06,
      "loss": 0.0775,
      "step": 722
    },
    {
      "epoch": 0.03733539891556933,
      "grad_norm": 7.320996284484863,
      "learning_rate": 9.981224583733493e-06,
      "loss": 0.0907,
      "step": 723
    },
    {
      "epoch": 0.037387038471469146,
      "grad_norm": 2.793825149536133,
      "learning_rate": 9.981153576749055e-06,
      "loss": 0.0628,
      "step": 724
    },
    {
      "epoch": 0.037438678027368964,
      "grad_norm": 16.38835334777832,
      "learning_rate": 9.98108243600034e-06,
      "loss": 0.1682,
      "step": 725
    },
    {
      "epoch": 0.03749031758326878,
      "grad_norm": 7.031982421875,
      "learning_rate": 9.981011161489262e-06,
      "loss": 0.1241,
      "step": 726
    },
    {
      "epoch": 0.0375419571391686,
      "grad_norm": 9.072881698608398,
      "learning_rate": 9.98093975321773e-06,
      "loss": 0.0934,
      "step": 727
    },
    {
      "epoch": 0.03759359669506842,
      "grad_norm": 9.17534065246582,
      "learning_rate": 9.980868211187667e-06,
      "loss": 0.0839,
      "step": 728
    },
    {
      "epoch": 0.037645236250968245,
      "grad_norm": 9.812263488769531,
      "learning_rate": 9.980796535400991e-06,
      "loss": 0.0902,
      "step": 729
    },
    {
      "epoch": 0.03769687580686806,
      "grad_norm": 17.71500587463379,
      "learning_rate": 9.980724725859627e-06,
      "loss": 0.1609,
      "step": 730
    },
    {
      "epoch": 0.03774851536276788,
      "grad_norm": 10.455251693725586,
      "learning_rate": 9.980652782565505e-06,
      "loss": 0.1199,
      "step": 731
    },
    {
      "epoch": 0.0378001549186677,
      "grad_norm": 11.522178649902344,
      "learning_rate": 9.980580705520555e-06,
      "loss": 0.114,
      "step": 732
    },
    {
      "epoch": 0.03785179447456752,
      "grad_norm": 13.685524940490723,
      "learning_rate": 9.980508494726712e-06,
      "loss": 0.1292,
      "step": 733
    },
    {
      "epoch": 0.03790343403046734,
      "grad_norm": 14.935537338256836,
      "learning_rate": 9.980436150185918e-06,
      "loss": 0.0905,
      "step": 734
    },
    {
      "epoch": 0.037955073586367155,
      "grad_norm": 10.648333549499512,
      "learning_rate": 9.980363671900114e-06,
      "loss": 0.1382,
      "step": 735
    },
    {
      "epoch": 0.038006713142266974,
      "grad_norm": 9.073678970336914,
      "learning_rate": 9.980291059871245e-06,
      "loss": 0.0775,
      "step": 736
    },
    {
      "epoch": 0.0380583526981668,
      "grad_norm": 5.507081985473633,
      "learning_rate": 9.980218314101264e-06,
      "loss": 0.1078,
      "step": 737
    },
    {
      "epoch": 0.03810999225406662,
      "grad_norm": 5.281740665435791,
      "learning_rate": 9.980145434592123e-06,
      "loss": 0.0885,
      "step": 738
    },
    {
      "epoch": 0.038161631809966436,
      "grad_norm": 4.932861804962158,
      "learning_rate": 9.98007242134578e-06,
      "loss": 0.0703,
      "step": 739
    },
    {
      "epoch": 0.038213271365866254,
      "grad_norm": 8.659173011779785,
      "learning_rate": 9.979999274364194e-06,
      "loss": 0.1387,
      "step": 740
    },
    {
      "epoch": 0.03826491092176607,
      "grad_norm": 4.144313335418701,
      "learning_rate": 9.97992599364933e-06,
      "loss": 0.111,
      "step": 741
    },
    {
      "epoch": 0.03831655047766589,
      "grad_norm": 4.0075554847717285,
      "learning_rate": 9.979852579203155e-06,
      "loss": 0.1043,
      "step": 742
    },
    {
      "epoch": 0.03836819003356571,
      "grad_norm": 2.8916232585906982,
      "learning_rate": 9.979779031027643e-06,
      "loss": 0.0766,
      "step": 743
    },
    {
      "epoch": 0.03841982958946553,
      "grad_norm": 6.32964563369751,
      "learning_rate": 9.979705349124767e-06,
      "loss": 0.079,
      "step": 744
    },
    {
      "epoch": 0.03847146914536535,
      "grad_norm": 7.275129795074463,
      "learning_rate": 9.979631533496507e-06,
      "loss": 0.0758,
      "step": 745
    },
    {
      "epoch": 0.03852310870126517,
      "grad_norm": 6.906919002532959,
      "learning_rate": 9.979557584144843e-06,
      "loss": 0.1083,
      "step": 746
    },
    {
      "epoch": 0.03857474825716499,
      "grad_norm": 3.8760201930999756,
      "learning_rate": 9.979483501071762e-06,
      "loss": 0.085,
      "step": 747
    },
    {
      "epoch": 0.03862638781306481,
      "grad_norm": 4.3025360107421875,
      "learning_rate": 9.979409284279256e-06,
      "loss": 0.1083,
      "step": 748
    },
    {
      "epoch": 0.038678027368964626,
      "grad_norm": 13.04561996459961,
      "learning_rate": 9.979334933769315e-06,
      "loss": 0.1383,
      "step": 749
    },
    {
      "epoch": 0.038729666924864445,
      "grad_norm": 20.441804885864258,
      "learning_rate": 9.979260449543934e-06,
      "loss": 0.1739,
      "step": 750
    },
    {
      "epoch": 0.03878130648076426,
      "grad_norm": 8.820724487304688,
      "learning_rate": 9.97918583160512e-06,
      "loss": 0.0839,
      "step": 751
    },
    {
      "epoch": 0.03883294603666408,
      "grad_norm": 4.344066143035889,
      "learning_rate": 9.979111079954868e-06,
      "loss": 0.1264,
      "step": 752
    },
    {
      "epoch": 0.03888458559256391,
      "grad_norm": 4.008427143096924,
      "learning_rate": 9.979036194595193e-06,
      "loss": 0.0696,
      "step": 753
    },
    {
      "epoch": 0.038936225148463725,
      "grad_norm": 19.598247528076172,
      "learning_rate": 9.9789611755281e-06,
      "loss": 0.158,
      "step": 754
    },
    {
      "epoch": 0.038987864704363544,
      "grad_norm": 6.3417253494262695,
      "learning_rate": 9.97888602275561e-06,
      "loss": 0.1218,
      "step": 755
    },
    {
      "epoch": 0.03903950426026336,
      "grad_norm": 10.189105987548828,
      "learning_rate": 9.978810736279734e-06,
      "loss": 0.1624,
      "step": 756
    },
    {
      "epoch": 0.03909114381616318,
      "grad_norm": 27.945903778076172,
      "learning_rate": 9.9787353161025e-06,
      "loss": 0.1486,
      "step": 757
    },
    {
      "epoch": 0.039142783372063,
      "grad_norm": 4.879355430603027,
      "learning_rate": 9.97865976222593e-06,
      "loss": 0.1322,
      "step": 758
    },
    {
      "epoch": 0.03919442292796282,
      "grad_norm": 10.178874015808105,
      "learning_rate": 9.97858407465205e-06,
      "loss": 0.076,
      "step": 759
    },
    {
      "epoch": 0.039246062483862636,
      "grad_norm": 12.60031795501709,
      "learning_rate": 9.978508253382899e-06,
      "loss": 0.0983,
      "step": 760
    },
    {
      "epoch": 0.03929770203976246,
      "grad_norm": 3.5435080528259277,
      "learning_rate": 9.978432298420509e-06,
      "loss": 0.0862,
      "step": 761
    },
    {
      "epoch": 0.03934934159566228,
      "grad_norm": 20.257747650146484,
      "learning_rate": 9.978356209766922e-06,
      "loss": 0.2023,
      "step": 762
    },
    {
      "epoch": 0.0394009811515621,
      "grad_norm": 10.860344886779785,
      "learning_rate": 9.978279987424179e-06,
      "loss": 0.1288,
      "step": 763
    },
    {
      "epoch": 0.039452620707461916,
      "grad_norm": 3.3792474269866943,
      "learning_rate": 9.97820363139433e-06,
      "loss": 0.0521,
      "step": 764
    },
    {
      "epoch": 0.039504260263361735,
      "grad_norm": 10.980175971984863,
      "learning_rate": 9.97812714167942e-06,
      "loss": 0.0749,
      "step": 765
    },
    {
      "epoch": 0.03955589981926155,
      "grad_norm": 10.645520210266113,
      "learning_rate": 9.978050518281509e-06,
      "loss": 0.2014,
      "step": 766
    },
    {
      "epoch": 0.03960753937516137,
      "grad_norm": 8.906400680541992,
      "learning_rate": 9.977973761202653e-06,
      "loss": 0.0794,
      "step": 767
    },
    {
      "epoch": 0.03965917893106119,
      "grad_norm": 20.042911529541016,
      "learning_rate": 9.97789687044491e-06,
      "loss": 0.1279,
      "step": 768
    },
    {
      "epoch": 0.039710818486961015,
      "grad_norm": 4.938220500946045,
      "learning_rate": 9.977819846010348e-06,
      "loss": 0.0581,
      "step": 769
    },
    {
      "epoch": 0.03976245804286083,
      "grad_norm": 13.411139488220215,
      "learning_rate": 9.977742687901036e-06,
      "loss": 0.0945,
      "step": 770
    },
    {
      "epoch": 0.03981409759876065,
      "grad_norm": 11.164210319519043,
      "learning_rate": 9.977665396119042e-06,
      "loss": 0.1149,
      "step": 771
    },
    {
      "epoch": 0.03986573715466047,
      "grad_norm": 11.549676895141602,
      "learning_rate": 9.977587970666447e-06,
      "loss": 0.104,
      "step": 772
    },
    {
      "epoch": 0.03991737671056029,
      "grad_norm": 8.810893058776855,
      "learning_rate": 9.977510411545326e-06,
      "loss": 0.073,
      "step": 773
    },
    {
      "epoch": 0.03996901626646011,
      "grad_norm": 4.197683334350586,
      "learning_rate": 9.977432718757764e-06,
      "loss": 0.0485,
      "step": 774
    },
    {
      "epoch": 0.040020655822359925,
      "grad_norm": 10.598067283630371,
      "learning_rate": 9.977354892305846e-06,
      "loss": 0.1453,
      "step": 775
    },
    {
      "epoch": 0.040072295378259744,
      "grad_norm": 17.110424041748047,
      "learning_rate": 9.977276932191664e-06,
      "loss": 0.2526,
      "step": 776
    },
    {
      "epoch": 0.04012393493415957,
      "grad_norm": 4.0764055252075195,
      "learning_rate": 9.977198838417309e-06,
      "loss": 0.0693,
      "step": 777
    },
    {
      "epoch": 0.04017557449005939,
      "grad_norm": 9.176712036132812,
      "learning_rate": 9.977120610984878e-06,
      "loss": 0.1199,
      "step": 778
    },
    {
      "epoch": 0.040227214045959206,
      "grad_norm": 3.3183534145355225,
      "learning_rate": 9.977042249896474e-06,
      "loss": 0.0405,
      "step": 779
    },
    {
      "epoch": 0.040278853601859024,
      "grad_norm": 6.270786285400391,
      "learning_rate": 9.976963755154201e-06,
      "loss": 0.0877,
      "step": 780
    },
    {
      "epoch": 0.04033049315775884,
      "grad_norm": 3.622434377670288,
      "learning_rate": 9.976885126760164e-06,
      "loss": 0.0706,
      "step": 781
    },
    {
      "epoch": 0.04038213271365866,
      "grad_norm": 7.73430061340332,
      "learning_rate": 9.97680636471648e-06,
      "loss": 0.09,
      "step": 782
    },
    {
      "epoch": 0.04043377226955848,
      "grad_norm": 16.712129592895508,
      "learning_rate": 9.97672746902526e-06,
      "loss": 0.1539,
      "step": 783
    },
    {
      "epoch": 0.0404854118254583,
      "grad_norm": 15.203065872192383,
      "learning_rate": 9.976648439688621e-06,
      "loss": 0.2242,
      "step": 784
    },
    {
      "epoch": 0.04053705138135812,
      "grad_norm": 8.82221794128418,
      "learning_rate": 9.97656927670869e-06,
      "loss": 0.0995,
      "step": 785
    },
    {
      "epoch": 0.04058869093725794,
      "grad_norm": 6.283070087432861,
      "learning_rate": 9.97648998008759e-06,
      "loss": 0.0602,
      "step": 786
    },
    {
      "epoch": 0.04064033049315776,
      "grad_norm": 4.23555326461792,
      "learning_rate": 9.97641054982745e-06,
      "loss": 0.0488,
      "step": 787
    },
    {
      "epoch": 0.04069197004905758,
      "grad_norm": 17.642248153686523,
      "learning_rate": 9.976330985930404e-06,
      "loss": 0.0812,
      "step": 788
    },
    {
      "epoch": 0.0407436096049574,
      "grad_norm": 9.071032524108887,
      "learning_rate": 9.97625128839859e-06,
      "loss": 0.0983,
      "step": 789
    },
    {
      "epoch": 0.040795249160857215,
      "grad_norm": 12.044506072998047,
      "learning_rate": 9.976171457234147e-06,
      "loss": 0.1119,
      "step": 790
    },
    {
      "epoch": 0.040846888716757034,
      "grad_norm": 37.076171875,
      "learning_rate": 9.976091492439217e-06,
      "loss": 0.1713,
      "step": 791
    },
    {
      "epoch": 0.04089852827265685,
      "grad_norm": 6.6822638511657715,
      "learning_rate": 9.976011394015952e-06,
      "loss": 0.0614,
      "step": 792
    },
    {
      "epoch": 0.04095016782855668,
      "grad_norm": 4.462831974029541,
      "learning_rate": 9.975931161966496e-06,
      "loss": 0.0631,
      "step": 793
    },
    {
      "epoch": 0.041001807384456496,
      "grad_norm": 4.618829727172852,
      "learning_rate": 9.97585079629301e-06,
      "loss": 0.087,
      "step": 794
    },
    {
      "epoch": 0.041053446940356314,
      "grad_norm": 6.905725955963135,
      "learning_rate": 9.975770296997649e-06,
      "loss": 0.0999,
      "step": 795
    },
    {
      "epoch": 0.04110508649625613,
      "grad_norm": 17.7667236328125,
      "learning_rate": 9.975689664082575e-06,
      "loss": 0.1835,
      "step": 796
    },
    {
      "epoch": 0.04115672605215595,
      "grad_norm": 11.432498931884766,
      "learning_rate": 9.975608897549955e-06,
      "loss": 0.1447,
      "step": 797
    },
    {
      "epoch": 0.04120836560805577,
      "grad_norm": 18.048215866088867,
      "learning_rate": 9.975527997401955e-06,
      "loss": 0.1115,
      "step": 798
    },
    {
      "epoch": 0.04126000516395559,
      "grad_norm": 16.136343002319336,
      "learning_rate": 9.97544696364075e-06,
      "loss": 0.0977,
      "step": 799
    },
    {
      "epoch": 0.041311644719855406,
      "grad_norm": 9.306234359741211,
      "learning_rate": 9.975365796268515e-06,
      "loss": 0.1534,
      "step": 800
    },
    {
      "epoch": 0.04136328427575523,
      "grad_norm": 3.0302329063415527,
      "learning_rate": 9.97528449528743e-06,
      "loss": 0.0483,
      "step": 801
    },
    {
      "epoch": 0.04141492383165505,
      "grad_norm": 6.289890289306641,
      "learning_rate": 9.97520306069968e-06,
      "loss": 0.144,
      "step": 802
    },
    {
      "epoch": 0.04146656338755487,
      "grad_norm": 3.0505173206329346,
      "learning_rate": 9.975121492507447e-06,
      "loss": 0.0684,
      "step": 803
    },
    {
      "epoch": 0.041518202943454686,
      "grad_norm": 4.164590835571289,
      "learning_rate": 9.975039790712926e-06,
      "loss": 0.0657,
      "step": 804
    },
    {
      "epoch": 0.041569842499354505,
      "grad_norm": 9.316987037658691,
      "learning_rate": 9.97495795531831e-06,
      "loss": 0.0862,
      "step": 805
    },
    {
      "epoch": 0.04162148205525432,
      "grad_norm": 10.040600776672363,
      "learning_rate": 9.974875986325794e-06,
      "loss": 0.0633,
      "step": 806
    },
    {
      "epoch": 0.04167312161115414,
      "grad_norm": 5.297354698181152,
      "learning_rate": 9.974793883737584e-06,
      "loss": 0.0702,
      "step": 807
    },
    {
      "epoch": 0.04172476116705396,
      "grad_norm": 7.815354824066162,
      "learning_rate": 9.97471164755588e-06,
      "loss": 0.069,
      "step": 808
    },
    {
      "epoch": 0.041776400722953785,
      "grad_norm": 21.722492218017578,
      "learning_rate": 9.974629277782892e-06,
      "loss": 0.0939,
      "step": 809
    },
    {
      "epoch": 0.041828040278853604,
      "grad_norm": 14.803404808044434,
      "learning_rate": 9.974546774420834e-06,
      "loss": 0.0885,
      "step": 810
    },
    {
      "epoch": 0.04187967983475342,
      "grad_norm": 10.625041961669922,
      "learning_rate": 9.97446413747192e-06,
      "loss": 0.0868,
      "step": 811
    },
    {
      "epoch": 0.04193131939065324,
      "grad_norm": 13.500225067138672,
      "learning_rate": 9.974381366938368e-06,
      "loss": 0.1104,
      "step": 812
    },
    {
      "epoch": 0.04198295894655306,
      "grad_norm": 6.089194297790527,
      "learning_rate": 9.974298462822402e-06,
      "loss": 0.1083,
      "step": 813
    },
    {
      "epoch": 0.04203459850245288,
      "grad_norm": 8.907486915588379,
      "learning_rate": 9.974215425126249e-06,
      "loss": 0.1151,
      "step": 814
    },
    {
      "epoch": 0.042086238058352696,
      "grad_norm": 6.423090934753418,
      "learning_rate": 9.974132253852138e-06,
      "loss": 0.1157,
      "step": 815
    },
    {
      "epoch": 0.042137877614252514,
      "grad_norm": 9.31933879852295,
      "learning_rate": 9.974048949002302e-06,
      "loss": 0.1553,
      "step": 816
    },
    {
      "epoch": 0.04218951717015234,
      "grad_norm": 16.927087783813477,
      "learning_rate": 9.97396551057898e-06,
      "loss": 0.0861,
      "step": 817
    },
    {
      "epoch": 0.04224115672605216,
      "grad_norm": 12.393399238586426,
      "learning_rate": 9.97388193858441e-06,
      "loss": 0.0618,
      "step": 818
    },
    {
      "epoch": 0.042292796281951976,
      "grad_norm": 8.104972839355469,
      "learning_rate": 9.973798233020838e-06,
      "loss": 0.1376,
      "step": 819
    },
    {
      "epoch": 0.042344435837851795,
      "grad_norm": 4.567582607269287,
      "learning_rate": 9.97371439389051e-06,
      "loss": 0.1014,
      "step": 820
    },
    {
      "epoch": 0.04239607539375161,
      "grad_norm": 15.61298656463623,
      "learning_rate": 9.973630421195679e-06,
      "loss": 0.1707,
      "step": 821
    },
    {
      "epoch": 0.04244771494965143,
      "grad_norm": 3.928539514541626,
      "learning_rate": 9.973546314938601e-06,
      "loss": 0.1187,
      "step": 822
    },
    {
      "epoch": 0.04249935450555125,
      "grad_norm": 2.859912633895874,
      "learning_rate": 9.973462075121534e-06,
      "loss": 0.0457,
      "step": 823
    },
    {
      "epoch": 0.04255099406145107,
      "grad_norm": 7.588155269622803,
      "learning_rate": 9.973377701746738e-06,
      "loss": 0.1128,
      "step": 824
    },
    {
      "epoch": 0.042602633617350893,
      "grad_norm": 5.014381408691406,
      "learning_rate": 9.973293194816481e-06,
      "loss": 0.1357,
      "step": 825
    },
    {
      "epoch": 0.04265427317325071,
      "grad_norm": 11.294792175292969,
      "learning_rate": 9.973208554333031e-06,
      "loss": 0.2035,
      "step": 826
    },
    {
      "epoch": 0.04270591272915053,
      "grad_norm": 2.954505205154419,
      "learning_rate": 9.973123780298663e-06,
      "loss": 0.0831,
      "step": 827
    },
    {
      "epoch": 0.04275755228505035,
      "grad_norm": 6.452744007110596,
      "learning_rate": 9.97303887271565e-06,
      "loss": 0.1043,
      "step": 828
    },
    {
      "epoch": 0.04280919184095017,
      "grad_norm": 4.041753768920898,
      "learning_rate": 9.972953831586276e-06,
      "loss": 0.0684,
      "step": 829
    },
    {
      "epoch": 0.042860831396849985,
      "grad_norm": 2.5805106163024902,
      "learning_rate": 9.972868656912822e-06,
      "loss": 0.0629,
      "step": 830
    },
    {
      "epoch": 0.042912470952749804,
      "grad_norm": 14.144672393798828,
      "learning_rate": 9.97278334869758e-06,
      "loss": 0.1051,
      "step": 831
    },
    {
      "epoch": 0.04296411050864962,
      "grad_norm": 5.992001056671143,
      "learning_rate": 9.972697906942832e-06,
      "loss": 0.0539,
      "step": 832
    },
    {
      "epoch": 0.04301575006454945,
      "grad_norm": 9.389202117919922,
      "learning_rate": 9.97261233165088e-06,
      "loss": 0.1196,
      "step": 833
    },
    {
      "epoch": 0.043067389620449266,
      "grad_norm": 7.014853477478027,
      "learning_rate": 9.97252662282402e-06,
      "loss": 0.0679,
      "step": 834
    },
    {
      "epoch": 0.043119029176349084,
      "grad_norm": 7.218689441680908,
      "learning_rate": 9.972440780464552e-06,
      "loss": 0.138,
      "step": 835
    },
    {
      "epoch": 0.0431706687322489,
      "grad_norm": 3.6840243339538574,
      "learning_rate": 9.972354804574783e-06,
      "loss": 0.054,
      "step": 836
    },
    {
      "epoch": 0.04322230828814872,
      "grad_norm": 5.094080448150635,
      "learning_rate": 9.972268695157021e-06,
      "loss": 0.0589,
      "step": 837
    },
    {
      "epoch": 0.04327394784404854,
      "grad_norm": 13.549171447753906,
      "learning_rate": 9.972182452213579e-06,
      "loss": 0.2558,
      "step": 838
    },
    {
      "epoch": 0.04332558739994836,
      "grad_norm": 18.32237434387207,
      "learning_rate": 9.972096075746773e-06,
      "loss": 0.1599,
      "step": 839
    },
    {
      "epoch": 0.043377226955848176,
      "grad_norm": 3.897803783416748,
      "learning_rate": 9.972009565758921e-06,
      "loss": 0.0647,
      "step": 840
    },
    {
      "epoch": 0.043428866511748,
      "grad_norm": 4.4356207847595215,
      "learning_rate": 9.97192292225235e-06,
      "loss": 0.0377,
      "step": 841
    },
    {
      "epoch": 0.04348050606764782,
      "grad_norm": 18.65250587463379,
      "learning_rate": 9.971836145229383e-06,
      "loss": 0.1467,
      "step": 842
    },
    {
      "epoch": 0.04353214562354764,
      "grad_norm": 9.984923362731934,
      "learning_rate": 9.971749234692349e-06,
      "loss": 0.0756,
      "step": 843
    },
    {
      "epoch": 0.04358378517944746,
      "grad_norm": 8.722428321838379,
      "learning_rate": 9.971662190643587e-06,
      "loss": 0.0633,
      "step": 844
    },
    {
      "epoch": 0.043635424735347275,
      "grad_norm": 7.131714344024658,
      "learning_rate": 9.97157501308543e-06,
      "loss": 0.1065,
      "step": 845
    },
    {
      "epoch": 0.043687064291247094,
      "grad_norm": 4.427618980407715,
      "learning_rate": 9.971487702020223e-06,
      "loss": 0.0602,
      "step": 846
    },
    {
      "epoch": 0.04373870384714691,
      "grad_norm": 4.171732425689697,
      "learning_rate": 9.971400257450308e-06,
      "loss": 0.0778,
      "step": 847
    },
    {
      "epoch": 0.04379034340304673,
      "grad_norm": 13.429222106933594,
      "learning_rate": 9.971312679378032e-06,
      "loss": 0.1527,
      "step": 848
    },
    {
      "epoch": 0.043841982958946556,
      "grad_norm": 21.4901180267334,
      "learning_rate": 9.97122496780575e-06,
      "loss": 0.069,
      "step": 849
    },
    {
      "epoch": 0.043893622514846374,
      "grad_norm": 12.748005867004395,
      "learning_rate": 9.971137122735817e-06,
      "loss": 0.0772,
      "step": 850
    },
    {
      "epoch": 0.04394526207074619,
      "grad_norm": 10.852156639099121,
      "learning_rate": 9.971049144170589e-06,
      "loss": 0.0954,
      "step": 851
    },
    {
      "epoch": 0.04399690162664601,
      "grad_norm": 21.671937942504883,
      "learning_rate": 9.970961032112432e-06,
      "loss": 0.161,
      "step": 852
    },
    {
      "epoch": 0.04404854118254583,
      "grad_norm": 8.625605583190918,
      "learning_rate": 9.97087278656371e-06,
      "loss": 0.0684,
      "step": 853
    },
    {
      "epoch": 0.04410018073844565,
      "grad_norm": 11.761564254760742,
      "learning_rate": 9.970784407526793e-06,
      "loss": 0.0792,
      "step": 854
    },
    {
      "epoch": 0.044151820294345466,
      "grad_norm": 5.737818241119385,
      "learning_rate": 9.970695895004057e-06,
      "loss": 0.0562,
      "step": 855
    },
    {
      "epoch": 0.04420345985024529,
      "grad_norm": 13.937824249267578,
      "learning_rate": 9.970607248997874e-06,
      "loss": 0.0984,
      "step": 856
    },
    {
      "epoch": 0.04425509940614511,
      "grad_norm": 2.1514644622802734,
      "learning_rate": 9.97051846951063e-06,
      "loss": 0.0304,
      "step": 857
    },
    {
      "epoch": 0.04430673896204493,
      "grad_norm": 12.4420166015625,
      "learning_rate": 9.970429556544705e-06,
      "loss": 0.1126,
      "step": 858
    },
    {
      "epoch": 0.044358378517944747,
      "grad_norm": 11.500563621520996,
      "learning_rate": 9.970340510102488e-06,
      "loss": 0.1208,
      "step": 859
    },
    {
      "epoch": 0.044410018073844565,
      "grad_norm": 46.162227630615234,
      "learning_rate": 9.97025133018637e-06,
      "loss": 0.2157,
      "step": 860
    },
    {
      "epoch": 0.04446165762974438,
      "grad_norm": 25.29465103149414,
      "learning_rate": 9.970162016798746e-06,
      "loss": 0.2365,
      "step": 861
    },
    {
      "epoch": 0.0445132971856442,
      "grad_norm": 9.484661102294922,
      "learning_rate": 9.970072569942015e-06,
      "loss": 0.2403,
      "step": 862
    },
    {
      "epoch": 0.04456493674154402,
      "grad_norm": 18.08078956604004,
      "learning_rate": 9.969982989618579e-06,
      "loss": 0.1693,
      "step": 863
    },
    {
      "epoch": 0.044616576297443845,
      "grad_norm": 10.932611465454102,
      "learning_rate": 9.969893275830845e-06,
      "loss": 0.0947,
      "step": 864
    },
    {
      "epoch": 0.044668215853343664,
      "grad_norm": 7.465144634246826,
      "learning_rate": 9.969803428581218e-06,
      "loss": 0.1104,
      "step": 865
    },
    {
      "epoch": 0.04471985540924348,
      "grad_norm": 10.430362701416016,
      "learning_rate": 9.969713447872113e-06,
      "loss": 0.1101,
      "step": 866
    },
    {
      "epoch": 0.0447714949651433,
      "grad_norm": 5.117002487182617,
      "learning_rate": 9.969623333705946e-06,
      "loss": 0.0701,
      "step": 867
    },
    {
      "epoch": 0.04482313452104312,
      "grad_norm": 3.797203779220581,
      "learning_rate": 9.96953308608514e-06,
      "loss": 0.069,
      "step": 868
    },
    {
      "epoch": 0.04487477407694294,
      "grad_norm": 20.687946319580078,
      "learning_rate": 9.969442705012113e-06,
      "loss": 0.1214,
      "step": 869
    },
    {
      "epoch": 0.044926413632842756,
      "grad_norm": 17.44894790649414,
      "learning_rate": 9.969352190489297e-06,
      "loss": 0.1702,
      "step": 870
    },
    {
      "epoch": 0.044978053188742574,
      "grad_norm": 4.799624919891357,
      "learning_rate": 9.96926154251912e-06,
      "loss": 0.0717,
      "step": 871
    },
    {
      "epoch": 0.0450296927446424,
      "grad_norm": 8.96164321899414,
      "learning_rate": 9.969170761104017e-06,
      "loss": 0.1146,
      "step": 872
    },
    {
      "epoch": 0.04508133230054222,
      "grad_norm": 6.189146041870117,
      "learning_rate": 9.969079846246426e-06,
      "loss": 0.074,
      "step": 873
    },
    {
      "epoch": 0.045132971856442036,
      "grad_norm": 2.227996587753296,
      "learning_rate": 9.968988797948788e-06,
      "loss": 0.0544,
      "step": 874
    },
    {
      "epoch": 0.045184611412341855,
      "grad_norm": 4.4484429359436035,
      "learning_rate": 9.968897616213549e-06,
      "loss": 0.0926,
      "step": 875
    },
    {
      "epoch": 0.04523625096824167,
      "grad_norm": 6.165963172912598,
      "learning_rate": 9.968806301043156e-06,
      "loss": 0.0572,
      "step": 876
    },
    {
      "epoch": 0.04528789052414149,
      "grad_norm": 6.131039142608643,
      "learning_rate": 9.968714852440062e-06,
      "loss": 0.1019,
      "step": 877
    },
    {
      "epoch": 0.04533953008004131,
      "grad_norm": 3.9879472255706787,
      "learning_rate": 9.968623270406725e-06,
      "loss": 0.0879,
      "step": 878
    },
    {
      "epoch": 0.04539116963594113,
      "grad_norm": 7.463747024536133,
      "learning_rate": 9.9685315549456e-06,
      "loss": 0.101,
      "step": 879
    },
    {
      "epoch": 0.045442809191840954,
      "grad_norm": 15.50757122039795,
      "learning_rate": 9.968439706059153e-06,
      "loss": 0.1272,
      "step": 880
    },
    {
      "epoch": 0.04549444874774077,
      "grad_norm": 7.287950038909912,
      "learning_rate": 9.968347723749849e-06,
      "loss": 0.0874,
      "step": 881
    },
    {
      "epoch": 0.04554608830364059,
      "grad_norm": 5.735359191894531,
      "learning_rate": 9.96825560802016e-06,
      "loss": 0.142,
      "step": 882
    },
    {
      "epoch": 0.04559772785954041,
      "grad_norm": 10.07146167755127,
      "learning_rate": 9.968163358872557e-06,
      "loss": 0.133,
      "step": 883
    },
    {
      "epoch": 0.04564936741544023,
      "grad_norm": 13.110625267028809,
      "learning_rate": 9.968070976309518e-06,
      "loss": 0.071,
      "step": 884
    },
    {
      "epoch": 0.045701006971340045,
      "grad_norm": 17.043792724609375,
      "learning_rate": 9.967978460333528e-06,
      "loss": 0.1711,
      "step": 885
    },
    {
      "epoch": 0.045752646527239864,
      "grad_norm": 6.937654972076416,
      "learning_rate": 9.967885810947064e-06,
      "loss": 0.0973,
      "step": 886
    },
    {
      "epoch": 0.04580428608313968,
      "grad_norm": 5.862943649291992,
      "learning_rate": 9.96779302815262e-06,
      "loss": 0.1229,
      "step": 887
    },
    {
      "epoch": 0.04585592563903951,
      "grad_norm": 5.186105728149414,
      "learning_rate": 9.967700111952685e-06,
      "loss": 0.0982,
      "step": 888
    },
    {
      "epoch": 0.045907565194939326,
      "grad_norm": 3.3940625190734863,
      "learning_rate": 9.967607062349755e-06,
      "loss": 0.0351,
      "step": 889
    },
    {
      "epoch": 0.045959204750839144,
      "grad_norm": 5.810781955718994,
      "learning_rate": 9.967513879346329e-06,
      "loss": 0.104,
      "step": 890
    },
    {
      "epoch": 0.04601084430673896,
      "grad_norm": 8.502846717834473,
      "learning_rate": 9.967420562944909e-06,
      "loss": 0.136,
      "step": 891
    },
    {
      "epoch": 0.04606248386263878,
      "grad_norm": 7.6176252365112305,
      "learning_rate": 9.967327113148e-06,
      "loss": 0.0543,
      "step": 892
    },
    {
      "epoch": 0.0461141234185386,
      "grad_norm": 10.15440845489502,
      "learning_rate": 9.967233529958114e-06,
      "loss": 0.0919,
      "step": 893
    },
    {
      "epoch": 0.04616576297443842,
      "grad_norm": 13.307798385620117,
      "learning_rate": 9.96713981337776e-06,
      "loss": 0.1082,
      "step": 894
    },
    {
      "epoch": 0.046217402530338236,
      "grad_norm": 14.547751426696777,
      "learning_rate": 9.967045963409458e-06,
      "loss": 0.0986,
      "step": 895
    },
    {
      "epoch": 0.04626904208623806,
      "grad_norm": 12.101489067077637,
      "learning_rate": 9.966951980055728e-06,
      "loss": 0.078,
      "step": 896
    },
    {
      "epoch": 0.04632068164213788,
      "grad_norm": 7.015307426452637,
      "learning_rate": 9.966857863319092e-06,
      "loss": 0.0853,
      "step": 897
    },
    {
      "epoch": 0.0463723211980377,
      "grad_norm": 10.912406921386719,
      "learning_rate": 9.96676361320208e-06,
      "loss": 0.1652,
      "step": 898
    },
    {
      "epoch": 0.04642396075393752,
      "grad_norm": 14.02098274230957,
      "learning_rate": 9.96666922970722e-06,
      "loss": 0.2262,
      "step": 899
    },
    {
      "epoch": 0.046475600309837335,
      "grad_norm": 15.92883014678955,
      "learning_rate": 9.966574712837049e-06,
      "loss": 0.1681,
      "step": 900
    },
    {
      "epoch": 0.046527239865737154,
      "grad_norm": 6.429821014404297,
      "learning_rate": 9.966480062594106e-06,
      "loss": 0.0619,
      "step": 901
    },
    {
      "epoch": 0.04657887942163697,
      "grad_norm": 12.450419425964355,
      "learning_rate": 9.966385278980929e-06,
      "loss": 0.1218,
      "step": 902
    },
    {
      "epoch": 0.04663051897753679,
      "grad_norm": 9.597604751586914,
      "learning_rate": 9.966290362000065e-06,
      "loss": 0.1281,
      "step": 903
    },
    {
      "epoch": 0.046682158533436616,
      "grad_norm": 13.726762771606445,
      "learning_rate": 9.966195311654063e-06,
      "loss": 0.1737,
      "step": 904
    },
    {
      "epoch": 0.046733798089336434,
      "grad_norm": 4.244523048400879,
      "learning_rate": 9.966100127945478e-06,
      "loss": 0.1068,
      "step": 905
    },
    {
      "epoch": 0.04678543764523625,
      "grad_norm": 7.635623455047607,
      "learning_rate": 9.966004810876861e-06,
      "loss": 0.112,
      "step": 906
    },
    {
      "epoch": 0.04683707720113607,
      "grad_norm": 20.392833709716797,
      "learning_rate": 9.965909360450778e-06,
      "loss": 0.0981,
      "step": 907
    },
    {
      "epoch": 0.04688871675703589,
      "grad_norm": 8.07287311553955,
      "learning_rate": 9.965813776669786e-06,
      "loss": 0.0783,
      "step": 908
    },
    {
      "epoch": 0.04694035631293571,
      "grad_norm": 5.917474269866943,
      "learning_rate": 9.965718059536456e-06,
      "loss": 0.079,
      "step": 909
    },
    {
      "epoch": 0.046991995868835526,
      "grad_norm": 7.905811309814453,
      "learning_rate": 9.965622209053356e-06,
      "loss": 0.1234,
      "step": 910
    },
    {
      "epoch": 0.047043635424735344,
      "grad_norm": 7.083068370819092,
      "learning_rate": 9.965526225223061e-06,
      "loss": 0.0858,
      "step": 911
    },
    {
      "epoch": 0.04709527498063517,
      "grad_norm": 4.612298965454102,
      "learning_rate": 9.96543010804815e-06,
      "loss": 0.084,
      "step": 912
    },
    {
      "epoch": 0.04714691453653499,
      "grad_norm": 10.781512260437012,
      "learning_rate": 9.9653338575312e-06,
      "loss": 0.1173,
      "step": 913
    },
    {
      "epoch": 0.04719855409243481,
      "grad_norm": 5.375583648681641,
      "learning_rate": 9.9652374736748e-06,
      "loss": 0.0943,
      "step": 914
    },
    {
      "epoch": 0.047250193648334625,
      "grad_norm": 17.156444549560547,
      "learning_rate": 9.965140956481537e-06,
      "loss": 0.1397,
      "step": 915
    },
    {
      "epoch": 0.04730183320423444,
      "grad_norm": 3.3991732597351074,
      "learning_rate": 9.965044305954002e-06,
      "loss": 0.0644,
      "step": 916
    },
    {
      "epoch": 0.04735347276013426,
      "grad_norm": 7.4486212730407715,
      "learning_rate": 9.964947522094792e-06,
      "loss": 0.0997,
      "step": 917
    },
    {
      "epoch": 0.04740511231603408,
      "grad_norm": 8.728047370910645,
      "learning_rate": 9.964850604906502e-06,
      "loss": 0.0551,
      "step": 918
    },
    {
      "epoch": 0.0474567518719339,
      "grad_norm": 10.38275146484375,
      "learning_rate": 9.96475355439174e-06,
      "loss": 0.0652,
      "step": 919
    },
    {
      "epoch": 0.047508391427833724,
      "grad_norm": 14.692436218261719,
      "learning_rate": 9.96465637055311e-06,
      "loss": 0.1639,
      "step": 920
    },
    {
      "epoch": 0.04756003098373354,
      "grad_norm": 17.322872161865234,
      "learning_rate": 9.96455905339322e-06,
      "loss": 0.1761,
      "step": 921
    },
    {
      "epoch": 0.04761167053963336,
      "grad_norm": 11.784518241882324,
      "learning_rate": 9.964461602914688e-06,
      "loss": 0.1484,
      "step": 922
    },
    {
      "epoch": 0.04766331009553318,
      "grad_norm": 9.855957984924316,
      "learning_rate": 9.964364019120124e-06,
      "loss": 0.0943,
      "step": 923
    },
    {
      "epoch": 0.047714949651433,
      "grad_norm": 6.033347129821777,
      "learning_rate": 9.964266302012156e-06,
      "loss": 0.0422,
      "step": 924
    },
    {
      "epoch": 0.047766589207332816,
      "grad_norm": 6.810412883758545,
      "learning_rate": 9.964168451593403e-06,
      "loss": 0.0547,
      "step": 925
    },
    {
      "epoch": 0.047818228763232634,
      "grad_norm": 10.484560012817383,
      "learning_rate": 9.964070467866495e-06,
      "loss": 0.1067,
      "step": 926
    },
    {
      "epoch": 0.04786986831913245,
      "grad_norm": 3.1187899112701416,
      "learning_rate": 9.963972350834062e-06,
      "loss": 0.0629,
      "step": 927
    },
    {
      "epoch": 0.04792150787503228,
      "grad_norm": 11.22729206085205,
      "learning_rate": 9.963874100498738e-06,
      "loss": 0.1496,
      "step": 928
    },
    {
      "epoch": 0.047973147430932096,
      "grad_norm": 7.338229656219482,
      "learning_rate": 9.963775716863164e-06,
      "loss": 0.0945,
      "step": 929
    },
    {
      "epoch": 0.048024786986831915,
      "grad_norm": 12.217137336730957,
      "learning_rate": 9.963677199929982e-06,
      "loss": 0.168,
      "step": 930
    },
    {
      "epoch": 0.04807642654273173,
      "grad_norm": 11.58438491821289,
      "learning_rate": 9.963578549701835e-06,
      "loss": 0.1317,
      "step": 931
    },
    {
      "epoch": 0.04812806609863155,
      "grad_norm": 10.187458038330078,
      "learning_rate": 9.963479766181374e-06,
      "loss": 0.0657,
      "step": 932
    },
    {
      "epoch": 0.04817970565453137,
      "grad_norm": 4.435177326202393,
      "learning_rate": 9.96338084937125e-06,
      "loss": 0.0529,
      "step": 933
    },
    {
      "epoch": 0.04823134521043119,
      "grad_norm": 4.478713512420654,
      "learning_rate": 9.96328179927412e-06,
      "loss": 0.0855,
      "step": 934
    },
    {
      "epoch": 0.04828298476633101,
      "grad_norm": 6.746525287628174,
      "learning_rate": 9.963182615892647e-06,
      "loss": 0.0833,
      "step": 935
    },
    {
      "epoch": 0.04833462432223083,
      "grad_norm": 13.77139949798584,
      "learning_rate": 9.963083299229491e-06,
      "loss": 0.0644,
      "step": 936
    },
    {
      "epoch": 0.04838626387813065,
      "grad_norm": 9.51870346069336,
      "learning_rate": 9.96298384928732e-06,
      "loss": 0.1266,
      "step": 937
    },
    {
      "epoch": 0.04843790343403047,
      "grad_norm": 7.060708045959473,
      "learning_rate": 9.962884266068804e-06,
      "loss": 0.0864,
      "step": 938
    },
    {
      "epoch": 0.04848954298993029,
      "grad_norm": 3.317643642425537,
      "learning_rate": 9.962784549576618e-06,
      "loss": 0.0517,
      "step": 939
    },
    {
      "epoch": 0.048541182545830105,
      "grad_norm": 12.902382850646973,
      "learning_rate": 9.96268469981344e-06,
      "loss": 0.172,
      "step": 940
    },
    {
      "epoch": 0.048592822101729924,
      "grad_norm": 3.9879536628723145,
      "learning_rate": 9.962584716781951e-06,
      "loss": 0.043,
      "step": 941
    },
    {
      "epoch": 0.04864446165762974,
      "grad_norm": 6.85207462310791,
      "learning_rate": 9.962484600484836e-06,
      "loss": 0.0566,
      "step": 942
    },
    {
      "epoch": 0.04869610121352956,
      "grad_norm": 6.381924152374268,
      "learning_rate": 9.962384350924784e-06,
      "loss": 0.0695,
      "step": 943
    },
    {
      "epoch": 0.048747740769429386,
      "grad_norm": 9.725566864013672,
      "learning_rate": 9.962283968104487e-06,
      "loss": 0.1154,
      "step": 944
    },
    {
      "epoch": 0.048799380325329204,
      "grad_norm": 15.946682929992676,
      "learning_rate": 9.96218345202664e-06,
      "loss": 0.1493,
      "step": 945
    },
    {
      "epoch": 0.04885101988122902,
      "grad_norm": 10.69935417175293,
      "learning_rate": 9.96208280269394e-06,
      "loss": 0.0692,
      "step": 946
    },
    {
      "epoch": 0.04890265943712884,
      "grad_norm": 10.50161361694336,
      "learning_rate": 9.961982020109095e-06,
      "loss": 0.0975,
      "step": 947
    },
    {
      "epoch": 0.04895429899302866,
      "grad_norm": 16.94684600830078,
      "learning_rate": 9.96188110427481e-06,
      "loss": 0.1007,
      "step": 948
    },
    {
      "epoch": 0.04900593854892848,
      "grad_norm": 10.500828742980957,
      "learning_rate": 9.96178005519379e-06,
      "loss": 0.1094,
      "step": 949
    },
    {
      "epoch": 0.049057578104828296,
      "grad_norm": 16.364707946777344,
      "learning_rate": 9.961678872868755e-06,
      "loss": 0.0858,
      "step": 950
    },
    {
      "epoch": 0.049109217660728115,
      "grad_norm": 19.817697525024414,
      "learning_rate": 9.961577557302419e-06,
      "loss": 0.2315,
      "step": 951
    },
    {
      "epoch": 0.04916085721662794,
      "grad_norm": 8.440033912658691,
      "learning_rate": 9.961476108497503e-06,
      "loss": 0.1122,
      "step": 952
    },
    {
      "epoch": 0.04921249677252776,
      "grad_norm": 4.382153511047363,
      "learning_rate": 9.961374526456733e-06,
      "loss": 0.0594,
      "step": 953
    },
    {
      "epoch": 0.04926413632842758,
      "grad_norm": 5.006882667541504,
      "learning_rate": 9.961272811182834e-06,
      "loss": 0.0661,
      "step": 954
    },
    {
      "epoch": 0.049315775884327395,
      "grad_norm": 6.186721324920654,
      "learning_rate": 9.961170962678539e-06,
      "loss": 0.0667,
      "step": 955
    },
    {
      "epoch": 0.049367415440227214,
      "grad_norm": 6.530024528503418,
      "learning_rate": 9.961068980946584e-06,
      "loss": 0.0648,
      "step": 956
    },
    {
      "epoch": 0.04941905499612703,
      "grad_norm": 5.2576375007629395,
      "learning_rate": 9.960966865989706e-06,
      "loss": 0.0812,
      "step": 957
    },
    {
      "epoch": 0.04947069455202685,
      "grad_norm": 19.271772384643555,
      "learning_rate": 9.960864617810647e-06,
      "loss": 0.0954,
      "step": 958
    },
    {
      "epoch": 0.04952233410792667,
      "grad_norm": 4.333558559417725,
      "learning_rate": 9.960762236412156e-06,
      "loss": 0.0897,
      "step": 959
    },
    {
      "epoch": 0.049573973663826494,
      "grad_norm": 5.439894199371338,
      "learning_rate": 9.960659721796978e-06,
      "loss": 0.074,
      "step": 960
    },
    {
      "epoch": 0.04962561321972631,
      "grad_norm": 8.184279441833496,
      "learning_rate": 9.960557073967869e-06,
      "loss": 0.1208,
      "step": 961
    },
    {
      "epoch": 0.04967725277562613,
      "grad_norm": 5.069379806518555,
      "learning_rate": 9.960454292927584e-06,
      "loss": 0.078,
      "step": 962
    },
    {
      "epoch": 0.04972889233152595,
      "grad_norm": 8.590649604797363,
      "learning_rate": 9.960351378678886e-06,
      "loss": 0.0711,
      "step": 963
    },
    {
      "epoch": 0.04978053188742577,
      "grad_norm": 8.992157936096191,
      "learning_rate": 9.960248331224532e-06,
      "loss": 0.1667,
      "step": 964
    },
    {
      "epoch": 0.049832171443325586,
      "grad_norm": 9.380717277526855,
      "learning_rate": 9.960145150567295e-06,
      "loss": 0.0971,
      "step": 965
    },
    {
      "epoch": 0.049883810999225404,
      "grad_norm": 6.108704566955566,
      "learning_rate": 9.960041836709947e-06,
      "loss": 0.0851,
      "step": 966
    },
    {
      "epoch": 0.04993545055512522,
      "grad_norm": 4.403829097747803,
      "learning_rate": 9.959938389655256e-06,
      "loss": 0.0737,
      "step": 967
    },
    {
      "epoch": 0.04998709011102505,
      "grad_norm": 3.4059555530548096,
      "learning_rate": 9.959834809406004e-06,
      "loss": 0.0492,
      "step": 968
    },
    {
      "epoch": 0.05003872966692487,
      "grad_norm": 10.863164901733398,
      "learning_rate": 9.959731095964974e-06,
      "loss": 0.1275,
      "step": 969
    },
    {
      "epoch": 0.050090369222824685,
      "grad_norm": 9.847679138183594,
      "learning_rate": 9.959627249334949e-06,
      "loss": 0.0891,
      "step": 970
    },
    {
      "epoch": 0.0501420087787245,
      "grad_norm": 27.966001510620117,
      "learning_rate": 9.959523269518718e-06,
      "loss": 0.1379,
      "step": 971
    },
    {
      "epoch": 0.05019364833462432,
      "grad_norm": 22.834850311279297,
      "learning_rate": 9.959419156519073e-06,
      "loss": 0.1399,
      "step": 972
    },
    {
      "epoch": 0.05024528789052414,
      "grad_norm": 5.0529375076293945,
      "learning_rate": 9.959314910338809e-06,
      "loss": 0.0816,
      "step": 973
    },
    {
      "epoch": 0.05029692744642396,
      "grad_norm": 6.090721130371094,
      "learning_rate": 9.959210530980728e-06,
      "loss": 0.1391,
      "step": 974
    },
    {
      "epoch": 0.05034856700232378,
      "grad_norm": 7.247451305389404,
      "learning_rate": 9.959106018447631e-06,
      "loss": 0.0966,
      "step": 975
    },
    {
      "epoch": 0.0504002065582236,
      "grad_norm": 6.551202297210693,
      "learning_rate": 9.959001372742325e-06,
      "loss": 0.0732,
      "step": 976
    },
    {
      "epoch": 0.05045184611412342,
      "grad_norm": 6.896840572357178,
      "learning_rate": 9.958896593867621e-06,
      "loss": 0.0766,
      "step": 977
    },
    {
      "epoch": 0.05050348567002324,
      "grad_norm": 13.101635932922363,
      "learning_rate": 9.958791681826332e-06,
      "loss": 0.1229,
      "step": 978
    },
    {
      "epoch": 0.05055512522592306,
      "grad_norm": 5.680718898773193,
      "learning_rate": 9.958686636621276e-06,
      "loss": 0.1027,
      "step": 979
    },
    {
      "epoch": 0.050606764781822876,
      "grad_norm": 3.7764272689819336,
      "learning_rate": 9.958581458255271e-06,
      "loss": 0.069,
      "step": 980
    },
    {
      "epoch": 0.050658404337722694,
      "grad_norm": 7.32869291305542,
      "learning_rate": 9.958476146731147e-06,
      "loss": 0.057,
      "step": 981
    },
    {
      "epoch": 0.05071004389362251,
      "grad_norm": 10.05722427368164,
      "learning_rate": 9.958370702051728e-06,
      "loss": 0.1089,
      "step": 982
    },
    {
      "epoch": 0.05076168344952233,
      "grad_norm": 13.773096084594727,
      "learning_rate": 9.958265124219845e-06,
      "loss": 0.1543,
      "step": 983
    },
    {
      "epoch": 0.050813323005422156,
      "grad_norm": 7.886497974395752,
      "learning_rate": 9.958159413238335e-06,
      "loss": 0.1026,
      "step": 984
    },
    {
      "epoch": 0.050864962561321975,
      "grad_norm": 27.10460662841797,
      "learning_rate": 9.958053569110036e-06,
      "loss": 0.2791,
      "step": 985
    },
    {
      "epoch": 0.05091660211722179,
      "grad_norm": 6.482113838195801,
      "learning_rate": 9.957947591837792e-06,
      "loss": 0.1079,
      "step": 986
    },
    {
      "epoch": 0.05096824167312161,
      "grad_norm": 5.136651992797852,
      "learning_rate": 9.957841481424447e-06,
      "loss": 0.0582,
      "step": 987
    },
    {
      "epoch": 0.05101988122902143,
      "grad_norm": 5.925328731536865,
      "learning_rate": 9.957735237872852e-06,
      "loss": 0.0768,
      "step": 988
    },
    {
      "epoch": 0.05107152078492125,
      "grad_norm": 6.507165908813477,
      "learning_rate": 9.957628861185859e-06,
      "loss": 0.1127,
      "step": 989
    },
    {
      "epoch": 0.05112316034082107,
      "grad_norm": 6.649933338165283,
      "learning_rate": 9.957522351366325e-06,
      "loss": 0.1051,
      "step": 990
    },
    {
      "epoch": 0.051174799896720885,
      "grad_norm": 7.771978378295898,
      "learning_rate": 9.957415708417108e-06,
      "loss": 0.088,
      "step": 991
    },
    {
      "epoch": 0.05122643945262071,
      "grad_norm": 15.413185119628906,
      "learning_rate": 9.957308932341077e-06,
      "loss": 0.1819,
      "step": 992
    },
    {
      "epoch": 0.05127807900852053,
      "grad_norm": 14.970609664916992,
      "learning_rate": 9.957202023141095e-06,
      "loss": 0.1245,
      "step": 993
    },
    {
      "epoch": 0.05132971856442035,
      "grad_norm": 4.743940830230713,
      "learning_rate": 9.957094980820033e-06,
      "loss": 0.0847,
      "step": 994
    },
    {
      "epoch": 0.051381358120320166,
      "grad_norm": 5.798779010772705,
      "learning_rate": 9.956987805380769e-06,
      "loss": 0.1401,
      "step": 995
    },
    {
      "epoch": 0.051432997676219984,
      "grad_norm": 20.263269424438477,
      "learning_rate": 9.956880496826177e-06,
      "loss": 0.1596,
      "step": 996
    },
    {
      "epoch": 0.0514846372321198,
      "grad_norm": 15.614092826843262,
      "learning_rate": 9.956773055159142e-06,
      "loss": 0.1054,
      "step": 997
    },
    {
      "epoch": 0.05153627678801962,
      "grad_norm": 6.600391864776611,
      "learning_rate": 9.956665480382547e-06,
      "loss": 0.1167,
      "step": 998
    },
    {
      "epoch": 0.05158791634391944,
      "grad_norm": 9.139406204223633,
      "learning_rate": 9.95655777249928e-06,
      "loss": 0.0873,
      "step": 999
    },
    {
      "epoch": 0.051639555899819264,
      "grad_norm": 16.481441497802734,
      "learning_rate": 9.956449931512238e-06,
      "loss": 0.2194,
      "step": 1000
    },
    {
      "epoch": 0.05169119545571908,
      "grad_norm": 11.570337295532227,
      "learning_rate": 9.956341957424312e-06,
      "loss": 0.1191,
      "step": 1001
    },
    {
      "epoch": 0.0517428350116189,
      "grad_norm": 9.509108543395996,
      "learning_rate": 9.956233850238404e-06,
      "loss": 0.0989,
      "step": 1002
    },
    {
      "epoch": 0.05179447456751872,
      "grad_norm": 6.320773124694824,
      "learning_rate": 9.956125609957417e-06,
      "loss": 0.0997,
      "step": 1003
    },
    {
      "epoch": 0.05184611412341854,
      "grad_norm": 25.902307510375977,
      "learning_rate": 9.956017236584258e-06,
      "loss": 0.1476,
      "step": 1004
    },
    {
      "epoch": 0.051897753679318356,
      "grad_norm": 24.61500358581543,
      "learning_rate": 9.955908730121835e-06,
      "loss": 0.2034,
      "step": 1005
    },
    {
      "epoch": 0.051949393235218175,
      "grad_norm": 14.049229621887207,
      "learning_rate": 9.955800090573064e-06,
      "loss": 0.1139,
      "step": 1006
    },
    {
      "epoch": 0.05200103279111799,
      "grad_norm": 8.833514213562012,
      "learning_rate": 9.955691317940863e-06,
      "loss": 0.0787,
      "step": 1007
    },
    {
      "epoch": 0.05205267234701782,
      "grad_norm": 4.528606414794922,
      "learning_rate": 9.955582412228151e-06,
      "loss": 0.1122,
      "step": 1008
    },
    {
      "epoch": 0.05210431190291764,
      "grad_norm": 4.34523868560791,
      "learning_rate": 9.955473373437852e-06,
      "loss": 0.0326,
      "step": 1009
    },
    {
      "epoch": 0.052155951458817455,
      "grad_norm": 4.2879557609558105,
      "learning_rate": 9.955364201572897e-06,
      "loss": 0.0516,
      "step": 1010
    },
    {
      "epoch": 0.052207591014717274,
      "grad_norm": 4.609678268432617,
      "learning_rate": 9.955254896636217e-06,
      "loss": 0.0743,
      "step": 1011
    },
    {
      "epoch": 0.05225923057061709,
      "grad_norm": 5.277791500091553,
      "learning_rate": 9.955145458630745e-06,
      "loss": 0.0898,
      "step": 1012
    },
    {
      "epoch": 0.05231087012651691,
      "grad_norm": 4.0711350440979,
      "learning_rate": 9.955035887559421e-06,
      "loss": 0.0403,
      "step": 1013
    },
    {
      "epoch": 0.05236250968241673,
      "grad_norm": 5.725537300109863,
      "learning_rate": 9.95492618342519e-06,
      "loss": 0.0741,
      "step": 1014
    },
    {
      "epoch": 0.05241414923831655,
      "grad_norm": 23.1632022857666,
      "learning_rate": 9.954816346230992e-06,
      "loss": 0.1741,
      "step": 1015
    },
    {
      "epoch": 0.05246578879421637,
      "grad_norm": 8.518187522888184,
      "learning_rate": 9.954706375979786e-06,
      "loss": 0.0915,
      "step": 1016
    },
    {
      "epoch": 0.05251742835011619,
      "grad_norm": 9.00818920135498,
      "learning_rate": 9.954596272674516e-06,
      "loss": 0.1466,
      "step": 1017
    },
    {
      "epoch": 0.05256906790601601,
      "grad_norm": 14.40527057647705,
      "learning_rate": 9.954486036318143e-06,
      "loss": 0.1883,
      "step": 1018
    },
    {
      "epoch": 0.05262070746191583,
      "grad_norm": 4.505095958709717,
      "learning_rate": 9.954375666913625e-06,
      "loss": 0.0976,
      "step": 1019
    },
    {
      "epoch": 0.052672347017815646,
      "grad_norm": 4.5201544761657715,
      "learning_rate": 9.954265164463928e-06,
      "loss": 0.0523,
      "step": 1020
    },
    {
      "epoch": 0.052723986573715464,
      "grad_norm": 12.65467357635498,
      "learning_rate": 9.954154528972019e-06,
      "loss": 0.1143,
      "step": 1021
    },
    {
      "epoch": 0.05277562612961528,
      "grad_norm": 11.090418815612793,
      "learning_rate": 9.95404376044087e-06,
      "loss": 0.0991,
      "step": 1022
    },
    {
      "epoch": 0.0528272656855151,
      "grad_norm": 23.522214889526367,
      "learning_rate": 9.953932858873454e-06,
      "loss": 0.1102,
      "step": 1023
    },
    {
      "epoch": 0.05287890524141493,
      "grad_norm": 10.948710441589355,
      "learning_rate": 9.95382182427275e-06,
      "loss": 0.0814,
      "step": 1024
    },
    {
      "epoch": 0.052930544797314745,
      "grad_norm": 10.393889427185059,
      "learning_rate": 9.95371065664174e-06,
      "loss": 0.1008,
      "step": 1025
    },
    {
      "epoch": 0.05298218435321456,
      "grad_norm": 11.750814437866211,
      "learning_rate": 9.953599355983407e-06,
      "loss": 0.107,
      "step": 1026
    },
    {
      "epoch": 0.05303382390911438,
      "grad_norm": 9.13149356842041,
      "learning_rate": 9.953487922300742e-06,
      "loss": 0.0895,
      "step": 1027
    },
    {
      "epoch": 0.0530854634650142,
      "grad_norm": 5.038003921508789,
      "learning_rate": 9.953376355596738e-06,
      "loss": 0.0461,
      "step": 1028
    },
    {
      "epoch": 0.05313710302091402,
      "grad_norm": 38.645835876464844,
      "learning_rate": 9.953264655874389e-06,
      "loss": 0.2629,
      "step": 1029
    },
    {
      "epoch": 0.05318874257681384,
      "grad_norm": 12.356998443603516,
      "learning_rate": 9.953152823136695e-06,
      "loss": 0.127,
      "step": 1030
    },
    {
      "epoch": 0.053240382132713655,
      "grad_norm": 8.27319049835205,
      "learning_rate": 9.95304085738666e-06,
      "loss": 0.1109,
      "step": 1031
    },
    {
      "epoch": 0.05329202168861348,
      "grad_norm": 20.264049530029297,
      "learning_rate": 9.952928758627291e-06,
      "loss": 0.1278,
      "step": 1032
    },
    {
      "epoch": 0.0533436612445133,
      "grad_norm": 5.335649013519287,
      "learning_rate": 9.952816526861598e-06,
      "loss": 0.0814,
      "step": 1033
    },
    {
      "epoch": 0.05339530080041312,
      "grad_norm": 6.851417541503906,
      "learning_rate": 9.952704162092595e-06,
      "loss": 0.1018,
      "step": 1034
    },
    {
      "epoch": 0.053446940356312936,
      "grad_norm": 14.031368255615234,
      "learning_rate": 9.952591664323298e-06,
      "loss": 0.1826,
      "step": 1035
    },
    {
      "epoch": 0.053498579912212754,
      "grad_norm": 33.364967346191406,
      "learning_rate": 9.95247903355673e-06,
      "loss": 0.286,
      "step": 1036
    },
    {
      "epoch": 0.05355021946811257,
      "grad_norm": 9.411717414855957,
      "learning_rate": 9.952366269795915e-06,
      "loss": 0.1368,
      "step": 1037
    },
    {
      "epoch": 0.05360185902401239,
      "grad_norm": 10.625096321105957,
      "learning_rate": 9.95225337304388e-06,
      "loss": 0.0936,
      "step": 1038
    },
    {
      "epoch": 0.05365349857991221,
      "grad_norm": 6.415925979614258,
      "learning_rate": 9.952140343303658e-06,
      "loss": 0.1154,
      "step": 1039
    },
    {
      "epoch": 0.053705138135812035,
      "grad_norm": 7.103929042816162,
      "learning_rate": 9.952027180578283e-06,
      "loss": 0.0854,
      "step": 1040
    },
    {
      "epoch": 0.05375677769171185,
      "grad_norm": 3.703270435333252,
      "learning_rate": 9.951913884870796e-06,
      "loss": 0.0648,
      "step": 1041
    },
    {
      "epoch": 0.05380841724761167,
      "grad_norm": 2.466475486755371,
      "learning_rate": 9.951800456184239e-06,
      "loss": 0.0564,
      "step": 1042
    },
    {
      "epoch": 0.05386005680351149,
      "grad_norm": 5.798369407653809,
      "learning_rate": 9.951686894521658e-06,
      "loss": 0.0799,
      "step": 1043
    },
    {
      "epoch": 0.05391169635941131,
      "grad_norm": 6.195906162261963,
      "learning_rate": 9.951573199886099e-06,
      "loss": 0.0754,
      "step": 1044
    },
    {
      "epoch": 0.05396333591531113,
      "grad_norm": 8.84292984008789,
      "learning_rate": 9.95145937228062e-06,
      "loss": 0.0561,
      "step": 1045
    },
    {
      "epoch": 0.054014975471210945,
      "grad_norm": 8.380972862243652,
      "learning_rate": 9.951345411708276e-06,
      "loss": 0.1232,
      "step": 1046
    },
    {
      "epoch": 0.05406661502711076,
      "grad_norm": 4.491415500640869,
      "learning_rate": 9.951231318172125e-06,
      "loss": 0.055,
      "step": 1047
    },
    {
      "epoch": 0.05411825458301059,
      "grad_norm": 5.491847515106201,
      "learning_rate": 9.951117091675235e-06,
      "loss": 0.087,
      "step": 1048
    },
    {
      "epoch": 0.05416989413891041,
      "grad_norm": 13.743888854980469,
      "learning_rate": 9.951002732220671e-06,
      "loss": 0.0924,
      "step": 1049
    },
    {
      "epoch": 0.054221533694810226,
      "grad_norm": 4.002910614013672,
      "learning_rate": 9.950888239811504e-06,
      "loss": 0.0952,
      "step": 1050
    },
    {
      "epoch": 0.054273173250710044,
      "grad_norm": 6.186213493347168,
      "learning_rate": 9.950773614450811e-06,
      "loss": 0.0964,
      "step": 1051
    },
    {
      "epoch": 0.05432481280660986,
      "grad_norm": 8.949538230895996,
      "learning_rate": 9.950658856141666e-06,
      "loss": 0.1113,
      "step": 1052
    },
    {
      "epoch": 0.05437645236250968,
      "grad_norm": 5.706902503967285,
      "learning_rate": 9.950543964887153e-06,
      "loss": 0.1328,
      "step": 1053
    },
    {
      "epoch": 0.0544280919184095,
      "grad_norm": 8.50443172454834,
      "learning_rate": 9.950428940690358e-06,
      "loss": 0.1143,
      "step": 1054
    },
    {
      "epoch": 0.05447973147430932,
      "grad_norm": 15.361687660217285,
      "learning_rate": 9.95031378355437e-06,
      "loss": 0.1227,
      "step": 1055
    },
    {
      "epoch": 0.05453137103020914,
      "grad_norm": 8.670136451721191,
      "learning_rate": 9.950198493482279e-06,
      "loss": 0.0835,
      "step": 1056
    },
    {
      "epoch": 0.05458301058610896,
      "grad_norm": 13.002370834350586,
      "learning_rate": 9.950083070477183e-06,
      "loss": 0.1202,
      "step": 1057
    },
    {
      "epoch": 0.05463465014200878,
      "grad_norm": 3.1082375049591064,
      "learning_rate": 9.94996751454218e-06,
      "loss": 0.0482,
      "step": 1058
    },
    {
      "epoch": 0.0546862896979086,
      "grad_norm": 7.076565265655518,
      "learning_rate": 9.949851825680377e-06,
      "loss": 0.0706,
      "step": 1059
    },
    {
      "epoch": 0.054737929253808416,
      "grad_norm": 6.469417572021484,
      "learning_rate": 9.949736003894877e-06,
      "loss": 0.0833,
      "step": 1060
    },
    {
      "epoch": 0.054789568809708235,
      "grad_norm": 9.60594654083252,
      "learning_rate": 9.94962004918879e-06,
      "loss": 0.0813,
      "step": 1061
    },
    {
      "epoch": 0.05484120836560805,
      "grad_norm": 8.795599937438965,
      "learning_rate": 9.949503961565232e-06,
      "loss": 0.1592,
      "step": 1062
    },
    {
      "epoch": 0.05489284792150787,
      "grad_norm": 2.828373670578003,
      "learning_rate": 9.94938774102732e-06,
      "loss": 0.0883,
      "step": 1063
    },
    {
      "epoch": 0.0549444874774077,
      "grad_norm": 3.1902809143066406,
      "learning_rate": 9.949271387578174e-06,
      "loss": 0.0708,
      "step": 1064
    },
    {
      "epoch": 0.054996127033307515,
      "grad_norm": 11.754432678222656,
      "learning_rate": 9.949154901220918e-06,
      "loss": 0.0808,
      "step": 1065
    },
    {
      "epoch": 0.055047766589207334,
      "grad_norm": 5.906667709350586,
      "learning_rate": 9.949038281958684e-06,
      "loss": 0.076,
      "step": 1066
    },
    {
      "epoch": 0.05509940614510715,
      "grad_norm": 16.761486053466797,
      "learning_rate": 9.9489215297946e-06,
      "loss": 0.1422,
      "step": 1067
    },
    {
      "epoch": 0.05515104570100697,
      "grad_norm": 25.70556640625,
      "learning_rate": 9.9488046447318e-06,
      "loss": 0.1672,
      "step": 1068
    },
    {
      "epoch": 0.05520268525690679,
      "grad_norm": 5.716439723968506,
      "learning_rate": 9.948687626773428e-06,
      "loss": 0.0756,
      "step": 1069
    },
    {
      "epoch": 0.05525432481280661,
      "grad_norm": 7.395629405975342,
      "learning_rate": 9.948570475922622e-06,
      "loss": 0.1183,
      "step": 1070
    },
    {
      "epoch": 0.05530596436870643,
      "grad_norm": 13.710158348083496,
      "learning_rate": 9.94845319218253e-06,
      "loss": 0.1305,
      "step": 1071
    },
    {
      "epoch": 0.05535760392460625,
      "grad_norm": 10.23421859741211,
      "learning_rate": 9.9483357755563e-06,
      "loss": 0.1043,
      "step": 1072
    },
    {
      "epoch": 0.05540924348050607,
      "grad_norm": 30.44868278503418,
      "learning_rate": 9.948218226047088e-06,
      "loss": 0.1593,
      "step": 1073
    },
    {
      "epoch": 0.05546088303640589,
      "grad_norm": 10.64648723602295,
      "learning_rate": 9.948100543658047e-06,
      "loss": 0.1153,
      "step": 1074
    },
    {
      "epoch": 0.055512522592305706,
      "grad_norm": 17.30192756652832,
      "learning_rate": 9.94798272839234e-06,
      "loss": 0.1293,
      "step": 1075
    },
    {
      "epoch": 0.055564162148205525,
      "grad_norm": 9.17283821105957,
      "learning_rate": 9.947864780253129e-06,
      "loss": 0.0967,
      "step": 1076
    },
    {
      "epoch": 0.05561580170410534,
      "grad_norm": 8.563742637634277,
      "learning_rate": 9.947746699243583e-06,
      "loss": 0.0465,
      "step": 1077
    },
    {
      "epoch": 0.05566744126000516,
      "grad_norm": 12.164257049560547,
      "learning_rate": 9.947628485366873e-06,
      "loss": 0.0989,
      "step": 1078
    },
    {
      "epoch": 0.05571908081590499,
      "grad_norm": 12.901748657226562,
      "learning_rate": 9.947510138626171e-06,
      "loss": 0.1195,
      "step": 1079
    },
    {
      "epoch": 0.055770720371804805,
      "grad_norm": 3.795806407928467,
      "learning_rate": 9.947391659024658e-06,
      "loss": 0.0792,
      "step": 1080
    },
    {
      "epoch": 0.05582235992770462,
      "grad_norm": 9.9214448928833,
      "learning_rate": 9.947273046565514e-06,
      "loss": 0.1245,
      "step": 1081
    },
    {
      "epoch": 0.05587399948360444,
      "grad_norm": 16.153884887695312,
      "learning_rate": 9.947154301251924e-06,
      "loss": 0.1795,
      "step": 1082
    },
    {
      "epoch": 0.05592563903950426,
      "grad_norm": 4.903212070465088,
      "learning_rate": 9.94703542308708e-06,
      "loss": 0.0653,
      "step": 1083
    },
    {
      "epoch": 0.05597727859540408,
      "grad_norm": 9.060792922973633,
      "learning_rate": 9.94691641207417e-06,
      "loss": 0.1179,
      "step": 1084
    },
    {
      "epoch": 0.0560289181513039,
      "grad_norm": 6.150825500488281,
      "learning_rate": 9.946797268216392e-06,
      "loss": 0.0635,
      "step": 1085
    },
    {
      "epoch": 0.056080557707203715,
      "grad_norm": 15.91894245147705,
      "learning_rate": 9.946677991516947e-06,
      "loss": 0.0916,
      "step": 1086
    },
    {
      "epoch": 0.05613219726310354,
      "grad_norm": 7.481376647949219,
      "learning_rate": 9.946558581979034e-06,
      "loss": 0.0877,
      "step": 1087
    },
    {
      "epoch": 0.05618383681900336,
      "grad_norm": 5.0303850173950195,
      "learning_rate": 9.946439039605862e-06,
      "loss": 0.0878,
      "step": 1088
    },
    {
      "epoch": 0.05623547637490318,
      "grad_norm": 3.572890043258667,
      "learning_rate": 9.946319364400644e-06,
      "loss": 0.0312,
      "step": 1089
    },
    {
      "epoch": 0.056287115930802996,
      "grad_norm": 3.4450387954711914,
      "learning_rate": 9.946199556366587e-06,
      "loss": 0.0574,
      "step": 1090
    },
    {
      "epoch": 0.056338755486702814,
      "grad_norm": 5.432600021362305,
      "learning_rate": 9.946079615506915e-06,
      "loss": 0.0575,
      "step": 1091
    },
    {
      "epoch": 0.05639039504260263,
      "grad_norm": 7.854936599731445,
      "learning_rate": 9.945959541824846e-06,
      "loss": 0.0828,
      "step": 1092
    },
    {
      "epoch": 0.05644203459850245,
      "grad_norm": 17.4731388092041,
      "learning_rate": 9.945839335323604e-06,
      "loss": 0.1735,
      "step": 1093
    },
    {
      "epoch": 0.05649367415440227,
      "grad_norm": 10.117325782775879,
      "learning_rate": 9.945718996006418e-06,
      "loss": 0.0789,
      "step": 1094
    },
    {
      "epoch": 0.056545313710302095,
      "grad_norm": 17.31111717224121,
      "learning_rate": 9.94559852387652e-06,
      "loss": 0.1016,
      "step": 1095
    },
    {
      "epoch": 0.05659695326620191,
      "grad_norm": 14.24697208404541,
      "learning_rate": 9.945477918937142e-06,
      "loss": 0.0876,
      "step": 1096
    },
    {
      "epoch": 0.05664859282210173,
      "grad_norm": 5.821711540222168,
      "learning_rate": 9.945357181191529e-06,
      "loss": 0.1154,
      "step": 1097
    },
    {
      "epoch": 0.05670023237800155,
      "grad_norm": 4.818850994110107,
      "learning_rate": 9.945236310642917e-06,
      "loss": 0.0975,
      "step": 1098
    },
    {
      "epoch": 0.05675187193390137,
      "grad_norm": 5.4996514320373535,
      "learning_rate": 9.945115307294554e-06,
      "loss": 0.1343,
      "step": 1099
    },
    {
      "epoch": 0.05680351148980119,
      "grad_norm": 10.644469261169434,
      "learning_rate": 9.94499417114969e-06,
      "loss": 0.136,
      "step": 1100
    },
    {
      "epoch": 0.056855151045701005,
      "grad_norm": 21.026870727539062,
      "learning_rate": 9.944872902211579e-06,
      "loss": 0.1654,
      "step": 1101
    },
    {
      "epoch": 0.056906790601600823,
      "grad_norm": 13.969194412231445,
      "learning_rate": 9.944751500483479e-06,
      "loss": 0.1395,
      "step": 1102
    },
    {
      "epoch": 0.05695843015750065,
      "grad_norm": 13.297613143920898,
      "learning_rate": 9.944629965968642e-06,
      "loss": 0.1476,
      "step": 1103
    },
    {
      "epoch": 0.05701006971340047,
      "grad_norm": 9.346051216125488,
      "learning_rate": 9.94450829867034e-06,
      "loss": 0.0896,
      "step": 1104
    },
    {
      "epoch": 0.057061709269300286,
      "grad_norm": 5.614561080932617,
      "learning_rate": 9.944386498591839e-06,
      "loss": 0.084,
      "step": 1105
    },
    {
      "epoch": 0.057113348825200104,
      "grad_norm": 17.393224716186523,
      "learning_rate": 9.944264565736406e-06,
      "loss": 0.122,
      "step": 1106
    },
    {
      "epoch": 0.05716498838109992,
      "grad_norm": 8.24782657623291,
      "learning_rate": 9.944142500107317e-06,
      "loss": 0.0974,
      "step": 1107
    },
    {
      "epoch": 0.05721662793699974,
      "grad_norm": 5.9843831062316895,
      "learning_rate": 9.944020301707853e-06,
      "loss": 0.0657,
      "step": 1108
    },
    {
      "epoch": 0.05726826749289956,
      "grad_norm": 5.49865198135376,
      "learning_rate": 9.943897970541291e-06,
      "loss": 0.0909,
      "step": 1109
    },
    {
      "epoch": 0.05731990704879938,
      "grad_norm": 15.59573745727539,
      "learning_rate": 9.94377550661092e-06,
      "loss": 0.1864,
      "step": 1110
    },
    {
      "epoch": 0.0573715466046992,
      "grad_norm": 14.423590660095215,
      "learning_rate": 9.943652909920027e-06,
      "loss": 0.1784,
      "step": 1111
    },
    {
      "epoch": 0.05742318616059902,
      "grad_norm": 9.068049430847168,
      "learning_rate": 9.943530180471904e-06,
      "loss": 0.1378,
      "step": 1112
    },
    {
      "epoch": 0.05747482571649884,
      "grad_norm": 5.396983623504639,
      "learning_rate": 9.943407318269846e-06,
      "loss": 0.054,
      "step": 1113
    },
    {
      "epoch": 0.05752646527239866,
      "grad_norm": 10.803695678710938,
      "learning_rate": 9.943284323317153e-06,
      "loss": 0.1221,
      "step": 1114
    },
    {
      "epoch": 0.057578104828298476,
      "grad_norm": 8.469614028930664,
      "learning_rate": 9.94316119561713e-06,
      "loss": 0.0891,
      "step": 1115
    },
    {
      "epoch": 0.057629744384198295,
      "grad_norm": 14.784948348999023,
      "learning_rate": 9.94303793517308e-06,
      "loss": 0.104,
      "step": 1116
    },
    {
      "epoch": 0.05768138394009811,
      "grad_norm": 3.317925453186035,
      "learning_rate": 9.942914541988315e-06,
      "loss": 0.0434,
      "step": 1117
    },
    {
      "epoch": 0.05773302349599793,
      "grad_norm": 7.20095157623291,
      "learning_rate": 9.94279101606615e-06,
      "loss": 0.0735,
      "step": 1118
    },
    {
      "epoch": 0.05778466305189776,
      "grad_norm": 17.488773345947266,
      "learning_rate": 9.9426673574099e-06,
      "loss": 0.2653,
      "step": 1119
    },
    {
      "epoch": 0.057836302607797575,
      "grad_norm": 25.986351013183594,
      "learning_rate": 9.942543566022883e-06,
      "loss": 0.2054,
      "step": 1120
    },
    {
      "epoch": 0.057887942163697394,
      "grad_norm": 4.9818830490112305,
      "learning_rate": 9.942419641908429e-06,
      "loss": 0.0843,
      "step": 1121
    },
    {
      "epoch": 0.05793958171959721,
      "grad_norm": 7.170565128326416,
      "learning_rate": 9.942295585069862e-06,
      "loss": 0.0566,
      "step": 1122
    },
    {
      "epoch": 0.05799122127549703,
      "grad_norm": 6.688936233520508,
      "learning_rate": 9.942171395510519e-06,
      "loss": 0.1005,
      "step": 1123
    },
    {
      "epoch": 0.05804286083139685,
      "grad_norm": 2.633913993835449,
      "learning_rate": 9.942047073233728e-06,
      "loss": 0.0291,
      "step": 1124
    },
    {
      "epoch": 0.05809450038729667,
      "grad_norm": 7.889233112335205,
      "learning_rate": 9.94192261824283e-06,
      "loss": 0.0779,
      "step": 1125
    },
    {
      "epoch": 0.058146139943196486,
      "grad_norm": 4.916865348815918,
      "learning_rate": 9.941798030541167e-06,
      "loss": 0.0633,
      "step": 1126
    },
    {
      "epoch": 0.05819777949909631,
      "grad_norm": 8.606673240661621,
      "learning_rate": 9.941673310132087e-06,
      "loss": 0.0901,
      "step": 1127
    },
    {
      "epoch": 0.05824941905499613,
      "grad_norm": 8.466205596923828,
      "learning_rate": 9.941548457018937e-06,
      "loss": 0.1041,
      "step": 1128
    },
    {
      "epoch": 0.05830105861089595,
      "grad_norm": 12.091532707214355,
      "learning_rate": 9.94142347120507e-06,
      "loss": 0.1018,
      "step": 1129
    },
    {
      "epoch": 0.058352698166795766,
      "grad_norm": 7.740886688232422,
      "learning_rate": 9.941298352693842e-06,
      "loss": 0.0763,
      "step": 1130
    },
    {
      "epoch": 0.058404337722695585,
      "grad_norm": 16.787857055664062,
      "learning_rate": 9.941173101488615e-06,
      "loss": 0.1174,
      "step": 1131
    },
    {
      "epoch": 0.0584559772785954,
      "grad_norm": 14.423918724060059,
      "learning_rate": 9.94104771759275e-06,
      "loss": 0.218,
      "step": 1132
    },
    {
      "epoch": 0.05850761683449522,
      "grad_norm": 8.326886177062988,
      "learning_rate": 9.940922201009618e-06,
      "loss": 0.0794,
      "step": 1133
    },
    {
      "epoch": 0.05855925639039504,
      "grad_norm": 15.0519380569458,
      "learning_rate": 9.940796551742583e-06,
      "loss": 0.1447,
      "step": 1134
    },
    {
      "epoch": 0.058610895946294865,
      "grad_norm": 6.625211238861084,
      "learning_rate": 9.940670769795024e-06,
      "loss": 0.0409,
      "step": 1135
    },
    {
      "epoch": 0.05866253550219468,
      "grad_norm": 3.1709840297698975,
      "learning_rate": 9.94054485517032e-06,
      "loss": 0.0494,
      "step": 1136
    },
    {
      "epoch": 0.0587141750580945,
      "grad_norm": 12.942554473876953,
      "learning_rate": 9.940418807871847e-06,
      "loss": 0.0864,
      "step": 1137
    },
    {
      "epoch": 0.05876581461399432,
      "grad_norm": 5.26187801361084,
      "learning_rate": 9.940292627902995e-06,
      "loss": 0.0937,
      "step": 1138
    },
    {
      "epoch": 0.05881745416989414,
      "grad_norm": 4.536093711853027,
      "learning_rate": 9.94016631526715e-06,
      "loss": 0.0725,
      "step": 1139
    },
    {
      "epoch": 0.05886909372579396,
      "grad_norm": 6.131861686706543,
      "learning_rate": 9.940039869967704e-06,
      "loss": 0.0834,
      "step": 1140
    },
    {
      "epoch": 0.058920733281693775,
      "grad_norm": 3.671354055404663,
      "learning_rate": 9.939913292008053e-06,
      "loss": 0.0604,
      "step": 1141
    },
    {
      "epoch": 0.058972372837593594,
      "grad_norm": 2.8117148876190186,
      "learning_rate": 9.939786581391596e-06,
      "loss": 0.0365,
      "step": 1142
    },
    {
      "epoch": 0.05902401239349342,
      "grad_norm": 4.821849822998047,
      "learning_rate": 9.939659738121736e-06,
      "loss": 0.0703,
      "step": 1143
    },
    {
      "epoch": 0.05907565194939324,
      "grad_norm": 21.95835304260254,
      "learning_rate": 9.93953276220188e-06,
      "loss": 0.1832,
      "step": 1144
    },
    {
      "epoch": 0.059127291505293056,
      "grad_norm": 11.441669464111328,
      "learning_rate": 9.939405653635436e-06,
      "loss": 0.1206,
      "step": 1145
    },
    {
      "epoch": 0.059178931061192874,
      "grad_norm": 12.222320556640625,
      "learning_rate": 9.939278412425819e-06,
      "loss": 0.1029,
      "step": 1146
    },
    {
      "epoch": 0.05923057061709269,
      "grad_norm": 13.483321189880371,
      "learning_rate": 9.939151038576444e-06,
      "loss": 0.1068,
      "step": 1147
    },
    {
      "epoch": 0.05928221017299251,
      "grad_norm": 11.69172477722168,
      "learning_rate": 9.939023532090733e-06,
      "loss": 0.0672,
      "step": 1148
    },
    {
      "epoch": 0.05933384972889233,
      "grad_norm": 9.6361665725708,
      "learning_rate": 9.938895892972112e-06,
      "loss": 0.152,
      "step": 1149
    },
    {
      "epoch": 0.05938548928479215,
      "grad_norm": 6.696620464324951,
      "learning_rate": 9.938768121224002e-06,
      "loss": 0.0444,
      "step": 1150
    },
    {
      "epoch": 0.05943712884069197,
      "grad_norm": 10.39874267578125,
      "learning_rate": 9.938640216849843e-06,
      "loss": 0.0803,
      "step": 1151
    },
    {
      "epoch": 0.05948876839659179,
      "grad_norm": 6.580800533294678,
      "learning_rate": 9.938512179853063e-06,
      "loss": 0.087,
      "step": 1152
    },
    {
      "epoch": 0.05954040795249161,
      "grad_norm": 6.945914268493652,
      "learning_rate": 9.938384010237104e-06,
      "loss": 0.0849,
      "step": 1153
    },
    {
      "epoch": 0.05959204750839143,
      "grad_norm": 6.734384059906006,
      "learning_rate": 9.938255708005406e-06,
      "loss": 0.0978,
      "step": 1154
    },
    {
      "epoch": 0.05964368706429125,
      "grad_norm": 2.750528573989868,
      "learning_rate": 9.938127273161415e-06,
      "loss": 0.0495,
      "step": 1155
    },
    {
      "epoch": 0.059695326620191065,
      "grad_norm": 8.581611633300781,
      "learning_rate": 9.93799870570858e-06,
      "loss": 0.1217,
      "step": 1156
    },
    {
      "epoch": 0.059746966176090884,
      "grad_norm": 9.268179893493652,
      "learning_rate": 9.937870005650353e-06,
      "loss": 0.1146,
      "step": 1157
    },
    {
      "epoch": 0.0597986057319907,
      "grad_norm": 7.547562122344971,
      "learning_rate": 9.937741172990191e-06,
      "loss": 0.0959,
      "step": 1158
    },
    {
      "epoch": 0.05985024528789053,
      "grad_norm": 5.667535781860352,
      "learning_rate": 9.937612207731555e-06,
      "loss": 0.0413,
      "step": 1159
    },
    {
      "epoch": 0.059901884843790346,
      "grad_norm": 5.789296627044678,
      "learning_rate": 9.937483109877906e-06,
      "loss": 0.1118,
      "step": 1160
    },
    {
      "epoch": 0.059953524399690164,
      "grad_norm": 10.504975318908691,
      "learning_rate": 9.93735387943271e-06,
      "loss": 0.0596,
      "step": 1161
    },
    {
      "epoch": 0.06000516395558998,
      "grad_norm": 7.312224864959717,
      "learning_rate": 9.93722451639944e-06,
      "loss": 0.0926,
      "step": 1162
    },
    {
      "epoch": 0.0600568035114898,
      "grad_norm": 15.602783203125,
      "learning_rate": 9.937095020781569e-06,
      "loss": 0.2134,
      "step": 1163
    },
    {
      "epoch": 0.06010844306738962,
      "grad_norm": 12.315022468566895,
      "learning_rate": 9.936965392582574e-06,
      "loss": 0.1659,
      "step": 1164
    },
    {
      "epoch": 0.06016008262328944,
      "grad_norm": 2.6122653484344482,
      "learning_rate": 9.936835631805935e-06,
      "loss": 0.0562,
      "step": 1165
    },
    {
      "epoch": 0.060211722179189256,
      "grad_norm": 10.390530586242676,
      "learning_rate": 9.936705738455139e-06,
      "loss": 0.0913,
      "step": 1166
    },
    {
      "epoch": 0.06026336173508908,
      "grad_norm": 6.191280364990234,
      "learning_rate": 9.936575712533672e-06,
      "loss": 0.0815,
      "step": 1167
    },
    {
      "epoch": 0.0603150012909889,
      "grad_norm": 8.103361129760742,
      "learning_rate": 9.936445554045028e-06,
      "loss": 0.1072,
      "step": 1168
    },
    {
      "epoch": 0.06036664084688872,
      "grad_norm": 7.3403520584106445,
      "learning_rate": 9.9363152629927e-06,
      "loss": 0.0873,
      "step": 1169
    },
    {
      "epoch": 0.060418280402788536,
      "grad_norm": 5.954051971435547,
      "learning_rate": 9.936184839380189e-06,
      "loss": 0.0665,
      "step": 1170
    },
    {
      "epoch": 0.060469919958688355,
      "grad_norm": 4.886270046234131,
      "learning_rate": 9.936054283210996e-06,
      "loss": 0.0934,
      "step": 1171
    },
    {
      "epoch": 0.06052155951458817,
      "grad_norm": 3.872629404067993,
      "learning_rate": 9.935923594488626e-06,
      "loss": 0.0893,
      "step": 1172
    },
    {
      "epoch": 0.06057319907048799,
      "grad_norm": 9.638045310974121,
      "learning_rate": 9.935792773216591e-06,
      "loss": 0.1271,
      "step": 1173
    },
    {
      "epoch": 0.06062483862638781,
      "grad_norm": 4.630978107452393,
      "learning_rate": 9.935661819398403e-06,
      "loss": 0.0891,
      "step": 1174
    },
    {
      "epoch": 0.060676478182287635,
      "grad_norm": 5.752591133117676,
      "learning_rate": 9.935530733037576e-06,
      "loss": 0.12,
      "step": 1175
    },
    {
      "epoch": 0.060728117738187454,
      "grad_norm": 14.640379905700684,
      "learning_rate": 9.935399514137636e-06,
      "loss": 0.1896,
      "step": 1176
    },
    {
      "epoch": 0.06077975729408727,
      "grad_norm": 11.278938293457031,
      "learning_rate": 9.935268162702104e-06,
      "loss": 0.1035,
      "step": 1177
    },
    {
      "epoch": 0.06083139684998709,
      "grad_norm": 4.458139896392822,
      "learning_rate": 9.935136678734503e-06,
      "loss": 0.1167,
      "step": 1178
    },
    {
      "epoch": 0.06088303640588691,
      "grad_norm": 6.115289688110352,
      "learning_rate": 9.93500506223837e-06,
      "loss": 0.1138,
      "step": 1179
    },
    {
      "epoch": 0.06093467596178673,
      "grad_norm": 8.208502769470215,
      "learning_rate": 9.934873313217237e-06,
      "loss": 0.109,
      "step": 1180
    },
    {
      "epoch": 0.060986315517686546,
      "grad_norm": 3.5345458984375,
      "learning_rate": 9.934741431674641e-06,
      "loss": 0.0659,
      "step": 1181
    },
    {
      "epoch": 0.061037955073586364,
      "grad_norm": 4.205630302429199,
      "learning_rate": 9.934609417614125e-06,
      "loss": 0.1159,
      "step": 1182
    },
    {
      "epoch": 0.06108959462948619,
      "grad_norm": 2.5363502502441406,
      "learning_rate": 9.934477271039235e-06,
      "loss": 0.0525,
      "step": 1183
    },
    {
      "epoch": 0.06114123418538601,
      "grad_norm": 4.502866744995117,
      "learning_rate": 9.934344991953518e-06,
      "loss": 0.0676,
      "step": 1184
    },
    {
      "epoch": 0.061192873741285826,
      "grad_norm": 3.367593288421631,
      "learning_rate": 9.934212580360528e-06,
      "loss": 0.0655,
      "step": 1185
    },
    {
      "epoch": 0.061244513297185645,
      "grad_norm": 10.19282341003418,
      "learning_rate": 9.934080036263817e-06,
      "loss": 0.0955,
      "step": 1186
    },
    {
      "epoch": 0.06129615285308546,
      "grad_norm": 7.235620975494385,
      "learning_rate": 9.933947359666947e-06,
      "loss": 0.0588,
      "step": 1187
    },
    {
      "epoch": 0.06134779240898528,
      "grad_norm": 3.440892457962036,
      "learning_rate": 9.933814550573482e-06,
      "loss": 0.0729,
      "step": 1188
    },
    {
      "epoch": 0.0613994319648851,
      "grad_norm": 7.231861114501953,
      "learning_rate": 9.933681608986986e-06,
      "loss": 0.0852,
      "step": 1189
    },
    {
      "epoch": 0.06145107152078492,
      "grad_norm": 8.988840103149414,
      "learning_rate": 9.933548534911032e-06,
      "loss": 0.1514,
      "step": 1190
    },
    {
      "epoch": 0.06150271107668474,
      "grad_norm": 5.542039394378662,
      "learning_rate": 9.933415328349191e-06,
      "loss": 0.114,
      "step": 1191
    },
    {
      "epoch": 0.06155435063258456,
      "grad_norm": 14.903416633605957,
      "learning_rate": 9.93328198930504e-06,
      "loss": 0.0791,
      "step": 1192
    },
    {
      "epoch": 0.06160599018848438,
      "grad_norm": 24.262592315673828,
      "learning_rate": 9.933148517782162e-06,
      "loss": 0.2083,
      "step": 1193
    },
    {
      "epoch": 0.0616576297443842,
      "grad_norm": 6.027246952056885,
      "learning_rate": 9.933014913784139e-06,
      "loss": 0.0924,
      "step": 1194
    },
    {
      "epoch": 0.06170926930028402,
      "grad_norm": 13.331570625305176,
      "learning_rate": 9.93288117731456e-06,
      "loss": 0.1125,
      "step": 1195
    },
    {
      "epoch": 0.061760908856183835,
      "grad_norm": 5.753688335418701,
      "learning_rate": 9.932747308377016e-06,
      "loss": 0.0769,
      "step": 1196
    },
    {
      "epoch": 0.061812548412083654,
      "grad_norm": 9.084490776062012,
      "learning_rate": 9.932613306975104e-06,
      "loss": 0.099,
      "step": 1197
    },
    {
      "epoch": 0.06186418796798347,
      "grad_norm": 3.1822590827941895,
      "learning_rate": 9.932479173112417e-06,
      "loss": 0.0339,
      "step": 1198
    },
    {
      "epoch": 0.0619158275238833,
      "grad_norm": 12.099540710449219,
      "learning_rate": 9.932344906792562e-06,
      "loss": 0.1188,
      "step": 1199
    },
    {
      "epoch": 0.061967467079783116,
      "grad_norm": 20.985008239746094,
      "learning_rate": 9.932210508019142e-06,
      "loss": 0.2271,
      "step": 1200
    },
    {
      "epoch": 0.062019106635682934,
      "grad_norm": 25.958675384521484,
      "learning_rate": 9.93207597679577e-06,
      "loss": 0.1206,
      "step": 1201
    },
    {
      "epoch": 0.06207074619158275,
      "grad_norm": 10.432710647583008,
      "learning_rate": 9.931941313126054e-06,
      "loss": 0.1423,
      "step": 1202
    },
    {
      "epoch": 0.06212238574748257,
      "grad_norm": 4.738035202026367,
      "learning_rate": 9.931806517013612e-06,
      "loss": 0.1405,
      "step": 1203
    },
    {
      "epoch": 0.06217402530338239,
      "grad_norm": 9.080729484558105,
      "learning_rate": 9.931671588462067e-06,
      "loss": 0.1084,
      "step": 1204
    },
    {
      "epoch": 0.06222566485928221,
      "grad_norm": 3.2312731742858887,
      "learning_rate": 9.931536527475036e-06,
      "loss": 0.071,
      "step": 1205
    },
    {
      "epoch": 0.062277304415182026,
      "grad_norm": 9.191733360290527,
      "learning_rate": 9.93140133405615e-06,
      "loss": 0.1053,
      "step": 1206
    },
    {
      "epoch": 0.06232894397108185,
      "grad_norm": 5.461610794067383,
      "learning_rate": 9.93126600820904e-06,
      "loss": 0.1036,
      "step": 1207
    },
    {
      "epoch": 0.06238058352698167,
      "grad_norm": 19.05320167541504,
      "learning_rate": 9.931130549937338e-06,
      "loss": 0.1777,
      "step": 1208
    },
    {
      "epoch": 0.06243222308288149,
      "grad_norm": 14.095467567443848,
      "learning_rate": 9.930994959244684e-06,
      "loss": 0.0931,
      "step": 1209
    },
    {
      "epoch": 0.06248386263878131,
      "grad_norm": 11.395224571228027,
      "learning_rate": 9.930859236134716e-06,
      "loss": 0.0792,
      "step": 1210
    },
    {
      "epoch": 0.06253550219468113,
      "grad_norm": 9.051433563232422,
      "learning_rate": 9.930723380611083e-06,
      "loss": 0.1047,
      "step": 1211
    },
    {
      "epoch": 0.06258714175058094,
      "grad_norm": 6.719902992248535,
      "learning_rate": 9.930587392677428e-06,
      "loss": 0.0437,
      "step": 1212
    },
    {
      "epoch": 0.06263878130648076,
      "grad_norm": 10.475923538208008,
      "learning_rate": 9.930451272337407e-06,
      "loss": 0.133,
      "step": 1213
    },
    {
      "epoch": 0.06269042086238058,
      "grad_norm": 4.261584281921387,
      "learning_rate": 9.930315019594672e-06,
      "loss": 0.0989,
      "step": 1214
    },
    {
      "epoch": 0.0627420604182804,
      "grad_norm": 8.923521041870117,
      "learning_rate": 9.930178634452887e-06,
      "loss": 0.074,
      "step": 1215
    },
    {
      "epoch": 0.06279369997418022,
      "grad_norm": 5.945504188537598,
      "learning_rate": 9.930042116915708e-06,
      "loss": 0.0784,
      "step": 1216
    },
    {
      "epoch": 0.06284533953008004,
      "grad_norm": 12.92665958404541,
      "learning_rate": 9.929905466986807e-06,
      "loss": 0.1696,
      "step": 1217
    },
    {
      "epoch": 0.06289697908597985,
      "grad_norm": 6.72512674331665,
      "learning_rate": 9.929768684669849e-06,
      "loss": 0.1063,
      "step": 1218
    },
    {
      "epoch": 0.06294861864187969,
      "grad_norm": 5.29183292388916,
      "learning_rate": 9.929631769968511e-06,
      "loss": 0.1045,
      "step": 1219
    },
    {
      "epoch": 0.0630002581977795,
      "grad_norm": 7.281365394592285,
      "learning_rate": 9.929494722886468e-06,
      "loss": 0.0746,
      "step": 1220
    },
    {
      "epoch": 0.06305189775367932,
      "grad_norm": 10.697446823120117,
      "learning_rate": 9.929357543427398e-06,
      "loss": 0.1021,
      "step": 1221
    },
    {
      "epoch": 0.06310353730957914,
      "grad_norm": 11.303498268127441,
      "learning_rate": 9.929220231594989e-06,
      "loss": 0.0992,
      "step": 1222
    },
    {
      "epoch": 0.06315517686547896,
      "grad_norm": 3.6598739624023438,
      "learning_rate": 9.929082787392926e-06,
      "loss": 0.076,
      "step": 1223
    },
    {
      "epoch": 0.06320681642137878,
      "grad_norm": 5.380300045013428,
      "learning_rate": 9.9289452108249e-06,
      "loss": 0.0806,
      "step": 1224
    },
    {
      "epoch": 0.0632584559772786,
      "grad_norm": 6.715560436248779,
      "learning_rate": 9.928807501894605e-06,
      "loss": 0.1175,
      "step": 1225
    },
    {
      "epoch": 0.06331009553317841,
      "grad_norm": 7.371181011199951,
      "learning_rate": 9.928669660605743e-06,
      "loss": 0.1264,
      "step": 1226
    },
    {
      "epoch": 0.06336173508907823,
      "grad_norm": 6.328311443328857,
      "learning_rate": 9.928531686962009e-06,
      "loss": 0.0814,
      "step": 1227
    },
    {
      "epoch": 0.06341337464497805,
      "grad_norm": 20.690204620361328,
      "learning_rate": 9.928393580967112e-06,
      "loss": 0.1363,
      "step": 1228
    },
    {
      "epoch": 0.06346501420087787,
      "grad_norm": 7.64021635055542,
      "learning_rate": 9.928255342624762e-06,
      "loss": 0.1269,
      "step": 1229
    },
    {
      "epoch": 0.06351665375677769,
      "grad_norm": 8.74015998840332,
      "learning_rate": 9.92811697193867e-06,
      "loss": 0.1124,
      "step": 1230
    },
    {
      "epoch": 0.0635682933126775,
      "grad_norm": 10.296818733215332,
      "learning_rate": 9.927978468912549e-06,
      "loss": 0.0999,
      "step": 1231
    },
    {
      "epoch": 0.06361993286857733,
      "grad_norm": 5.03084135055542,
      "learning_rate": 9.927839833550124e-06,
      "loss": 0.0813,
      "step": 1232
    },
    {
      "epoch": 0.06367157242447714,
      "grad_norm": 5.568746566772461,
      "learning_rate": 9.927701065855112e-06,
      "loss": 0.1013,
      "step": 1233
    },
    {
      "epoch": 0.06372321198037696,
      "grad_norm": 5.598298072814941,
      "learning_rate": 9.927562165831242e-06,
      "loss": 0.1015,
      "step": 1234
    },
    {
      "epoch": 0.0637748515362768,
      "grad_norm": 3.417015314102173,
      "learning_rate": 9.927423133482245e-06,
      "loss": 0.0536,
      "step": 1235
    },
    {
      "epoch": 0.06382649109217661,
      "grad_norm": 7.917229175567627,
      "learning_rate": 9.927283968811856e-06,
      "loss": 0.0693,
      "step": 1236
    },
    {
      "epoch": 0.06387813064807643,
      "grad_norm": 7.341648578643799,
      "learning_rate": 9.927144671823807e-06,
      "loss": 0.0707,
      "step": 1237
    },
    {
      "epoch": 0.06392977020397625,
      "grad_norm": 18.99108123779297,
      "learning_rate": 9.927005242521842e-06,
      "loss": 0.0998,
      "step": 1238
    },
    {
      "epoch": 0.06398140975987607,
      "grad_norm": 11.819704055786133,
      "learning_rate": 9.926865680909706e-06,
      "loss": 0.2069,
      "step": 1239
    },
    {
      "epoch": 0.06403304931577589,
      "grad_norm": 3.923874855041504,
      "learning_rate": 9.926725986991145e-06,
      "loss": 0.1061,
      "step": 1240
    },
    {
      "epoch": 0.0640846888716757,
      "grad_norm": 20.79497718811035,
      "learning_rate": 9.92658616076991e-06,
      "loss": 0.1169,
      "step": 1241
    },
    {
      "epoch": 0.06413632842757552,
      "grad_norm": 4.6531195640563965,
      "learning_rate": 9.92644620224976e-06,
      "loss": 0.0728,
      "step": 1242
    },
    {
      "epoch": 0.06418796798347534,
      "grad_norm": 7.07065486907959,
      "learning_rate": 9.926306111434447e-06,
      "loss": 0.1238,
      "step": 1243
    },
    {
      "epoch": 0.06423960753937516,
      "grad_norm": 7.962484836578369,
      "learning_rate": 9.926165888327737e-06,
      "loss": 0.1168,
      "step": 1244
    },
    {
      "epoch": 0.06429124709527498,
      "grad_norm": 5.358644008636475,
      "learning_rate": 9.926025532933396e-06,
      "loss": 0.0737,
      "step": 1245
    },
    {
      "epoch": 0.0643428866511748,
      "grad_norm": 8.054380416870117,
      "learning_rate": 9.925885045255193e-06,
      "loss": 0.12,
      "step": 1246
    },
    {
      "epoch": 0.06439452620707461,
      "grad_norm": 14.977758407592773,
      "learning_rate": 9.925744425296899e-06,
      "loss": 0.1214,
      "step": 1247
    },
    {
      "epoch": 0.06444616576297443,
      "grad_norm": 3.195209264755249,
      "learning_rate": 9.92560367306229e-06,
      "loss": 0.0563,
      "step": 1248
    },
    {
      "epoch": 0.06449780531887425,
      "grad_norm": 14.554932594299316,
      "learning_rate": 9.925462788555147e-06,
      "loss": 0.1054,
      "step": 1249
    },
    {
      "epoch": 0.06454944487477407,
      "grad_norm": 2.438035726547241,
      "learning_rate": 9.925321771779252e-06,
      "loss": 0.0702,
      "step": 1250
    },
    {
      "epoch": 0.0646010844306739,
      "grad_norm": 4.755638599395752,
      "learning_rate": 9.925180622738396e-06,
      "loss": 0.0879,
      "step": 1251
    },
    {
      "epoch": 0.06465272398657372,
      "grad_norm": 5.695033073425293,
      "learning_rate": 9.925039341436364e-06,
      "loss": 0.0622,
      "step": 1252
    },
    {
      "epoch": 0.06470436354247354,
      "grad_norm": 9.509181022644043,
      "learning_rate": 9.924897927876954e-06,
      "loss": 0.1261,
      "step": 1253
    },
    {
      "epoch": 0.06475600309837336,
      "grad_norm": 10.48636245727539,
      "learning_rate": 9.92475638206396e-06,
      "loss": 0.0995,
      "step": 1254
    },
    {
      "epoch": 0.06480764265427318,
      "grad_norm": 7.4233622550964355,
      "learning_rate": 9.924614704001185e-06,
      "loss": 0.1058,
      "step": 1255
    },
    {
      "epoch": 0.064859282210173,
      "grad_norm": 4.27882719039917,
      "learning_rate": 9.924472893692434e-06,
      "loss": 0.0928,
      "step": 1256
    },
    {
      "epoch": 0.06491092176607281,
      "grad_norm": 12.355961799621582,
      "learning_rate": 9.924330951141516e-06,
      "loss": 0.0571,
      "step": 1257
    },
    {
      "epoch": 0.06496256132197263,
      "grad_norm": 2.334402322769165,
      "learning_rate": 9.924188876352242e-06,
      "loss": 0.044,
      "step": 1258
    },
    {
      "epoch": 0.06501420087787245,
      "grad_norm": 7.515434265136719,
      "learning_rate": 9.924046669328424e-06,
      "loss": 0.1038,
      "step": 1259
    },
    {
      "epoch": 0.06506584043377227,
      "grad_norm": 4.052685737609863,
      "learning_rate": 9.923904330073885e-06,
      "loss": 0.0661,
      "step": 1260
    },
    {
      "epoch": 0.06511747998967209,
      "grad_norm": 8.571357727050781,
      "learning_rate": 9.923761858592446e-06,
      "loss": 0.1322,
      "step": 1261
    },
    {
      "epoch": 0.0651691195455719,
      "grad_norm": 8.839433670043945,
      "learning_rate": 9.923619254887933e-06,
      "loss": 0.1346,
      "step": 1262
    },
    {
      "epoch": 0.06522075910147172,
      "grad_norm": 7.773127555847168,
      "learning_rate": 9.923476518964175e-06,
      "loss": 0.1273,
      "step": 1263
    },
    {
      "epoch": 0.06527239865737154,
      "grad_norm": 10.597262382507324,
      "learning_rate": 9.923333650825006e-06,
      "loss": 0.1653,
      "step": 1264
    },
    {
      "epoch": 0.06532403821327136,
      "grad_norm": 6.57686710357666,
      "learning_rate": 9.923190650474262e-06,
      "loss": 0.1035,
      "step": 1265
    },
    {
      "epoch": 0.06537567776917118,
      "grad_norm": 12.344437599182129,
      "learning_rate": 9.923047517915783e-06,
      "loss": 0.1102,
      "step": 1266
    },
    {
      "epoch": 0.06542731732507101,
      "grad_norm": 8.137676239013672,
      "learning_rate": 9.922904253153413e-06,
      "loss": 0.0878,
      "step": 1267
    },
    {
      "epoch": 0.06547895688097083,
      "grad_norm": 7.956837177276611,
      "learning_rate": 9.922760856191e-06,
      "loss": 0.0694,
      "step": 1268
    },
    {
      "epoch": 0.06553059643687065,
      "grad_norm": 13.860042572021484,
      "learning_rate": 9.922617327032393e-06,
      "loss": 0.1879,
      "step": 1269
    },
    {
      "epoch": 0.06558223599277047,
      "grad_norm": 6.2405571937561035,
      "learning_rate": 9.922473665681446e-06,
      "loss": 0.13,
      "step": 1270
    },
    {
      "epoch": 0.06563387554867028,
      "grad_norm": 5.659235000610352,
      "learning_rate": 9.92232987214202e-06,
      "loss": 0.0904,
      "step": 1271
    },
    {
      "epoch": 0.0656855151045701,
      "grad_norm": 7.544562816619873,
      "learning_rate": 9.922185946417974e-06,
      "loss": 0.0945,
      "step": 1272
    },
    {
      "epoch": 0.06573715466046992,
      "grad_norm": 4.14890193939209,
      "learning_rate": 9.922041888513175e-06,
      "loss": 0.1016,
      "step": 1273
    },
    {
      "epoch": 0.06578879421636974,
      "grad_norm": 6.5902419090271,
      "learning_rate": 9.921897698431488e-06,
      "loss": 0.0775,
      "step": 1274
    },
    {
      "epoch": 0.06584043377226956,
      "grad_norm": 5.426565647125244,
      "learning_rate": 9.92175337617679e-06,
      "loss": 0.1024,
      "step": 1275
    },
    {
      "epoch": 0.06589207332816938,
      "grad_norm": 3.19451904296875,
      "learning_rate": 9.921608921752953e-06,
      "loss": 0.0751,
      "step": 1276
    },
    {
      "epoch": 0.0659437128840692,
      "grad_norm": 6.372793674468994,
      "learning_rate": 9.921464335163856e-06,
      "loss": 0.0838,
      "step": 1277
    },
    {
      "epoch": 0.06599535243996901,
      "grad_norm": 5.280681610107422,
      "learning_rate": 9.921319616413382e-06,
      "loss": 0.0763,
      "step": 1278
    },
    {
      "epoch": 0.06604699199586883,
      "grad_norm": 4.356475353240967,
      "learning_rate": 9.92117476550542e-06,
      "loss": 0.0571,
      "step": 1279
    },
    {
      "epoch": 0.06609863155176865,
      "grad_norm": 7.970991611480713,
      "learning_rate": 9.921029782443856e-06,
      "loss": 0.078,
      "step": 1280
    },
    {
      "epoch": 0.06615027110766847,
      "grad_norm": 4.028067588806152,
      "learning_rate": 9.920884667232586e-06,
      "loss": 0.075,
      "step": 1281
    },
    {
      "epoch": 0.0662019106635683,
      "grad_norm": 23.28201675415039,
      "learning_rate": 9.920739419875507e-06,
      "loss": 0.2263,
      "step": 1282
    },
    {
      "epoch": 0.06625355021946812,
      "grad_norm": 8.166048049926758,
      "learning_rate": 9.920594040376519e-06,
      "loss": 0.0868,
      "step": 1283
    },
    {
      "epoch": 0.06630518977536794,
      "grad_norm": 3.7279586791992188,
      "learning_rate": 9.920448528739526e-06,
      "loss": 0.0715,
      "step": 1284
    },
    {
      "epoch": 0.06635682933126776,
      "grad_norm": 8.629448890686035,
      "learning_rate": 9.920302884968435e-06,
      "loss": 0.0605,
      "step": 1285
    },
    {
      "epoch": 0.06640846888716757,
      "grad_norm": 3.7924129962921143,
      "learning_rate": 9.920157109067157e-06,
      "loss": 0.1002,
      "step": 1286
    },
    {
      "epoch": 0.06646010844306739,
      "grad_norm": 5.039967060089111,
      "learning_rate": 9.920011201039607e-06,
      "loss": 0.0679,
      "step": 1287
    },
    {
      "epoch": 0.06651174799896721,
      "grad_norm": 7.205632209777832,
      "learning_rate": 9.919865160889703e-06,
      "loss": 0.0566,
      "step": 1288
    },
    {
      "epoch": 0.06656338755486703,
      "grad_norm": 7.976690769195557,
      "learning_rate": 9.919718988621369e-06,
      "loss": 0.1368,
      "step": 1289
    },
    {
      "epoch": 0.06661502711076685,
      "grad_norm": 10.022822380065918,
      "learning_rate": 9.919572684238528e-06,
      "loss": 0.105,
      "step": 1290
    },
    {
      "epoch": 0.06666666666666667,
      "grad_norm": 15.994638442993164,
      "learning_rate": 9.919426247745107e-06,
      "loss": 0.1651,
      "step": 1291
    },
    {
      "epoch": 0.06671830622256648,
      "grad_norm": 7.656363010406494,
      "learning_rate": 9.919279679145045e-06,
      "loss": 0.1345,
      "step": 1292
    },
    {
      "epoch": 0.0667699457784663,
      "grad_norm": 8.473938941955566,
      "learning_rate": 9.91913297844227e-06,
      "loss": 0.1119,
      "step": 1293
    },
    {
      "epoch": 0.06682158533436612,
      "grad_norm": 2.521805763244629,
      "learning_rate": 9.918986145640725e-06,
      "loss": 0.0663,
      "step": 1294
    },
    {
      "epoch": 0.06687322489026594,
      "grad_norm": 5.536164283752441,
      "learning_rate": 9.918839180744355e-06,
      "loss": 0.0733,
      "step": 1295
    },
    {
      "epoch": 0.06692486444616576,
      "grad_norm": 2.302006483078003,
      "learning_rate": 9.918692083757105e-06,
      "loss": 0.0497,
      "step": 1296
    },
    {
      "epoch": 0.06697650400206558,
      "grad_norm": 8.535494804382324,
      "learning_rate": 9.918544854682926e-06,
      "loss": 0.0632,
      "step": 1297
    },
    {
      "epoch": 0.06702814355796541,
      "grad_norm": 6.7510294914245605,
      "learning_rate": 9.91839749352577e-06,
      "loss": 0.104,
      "step": 1298
    },
    {
      "epoch": 0.06707978311386523,
      "grad_norm": 6.219775199890137,
      "learning_rate": 9.918250000289593e-06,
      "loss": 0.0818,
      "step": 1299
    },
    {
      "epoch": 0.06713142266976505,
      "grad_norm": 11.856069564819336,
      "learning_rate": 9.91810237497836e-06,
      "loss": 0.1407,
      "step": 1300
    },
    {
      "epoch": 0.06718306222566486,
      "grad_norm": 7.039827823638916,
      "learning_rate": 9.917954617596033e-06,
      "loss": 0.0644,
      "step": 1301
    },
    {
      "epoch": 0.06723470178156468,
      "grad_norm": 6.672121524810791,
      "learning_rate": 9.917806728146579e-06,
      "loss": 0.0885,
      "step": 1302
    },
    {
      "epoch": 0.0672863413374645,
      "grad_norm": 4.981528282165527,
      "learning_rate": 9.917658706633972e-06,
      "loss": 0.0885,
      "step": 1303
    },
    {
      "epoch": 0.06733798089336432,
      "grad_norm": 10.609665870666504,
      "learning_rate": 9.917510553062184e-06,
      "loss": 0.1374,
      "step": 1304
    },
    {
      "epoch": 0.06738962044926414,
      "grad_norm": 16.03255844116211,
      "learning_rate": 9.917362267435195e-06,
      "loss": 0.1124,
      "step": 1305
    },
    {
      "epoch": 0.06744126000516396,
      "grad_norm": 18.863704681396484,
      "learning_rate": 9.917213849756988e-06,
      "loss": 0.2367,
      "step": 1306
    },
    {
      "epoch": 0.06749289956106377,
      "grad_norm": 9.5030517578125,
      "learning_rate": 9.917065300031548e-06,
      "loss": 0.1249,
      "step": 1307
    },
    {
      "epoch": 0.06754453911696359,
      "grad_norm": 7.682310581207275,
      "learning_rate": 9.916916618262862e-06,
      "loss": 0.1443,
      "step": 1308
    },
    {
      "epoch": 0.06759617867286341,
      "grad_norm": 3.459469795227051,
      "learning_rate": 9.916767804454926e-06,
      "loss": 0.0744,
      "step": 1309
    },
    {
      "epoch": 0.06764781822876323,
      "grad_norm": 4.083858489990234,
      "learning_rate": 9.916618858611735e-06,
      "loss": 0.0753,
      "step": 1310
    },
    {
      "epoch": 0.06769945778466305,
      "grad_norm": 7.9067487716674805,
      "learning_rate": 9.916469780737287e-06,
      "loss": 0.1188,
      "step": 1311
    },
    {
      "epoch": 0.06775109734056287,
      "grad_norm": 6.65146017074585,
      "learning_rate": 9.91632057083559e-06,
      "loss": 0.0543,
      "step": 1312
    },
    {
      "epoch": 0.06780273689646268,
      "grad_norm": 15.28924560546875,
      "learning_rate": 9.916171228910645e-06,
      "loss": 0.2237,
      "step": 1313
    },
    {
      "epoch": 0.06785437645236252,
      "grad_norm": 8.036544799804688,
      "learning_rate": 9.916021754966467e-06,
      "loss": 0.1465,
      "step": 1314
    },
    {
      "epoch": 0.06790601600826233,
      "grad_norm": 12.149577140808105,
      "learning_rate": 9.915872149007066e-06,
      "loss": 0.1593,
      "step": 1315
    },
    {
      "epoch": 0.06795765556416215,
      "grad_norm": 5.963313102722168,
      "learning_rate": 9.915722411036464e-06,
      "loss": 0.0629,
      "step": 1316
    },
    {
      "epoch": 0.06800929512006197,
      "grad_norm": 9.893052101135254,
      "learning_rate": 9.91557254105868e-06,
      "loss": 0.1598,
      "step": 1317
    },
    {
      "epoch": 0.06806093467596179,
      "grad_norm": 8.609441757202148,
      "learning_rate": 9.915422539077738e-06,
      "loss": 0.1415,
      "step": 1318
    },
    {
      "epoch": 0.06811257423186161,
      "grad_norm": 5.389427661895752,
      "learning_rate": 9.915272405097666e-06,
      "loss": 0.0898,
      "step": 1319
    },
    {
      "epoch": 0.06816421378776143,
      "grad_norm": 8.800772666931152,
      "learning_rate": 9.915122139122498e-06,
      "loss": 0.1628,
      "step": 1320
    },
    {
      "epoch": 0.06821585334366125,
      "grad_norm": 27.063777923583984,
      "learning_rate": 9.914971741156265e-06,
      "loss": 0.2064,
      "step": 1321
    },
    {
      "epoch": 0.06826749289956106,
      "grad_norm": 11.166287422180176,
      "learning_rate": 9.91482121120301e-06,
      "loss": 0.1074,
      "step": 1322
    },
    {
      "epoch": 0.06831913245546088,
      "grad_norm": 7.151319980621338,
      "learning_rate": 9.914670549266775e-06,
      "loss": 0.1309,
      "step": 1323
    },
    {
      "epoch": 0.0683707720113607,
      "grad_norm": 4.049313068389893,
      "learning_rate": 9.914519755351602e-06,
      "loss": 0.0581,
      "step": 1324
    },
    {
      "epoch": 0.06842241156726052,
      "grad_norm": 3.8050293922424316,
      "learning_rate": 9.914368829461546e-06,
      "loss": 0.0804,
      "step": 1325
    },
    {
      "epoch": 0.06847405112316034,
      "grad_norm": 14.757622718811035,
      "learning_rate": 9.914217771600654e-06,
      "loss": 0.1507,
      "step": 1326
    },
    {
      "epoch": 0.06852569067906016,
      "grad_norm": 7.211677074432373,
      "learning_rate": 9.914066581772988e-06,
      "loss": 0.1076,
      "step": 1327
    },
    {
      "epoch": 0.06857733023495997,
      "grad_norm": 10.425911903381348,
      "learning_rate": 9.913915259982604e-06,
      "loss": 0.0803,
      "step": 1328
    },
    {
      "epoch": 0.06862896979085979,
      "grad_norm": 4.588395118713379,
      "learning_rate": 9.913763806233568e-06,
      "loss": 0.1,
      "step": 1329
    },
    {
      "epoch": 0.06868060934675962,
      "grad_norm": 4.689230918884277,
      "learning_rate": 9.913612220529946e-06,
      "loss": 0.0965,
      "step": 1330
    },
    {
      "epoch": 0.06873224890265944,
      "grad_norm": 20.405303955078125,
      "learning_rate": 9.913460502875808e-06,
      "loss": 0.1383,
      "step": 1331
    },
    {
      "epoch": 0.06878388845855926,
      "grad_norm": 7.568155288696289,
      "learning_rate": 9.913308653275232e-06,
      "loss": 0.0856,
      "step": 1332
    },
    {
      "epoch": 0.06883552801445908,
      "grad_norm": 5.780739784240723,
      "learning_rate": 9.91315667173229e-06,
      "loss": 0.0939,
      "step": 1333
    },
    {
      "epoch": 0.0688871675703589,
      "grad_norm": 2.367008686065674,
      "learning_rate": 9.913004558251071e-06,
      "loss": 0.0346,
      "step": 1334
    },
    {
      "epoch": 0.06893880712625872,
      "grad_norm": 5.467902183532715,
      "learning_rate": 9.912852312835652e-06,
      "loss": 0.066,
      "step": 1335
    },
    {
      "epoch": 0.06899044668215853,
      "grad_norm": 7.9226837158203125,
      "learning_rate": 9.912699935490126e-06,
      "loss": 0.054,
      "step": 1336
    },
    {
      "epoch": 0.06904208623805835,
      "grad_norm": 6.699788570404053,
      "learning_rate": 9.912547426218582e-06,
      "loss": 0.0575,
      "step": 1337
    },
    {
      "epoch": 0.06909372579395817,
      "grad_norm": 5.843038558959961,
      "learning_rate": 9.91239478502512e-06,
      "loss": 0.0616,
      "step": 1338
    },
    {
      "epoch": 0.06914536534985799,
      "grad_norm": 12.948184967041016,
      "learning_rate": 9.912242011913833e-06,
      "loss": 0.1111,
      "step": 1339
    },
    {
      "epoch": 0.06919700490575781,
      "grad_norm": 7.509604454040527,
      "learning_rate": 9.912089106888829e-06,
      "loss": 0.0947,
      "step": 1340
    },
    {
      "epoch": 0.06924864446165763,
      "grad_norm": 11.551898956298828,
      "learning_rate": 9.91193606995421e-06,
      "loss": 0.1464,
      "step": 1341
    },
    {
      "epoch": 0.06930028401755745,
      "grad_norm": 4.977790355682373,
      "learning_rate": 9.91178290111409e-06,
      "loss": 0.0639,
      "step": 1342
    },
    {
      "epoch": 0.06935192357345726,
      "grad_norm": 5.4888916015625,
      "learning_rate": 9.911629600372578e-06,
      "loss": 0.1141,
      "step": 1343
    },
    {
      "epoch": 0.06940356312935708,
      "grad_norm": 8.371848106384277,
      "learning_rate": 9.911476167733793e-06,
      "loss": 0.1212,
      "step": 1344
    },
    {
      "epoch": 0.0694552026852569,
      "grad_norm": 6.015946388244629,
      "learning_rate": 9.911322603201855e-06,
      "loss": 0.1438,
      "step": 1345
    },
    {
      "epoch": 0.06950684224115673,
      "grad_norm": 3.949768304824829,
      "learning_rate": 9.911168906780888e-06,
      "loss": 0.0667,
      "step": 1346
    },
    {
      "epoch": 0.06955848179705655,
      "grad_norm": 10.267816543579102,
      "learning_rate": 9.911015078475019e-06,
      "loss": 0.1023,
      "step": 1347
    },
    {
      "epoch": 0.06961012135295637,
      "grad_norm": 7.508604526519775,
      "learning_rate": 9.910861118288377e-06,
      "loss": 0.1187,
      "step": 1348
    },
    {
      "epoch": 0.06966176090885619,
      "grad_norm": 4.787530899047852,
      "learning_rate": 9.910707026225101e-06,
      "loss": 0.0496,
      "step": 1349
    },
    {
      "epoch": 0.069713400464756,
      "grad_norm": 2.990565776824951,
      "learning_rate": 9.910552802289326e-06,
      "loss": 0.035,
      "step": 1350
    },
    {
      "epoch": 0.06976504002065582,
      "grad_norm": 4.355076789855957,
      "learning_rate": 9.910398446485192e-06,
      "loss": 0.0613,
      "step": 1351
    },
    {
      "epoch": 0.06981667957655564,
      "grad_norm": 4.270380973815918,
      "learning_rate": 9.910243958816847e-06,
      "loss": 0.0887,
      "step": 1352
    },
    {
      "epoch": 0.06986831913245546,
      "grad_norm": 3.332289934158325,
      "learning_rate": 9.91008933928844e-06,
      "loss": 0.0973,
      "step": 1353
    },
    {
      "epoch": 0.06991995868835528,
      "grad_norm": 6.143850803375244,
      "learning_rate": 9.909934587904121e-06,
      "loss": 0.0704,
      "step": 1354
    },
    {
      "epoch": 0.0699715982442551,
      "grad_norm": 4.577143669128418,
      "learning_rate": 9.909779704668046e-06,
      "loss": 0.0724,
      "step": 1355
    },
    {
      "epoch": 0.07002323780015492,
      "grad_norm": 5.497023582458496,
      "learning_rate": 9.909624689584376e-06,
      "loss": 0.0666,
      "step": 1356
    },
    {
      "epoch": 0.07007487735605474,
      "grad_norm": 6.708907604217529,
      "learning_rate": 9.909469542657271e-06,
      "loss": 0.1257,
      "step": 1357
    },
    {
      "epoch": 0.07012651691195455,
      "grad_norm": 6.615805149078369,
      "learning_rate": 9.909314263890899e-06,
      "loss": 0.1794,
      "step": 1358
    },
    {
      "epoch": 0.07017815646785437,
      "grad_norm": 5.423852920532227,
      "learning_rate": 9.90915885328943e-06,
      "loss": 0.0664,
      "step": 1359
    },
    {
      "epoch": 0.07022979602375419,
      "grad_norm": 6.149082183837891,
      "learning_rate": 9.909003310857039e-06,
      "loss": 0.0662,
      "step": 1360
    },
    {
      "epoch": 0.07028143557965401,
      "grad_norm": 5.271681785583496,
      "learning_rate": 9.908847636597899e-06,
      "loss": 0.1026,
      "step": 1361
    },
    {
      "epoch": 0.07033307513555384,
      "grad_norm": 39.524391174316406,
      "learning_rate": 9.908691830516194e-06,
      "loss": 0.1552,
      "step": 1362
    },
    {
      "epoch": 0.07038471469145366,
      "grad_norm": 8.136341094970703,
      "learning_rate": 9.908535892616106e-06,
      "loss": 0.1433,
      "step": 1363
    },
    {
      "epoch": 0.07043635424735348,
      "grad_norm": 9.002847671508789,
      "learning_rate": 9.908379822901823e-06,
      "loss": 0.1091,
      "step": 1364
    },
    {
      "epoch": 0.0704879938032533,
      "grad_norm": 5.843984127044678,
      "learning_rate": 9.908223621377535e-06,
      "loss": 0.1083,
      "step": 1365
    },
    {
      "epoch": 0.07053963335915311,
      "grad_norm": 18.135719299316406,
      "learning_rate": 9.90806728804744e-06,
      "loss": 0.1118,
      "step": 1366
    },
    {
      "epoch": 0.07059127291505293,
      "grad_norm": 3.7867112159729004,
      "learning_rate": 9.907910822915733e-06,
      "loss": 0.0426,
      "step": 1367
    },
    {
      "epoch": 0.07064291247095275,
      "grad_norm": 6.340321063995361,
      "learning_rate": 9.90775422598662e-06,
      "loss": 0.0844,
      "step": 1368
    },
    {
      "epoch": 0.07069455202685257,
      "grad_norm": 4.155909061431885,
      "learning_rate": 9.907597497264298e-06,
      "loss": 0.0811,
      "step": 1369
    },
    {
      "epoch": 0.07074619158275239,
      "grad_norm": 4.235940456390381,
      "learning_rate": 9.907440636752983e-06,
      "loss": 0.0969,
      "step": 1370
    },
    {
      "epoch": 0.0707978311386522,
      "grad_norm": 20.91172218322754,
      "learning_rate": 9.907283644456886e-06,
      "loss": 0.2669,
      "step": 1371
    },
    {
      "epoch": 0.07084947069455202,
      "grad_norm": 7.226406097412109,
      "learning_rate": 9.907126520380222e-06,
      "loss": 0.0918,
      "step": 1372
    },
    {
      "epoch": 0.07090111025045184,
      "grad_norm": 7.369819164276123,
      "learning_rate": 9.90696926452721e-06,
      "loss": 0.0781,
      "step": 1373
    },
    {
      "epoch": 0.07095274980635166,
      "grad_norm": 3.830512285232544,
      "learning_rate": 9.906811876902075e-06,
      "loss": 0.1382,
      "step": 1374
    },
    {
      "epoch": 0.07100438936225148,
      "grad_norm": 7.348693370819092,
      "learning_rate": 9.90665435750904e-06,
      "loss": 0.096,
      "step": 1375
    },
    {
      "epoch": 0.0710560289181513,
      "grad_norm": 1.69636070728302,
      "learning_rate": 9.90649670635234e-06,
      "loss": 0.0411,
      "step": 1376
    },
    {
      "epoch": 0.07110766847405112,
      "grad_norm": 12.272241592407227,
      "learning_rate": 9.906338923436203e-06,
      "loss": 0.1486,
      "step": 1377
    },
    {
      "epoch": 0.07115930802995095,
      "grad_norm": 8.255807876586914,
      "learning_rate": 9.90618100876487e-06,
      "loss": 0.0549,
      "step": 1378
    },
    {
      "epoch": 0.07121094758585077,
      "grad_norm": 3.503892183303833,
      "learning_rate": 9.90602296234258e-06,
      "loss": 0.0527,
      "step": 1379
    },
    {
      "epoch": 0.07126258714175059,
      "grad_norm": 8.46487045288086,
      "learning_rate": 9.905864784173578e-06,
      "loss": 0.0838,
      "step": 1380
    },
    {
      "epoch": 0.0713142266976504,
      "grad_norm": 6.217231750488281,
      "learning_rate": 9.90570647426211e-06,
      "loss": 0.0949,
      "step": 1381
    },
    {
      "epoch": 0.07136586625355022,
      "grad_norm": 10.362788200378418,
      "learning_rate": 9.905548032612429e-06,
      "loss": 0.1522,
      "step": 1382
    },
    {
      "epoch": 0.07141750580945004,
      "grad_norm": 4.073149681091309,
      "learning_rate": 9.90538945922879e-06,
      "loss": 0.0704,
      "step": 1383
    },
    {
      "epoch": 0.07146914536534986,
      "grad_norm": 10.534314155578613,
      "learning_rate": 9.90523075411545e-06,
      "loss": 0.1062,
      "step": 1384
    },
    {
      "epoch": 0.07152078492124968,
      "grad_norm": 13.903188705444336,
      "learning_rate": 9.905071917276675e-06,
      "loss": 0.1594,
      "step": 1385
    },
    {
      "epoch": 0.0715724244771495,
      "grad_norm": 9.87350082397461,
      "learning_rate": 9.904912948716723e-06,
      "loss": 0.114,
      "step": 1386
    },
    {
      "epoch": 0.07162406403304931,
      "grad_norm": 10.744388580322266,
      "learning_rate": 9.90475384843987e-06,
      "loss": 0.094,
      "step": 1387
    },
    {
      "epoch": 0.07167570358894913,
      "grad_norm": 6.064027309417725,
      "learning_rate": 9.904594616450383e-06,
      "loss": 0.0716,
      "step": 1388
    },
    {
      "epoch": 0.07172734314484895,
      "grad_norm": 6.393805027008057,
      "learning_rate": 9.90443525275254e-06,
      "loss": 0.1029,
      "step": 1389
    },
    {
      "epoch": 0.07177898270074877,
      "grad_norm": 7.3625264167785645,
      "learning_rate": 9.904275757350623e-06,
      "loss": 0.0783,
      "step": 1390
    },
    {
      "epoch": 0.07183062225664859,
      "grad_norm": 3.916205644607544,
      "learning_rate": 9.904116130248913e-06,
      "loss": 0.0649,
      "step": 1391
    },
    {
      "epoch": 0.0718822618125484,
      "grad_norm": 7.671761512756348,
      "learning_rate": 9.903956371451698e-06,
      "loss": 0.1005,
      "step": 1392
    },
    {
      "epoch": 0.07193390136844822,
      "grad_norm": 3.5519344806671143,
      "learning_rate": 9.903796480963264e-06,
      "loss": 0.0758,
      "step": 1393
    },
    {
      "epoch": 0.07198554092434806,
      "grad_norm": 4.115356922149658,
      "learning_rate": 9.90363645878791e-06,
      "loss": 0.056,
      "step": 1394
    },
    {
      "epoch": 0.07203718048024788,
      "grad_norm": 5.906105995178223,
      "learning_rate": 9.90347630492993e-06,
      "loss": 0.072,
      "step": 1395
    },
    {
      "epoch": 0.0720888200361477,
      "grad_norm": 18.071977615356445,
      "learning_rate": 9.903316019393625e-06,
      "loss": 0.2106,
      "step": 1396
    },
    {
      "epoch": 0.07214045959204751,
      "grad_norm": 5.8592915534973145,
      "learning_rate": 9.903155602183302e-06,
      "loss": 0.1246,
      "step": 1397
    },
    {
      "epoch": 0.07219209914794733,
      "grad_norm": 4.405320167541504,
      "learning_rate": 9.902995053303266e-06,
      "loss": 0.0691,
      "step": 1398
    },
    {
      "epoch": 0.07224373870384715,
      "grad_norm": 9.566997528076172,
      "learning_rate": 9.902834372757829e-06,
      "loss": 0.1258,
      "step": 1399
    },
    {
      "epoch": 0.07229537825974697,
      "grad_norm": 8.157554626464844,
      "learning_rate": 9.902673560551307e-06,
      "loss": 0.0982,
      "step": 1400
    },
    {
      "epoch": 0.07234701781564679,
      "grad_norm": 8.82274055480957,
      "learning_rate": 9.902512616688017e-06,
      "loss": 0.069,
      "step": 1401
    },
    {
      "epoch": 0.0723986573715466,
      "grad_norm": 10.751070976257324,
      "learning_rate": 9.902351541172282e-06,
      "loss": 0.1155,
      "step": 1402
    },
    {
      "epoch": 0.07245029692744642,
      "grad_norm": 4.588295936584473,
      "learning_rate": 9.902190334008428e-06,
      "loss": 0.0849,
      "step": 1403
    },
    {
      "epoch": 0.07250193648334624,
      "grad_norm": 10.583535194396973,
      "learning_rate": 9.902028995200781e-06,
      "loss": 0.1451,
      "step": 1404
    },
    {
      "epoch": 0.07255357603924606,
      "grad_norm": 20.56202507019043,
      "learning_rate": 9.90186752475368e-06,
      "loss": 0.0504,
      "step": 1405
    },
    {
      "epoch": 0.07260521559514588,
      "grad_norm": 4.189383029937744,
      "learning_rate": 9.901705922671454e-06,
      "loss": 0.0506,
      "step": 1406
    },
    {
      "epoch": 0.0726568551510457,
      "grad_norm": 10.40279769897461,
      "learning_rate": 9.901544188958445e-06,
      "loss": 0.1026,
      "step": 1407
    },
    {
      "epoch": 0.07270849470694551,
      "grad_norm": 2.0267105102539062,
      "learning_rate": 9.901382323618997e-06,
      "loss": 0.0517,
      "step": 1408
    },
    {
      "epoch": 0.07276013426284533,
      "grad_norm": 17.865154266357422,
      "learning_rate": 9.901220326657459e-06,
      "loss": 0.1429,
      "step": 1409
    },
    {
      "epoch": 0.07281177381874517,
      "grad_norm": 8.260114669799805,
      "learning_rate": 9.901058198078178e-06,
      "loss": 0.0809,
      "step": 1410
    },
    {
      "epoch": 0.07286341337464498,
      "grad_norm": 4.385097980499268,
      "learning_rate": 9.900895937885508e-06,
      "loss": 0.0747,
      "step": 1411
    },
    {
      "epoch": 0.0729150529305448,
      "grad_norm": 6.666133880615234,
      "learning_rate": 9.900733546083808e-06,
      "loss": 0.0605,
      "step": 1412
    },
    {
      "epoch": 0.07296669248644462,
      "grad_norm": 3.268667459487915,
      "learning_rate": 9.900571022677437e-06,
      "loss": 0.0512,
      "step": 1413
    },
    {
      "epoch": 0.07301833204234444,
      "grad_norm": 6.059147834777832,
      "learning_rate": 9.900408367670763e-06,
      "loss": 0.0453,
      "step": 1414
    },
    {
      "epoch": 0.07306997159824426,
      "grad_norm": 4.692196369171143,
      "learning_rate": 9.900245581068148e-06,
      "loss": 0.0755,
      "step": 1415
    },
    {
      "epoch": 0.07312161115414408,
      "grad_norm": 21.003406524658203,
      "learning_rate": 9.900082662873969e-06,
      "loss": 0.1545,
      "step": 1416
    },
    {
      "epoch": 0.0731732507100439,
      "grad_norm": 3.814271926879883,
      "learning_rate": 9.899919613092598e-06,
      "loss": 0.0431,
      "step": 1417
    },
    {
      "epoch": 0.07322489026594371,
      "grad_norm": 18.252321243286133,
      "learning_rate": 9.899756431728415e-06,
      "loss": 0.1112,
      "step": 1418
    },
    {
      "epoch": 0.07327652982184353,
      "grad_norm": 10.341623306274414,
      "learning_rate": 9.8995931187858e-06,
      "loss": 0.1274,
      "step": 1419
    },
    {
      "epoch": 0.07332816937774335,
      "grad_norm": 9.239235877990723,
      "learning_rate": 9.899429674269142e-06,
      "loss": 0.1269,
      "step": 1420
    },
    {
      "epoch": 0.07337980893364317,
      "grad_norm": 6.625662803649902,
      "learning_rate": 9.899266098182827e-06,
      "loss": 0.0664,
      "step": 1421
    },
    {
      "epoch": 0.07343144848954299,
      "grad_norm": 2.3173179626464844,
      "learning_rate": 9.89910239053125e-06,
      "loss": 0.0118,
      "step": 1422
    },
    {
      "epoch": 0.0734830880454428,
      "grad_norm": 10.450085639953613,
      "learning_rate": 9.898938551318807e-06,
      "loss": 0.0878,
      "step": 1423
    },
    {
      "epoch": 0.07353472760134262,
      "grad_norm": 8.23825454711914,
      "learning_rate": 9.898774580549895e-06,
      "loss": 0.0869,
      "step": 1424
    },
    {
      "epoch": 0.07358636715724244,
      "grad_norm": 11.435626983642578,
      "learning_rate": 9.89861047822892e-06,
      "loss": 0.0634,
      "step": 1425
    },
    {
      "epoch": 0.07363800671314227,
      "grad_norm": 7.565819263458252,
      "learning_rate": 9.898446244360287e-06,
      "loss": 0.1266,
      "step": 1426
    },
    {
      "epoch": 0.07368964626904209,
      "grad_norm": 7.757447719573975,
      "learning_rate": 9.898281878948408e-06,
      "loss": 0.1529,
      "step": 1427
    },
    {
      "epoch": 0.07374128582494191,
      "grad_norm": 8.126981735229492,
      "learning_rate": 9.898117381997696e-06,
      "loss": 0.1055,
      "step": 1428
    },
    {
      "epoch": 0.07379292538084173,
      "grad_norm": 5.061715602874756,
      "learning_rate": 9.89795275351257e-06,
      "loss": 0.0692,
      "step": 1429
    },
    {
      "epoch": 0.07384456493674155,
      "grad_norm": 3.4910943508148193,
      "learning_rate": 9.89778799349745e-06,
      "loss": 0.0994,
      "step": 1430
    },
    {
      "epoch": 0.07389620449264137,
      "grad_norm": 13.545804977416992,
      "learning_rate": 9.897623101956757e-06,
      "loss": 0.0462,
      "step": 1431
    },
    {
      "epoch": 0.07394784404854118,
      "grad_norm": 5.253116607666016,
      "learning_rate": 9.897458078894925e-06,
      "loss": 0.067,
      "step": 1432
    },
    {
      "epoch": 0.073999483604441,
      "grad_norm": 4.77596378326416,
      "learning_rate": 9.897292924316381e-06,
      "loss": 0.0765,
      "step": 1433
    },
    {
      "epoch": 0.07405112316034082,
      "grad_norm": 8.27817440032959,
      "learning_rate": 9.897127638225563e-06,
      "loss": 0.1042,
      "step": 1434
    },
    {
      "epoch": 0.07410276271624064,
      "grad_norm": 20.247007369995117,
      "learning_rate": 9.896962220626907e-06,
      "loss": 0.245,
      "step": 1435
    },
    {
      "epoch": 0.07415440227214046,
      "grad_norm": 8.865907669067383,
      "learning_rate": 9.896796671524858e-06,
      "loss": 0.1347,
      "step": 1436
    },
    {
      "epoch": 0.07420604182804028,
      "grad_norm": 11.302104949951172,
      "learning_rate": 9.896630990923858e-06,
      "loss": 0.0916,
      "step": 1437
    },
    {
      "epoch": 0.0742576813839401,
      "grad_norm": 5.079559803009033,
      "learning_rate": 9.89646517882836e-06,
      "loss": 0.0625,
      "step": 1438
    },
    {
      "epoch": 0.07430932093983991,
      "grad_norm": 7.780796051025391,
      "learning_rate": 9.896299235242814e-06,
      "loss": 0.101,
      "step": 1439
    },
    {
      "epoch": 0.07436096049573973,
      "grad_norm": 2.7789483070373535,
      "learning_rate": 9.89613316017168e-06,
      "loss": 0.0476,
      "step": 1440
    },
    {
      "epoch": 0.07441260005163955,
      "grad_norm": 9.445821762084961,
      "learning_rate": 9.895966953619411e-06,
      "loss": 0.1024,
      "step": 1441
    },
    {
      "epoch": 0.07446423960753938,
      "grad_norm": 15.584931373596191,
      "learning_rate": 9.895800615590477e-06,
      "loss": 0.1973,
      "step": 1442
    },
    {
      "epoch": 0.0745158791634392,
      "grad_norm": 12.551444053649902,
      "learning_rate": 9.895634146089343e-06,
      "loss": 0.1018,
      "step": 1443
    },
    {
      "epoch": 0.07456751871933902,
      "grad_norm": 13.22176456451416,
      "learning_rate": 9.895467545120476e-06,
      "loss": 0.1369,
      "step": 1444
    },
    {
      "epoch": 0.07461915827523884,
      "grad_norm": 7.886132717132568,
      "learning_rate": 9.895300812688355e-06,
      "loss": 0.0773,
      "step": 1445
    },
    {
      "epoch": 0.07467079783113865,
      "grad_norm": 9.36072826385498,
      "learning_rate": 9.895133948797454e-06,
      "loss": 0.1381,
      "step": 1446
    },
    {
      "epoch": 0.07472243738703847,
      "grad_norm": 13.242191314697266,
      "learning_rate": 9.894966953452255e-06,
      "loss": 0.0981,
      "step": 1447
    },
    {
      "epoch": 0.07477407694293829,
      "grad_norm": 4.233486175537109,
      "learning_rate": 9.894799826657242e-06,
      "loss": 0.0925,
      "step": 1448
    },
    {
      "epoch": 0.07482571649883811,
      "grad_norm": 4.473855018615723,
      "learning_rate": 9.894632568416905e-06,
      "loss": 0.0669,
      "step": 1449
    },
    {
      "epoch": 0.07487735605473793,
      "grad_norm": 5.834971904754639,
      "learning_rate": 9.894465178735733e-06,
      "loss": 0.058,
      "step": 1450
    },
    {
      "epoch": 0.07492899561063775,
      "grad_norm": 2.769235849380493,
      "learning_rate": 9.894297657618223e-06,
      "loss": 0.0614,
      "step": 1451
    },
    {
      "epoch": 0.07498063516653757,
      "grad_norm": 7.935806751251221,
      "learning_rate": 9.894130005068872e-06,
      "loss": 0.1108,
      "step": 1452
    },
    {
      "epoch": 0.07503227472243738,
      "grad_norm": 6.011410236358643,
      "learning_rate": 9.893962221092184e-06,
      "loss": 0.1046,
      "step": 1453
    },
    {
      "epoch": 0.0750839142783372,
      "grad_norm": 12.028877258300781,
      "learning_rate": 9.893794305692663e-06,
      "loss": 0.1131,
      "step": 1454
    },
    {
      "epoch": 0.07513555383423702,
      "grad_norm": 5.340762138366699,
      "learning_rate": 9.893626258874818e-06,
      "loss": 0.0395,
      "step": 1455
    },
    {
      "epoch": 0.07518719339013684,
      "grad_norm": 6.947221279144287,
      "learning_rate": 9.893458080643164e-06,
      "loss": 0.083,
      "step": 1456
    },
    {
      "epoch": 0.07523883294603666,
      "grad_norm": 2.8537750244140625,
      "learning_rate": 9.893289771002215e-06,
      "loss": 0.0431,
      "step": 1457
    },
    {
      "epoch": 0.07529047250193649,
      "grad_norm": 7.616004943847656,
      "learning_rate": 9.893121329956492e-06,
      "loss": 0.0872,
      "step": 1458
    },
    {
      "epoch": 0.07534211205783631,
      "grad_norm": 8.377547264099121,
      "learning_rate": 9.892952757510518e-06,
      "loss": 0.0966,
      "step": 1459
    },
    {
      "epoch": 0.07539375161373613,
      "grad_norm": 8.090509414672852,
      "learning_rate": 9.89278405366882e-06,
      "loss": 0.1576,
      "step": 1460
    },
    {
      "epoch": 0.07544539116963594,
      "grad_norm": 4.312973499298096,
      "learning_rate": 9.89261521843593e-06,
      "loss": 0.0646,
      "step": 1461
    },
    {
      "epoch": 0.07549703072553576,
      "grad_norm": 5.347316741943359,
      "learning_rate": 9.892446251816378e-06,
      "loss": 0.0938,
      "step": 1462
    },
    {
      "epoch": 0.07554867028143558,
      "grad_norm": 6.790285587310791,
      "learning_rate": 9.892277153814704e-06,
      "loss": 0.0726,
      "step": 1463
    },
    {
      "epoch": 0.0756003098373354,
      "grad_norm": 5.863140106201172,
      "learning_rate": 9.892107924435449e-06,
      "loss": 0.0713,
      "step": 1464
    },
    {
      "epoch": 0.07565194939323522,
      "grad_norm": 7.115054130554199,
      "learning_rate": 9.891938563683157e-06,
      "loss": 0.0671,
      "step": 1465
    },
    {
      "epoch": 0.07570358894913504,
      "grad_norm": 8.285832405090332,
      "learning_rate": 9.891769071562376e-06,
      "loss": 0.0987,
      "step": 1466
    },
    {
      "epoch": 0.07575522850503486,
      "grad_norm": 8.662095069885254,
      "learning_rate": 9.891599448077659e-06,
      "loss": 0.0882,
      "step": 1467
    },
    {
      "epoch": 0.07580686806093467,
      "grad_norm": 12.331743240356445,
      "learning_rate": 9.891429693233556e-06,
      "loss": 0.1056,
      "step": 1468
    },
    {
      "epoch": 0.07585850761683449,
      "grad_norm": 2.9506211280822754,
      "learning_rate": 9.891259807034633e-06,
      "loss": 0.0459,
      "step": 1469
    },
    {
      "epoch": 0.07591014717273431,
      "grad_norm": 6.1873064041137695,
      "learning_rate": 9.891089789485447e-06,
      "loss": 0.1281,
      "step": 1470
    },
    {
      "epoch": 0.07596178672863413,
      "grad_norm": 4.032969951629639,
      "learning_rate": 9.890919640590566e-06,
      "loss": 0.0663,
      "step": 1471
    },
    {
      "epoch": 0.07601342628453395,
      "grad_norm": 13.026461601257324,
      "learning_rate": 9.890749360354557e-06,
      "loss": 0.1478,
      "step": 1472
    },
    {
      "epoch": 0.07606506584043377,
      "grad_norm": 10.88941478729248,
      "learning_rate": 9.890578948781995e-06,
      "loss": 0.0721,
      "step": 1473
    },
    {
      "epoch": 0.0761167053963336,
      "grad_norm": 6.002615451812744,
      "learning_rate": 9.890408405877456e-06,
      "loss": 0.0925,
      "step": 1474
    },
    {
      "epoch": 0.07616834495223342,
      "grad_norm": 6.8716721534729,
      "learning_rate": 9.890237731645518e-06,
      "loss": 0.1015,
      "step": 1475
    },
    {
      "epoch": 0.07621998450813323,
      "grad_norm": 2.4050464630126953,
      "learning_rate": 9.890066926090766e-06,
      "loss": 0.0395,
      "step": 1476
    },
    {
      "epoch": 0.07627162406403305,
      "grad_norm": 28.239473342895508,
      "learning_rate": 9.889895989217786e-06,
      "loss": 0.2107,
      "step": 1477
    },
    {
      "epoch": 0.07632326361993287,
      "grad_norm": 7.384927272796631,
      "learning_rate": 9.889724921031169e-06,
      "loss": 0.0771,
      "step": 1478
    },
    {
      "epoch": 0.07637490317583269,
      "grad_norm": 10.251492500305176,
      "learning_rate": 9.889553721535508e-06,
      "loss": 0.1411,
      "step": 1479
    },
    {
      "epoch": 0.07642654273173251,
      "grad_norm": 11.797566413879395,
      "learning_rate": 9.8893823907354e-06,
      "loss": 0.0979,
      "step": 1480
    },
    {
      "epoch": 0.07647818228763233,
      "grad_norm": 4.84760856628418,
      "learning_rate": 9.889210928635448e-06,
      "loss": 0.0822,
      "step": 1481
    },
    {
      "epoch": 0.07652982184353214,
      "grad_norm": 10.370800971984863,
      "learning_rate": 9.889039335240256e-06,
      "loss": 0.0806,
      "step": 1482
    },
    {
      "epoch": 0.07658146139943196,
      "grad_norm": 6.509635925292969,
      "learning_rate": 9.888867610554429e-06,
      "loss": 0.0653,
      "step": 1483
    },
    {
      "epoch": 0.07663310095533178,
      "grad_norm": 4.364871501922607,
      "learning_rate": 9.888695754582583e-06,
      "loss": 0.1035,
      "step": 1484
    },
    {
      "epoch": 0.0766847405112316,
      "grad_norm": 15.824504852294922,
      "learning_rate": 9.888523767329329e-06,
      "loss": 0.1336,
      "step": 1485
    },
    {
      "epoch": 0.07673638006713142,
      "grad_norm": 6.411131858825684,
      "learning_rate": 9.888351648799288e-06,
      "loss": 0.0856,
      "step": 1486
    },
    {
      "epoch": 0.07678801962303124,
      "grad_norm": 9.850699424743652,
      "learning_rate": 9.888179398997083e-06,
      "loss": 0.0913,
      "step": 1487
    },
    {
      "epoch": 0.07683965917893106,
      "grad_norm": 8.082440376281738,
      "learning_rate": 9.888007017927335e-06,
      "loss": 0.0943,
      "step": 1488
    },
    {
      "epoch": 0.07689129873483087,
      "grad_norm": 12.189684867858887,
      "learning_rate": 9.887834505594678e-06,
      "loss": 0.1166,
      "step": 1489
    },
    {
      "epoch": 0.0769429382907307,
      "grad_norm": 15.967485427856445,
      "learning_rate": 9.887661862003741e-06,
      "loss": 0.2187,
      "step": 1490
    },
    {
      "epoch": 0.07699457784663052,
      "grad_norm": 5.966521739959717,
      "learning_rate": 9.887489087159165e-06,
      "loss": 0.0821,
      "step": 1491
    },
    {
      "epoch": 0.07704621740253034,
      "grad_norm": 2.8560187816619873,
      "learning_rate": 9.887316181065584e-06,
      "loss": 0.0606,
      "step": 1492
    },
    {
      "epoch": 0.07709785695843016,
      "grad_norm": 7.084949493408203,
      "learning_rate": 9.887143143727645e-06,
      "loss": 0.0642,
      "step": 1493
    },
    {
      "epoch": 0.07714949651432998,
      "grad_norm": 5.893911361694336,
      "learning_rate": 9.886969975149993e-06,
      "loss": 0.0762,
      "step": 1494
    },
    {
      "epoch": 0.0772011360702298,
      "grad_norm": 3.5023701190948486,
      "learning_rate": 9.88679667533728e-06,
      "loss": 0.0275,
      "step": 1495
    },
    {
      "epoch": 0.07725277562612962,
      "grad_norm": 5.2905755043029785,
      "learning_rate": 9.88662324429416e-06,
      "loss": 0.0921,
      "step": 1496
    },
    {
      "epoch": 0.07730441518202943,
      "grad_norm": 5.245697498321533,
      "learning_rate": 9.886449682025288e-06,
      "loss": 0.0765,
      "step": 1497
    },
    {
      "epoch": 0.07735605473792925,
      "grad_norm": 9.473251342773438,
      "learning_rate": 9.886275988535324e-06,
      "loss": 0.0694,
      "step": 1498
    },
    {
      "epoch": 0.07740769429382907,
      "grad_norm": 5.2640790939331055,
      "learning_rate": 9.886102163828937e-06,
      "loss": 0.0588,
      "step": 1499
    },
    {
      "epoch": 0.07745933384972889,
      "grad_norm": 16.28326988220215,
      "learning_rate": 9.88592820791079e-06,
      "loss": 0.1014,
      "step": 1500
    },
    {
      "epoch": 0.07751097340562871,
      "grad_norm": 5.4703755378723145,
      "learning_rate": 9.885754120785559e-06,
      "loss": 0.0427,
      "step": 1501
    },
    {
      "epoch": 0.07756261296152853,
      "grad_norm": 9.543052673339844,
      "learning_rate": 9.885579902457916e-06,
      "loss": 0.0918,
      "step": 1502
    },
    {
      "epoch": 0.07761425251742834,
      "grad_norm": 9.783391952514648,
      "learning_rate": 9.88540555293254e-06,
      "loss": 0.0817,
      "step": 1503
    },
    {
      "epoch": 0.07766589207332816,
      "grad_norm": 5.2097086906433105,
      "learning_rate": 9.885231072214112e-06,
      "loss": 0.0765,
      "step": 1504
    },
    {
      "epoch": 0.077717531629228,
      "grad_norm": 8.109561920166016,
      "learning_rate": 9.885056460307318e-06,
      "loss": 0.0922,
      "step": 1505
    },
    {
      "epoch": 0.07776917118512781,
      "grad_norm": 4.726545333862305,
      "learning_rate": 9.88488171721685e-06,
      "loss": 0.1188,
      "step": 1506
    },
    {
      "epoch": 0.07782081074102763,
      "grad_norm": 5.356204509735107,
      "learning_rate": 9.884706842947397e-06,
      "loss": 0.0228,
      "step": 1507
    },
    {
      "epoch": 0.07787245029692745,
      "grad_norm": 5.914389133453369,
      "learning_rate": 9.884531837503657e-06,
      "loss": 0.0569,
      "step": 1508
    },
    {
      "epoch": 0.07792408985282727,
      "grad_norm": 14.898913383483887,
      "learning_rate": 9.884356700890327e-06,
      "loss": 0.1631,
      "step": 1509
    },
    {
      "epoch": 0.07797572940872709,
      "grad_norm": 11.437728881835938,
      "learning_rate": 9.884181433112114e-06,
      "loss": 0.0769,
      "step": 1510
    },
    {
      "epoch": 0.0780273689646269,
      "grad_norm": 6.547587871551514,
      "learning_rate": 9.884006034173722e-06,
      "loss": 0.1166,
      "step": 1511
    },
    {
      "epoch": 0.07807900852052672,
      "grad_norm": 12.46527099609375,
      "learning_rate": 9.883830504079861e-06,
      "loss": 0.0743,
      "step": 1512
    },
    {
      "epoch": 0.07813064807642654,
      "grad_norm": 8.935065269470215,
      "learning_rate": 9.88365484283525e-06,
      "loss": 0.1014,
      "step": 1513
    },
    {
      "epoch": 0.07818228763232636,
      "grad_norm": 3.626516819000244,
      "learning_rate": 9.883479050444595e-06,
      "loss": 0.0805,
      "step": 1514
    },
    {
      "epoch": 0.07823392718822618,
      "grad_norm": 5.138033390045166,
      "learning_rate": 9.883303126912628e-06,
      "loss": 0.0839,
      "step": 1515
    },
    {
      "epoch": 0.078285566744126,
      "grad_norm": 3.235045909881592,
      "learning_rate": 9.883127072244068e-06,
      "loss": 0.0792,
      "step": 1516
    },
    {
      "epoch": 0.07833720630002582,
      "grad_norm": 5.060999393463135,
      "learning_rate": 9.882950886443644e-06,
      "loss": 0.0896,
      "step": 1517
    },
    {
      "epoch": 0.07838884585592563,
      "grad_norm": 5.156051158905029,
      "learning_rate": 9.882774569516085e-06,
      "loss": 0.0892,
      "step": 1518
    },
    {
      "epoch": 0.07844048541182545,
      "grad_norm": 7.339539051055908,
      "learning_rate": 9.88259812146613e-06,
      "loss": 0.0839,
      "step": 1519
    },
    {
      "epoch": 0.07849212496772527,
      "grad_norm": 5.8548903465271,
      "learning_rate": 9.882421542298512e-06,
      "loss": 0.0902,
      "step": 1520
    },
    {
      "epoch": 0.0785437645236251,
      "grad_norm": 2.01651930809021,
      "learning_rate": 9.882244832017979e-06,
      "loss": 0.0193,
      "step": 1521
    },
    {
      "epoch": 0.07859540407952492,
      "grad_norm": 5.731785297393799,
      "learning_rate": 9.882067990629272e-06,
      "loss": 0.043,
      "step": 1522
    },
    {
      "epoch": 0.07864704363542474,
      "grad_norm": 19.1389102935791,
      "learning_rate": 9.881891018137142e-06,
      "loss": 0.1049,
      "step": 1523
    },
    {
      "epoch": 0.07869868319132456,
      "grad_norm": 10.94605827331543,
      "learning_rate": 9.88171391454634e-06,
      "loss": 0.1506,
      "step": 1524
    },
    {
      "epoch": 0.07875032274722438,
      "grad_norm": 7.163397312164307,
      "learning_rate": 9.881536679861624e-06,
      "loss": 0.0593,
      "step": 1525
    },
    {
      "epoch": 0.0788019623031242,
      "grad_norm": 8.335151672363281,
      "learning_rate": 9.88135931408775e-06,
      "loss": 0.0602,
      "step": 1526
    },
    {
      "epoch": 0.07885360185902401,
      "grad_norm": 10.452160835266113,
      "learning_rate": 9.881181817229484e-06,
      "loss": 0.0924,
      "step": 1527
    },
    {
      "epoch": 0.07890524141492383,
      "grad_norm": 15.963120460510254,
      "learning_rate": 9.881004189291591e-06,
      "loss": 0.1389,
      "step": 1528
    },
    {
      "epoch": 0.07895688097082365,
      "grad_norm": 14.23031997680664,
      "learning_rate": 9.880826430278842e-06,
      "loss": 0.1316,
      "step": 1529
    },
    {
      "epoch": 0.07900852052672347,
      "grad_norm": 5.971617698669434,
      "learning_rate": 9.88064854019601e-06,
      "loss": 0.0575,
      "step": 1530
    },
    {
      "epoch": 0.07906016008262329,
      "grad_norm": 4.010227203369141,
      "learning_rate": 9.880470519047873e-06,
      "loss": 0.0301,
      "step": 1531
    },
    {
      "epoch": 0.0791117996385231,
      "grad_norm": 6.672082901000977,
      "learning_rate": 9.880292366839211e-06,
      "loss": 0.0432,
      "step": 1532
    },
    {
      "epoch": 0.07916343919442292,
      "grad_norm": 40.38309097290039,
      "learning_rate": 9.880114083574808e-06,
      "loss": 0.1073,
      "step": 1533
    },
    {
      "epoch": 0.07921507875032274,
      "grad_norm": 6.665011882781982,
      "learning_rate": 9.87993566925945e-06,
      "loss": 0.1049,
      "step": 1534
    },
    {
      "epoch": 0.07926671830622256,
      "grad_norm": 12.58816909790039,
      "learning_rate": 9.879757123897932e-06,
      "loss": 0.1606,
      "step": 1535
    },
    {
      "epoch": 0.07931835786212238,
      "grad_norm": 12.618379592895508,
      "learning_rate": 9.879578447495046e-06,
      "loss": 0.0772,
      "step": 1536
    },
    {
      "epoch": 0.07936999741802221,
      "grad_norm": 18.548120498657227,
      "learning_rate": 9.87939964005559e-06,
      "loss": 0.1291,
      "step": 1537
    },
    {
      "epoch": 0.07942163697392203,
      "grad_norm": 6.974266529083252,
      "learning_rate": 9.879220701584366e-06,
      "loss": 0.0749,
      "step": 1538
    },
    {
      "epoch": 0.07947327652982185,
      "grad_norm": 15.310351371765137,
      "learning_rate": 9.879041632086179e-06,
      "loss": 0.1175,
      "step": 1539
    },
    {
      "epoch": 0.07952491608572167,
      "grad_norm": 11.017727851867676,
      "learning_rate": 9.878862431565839e-06,
      "loss": 0.1389,
      "step": 1540
    },
    {
      "epoch": 0.07957655564162149,
      "grad_norm": 12.546806335449219,
      "learning_rate": 9.87868310002816e-06,
      "loss": 0.2236,
      "step": 1541
    },
    {
      "epoch": 0.0796281951975213,
      "grad_norm": 9.296097755432129,
      "learning_rate": 9.878503637477952e-06,
      "loss": 0.1376,
      "step": 1542
    },
    {
      "epoch": 0.07967983475342112,
      "grad_norm": 9.833792686462402,
      "learning_rate": 9.878324043920038e-06,
      "loss": 0.0861,
      "step": 1543
    },
    {
      "epoch": 0.07973147430932094,
      "grad_norm": 10.89438247680664,
      "learning_rate": 9.878144319359243e-06,
      "loss": 0.0787,
      "step": 1544
    },
    {
      "epoch": 0.07978311386522076,
      "grad_norm": 6.690000534057617,
      "learning_rate": 9.877964463800387e-06,
      "loss": 0.0888,
      "step": 1545
    },
    {
      "epoch": 0.07983475342112058,
      "grad_norm": 4.659882545471191,
      "learning_rate": 9.877784477248307e-06,
      "loss": 0.1144,
      "step": 1546
    },
    {
      "epoch": 0.0798863929770204,
      "grad_norm": 6.463596343994141,
      "learning_rate": 9.877604359707831e-06,
      "loss": 0.0763,
      "step": 1547
    },
    {
      "epoch": 0.07993803253292021,
      "grad_norm": 3.089244842529297,
      "learning_rate": 9.877424111183798e-06,
      "loss": 0.0758,
      "step": 1548
    },
    {
      "epoch": 0.07998967208882003,
      "grad_norm": 8.368441581726074,
      "learning_rate": 9.87724373168105e-06,
      "loss": 0.1061,
      "step": 1549
    },
    {
      "epoch": 0.08004131164471985,
      "grad_norm": 3.946207284927368,
      "learning_rate": 9.877063221204429e-06,
      "loss": 0.0636,
      "step": 1550
    },
    {
      "epoch": 0.08009295120061967,
      "grad_norm": 8.067086219787598,
      "learning_rate": 9.876882579758781e-06,
      "loss": 0.0912,
      "step": 1551
    },
    {
      "epoch": 0.08014459075651949,
      "grad_norm": 10.05697250366211,
      "learning_rate": 9.87670180734896e-06,
      "loss": 0.0816,
      "step": 1552
    },
    {
      "epoch": 0.08019623031241932,
      "grad_norm": 10.876663208007812,
      "learning_rate": 9.876520903979818e-06,
      "loss": 0.0746,
      "step": 1553
    },
    {
      "epoch": 0.08024786986831914,
      "grad_norm": 7.866598129272461,
      "learning_rate": 9.876339869656215e-06,
      "loss": 0.0913,
      "step": 1554
    },
    {
      "epoch": 0.08029950942421896,
      "grad_norm": 17.315135955810547,
      "learning_rate": 9.876158704383013e-06,
      "loss": 0.2292,
      "step": 1555
    },
    {
      "epoch": 0.08035114898011878,
      "grad_norm": 7.926328182220459,
      "learning_rate": 9.875977408165073e-06,
      "loss": 0.0821,
      "step": 1556
    },
    {
      "epoch": 0.0804027885360186,
      "grad_norm": 16.97101593017578,
      "learning_rate": 9.875795981007268e-06,
      "loss": 0.1805,
      "step": 1557
    },
    {
      "epoch": 0.08045442809191841,
      "grad_norm": 9.554069519042969,
      "learning_rate": 9.875614422914466e-06,
      "loss": 0.0677,
      "step": 1558
    },
    {
      "epoch": 0.08050606764781823,
      "grad_norm": 3.732455253601074,
      "learning_rate": 9.875432733891547e-06,
      "loss": 0.0753,
      "step": 1559
    },
    {
      "epoch": 0.08055770720371805,
      "grad_norm": 7.083679676055908,
      "learning_rate": 9.875250913943388e-06,
      "loss": 0.0622,
      "step": 1560
    },
    {
      "epoch": 0.08060934675961787,
      "grad_norm": 7.894657611846924,
      "learning_rate": 9.87506896307487e-06,
      "loss": 0.0829,
      "step": 1561
    },
    {
      "epoch": 0.08066098631551769,
      "grad_norm": 6.46352481842041,
      "learning_rate": 9.874886881290882e-06,
      "loss": 0.0862,
      "step": 1562
    },
    {
      "epoch": 0.0807126258714175,
      "grad_norm": 6.58678674697876,
      "learning_rate": 9.874704668596312e-06,
      "loss": 0.0817,
      "step": 1563
    },
    {
      "epoch": 0.08076426542731732,
      "grad_norm": 4.636165142059326,
      "learning_rate": 9.874522324996054e-06,
      "loss": 0.052,
      "step": 1564
    },
    {
      "epoch": 0.08081590498321714,
      "grad_norm": 17.254531860351562,
      "learning_rate": 9.874339850495003e-06,
      "loss": 0.1388,
      "step": 1565
    },
    {
      "epoch": 0.08086754453911696,
      "grad_norm": 6.634430885314941,
      "learning_rate": 9.874157245098062e-06,
      "loss": 0.1198,
      "step": 1566
    },
    {
      "epoch": 0.08091918409501678,
      "grad_norm": 4.608362674713135,
      "learning_rate": 9.873974508810131e-06,
      "loss": 0.0611,
      "step": 1567
    },
    {
      "epoch": 0.0809708236509166,
      "grad_norm": 9.408345222473145,
      "learning_rate": 9.87379164163612e-06,
      "loss": 0.1297,
      "step": 1568
    },
    {
      "epoch": 0.08102246320681643,
      "grad_norm": 4.371613025665283,
      "learning_rate": 9.873608643580939e-06,
      "loss": 0.093,
      "step": 1569
    },
    {
      "epoch": 0.08107410276271625,
      "grad_norm": 4.093642711639404,
      "learning_rate": 9.873425514649502e-06,
      "loss": 0.052,
      "step": 1570
    },
    {
      "epoch": 0.08112574231861606,
      "grad_norm": 16.078113555908203,
      "learning_rate": 9.873242254846727e-06,
      "loss": 0.1165,
      "step": 1571
    },
    {
      "epoch": 0.08117738187451588,
      "grad_norm": 4.727616786956787,
      "learning_rate": 9.873058864177537e-06,
      "loss": 0.0715,
      "step": 1572
    },
    {
      "epoch": 0.0812290214304157,
      "grad_norm": 6.562779426574707,
      "learning_rate": 9.872875342646853e-06,
      "loss": 0.0775,
      "step": 1573
    },
    {
      "epoch": 0.08128066098631552,
      "grad_norm": 11.935685157775879,
      "learning_rate": 9.872691690259603e-06,
      "loss": 0.1171,
      "step": 1574
    },
    {
      "epoch": 0.08133230054221534,
      "grad_norm": 6.318705081939697,
      "learning_rate": 9.872507907020725e-06,
      "loss": 0.0995,
      "step": 1575
    },
    {
      "epoch": 0.08138394009811516,
      "grad_norm": 4.278930187225342,
      "learning_rate": 9.872323992935148e-06,
      "loss": 0.0939,
      "step": 1576
    },
    {
      "epoch": 0.08143557965401498,
      "grad_norm": 8.477849960327148,
      "learning_rate": 9.872139948007816e-06,
      "loss": 0.0814,
      "step": 1577
    },
    {
      "epoch": 0.0814872192099148,
      "grad_norm": 5.852008819580078,
      "learning_rate": 9.871955772243667e-06,
      "loss": 0.0419,
      "step": 1578
    },
    {
      "epoch": 0.08153885876581461,
      "grad_norm": 3.035065174102783,
      "learning_rate": 9.871771465647648e-06,
      "loss": 0.0725,
      "step": 1579
    },
    {
      "epoch": 0.08159049832171443,
      "grad_norm": 9.51020622253418,
      "learning_rate": 9.87158702822471e-06,
      "loss": 0.0453,
      "step": 1580
    },
    {
      "epoch": 0.08164213787761425,
      "grad_norm": 5.289865970611572,
      "learning_rate": 9.871402459979804e-06,
      "loss": 0.0626,
      "step": 1581
    },
    {
      "epoch": 0.08169377743351407,
      "grad_norm": 13.279815673828125,
      "learning_rate": 9.871217760917889e-06,
      "loss": 0.1087,
      "step": 1582
    },
    {
      "epoch": 0.08174541698941389,
      "grad_norm": 17.11421775817871,
      "learning_rate": 9.871032931043921e-06,
      "loss": 0.1032,
      "step": 1583
    },
    {
      "epoch": 0.0817970565453137,
      "grad_norm": 7.371872901916504,
      "learning_rate": 9.870847970362866e-06,
      "loss": 0.1443,
      "step": 1584
    },
    {
      "epoch": 0.08184869610121354,
      "grad_norm": 9.716896057128906,
      "learning_rate": 9.870662878879692e-06,
      "loss": 0.1465,
      "step": 1585
    },
    {
      "epoch": 0.08190033565711335,
      "grad_norm": 8.965606689453125,
      "learning_rate": 9.870477656599366e-06,
      "loss": 0.0929,
      "step": 1586
    },
    {
      "epoch": 0.08195197521301317,
      "grad_norm": 2.738574743270874,
      "learning_rate": 9.870292303526865e-06,
      "loss": 0.0439,
      "step": 1587
    },
    {
      "epoch": 0.08200361476891299,
      "grad_norm": 3.5564348697662354,
      "learning_rate": 9.870106819667165e-06,
      "loss": 0.0824,
      "step": 1588
    },
    {
      "epoch": 0.08205525432481281,
      "grad_norm": 3.361772298812866,
      "learning_rate": 9.869921205025247e-06,
      "loss": 0.0997,
      "step": 1589
    },
    {
      "epoch": 0.08210689388071263,
      "grad_norm": 9.025992393493652,
      "learning_rate": 9.869735459606098e-06,
      "loss": 0.101,
      "step": 1590
    },
    {
      "epoch": 0.08215853343661245,
      "grad_norm": 11.976356506347656,
      "learning_rate": 9.869549583414701e-06,
      "loss": 0.1011,
      "step": 1591
    },
    {
      "epoch": 0.08221017299251226,
      "grad_norm": 3.784984588623047,
      "learning_rate": 9.869363576456053e-06,
      "loss": 0.0845,
      "step": 1592
    },
    {
      "epoch": 0.08226181254841208,
      "grad_norm": 8.159526824951172,
      "learning_rate": 9.869177438735146e-06,
      "loss": 0.1417,
      "step": 1593
    },
    {
      "epoch": 0.0823134521043119,
      "grad_norm": 5.973226547241211,
      "learning_rate": 9.868991170256979e-06,
      "loss": 0.0681,
      "step": 1594
    },
    {
      "epoch": 0.08236509166021172,
      "grad_norm": 4.259990692138672,
      "learning_rate": 9.868804771026554e-06,
      "loss": 0.037,
      "step": 1595
    },
    {
      "epoch": 0.08241673121611154,
      "grad_norm": 4.798079490661621,
      "learning_rate": 9.868618241048875e-06,
      "loss": 0.0642,
      "step": 1596
    },
    {
      "epoch": 0.08246837077201136,
      "grad_norm": 3.1189217567443848,
      "learning_rate": 9.868431580328955e-06,
      "loss": 0.0785,
      "step": 1597
    },
    {
      "epoch": 0.08252001032791118,
      "grad_norm": 4.6578898429870605,
      "learning_rate": 9.868244788871803e-06,
      "loss": 0.0482,
      "step": 1598
    },
    {
      "epoch": 0.082571649883811,
      "grad_norm": 3.2252321243286133,
      "learning_rate": 9.868057866682438e-06,
      "loss": 0.0352,
      "step": 1599
    },
    {
      "epoch": 0.08262328943971081,
      "grad_norm": 2.9115312099456787,
      "learning_rate": 9.867870813765878e-06,
      "loss": 0.042,
      "step": 1600
    },
    {
      "epoch": 0.08267492899561064,
      "grad_norm": 15.538789749145508,
      "learning_rate": 9.867683630127146e-06,
      "loss": 0.1261,
      "step": 1601
    },
    {
      "epoch": 0.08272656855151046,
      "grad_norm": 6.348582744598389,
      "learning_rate": 9.86749631577127e-06,
      "loss": 0.0545,
      "step": 1602
    },
    {
      "epoch": 0.08277820810741028,
      "grad_norm": 4.878490447998047,
      "learning_rate": 9.867308870703278e-06,
      "loss": 0.0464,
      "step": 1603
    },
    {
      "epoch": 0.0828298476633101,
      "grad_norm": 27.70543098449707,
      "learning_rate": 9.867121294928204e-06,
      "loss": 0.2535,
      "step": 1604
    },
    {
      "epoch": 0.08288148721920992,
      "grad_norm": 11.833864212036133,
      "learning_rate": 9.866933588451087e-06,
      "loss": 0.1646,
      "step": 1605
    },
    {
      "epoch": 0.08293312677510974,
      "grad_norm": 10.520862579345703,
      "learning_rate": 9.866745751276967e-06,
      "loss": 0.1013,
      "step": 1606
    },
    {
      "epoch": 0.08298476633100955,
      "grad_norm": 14.20620059967041,
      "learning_rate": 9.866557783410887e-06,
      "loss": 0.1614,
      "step": 1607
    },
    {
      "epoch": 0.08303640588690937,
      "grad_norm": 5.492526054382324,
      "learning_rate": 9.866369684857894e-06,
      "loss": 0.0616,
      "step": 1608
    },
    {
      "epoch": 0.08308804544280919,
      "grad_norm": 2.560492753982544,
      "learning_rate": 9.866181455623043e-06,
      "loss": 0.0435,
      "step": 1609
    },
    {
      "epoch": 0.08313968499870901,
      "grad_norm": 9.104405403137207,
      "learning_rate": 9.865993095711387e-06,
      "loss": 0.1313,
      "step": 1610
    },
    {
      "epoch": 0.08319132455460883,
      "grad_norm": 8.874317169189453,
      "learning_rate": 9.865804605127982e-06,
      "loss": 0.0587,
      "step": 1611
    },
    {
      "epoch": 0.08324296411050865,
      "grad_norm": 4.579291820526123,
      "learning_rate": 9.865615983877891e-06,
      "loss": 0.1066,
      "step": 1612
    },
    {
      "epoch": 0.08329460366640846,
      "grad_norm": 17.396869659423828,
      "learning_rate": 9.86542723196618e-06,
      "loss": 0.1067,
      "step": 1613
    },
    {
      "epoch": 0.08334624322230828,
      "grad_norm": 9.297853469848633,
      "learning_rate": 9.865238349397918e-06,
      "loss": 0.0907,
      "step": 1614
    },
    {
      "epoch": 0.0833978827782081,
      "grad_norm": 10.54130744934082,
      "learning_rate": 9.865049336178178e-06,
      "loss": 0.1266,
      "step": 1615
    },
    {
      "epoch": 0.08344952233410792,
      "grad_norm": 29.424470901489258,
      "learning_rate": 9.864860192312034e-06,
      "loss": 0.4209,
      "step": 1616
    },
    {
      "epoch": 0.08350116189000775,
      "grad_norm": 7.045882225036621,
      "learning_rate": 9.864670917804564e-06,
      "loss": 0.0698,
      "step": 1617
    },
    {
      "epoch": 0.08355280144590757,
      "grad_norm": 7.579276084899902,
      "learning_rate": 9.864481512660853e-06,
      "loss": 0.0968,
      "step": 1618
    },
    {
      "epoch": 0.08360444100180739,
      "grad_norm": 14.35197639465332,
      "learning_rate": 9.864291976885988e-06,
      "loss": 0.1874,
      "step": 1619
    },
    {
      "epoch": 0.08365608055770721,
      "grad_norm": 6.296283721923828,
      "learning_rate": 9.864102310485058e-06,
      "loss": 0.0977,
      "step": 1620
    },
    {
      "epoch": 0.08370772011360703,
      "grad_norm": 5.095728397369385,
      "learning_rate": 9.863912513463155e-06,
      "loss": 0.0817,
      "step": 1621
    },
    {
      "epoch": 0.08375935966950684,
      "grad_norm": 8.2838716506958,
      "learning_rate": 9.863722585825376e-06,
      "loss": 0.0659,
      "step": 1622
    },
    {
      "epoch": 0.08381099922540666,
      "grad_norm": 9.795034408569336,
      "learning_rate": 9.863532527576823e-06,
      "loss": 0.17,
      "step": 1623
    },
    {
      "epoch": 0.08386263878130648,
      "grad_norm": 14.222555160522461,
      "learning_rate": 9.8633423387226e-06,
      "loss": 0.1143,
      "step": 1624
    },
    {
      "epoch": 0.0839142783372063,
      "grad_norm": 10.197482109069824,
      "learning_rate": 9.863152019267812e-06,
      "loss": 0.1293,
      "step": 1625
    },
    {
      "epoch": 0.08396591789310612,
      "grad_norm": 2.7547242641448975,
      "learning_rate": 9.86296156921757e-06,
      "loss": 0.1007,
      "step": 1626
    },
    {
      "epoch": 0.08401755744900594,
      "grad_norm": 5.112005233764648,
      "learning_rate": 9.862770988576992e-06,
      "loss": 0.0954,
      "step": 1627
    },
    {
      "epoch": 0.08406919700490575,
      "grad_norm": 2.4318811893463135,
      "learning_rate": 9.862580277351192e-06,
      "loss": 0.0618,
      "step": 1628
    },
    {
      "epoch": 0.08412083656080557,
      "grad_norm": 6.29991340637207,
      "learning_rate": 9.862389435545294e-06,
      "loss": 0.0976,
      "step": 1629
    },
    {
      "epoch": 0.08417247611670539,
      "grad_norm": 3.1610946655273438,
      "learning_rate": 9.86219846316442e-06,
      "loss": 0.05,
      "step": 1630
    },
    {
      "epoch": 0.08422411567260521,
      "grad_norm": 3.029236316680908,
      "learning_rate": 9.8620073602137e-06,
      "loss": 0.0938,
      "step": 1631
    },
    {
      "epoch": 0.08427575522850503,
      "grad_norm": 4.5872578620910645,
      "learning_rate": 9.861816126698268e-06,
      "loss": 0.0509,
      "step": 1632
    },
    {
      "epoch": 0.08432739478440486,
      "grad_norm": 4.084672451019287,
      "learning_rate": 9.861624762623256e-06,
      "loss": 0.0751,
      "step": 1633
    },
    {
      "epoch": 0.08437903434030468,
      "grad_norm": 20.272663116455078,
      "learning_rate": 9.861433267993805e-06,
      "loss": 0.1505,
      "step": 1634
    },
    {
      "epoch": 0.0844306738962045,
      "grad_norm": 9.579610824584961,
      "learning_rate": 9.861241642815056e-06,
      "loss": 0.055,
      "step": 1635
    },
    {
      "epoch": 0.08448231345210432,
      "grad_norm": 16.67690086364746,
      "learning_rate": 9.861049887092156e-06,
      "loss": 0.1577,
      "step": 1636
    },
    {
      "epoch": 0.08453395300800413,
      "grad_norm": 28.070781707763672,
      "learning_rate": 9.860858000830254e-06,
      "loss": 0.2432,
      "step": 1637
    },
    {
      "epoch": 0.08458559256390395,
      "grad_norm": 8.283512115478516,
      "learning_rate": 9.8606659840345e-06,
      "loss": 0.0909,
      "step": 1638
    },
    {
      "epoch": 0.08463723211980377,
      "grad_norm": 4.815789222717285,
      "learning_rate": 9.860473836710058e-06,
      "loss": 0.056,
      "step": 1639
    },
    {
      "epoch": 0.08468887167570359,
      "grad_norm": 9.729043006896973,
      "learning_rate": 9.86028155886208e-06,
      "loss": 0.1248,
      "step": 1640
    },
    {
      "epoch": 0.08474051123160341,
      "grad_norm": 5.7178239822387695,
      "learning_rate": 9.860089150495734e-06,
      "loss": 0.0804,
      "step": 1641
    },
    {
      "epoch": 0.08479215078750323,
      "grad_norm": 11.101826667785645,
      "learning_rate": 9.859896611616187e-06,
      "loss": 0.0563,
      "step": 1642
    },
    {
      "epoch": 0.08484379034340304,
      "grad_norm": 14.085371017456055,
      "learning_rate": 9.859703942228605e-06,
      "loss": 0.1109,
      "step": 1643
    },
    {
      "epoch": 0.08489542989930286,
      "grad_norm": 9.614492416381836,
      "learning_rate": 9.859511142338167e-06,
      "loss": 0.1172,
      "step": 1644
    },
    {
      "epoch": 0.08494706945520268,
      "grad_norm": 16.946683883666992,
      "learning_rate": 9.859318211950049e-06,
      "loss": 0.2637,
      "step": 1645
    },
    {
      "epoch": 0.0849987090111025,
      "grad_norm": 10.54627799987793,
      "learning_rate": 9.859125151069431e-06,
      "loss": 0.083,
      "step": 1646
    },
    {
      "epoch": 0.08505034856700232,
      "grad_norm": 5.6411614418029785,
      "learning_rate": 9.858931959701498e-06,
      "loss": 0.0975,
      "step": 1647
    },
    {
      "epoch": 0.08510198812290214,
      "grad_norm": 3.5365278720855713,
      "learning_rate": 9.858738637851437e-06,
      "loss": 0.0208,
      "step": 1648
    },
    {
      "epoch": 0.08515362767880197,
      "grad_norm": 6.792642116546631,
      "learning_rate": 9.858545185524443e-06,
      "loss": 0.0664,
      "step": 1649
    },
    {
      "epoch": 0.08520526723470179,
      "grad_norm": 5.6763505935668945,
      "learning_rate": 9.858351602725706e-06,
      "loss": 0.0722,
      "step": 1650
    },
    {
      "epoch": 0.0852569067906016,
      "grad_norm": 3.30926251411438,
      "learning_rate": 9.858157889460427e-06,
      "loss": 0.0584,
      "step": 1651
    },
    {
      "epoch": 0.08530854634650142,
      "grad_norm": 2.4739391803741455,
      "learning_rate": 9.857964045733807e-06,
      "loss": 0.0654,
      "step": 1652
    },
    {
      "epoch": 0.08536018590240124,
      "grad_norm": 4.001121520996094,
      "learning_rate": 9.857770071551055e-06,
      "loss": 0.0914,
      "step": 1653
    },
    {
      "epoch": 0.08541182545830106,
      "grad_norm": 9.836386680603027,
      "learning_rate": 9.857575966917376e-06,
      "loss": 0.0784,
      "step": 1654
    },
    {
      "epoch": 0.08546346501420088,
      "grad_norm": 18.980026245117188,
      "learning_rate": 9.857381731837986e-06,
      "loss": 0.1612,
      "step": 1655
    },
    {
      "epoch": 0.0855151045701007,
      "grad_norm": 30.142732620239258,
      "learning_rate": 9.857187366318096e-06,
      "loss": 0.1261,
      "step": 1656
    },
    {
      "epoch": 0.08556674412600052,
      "grad_norm": 14.156326293945312,
      "learning_rate": 9.856992870362929e-06,
      "loss": 0.1972,
      "step": 1657
    },
    {
      "epoch": 0.08561838368190033,
      "grad_norm": 14.704541206359863,
      "learning_rate": 9.856798243977708e-06,
      "loss": 0.1786,
      "step": 1658
    },
    {
      "epoch": 0.08567002323780015,
      "grad_norm": 3.36936092376709,
      "learning_rate": 9.85660348716766e-06,
      "loss": 0.0773,
      "step": 1659
    },
    {
      "epoch": 0.08572166279369997,
      "grad_norm": 8.380145072937012,
      "learning_rate": 9.856408599938012e-06,
      "loss": 0.0778,
      "step": 1660
    },
    {
      "epoch": 0.08577330234959979,
      "grad_norm": 5.053802013397217,
      "learning_rate": 9.856213582294e-06,
      "loss": 0.0804,
      "step": 1661
    },
    {
      "epoch": 0.08582494190549961,
      "grad_norm": 5.100456714630127,
      "learning_rate": 9.856018434240862e-06,
      "loss": 0.1094,
      "step": 1662
    },
    {
      "epoch": 0.08587658146139943,
      "grad_norm": 17.73405647277832,
      "learning_rate": 9.855823155783836e-06,
      "loss": 0.1772,
      "step": 1663
    },
    {
      "epoch": 0.08592822101729924,
      "grad_norm": 7.981858253479004,
      "learning_rate": 9.855627746928167e-06,
      "loss": 0.069,
      "step": 1664
    },
    {
      "epoch": 0.08597986057319908,
      "grad_norm": 13.957771301269531,
      "learning_rate": 9.855432207679102e-06,
      "loss": 0.1677,
      "step": 1665
    },
    {
      "epoch": 0.0860315001290989,
      "grad_norm": 3.574620008468628,
      "learning_rate": 9.855236538041893e-06,
      "loss": 0.1197,
      "step": 1666
    },
    {
      "epoch": 0.08608313968499871,
      "grad_norm": 3.4857044219970703,
      "learning_rate": 9.855040738021795e-06,
      "loss": 0.0863,
      "step": 1667
    },
    {
      "epoch": 0.08613477924089853,
      "grad_norm": 9.835796356201172,
      "learning_rate": 9.854844807624063e-06,
      "loss": 0.1218,
      "step": 1668
    },
    {
      "epoch": 0.08618641879679835,
      "grad_norm": 3.499135971069336,
      "learning_rate": 9.854648746853963e-06,
      "loss": 0.0803,
      "step": 1669
    },
    {
      "epoch": 0.08623805835269817,
      "grad_norm": 6.2881879806518555,
      "learning_rate": 9.854452555716755e-06,
      "loss": 0.0774,
      "step": 1670
    },
    {
      "epoch": 0.08628969790859799,
      "grad_norm": 8.11571979522705,
      "learning_rate": 9.854256234217713e-06,
      "loss": 0.0641,
      "step": 1671
    },
    {
      "epoch": 0.0863413374644978,
      "grad_norm": 4.20208740234375,
      "learning_rate": 9.854059782362104e-06,
      "loss": 0.0957,
      "step": 1672
    },
    {
      "epoch": 0.08639297702039762,
      "grad_norm": 10.813671112060547,
      "learning_rate": 9.853863200155206e-06,
      "loss": 0.1106,
      "step": 1673
    },
    {
      "epoch": 0.08644461657629744,
      "grad_norm": 4.746212005615234,
      "learning_rate": 9.853666487602301e-06,
      "loss": 0.1031,
      "step": 1674
    },
    {
      "epoch": 0.08649625613219726,
      "grad_norm": 3.310375928878784,
      "learning_rate": 9.853469644708665e-06,
      "loss": 0.0468,
      "step": 1675
    },
    {
      "epoch": 0.08654789568809708,
      "grad_norm": 8.869812965393066,
      "learning_rate": 9.853272671479588e-06,
      "loss": 0.112,
      "step": 1676
    },
    {
      "epoch": 0.0865995352439969,
      "grad_norm": 12.56878662109375,
      "learning_rate": 9.85307556792036e-06,
      "loss": 0.1052,
      "step": 1677
    },
    {
      "epoch": 0.08665117479989672,
      "grad_norm": 21.73412322998047,
      "learning_rate": 9.852878334036274e-06,
      "loss": 0.1097,
      "step": 1678
    },
    {
      "epoch": 0.08670281435579653,
      "grad_norm": 12.53563117980957,
      "learning_rate": 9.852680969832623e-06,
      "loss": 0.1591,
      "step": 1679
    },
    {
      "epoch": 0.08675445391169635,
      "grad_norm": 7.018848419189453,
      "learning_rate": 9.85248347531471e-06,
      "loss": 0.0966,
      "step": 1680
    },
    {
      "epoch": 0.08680609346759618,
      "grad_norm": 12.492026329040527,
      "learning_rate": 9.852285850487838e-06,
      "loss": 0.1928,
      "step": 1681
    },
    {
      "epoch": 0.086857733023496,
      "grad_norm": 24.83223533630371,
      "learning_rate": 9.852088095357315e-06,
      "loss": 0.3549,
      "step": 1682
    },
    {
      "epoch": 0.08690937257939582,
      "grad_norm": 7.454451560974121,
      "learning_rate": 9.85189020992845e-06,
      "loss": 0.0387,
      "step": 1683
    },
    {
      "epoch": 0.08696101213529564,
      "grad_norm": 23.56189727783203,
      "learning_rate": 9.851692194206556e-06,
      "loss": 0.1833,
      "step": 1684
    },
    {
      "epoch": 0.08701265169119546,
      "grad_norm": 4.013763427734375,
      "learning_rate": 9.851494048196956e-06,
      "loss": 0.0473,
      "step": 1685
    },
    {
      "epoch": 0.08706429124709528,
      "grad_norm": 10.731250762939453,
      "learning_rate": 9.851295771904965e-06,
      "loss": 0.1122,
      "step": 1686
    },
    {
      "epoch": 0.0871159308029951,
      "grad_norm": 5.049229621887207,
      "learning_rate": 9.851097365335908e-06,
      "loss": 0.0979,
      "step": 1687
    },
    {
      "epoch": 0.08716757035889491,
      "grad_norm": 12.049827575683594,
      "learning_rate": 9.850898828495118e-06,
      "loss": 0.1102,
      "step": 1688
    },
    {
      "epoch": 0.08721920991479473,
      "grad_norm": 5.580020904541016,
      "learning_rate": 9.850700161387922e-06,
      "loss": 0.0765,
      "step": 1689
    },
    {
      "epoch": 0.08727084947069455,
      "grad_norm": 10.15054988861084,
      "learning_rate": 9.850501364019656e-06,
      "loss": 0.1166,
      "step": 1690
    },
    {
      "epoch": 0.08732248902659437,
      "grad_norm": 3.79803466796875,
      "learning_rate": 9.85030243639566e-06,
      "loss": 0.0684,
      "step": 1691
    },
    {
      "epoch": 0.08737412858249419,
      "grad_norm": 8.494483947753906,
      "learning_rate": 9.850103378521274e-06,
      "loss": 0.0965,
      "step": 1692
    },
    {
      "epoch": 0.087425768138394,
      "grad_norm": 2.7579357624053955,
      "learning_rate": 9.849904190401844e-06,
      "loss": 0.0782,
      "step": 1693
    },
    {
      "epoch": 0.08747740769429382,
      "grad_norm": 8.46857738494873,
      "learning_rate": 9.84970487204272e-06,
      "loss": 0.1061,
      "step": 1694
    },
    {
      "epoch": 0.08752904725019364,
      "grad_norm": 4.205897331237793,
      "learning_rate": 9.849505423449253e-06,
      "loss": 0.0779,
      "step": 1695
    },
    {
      "epoch": 0.08758068680609346,
      "grad_norm": 5.345470905303955,
      "learning_rate": 9.8493058446268e-06,
      "loss": 0.079,
      "step": 1696
    },
    {
      "epoch": 0.08763232636199329,
      "grad_norm": 6.173393249511719,
      "learning_rate": 9.849106135580722e-06,
      "loss": 0.0433,
      "step": 1697
    },
    {
      "epoch": 0.08768396591789311,
      "grad_norm": 3.3123373985290527,
      "learning_rate": 9.848906296316378e-06,
      "loss": 0.0355,
      "step": 1698
    },
    {
      "epoch": 0.08773560547379293,
      "grad_norm": 6.623047351837158,
      "learning_rate": 9.848706326839139e-06,
      "loss": 0.0523,
      "step": 1699
    },
    {
      "epoch": 0.08778724502969275,
      "grad_norm": 6.054687023162842,
      "learning_rate": 9.84850622715437e-06,
      "loss": 0.0746,
      "step": 1700
    },
    {
      "epoch": 0.08783888458559257,
      "grad_norm": 7.530181407928467,
      "learning_rate": 9.848305997267451e-06,
      "loss": 0.0919,
      "step": 1701
    },
    {
      "epoch": 0.08789052414149238,
      "grad_norm": 5.139504909515381,
      "learning_rate": 9.848105637183753e-06,
      "loss": 0.0905,
      "step": 1702
    },
    {
      "epoch": 0.0879421636973922,
      "grad_norm": 3.1062676906585693,
      "learning_rate": 9.84790514690866e-06,
      "loss": 0.078,
      "step": 1703
    },
    {
      "epoch": 0.08799380325329202,
      "grad_norm": 8.131316184997559,
      "learning_rate": 9.847704526447555e-06,
      "loss": 0.1361,
      "step": 1704
    },
    {
      "epoch": 0.08804544280919184,
      "grad_norm": 11.234710693359375,
      "learning_rate": 9.847503775805825e-06,
      "loss": 0.1366,
      "step": 1705
    },
    {
      "epoch": 0.08809708236509166,
      "grad_norm": 8.701875686645508,
      "learning_rate": 9.847302894988861e-06,
      "loss": 0.0892,
      "step": 1706
    },
    {
      "epoch": 0.08814872192099148,
      "grad_norm": 17.83285903930664,
      "learning_rate": 9.847101884002056e-06,
      "loss": 0.1677,
      "step": 1707
    },
    {
      "epoch": 0.0882003614768913,
      "grad_norm": 15.233487129211426,
      "learning_rate": 9.846900742850814e-06,
      "loss": 0.0572,
      "step": 1708
    },
    {
      "epoch": 0.08825200103279111,
      "grad_norm": 8.639863014221191,
      "learning_rate": 9.846699471540529e-06,
      "loss": 0.0505,
      "step": 1709
    },
    {
      "epoch": 0.08830364058869093,
      "grad_norm": 7.126918315887451,
      "learning_rate": 9.846498070076609e-06,
      "loss": 0.1109,
      "step": 1710
    },
    {
      "epoch": 0.08835528014459075,
      "grad_norm": 12.139455795288086,
      "learning_rate": 9.846296538464464e-06,
      "loss": 0.1326,
      "step": 1711
    },
    {
      "epoch": 0.08840691970049058,
      "grad_norm": 3.6967175006866455,
      "learning_rate": 9.846094876709505e-06,
      "loss": 0.0744,
      "step": 1712
    },
    {
      "epoch": 0.0884585592563904,
      "grad_norm": 6.022879600524902,
      "learning_rate": 9.845893084817144e-06,
      "loss": 0.1085,
      "step": 1713
    },
    {
      "epoch": 0.08851019881229022,
      "grad_norm": 7.377680778503418,
      "learning_rate": 9.845691162792805e-06,
      "loss": 0.1231,
      "step": 1714
    },
    {
      "epoch": 0.08856183836819004,
      "grad_norm": 9.667302131652832,
      "learning_rate": 9.845489110641908e-06,
      "loss": 0.0708,
      "step": 1715
    },
    {
      "epoch": 0.08861347792408986,
      "grad_norm": 7.058122158050537,
      "learning_rate": 9.845286928369881e-06,
      "loss": 0.0621,
      "step": 1716
    },
    {
      "epoch": 0.08866511747998967,
      "grad_norm": 5.288403511047363,
      "learning_rate": 9.84508461598215e-06,
      "loss": 0.1303,
      "step": 1717
    },
    {
      "epoch": 0.08871675703588949,
      "grad_norm": 4.258204460144043,
      "learning_rate": 9.844882173484149e-06,
      "loss": 0.0583,
      "step": 1718
    },
    {
      "epoch": 0.08876839659178931,
      "grad_norm": 9.808914184570312,
      "learning_rate": 9.844679600881315e-06,
      "loss": 0.0933,
      "step": 1719
    },
    {
      "epoch": 0.08882003614768913,
      "grad_norm": 5.588613510131836,
      "learning_rate": 9.84447689817909e-06,
      "loss": 0.0595,
      "step": 1720
    },
    {
      "epoch": 0.08887167570358895,
      "grad_norm": 13.87211799621582,
      "learning_rate": 9.844274065382912e-06,
      "loss": 0.1792,
      "step": 1721
    },
    {
      "epoch": 0.08892331525948877,
      "grad_norm": 3.135634183883667,
      "learning_rate": 9.844071102498234e-06,
      "loss": 0.0801,
      "step": 1722
    },
    {
      "epoch": 0.08897495481538858,
      "grad_norm": 4.6009626388549805,
      "learning_rate": 9.843868009530503e-06,
      "loss": 0.0839,
      "step": 1723
    },
    {
      "epoch": 0.0890265943712884,
      "grad_norm": 2.228538990020752,
      "learning_rate": 9.843664786485172e-06,
      "loss": 0.0275,
      "step": 1724
    },
    {
      "epoch": 0.08907823392718822,
      "grad_norm": 4.8577656745910645,
      "learning_rate": 9.843461433367702e-06,
      "loss": 0.0865,
      "step": 1725
    },
    {
      "epoch": 0.08912987348308804,
      "grad_norm": 13.217398643493652,
      "learning_rate": 9.84325795018355e-06,
      "loss": 0.1171,
      "step": 1726
    },
    {
      "epoch": 0.08918151303898786,
      "grad_norm": 28.307193756103516,
      "learning_rate": 9.843054336938183e-06,
      "loss": 0.1062,
      "step": 1727
    },
    {
      "epoch": 0.08923315259488769,
      "grad_norm": 13.095776557922363,
      "learning_rate": 9.842850593637068e-06,
      "loss": 0.132,
      "step": 1728
    },
    {
      "epoch": 0.08928479215078751,
      "grad_norm": 29.79486846923828,
      "learning_rate": 9.842646720285675e-06,
      "loss": 0.1165,
      "step": 1729
    },
    {
      "epoch": 0.08933643170668733,
      "grad_norm": 5.077782154083252,
      "learning_rate": 9.84244271688948e-06,
      "loss": 0.0968,
      "step": 1730
    },
    {
      "epoch": 0.08938807126258715,
      "grad_norm": 14.302273750305176,
      "learning_rate": 9.842238583453963e-06,
      "loss": 0.1436,
      "step": 1731
    },
    {
      "epoch": 0.08943971081848696,
      "grad_norm": 6.998988628387451,
      "learning_rate": 9.842034319984602e-06,
      "loss": 0.1135,
      "step": 1732
    },
    {
      "epoch": 0.08949135037438678,
      "grad_norm": 6.453922271728516,
      "learning_rate": 9.841829926486886e-06,
      "loss": 0.0878,
      "step": 1733
    },
    {
      "epoch": 0.0895429899302866,
      "grad_norm": 5.74137020111084,
      "learning_rate": 9.841625402966301e-06,
      "loss": 0.0771,
      "step": 1734
    },
    {
      "epoch": 0.08959462948618642,
      "grad_norm": 9.513408660888672,
      "learning_rate": 9.841420749428341e-06,
      "loss": 0.0918,
      "step": 1735
    },
    {
      "epoch": 0.08964626904208624,
      "grad_norm": 5.537482738494873,
      "learning_rate": 9.841215965878503e-06,
      "loss": 0.089,
      "step": 1736
    },
    {
      "epoch": 0.08969790859798606,
      "grad_norm": 3.6365134716033936,
      "learning_rate": 9.841011052322283e-06,
      "loss": 0.0598,
      "step": 1737
    },
    {
      "epoch": 0.08974954815388587,
      "grad_norm": 5.558864593505859,
      "learning_rate": 9.840806008765185e-06,
      "loss": 0.0622,
      "step": 1738
    },
    {
      "epoch": 0.0898011877097857,
      "grad_norm": 3.6402924060821533,
      "learning_rate": 9.840600835212716e-06,
      "loss": 0.0593,
      "step": 1739
    },
    {
      "epoch": 0.08985282726568551,
      "grad_norm": 5.003504276275635,
      "learning_rate": 9.840395531670387e-06,
      "loss": 0.0841,
      "step": 1740
    },
    {
      "epoch": 0.08990446682158533,
      "grad_norm": 11.903522491455078,
      "learning_rate": 9.840190098143709e-06,
      "loss": 0.1028,
      "step": 1741
    },
    {
      "epoch": 0.08995610637748515,
      "grad_norm": 2.954280376434326,
      "learning_rate": 9.839984534638198e-06,
      "loss": 0.059,
      "step": 1742
    },
    {
      "epoch": 0.09000774593338497,
      "grad_norm": 16.52904510498047,
      "learning_rate": 9.839778841159377e-06,
      "loss": 0.0998,
      "step": 1743
    },
    {
      "epoch": 0.0900593854892848,
      "grad_norm": 3.3990962505340576,
      "learning_rate": 9.839573017712768e-06,
      "loss": 0.0391,
      "step": 1744
    },
    {
      "epoch": 0.09011102504518462,
      "grad_norm": 14.730329513549805,
      "learning_rate": 9.839367064303898e-06,
      "loss": 0.1564,
      "step": 1745
    },
    {
      "epoch": 0.09016266460108444,
      "grad_norm": 13.882684707641602,
      "learning_rate": 9.839160980938297e-06,
      "loss": 0.1169,
      "step": 1746
    },
    {
      "epoch": 0.09021430415698425,
      "grad_norm": 5.362858772277832,
      "learning_rate": 9.838954767621502e-06,
      "loss": 0.0416,
      "step": 1747
    },
    {
      "epoch": 0.09026594371288407,
      "grad_norm": 8.111239433288574,
      "learning_rate": 9.838748424359049e-06,
      "loss": 0.1211,
      "step": 1748
    },
    {
      "epoch": 0.09031758326878389,
      "grad_norm": 5.569311141967773,
      "learning_rate": 9.838541951156478e-06,
      "loss": 0.0943,
      "step": 1749
    },
    {
      "epoch": 0.09036922282468371,
      "grad_norm": 9.021713256835938,
      "learning_rate": 9.838335348019334e-06,
      "loss": 0.1077,
      "step": 1750
    },
    {
      "epoch": 0.09042086238058353,
      "grad_norm": 17.759279251098633,
      "learning_rate": 9.838128614953168e-06,
      "loss": 0.1499,
      "step": 1751
    },
    {
      "epoch": 0.09047250193648335,
      "grad_norm": 10.977428436279297,
      "learning_rate": 9.837921751963528e-06,
      "loss": 0.082,
      "step": 1752
    },
    {
      "epoch": 0.09052414149238316,
      "grad_norm": 6.072782039642334,
      "learning_rate": 9.837714759055972e-06,
      "loss": 0.1243,
      "step": 1753
    },
    {
      "epoch": 0.09057578104828298,
      "grad_norm": 5.877910137176514,
      "learning_rate": 9.837507636236058e-06,
      "loss": 0.0546,
      "step": 1754
    },
    {
      "epoch": 0.0906274206041828,
      "grad_norm": 4.969249725341797,
      "learning_rate": 9.837300383509346e-06,
      "loss": 0.0871,
      "step": 1755
    },
    {
      "epoch": 0.09067906016008262,
      "grad_norm": 20.567020416259766,
      "learning_rate": 9.837093000881404e-06,
      "loss": 0.0903,
      "step": 1756
    },
    {
      "epoch": 0.09073069971598244,
      "grad_norm": 9.79025936126709,
      "learning_rate": 9.836885488357799e-06,
      "loss": 0.1175,
      "step": 1757
    },
    {
      "epoch": 0.09078233927188226,
      "grad_norm": 11.21975040435791,
      "learning_rate": 9.836677845944104e-06,
      "loss": 0.0767,
      "step": 1758
    },
    {
      "epoch": 0.09083397882778207,
      "grad_norm": 7.871028900146484,
      "learning_rate": 9.836470073645896e-06,
      "loss": 0.1113,
      "step": 1759
    },
    {
      "epoch": 0.09088561838368191,
      "grad_norm": 15.622942924499512,
      "learning_rate": 9.836262171468756e-06,
      "loss": 0.1447,
      "step": 1760
    },
    {
      "epoch": 0.09093725793958173,
      "grad_norm": 8.879650115966797,
      "learning_rate": 9.836054139418264e-06,
      "loss": 0.1049,
      "step": 1761
    },
    {
      "epoch": 0.09098889749548154,
      "grad_norm": 3.556478977203369,
      "learning_rate": 9.835845977500007e-06,
      "loss": 0.0614,
      "step": 1762
    },
    {
      "epoch": 0.09104053705138136,
      "grad_norm": 8.62171745300293,
      "learning_rate": 9.835637685719575e-06,
      "loss": 0.0612,
      "step": 1763
    },
    {
      "epoch": 0.09109217660728118,
      "grad_norm": 8.478828430175781,
      "learning_rate": 9.835429264082564e-06,
      "loss": 0.1066,
      "step": 1764
    },
    {
      "epoch": 0.091143816163181,
      "grad_norm": 15.574323654174805,
      "learning_rate": 9.835220712594568e-06,
      "loss": 0.1341,
      "step": 1765
    },
    {
      "epoch": 0.09119545571908082,
      "grad_norm": 2.755779504776001,
      "learning_rate": 9.83501203126119e-06,
      "loss": 0.042,
      "step": 1766
    },
    {
      "epoch": 0.09124709527498064,
      "grad_norm": 4.717486381530762,
      "learning_rate": 9.834803220088031e-06,
      "loss": 0.0895,
      "step": 1767
    },
    {
      "epoch": 0.09129873483088045,
      "grad_norm": 2.8277406692504883,
      "learning_rate": 9.834594279080701e-06,
      "loss": 0.08,
      "step": 1768
    },
    {
      "epoch": 0.09135037438678027,
      "grad_norm": 5.761186122894287,
      "learning_rate": 9.83438520824481e-06,
      "loss": 0.0871,
      "step": 1769
    },
    {
      "epoch": 0.09140201394268009,
      "grad_norm": 4.235811233520508,
      "learning_rate": 9.834176007585972e-06,
      "loss": 0.0543,
      "step": 1770
    },
    {
      "epoch": 0.09145365349857991,
      "grad_norm": 3.7856736183166504,
      "learning_rate": 9.833966677109805e-06,
      "loss": 0.0506,
      "step": 1771
    },
    {
      "epoch": 0.09150529305447973,
      "grad_norm": 8.421631813049316,
      "learning_rate": 9.83375721682193e-06,
      "loss": 0.0909,
      "step": 1772
    },
    {
      "epoch": 0.09155693261037955,
      "grad_norm": 4.535943984985352,
      "learning_rate": 9.833547626727972e-06,
      "loss": 0.0934,
      "step": 1773
    },
    {
      "epoch": 0.09160857216627936,
      "grad_norm": 10.975719451904297,
      "learning_rate": 9.833337906833562e-06,
      "loss": 0.1704,
      "step": 1774
    },
    {
      "epoch": 0.09166021172217918,
      "grad_norm": 9.396533012390137,
      "learning_rate": 9.833128057144328e-06,
      "loss": 0.0717,
      "step": 1775
    },
    {
      "epoch": 0.09171185127807902,
      "grad_norm": 11.612809181213379,
      "learning_rate": 9.832918077665907e-06,
      "loss": 0.1787,
      "step": 1776
    },
    {
      "epoch": 0.09176349083397883,
      "grad_norm": 4.072597026824951,
      "learning_rate": 9.832707968403938e-06,
      "loss": 0.0476,
      "step": 1777
    },
    {
      "epoch": 0.09181513038987865,
      "grad_norm": 6.925817012786865,
      "learning_rate": 9.832497729364063e-06,
      "loss": 0.0626,
      "step": 1778
    },
    {
      "epoch": 0.09186676994577847,
      "grad_norm": 4.788794994354248,
      "learning_rate": 9.832287360551929e-06,
      "loss": 0.0848,
      "step": 1779
    },
    {
      "epoch": 0.09191840950167829,
      "grad_norm": 15.535964012145996,
      "learning_rate": 9.832076861973182e-06,
      "loss": 0.1573,
      "step": 1780
    },
    {
      "epoch": 0.09197004905757811,
      "grad_norm": 4.139308452606201,
      "learning_rate": 9.831866233633478e-06,
      "loss": 0.0632,
      "step": 1781
    },
    {
      "epoch": 0.09202168861347793,
      "grad_norm": 11.172552108764648,
      "learning_rate": 9.83165547553847e-06,
      "loss": 0.1221,
      "step": 1782
    },
    {
      "epoch": 0.09207332816937774,
      "grad_norm": 6.416905403137207,
      "learning_rate": 9.831444587693823e-06,
      "loss": 0.0844,
      "step": 1783
    },
    {
      "epoch": 0.09212496772527756,
      "grad_norm": 6.871948719024658,
      "learning_rate": 9.831233570105195e-06,
      "loss": 0.1003,
      "step": 1784
    },
    {
      "epoch": 0.09217660728117738,
      "grad_norm": 4.996333599090576,
      "learning_rate": 9.831022422778255e-06,
      "loss": 0.1186,
      "step": 1785
    },
    {
      "epoch": 0.0922282468370772,
      "grad_norm": 3.5580170154571533,
      "learning_rate": 9.830811145718674e-06,
      "loss": 0.0592,
      "step": 1786
    },
    {
      "epoch": 0.09227988639297702,
      "grad_norm": 18.44939613342285,
      "learning_rate": 9.830599738932123e-06,
      "loss": 0.1163,
      "step": 1787
    },
    {
      "epoch": 0.09233152594887684,
      "grad_norm": 7.999111652374268,
      "learning_rate": 9.83038820242428e-06,
      "loss": 0.0916,
      "step": 1788
    },
    {
      "epoch": 0.09238316550477665,
      "grad_norm": 5.1933817863464355,
      "learning_rate": 9.830176536200826e-06,
      "loss": 0.0943,
      "step": 1789
    },
    {
      "epoch": 0.09243480506067647,
      "grad_norm": 3.4781298637390137,
      "learning_rate": 9.829964740267448e-06,
      "loss": 0.0677,
      "step": 1790
    },
    {
      "epoch": 0.09248644461657629,
      "grad_norm": 3.352573871612549,
      "learning_rate": 9.829752814629828e-06,
      "loss": 0.0852,
      "step": 1791
    },
    {
      "epoch": 0.09253808417247612,
      "grad_norm": 11.4541654586792,
      "learning_rate": 9.82954075929366e-06,
      "loss": 0.0975,
      "step": 1792
    },
    {
      "epoch": 0.09258972372837594,
      "grad_norm": 7.89384651184082,
      "learning_rate": 9.82932857426464e-06,
      "loss": 0.0651,
      "step": 1793
    },
    {
      "epoch": 0.09264136328427576,
      "grad_norm": 6.244324684143066,
      "learning_rate": 9.829116259548465e-06,
      "loss": 0.1155,
      "step": 1794
    },
    {
      "epoch": 0.09269300284017558,
      "grad_norm": 3.333156108856201,
      "learning_rate": 9.828903815150835e-06,
      "loss": 0.105,
      "step": 1795
    },
    {
      "epoch": 0.0927446423960754,
      "grad_norm": 14.635869979858398,
      "learning_rate": 9.828691241077457e-06,
      "loss": 0.1729,
      "step": 1796
    },
    {
      "epoch": 0.09279628195197522,
      "grad_norm": 2.689655303955078,
      "learning_rate": 9.828478537334038e-06,
      "loss": 0.0731,
      "step": 1797
    },
    {
      "epoch": 0.09284792150787503,
      "grad_norm": 5.8939337730407715,
      "learning_rate": 9.82826570392629e-06,
      "loss": 0.0737,
      "step": 1798
    },
    {
      "epoch": 0.09289956106377485,
      "grad_norm": 9.114204406738281,
      "learning_rate": 9.828052740859929e-06,
      "loss": 0.0777,
      "step": 1799
    },
    {
      "epoch": 0.09295120061967467,
      "grad_norm": 12.59143352508545,
      "learning_rate": 9.827839648140673e-06,
      "loss": 0.1508,
      "step": 1800
    },
    {
      "epoch": 0.09300284017557449,
      "grad_norm": 10.034825325012207,
      "learning_rate": 9.827626425774247e-06,
      "loss": 0.0888,
      "step": 1801
    },
    {
      "epoch": 0.09305447973147431,
      "grad_norm": 4.873637676239014,
      "learning_rate": 9.827413073766374e-06,
      "loss": 0.0723,
      "step": 1802
    },
    {
      "epoch": 0.09310611928737413,
      "grad_norm": 4.191925048828125,
      "learning_rate": 9.827199592122785e-06,
      "loss": 0.0921,
      "step": 1803
    },
    {
      "epoch": 0.09315775884327394,
      "grad_norm": 17.463207244873047,
      "learning_rate": 9.826985980849213e-06,
      "loss": 0.0792,
      "step": 1804
    },
    {
      "epoch": 0.09320939839917376,
      "grad_norm": 13.029545783996582,
      "learning_rate": 9.826772239951392e-06,
      "loss": 0.0853,
      "step": 1805
    },
    {
      "epoch": 0.09326103795507358,
      "grad_norm": 10.91751766204834,
      "learning_rate": 9.826558369435065e-06,
      "loss": 0.0637,
      "step": 1806
    },
    {
      "epoch": 0.0933126775109734,
      "grad_norm": 5.025144100189209,
      "learning_rate": 9.826344369305973e-06,
      "loss": 0.0619,
      "step": 1807
    },
    {
      "epoch": 0.09336431706687323,
      "grad_norm": 7.501998424530029,
      "learning_rate": 9.826130239569865e-06,
      "loss": 0.0622,
      "step": 1808
    },
    {
      "epoch": 0.09341595662277305,
      "grad_norm": 9.078734397888184,
      "learning_rate": 9.825915980232488e-06,
      "loss": 0.0437,
      "step": 1809
    },
    {
      "epoch": 0.09346759617867287,
      "grad_norm": 5.342756748199463,
      "learning_rate": 9.825701591299598e-06,
      "loss": 0.1031,
      "step": 1810
    },
    {
      "epoch": 0.09351923573457269,
      "grad_norm": 5.520531177520752,
      "learning_rate": 9.825487072776954e-06,
      "loss": 0.0687,
      "step": 1811
    },
    {
      "epoch": 0.0935708752904725,
      "grad_norm": 4.900416851043701,
      "learning_rate": 9.825272424670311e-06,
      "loss": 0.0762,
      "step": 1812
    },
    {
      "epoch": 0.09362251484637232,
      "grad_norm": 5.583768367767334,
      "learning_rate": 9.825057646985438e-06,
      "loss": 0.0664,
      "step": 1813
    },
    {
      "epoch": 0.09367415440227214,
      "grad_norm": 6.705413341522217,
      "learning_rate": 9.824842739728102e-06,
      "loss": 0.0805,
      "step": 1814
    },
    {
      "epoch": 0.09372579395817196,
      "grad_norm": 5.568469524383545,
      "learning_rate": 9.824627702904075e-06,
      "loss": 0.0371,
      "step": 1815
    },
    {
      "epoch": 0.09377743351407178,
      "grad_norm": 7.870848655700684,
      "learning_rate": 9.824412536519127e-06,
      "loss": 0.0796,
      "step": 1816
    },
    {
      "epoch": 0.0938290730699716,
      "grad_norm": 9.060432434082031,
      "learning_rate": 9.824197240579041e-06,
      "loss": 0.0454,
      "step": 1817
    },
    {
      "epoch": 0.09388071262587142,
      "grad_norm": 7.044742584228516,
      "learning_rate": 9.823981815089596e-06,
      "loss": 0.1139,
      "step": 1818
    },
    {
      "epoch": 0.09393235218177123,
      "grad_norm": 10.982089042663574,
      "learning_rate": 9.823766260056578e-06,
      "loss": 0.0566,
      "step": 1819
    },
    {
      "epoch": 0.09398399173767105,
      "grad_norm": 13.262882232666016,
      "learning_rate": 9.823550575485775e-06,
      "loss": 0.076,
      "step": 1820
    },
    {
      "epoch": 0.09403563129357087,
      "grad_norm": 5.834909439086914,
      "learning_rate": 9.82333476138298e-06,
      "loss": 0.0754,
      "step": 1821
    },
    {
      "epoch": 0.09408727084947069,
      "grad_norm": 7.964276313781738,
      "learning_rate": 9.823118817753986e-06,
      "loss": 0.0719,
      "step": 1822
    },
    {
      "epoch": 0.09413891040537051,
      "grad_norm": 7.014941692352295,
      "learning_rate": 9.822902744604596e-06,
      "loss": 0.0853,
      "step": 1823
    },
    {
      "epoch": 0.09419054996127034,
      "grad_norm": 3.454063892364502,
      "learning_rate": 9.82268654194061e-06,
      "loss": 0.0804,
      "step": 1824
    },
    {
      "epoch": 0.09424218951717016,
      "grad_norm": 6.168149471282959,
      "learning_rate": 9.822470209767834e-06,
      "loss": 0.0591,
      "step": 1825
    },
    {
      "epoch": 0.09429382907306998,
      "grad_norm": 35.77104568481445,
      "learning_rate": 9.822253748092076e-06,
      "loss": 0.3249,
      "step": 1826
    },
    {
      "epoch": 0.0943454686289698,
      "grad_norm": 9.165575981140137,
      "learning_rate": 9.822037156919153e-06,
      "loss": 0.1323,
      "step": 1827
    },
    {
      "epoch": 0.09439710818486961,
      "grad_norm": 8.557072639465332,
      "learning_rate": 9.821820436254878e-06,
      "loss": 0.0837,
      "step": 1828
    },
    {
      "epoch": 0.09444874774076943,
      "grad_norm": 3.277454376220703,
      "learning_rate": 9.82160358610507e-06,
      "loss": 0.0568,
      "step": 1829
    },
    {
      "epoch": 0.09450038729666925,
      "grad_norm": 10.797391891479492,
      "learning_rate": 9.821386606475555e-06,
      "loss": 0.1037,
      "step": 1830
    },
    {
      "epoch": 0.09455202685256907,
      "grad_norm": 12.849955558776855,
      "learning_rate": 9.82116949737216e-06,
      "loss": 0.0998,
      "step": 1831
    },
    {
      "epoch": 0.09460366640846889,
      "grad_norm": 3.663731813430786,
      "learning_rate": 9.820952258800712e-06,
      "loss": 0.0812,
      "step": 1832
    },
    {
      "epoch": 0.0946553059643687,
      "grad_norm": 4.015179634094238,
      "learning_rate": 9.820734890767048e-06,
      "loss": 0.115,
      "step": 1833
    },
    {
      "epoch": 0.09470694552026852,
      "grad_norm": 4.741633892059326,
      "learning_rate": 9.820517393277003e-06,
      "loss": 0.0727,
      "step": 1834
    },
    {
      "epoch": 0.09475858507616834,
      "grad_norm": 3.3521902561187744,
      "learning_rate": 9.82029976633642e-06,
      "loss": 0.0462,
      "step": 1835
    },
    {
      "epoch": 0.09481022463206816,
      "grad_norm": 3.564316511154175,
      "learning_rate": 9.820082009951141e-06,
      "loss": 0.0687,
      "step": 1836
    },
    {
      "epoch": 0.09486186418796798,
      "grad_norm": 2.3372137546539307,
      "learning_rate": 9.819864124127014e-06,
      "loss": 0.0428,
      "step": 1837
    },
    {
      "epoch": 0.0949135037438678,
      "grad_norm": 4.530689716339111,
      "learning_rate": 9.819646108869891e-06,
      "loss": 0.097,
      "step": 1838
    },
    {
      "epoch": 0.09496514329976762,
      "grad_norm": 3.3220248222351074,
      "learning_rate": 9.819427964185625e-06,
      "loss": 0.069,
      "step": 1839
    },
    {
      "epoch": 0.09501678285566745,
      "grad_norm": 10.8014554977417,
      "learning_rate": 9.819209690080079e-06,
      "loss": 0.1152,
      "step": 1840
    },
    {
      "epoch": 0.09506842241156727,
      "grad_norm": 11.540700912475586,
      "learning_rate": 9.818991286559107e-06,
      "loss": 0.0853,
      "step": 1841
    },
    {
      "epoch": 0.09512006196746708,
      "grad_norm": 7.560811519622803,
      "learning_rate": 9.81877275362858e-06,
      "loss": 0.0937,
      "step": 1842
    },
    {
      "epoch": 0.0951717015233669,
      "grad_norm": 7.112672805786133,
      "learning_rate": 9.818554091294363e-06,
      "loss": 0.0905,
      "step": 1843
    },
    {
      "epoch": 0.09522334107926672,
      "grad_norm": 5.824686527252197,
      "learning_rate": 9.818335299562329e-06,
      "loss": 0.0581,
      "step": 1844
    },
    {
      "epoch": 0.09527498063516654,
      "grad_norm": 2.525775909423828,
      "learning_rate": 9.818116378438355e-06,
      "loss": 0.0415,
      "step": 1845
    },
    {
      "epoch": 0.09532662019106636,
      "grad_norm": 8.731331825256348,
      "learning_rate": 9.817897327928319e-06,
      "loss": 0.0711,
      "step": 1846
    },
    {
      "epoch": 0.09537825974696618,
      "grad_norm": 12.886061668395996,
      "learning_rate": 9.8176781480381e-06,
      "loss": 0.1671,
      "step": 1847
    },
    {
      "epoch": 0.095429899302866,
      "grad_norm": 7.124521255493164,
      "learning_rate": 9.81745883877359e-06,
      "loss": 0.0698,
      "step": 1848
    },
    {
      "epoch": 0.09548153885876581,
      "grad_norm": 6.244156360626221,
      "learning_rate": 9.817239400140675e-06,
      "loss": 0.0989,
      "step": 1849
    },
    {
      "epoch": 0.09553317841466563,
      "grad_norm": 17.1066951751709,
      "learning_rate": 9.817019832145247e-06,
      "loss": 0.1634,
      "step": 1850
    },
    {
      "epoch": 0.09558481797056545,
      "grad_norm": 5.6550445556640625,
      "learning_rate": 9.816800134793205e-06,
      "loss": 0.1273,
      "step": 1851
    },
    {
      "epoch": 0.09563645752646527,
      "grad_norm": 10.198446273803711,
      "learning_rate": 9.816580308090446e-06,
      "loss": 0.0739,
      "step": 1852
    },
    {
      "epoch": 0.09568809708236509,
      "grad_norm": 11.96576976776123,
      "learning_rate": 9.816360352042875e-06,
      "loss": 0.1452,
      "step": 1853
    },
    {
      "epoch": 0.0957397366382649,
      "grad_norm": 2.7215030193328857,
      "learning_rate": 9.816140266656397e-06,
      "loss": 0.0321,
      "step": 1854
    },
    {
      "epoch": 0.09579137619416472,
      "grad_norm": 4.893381595611572,
      "learning_rate": 9.815920051936925e-06,
      "loss": 0.0999,
      "step": 1855
    },
    {
      "epoch": 0.09584301575006456,
      "grad_norm": 4.623499870300293,
      "learning_rate": 9.81569970789037e-06,
      "loss": 0.0668,
      "step": 1856
    },
    {
      "epoch": 0.09589465530596437,
      "grad_norm": 8.872573852539062,
      "learning_rate": 9.81547923452265e-06,
      "loss": 0.1303,
      "step": 1857
    },
    {
      "epoch": 0.09594629486186419,
      "grad_norm": 10.631209373474121,
      "learning_rate": 9.815258631839686e-06,
      "loss": 0.1556,
      "step": 1858
    },
    {
      "epoch": 0.09599793441776401,
      "grad_norm": 2.745466947555542,
      "learning_rate": 9.815037899847403e-06,
      "loss": 0.0553,
      "step": 1859
    },
    {
      "epoch": 0.09604957397366383,
      "grad_norm": 8.43391227722168,
      "learning_rate": 9.814817038551727e-06,
      "loss": 0.1458,
      "step": 1860
    },
    {
      "epoch": 0.09610121352956365,
      "grad_norm": 5.506300926208496,
      "learning_rate": 9.814596047958587e-06,
      "loss": 0.0938,
      "step": 1861
    },
    {
      "epoch": 0.09615285308546347,
      "grad_norm": 7.8981242179870605,
      "learning_rate": 9.814374928073922e-06,
      "loss": 0.0805,
      "step": 1862
    },
    {
      "epoch": 0.09620449264136328,
      "grad_norm": 22.32009506225586,
      "learning_rate": 9.814153678903668e-06,
      "loss": 0.183,
      "step": 1863
    },
    {
      "epoch": 0.0962561321972631,
      "grad_norm": 10.193687438964844,
      "learning_rate": 9.813932300453766e-06,
      "loss": 0.0914,
      "step": 1864
    },
    {
      "epoch": 0.09630777175316292,
      "grad_norm": 12.537751197814941,
      "learning_rate": 9.81371079273016e-06,
      "loss": 0.0742,
      "step": 1865
    },
    {
      "epoch": 0.09635941130906274,
      "grad_norm": 6.621200084686279,
      "learning_rate": 9.813489155738801e-06,
      "loss": 0.1382,
      "step": 1866
    },
    {
      "epoch": 0.09641105086496256,
      "grad_norm": 5.677754878997803,
      "learning_rate": 9.813267389485639e-06,
      "loss": 0.0826,
      "step": 1867
    },
    {
      "epoch": 0.09646269042086238,
      "grad_norm": 7.041050910949707,
      "learning_rate": 9.81304549397663e-06,
      "loss": 0.1221,
      "step": 1868
    },
    {
      "epoch": 0.0965143299767622,
      "grad_norm": 5.4804277420043945,
      "learning_rate": 9.812823469217733e-06,
      "loss": 0.0826,
      "step": 1869
    },
    {
      "epoch": 0.09656596953266201,
      "grad_norm": 5.079614162445068,
      "learning_rate": 9.81260131521491e-06,
      "loss": 0.0844,
      "step": 1870
    },
    {
      "epoch": 0.09661760908856183,
      "grad_norm": 5.21281099319458,
      "learning_rate": 9.812379031974125e-06,
      "loss": 0.0975,
      "step": 1871
    },
    {
      "epoch": 0.09666924864446166,
      "grad_norm": 7.680988788604736,
      "learning_rate": 9.81215661950135e-06,
      "loss": 0.0916,
      "step": 1872
    },
    {
      "epoch": 0.09672088820036148,
      "grad_norm": 4.600657939910889,
      "learning_rate": 9.811934077802556e-06,
      "loss": 0.0548,
      "step": 1873
    },
    {
      "epoch": 0.0967725277562613,
      "grad_norm": 3.1328094005584717,
      "learning_rate": 9.81171140688372e-06,
      "loss": 0.0632,
      "step": 1874
    },
    {
      "epoch": 0.09682416731216112,
      "grad_norm": 5.855184078216553,
      "learning_rate": 9.811488606750821e-06,
      "loss": 0.0917,
      "step": 1875
    },
    {
      "epoch": 0.09687580686806094,
      "grad_norm": 18.89946937561035,
      "learning_rate": 9.811265677409842e-06,
      "loss": 0.0998,
      "step": 1876
    },
    {
      "epoch": 0.09692744642396076,
      "grad_norm": 6.200829029083252,
      "learning_rate": 9.81104261886677e-06,
      "loss": 0.0337,
      "step": 1877
    },
    {
      "epoch": 0.09697908597986057,
      "grad_norm": 11.99652099609375,
      "learning_rate": 9.810819431127596e-06,
      "loss": 0.0674,
      "step": 1878
    },
    {
      "epoch": 0.09703072553576039,
      "grad_norm": 17.244298934936523,
      "learning_rate": 9.810596114198313e-06,
      "loss": 0.1302,
      "step": 1879
    },
    {
      "epoch": 0.09708236509166021,
      "grad_norm": 4.602144241333008,
      "learning_rate": 9.810372668084916e-06,
      "loss": 0.0701,
      "step": 1880
    },
    {
      "epoch": 0.09713400464756003,
      "grad_norm": 11.134688377380371,
      "learning_rate": 9.810149092793407e-06,
      "loss": 0.0808,
      "step": 1881
    },
    {
      "epoch": 0.09718564420345985,
      "grad_norm": 6.981633186340332,
      "learning_rate": 9.80992538832979e-06,
      "loss": 0.1082,
      "step": 1882
    },
    {
      "epoch": 0.09723728375935967,
      "grad_norm": 9.001764297485352,
      "learning_rate": 9.809701554700073e-06,
      "loss": 0.0566,
      "step": 1883
    },
    {
      "epoch": 0.09728892331525948,
      "grad_norm": 15.47097396850586,
      "learning_rate": 9.809477591910264e-06,
      "loss": 0.1216,
      "step": 1884
    },
    {
      "epoch": 0.0973405628711593,
      "grad_norm": 23.533655166625977,
      "learning_rate": 9.809253499966382e-06,
      "loss": 0.2238,
      "step": 1885
    },
    {
      "epoch": 0.09739220242705912,
      "grad_norm": 6.598683834075928,
      "learning_rate": 9.80902927887444e-06,
      "loss": 0.0957,
      "step": 1886
    },
    {
      "epoch": 0.09744384198295894,
      "grad_norm": 8.178055763244629,
      "learning_rate": 9.808804928640461e-06,
      "loss": 0.0733,
      "step": 1887
    },
    {
      "epoch": 0.09749548153885877,
      "grad_norm": 9.383674621582031,
      "learning_rate": 9.808580449270471e-06,
      "loss": 0.1236,
      "step": 1888
    },
    {
      "epoch": 0.09754712109475859,
      "grad_norm": 5.315759658813477,
      "learning_rate": 9.808355840770498e-06,
      "loss": 0.0779,
      "step": 1889
    },
    {
      "epoch": 0.09759876065065841,
      "grad_norm": 5.115828990936279,
      "learning_rate": 9.808131103146571e-06,
      "loss": 0.0937,
      "step": 1890
    },
    {
      "epoch": 0.09765040020655823,
      "grad_norm": 5.278205871582031,
      "learning_rate": 9.80790623640473e-06,
      "loss": 0.0869,
      "step": 1891
    },
    {
      "epoch": 0.09770203976245805,
      "grad_norm": 7.2783403396606445,
      "learning_rate": 9.807681240551008e-06,
      "loss": 0.0557,
      "step": 1892
    },
    {
      "epoch": 0.09775367931835786,
      "grad_norm": 11.813129425048828,
      "learning_rate": 9.80745611559145e-06,
      "loss": 0.1437,
      "step": 1893
    },
    {
      "epoch": 0.09780531887425768,
      "grad_norm": 3.026660919189453,
      "learning_rate": 9.807230861532102e-06,
      "loss": 0.0595,
      "step": 1894
    },
    {
      "epoch": 0.0978569584301575,
      "grad_norm": 10.486563682556152,
      "learning_rate": 9.807005478379013e-06,
      "loss": 0.1284,
      "step": 1895
    },
    {
      "epoch": 0.09790859798605732,
      "grad_norm": 2.574282646179199,
      "learning_rate": 9.806779966138233e-06,
      "loss": 0.0583,
      "step": 1896
    },
    {
      "epoch": 0.09796023754195714,
      "grad_norm": 7.1395368576049805,
      "learning_rate": 9.80655432481582e-06,
      "loss": 0.0854,
      "step": 1897
    },
    {
      "epoch": 0.09801187709785696,
      "grad_norm": 6.042491912841797,
      "learning_rate": 9.806328554417833e-06,
      "loss": 0.1126,
      "step": 1898
    },
    {
      "epoch": 0.09806351665375677,
      "grad_norm": 16.955158233642578,
      "learning_rate": 9.806102654950333e-06,
      "loss": 0.1488,
      "step": 1899
    },
    {
      "epoch": 0.09811515620965659,
      "grad_norm": 10.949597358703613,
      "learning_rate": 9.80587662641939e-06,
      "loss": 0.0815,
      "step": 1900
    },
    {
      "epoch": 0.09816679576555641,
      "grad_norm": 3.743131637573242,
      "learning_rate": 9.805650468831073e-06,
      "loss": 0.0626,
      "step": 1901
    },
    {
      "epoch": 0.09821843532145623,
      "grad_norm": 2.661734104156494,
      "learning_rate": 9.805424182191452e-06,
      "loss": 0.0887,
      "step": 1902
    },
    {
      "epoch": 0.09827007487735605,
      "grad_norm": 8.654378890991211,
      "learning_rate": 9.805197766506606e-06,
      "loss": 0.1067,
      "step": 1903
    },
    {
      "epoch": 0.09832171443325588,
      "grad_norm": 10.484905242919922,
      "learning_rate": 9.804971221782618e-06,
      "loss": 0.0968,
      "step": 1904
    },
    {
      "epoch": 0.0983733539891557,
      "grad_norm": 16.19354820251465,
      "learning_rate": 9.804744548025567e-06,
      "loss": 0.1329,
      "step": 1905
    },
    {
      "epoch": 0.09842499354505552,
      "grad_norm": 9.296616554260254,
      "learning_rate": 9.80451774524154e-06,
      "loss": 0.1011,
      "step": 1906
    },
    {
      "epoch": 0.09847663310095534,
      "grad_norm": 14.93027400970459,
      "learning_rate": 9.804290813436632e-06,
      "loss": 0.1349,
      "step": 1907
    },
    {
      "epoch": 0.09852827265685515,
      "grad_norm": 12.33952808380127,
      "learning_rate": 9.804063752616933e-06,
      "loss": 0.1237,
      "step": 1908
    },
    {
      "epoch": 0.09857991221275497,
      "grad_norm": 2.5484187602996826,
      "learning_rate": 9.803836562788542e-06,
      "loss": 0.0648,
      "step": 1909
    },
    {
      "epoch": 0.09863155176865479,
      "grad_norm": 4.595915794372559,
      "learning_rate": 9.803609243957559e-06,
      "loss": 0.0522,
      "step": 1910
    },
    {
      "epoch": 0.09868319132455461,
      "grad_norm": 2.551128387451172,
      "learning_rate": 9.803381796130092e-06,
      "loss": 0.0417,
      "step": 1911
    },
    {
      "epoch": 0.09873483088045443,
      "grad_norm": 3.8467905521392822,
      "learning_rate": 9.803154219312245e-06,
      "loss": 0.0955,
      "step": 1912
    },
    {
      "epoch": 0.09878647043635425,
      "grad_norm": 5.139723300933838,
      "learning_rate": 9.80292651351013e-06,
      "loss": 0.091,
      "step": 1913
    },
    {
      "epoch": 0.09883810999225406,
      "grad_norm": 7.910643100738525,
      "learning_rate": 9.802698678729864e-06,
      "loss": 0.0881,
      "step": 1914
    },
    {
      "epoch": 0.09888974954815388,
      "grad_norm": 10.538387298583984,
      "learning_rate": 9.80247071497756e-06,
      "loss": 0.1407,
      "step": 1915
    },
    {
      "epoch": 0.0989413891040537,
      "grad_norm": 5.810377597808838,
      "learning_rate": 9.802242622259349e-06,
      "loss": 0.0919,
      "step": 1916
    },
    {
      "epoch": 0.09899302865995352,
      "grad_norm": 5.246915340423584,
      "learning_rate": 9.802014400581346e-06,
      "loss": 0.0923,
      "step": 1917
    },
    {
      "epoch": 0.09904466821585334,
      "grad_norm": 5.7014336585998535,
      "learning_rate": 9.801786049949687e-06,
      "loss": 0.1036,
      "step": 1918
    },
    {
      "epoch": 0.09909630777175317,
      "grad_norm": 2.25656795501709,
      "learning_rate": 9.801557570370501e-06,
      "loss": 0.0302,
      "step": 1919
    },
    {
      "epoch": 0.09914794732765299,
      "grad_norm": 7.853806018829346,
      "learning_rate": 9.801328961849924e-06,
      "loss": 0.0691,
      "step": 1920
    },
    {
      "epoch": 0.0991995868835528,
      "grad_norm": 4.8204450607299805,
      "learning_rate": 9.801100224394095e-06,
      "loss": 0.0648,
      "step": 1921
    },
    {
      "epoch": 0.09925122643945263,
      "grad_norm": 4.457400321960449,
      "learning_rate": 9.800871358009158e-06,
      "loss": 0.0824,
      "step": 1922
    },
    {
      "epoch": 0.09930286599535244,
      "grad_norm": 6.8608717918396,
      "learning_rate": 9.800642362701255e-06,
      "loss": 0.0761,
      "step": 1923
    },
    {
      "epoch": 0.09935450555125226,
      "grad_norm": 2.401528835296631,
      "learning_rate": 9.800413238476542e-06,
      "loss": 0.0459,
      "step": 1924
    },
    {
      "epoch": 0.09940614510715208,
      "grad_norm": 4.328356742858887,
      "learning_rate": 9.800183985341166e-06,
      "loss": 0.0906,
      "step": 1925
    },
    {
      "epoch": 0.0994577846630519,
      "grad_norm": 11.808272361755371,
      "learning_rate": 9.799954603301285e-06,
      "loss": 0.1239,
      "step": 1926
    },
    {
      "epoch": 0.09950942421895172,
      "grad_norm": 6.682941436767578,
      "learning_rate": 9.799725092363062e-06,
      "loss": 0.0564,
      "step": 1927
    },
    {
      "epoch": 0.09956106377485154,
      "grad_norm": 18.07846450805664,
      "learning_rate": 9.799495452532654e-06,
      "loss": 0.1003,
      "step": 1928
    },
    {
      "epoch": 0.09961270333075135,
      "grad_norm": 7.552882671356201,
      "learning_rate": 9.799265683816234e-06,
      "loss": 0.0656,
      "step": 1929
    },
    {
      "epoch": 0.09966434288665117,
      "grad_norm": 3.980531692504883,
      "learning_rate": 9.799035786219967e-06,
      "loss": 0.0441,
      "step": 1930
    },
    {
      "epoch": 0.09971598244255099,
      "grad_norm": 4.45389986038208,
      "learning_rate": 9.798805759750033e-06,
      "loss": 0.0863,
      "step": 1931
    },
    {
      "epoch": 0.09976762199845081,
      "grad_norm": 28.087955474853516,
      "learning_rate": 9.798575604412604e-06,
      "loss": 0.1762,
      "step": 1932
    },
    {
      "epoch": 0.09981926155435063,
      "grad_norm": 5.487453937530518,
      "learning_rate": 9.798345320213862e-06,
      "loss": 0.0853,
      "step": 1933
    },
    {
      "epoch": 0.09987090111025045,
      "grad_norm": 4.8082427978515625,
      "learning_rate": 9.79811490715999e-06,
      "loss": 0.1003,
      "step": 1934
    },
    {
      "epoch": 0.09992254066615028,
      "grad_norm": 9.095142364501953,
      "learning_rate": 9.79788436525718e-06,
      "loss": 0.0955,
      "step": 1935
    },
    {
      "epoch": 0.0999741802220501,
      "grad_norm": 9.00161075592041,
      "learning_rate": 9.797653694511615e-06,
      "loss": 0.1013,
      "step": 1936
    },
    {
      "epoch": 0.10002581977794991,
      "grad_norm": 3.8558731079101562,
      "learning_rate": 9.797422894929498e-06,
      "loss": 0.0517,
      "step": 1937
    },
    {
      "epoch": 0.10002581977794991,
      "eval_loss": 0.0871032103896141,
      "eval_runtime": 34.1823,
      "eval_samples_per_second": 45.784,
      "eval_steps_per_second": 5.734,
      "step": 1937
    },
    {
      "epoch": 0.10007745933384973,
      "grad_norm": 4.251840591430664,
      "learning_rate": 9.797191966517021e-06,
      "loss": 0.0818,
      "step": 1938
    },
    {
      "epoch": 0.10012909888974955,
      "grad_norm": 2.162189245223999,
      "learning_rate": 9.796960909280389e-06,
      "loss": 0.0469,
      "step": 1939
    },
    {
      "epoch": 0.10018073844564937,
      "grad_norm": 3.7780263423919678,
      "learning_rate": 9.796729723225804e-06,
      "loss": 0.0738,
      "step": 1940
    },
    {
      "epoch": 0.10023237800154919,
      "grad_norm": 4.518858909606934,
      "learning_rate": 9.796498408359476e-06,
      "loss": 0.0854,
      "step": 1941
    },
    {
      "epoch": 0.100284017557449,
      "grad_norm": 4.142231464385986,
      "learning_rate": 9.796266964687616e-06,
      "loss": 0.0942,
      "step": 1942
    },
    {
      "epoch": 0.10033565711334883,
      "grad_norm": 21.666763305664062,
      "learning_rate": 9.79603539221644e-06,
      "loss": 0.1473,
      "step": 1943
    },
    {
      "epoch": 0.10038729666924864,
      "grad_norm": 13.112496376037598,
      "learning_rate": 9.795803690952167e-06,
      "loss": 0.1161,
      "step": 1944
    },
    {
      "epoch": 0.10043893622514846,
      "grad_norm": 10.936271667480469,
      "learning_rate": 9.795571860901015e-06,
      "loss": 0.0961,
      "step": 1945
    },
    {
      "epoch": 0.10049057578104828,
      "grad_norm": 14.921509742736816,
      "learning_rate": 9.795339902069217e-06,
      "loss": 0.1189,
      "step": 1946
    },
    {
      "epoch": 0.1005422153369481,
      "grad_norm": 17.235944747924805,
      "learning_rate": 9.795107814462994e-06,
      "loss": 0.1362,
      "step": 1947
    },
    {
      "epoch": 0.10059385489284792,
      "grad_norm": 8.77182388305664,
      "learning_rate": 9.794875598088585e-06,
      "loss": 0.0724,
      "step": 1948
    },
    {
      "epoch": 0.10064549444874774,
      "grad_norm": 10.630849838256836,
      "learning_rate": 9.794643252952223e-06,
      "loss": 0.0591,
      "step": 1949
    },
    {
      "epoch": 0.10069713400464755,
      "grad_norm": 5.080479621887207,
      "learning_rate": 9.794410779060147e-06,
      "loss": 0.0998,
      "step": 1950
    },
    {
      "epoch": 0.10074877356054739,
      "grad_norm": 9.519889831542969,
      "learning_rate": 9.794178176418601e-06,
      "loss": 0.1414,
      "step": 1951
    },
    {
      "epoch": 0.1008004131164472,
      "grad_norm": 8.21805191040039,
      "learning_rate": 9.79394544503383e-06,
      "loss": 0.1055,
      "step": 1952
    },
    {
      "epoch": 0.10085205267234702,
      "grad_norm": 11.321654319763184,
      "learning_rate": 9.793712584912087e-06,
      "loss": 0.1064,
      "step": 1953
    },
    {
      "epoch": 0.10090369222824684,
      "grad_norm": 7.448075294494629,
      "learning_rate": 9.793479596059622e-06,
      "loss": 0.0779,
      "step": 1954
    },
    {
      "epoch": 0.10095533178414666,
      "grad_norm": 8.96998119354248,
      "learning_rate": 9.793246478482692e-06,
      "loss": 0.1146,
      "step": 1955
    },
    {
      "epoch": 0.10100697134004648,
      "grad_norm": 5.400252342224121,
      "learning_rate": 9.793013232187558e-06,
      "loss": 0.0559,
      "step": 1956
    },
    {
      "epoch": 0.1010586108959463,
      "grad_norm": 3.6698949337005615,
      "learning_rate": 9.792779857180485e-06,
      "loss": 0.0549,
      "step": 1957
    },
    {
      "epoch": 0.10111025045184611,
      "grad_norm": 15.967495918273926,
      "learning_rate": 9.792546353467737e-06,
      "loss": 0.0965,
      "step": 1958
    },
    {
      "epoch": 0.10116189000774593,
      "grad_norm": 4.44314432144165,
      "learning_rate": 9.792312721055586e-06,
      "loss": 0.03,
      "step": 1959
    },
    {
      "epoch": 0.10121352956364575,
      "grad_norm": 3.694298028945923,
      "learning_rate": 9.792078959950307e-06,
      "loss": 0.0532,
      "step": 1960
    },
    {
      "epoch": 0.10126516911954557,
      "grad_norm": 7.103492736816406,
      "learning_rate": 9.791845070158175e-06,
      "loss": 0.0343,
      "step": 1961
    },
    {
      "epoch": 0.10131680867544539,
      "grad_norm": 5.511290550231934,
      "learning_rate": 9.791611051685474e-06,
      "loss": 0.0502,
      "step": 1962
    },
    {
      "epoch": 0.1013684482313452,
      "grad_norm": 29.927413940429688,
      "learning_rate": 9.791376904538487e-06,
      "loss": 0.2113,
      "step": 1963
    },
    {
      "epoch": 0.10142008778724503,
      "grad_norm": 6.292163848876953,
      "learning_rate": 9.7911426287235e-06,
      "loss": 0.0457,
      "step": 1964
    },
    {
      "epoch": 0.10147172734314484,
      "grad_norm": 5.619717121124268,
      "learning_rate": 9.790908224246808e-06,
      "loss": 0.0818,
      "step": 1965
    },
    {
      "epoch": 0.10152336689904466,
      "grad_norm": 4.767144203186035,
      "learning_rate": 9.7906736911147e-06,
      "loss": 0.0575,
      "step": 1966
    },
    {
      "epoch": 0.1015750064549445,
      "grad_norm": 5.287398338317871,
      "learning_rate": 9.79043902933348e-06,
      "loss": 0.0538,
      "step": 1967
    },
    {
      "epoch": 0.10162664601084431,
      "grad_norm": 9.71335506439209,
      "learning_rate": 9.790204238909447e-06,
      "loss": 0.0808,
      "step": 1968
    },
    {
      "epoch": 0.10167828556674413,
      "grad_norm": 12.975753784179688,
      "learning_rate": 9.789969319848907e-06,
      "loss": 0.1312,
      "step": 1969
    },
    {
      "epoch": 0.10172992512264395,
      "grad_norm": 15.913261413574219,
      "learning_rate": 9.789734272158166e-06,
      "loss": 0.1743,
      "step": 1970
    },
    {
      "epoch": 0.10178156467854377,
      "grad_norm": 5.520126819610596,
      "learning_rate": 9.78949909584354e-06,
      "loss": 0.0984,
      "step": 1971
    },
    {
      "epoch": 0.10183320423444359,
      "grad_norm": 4.64992094039917,
      "learning_rate": 9.789263790911341e-06,
      "loss": 0.0939,
      "step": 1972
    },
    {
      "epoch": 0.1018848437903434,
      "grad_norm": 3.552199363708496,
      "learning_rate": 9.789028357367892e-06,
      "loss": 0.0509,
      "step": 1973
    },
    {
      "epoch": 0.10193648334624322,
      "grad_norm": 13.985942840576172,
      "learning_rate": 9.78879279521951e-06,
      "loss": 0.1013,
      "step": 1974
    },
    {
      "epoch": 0.10198812290214304,
      "grad_norm": 20.859432220458984,
      "learning_rate": 9.788557104472522e-06,
      "loss": 0.1899,
      "step": 1975
    },
    {
      "epoch": 0.10203976245804286,
      "grad_norm": 8.610810279846191,
      "learning_rate": 9.788321285133262e-06,
      "loss": 0.1098,
      "step": 1976
    },
    {
      "epoch": 0.10209140201394268,
      "grad_norm": 7.065051078796387,
      "learning_rate": 9.788085337208058e-06,
      "loss": 0.0737,
      "step": 1977
    },
    {
      "epoch": 0.1021430415698425,
      "grad_norm": 14.392461776733398,
      "learning_rate": 9.787849260703247e-06,
      "loss": 0.0686,
      "step": 1978
    },
    {
      "epoch": 0.10219468112574231,
      "grad_norm": 4.394903659820557,
      "learning_rate": 9.787613055625168e-06,
      "loss": 0.0637,
      "step": 1979
    },
    {
      "epoch": 0.10224632068164213,
      "grad_norm": 2.2038514614105225,
      "learning_rate": 9.787376721980169e-06,
      "loss": 0.0615,
      "step": 1980
    },
    {
      "epoch": 0.10229796023754195,
      "grad_norm": 4.440915584564209,
      "learning_rate": 9.78714025977459e-06,
      "loss": 0.0394,
      "step": 1981
    },
    {
      "epoch": 0.10234959979344177,
      "grad_norm": 4.2257232666015625,
      "learning_rate": 9.786903669014785e-06,
      "loss": 0.076,
      "step": 1982
    },
    {
      "epoch": 0.1024012393493416,
      "grad_norm": 13.155374526977539,
      "learning_rate": 9.786666949707104e-06,
      "loss": 0.1311,
      "step": 1983
    },
    {
      "epoch": 0.10245287890524142,
      "grad_norm": 5.2980546951293945,
      "learning_rate": 9.78643010185791e-06,
      "loss": 0.0852,
      "step": 1984
    },
    {
      "epoch": 0.10250451846114124,
      "grad_norm": 3.1543400287628174,
      "learning_rate": 9.786193125473556e-06,
      "loss": 0.0543,
      "step": 1985
    },
    {
      "epoch": 0.10255615801704106,
      "grad_norm": 5.613210678100586,
      "learning_rate": 9.78595602056041e-06,
      "loss": 0.073,
      "step": 1986
    },
    {
      "epoch": 0.10260779757294088,
      "grad_norm": 5.689126491546631,
      "learning_rate": 9.785718787124839e-06,
      "loss": 0.0667,
      "step": 1987
    },
    {
      "epoch": 0.1026594371288407,
      "grad_norm": 6.151695251464844,
      "learning_rate": 9.785481425173212e-06,
      "loss": 0.0419,
      "step": 1988
    },
    {
      "epoch": 0.10271107668474051,
      "grad_norm": 5.458559513092041,
      "learning_rate": 9.785243934711903e-06,
      "loss": 0.0566,
      "step": 1989
    },
    {
      "epoch": 0.10276271624064033,
      "grad_norm": 7.302018165588379,
      "learning_rate": 9.785006315747293e-06,
      "loss": 0.1714,
      "step": 1990
    },
    {
      "epoch": 0.10281435579654015,
      "grad_norm": 15.574871063232422,
      "learning_rate": 9.78476856828576e-06,
      "loss": 0.0606,
      "step": 1991
    },
    {
      "epoch": 0.10286599535243997,
      "grad_norm": 35.66671371459961,
      "learning_rate": 9.784530692333689e-06,
      "loss": 0.0483,
      "step": 1992
    },
    {
      "epoch": 0.10291763490833979,
      "grad_norm": 14.461953163146973,
      "learning_rate": 9.78429268789747e-06,
      "loss": 0.0865,
      "step": 1993
    },
    {
      "epoch": 0.1029692744642396,
      "grad_norm": 3.5178775787353516,
      "learning_rate": 9.784054554983491e-06,
      "loss": 0.0468,
      "step": 1994
    },
    {
      "epoch": 0.10302091402013942,
      "grad_norm": 6.058546543121338,
      "learning_rate": 9.783816293598148e-06,
      "loss": 0.0832,
      "step": 1995
    },
    {
      "epoch": 0.10307255357603924,
      "grad_norm": 8.544294357299805,
      "learning_rate": 9.78357790374784e-06,
      "loss": 0.0763,
      "step": 1996
    },
    {
      "epoch": 0.10312419313193906,
      "grad_norm": 10.001501083374023,
      "learning_rate": 9.78333938543897e-06,
      "loss": 0.1648,
      "step": 1997
    },
    {
      "epoch": 0.10317583268783888,
      "grad_norm": 3.9284937381744385,
      "learning_rate": 9.783100738677941e-06,
      "loss": 0.0886,
      "step": 1998
    },
    {
      "epoch": 0.10322747224373871,
      "grad_norm": 10.112072944641113,
      "learning_rate": 9.782861963471162e-06,
      "loss": 0.0483,
      "step": 1999
    },
    {
      "epoch": 0.10327911179963853,
      "grad_norm": 5.806304931640625,
      "learning_rate": 9.782623059825046e-06,
      "loss": 0.0324,
      "step": 2000
    },
    {
      "epoch": 0.10333075135553835,
      "grad_norm": 7.642569541931152,
      "learning_rate": 9.782384027746007e-06,
      "loss": 0.0941,
      "step": 2001
    },
    {
      "epoch": 0.10338239091143817,
      "grad_norm": 3.299469470977783,
      "learning_rate": 9.782144867240464e-06,
      "loss": 0.0357,
      "step": 2002
    },
    {
      "epoch": 0.10343403046733798,
      "grad_norm": 9.750215530395508,
      "learning_rate": 9.781905578314844e-06,
      "loss": 0.1061,
      "step": 2003
    },
    {
      "epoch": 0.1034856700232378,
      "grad_norm": 4.331452369689941,
      "learning_rate": 9.781666160975566e-06,
      "loss": 0.0735,
      "step": 2004
    },
    {
      "epoch": 0.10353730957913762,
      "grad_norm": 3.3786838054656982,
      "learning_rate": 9.781426615229065e-06,
      "loss": 0.0811,
      "step": 2005
    },
    {
      "epoch": 0.10358894913503744,
      "grad_norm": 36.41523742675781,
      "learning_rate": 9.781186941081769e-06,
      "loss": 0.1637,
      "step": 2006
    },
    {
      "epoch": 0.10364058869093726,
      "grad_norm": 4.351054668426514,
      "learning_rate": 9.780947138540118e-06,
      "loss": 0.0868,
      "step": 2007
    },
    {
      "epoch": 0.10369222824683708,
      "grad_norm": 5.689184665679932,
      "learning_rate": 9.780707207610549e-06,
      "loss": 0.0632,
      "step": 2008
    },
    {
      "epoch": 0.1037438678027369,
      "grad_norm": 7.827999591827393,
      "learning_rate": 9.780467148299508e-06,
      "loss": 0.0891,
      "step": 2009
    },
    {
      "epoch": 0.10379550735863671,
      "grad_norm": 5.224191188812256,
      "learning_rate": 9.78022696061344e-06,
      "loss": 0.0755,
      "step": 2010
    },
    {
      "epoch": 0.10384714691453653,
      "grad_norm": 4.430156707763672,
      "learning_rate": 9.779986644558793e-06,
      "loss": 0.0916,
      "step": 2011
    },
    {
      "epoch": 0.10389878647043635,
      "grad_norm": 8.777790069580078,
      "learning_rate": 9.779746200142023e-06,
      "loss": 0.101,
      "step": 2012
    },
    {
      "epoch": 0.10395042602633617,
      "grad_norm": 3.7633581161499023,
      "learning_rate": 9.779505627369588e-06,
      "loss": 0.0523,
      "step": 2013
    },
    {
      "epoch": 0.10400206558223599,
      "grad_norm": 4.409654140472412,
      "learning_rate": 9.779264926247944e-06,
      "loss": 0.0566,
      "step": 2014
    },
    {
      "epoch": 0.10405370513813582,
      "grad_norm": 7.17781400680542,
      "learning_rate": 9.77902409678356e-06,
      "loss": 0.1242,
      "step": 2015
    },
    {
      "epoch": 0.10410534469403564,
      "grad_norm": 11.272137641906738,
      "learning_rate": 9.7787831389829e-06,
      "loss": 0.0622,
      "step": 2016
    },
    {
      "epoch": 0.10415698424993546,
      "grad_norm": 5.911422252655029,
      "learning_rate": 9.778542052852436e-06,
      "loss": 0.0517,
      "step": 2017
    },
    {
      "epoch": 0.10420862380583527,
      "grad_norm": 4.582866191864014,
      "learning_rate": 9.77830083839864e-06,
      "loss": 0.1011,
      "step": 2018
    },
    {
      "epoch": 0.10426026336173509,
      "grad_norm": 6.585960865020752,
      "learning_rate": 9.778059495627993e-06,
      "loss": 0.0875,
      "step": 2019
    },
    {
      "epoch": 0.10431190291763491,
      "grad_norm": 4.37938928604126,
      "learning_rate": 9.777818024546973e-06,
      "loss": 0.0838,
      "step": 2020
    },
    {
      "epoch": 0.10436354247353473,
      "grad_norm": 4.191550254821777,
      "learning_rate": 9.777576425162068e-06,
      "loss": 0.0779,
      "step": 2021
    },
    {
      "epoch": 0.10441518202943455,
      "grad_norm": 12.772279739379883,
      "learning_rate": 9.777334697479762e-06,
      "loss": 0.1556,
      "step": 2022
    },
    {
      "epoch": 0.10446682158533437,
      "grad_norm": 11.821547508239746,
      "learning_rate": 9.777092841506547e-06,
      "loss": 0.1473,
      "step": 2023
    },
    {
      "epoch": 0.10451846114123418,
      "grad_norm": 4.903713703155518,
      "learning_rate": 9.77685085724892e-06,
      "loss": 0.077,
      "step": 2024
    },
    {
      "epoch": 0.104570100697134,
      "grad_norm": 11.710458755493164,
      "learning_rate": 9.776608744713379e-06,
      "loss": 0.0963,
      "step": 2025
    },
    {
      "epoch": 0.10462174025303382,
      "grad_norm": 16.352083206176758,
      "learning_rate": 9.776366503906423e-06,
      "loss": 0.1102,
      "step": 2026
    },
    {
      "epoch": 0.10467337980893364,
      "grad_norm": 5.119138717651367,
      "learning_rate": 9.776124134834559e-06,
      "loss": 0.0638,
      "step": 2027
    },
    {
      "epoch": 0.10472501936483346,
      "grad_norm": 6.373201370239258,
      "learning_rate": 9.775881637504296e-06,
      "loss": 0.0741,
      "step": 2028
    },
    {
      "epoch": 0.10477665892073328,
      "grad_norm": 20.67069435119629,
      "learning_rate": 9.775639011922146e-06,
      "loss": 0.1836,
      "step": 2029
    },
    {
      "epoch": 0.1048282984766331,
      "grad_norm": 4.027720928192139,
      "learning_rate": 9.775396258094626e-06,
      "loss": 0.0826,
      "step": 2030
    },
    {
      "epoch": 0.10487993803253293,
      "grad_norm": 1.8565670251846313,
      "learning_rate": 9.77515337602825e-06,
      "loss": 0.0215,
      "step": 2031
    },
    {
      "epoch": 0.10493157758843275,
      "grad_norm": 3.727696180343628,
      "learning_rate": 9.774910365729544e-06,
      "loss": 0.0524,
      "step": 2032
    },
    {
      "epoch": 0.10498321714433256,
      "grad_norm": 3.11360239982605,
      "learning_rate": 9.774667227205034e-06,
      "loss": 0.0546,
      "step": 2033
    },
    {
      "epoch": 0.10503485670023238,
      "grad_norm": 6.251511573791504,
      "learning_rate": 9.774423960461247e-06,
      "loss": 0.0535,
      "step": 2034
    },
    {
      "epoch": 0.1050864962561322,
      "grad_norm": 6.443672180175781,
      "learning_rate": 9.774180565504719e-06,
      "loss": 0.0495,
      "step": 2035
    },
    {
      "epoch": 0.10513813581203202,
      "grad_norm": 8.704218864440918,
      "learning_rate": 9.773937042341984e-06,
      "loss": 0.0838,
      "step": 2036
    },
    {
      "epoch": 0.10518977536793184,
      "grad_norm": 10.431185722351074,
      "learning_rate": 9.773693390979582e-06,
      "loss": 0.0877,
      "step": 2037
    },
    {
      "epoch": 0.10524141492383166,
      "grad_norm": 3.288109540939331,
      "learning_rate": 9.773449611424054e-06,
      "loss": 0.0681,
      "step": 2038
    },
    {
      "epoch": 0.10529305447973147,
      "grad_norm": 3.204777956008911,
      "learning_rate": 9.773205703681952e-06,
      "loss": 0.0488,
      "step": 2039
    },
    {
      "epoch": 0.10534469403563129,
      "grad_norm": 7.992249011993408,
      "learning_rate": 9.77296166775982e-06,
      "loss": 0.0832,
      "step": 2040
    },
    {
      "epoch": 0.10539633359153111,
      "grad_norm": 23.229520797729492,
      "learning_rate": 9.772717503664214e-06,
      "loss": 0.1686,
      "step": 2041
    },
    {
      "epoch": 0.10544797314743093,
      "grad_norm": 8.382665634155273,
      "learning_rate": 9.77247321140169e-06,
      "loss": 0.0563,
      "step": 2042
    },
    {
      "epoch": 0.10549961270333075,
      "grad_norm": 18.887012481689453,
      "learning_rate": 9.77222879097881e-06,
      "loss": 0.1567,
      "step": 2043
    },
    {
      "epoch": 0.10555125225923057,
      "grad_norm": 11.232571601867676,
      "learning_rate": 9.771984242402136e-06,
      "loss": 0.0835,
      "step": 2044
    },
    {
      "epoch": 0.10560289181513038,
      "grad_norm": 8.03754997253418,
      "learning_rate": 9.771739565678235e-06,
      "loss": 0.0581,
      "step": 2045
    },
    {
      "epoch": 0.1056545313710302,
      "grad_norm": 3.4964401721954346,
      "learning_rate": 9.771494760813678e-06,
      "loss": 0.0756,
      "step": 2046
    },
    {
      "epoch": 0.10570617092693003,
      "grad_norm": 11.853285789489746,
      "learning_rate": 9.77124982781504e-06,
      "loss": 0.0935,
      "step": 2047
    },
    {
      "epoch": 0.10575781048282985,
      "grad_norm": 4.26107931137085,
      "learning_rate": 9.771004766688897e-06,
      "loss": 0.0569,
      "step": 2048
    },
    {
      "epoch": 0.10580945003872967,
      "grad_norm": 3.437391996383667,
      "learning_rate": 9.770759577441831e-06,
      "loss": 0.0676,
      "step": 2049
    },
    {
      "epoch": 0.10586108959462949,
      "grad_norm": 2.3016724586486816,
      "learning_rate": 9.770514260080425e-06,
      "loss": 0.0329,
      "step": 2050
    },
    {
      "epoch": 0.10591272915052931,
      "grad_norm": 4.373770236968994,
      "learning_rate": 9.770268814611268e-06,
      "loss": 0.0674,
      "step": 2051
    },
    {
      "epoch": 0.10596436870642913,
      "grad_norm": 8.030184745788574,
      "learning_rate": 9.770023241040951e-06,
      "loss": 0.2272,
      "step": 2052
    },
    {
      "epoch": 0.10601600826232895,
      "grad_norm": 18.8969669342041,
      "learning_rate": 9.769777539376069e-06,
      "loss": 0.1251,
      "step": 2053
    },
    {
      "epoch": 0.10606764781822876,
      "grad_norm": 1.6809977293014526,
      "learning_rate": 9.769531709623217e-06,
      "loss": 0.034,
      "step": 2054
    },
    {
      "epoch": 0.10611928737412858,
      "grad_norm": 3.9639079570770264,
      "learning_rate": 9.769285751789002e-06,
      "loss": 0.0575,
      "step": 2055
    },
    {
      "epoch": 0.1061709269300284,
      "grad_norm": 1.7444920539855957,
      "learning_rate": 9.769039665880024e-06,
      "loss": 0.0234,
      "step": 2056
    },
    {
      "epoch": 0.10622256648592822,
      "grad_norm": 4.807048320770264,
      "learning_rate": 9.768793451902895e-06,
      "loss": 0.0863,
      "step": 2057
    },
    {
      "epoch": 0.10627420604182804,
      "grad_norm": 3.4211246967315674,
      "learning_rate": 9.768547109864224e-06,
      "loss": 0.0585,
      "step": 2058
    },
    {
      "epoch": 0.10632584559772786,
      "grad_norm": 11.862258911132812,
      "learning_rate": 9.768300639770628e-06,
      "loss": 0.2281,
      "step": 2059
    },
    {
      "epoch": 0.10637748515362767,
      "grad_norm": 19.339588165283203,
      "learning_rate": 9.768054041628726e-06,
      "loss": 0.1306,
      "step": 2060
    },
    {
      "epoch": 0.10642912470952749,
      "grad_norm": 6.975948810577393,
      "learning_rate": 9.76780731544514e-06,
      "loss": 0.0643,
      "step": 2061
    },
    {
      "epoch": 0.10648076426542731,
      "grad_norm": 2.649651288986206,
      "learning_rate": 9.767560461226492e-06,
      "loss": 0.0226,
      "step": 2062
    },
    {
      "epoch": 0.10653240382132714,
      "grad_norm": 3.3834307193756104,
      "learning_rate": 9.767313478979416e-06,
      "loss": 0.055,
      "step": 2063
    },
    {
      "epoch": 0.10658404337722696,
      "grad_norm": 16.137128829956055,
      "learning_rate": 9.767066368710544e-06,
      "loss": 0.1431,
      "step": 2064
    },
    {
      "epoch": 0.10663568293312678,
      "grad_norm": 3.322272539138794,
      "learning_rate": 9.766819130426509e-06,
      "loss": 0.0696,
      "step": 2065
    },
    {
      "epoch": 0.1066873224890266,
      "grad_norm": 7.925389289855957,
      "learning_rate": 9.766571764133953e-06,
      "loss": 0.234,
      "step": 2066
    },
    {
      "epoch": 0.10673896204492642,
      "grad_norm": 12.332914352416992,
      "learning_rate": 9.766324269839515e-06,
      "loss": 0.1621,
      "step": 2067
    },
    {
      "epoch": 0.10679060160082623,
      "grad_norm": 4.431829452514648,
      "learning_rate": 9.766076647549846e-06,
      "loss": 0.0488,
      "step": 2068
    },
    {
      "epoch": 0.10684224115672605,
      "grad_norm": 18.694639205932617,
      "learning_rate": 9.765828897271595e-06,
      "loss": 0.2076,
      "step": 2069
    },
    {
      "epoch": 0.10689388071262587,
      "grad_norm": 6.831714153289795,
      "learning_rate": 9.76558101901141e-06,
      "loss": 0.0615,
      "step": 2070
    },
    {
      "epoch": 0.10694552026852569,
      "grad_norm": 3.7078731060028076,
      "learning_rate": 9.765333012775954e-06,
      "loss": 0.0631,
      "step": 2071
    },
    {
      "epoch": 0.10699715982442551,
      "grad_norm": 2.9625089168548584,
      "learning_rate": 9.765084878571884e-06,
      "loss": 0.0739,
      "step": 2072
    },
    {
      "epoch": 0.10704879938032533,
      "grad_norm": 5.413693904876709,
      "learning_rate": 9.764836616405861e-06,
      "loss": 0.0746,
      "step": 2073
    },
    {
      "epoch": 0.10710043893622515,
      "grad_norm": 18.310897827148438,
      "learning_rate": 9.764588226284557e-06,
      "loss": 0.1059,
      "step": 2074
    },
    {
      "epoch": 0.10715207849212496,
      "grad_norm": 8.432944297790527,
      "learning_rate": 9.76433970821464e-06,
      "loss": 0.0958,
      "step": 2075
    },
    {
      "epoch": 0.10720371804802478,
      "grad_norm": 5.505005359649658,
      "learning_rate": 9.764091062202782e-06,
      "loss": 0.1318,
      "step": 2076
    },
    {
      "epoch": 0.1072553576039246,
      "grad_norm": 13.190509796142578,
      "learning_rate": 9.763842288255662e-06,
      "loss": 0.1397,
      "step": 2077
    },
    {
      "epoch": 0.10730699715982442,
      "grad_norm": 5.8679914474487305,
      "learning_rate": 9.76359338637996e-06,
      "loss": 0.0877,
      "step": 2078
    },
    {
      "epoch": 0.10735863671572425,
      "grad_norm": 16.817384719848633,
      "learning_rate": 9.76334435658236e-06,
      "loss": 0.0941,
      "step": 2079
    },
    {
      "epoch": 0.10741027627162407,
      "grad_norm": 22.88298797607422,
      "learning_rate": 9.76309519886955e-06,
      "loss": 0.0953,
      "step": 2080
    },
    {
      "epoch": 0.10746191582752389,
      "grad_norm": 6.573721885681152,
      "learning_rate": 9.762845913248222e-06,
      "loss": 0.1026,
      "step": 2081
    },
    {
      "epoch": 0.1075135553834237,
      "grad_norm": 3.5872700214385986,
      "learning_rate": 9.762596499725067e-06,
      "loss": 0.0384,
      "step": 2082
    },
    {
      "epoch": 0.10756519493932352,
      "grad_norm": 13.002192497253418,
      "learning_rate": 9.762346958306787e-06,
      "loss": 0.1186,
      "step": 2083
    },
    {
      "epoch": 0.10761683449522334,
      "grad_norm": 11.97147274017334,
      "learning_rate": 9.762097289000079e-06,
      "loss": 0.13,
      "step": 2084
    },
    {
      "epoch": 0.10766847405112316,
      "grad_norm": 6.521052837371826,
      "learning_rate": 9.76184749181165e-06,
      "loss": 0.0942,
      "step": 2085
    },
    {
      "epoch": 0.10772011360702298,
      "grad_norm": 28.85398292541504,
      "learning_rate": 9.761597566748208e-06,
      "loss": 0.2997,
      "step": 2086
    },
    {
      "epoch": 0.1077717531629228,
      "grad_norm": 8.573394775390625,
      "learning_rate": 9.761347513816463e-06,
      "loss": 0.0639,
      "step": 2087
    },
    {
      "epoch": 0.10782339271882262,
      "grad_norm": 5.510049343109131,
      "learning_rate": 9.761097333023133e-06,
      "loss": 0.0707,
      "step": 2088
    },
    {
      "epoch": 0.10787503227472243,
      "grad_norm": 6.147914409637451,
      "learning_rate": 9.760847024374933e-06,
      "loss": 0.1194,
      "step": 2089
    },
    {
      "epoch": 0.10792667183062225,
      "grad_norm": 3.9046754837036133,
      "learning_rate": 9.760596587878586e-06,
      "loss": 0.0789,
      "step": 2090
    },
    {
      "epoch": 0.10797831138652207,
      "grad_norm": 5.170952796936035,
      "learning_rate": 9.760346023540819e-06,
      "loss": 0.0772,
      "step": 2091
    },
    {
      "epoch": 0.10802995094242189,
      "grad_norm": 10.47472858428955,
      "learning_rate": 9.760095331368357e-06,
      "loss": 0.1171,
      "step": 2092
    },
    {
      "epoch": 0.10808159049832171,
      "grad_norm": 13.118064880371094,
      "learning_rate": 9.759844511367937e-06,
      "loss": 0.0988,
      "step": 2093
    },
    {
      "epoch": 0.10813323005422153,
      "grad_norm": 4.261368751525879,
      "learning_rate": 9.75959356354629e-06,
      "loss": 0.0877,
      "step": 2094
    },
    {
      "epoch": 0.10818486961012136,
      "grad_norm": 4.844841480255127,
      "learning_rate": 9.759342487910158e-06,
      "loss": 0.0339,
      "step": 2095
    },
    {
      "epoch": 0.10823650916602118,
      "grad_norm": 13.524516105651855,
      "learning_rate": 9.75909128446628e-06,
      "loss": 0.1419,
      "step": 2096
    },
    {
      "epoch": 0.108288148721921,
      "grad_norm": 8.886316299438477,
      "learning_rate": 9.758839953221406e-06,
      "loss": 0.0874,
      "step": 2097
    },
    {
      "epoch": 0.10833978827782081,
      "grad_norm": 8.651788711547852,
      "learning_rate": 9.758588494182282e-06,
      "loss": 0.1419,
      "step": 2098
    },
    {
      "epoch": 0.10839142783372063,
      "grad_norm": 6.159842014312744,
      "learning_rate": 9.758336907355664e-06,
      "loss": 0.0831,
      "step": 2099
    },
    {
      "epoch": 0.10844306738962045,
      "grad_norm": 5.27017879486084,
      "learning_rate": 9.758085192748303e-06,
      "loss": 0.0667,
      "step": 2100
    },
    {
      "epoch": 0.10849470694552027,
      "grad_norm": 2.543051242828369,
      "learning_rate": 9.757833350366964e-06,
      "loss": 0.0519,
      "step": 2101
    },
    {
      "epoch": 0.10854634650142009,
      "grad_norm": 3.373077630996704,
      "learning_rate": 9.757581380218407e-06,
      "loss": 0.0712,
      "step": 2102
    },
    {
      "epoch": 0.1085979860573199,
      "grad_norm": 10.89390754699707,
      "learning_rate": 9.7573292823094e-06,
      "loss": 0.0754,
      "step": 2103
    },
    {
      "epoch": 0.10864962561321972,
      "grad_norm": 6.665633678436279,
      "learning_rate": 9.757077056646711e-06,
      "loss": 0.0875,
      "step": 2104
    },
    {
      "epoch": 0.10870126516911954,
      "grad_norm": 3.5958657264709473,
      "learning_rate": 9.756824703237114e-06,
      "loss": 0.0464,
      "step": 2105
    },
    {
      "epoch": 0.10875290472501936,
      "grad_norm": 4.116170883178711,
      "learning_rate": 9.756572222087386e-06,
      "loss": 0.0476,
      "step": 2106
    },
    {
      "epoch": 0.10880454428091918,
      "grad_norm": 4.5543532371521,
      "learning_rate": 9.756319613204309e-06,
      "loss": 0.0345,
      "step": 2107
    },
    {
      "epoch": 0.108856183836819,
      "grad_norm": 21.793842315673828,
      "learning_rate": 9.756066876594663e-06,
      "loss": 0.1596,
      "step": 2108
    },
    {
      "epoch": 0.10890782339271882,
      "grad_norm": 6.7096476554870605,
      "learning_rate": 9.755814012265236e-06,
      "loss": 0.0781,
      "step": 2109
    },
    {
      "epoch": 0.10895946294861864,
      "grad_norm": 5.751611709594727,
      "learning_rate": 9.755561020222822e-06,
      "loss": 0.0494,
      "step": 2110
    },
    {
      "epoch": 0.10901110250451847,
      "grad_norm": 10.916133880615234,
      "learning_rate": 9.75530790047421e-06,
      "loss": 0.1192,
      "step": 2111
    },
    {
      "epoch": 0.10906274206041829,
      "grad_norm": 5.660112380981445,
      "learning_rate": 9.755054653026198e-06,
      "loss": 0.0431,
      "step": 2112
    },
    {
      "epoch": 0.1091143816163181,
      "grad_norm": 10.03278923034668,
      "learning_rate": 9.75480127788559e-06,
      "loss": 0.0404,
      "step": 2113
    },
    {
      "epoch": 0.10916602117221792,
      "grad_norm": 20.372529983520508,
      "learning_rate": 9.754547775059187e-06,
      "loss": 0.1511,
      "step": 2114
    },
    {
      "epoch": 0.10921766072811774,
      "grad_norm": 9.500533103942871,
      "learning_rate": 9.7542941445538e-06,
      "loss": 0.0983,
      "step": 2115
    },
    {
      "epoch": 0.10926930028401756,
      "grad_norm": 4.202193737030029,
      "learning_rate": 9.754040386376235e-06,
      "loss": 0.0835,
      "step": 2116
    },
    {
      "epoch": 0.10932093983991738,
      "grad_norm": 14.881027221679688,
      "learning_rate": 9.753786500533312e-06,
      "loss": 0.075,
      "step": 2117
    },
    {
      "epoch": 0.1093725793958172,
      "grad_norm": 6.769052982330322,
      "learning_rate": 9.753532487031842e-06,
      "loss": 0.1202,
      "step": 2118
    },
    {
      "epoch": 0.10942421895171701,
      "grad_norm": 6.516646385192871,
      "learning_rate": 9.753278345878654e-06,
      "loss": 0.1181,
      "step": 2119
    },
    {
      "epoch": 0.10947585850761683,
      "grad_norm": 2.7799127101898193,
      "learning_rate": 9.753024077080568e-06,
      "loss": 0.0406,
      "step": 2120
    },
    {
      "epoch": 0.10952749806351665,
      "grad_norm": 6.271546840667725,
      "learning_rate": 9.752769680644414e-06,
      "loss": 0.0857,
      "step": 2121
    },
    {
      "epoch": 0.10957913761941647,
      "grad_norm": 4.846884250640869,
      "learning_rate": 9.752515156577023e-06,
      "loss": 0.0799,
      "step": 2122
    },
    {
      "epoch": 0.10963077717531629,
      "grad_norm": 6.9723405838012695,
      "learning_rate": 9.752260504885229e-06,
      "loss": 0.0445,
      "step": 2123
    },
    {
      "epoch": 0.1096824167312161,
      "grad_norm": 6.017851829528809,
      "learning_rate": 9.752005725575871e-06,
      "loss": 0.0627,
      "step": 2124
    },
    {
      "epoch": 0.10973405628711592,
      "grad_norm": 8.299386978149414,
      "learning_rate": 9.751750818655792e-06,
      "loss": 0.0771,
      "step": 2125
    },
    {
      "epoch": 0.10978569584301574,
      "grad_norm": 6.012026786804199,
      "learning_rate": 9.751495784131834e-06,
      "loss": 0.0978,
      "step": 2126
    },
    {
      "epoch": 0.10983733539891558,
      "grad_norm": 2.9087915420532227,
      "learning_rate": 9.75124062201085e-06,
      "loss": 0.0483,
      "step": 2127
    },
    {
      "epoch": 0.1098889749548154,
      "grad_norm": 6.7258219718933105,
      "learning_rate": 9.750985332299691e-06,
      "loss": 0.0991,
      "step": 2128
    },
    {
      "epoch": 0.10994061451071521,
      "grad_norm": 3.208906412124634,
      "learning_rate": 9.750729915005213e-06,
      "loss": 0.0779,
      "step": 2129
    },
    {
      "epoch": 0.10999225406661503,
      "grad_norm": 10.359899520874023,
      "learning_rate": 9.750474370134272e-06,
      "loss": 0.0999,
      "step": 2130
    },
    {
      "epoch": 0.11004389362251485,
      "grad_norm": 12.22747802734375,
      "learning_rate": 9.750218697693732e-06,
      "loss": 0.1374,
      "step": 2131
    },
    {
      "epoch": 0.11009553317841467,
      "grad_norm": 15.83980941772461,
      "learning_rate": 9.74996289769046e-06,
      "loss": 0.1653,
      "step": 2132
    },
    {
      "epoch": 0.11014717273431449,
      "grad_norm": 17.476655960083008,
      "learning_rate": 9.749706970131325e-06,
      "loss": 0.1134,
      "step": 2133
    },
    {
      "epoch": 0.1101988122902143,
      "grad_norm": 5.416986465454102,
      "learning_rate": 9.749450915023198e-06,
      "loss": 0.0636,
      "step": 2134
    },
    {
      "epoch": 0.11025045184611412,
      "grad_norm": 17.826244354248047,
      "learning_rate": 9.749194732372955e-06,
      "loss": 0.1536,
      "step": 2135
    },
    {
      "epoch": 0.11030209140201394,
      "grad_norm": 8.771352767944336,
      "learning_rate": 9.748938422187479e-06,
      "loss": 0.1287,
      "step": 2136
    },
    {
      "epoch": 0.11035373095791376,
      "grad_norm": 4.171964168548584,
      "learning_rate": 9.74868198447365e-06,
      "loss": 0.0814,
      "step": 2137
    },
    {
      "epoch": 0.11040537051381358,
      "grad_norm": 5.210385799407959,
      "learning_rate": 9.748425419238356e-06,
      "loss": 0.0584,
      "step": 2138
    },
    {
      "epoch": 0.1104570100697134,
      "grad_norm": 4.26218843460083,
      "learning_rate": 9.748168726488485e-06,
      "loss": 0.1041,
      "step": 2139
    },
    {
      "epoch": 0.11050864962561321,
      "grad_norm": 6.511321067810059,
      "learning_rate": 9.747911906230932e-06,
      "loss": 0.1091,
      "step": 2140
    },
    {
      "epoch": 0.11056028918151303,
      "grad_norm": 3.071866512298584,
      "learning_rate": 9.747654958472592e-06,
      "loss": 0.0327,
      "step": 2141
    },
    {
      "epoch": 0.11061192873741287,
      "grad_norm": 9.81618881225586,
      "learning_rate": 9.747397883220366e-06,
      "loss": 0.0673,
      "step": 2142
    },
    {
      "epoch": 0.11066356829331268,
      "grad_norm": 8.447175979614258,
      "learning_rate": 9.747140680481157e-06,
      "loss": 0.0834,
      "step": 2143
    },
    {
      "epoch": 0.1107152078492125,
      "grad_norm": 17.74113655090332,
      "learning_rate": 9.746883350261873e-06,
      "loss": 0.1235,
      "step": 2144
    },
    {
      "epoch": 0.11076684740511232,
      "grad_norm": 1.5690646171569824,
      "learning_rate": 9.746625892569426e-06,
      "loss": 0.0299,
      "step": 2145
    },
    {
      "epoch": 0.11081848696101214,
      "grad_norm": 6.73326301574707,
      "learning_rate": 9.746368307410724e-06,
      "loss": 0.0467,
      "step": 2146
    },
    {
      "epoch": 0.11087012651691196,
      "grad_norm": 6.625162124633789,
      "learning_rate": 9.74611059479269e-06,
      "loss": 0.062,
      "step": 2147
    },
    {
      "epoch": 0.11092176607281178,
      "grad_norm": 5.861991882324219,
      "learning_rate": 9.745852754722243e-06,
      "loss": 0.0588,
      "step": 2148
    },
    {
      "epoch": 0.1109734056287116,
      "grad_norm": 5.935067176818848,
      "learning_rate": 9.745594787206304e-06,
      "loss": 0.1196,
      "step": 2149
    },
    {
      "epoch": 0.11102504518461141,
      "grad_norm": 8.87548828125,
      "learning_rate": 9.745336692251806e-06,
      "loss": 0.1329,
      "step": 2150
    },
    {
      "epoch": 0.11107668474051123,
      "grad_norm": 11.12426471710205,
      "learning_rate": 9.745078469865673e-06,
      "loss": 0.0787,
      "step": 2151
    },
    {
      "epoch": 0.11112832429641105,
      "grad_norm": 18.462793350219727,
      "learning_rate": 9.744820120054848e-06,
      "loss": 0.1077,
      "step": 2152
    },
    {
      "epoch": 0.11117996385231087,
      "grad_norm": 5.673439025878906,
      "learning_rate": 9.74456164282626e-06,
      "loss": 0.0708,
      "step": 2153
    },
    {
      "epoch": 0.11123160340821069,
      "grad_norm": 5.191955089569092,
      "learning_rate": 9.744303038186857e-06,
      "loss": 0.107,
      "step": 2154
    },
    {
      "epoch": 0.1112832429641105,
      "grad_norm": 6.443633079528809,
      "learning_rate": 9.744044306143578e-06,
      "loss": 0.0561,
      "step": 2155
    },
    {
      "epoch": 0.11133488252001032,
      "grad_norm": 11.940361976623535,
      "learning_rate": 9.743785446703376e-06,
      "loss": 0.1085,
      "step": 2156
    },
    {
      "epoch": 0.11138652207591014,
      "grad_norm": 8.522798538208008,
      "learning_rate": 9.7435264598732e-06,
      "loss": 0.0579,
      "step": 2157
    },
    {
      "epoch": 0.11143816163180997,
      "grad_norm": 5.714020729064941,
      "learning_rate": 9.743267345660004e-06,
      "loss": 0.0789,
      "step": 2158
    },
    {
      "epoch": 0.11148980118770979,
      "grad_norm": 16.13605308532715,
      "learning_rate": 9.743008104070747e-06,
      "loss": 0.1568,
      "step": 2159
    },
    {
      "epoch": 0.11154144074360961,
      "grad_norm": 38.641815185546875,
      "learning_rate": 9.742748735112392e-06,
      "loss": 0.1095,
      "step": 2160
    },
    {
      "epoch": 0.11159308029950943,
      "grad_norm": 2.3577444553375244,
      "learning_rate": 9.742489238791902e-06,
      "loss": 0.0422,
      "step": 2161
    },
    {
      "epoch": 0.11164471985540925,
      "grad_norm": 7.300471305847168,
      "learning_rate": 9.742229615116247e-06,
      "loss": 0.0806,
      "step": 2162
    },
    {
      "epoch": 0.11169635941130907,
      "grad_norm": 12.916258811950684,
      "learning_rate": 9.7419698640924e-06,
      "loss": 0.1018,
      "step": 2163
    },
    {
      "epoch": 0.11174799896720888,
      "grad_norm": 5.118677139282227,
      "learning_rate": 9.741709985727331e-06,
      "loss": 0.0408,
      "step": 2164
    },
    {
      "epoch": 0.1117996385231087,
      "grad_norm": 7.8203253746032715,
      "learning_rate": 9.741449980028028e-06,
      "loss": 0.0631,
      "step": 2165
    },
    {
      "epoch": 0.11185127807900852,
      "grad_norm": 13.096534729003906,
      "learning_rate": 9.741189847001464e-06,
      "loss": 0.0457,
      "step": 2166
    },
    {
      "epoch": 0.11190291763490834,
      "grad_norm": 12.838448524475098,
      "learning_rate": 9.74092958665463e-06,
      "loss": 0.0779,
      "step": 2167
    },
    {
      "epoch": 0.11195455719080816,
      "grad_norm": 15.374188423156738,
      "learning_rate": 9.740669198994514e-06,
      "loss": 0.1261,
      "step": 2168
    },
    {
      "epoch": 0.11200619674670798,
      "grad_norm": 13.27197551727295,
      "learning_rate": 9.740408684028108e-06,
      "loss": 0.1693,
      "step": 2169
    },
    {
      "epoch": 0.1120578363026078,
      "grad_norm": 23.663053512573242,
      "learning_rate": 9.740148041762409e-06,
      "loss": 0.1397,
      "step": 2170
    },
    {
      "epoch": 0.11210947585850761,
      "grad_norm": 12.833771705627441,
      "learning_rate": 9.739887272204413e-06,
      "loss": 0.1882,
      "step": 2171
    },
    {
      "epoch": 0.11216111541440743,
      "grad_norm": 16.536592483520508,
      "learning_rate": 9.739626375361126e-06,
      "loss": 0.0994,
      "step": 2172
    },
    {
      "epoch": 0.11221275497030725,
      "grad_norm": 2.816948890686035,
      "learning_rate": 9.739365351239553e-06,
      "loss": 0.0319,
      "step": 2173
    },
    {
      "epoch": 0.11226439452620708,
      "grad_norm": 9.92776107788086,
      "learning_rate": 9.739104199846704e-06,
      "loss": 0.0948,
      "step": 2174
    },
    {
      "epoch": 0.1123160340821069,
      "grad_norm": 2.929654359817505,
      "learning_rate": 9.738842921189593e-06,
      "loss": 0.0406,
      "step": 2175
    },
    {
      "epoch": 0.11236767363800672,
      "grad_norm": 15.991009712219238,
      "learning_rate": 9.738581515275234e-06,
      "loss": 0.1425,
      "step": 2176
    },
    {
      "epoch": 0.11241931319390654,
      "grad_norm": 6.163272380828857,
      "learning_rate": 9.738319982110646e-06,
      "loss": 0.0843,
      "step": 2177
    },
    {
      "epoch": 0.11247095274980635,
      "grad_norm": 4.0493597984313965,
      "learning_rate": 9.738058321702857e-06,
      "loss": 0.0623,
      "step": 2178
    },
    {
      "epoch": 0.11252259230570617,
      "grad_norm": 3.4684197902679443,
      "learning_rate": 9.73779653405889e-06,
      "loss": 0.055,
      "step": 2179
    },
    {
      "epoch": 0.11257423186160599,
      "grad_norm": 4.526513576507568,
      "learning_rate": 9.737534619185776e-06,
      "loss": 0.051,
      "step": 2180
    },
    {
      "epoch": 0.11262587141750581,
      "grad_norm": 10.505648612976074,
      "learning_rate": 9.737272577090548e-06,
      "loss": 0.0921,
      "step": 2181
    },
    {
      "epoch": 0.11267751097340563,
      "grad_norm": 3.415388584136963,
      "learning_rate": 9.737010407780242e-06,
      "loss": 0.045,
      "step": 2182
    },
    {
      "epoch": 0.11272915052930545,
      "grad_norm": 6.06058406829834,
      "learning_rate": 9.736748111261902e-06,
      "loss": 0.0333,
      "step": 2183
    },
    {
      "epoch": 0.11278079008520527,
      "grad_norm": 11.058216094970703,
      "learning_rate": 9.736485687542567e-06,
      "loss": 0.082,
      "step": 2184
    },
    {
      "epoch": 0.11283242964110508,
      "grad_norm": 11.906684875488281,
      "learning_rate": 9.736223136629289e-06,
      "loss": 0.1207,
      "step": 2185
    },
    {
      "epoch": 0.1128840691970049,
      "grad_norm": 7.429265975952148,
      "learning_rate": 9.735960458529114e-06,
      "loss": 0.0688,
      "step": 2186
    },
    {
      "epoch": 0.11293570875290472,
      "grad_norm": 15.807171821594238,
      "learning_rate": 9.7356976532491e-06,
      "loss": 0.3971,
      "step": 2187
    },
    {
      "epoch": 0.11298734830880454,
      "grad_norm": 10.129935264587402,
      "learning_rate": 9.7354347207963e-06,
      "loss": 0.1174,
      "step": 2188
    },
    {
      "epoch": 0.11303898786470436,
      "grad_norm": 2.3207437992095947,
      "learning_rate": 9.735171661177776e-06,
      "loss": 0.0506,
      "step": 2189
    },
    {
      "epoch": 0.11309062742060419,
      "grad_norm": 3.812859535217285,
      "learning_rate": 9.734908474400598e-06,
      "loss": 0.051,
      "step": 2190
    },
    {
      "epoch": 0.11314226697650401,
      "grad_norm": 18.167192459106445,
      "learning_rate": 9.734645160471825e-06,
      "loss": 0.3071,
      "step": 2191
    },
    {
      "epoch": 0.11319390653240383,
      "grad_norm": 9.720836639404297,
      "learning_rate": 9.734381719398534e-06,
      "loss": 0.0913,
      "step": 2192
    },
    {
      "epoch": 0.11324554608830364,
      "grad_norm": 6.713670253753662,
      "learning_rate": 9.734118151187797e-06,
      "loss": 0.1002,
      "step": 2193
    },
    {
      "epoch": 0.11329718564420346,
      "grad_norm": 7.859821796417236,
      "learning_rate": 9.733854455846692e-06,
      "loss": 0.0854,
      "step": 2194
    },
    {
      "epoch": 0.11334882520010328,
      "grad_norm": 7.923798561096191,
      "learning_rate": 9.733590633382302e-06,
      "loss": 0.0382,
      "step": 2195
    },
    {
      "epoch": 0.1134004647560031,
      "grad_norm": 42.5020751953125,
      "learning_rate": 9.733326683801709e-06,
      "loss": 0.3655,
      "step": 2196
    },
    {
      "epoch": 0.11345210431190292,
      "grad_norm": 27.46475601196289,
      "learning_rate": 9.733062607112004e-06,
      "loss": 0.2893,
      "step": 2197
    },
    {
      "epoch": 0.11350374386780274,
      "grad_norm": 21.494335174560547,
      "learning_rate": 9.732798403320276e-06,
      "loss": 0.2162,
      "step": 2198
    },
    {
      "epoch": 0.11355538342370256,
      "grad_norm": 23.280000686645508,
      "learning_rate": 9.732534072433621e-06,
      "loss": 0.2708,
      "step": 2199
    },
    {
      "epoch": 0.11360702297960237,
      "grad_norm": 13.37463092803955,
      "learning_rate": 9.732269614459139e-06,
      "loss": 0.2194,
      "step": 2200
    },
    {
      "epoch": 0.11365866253550219,
      "grad_norm": 14.857880592346191,
      "learning_rate": 9.732005029403929e-06,
      "loss": 0.1841,
      "step": 2201
    },
    {
      "epoch": 0.11371030209140201,
      "grad_norm": 5.486883163452148,
      "learning_rate": 9.731740317275098e-06,
      "loss": 0.1299,
      "step": 2202
    },
    {
      "epoch": 0.11376194164730183,
      "grad_norm": 14.090507507324219,
      "learning_rate": 9.731475478079752e-06,
      "loss": 0.166,
      "step": 2203
    },
    {
      "epoch": 0.11381358120320165,
      "grad_norm": 16.459945678710938,
      "learning_rate": 9.731210511825008e-06,
      "loss": 0.1761,
      "step": 2204
    },
    {
      "epoch": 0.11386522075910147,
      "grad_norm": 18.00110626220703,
      "learning_rate": 9.730945418517976e-06,
      "loss": 0.1037,
      "step": 2205
    },
    {
      "epoch": 0.1139168603150013,
      "grad_norm": 7.429619312286377,
      "learning_rate": 9.730680198165779e-06,
      "loss": 0.1373,
      "step": 2206
    },
    {
      "epoch": 0.11396849987090112,
      "grad_norm": 9.76856517791748,
      "learning_rate": 9.730414850775536e-06,
      "loss": 0.1003,
      "step": 2207
    },
    {
      "epoch": 0.11402013942680093,
      "grad_norm": 29.092653274536133,
      "learning_rate": 9.730149376354375e-06,
      "loss": 0.2122,
      "step": 2208
    },
    {
      "epoch": 0.11407177898270075,
      "grad_norm": 16.067989349365234,
      "learning_rate": 9.729883774909423e-06,
      "loss": 0.2021,
      "step": 2209
    },
    {
      "epoch": 0.11412341853860057,
      "grad_norm": 7.238290309906006,
      "learning_rate": 9.729618046447816e-06,
      "loss": 0.1124,
      "step": 2210
    },
    {
      "epoch": 0.11417505809450039,
      "grad_norm": 10.222694396972656,
      "learning_rate": 9.729352190976686e-06,
      "loss": 0.091,
      "step": 2211
    },
    {
      "epoch": 0.11422669765040021,
      "grad_norm": 4.076258182525635,
      "learning_rate": 9.729086208503174e-06,
      "loss": 0.0615,
      "step": 2212
    },
    {
      "epoch": 0.11427833720630003,
      "grad_norm": 17.148096084594727,
      "learning_rate": 9.728820099034424e-06,
      "loss": 0.1988,
      "step": 2213
    },
    {
      "epoch": 0.11432997676219984,
      "grad_norm": 11.148979187011719,
      "learning_rate": 9.728553862577579e-06,
      "loss": 0.1649,
      "step": 2214
    },
    {
      "epoch": 0.11438161631809966,
      "grad_norm": 7.7764506340026855,
      "learning_rate": 9.72828749913979e-06,
      "loss": 0.067,
      "step": 2215
    },
    {
      "epoch": 0.11443325587399948,
      "grad_norm": 6.100424766540527,
      "learning_rate": 9.728021008728212e-06,
      "loss": 0.1025,
      "step": 2216
    },
    {
      "epoch": 0.1144848954298993,
      "grad_norm": 8.306342124938965,
      "learning_rate": 9.727754391349997e-06,
      "loss": 0.1613,
      "step": 2217
    },
    {
      "epoch": 0.11453653498579912,
      "grad_norm": 6.538282871246338,
      "learning_rate": 9.727487647012311e-06,
      "loss": 0.0919,
      "step": 2218
    },
    {
      "epoch": 0.11458817454169894,
      "grad_norm": 11.223798751831055,
      "learning_rate": 9.727220775722311e-06,
      "loss": 0.1072,
      "step": 2219
    },
    {
      "epoch": 0.11463981409759876,
      "grad_norm": 7.08054780960083,
      "learning_rate": 9.726953777487166e-06,
      "loss": 0.1929,
      "step": 2220
    },
    {
      "epoch": 0.11469145365349857,
      "grad_norm": 5.507378578186035,
      "learning_rate": 9.726686652314045e-06,
      "loss": 0.1118,
      "step": 2221
    },
    {
      "epoch": 0.1147430932093984,
      "grad_norm": 14.44413948059082,
      "learning_rate": 9.726419400210126e-06,
      "loss": 0.1759,
      "step": 2222
    },
    {
      "epoch": 0.11479473276529822,
      "grad_norm": 6.235458850860596,
      "learning_rate": 9.72615202118258e-06,
      "loss": 0.0942,
      "step": 2223
    },
    {
      "epoch": 0.11484637232119804,
      "grad_norm": 20.074003219604492,
      "learning_rate": 9.725884515238589e-06,
      "loss": 0.1772,
      "step": 2224
    },
    {
      "epoch": 0.11489801187709786,
      "grad_norm": 5.937751293182373,
      "learning_rate": 9.725616882385339e-06,
      "loss": 0.0703,
      "step": 2225
    },
    {
      "epoch": 0.11494965143299768,
      "grad_norm": 7.22321891784668,
      "learning_rate": 9.725349122630014e-06,
      "loss": 0.1104,
      "step": 2226
    },
    {
      "epoch": 0.1150012909888975,
      "grad_norm": 11.319555282592773,
      "learning_rate": 9.725081235979805e-06,
      "loss": 0.0947,
      "step": 2227
    },
    {
      "epoch": 0.11505293054479732,
      "grad_norm": 10.534887313842773,
      "learning_rate": 9.724813222441908e-06,
      "loss": 0.1686,
      "step": 2228
    },
    {
      "epoch": 0.11510457010069713,
      "grad_norm": 10.271413803100586,
      "learning_rate": 9.724545082023518e-06,
      "loss": 0.1512,
      "step": 2229
    },
    {
      "epoch": 0.11515620965659695,
      "grad_norm": 7.21909761428833,
      "learning_rate": 9.724276814731837e-06,
      "loss": 0.1154,
      "step": 2230
    },
    {
      "epoch": 0.11520784921249677,
      "grad_norm": 7.2498955726623535,
      "learning_rate": 9.724008420574066e-06,
      "loss": 0.1229,
      "step": 2231
    },
    {
      "epoch": 0.11525948876839659,
      "grad_norm": 7.847073078155518,
      "learning_rate": 9.723739899557418e-06,
      "loss": 0.1004,
      "step": 2232
    },
    {
      "epoch": 0.11531112832429641,
      "grad_norm": 10.267163276672363,
      "learning_rate": 9.723471251689098e-06,
      "loss": 0.1171,
      "step": 2233
    },
    {
      "epoch": 0.11536276788019623,
      "grad_norm": 7.424792766571045,
      "learning_rate": 9.723202476976325e-06,
      "loss": 0.1274,
      "step": 2234
    },
    {
      "epoch": 0.11541440743609604,
      "grad_norm": 4.786654949188232,
      "learning_rate": 9.722933575426315e-06,
      "loss": 0.0877,
      "step": 2235
    },
    {
      "epoch": 0.11546604699199586,
      "grad_norm": 5.672687530517578,
      "learning_rate": 9.722664547046287e-06,
      "loss": 0.1143,
      "step": 2236
    },
    {
      "epoch": 0.11551768654789568,
      "grad_norm": 26.47737693786621,
      "learning_rate": 9.722395391843469e-06,
      "loss": 0.1915,
      "step": 2237
    },
    {
      "epoch": 0.11556932610379551,
      "grad_norm": 4.70955753326416,
      "learning_rate": 9.722126109825088e-06,
      "loss": 0.0557,
      "step": 2238
    },
    {
      "epoch": 0.11562096565969533,
      "grad_norm": 6.13336181640625,
      "learning_rate": 9.721856700998374e-06,
      "loss": 0.0342,
      "step": 2239
    },
    {
      "epoch": 0.11567260521559515,
      "grad_norm": 3.7183146476745605,
      "learning_rate": 9.72158716537056e-06,
      "loss": 0.0783,
      "step": 2240
    },
    {
      "epoch": 0.11572424477149497,
      "grad_norm": 7.013020038604736,
      "learning_rate": 9.721317502948888e-06,
      "loss": 0.0662,
      "step": 2241
    },
    {
      "epoch": 0.11577588432739479,
      "grad_norm": 6.472658634185791,
      "learning_rate": 9.7210477137406e-06,
      "loss": 0.0832,
      "step": 2242
    },
    {
      "epoch": 0.1158275238832946,
      "grad_norm": 6.178666591644287,
      "learning_rate": 9.720777797752937e-06,
      "loss": 0.0842,
      "step": 2243
    },
    {
      "epoch": 0.11587916343919442,
      "grad_norm": 7.263901710510254,
      "learning_rate": 9.72050775499315e-06,
      "loss": 0.0779,
      "step": 2244
    },
    {
      "epoch": 0.11593080299509424,
      "grad_norm": 16.57001304626465,
      "learning_rate": 9.720237585468489e-06,
      "loss": 0.1416,
      "step": 2245
    },
    {
      "epoch": 0.11598244255099406,
      "grad_norm": 11.435973167419434,
      "learning_rate": 9.719967289186211e-06,
      "loss": 0.1519,
      "step": 2246
    },
    {
      "epoch": 0.11603408210689388,
      "grad_norm": 17.130216598510742,
      "learning_rate": 9.719696866153574e-06,
      "loss": 0.1385,
      "step": 2247
    },
    {
      "epoch": 0.1160857216627937,
      "grad_norm": 20.21950340270996,
      "learning_rate": 9.719426316377837e-06,
      "loss": 0.2311,
      "step": 2248
    },
    {
      "epoch": 0.11613736121869352,
      "grad_norm": 10.714497566223145,
      "learning_rate": 9.719155639866273e-06,
      "loss": 0.1302,
      "step": 2249
    },
    {
      "epoch": 0.11618900077459333,
      "grad_norm": 18.659183502197266,
      "learning_rate": 9.718884836626142e-06,
      "loss": 0.1554,
      "step": 2250
    },
    {
      "epoch": 0.11624064033049315,
      "grad_norm": 8.360527992248535,
      "learning_rate": 9.718613906664722e-06,
      "loss": 0.1071,
      "step": 2251
    },
    {
      "epoch": 0.11629227988639297,
      "grad_norm": 5.729193210601807,
      "learning_rate": 9.718342849989287e-06,
      "loss": 0.0998,
      "step": 2252
    },
    {
      "epoch": 0.11634391944229279,
      "grad_norm": 4.576088905334473,
      "learning_rate": 9.718071666607115e-06,
      "loss": 0.0708,
      "step": 2253
    },
    {
      "epoch": 0.11639555899819262,
      "grad_norm": 14.730917930603027,
      "learning_rate": 9.717800356525487e-06,
      "loss": 0.1171,
      "step": 2254
    },
    {
      "epoch": 0.11644719855409244,
      "grad_norm": 27.544933319091797,
      "learning_rate": 9.717528919751693e-06,
      "loss": 0.1523,
      "step": 2255
    },
    {
      "epoch": 0.11649883810999226,
      "grad_norm": 16.13861846923828,
      "learning_rate": 9.717257356293017e-06,
      "loss": 0.1805,
      "step": 2256
    },
    {
      "epoch": 0.11655047766589208,
      "grad_norm": 15.97384262084961,
      "learning_rate": 9.716985666156757e-06,
      "loss": 0.1793,
      "step": 2257
    },
    {
      "epoch": 0.1166021172217919,
      "grad_norm": 8.20817756652832,
      "learning_rate": 9.716713849350206e-06,
      "loss": 0.1317,
      "step": 2258
    },
    {
      "epoch": 0.11665375677769171,
      "grad_norm": 8.05629825592041,
      "learning_rate": 9.716441905880664e-06,
      "loss": 0.0885,
      "step": 2259
    },
    {
      "epoch": 0.11670539633359153,
      "grad_norm": 19.04828643798828,
      "learning_rate": 9.716169835755434e-06,
      "loss": 0.1083,
      "step": 2260
    },
    {
      "epoch": 0.11675703588949135,
      "grad_norm": 21.06294059753418,
      "learning_rate": 9.715897638981821e-06,
      "loss": 0.1197,
      "step": 2261
    },
    {
      "epoch": 0.11680867544539117,
      "grad_norm": 9.772872924804688,
      "learning_rate": 9.715625315567136e-06,
      "loss": 0.152,
      "step": 2262
    },
    {
      "epoch": 0.11686031500129099,
      "grad_norm": 11.94089412689209,
      "learning_rate": 9.715352865518693e-06,
      "loss": 0.121,
      "step": 2263
    },
    {
      "epoch": 0.1169119545571908,
      "grad_norm": 12.643773078918457,
      "learning_rate": 9.715080288843805e-06,
      "loss": 0.1504,
      "step": 2264
    },
    {
      "epoch": 0.11696359411309062,
      "grad_norm": 16.228893280029297,
      "learning_rate": 9.714807585549793e-06,
      "loss": 0.203,
      "step": 2265
    },
    {
      "epoch": 0.11701523366899044,
      "grad_norm": 6.24453067779541,
      "learning_rate": 9.714534755643982e-06,
      "loss": 0.0864,
      "step": 2266
    },
    {
      "epoch": 0.11706687322489026,
      "grad_norm": 9.784741401672363,
      "learning_rate": 9.714261799133698e-06,
      "loss": 0.0961,
      "step": 2267
    },
    {
      "epoch": 0.11711851278079008,
      "grad_norm": 7.6266913414001465,
      "learning_rate": 9.71398871602627e-06,
      "loss": 0.0803,
      "step": 2268
    },
    {
      "epoch": 0.1171701523366899,
      "grad_norm": 2.291334629058838,
      "learning_rate": 9.71371550632903e-06,
      "loss": 0.0513,
      "step": 2269
    },
    {
      "epoch": 0.11722179189258973,
      "grad_norm": 10.392115592956543,
      "learning_rate": 9.713442170049319e-06,
      "loss": 0.2072,
      "step": 2270
    },
    {
      "epoch": 0.11727343144848955,
      "grad_norm": 13.38952350616455,
      "learning_rate": 9.713168707194473e-06,
      "loss": 0.0979,
      "step": 2271
    },
    {
      "epoch": 0.11732507100438937,
      "grad_norm": 3.8729326725006104,
      "learning_rate": 9.712895117771838e-06,
      "loss": 0.0722,
      "step": 2272
    },
    {
      "epoch": 0.11737671056028919,
      "grad_norm": 4.1583099365234375,
      "learning_rate": 9.712621401788762e-06,
      "loss": 0.0783,
      "step": 2273
    },
    {
      "epoch": 0.117428350116189,
      "grad_norm": 7.188017845153809,
      "learning_rate": 9.712347559252592e-06,
      "loss": 0.0946,
      "step": 2274
    },
    {
      "epoch": 0.11747998967208882,
      "grad_norm": 8.47387409210205,
      "learning_rate": 9.712073590170685e-06,
      "loss": 0.1399,
      "step": 2275
    },
    {
      "epoch": 0.11753162922798864,
      "grad_norm": 9.953566551208496,
      "learning_rate": 9.711799494550396e-06,
      "loss": 0.101,
      "step": 2276
    },
    {
      "epoch": 0.11758326878388846,
      "grad_norm": 5.734143257141113,
      "learning_rate": 9.711525272399084e-06,
      "loss": 0.0537,
      "step": 2277
    },
    {
      "epoch": 0.11763490833978828,
      "grad_norm": 23.32516860961914,
      "learning_rate": 9.711250923724118e-06,
      "loss": 0.1267,
      "step": 2278
    },
    {
      "epoch": 0.1176865478956881,
      "grad_norm": 10.245033264160156,
      "learning_rate": 9.710976448532863e-06,
      "loss": 0.0557,
      "step": 2279
    },
    {
      "epoch": 0.11773818745158791,
      "grad_norm": 10.164896965026855,
      "learning_rate": 9.710701846832688e-06,
      "loss": 0.0547,
      "step": 2280
    },
    {
      "epoch": 0.11778982700748773,
      "grad_norm": 2.3997764587402344,
      "learning_rate": 9.710427118630969e-06,
      "loss": 0.0709,
      "step": 2281
    },
    {
      "epoch": 0.11784146656338755,
      "grad_norm": 4.568481922149658,
      "learning_rate": 9.710152263935085e-06,
      "loss": 0.0737,
      "step": 2282
    },
    {
      "epoch": 0.11789310611928737,
      "grad_norm": 10.809128761291504,
      "learning_rate": 9.709877282752414e-06,
      "loss": 0.0809,
      "step": 2283
    },
    {
      "epoch": 0.11794474567518719,
      "grad_norm": 5.179327011108398,
      "learning_rate": 9.70960217509034e-06,
      "loss": 0.1109,
      "step": 2284
    },
    {
      "epoch": 0.117996385231087,
      "grad_norm": 3.9379167556762695,
      "learning_rate": 9.709326940956255e-06,
      "loss": 0.0724,
      "step": 2285
    },
    {
      "epoch": 0.11804802478698684,
      "grad_norm": 2.8654749393463135,
      "learning_rate": 9.709051580357545e-06,
      "loss": 0.072,
      "step": 2286
    },
    {
      "epoch": 0.11809966434288666,
      "grad_norm": 6.107388019561768,
      "learning_rate": 9.708776093301608e-06,
      "loss": 0.0784,
      "step": 2287
    },
    {
      "epoch": 0.11815130389878647,
      "grad_norm": 3.0791120529174805,
      "learning_rate": 9.708500479795842e-06,
      "loss": 0.0345,
      "step": 2288
    },
    {
      "epoch": 0.1182029434546863,
      "grad_norm": 4.918750762939453,
      "learning_rate": 9.708224739847645e-06,
      "loss": 0.0877,
      "step": 2289
    },
    {
      "epoch": 0.11825458301058611,
      "grad_norm": 10.124784469604492,
      "learning_rate": 9.707948873464426e-06,
      "loss": 0.0871,
      "step": 2290
    },
    {
      "epoch": 0.11830622256648593,
      "grad_norm": 5.1261067390441895,
      "learning_rate": 9.707672880653591e-06,
      "loss": 0.0531,
      "step": 2291
    },
    {
      "epoch": 0.11835786212238575,
      "grad_norm": 6.642049312591553,
      "learning_rate": 9.707396761422552e-06,
      "loss": 0.0756,
      "step": 2292
    },
    {
      "epoch": 0.11840950167828557,
      "grad_norm": 22.21722412109375,
      "learning_rate": 9.707120515778723e-06,
      "loss": 0.1444,
      "step": 2293
    },
    {
      "epoch": 0.11846114123418539,
      "grad_norm": 15.815106391906738,
      "learning_rate": 9.706844143729523e-06,
      "loss": 0.0747,
      "step": 2294
    },
    {
      "epoch": 0.1185127807900852,
      "grad_norm": 14.655130386352539,
      "learning_rate": 9.706567645282373e-06,
      "loss": 0.0965,
      "step": 2295
    },
    {
      "epoch": 0.11856442034598502,
      "grad_norm": 7.237361431121826,
      "learning_rate": 9.7062910204447e-06,
      "loss": 0.0672,
      "step": 2296
    },
    {
      "epoch": 0.11861605990188484,
      "grad_norm": 8.67096996307373,
      "learning_rate": 9.706014269223932e-06,
      "loss": 0.0985,
      "step": 2297
    },
    {
      "epoch": 0.11866769945778466,
      "grad_norm": 7.317073345184326,
      "learning_rate": 9.705737391627498e-06,
      "loss": 0.057,
      "step": 2298
    },
    {
      "epoch": 0.11871933901368448,
      "grad_norm": 9.991825103759766,
      "learning_rate": 9.705460387662836e-06,
      "loss": 0.1067,
      "step": 2299
    },
    {
      "epoch": 0.1187709785695843,
      "grad_norm": 3.1468913555145264,
      "learning_rate": 9.705183257337385e-06,
      "loss": 0.04,
      "step": 2300
    },
    {
      "epoch": 0.11882261812548411,
      "grad_norm": 19.312976837158203,
      "learning_rate": 9.704906000658587e-06,
      "loss": 0.1393,
      "step": 2301
    },
    {
      "epoch": 0.11887425768138395,
      "grad_norm": 6.334941387176514,
      "learning_rate": 9.704628617633884e-06,
      "loss": 0.0768,
      "step": 2302
    },
    {
      "epoch": 0.11892589723728376,
      "grad_norm": 26.089441299438477,
      "learning_rate": 9.70435110827073e-06,
      "loss": 0.1808,
      "step": 2303
    },
    {
      "epoch": 0.11897753679318358,
      "grad_norm": 11.365703582763672,
      "learning_rate": 9.704073472576575e-06,
      "loss": 0.2328,
      "step": 2304
    },
    {
      "epoch": 0.1190291763490834,
      "grad_norm": 12.55207347869873,
      "learning_rate": 9.703795710558875e-06,
      "loss": 0.0994,
      "step": 2305
    },
    {
      "epoch": 0.11908081590498322,
      "grad_norm": 7.369563579559326,
      "learning_rate": 9.703517822225086e-06,
      "loss": 0.0872,
      "step": 2306
    },
    {
      "epoch": 0.11913245546088304,
      "grad_norm": 10.00648021697998,
      "learning_rate": 9.703239807582675e-06,
      "loss": 0.1505,
      "step": 2307
    },
    {
      "epoch": 0.11918409501678286,
      "grad_norm": 7.587038516998291,
      "learning_rate": 9.702961666639106e-06,
      "loss": 0.1112,
      "step": 2308
    },
    {
      "epoch": 0.11923573457268268,
      "grad_norm": 11.053899765014648,
      "learning_rate": 9.702683399401847e-06,
      "loss": 0.0994,
      "step": 2309
    },
    {
      "epoch": 0.1192873741285825,
      "grad_norm": 5.144514560699463,
      "learning_rate": 9.702405005878372e-06,
      "loss": 0.0758,
      "step": 2310
    },
    {
      "epoch": 0.11933901368448231,
      "grad_norm": 5.521047592163086,
      "learning_rate": 9.702126486076156e-06,
      "loss": 0.0962,
      "step": 2311
    },
    {
      "epoch": 0.11939065324038213,
      "grad_norm": 8.011492729187012,
      "learning_rate": 9.70184784000268e-06,
      "loss": 0.109,
      "step": 2312
    },
    {
      "epoch": 0.11944229279628195,
      "grad_norm": 4.72354793548584,
      "learning_rate": 9.701569067665426e-06,
      "loss": 0.0729,
      "step": 2313
    },
    {
      "epoch": 0.11949393235218177,
      "grad_norm": 7.839730262756348,
      "learning_rate": 9.701290169071878e-06,
      "loss": 0.0755,
      "step": 2314
    },
    {
      "epoch": 0.11954557190808159,
      "grad_norm": 13.97143268585205,
      "learning_rate": 9.70101114422953e-06,
      "loss": 0.1487,
      "step": 2315
    },
    {
      "epoch": 0.1195972114639814,
      "grad_norm": 6.346673488616943,
      "learning_rate": 9.700731993145871e-06,
      "loss": 0.0793,
      "step": 2316
    },
    {
      "epoch": 0.11964885101988122,
      "grad_norm": 25.694822311401367,
      "learning_rate": 9.7004527158284e-06,
      "loss": 0.159,
      "step": 2317
    },
    {
      "epoch": 0.11970049057578105,
      "grad_norm": 7.470319747924805,
      "learning_rate": 9.700173312284614e-06,
      "loss": 0.128,
      "step": 2318
    },
    {
      "epoch": 0.11975213013168087,
      "grad_norm": 4.163618087768555,
      "learning_rate": 9.699893782522018e-06,
      "loss": 0.0505,
      "step": 2319
    },
    {
      "epoch": 0.11980376968758069,
      "grad_norm": 7.459300994873047,
      "learning_rate": 9.699614126548121e-06,
      "loss": 0.0985,
      "step": 2320
    },
    {
      "epoch": 0.11985540924348051,
      "grad_norm": 6.844574451446533,
      "learning_rate": 9.699334344370428e-06,
      "loss": 0.084,
      "step": 2321
    },
    {
      "epoch": 0.11990704879938033,
      "grad_norm": 5.479592323303223,
      "learning_rate": 9.699054435996453e-06,
      "loss": 0.0426,
      "step": 2322
    },
    {
      "epoch": 0.11995868835528015,
      "grad_norm": 4.987851142883301,
      "learning_rate": 9.698774401433716e-06,
      "loss": 0.0941,
      "step": 2323
    },
    {
      "epoch": 0.12001032791117996,
      "grad_norm": 9.216278076171875,
      "learning_rate": 9.698494240689735e-06,
      "loss": 0.0852,
      "step": 2324
    },
    {
      "epoch": 0.12006196746707978,
      "grad_norm": 12.225676536560059,
      "learning_rate": 9.698213953772033e-06,
      "loss": 0.1132,
      "step": 2325
    },
    {
      "epoch": 0.1201136070229796,
      "grad_norm": 12.75866413116455,
      "learning_rate": 9.697933540688138e-06,
      "loss": 0.1438,
      "step": 2326
    },
    {
      "epoch": 0.12016524657887942,
      "grad_norm": 2.1520731449127197,
      "learning_rate": 9.697653001445579e-06,
      "loss": 0.0488,
      "step": 2327
    },
    {
      "epoch": 0.12021688613477924,
      "grad_norm": 7.919856548309326,
      "learning_rate": 9.697372336051892e-06,
      "loss": 0.1335,
      "step": 2328
    },
    {
      "epoch": 0.12026852569067906,
      "grad_norm": 12.927034378051758,
      "learning_rate": 9.697091544514611e-06,
      "loss": 0.1142,
      "step": 2329
    },
    {
      "epoch": 0.12032016524657888,
      "grad_norm": 10.32867431640625,
      "learning_rate": 9.696810626841277e-06,
      "loss": 0.0896,
      "step": 2330
    },
    {
      "epoch": 0.1203718048024787,
      "grad_norm": 4.757408142089844,
      "learning_rate": 9.696529583039438e-06,
      "loss": 0.0642,
      "step": 2331
    },
    {
      "epoch": 0.12042344435837851,
      "grad_norm": 2.121429204940796,
      "learning_rate": 9.696248413116634e-06,
      "loss": 0.0444,
      "step": 2332
    },
    {
      "epoch": 0.12047508391427833,
      "grad_norm": 24.081787109375,
      "learning_rate": 9.69596711708042e-06,
      "loss": 0.2195,
      "step": 2333
    },
    {
      "epoch": 0.12052672347017816,
      "grad_norm": 3.5328304767608643,
      "learning_rate": 9.695685694938351e-06,
      "loss": 0.0652,
      "step": 2334
    },
    {
      "epoch": 0.12057836302607798,
      "grad_norm": 31.85869026184082,
      "learning_rate": 9.695404146697981e-06,
      "loss": 0.2221,
      "step": 2335
    },
    {
      "epoch": 0.1206300025819778,
      "grad_norm": 5.891547203063965,
      "learning_rate": 9.695122472366873e-06,
      "loss": 0.0712,
      "step": 2336
    },
    {
      "epoch": 0.12068164213787762,
      "grad_norm": 6.707433223724365,
      "learning_rate": 9.694840671952589e-06,
      "loss": 0.1136,
      "step": 2337
    },
    {
      "epoch": 0.12073328169377744,
      "grad_norm": 28.46772575378418,
      "learning_rate": 9.6945587454627e-06,
      "loss": 0.0992,
      "step": 2338
    },
    {
      "epoch": 0.12078492124967725,
      "grad_norm": 7.644097805023193,
      "learning_rate": 9.694276692904771e-06,
      "loss": 0.0777,
      "step": 2339
    },
    {
      "epoch": 0.12083656080557707,
      "grad_norm": 2.488215208053589,
      "learning_rate": 9.693994514286385e-06,
      "loss": 0.0324,
      "step": 2340
    },
    {
      "epoch": 0.12088820036147689,
      "grad_norm": 6.970714569091797,
      "learning_rate": 9.69371220961511e-06,
      "loss": 0.0538,
      "step": 2341
    },
    {
      "epoch": 0.12093983991737671,
      "grad_norm": 7.994577884674072,
      "learning_rate": 9.693429778898534e-06,
      "loss": 0.1152,
      "step": 2342
    },
    {
      "epoch": 0.12099147947327653,
      "grad_norm": 9.848847389221191,
      "learning_rate": 9.693147222144239e-06,
      "loss": 0.0742,
      "step": 2343
    },
    {
      "epoch": 0.12104311902917635,
      "grad_norm": 10.724406242370605,
      "learning_rate": 9.692864539359813e-06,
      "loss": 0.0585,
      "step": 2344
    },
    {
      "epoch": 0.12109475858507616,
      "grad_norm": 10.343335151672363,
      "learning_rate": 9.692581730552846e-06,
      "loss": 0.1788,
      "step": 2345
    },
    {
      "epoch": 0.12114639814097598,
      "grad_norm": 3.5361433029174805,
      "learning_rate": 9.692298795730935e-06,
      "loss": 0.0437,
      "step": 2346
    },
    {
      "epoch": 0.1211980376968758,
      "grad_norm": 13.338478088378906,
      "learning_rate": 9.692015734901676e-06,
      "loss": 0.0911,
      "step": 2347
    },
    {
      "epoch": 0.12124967725277562,
      "grad_norm": 8.836647987365723,
      "learning_rate": 9.69173254807267e-06,
      "loss": 0.1028,
      "step": 2348
    },
    {
      "epoch": 0.12130131680867545,
      "grad_norm": 13.712032318115234,
      "learning_rate": 9.691449235251525e-06,
      "loss": 0.1488,
      "step": 2349
    },
    {
      "epoch": 0.12135295636457527,
      "grad_norm": 14.106703758239746,
      "learning_rate": 9.691165796445845e-06,
      "loss": 0.1377,
      "step": 2350
    },
    {
      "epoch": 0.12140459592047509,
      "grad_norm": 12.10207462310791,
      "learning_rate": 9.690882231663243e-06,
      "loss": 0.1171,
      "step": 2351
    },
    {
      "epoch": 0.12145623547637491,
      "grad_norm": 4.476017475128174,
      "learning_rate": 9.690598540911335e-06,
      "loss": 0.0353,
      "step": 2352
    },
    {
      "epoch": 0.12150787503227473,
      "grad_norm": 14.29375171661377,
      "learning_rate": 9.690314724197738e-06,
      "loss": 0.1908,
      "step": 2353
    },
    {
      "epoch": 0.12155951458817454,
      "grad_norm": 7.190357685089111,
      "learning_rate": 9.690030781530073e-06,
      "loss": 0.1063,
      "step": 2354
    },
    {
      "epoch": 0.12161115414407436,
      "grad_norm": 3.8772048950195312,
      "learning_rate": 9.689746712915967e-06,
      "loss": 0.0409,
      "step": 2355
    },
    {
      "epoch": 0.12166279369997418,
      "grad_norm": 10.194367408752441,
      "learning_rate": 9.689462518363048e-06,
      "loss": 0.1087,
      "step": 2356
    },
    {
      "epoch": 0.121714433255874,
      "grad_norm": 3.267228126525879,
      "learning_rate": 9.689178197878945e-06,
      "loss": 0.092,
      "step": 2357
    },
    {
      "epoch": 0.12176607281177382,
      "grad_norm": 3.355398416519165,
      "learning_rate": 9.688893751471297e-06,
      "loss": 0.0698,
      "step": 2358
    },
    {
      "epoch": 0.12181771236767364,
      "grad_norm": 10.819433212280273,
      "learning_rate": 9.688609179147741e-06,
      "loss": 0.0811,
      "step": 2359
    },
    {
      "epoch": 0.12186935192357345,
      "grad_norm": 9.845712661743164,
      "learning_rate": 9.688324480915921e-06,
      "loss": 0.1722,
      "step": 2360
    },
    {
      "epoch": 0.12192099147947327,
      "grad_norm": 7.918739318847656,
      "learning_rate": 9.688039656783477e-06,
      "loss": 0.1491,
      "step": 2361
    },
    {
      "epoch": 0.12197263103537309,
      "grad_norm": 8.249861717224121,
      "learning_rate": 9.687754706758062e-06,
      "loss": 0.121,
      "step": 2362
    },
    {
      "epoch": 0.12202427059127291,
      "grad_norm": 8.28043270111084,
      "learning_rate": 9.687469630847325e-06,
      "loss": 0.1078,
      "step": 2363
    },
    {
      "epoch": 0.12207591014717273,
      "grad_norm": 3.258941888809204,
      "learning_rate": 9.687184429058924e-06,
      "loss": 0.0655,
      "step": 2364
    },
    {
      "epoch": 0.12212754970307256,
      "grad_norm": 9.577159881591797,
      "learning_rate": 9.686899101400519e-06,
      "loss": 0.1415,
      "step": 2365
    },
    {
      "epoch": 0.12217918925897238,
      "grad_norm": 5.609995365142822,
      "learning_rate": 9.68661364787977e-06,
      "loss": 0.106,
      "step": 2366
    },
    {
      "epoch": 0.1222308288148722,
      "grad_norm": 7.711476802825928,
      "learning_rate": 9.686328068504343e-06,
      "loss": 0.0599,
      "step": 2367
    },
    {
      "epoch": 0.12228246837077202,
      "grad_norm": 3.915482997894287,
      "learning_rate": 9.686042363281904e-06,
      "loss": 0.0639,
      "step": 2368
    },
    {
      "epoch": 0.12233410792667183,
      "grad_norm": 8.660676956176758,
      "learning_rate": 9.685756532220131e-06,
      "loss": 0.1089,
      "step": 2369
    },
    {
      "epoch": 0.12238574748257165,
      "grad_norm": 7.718691825866699,
      "learning_rate": 9.685470575326698e-06,
      "loss": 0.1346,
      "step": 2370
    },
    {
      "epoch": 0.12243738703847147,
      "grad_norm": 3.1297900676727295,
      "learning_rate": 9.685184492609281e-06,
      "loss": 0.0459,
      "step": 2371
    },
    {
      "epoch": 0.12248902659437129,
      "grad_norm": 8.130016326904297,
      "learning_rate": 9.684898284075566e-06,
      "loss": 0.1086,
      "step": 2372
    },
    {
      "epoch": 0.12254066615027111,
      "grad_norm": 3.5146098136901855,
      "learning_rate": 9.684611949733236e-06,
      "loss": 0.0664,
      "step": 2373
    },
    {
      "epoch": 0.12259230570617093,
      "grad_norm": 7.187399387359619,
      "learning_rate": 9.684325489589983e-06,
      "loss": 0.0708,
      "step": 2374
    },
    {
      "epoch": 0.12264394526207074,
      "grad_norm": 4.641982078552246,
      "learning_rate": 9.684038903653498e-06,
      "loss": 0.0816,
      "step": 2375
    },
    {
      "epoch": 0.12269558481797056,
      "grad_norm": 4.024410724639893,
      "learning_rate": 9.683752191931476e-06,
      "loss": 0.0728,
      "step": 2376
    },
    {
      "epoch": 0.12274722437387038,
      "grad_norm": 2.9543073177337646,
      "learning_rate": 9.68346535443162e-06,
      "loss": 0.0247,
      "step": 2377
    },
    {
      "epoch": 0.1227988639297702,
      "grad_norm": 10.729937553405762,
      "learning_rate": 9.68317839116163e-06,
      "loss": 0.0829,
      "step": 2378
    },
    {
      "epoch": 0.12285050348567002,
      "grad_norm": 6.931087017059326,
      "learning_rate": 9.68289130212921e-06,
      "loss": 0.104,
      "step": 2379
    },
    {
      "epoch": 0.12290214304156984,
      "grad_norm": 19.268808364868164,
      "learning_rate": 9.682604087342075e-06,
      "loss": 0.1403,
      "step": 2380
    },
    {
      "epoch": 0.12295378259746967,
      "grad_norm": 4.970942497253418,
      "learning_rate": 9.682316746807936e-06,
      "loss": 0.0512,
      "step": 2381
    },
    {
      "epoch": 0.12300542215336949,
      "grad_norm": 4.349081516265869,
      "learning_rate": 9.682029280534506e-06,
      "loss": 0.0317,
      "step": 2382
    },
    {
      "epoch": 0.1230570617092693,
      "grad_norm": 4.43085241317749,
      "learning_rate": 9.681741688529506e-06,
      "loss": 0.041,
      "step": 2383
    },
    {
      "epoch": 0.12310870126516912,
      "grad_norm": 5.557555675506592,
      "learning_rate": 9.681453970800663e-06,
      "loss": 0.0815,
      "step": 2384
    },
    {
      "epoch": 0.12316034082106894,
      "grad_norm": 11.016518592834473,
      "learning_rate": 9.681166127355698e-06,
      "loss": 0.064,
      "step": 2385
    },
    {
      "epoch": 0.12321198037696876,
      "grad_norm": 4.204098701477051,
      "learning_rate": 9.680878158202343e-06,
      "loss": 0.0434,
      "step": 2386
    },
    {
      "epoch": 0.12326361993286858,
      "grad_norm": 6.29036808013916,
      "learning_rate": 9.680590063348333e-06,
      "loss": 0.093,
      "step": 2387
    },
    {
      "epoch": 0.1233152594887684,
      "grad_norm": 8.14383602142334,
      "learning_rate": 9.680301842801402e-06,
      "loss": 0.061,
      "step": 2388
    },
    {
      "epoch": 0.12336689904466822,
      "grad_norm": 4.409150123596191,
      "learning_rate": 9.680013496569289e-06,
      "loss": 0.0683,
      "step": 2389
    },
    {
      "epoch": 0.12341853860056803,
      "grad_norm": 7.719663143157959,
      "learning_rate": 9.679725024659741e-06,
      "loss": 0.0634,
      "step": 2390
    },
    {
      "epoch": 0.12347017815646785,
      "grad_norm": 10.163674354553223,
      "learning_rate": 9.679436427080501e-06,
      "loss": 0.0892,
      "step": 2391
    },
    {
      "epoch": 0.12352181771236767,
      "grad_norm": 6.936243057250977,
      "learning_rate": 9.679147703839323e-06,
      "loss": 0.0436,
      "step": 2392
    },
    {
      "epoch": 0.12357345726826749,
      "grad_norm": 6.114467144012451,
      "learning_rate": 9.678858854943957e-06,
      "loss": 0.0903,
      "step": 2393
    },
    {
      "epoch": 0.12362509682416731,
      "grad_norm": 12.556259155273438,
      "learning_rate": 9.67856988040216e-06,
      "loss": 0.0469,
      "step": 2394
    },
    {
      "epoch": 0.12367673638006713,
      "grad_norm": 7.482909202575684,
      "learning_rate": 9.678280780221694e-06,
      "loss": 0.0646,
      "step": 2395
    },
    {
      "epoch": 0.12372837593596694,
      "grad_norm": 5.624312877655029,
      "learning_rate": 9.67799155441032e-06,
      "loss": 0.0839,
      "step": 2396
    },
    {
      "epoch": 0.12378001549186678,
      "grad_norm": 18.990741729736328,
      "learning_rate": 9.677702202975807e-06,
      "loss": 0.0996,
      "step": 2397
    },
    {
      "epoch": 0.1238316550477666,
      "grad_norm": 7.289380073547363,
      "learning_rate": 9.677412725925926e-06,
      "loss": 0.1259,
      "step": 2398
    },
    {
      "epoch": 0.12388329460366641,
      "grad_norm": 7.137175559997559,
      "learning_rate": 9.677123123268448e-06,
      "loss": 0.0692,
      "step": 2399
    },
    {
      "epoch": 0.12393493415956623,
      "grad_norm": 8.650971412658691,
      "learning_rate": 9.676833395011149e-06,
      "loss": 0.1353,
      "step": 2400
    },
    {
      "epoch": 0.12398657371546605,
      "grad_norm": 22.188215255737305,
      "learning_rate": 9.676543541161815e-06,
      "loss": 0.1721,
      "step": 2401
    },
    {
      "epoch": 0.12403821327136587,
      "grad_norm": 15.917112350463867,
      "learning_rate": 9.676253561728224e-06,
      "loss": 0.2553,
      "step": 2402
    },
    {
      "epoch": 0.12408985282726569,
      "grad_norm": 16.176633834838867,
      "learning_rate": 9.675963456718167e-06,
      "loss": 0.1473,
      "step": 2403
    },
    {
      "epoch": 0.1241414923831655,
      "grad_norm": 5.359736919403076,
      "learning_rate": 9.675673226139432e-06,
      "loss": 0.0657,
      "step": 2404
    },
    {
      "epoch": 0.12419313193906532,
      "grad_norm": 7.106653690338135,
      "learning_rate": 9.675382869999817e-06,
      "loss": 0.0801,
      "step": 2405
    },
    {
      "epoch": 0.12424477149496514,
      "grad_norm": 5.365082263946533,
      "learning_rate": 9.675092388307113e-06,
      "loss": 0.082,
      "step": 2406
    },
    {
      "epoch": 0.12429641105086496,
      "grad_norm": 3.182598829269409,
      "learning_rate": 9.674801781069123e-06,
      "loss": 0.0728,
      "step": 2407
    },
    {
      "epoch": 0.12434805060676478,
      "grad_norm": 5.277021408081055,
      "learning_rate": 9.674511048293655e-06,
      "loss": 0.0423,
      "step": 2408
    },
    {
      "epoch": 0.1243996901626646,
      "grad_norm": 3.2600064277648926,
      "learning_rate": 9.67422018998851e-06,
      "loss": 0.0671,
      "step": 2409
    },
    {
      "epoch": 0.12445132971856442,
      "grad_norm": 4.2733612060546875,
      "learning_rate": 9.673929206161504e-06,
      "loss": 0.0587,
      "step": 2410
    },
    {
      "epoch": 0.12450296927446423,
      "grad_norm": 2.6689131259918213,
      "learning_rate": 9.673638096820447e-06,
      "loss": 0.0697,
      "step": 2411
    },
    {
      "epoch": 0.12455460883036405,
      "grad_norm": 4.085305690765381,
      "learning_rate": 9.67334686197316e-06,
      "loss": 0.0776,
      "step": 2412
    },
    {
      "epoch": 0.12460624838626388,
      "grad_norm": 11.887299537658691,
      "learning_rate": 9.673055501627463e-06,
      "loss": 0.1421,
      "step": 2413
    },
    {
      "epoch": 0.1246578879421637,
      "grad_norm": 12.639544486999512,
      "learning_rate": 9.672764015791176e-06,
      "loss": 0.113,
      "step": 2414
    },
    {
      "epoch": 0.12470952749806352,
      "grad_norm": 12.96092414855957,
      "learning_rate": 9.672472404472133e-06,
      "loss": 0.1235,
      "step": 2415
    },
    {
      "epoch": 0.12476116705396334,
      "grad_norm": 7.828914165496826,
      "learning_rate": 9.672180667678161e-06,
      "loss": 0.0684,
      "step": 2416
    },
    {
      "epoch": 0.12481280660986316,
      "grad_norm": 11.91541862487793,
      "learning_rate": 9.671888805417096e-06,
      "loss": 0.1858,
      "step": 2417
    },
    {
      "epoch": 0.12486444616576298,
      "grad_norm": 4.193929195404053,
      "learning_rate": 9.671596817696774e-06,
      "loss": 0.0507,
      "step": 2418
    },
    {
      "epoch": 0.1249160857216628,
      "grad_norm": 11.855626106262207,
      "learning_rate": 9.671304704525037e-06,
      "loss": 0.1443,
      "step": 2419
    },
    {
      "epoch": 0.12496772527756261,
      "grad_norm": 23.452770233154297,
      "learning_rate": 9.671012465909731e-06,
      "loss": 0.0388,
      "step": 2420
    },
    {
      "epoch": 0.12501936483346243,
      "grad_norm": 6.89123010635376,
      "learning_rate": 9.670720101858702e-06,
      "loss": 0.0936,
      "step": 2421
    },
    {
      "epoch": 0.12507100438936225,
      "grad_norm": 3.985356330871582,
      "learning_rate": 9.6704276123798e-06,
      "loss": 0.0901,
      "step": 2422
    },
    {
      "epoch": 0.12512264394526207,
      "grad_norm": 23.873390197753906,
      "learning_rate": 9.670134997480882e-06,
      "loss": 0.1965,
      "step": 2423
    },
    {
      "epoch": 0.1251742835011619,
      "grad_norm": 5.943964004516602,
      "learning_rate": 9.669842257169804e-06,
      "loss": 0.0593,
      "step": 2424
    },
    {
      "epoch": 0.1252259230570617,
      "grad_norm": 7.296074867248535,
      "learning_rate": 9.669549391454429e-06,
      "loss": 0.0852,
      "step": 2425
    },
    {
      "epoch": 0.12527756261296152,
      "grad_norm": 5.300397872924805,
      "learning_rate": 9.669256400342622e-06,
      "loss": 0.0876,
      "step": 2426
    },
    {
      "epoch": 0.12532920216886134,
      "grad_norm": 9.2325439453125,
      "learning_rate": 9.668963283842247e-06,
      "loss": 0.091,
      "step": 2427
    },
    {
      "epoch": 0.12538084172476116,
      "grad_norm": 10.713523864746094,
      "learning_rate": 9.668670041961181e-06,
      "loss": 0.1173,
      "step": 2428
    },
    {
      "epoch": 0.12543248128066098,
      "grad_norm": 3.2672102451324463,
      "learning_rate": 9.668376674707294e-06,
      "loss": 0.0577,
      "step": 2429
    },
    {
      "epoch": 0.1254841208365608,
      "grad_norm": 3.777674436569214,
      "learning_rate": 9.668083182088468e-06,
      "loss": 0.0748,
      "step": 2430
    },
    {
      "epoch": 0.12553576039246062,
      "grad_norm": 3.37672758102417,
      "learning_rate": 9.66778956411258e-06,
      "loss": 0.0448,
      "step": 2431
    },
    {
      "epoch": 0.12558739994836043,
      "grad_norm": 18.898996353149414,
      "learning_rate": 9.66749582078752e-06,
      "loss": 0.1547,
      "step": 2432
    },
    {
      "epoch": 0.12563903950426025,
      "grad_norm": 7.287028789520264,
      "learning_rate": 9.667201952121171e-06,
      "loss": 0.0756,
      "step": 2433
    },
    {
      "epoch": 0.12569067906016007,
      "grad_norm": 5.676045894622803,
      "learning_rate": 9.666907958121428e-06,
      "loss": 0.1125,
      "step": 2434
    },
    {
      "epoch": 0.1257423186160599,
      "grad_norm": 4.809650421142578,
      "learning_rate": 9.666613838796186e-06,
      "loss": 0.0711,
      "step": 2435
    },
    {
      "epoch": 0.1257939581719597,
      "grad_norm": 5.4055328369140625,
      "learning_rate": 9.666319594153342e-06,
      "loss": 0.0785,
      "step": 2436
    },
    {
      "epoch": 0.12584559772785955,
      "grad_norm": 10.142650604248047,
      "learning_rate": 9.666025224200797e-06,
      "loss": 0.0671,
      "step": 2437
    },
    {
      "epoch": 0.12589723728375937,
      "grad_norm": 2.5391902923583984,
      "learning_rate": 9.665730728946458e-06,
      "loss": 0.0573,
      "step": 2438
    },
    {
      "epoch": 0.1259488768396592,
      "grad_norm": 11.59704875946045,
      "learning_rate": 9.665436108398232e-06,
      "loss": 0.0987,
      "step": 2439
    },
    {
      "epoch": 0.126000516395559,
      "grad_norm": 2.534247875213623,
      "learning_rate": 9.66514136256403e-06,
      "loss": 0.0519,
      "step": 2440
    },
    {
      "epoch": 0.12605215595145883,
      "grad_norm": 5.638066291809082,
      "learning_rate": 9.664846491451772e-06,
      "loss": 0.0597,
      "step": 2441
    },
    {
      "epoch": 0.12610379550735865,
      "grad_norm": 4.069097995758057,
      "learning_rate": 9.66455149506937e-06,
      "loss": 0.0468,
      "step": 2442
    },
    {
      "epoch": 0.12615543506325846,
      "grad_norm": 2.946044921875,
      "learning_rate": 9.664256373424749e-06,
      "loss": 0.0613,
      "step": 2443
    },
    {
      "epoch": 0.12620707461915828,
      "grad_norm": 2.8167684078216553,
      "learning_rate": 9.663961126525834e-06,
      "loss": 0.0628,
      "step": 2444
    },
    {
      "epoch": 0.1262587141750581,
      "grad_norm": 13.30502700805664,
      "learning_rate": 9.663665754380553e-06,
      "loss": 0.1044,
      "step": 2445
    },
    {
      "epoch": 0.12631035373095792,
      "grad_norm": 6.020079135894775,
      "learning_rate": 9.66337025699684e-06,
      "loss": 0.1072,
      "step": 2446
    },
    {
      "epoch": 0.12636199328685774,
      "grad_norm": 45.45231628417969,
      "learning_rate": 9.663074634382627e-06,
      "loss": 0.1474,
      "step": 2447
    },
    {
      "epoch": 0.12641363284275756,
      "grad_norm": 8.204263687133789,
      "learning_rate": 9.662778886545857e-06,
      "loss": 0.1049,
      "step": 2448
    },
    {
      "epoch": 0.12646527239865737,
      "grad_norm": 5.073081970214844,
      "learning_rate": 9.662483013494468e-06,
      "loss": 0.0541,
      "step": 2449
    },
    {
      "epoch": 0.1265169119545572,
      "grad_norm": 5.023939609527588,
      "learning_rate": 9.662187015236406e-06,
      "loss": 0.0708,
      "step": 2450
    },
    {
      "epoch": 0.126568551510457,
      "grad_norm": 10.348799705505371,
      "learning_rate": 9.661890891779622e-06,
      "loss": 0.0577,
      "step": 2451
    },
    {
      "epoch": 0.12662019106635683,
      "grad_norm": 4.083311557769775,
      "learning_rate": 9.661594643132065e-06,
      "loss": 0.0603,
      "step": 2452
    },
    {
      "epoch": 0.12667183062225665,
      "grad_norm": 7.408773899078369,
      "learning_rate": 9.661298269301692e-06,
      "loss": 0.0589,
      "step": 2453
    },
    {
      "epoch": 0.12672347017815647,
      "grad_norm": 7.795835494995117,
      "learning_rate": 9.661001770296466e-06,
      "loss": 0.0628,
      "step": 2454
    },
    {
      "epoch": 0.12677510973405628,
      "grad_norm": 6.6792778968811035,
      "learning_rate": 9.660705146124341e-06,
      "loss": 0.1137,
      "step": 2455
    },
    {
      "epoch": 0.1268267492899561,
      "grad_norm": 5.662730693817139,
      "learning_rate": 9.660408396793289e-06,
      "loss": 0.0698,
      "step": 2456
    },
    {
      "epoch": 0.12687838884585592,
      "grad_norm": 3.8438565731048584,
      "learning_rate": 9.660111522311275e-06,
      "loss": 0.0378,
      "step": 2457
    },
    {
      "epoch": 0.12693002840175574,
      "grad_norm": 4.256504058837891,
      "learning_rate": 9.659814522686275e-06,
      "loss": 0.0476,
      "step": 2458
    },
    {
      "epoch": 0.12698166795765556,
      "grad_norm": 8.03322696685791,
      "learning_rate": 9.659517397926263e-06,
      "loss": 0.0442,
      "step": 2459
    },
    {
      "epoch": 0.12703330751355538,
      "grad_norm": 6.160405158996582,
      "learning_rate": 9.659220148039216e-06,
      "loss": 0.0916,
      "step": 2460
    },
    {
      "epoch": 0.1270849470694552,
      "grad_norm": 5.549685478210449,
      "learning_rate": 9.658922773033118e-06,
      "loss": 0.0586,
      "step": 2461
    },
    {
      "epoch": 0.127136586625355,
      "grad_norm": 9.448091506958008,
      "learning_rate": 9.658625272915956e-06,
      "loss": 0.1413,
      "step": 2462
    },
    {
      "epoch": 0.12718822618125483,
      "grad_norm": 15.244924545288086,
      "learning_rate": 9.658327647695718e-06,
      "loss": 0.0594,
      "step": 2463
    },
    {
      "epoch": 0.12723986573715465,
      "grad_norm": 10.197968482971191,
      "learning_rate": 9.658029897380397e-06,
      "loss": 0.1137,
      "step": 2464
    },
    {
      "epoch": 0.12729150529305447,
      "grad_norm": 5.1688385009765625,
      "learning_rate": 9.657732021977986e-06,
      "loss": 0.083,
      "step": 2465
    },
    {
      "epoch": 0.1273431448489543,
      "grad_norm": 5.379540920257568,
      "learning_rate": 9.657434021496488e-06,
      "loss": 0.0684,
      "step": 2466
    },
    {
      "epoch": 0.1273947844048541,
      "grad_norm": 18.724605560302734,
      "learning_rate": 9.657135895943903e-06,
      "loss": 0.2105,
      "step": 2467
    },
    {
      "epoch": 0.12744642396075392,
      "grad_norm": 9.7989501953125,
      "learning_rate": 9.65683764532824e-06,
      "loss": 0.0986,
      "step": 2468
    },
    {
      "epoch": 0.12749806351665377,
      "grad_norm": 13.488003730773926,
      "learning_rate": 9.656539269657502e-06,
      "loss": 0.1174,
      "step": 2469
    },
    {
      "epoch": 0.1275497030725536,
      "grad_norm": 13.75413703918457,
      "learning_rate": 9.656240768939708e-06,
      "loss": 0.1003,
      "step": 2470
    },
    {
      "epoch": 0.1276013426284534,
      "grad_norm": 6.402712345123291,
      "learning_rate": 9.65594214318287e-06,
      "loss": 0.1004,
      "step": 2471
    },
    {
      "epoch": 0.12765298218435323,
      "grad_norm": 8.22070598602295,
      "learning_rate": 9.655643392395011e-06,
      "loss": 0.0645,
      "step": 2472
    },
    {
      "epoch": 0.12770462174025304,
      "grad_norm": 10.006181716918945,
      "learning_rate": 9.65534451658415e-06,
      "loss": 0.1837,
      "step": 2473
    },
    {
      "epoch": 0.12775626129615286,
      "grad_norm": 7.070072174072266,
      "learning_rate": 9.655045515758317e-06,
      "loss": 0.101,
      "step": 2474
    },
    {
      "epoch": 0.12780790085205268,
      "grad_norm": 15.63623046875,
      "learning_rate": 9.654746389925536e-06,
      "loss": 0.1335,
      "step": 2475
    },
    {
      "epoch": 0.1278595404079525,
      "grad_norm": 6.544208526611328,
      "learning_rate": 9.654447139093844e-06,
      "loss": 0.0693,
      "step": 2476
    },
    {
      "epoch": 0.12791117996385232,
      "grad_norm": 7.462660789489746,
      "learning_rate": 9.654147763271275e-06,
      "loss": 0.0638,
      "step": 2477
    },
    {
      "epoch": 0.12796281951975214,
      "grad_norm": 9.58002758026123,
      "learning_rate": 9.653848262465871e-06,
      "loss": 0.1018,
      "step": 2478
    },
    {
      "epoch": 0.12801445907565195,
      "grad_norm": 11.597204208374023,
      "learning_rate": 9.653548636685672e-06,
      "loss": 0.1225,
      "step": 2479
    },
    {
      "epoch": 0.12806609863155177,
      "grad_norm": 5.370260715484619,
      "learning_rate": 9.653248885938726e-06,
      "loss": 0.0636,
      "step": 2480
    },
    {
      "epoch": 0.1281177381874516,
      "grad_norm": 9.490228652954102,
      "learning_rate": 9.65294901023308e-06,
      "loss": 0.1082,
      "step": 2481
    },
    {
      "epoch": 0.1281693777433514,
      "grad_norm": 3.6843791007995605,
      "learning_rate": 9.65264900957679e-06,
      "loss": 0.0702,
      "step": 2482
    },
    {
      "epoch": 0.12822101729925123,
      "grad_norm": 8.762184143066406,
      "learning_rate": 9.652348883977913e-06,
      "loss": 0.0793,
      "step": 2483
    },
    {
      "epoch": 0.12827265685515105,
      "grad_norm": 5.159938812255859,
      "learning_rate": 9.652048633444504e-06,
      "loss": 0.0657,
      "step": 2484
    },
    {
      "epoch": 0.12832429641105086,
      "grad_norm": 14.643305778503418,
      "learning_rate": 9.65174825798463e-06,
      "loss": 0.1568,
      "step": 2485
    },
    {
      "epoch": 0.12837593596695068,
      "grad_norm": 9.318314552307129,
      "learning_rate": 9.651447757606354e-06,
      "loss": 0.0903,
      "step": 2486
    },
    {
      "epoch": 0.1284275755228505,
      "grad_norm": 4.450388431549072,
      "learning_rate": 9.65114713231775e-06,
      "loss": 0.0416,
      "step": 2487
    },
    {
      "epoch": 0.12847921507875032,
      "grad_norm": 2.7697179317474365,
      "learning_rate": 9.650846382126885e-06,
      "loss": 0.0549,
      "step": 2488
    },
    {
      "epoch": 0.12853085463465014,
      "grad_norm": 13.44144344329834,
      "learning_rate": 9.65054550704184e-06,
      "loss": 0.211,
      "step": 2489
    },
    {
      "epoch": 0.12858249419054996,
      "grad_norm": 7.607124328613281,
      "learning_rate": 9.650244507070695e-06,
      "loss": 0.1241,
      "step": 2490
    },
    {
      "epoch": 0.12863413374644977,
      "grad_norm": 3.645751476287842,
      "learning_rate": 9.649943382221532e-06,
      "loss": 0.0601,
      "step": 2491
    },
    {
      "epoch": 0.1286857733023496,
      "grad_norm": 6.309833526611328,
      "learning_rate": 9.649642132502434e-06,
      "loss": 0.1478,
      "step": 2492
    },
    {
      "epoch": 0.1287374128582494,
      "grad_norm": 5.6513285636901855,
      "learning_rate": 9.649340757921497e-06,
      "loss": 0.154,
      "step": 2493
    },
    {
      "epoch": 0.12878905241414923,
      "grad_norm": 4.783064842224121,
      "learning_rate": 9.64903925848681e-06,
      "loss": 0.0873,
      "step": 2494
    },
    {
      "epoch": 0.12884069197004905,
      "grad_norm": 7.455208778381348,
      "learning_rate": 9.648737634206469e-06,
      "loss": 0.0745,
      "step": 2495
    },
    {
      "epoch": 0.12889233152594887,
      "grad_norm": 3.631751298904419,
      "learning_rate": 9.648435885088575e-06,
      "loss": 0.0672,
      "step": 2496
    },
    {
      "epoch": 0.12894397108184869,
      "grad_norm": 8.357789993286133,
      "learning_rate": 9.648134011141234e-06,
      "loss": 0.1118,
      "step": 2497
    },
    {
      "epoch": 0.1289956106377485,
      "grad_norm": 7.022691249847412,
      "learning_rate": 9.64783201237255e-06,
      "loss": 0.0965,
      "step": 2498
    },
    {
      "epoch": 0.12904725019364832,
      "grad_norm": 11.076321601867676,
      "learning_rate": 9.647529888790633e-06,
      "loss": 0.1167,
      "step": 2499
    },
    {
      "epoch": 0.12909888974954814,
      "grad_norm": 20.170663833618164,
      "learning_rate": 9.647227640403593e-06,
      "loss": 0.2072,
      "step": 2500
    },
    {
      "epoch": 0.129150529305448,
      "grad_norm": 4.17242431640625,
      "learning_rate": 9.646925267219553e-06,
      "loss": 0.0678,
      "step": 2501
    },
    {
      "epoch": 0.1292021688613478,
      "grad_norm": 9.938359260559082,
      "learning_rate": 9.646622769246629e-06,
      "loss": 0.14,
      "step": 2502
    },
    {
      "epoch": 0.12925380841724762,
      "grad_norm": 4.993642807006836,
      "learning_rate": 9.646320146492945e-06,
      "loss": 0.0475,
      "step": 2503
    },
    {
      "epoch": 0.12930544797314744,
      "grad_norm": 2.1381590366363525,
      "learning_rate": 9.646017398966627e-06,
      "loss": 0.0306,
      "step": 2504
    },
    {
      "epoch": 0.12935708752904726,
      "grad_norm": 2.6727869510650635,
      "learning_rate": 9.645714526675806e-06,
      "loss": 0.0561,
      "step": 2505
    },
    {
      "epoch": 0.12940872708494708,
      "grad_norm": 5.463906764984131,
      "learning_rate": 9.645411529628617e-06,
      "loss": 0.0639,
      "step": 2506
    },
    {
      "epoch": 0.1294603666408469,
      "grad_norm": 5.484694957733154,
      "learning_rate": 9.645108407833193e-06,
      "loss": 0.0935,
      "step": 2507
    },
    {
      "epoch": 0.12951200619674672,
      "grad_norm": 13.493364334106445,
      "learning_rate": 9.644805161297675e-06,
      "loss": 0.0728,
      "step": 2508
    },
    {
      "epoch": 0.12956364575264653,
      "grad_norm": 8.581274032592773,
      "learning_rate": 9.644501790030208e-06,
      "loss": 0.0565,
      "step": 2509
    },
    {
      "epoch": 0.12961528530854635,
      "grad_norm": 3.5533559322357178,
      "learning_rate": 9.644198294038937e-06,
      "loss": 0.0802,
      "step": 2510
    },
    {
      "epoch": 0.12966692486444617,
      "grad_norm": 9.363054275512695,
      "learning_rate": 9.643894673332015e-06,
      "loss": 0.0857,
      "step": 2511
    },
    {
      "epoch": 0.129718564420346,
      "grad_norm": 8.05274772644043,
      "learning_rate": 9.643590927917595e-06,
      "loss": 0.0802,
      "step": 2512
    },
    {
      "epoch": 0.1297702039762458,
      "grad_norm": 18.85692024230957,
      "learning_rate": 9.64328705780383e-06,
      "loss": 0.0979,
      "step": 2513
    },
    {
      "epoch": 0.12982184353214563,
      "grad_norm": 10.833379745483398,
      "learning_rate": 9.642983062998883e-06,
      "loss": 0.1428,
      "step": 2514
    },
    {
      "epoch": 0.12987348308804544,
      "grad_norm": 8.152714729309082,
      "learning_rate": 9.642678943510915e-06,
      "loss": 0.1122,
      "step": 2515
    },
    {
      "epoch": 0.12992512264394526,
      "grad_norm": 6.601385593414307,
      "learning_rate": 9.642374699348098e-06,
      "loss": 0.1695,
      "step": 2516
    },
    {
      "epoch": 0.12997676219984508,
      "grad_norm": 13.78110122680664,
      "learning_rate": 9.642070330518595e-06,
      "loss": 0.1053,
      "step": 2517
    },
    {
      "epoch": 0.1300284017557449,
      "grad_norm": 3.6589691638946533,
      "learning_rate": 9.641765837030588e-06,
      "loss": 0.0316,
      "step": 2518
    },
    {
      "epoch": 0.13008004131164472,
      "grad_norm": 5.829361438751221,
      "learning_rate": 9.641461218892247e-06,
      "loss": 0.0626,
      "step": 2519
    },
    {
      "epoch": 0.13013168086754454,
      "grad_norm": 6.078234672546387,
      "learning_rate": 9.641156476111756e-06,
      "loss": 0.096,
      "step": 2520
    },
    {
      "epoch": 0.13018332042344435,
      "grad_norm": 4.63171911239624,
      "learning_rate": 9.640851608697297e-06,
      "loss": 0.0731,
      "step": 2521
    },
    {
      "epoch": 0.13023495997934417,
      "grad_norm": 4.2525787353515625,
      "learning_rate": 9.640546616657056e-06,
      "loss": 0.0487,
      "step": 2522
    },
    {
      "epoch": 0.130286599535244,
      "grad_norm": 3.850069999694824,
      "learning_rate": 9.640241499999226e-06,
      "loss": 0.0653,
      "step": 2523
    },
    {
      "epoch": 0.1303382390911438,
      "grad_norm": 3.7463173866271973,
      "learning_rate": 9.639936258731998e-06,
      "loss": 0.0476,
      "step": 2524
    },
    {
      "epoch": 0.13038987864704363,
      "grad_norm": 5.037931442260742,
      "learning_rate": 9.639630892863571e-06,
      "loss": 0.0869,
      "step": 2525
    },
    {
      "epoch": 0.13044151820294345,
      "grad_norm": 3.5341169834136963,
      "learning_rate": 9.639325402402144e-06,
      "loss": 0.0605,
      "step": 2526
    },
    {
      "epoch": 0.13049315775884326,
      "grad_norm": 6.913913726806641,
      "learning_rate": 9.639019787355923e-06,
      "loss": 0.1261,
      "step": 2527
    },
    {
      "epoch": 0.13054479731474308,
      "grad_norm": 11.060751914978027,
      "learning_rate": 9.638714047733112e-06,
      "loss": 0.0422,
      "step": 2528
    },
    {
      "epoch": 0.1305964368706429,
      "grad_norm": 2.4349262714385986,
      "learning_rate": 9.63840818354192e-06,
      "loss": 0.0572,
      "step": 2529
    },
    {
      "epoch": 0.13064807642654272,
      "grad_norm": 8.010625839233398,
      "learning_rate": 9.638102194790567e-06,
      "loss": 0.0731,
      "step": 2530
    },
    {
      "epoch": 0.13069971598244254,
      "grad_norm": 3.0417640209198,
      "learning_rate": 9.637796081487263e-06,
      "loss": 0.0465,
      "step": 2531
    },
    {
      "epoch": 0.13075135553834236,
      "grad_norm": 11.397322654724121,
      "learning_rate": 9.637489843640233e-06,
      "loss": 0.0838,
      "step": 2532
    },
    {
      "epoch": 0.1308029950942422,
      "grad_norm": 1.3889784812927246,
      "learning_rate": 9.6371834812577e-06,
      "loss": 0.0212,
      "step": 2533
    },
    {
      "epoch": 0.13085463465014202,
      "grad_norm": 5.168978214263916,
      "learning_rate": 9.636876994347888e-06,
      "loss": 0.0711,
      "step": 2534
    },
    {
      "epoch": 0.13090627420604184,
      "grad_norm": 5.230069160461426,
      "learning_rate": 9.636570382919031e-06,
      "loss": 0.0716,
      "step": 2535
    },
    {
      "epoch": 0.13095791376194166,
      "grad_norm": 5.173567771911621,
      "learning_rate": 9.636263646979362e-06,
      "loss": 0.098,
      "step": 2536
    },
    {
      "epoch": 0.13100955331784148,
      "grad_norm": 15.038141250610352,
      "learning_rate": 9.635956786537119e-06,
      "loss": 0.125,
      "step": 2537
    },
    {
      "epoch": 0.1310611928737413,
      "grad_norm": 7.575570583343506,
      "learning_rate": 9.63564980160054e-06,
      "loss": 0.1692,
      "step": 2538
    },
    {
      "epoch": 0.1311128324296411,
      "grad_norm": 6.646186828613281,
      "learning_rate": 9.63534269217787e-06,
      "loss": 0.0757,
      "step": 2539
    },
    {
      "epoch": 0.13116447198554093,
      "grad_norm": 10.23719310760498,
      "learning_rate": 9.635035458277355e-06,
      "loss": 0.1255,
      "step": 2540
    },
    {
      "epoch": 0.13121611154144075,
      "grad_norm": 6.843015670776367,
      "learning_rate": 9.634728099907247e-06,
      "loss": 0.0862,
      "step": 2541
    },
    {
      "epoch": 0.13126775109734057,
      "grad_norm": 2.769355535507202,
      "learning_rate": 9.634420617075801e-06,
      "loss": 0.072,
      "step": 2542
    },
    {
      "epoch": 0.1313193906532404,
      "grad_norm": 8.641337394714355,
      "learning_rate": 9.634113009791271e-06,
      "loss": 0.0739,
      "step": 2543
    },
    {
      "epoch": 0.1313710302091402,
      "grad_norm": 6.068314552307129,
      "learning_rate": 9.63380527806192e-06,
      "loss": 0.069,
      "step": 2544
    },
    {
      "epoch": 0.13142266976504002,
      "grad_norm": 6.529634952545166,
      "learning_rate": 9.63349742189601e-06,
      "loss": 0.0862,
      "step": 2545
    },
    {
      "epoch": 0.13147430932093984,
      "grad_norm": 4.313961029052734,
      "learning_rate": 9.633189441301812e-06,
      "loss": 0.0871,
      "step": 2546
    },
    {
      "epoch": 0.13152594887683966,
      "grad_norm": 3.185741901397705,
      "learning_rate": 9.63288133628759e-06,
      "loss": 0.0534,
      "step": 2547
    },
    {
      "epoch": 0.13157758843273948,
      "grad_norm": 15.054586410522461,
      "learning_rate": 9.632573106861622e-06,
      "loss": 0.2333,
      "step": 2548
    },
    {
      "epoch": 0.1316292279886393,
      "grad_norm": 8.52761173248291,
      "learning_rate": 9.632264753032186e-06,
      "loss": 0.1158,
      "step": 2549
    },
    {
      "epoch": 0.13168086754453912,
      "grad_norm": 7.4901041984558105,
      "learning_rate": 9.631956274807561e-06,
      "loss": 0.105,
      "step": 2550
    },
    {
      "epoch": 0.13173250710043893,
      "grad_norm": 5.211860179901123,
      "learning_rate": 9.631647672196031e-06,
      "loss": 0.1263,
      "step": 2551
    },
    {
      "epoch": 0.13178414665633875,
      "grad_norm": 2.4992542266845703,
      "learning_rate": 9.631338945205885e-06,
      "loss": 0.0369,
      "step": 2552
    },
    {
      "epoch": 0.13183578621223857,
      "grad_norm": 7.706208229064941,
      "learning_rate": 9.63103009384541e-06,
      "loss": 0.1018,
      "step": 2553
    },
    {
      "epoch": 0.1318874257681384,
      "grad_norm": 7.43739652633667,
      "learning_rate": 9.630721118122903e-06,
      "loss": 0.0695,
      "step": 2554
    },
    {
      "epoch": 0.1319390653240382,
      "grad_norm": 11.444512367248535,
      "learning_rate": 9.63041201804666e-06,
      "loss": 0.2654,
      "step": 2555
    },
    {
      "epoch": 0.13199070487993803,
      "grad_norm": 9.041345596313477,
      "learning_rate": 9.630102793624982e-06,
      "loss": 0.1412,
      "step": 2556
    },
    {
      "epoch": 0.13204234443583784,
      "grad_norm": 3.423325538635254,
      "learning_rate": 9.629793444866171e-06,
      "loss": 0.0444,
      "step": 2557
    },
    {
      "epoch": 0.13209398399173766,
      "grad_norm": 4.971131324768066,
      "learning_rate": 9.629483971778539e-06,
      "loss": 0.0539,
      "step": 2558
    },
    {
      "epoch": 0.13214562354763748,
      "grad_norm": 4.170564651489258,
      "learning_rate": 9.629174374370392e-06,
      "loss": 0.0551,
      "step": 2559
    },
    {
      "epoch": 0.1321972631035373,
      "grad_norm": 16.4534912109375,
      "learning_rate": 9.628864652650045e-06,
      "loss": 0.146,
      "step": 2560
    },
    {
      "epoch": 0.13224890265943712,
      "grad_norm": 11.205957412719727,
      "learning_rate": 9.628554806625817e-06,
      "loss": 0.2105,
      "step": 2561
    },
    {
      "epoch": 0.13230054221533694,
      "grad_norm": 3.680671215057373,
      "learning_rate": 9.628244836306026e-06,
      "loss": 0.0737,
      "step": 2562
    },
    {
      "epoch": 0.13235218177123675,
      "grad_norm": 6.547028064727783,
      "learning_rate": 9.627934741698999e-06,
      "loss": 0.1181,
      "step": 2563
    },
    {
      "epoch": 0.1324038213271366,
      "grad_norm": 6.751556873321533,
      "learning_rate": 9.627624522813062e-06,
      "loss": 0.1141,
      "step": 2564
    },
    {
      "epoch": 0.13245546088303642,
      "grad_norm": 6.8021416664123535,
      "learning_rate": 9.627314179656544e-06,
      "loss": 0.1395,
      "step": 2565
    },
    {
      "epoch": 0.13250710043893624,
      "grad_norm": 7.344474792480469,
      "learning_rate": 9.62700371223778e-06,
      "loss": 0.1321,
      "step": 2566
    },
    {
      "epoch": 0.13255873999483606,
      "grad_norm": 16.169952392578125,
      "learning_rate": 9.62669312056511e-06,
      "loss": 0.1163,
      "step": 2567
    },
    {
      "epoch": 0.13261037955073587,
      "grad_norm": 27.560102462768555,
      "learning_rate": 9.62638240464687e-06,
      "loss": 0.0715,
      "step": 2568
    },
    {
      "epoch": 0.1326620191066357,
      "grad_norm": 4.606846809387207,
      "learning_rate": 9.626071564491408e-06,
      "loss": 0.0643,
      "step": 2569
    },
    {
      "epoch": 0.1327136586625355,
      "grad_norm": 4.2896504402160645,
      "learning_rate": 9.625760600107069e-06,
      "loss": 0.0789,
      "step": 2570
    },
    {
      "epoch": 0.13276529821843533,
      "grad_norm": 4.489531517028809,
      "learning_rate": 9.625449511502204e-06,
      "loss": 0.0925,
      "step": 2571
    },
    {
      "epoch": 0.13281693777433515,
      "grad_norm": 11.943916320800781,
      "learning_rate": 9.625138298685169e-06,
      "loss": 0.1123,
      "step": 2572
    },
    {
      "epoch": 0.13286857733023497,
      "grad_norm": 4.684223651885986,
      "learning_rate": 9.624826961664317e-06,
      "loss": 0.0787,
      "step": 2573
    },
    {
      "epoch": 0.13292021688613478,
      "grad_norm": 4.627407073974609,
      "learning_rate": 9.624515500448012e-06,
      "loss": 0.0676,
      "step": 2574
    },
    {
      "epoch": 0.1329718564420346,
      "grad_norm": 3.636462450027466,
      "learning_rate": 9.62420391504462e-06,
      "loss": 0.0736,
      "step": 2575
    },
    {
      "epoch": 0.13302349599793442,
      "grad_norm": 5.359170913696289,
      "learning_rate": 9.623892205462503e-06,
      "loss": 0.0985,
      "step": 2576
    },
    {
      "epoch": 0.13307513555383424,
      "grad_norm": 3.766188144683838,
      "learning_rate": 9.623580371710035e-06,
      "loss": 0.0851,
      "step": 2577
    },
    {
      "epoch": 0.13312677510973406,
      "grad_norm": 3.2804205417633057,
      "learning_rate": 9.623268413795589e-06,
      "loss": 0.0831,
      "step": 2578
    },
    {
      "epoch": 0.13317841466563388,
      "grad_norm": 5.23653507232666,
      "learning_rate": 9.622956331727544e-06,
      "loss": 0.0603,
      "step": 2579
    },
    {
      "epoch": 0.1332300542215337,
      "grad_norm": 12.997016906738281,
      "learning_rate": 9.622644125514278e-06,
      "loss": 0.1119,
      "step": 2580
    },
    {
      "epoch": 0.1332816937774335,
      "grad_norm": 4.402700901031494,
      "learning_rate": 9.622331795164177e-06,
      "loss": 0.0534,
      "step": 2581
    },
    {
      "epoch": 0.13333333333333333,
      "grad_norm": 7.317158222198486,
      "learning_rate": 9.622019340685627e-06,
      "loss": 0.0873,
      "step": 2582
    },
    {
      "epoch": 0.13338497288923315,
      "grad_norm": 6.468391418457031,
      "learning_rate": 9.62170676208702e-06,
      "loss": 0.0973,
      "step": 2583
    },
    {
      "epoch": 0.13343661244513297,
      "grad_norm": 26.122102737426758,
      "learning_rate": 9.62139405937675e-06,
      "loss": 0.1088,
      "step": 2584
    },
    {
      "epoch": 0.1334882520010328,
      "grad_norm": 8.42949390411377,
      "learning_rate": 9.621081232563212e-06,
      "loss": 0.145,
      "step": 2585
    },
    {
      "epoch": 0.1335398915569326,
      "grad_norm": 18.748414993286133,
      "learning_rate": 9.62076828165481e-06,
      "loss": 0.0574,
      "step": 2586
    },
    {
      "epoch": 0.13359153111283242,
      "grad_norm": 11.419737815856934,
      "learning_rate": 9.620455206659946e-06,
      "loss": 0.0974,
      "step": 2587
    },
    {
      "epoch": 0.13364317066873224,
      "grad_norm": 14.515474319458008,
      "learning_rate": 9.620142007587028e-06,
      "loss": 0.205,
      "step": 2588
    },
    {
      "epoch": 0.13369481022463206,
      "grad_norm": 10.737327575683594,
      "learning_rate": 9.619828684444465e-06,
      "loss": 0.1939,
      "step": 2589
    },
    {
      "epoch": 0.13374644978053188,
      "grad_norm": 3.8698770999908447,
      "learning_rate": 9.619515237240673e-06,
      "loss": 0.0436,
      "step": 2590
    },
    {
      "epoch": 0.1337980893364317,
      "grad_norm": 4.364255428314209,
      "learning_rate": 9.61920166598407e-06,
      "loss": 0.0832,
      "step": 2591
    },
    {
      "epoch": 0.13384972889233152,
      "grad_norm": 5.724700450897217,
      "learning_rate": 9.618887970683076e-06,
      "loss": 0.0979,
      "step": 2592
    },
    {
      "epoch": 0.13390136844823133,
      "grad_norm": 5.3790364265441895,
      "learning_rate": 9.618574151346113e-06,
      "loss": 0.0954,
      "step": 2593
    },
    {
      "epoch": 0.13395300800413115,
      "grad_norm": 20.540279388427734,
      "learning_rate": 9.61826020798161e-06,
      "loss": 0.1509,
      "step": 2594
    },
    {
      "epoch": 0.13400464756003097,
      "grad_norm": 7.884302616119385,
      "learning_rate": 9.617946140597998e-06,
      "loss": 0.0949,
      "step": 2595
    },
    {
      "epoch": 0.13405628711593082,
      "grad_norm": 3.047114372253418,
      "learning_rate": 9.61763194920371e-06,
      "loss": 0.077,
      "step": 2596
    },
    {
      "epoch": 0.13410792667183064,
      "grad_norm": 8.215514183044434,
      "learning_rate": 9.617317633807184e-06,
      "loss": 0.0742,
      "step": 2597
    },
    {
      "epoch": 0.13415956622773045,
      "grad_norm": 21.03900718688965,
      "learning_rate": 9.617003194416858e-06,
      "loss": 0.128,
      "step": 2598
    },
    {
      "epoch": 0.13421120578363027,
      "grad_norm": 6.911365985870361,
      "learning_rate": 9.616688631041182e-06,
      "loss": 0.0768,
      "step": 2599
    },
    {
      "epoch": 0.1342628453395301,
      "grad_norm": 4.025300979614258,
      "learning_rate": 9.616373943688598e-06,
      "loss": 0.0598,
      "step": 2600
    },
    {
      "epoch": 0.1343144848954299,
      "grad_norm": 2.6301934719085693,
      "learning_rate": 9.616059132367557e-06,
      "loss": 0.0708,
      "step": 2601
    },
    {
      "epoch": 0.13436612445132973,
      "grad_norm": 14.183063507080078,
      "learning_rate": 9.615744197086517e-06,
      "loss": 0.1146,
      "step": 2602
    },
    {
      "epoch": 0.13441776400722955,
      "grad_norm": 3.5525496006011963,
      "learning_rate": 9.61542913785393e-06,
      "loss": 0.0621,
      "step": 2603
    },
    {
      "epoch": 0.13446940356312936,
      "grad_norm": 22.95645523071289,
      "learning_rate": 9.615113954678261e-06,
      "loss": 0.2461,
      "step": 2604
    },
    {
      "epoch": 0.13452104311902918,
      "grad_norm": 7.479804515838623,
      "learning_rate": 9.614798647567972e-06,
      "loss": 0.0908,
      "step": 2605
    },
    {
      "epoch": 0.134572682674929,
      "grad_norm": 13.737288475036621,
      "learning_rate": 9.61448321653153e-06,
      "loss": 0.1034,
      "step": 2606
    },
    {
      "epoch": 0.13462432223082882,
      "grad_norm": 2.5322158336639404,
      "learning_rate": 9.614167661577404e-06,
      "loss": 0.0215,
      "step": 2607
    },
    {
      "epoch": 0.13467596178672864,
      "grad_norm": 4.463973045349121,
      "learning_rate": 9.613851982714073e-06,
      "loss": 0.0781,
      "step": 2608
    },
    {
      "epoch": 0.13472760134262846,
      "grad_norm": 3.1884727478027344,
      "learning_rate": 9.61353617995001e-06,
      "loss": 0.0747,
      "step": 2609
    },
    {
      "epoch": 0.13477924089852827,
      "grad_norm": 8.861504554748535,
      "learning_rate": 9.613220253293698e-06,
      "loss": 0.0565,
      "step": 2610
    },
    {
      "epoch": 0.1348308804544281,
      "grad_norm": 20.135202407836914,
      "learning_rate": 9.61290420275362e-06,
      "loss": 0.0922,
      "step": 2611
    },
    {
      "epoch": 0.1348825200103279,
      "grad_norm": 10.223929405212402,
      "learning_rate": 9.61258802833826e-06,
      "loss": 0.0863,
      "step": 2612
    },
    {
      "epoch": 0.13493415956622773,
      "grad_norm": 6.9574174880981445,
      "learning_rate": 9.612271730056113e-06,
      "loss": 0.0729,
      "step": 2613
    },
    {
      "epoch": 0.13498579912212755,
      "grad_norm": 4.667187690734863,
      "learning_rate": 9.611955307915672e-06,
      "loss": 0.0492,
      "step": 2614
    },
    {
      "epoch": 0.13503743867802737,
      "grad_norm": 4.369653224945068,
      "learning_rate": 9.611638761925435e-06,
      "loss": 0.0579,
      "step": 2615
    },
    {
      "epoch": 0.13508907823392718,
      "grad_norm": 4.709934234619141,
      "learning_rate": 9.611322092093899e-06,
      "loss": 0.0509,
      "step": 2616
    },
    {
      "epoch": 0.135140717789827,
      "grad_norm": 4.808836936950684,
      "learning_rate": 9.61100529842957e-06,
      "loss": 0.0817,
      "step": 2617
    },
    {
      "epoch": 0.13519235734572682,
      "grad_norm": 14.328741073608398,
      "learning_rate": 9.610688380940957e-06,
      "loss": 0.1926,
      "step": 2618
    },
    {
      "epoch": 0.13524399690162664,
      "grad_norm": 7.055743217468262,
      "learning_rate": 9.610371339636569e-06,
      "loss": 0.0748,
      "step": 2619
    },
    {
      "epoch": 0.13529563645752646,
      "grad_norm": 8.68529224395752,
      "learning_rate": 9.610054174524917e-06,
      "loss": 0.0681,
      "step": 2620
    },
    {
      "epoch": 0.13534727601342628,
      "grad_norm": 24.517913818359375,
      "learning_rate": 9.609736885614523e-06,
      "loss": 0.259,
      "step": 2621
    },
    {
      "epoch": 0.1353989155693261,
      "grad_norm": 8.28956127166748,
      "learning_rate": 9.609419472913906e-06,
      "loss": 0.0983,
      "step": 2622
    },
    {
      "epoch": 0.1354505551252259,
      "grad_norm": 29.302717208862305,
      "learning_rate": 9.609101936431588e-06,
      "loss": 0.0831,
      "step": 2623
    },
    {
      "epoch": 0.13550219468112573,
      "grad_norm": 3.482367515563965,
      "learning_rate": 9.608784276176097e-06,
      "loss": 0.0157,
      "step": 2624
    },
    {
      "epoch": 0.13555383423702555,
      "grad_norm": 9.027856826782227,
      "learning_rate": 9.608466492155966e-06,
      "loss": 0.0799,
      "step": 2625
    },
    {
      "epoch": 0.13560547379292537,
      "grad_norm": 5.1726460456848145,
      "learning_rate": 9.608148584379724e-06,
      "loss": 0.0876,
      "step": 2626
    },
    {
      "epoch": 0.1356571133488252,
      "grad_norm": 6.425196170806885,
      "learning_rate": 9.607830552855913e-06,
      "loss": 0.0526,
      "step": 2627
    },
    {
      "epoch": 0.13570875290472503,
      "grad_norm": 4.053124904632568,
      "learning_rate": 9.607512397593068e-06,
      "loss": 0.0617,
      "step": 2628
    },
    {
      "epoch": 0.13576039246062485,
      "grad_norm": 6.848910331726074,
      "learning_rate": 9.60719411859974e-06,
      "loss": 0.0891,
      "step": 2629
    },
    {
      "epoch": 0.13581203201652467,
      "grad_norm": 9.918862342834473,
      "learning_rate": 9.606875715884469e-06,
      "loss": 0.1157,
      "step": 2630
    },
    {
      "epoch": 0.1358636715724245,
      "grad_norm": 9.728243827819824,
      "learning_rate": 9.606557189455808e-06,
      "loss": 0.1574,
      "step": 2631
    },
    {
      "epoch": 0.1359153111283243,
      "grad_norm": 2.261666774749756,
      "learning_rate": 9.606238539322312e-06,
      "loss": 0.0403,
      "step": 2632
    },
    {
      "epoch": 0.13596695068422412,
      "grad_norm": 7.453363418579102,
      "learning_rate": 9.605919765492539e-06,
      "loss": 0.0735,
      "step": 2633
    },
    {
      "epoch": 0.13601859024012394,
      "grad_norm": 7.168885707855225,
      "learning_rate": 9.605600867975047e-06,
      "loss": 0.1191,
      "step": 2634
    },
    {
      "epoch": 0.13607022979602376,
      "grad_norm": 3.867342472076416,
      "learning_rate": 9.6052818467784e-06,
      "loss": 0.0815,
      "step": 2635
    },
    {
      "epoch": 0.13612186935192358,
      "grad_norm": 5.3179216384887695,
      "learning_rate": 9.604962701911165e-06,
      "loss": 0.0667,
      "step": 2636
    },
    {
      "epoch": 0.1361735089078234,
      "grad_norm": 12.638437271118164,
      "learning_rate": 9.604643433381911e-06,
      "loss": 0.1215,
      "step": 2637
    },
    {
      "epoch": 0.13622514846372322,
      "grad_norm": 4.209962368011475,
      "learning_rate": 9.604324041199216e-06,
      "loss": 0.0412,
      "step": 2638
    },
    {
      "epoch": 0.13627678801962304,
      "grad_norm": 5.134469985961914,
      "learning_rate": 9.604004525371653e-06,
      "loss": 0.0682,
      "step": 2639
    },
    {
      "epoch": 0.13632842757552285,
      "grad_norm": 6.614873886108398,
      "learning_rate": 9.603684885907804e-06,
      "loss": 0.0739,
      "step": 2640
    },
    {
      "epoch": 0.13638006713142267,
      "grad_norm": 22.323890686035156,
      "learning_rate": 9.603365122816251e-06,
      "loss": 0.1964,
      "step": 2641
    },
    {
      "epoch": 0.1364317066873225,
      "grad_norm": 2.558115243911743,
      "learning_rate": 9.603045236105583e-06,
      "loss": 0.0623,
      "step": 2642
    },
    {
      "epoch": 0.1364833462432223,
      "grad_norm": 4.7577900886535645,
      "learning_rate": 9.602725225784391e-06,
      "loss": 0.0442,
      "step": 2643
    },
    {
      "epoch": 0.13653498579912213,
      "grad_norm": 26.65488624572754,
      "learning_rate": 9.602405091861267e-06,
      "loss": 0.0879,
      "step": 2644
    },
    {
      "epoch": 0.13658662535502195,
      "grad_norm": 8.36832046508789,
      "learning_rate": 9.602084834344804e-06,
      "loss": 0.1071,
      "step": 2645
    },
    {
      "epoch": 0.13663826491092176,
      "grad_norm": 40.27882385253906,
      "learning_rate": 9.60176445324361e-06,
      "loss": 0.2317,
      "step": 2646
    },
    {
      "epoch": 0.13668990446682158,
      "grad_norm": 4.81992769241333,
      "learning_rate": 9.601443948566284e-06,
      "loss": 0.054,
      "step": 2647
    },
    {
      "epoch": 0.1367415440227214,
      "grad_norm": 4.519228458404541,
      "learning_rate": 9.601123320321433e-06,
      "loss": 0.0717,
      "step": 2648
    },
    {
      "epoch": 0.13679318357862122,
      "grad_norm": 6.601688861846924,
      "learning_rate": 9.600802568517667e-06,
      "loss": 0.0918,
      "step": 2649
    },
    {
      "epoch": 0.13684482313452104,
      "grad_norm": 20.964048385620117,
      "learning_rate": 9.600481693163602e-06,
      "loss": 0.1597,
      "step": 2650
    },
    {
      "epoch": 0.13689646269042086,
      "grad_norm": 7.606208801269531,
      "learning_rate": 9.600160694267852e-06,
      "loss": 0.067,
      "step": 2651
    },
    {
      "epoch": 0.13694810224632067,
      "grad_norm": 4.050169467926025,
      "learning_rate": 9.599839571839038e-06,
      "loss": 0.0986,
      "step": 2652
    },
    {
      "epoch": 0.1369997418022205,
      "grad_norm": 8.879363059997559,
      "learning_rate": 9.599518325885783e-06,
      "loss": 0.1183,
      "step": 2653
    },
    {
      "epoch": 0.1370513813581203,
      "grad_norm": 4.528372764587402,
      "learning_rate": 9.599196956416716e-06,
      "loss": 0.0789,
      "step": 2654
    },
    {
      "epoch": 0.13710302091402013,
      "grad_norm": 26.28464126586914,
      "learning_rate": 9.598875463440464e-06,
      "loss": 0.1559,
      "step": 2655
    },
    {
      "epoch": 0.13715466046991995,
      "grad_norm": 11.653474807739258,
      "learning_rate": 9.598553846965662e-06,
      "loss": 0.1326,
      "step": 2656
    },
    {
      "epoch": 0.13720630002581977,
      "grad_norm": 3.4576773643493652,
      "learning_rate": 9.598232107000947e-06,
      "loss": 0.0598,
      "step": 2657
    },
    {
      "epoch": 0.13725793958171958,
      "grad_norm": 2.9472763538360596,
      "learning_rate": 9.597910243554958e-06,
      "loss": 0.0612,
      "step": 2658
    },
    {
      "epoch": 0.1373095791376194,
      "grad_norm": 3.7325966358184814,
      "learning_rate": 9.59758825663634e-06,
      "loss": 0.0624,
      "step": 2659
    },
    {
      "epoch": 0.13736121869351925,
      "grad_norm": 9.64785385131836,
      "learning_rate": 9.597266146253737e-06,
      "loss": 0.1258,
      "step": 2660
    },
    {
      "epoch": 0.13741285824941907,
      "grad_norm": 10.438996315002441,
      "learning_rate": 9.596943912415802e-06,
      "loss": 0.078,
      "step": 2661
    },
    {
      "epoch": 0.13746449780531889,
      "grad_norm": 8.604681968688965,
      "learning_rate": 9.596621555131185e-06,
      "loss": 0.0904,
      "step": 2662
    },
    {
      "epoch": 0.1375161373612187,
      "grad_norm": 5.488528251647949,
      "learning_rate": 9.596299074408547e-06,
      "loss": 0.0873,
      "step": 2663
    },
    {
      "epoch": 0.13756777691711852,
      "grad_norm": 5.267910003662109,
      "learning_rate": 9.595976470256543e-06,
      "loss": 0.0529,
      "step": 2664
    },
    {
      "epoch": 0.13761941647301834,
      "grad_norm": 10.213601112365723,
      "learning_rate": 9.59565374268384e-06,
      "loss": 0.1826,
      "step": 2665
    },
    {
      "epoch": 0.13767105602891816,
      "grad_norm": 14.176304817199707,
      "learning_rate": 9.595330891699104e-06,
      "loss": 0.1118,
      "step": 2666
    },
    {
      "epoch": 0.13772269558481798,
      "grad_norm": 9.153640747070312,
      "learning_rate": 9.595007917311003e-06,
      "loss": 0.1514,
      "step": 2667
    },
    {
      "epoch": 0.1377743351407178,
      "grad_norm": 7.386855602264404,
      "learning_rate": 9.594684819528211e-06,
      "loss": 0.1192,
      "step": 2668
    },
    {
      "epoch": 0.13782597469661761,
      "grad_norm": 3.6340065002441406,
      "learning_rate": 9.594361598359404e-06,
      "loss": 0.0514,
      "step": 2669
    },
    {
      "epoch": 0.13787761425251743,
      "grad_norm": 5.973364353179932,
      "learning_rate": 9.594038253813264e-06,
      "loss": 0.1028,
      "step": 2670
    },
    {
      "epoch": 0.13792925380841725,
      "grad_norm": 9.825709342956543,
      "learning_rate": 9.59371478589847e-06,
      "loss": 0.0868,
      "step": 2671
    },
    {
      "epoch": 0.13798089336431707,
      "grad_norm": 3.1844425201416016,
      "learning_rate": 9.593391194623715e-06,
      "loss": 0.0786,
      "step": 2672
    },
    {
      "epoch": 0.1380325329202169,
      "grad_norm": 11.369342803955078,
      "learning_rate": 9.593067479997682e-06,
      "loss": 0.1412,
      "step": 2673
    },
    {
      "epoch": 0.1380841724761167,
      "grad_norm": 8.237730026245117,
      "learning_rate": 9.592743642029068e-06,
      "loss": 0.1054,
      "step": 2674
    },
    {
      "epoch": 0.13813581203201653,
      "grad_norm": 3.2943413257598877,
      "learning_rate": 9.592419680726568e-06,
      "loss": 0.047,
      "step": 2675
    },
    {
      "epoch": 0.13818745158791634,
      "grad_norm": 2.7410948276519775,
      "learning_rate": 9.592095596098882e-06,
      "loss": 0.0546,
      "step": 2676
    },
    {
      "epoch": 0.13823909114381616,
      "grad_norm": 11.035654067993164,
      "learning_rate": 9.591771388154712e-06,
      "loss": 0.1032,
      "step": 2677
    },
    {
      "epoch": 0.13829073069971598,
      "grad_norm": 8.90149211883545,
      "learning_rate": 9.591447056902767e-06,
      "loss": 0.1363,
      "step": 2678
    },
    {
      "epoch": 0.1383423702556158,
      "grad_norm": 3.3391635417938232,
      "learning_rate": 9.591122602351752e-06,
      "loss": 0.0819,
      "step": 2679
    },
    {
      "epoch": 0.13839400981151562,
      "grad_norm": 3.151822090148926,
      "learning_rate": 9.590798024510387e-06,
      "loss": 0.0574,
      "step": 2680
    },
    {
      "epoch": 0.13844564936741544,
      "grad_norm": 6.5341901779174805,
      "learning_rate": 9.59047332338738e-06,
      "loss": 0.0744,
      "step": 2681
    },
    {
      "epoch": 0.13849728892331525,
      "grad_norm": 8.318450927734375,
      "learning_rate": 9.590148498991456e-06,
      "loss": 0.0337,
      "step": 2682
    },
    {
      "epoch": 0.13854892847921507,
      "grad_norm": 9.104928970336914,
      "learning_rate": 9.589823551331337e-06,
      "loss": 0.1196,
      "step": 2683
    },
    {
      "epoch": 0.1386005680351149,
      "grad_norm": 3.6162662506103516,
      "learning_rate": 9.589498480415747e-06,
      "loss": 0.1235,
      "step": 2684
    },
    {
      "epoch": 0.1386522075910147,
      "grad_norm": 2.6793439388275146,
      "learning_rate": 9.589173286253416e-06,
      "loss": 0.0473,
      "step": 2685
    },
    {
      "epoch": 0.13870384714691453,
      "grad_norm": 5.906521797180176,
      "learning_rate": 9.58884796885308e-06,
      "loss": 0.0564,
      "step": 2686
    },
    {
      "epoch": 0.13875548670281435,
      "grad_norm": 3.035371780395508,
      "learning_rate": 9.588522528223475e-06,
      "loss": 0.0548,
      "step": 2687
    },
    {
      "epoch": 0.13880712625871416,
      "grad_norm": 4.439809322357178,
      "learning_rate": 9.588196964373336e-06,
      "loss": 0.0505,
      "step": 2688
    },
    {
      "epoch": 0.13885876581461398,
      "grad_norm": 15.910266876220703,
      "learning_rate": 9.587871277311407e-06,
      "loss": 0.2682,
      "step": 2689
    },
    {
      "epoch": 0.1389104053705138,
      "grad_norm": 11.69151496887207,
      "learning_rate": 9.587545467046436e-06,
      "loss": 0.1365,
      "step": 2690
    },
    {
      "epoch": 0.13896204492641362,
      "grad_norm": 4.094503402709961,
      "learning_rate": 9.587219533587172e-06,
      "loss": 0.0532,
      "step": 2691
    },
    {
      "epoch": 0.13901368448231347,
      "grad_norm": 7.772972106933594,
      "learning_rate": 9.586893476942367e-06,
      "loss": 0.0931,
      "step": 2692
    },
    {
      "epoch": 0.13906532403821328,
      "grad_norm": 16.114505767822266,
      "learning_rate": 9.586567297120776e-06,
      "loss": 0.1573,
      "step": 2693
    },
    {
      "epoch": 0.1391169635941131,
      "grad_norm": 3.050840377807617,
      "learning_rate": 9.58624099413116e-06,
      "loss": 0.037,
      "step": 2694
    },
    {
      "epoch": 0.13916860315001292,
      "grad_norm": 3.7094902992248535,
      "learning_rate": 9.58591456798228e-06,
      "loss": 0.0869,
      "step": 2695
    },
    {
      "epoch": 0.13922024270591274,
      "grad_norm": 4.906955718994141,
      "learning_rate": 9.585588018682905e-06,
      "loss": 0.0782,
      "step": 2696
    },
    {
      "epoch": 0.13927188226181256,
      "grad_norm": 18.051361083984375,
      "learning_rate": 9.585261346241798e-06,
      "loss": 0.1263,
      "step": 2697
    },
    {
      "epoch": 0.13932352181771238,
      "grad_norm": 7.996084213256836,
      "learning_rate": 9.584934550667739e-06,
      "loss": 0.0941,
      "step": 2698
    },
    {
      "epoch": 0.1393751613736122,
      "grad_norm": 7.995449066162109,
      "learning_rate": 9.584607631969499e-06,
      "loss": 0.0811,
      "step": 2699
    },
    {
      "epoch": 0.139426800929512,
      "grad_norm": 3.827314853668213,
      "learning_rate": 9.584280590155858e-06,
      "loss": 0.0766,
      "step": 2700
    },
    {
      "epoch": 0.13947844048541183,
      "grad_norm": 9.715757369995117,
      "learning_rate": 9.5839534252356e-06,
      "loss": 0.1067,
      "step": 2701
    },
    {
      "epoch": 0.13953008004131165,
      "grad_norm": 5.291261672973633,
      "learning_rate": 9.583626137217506e-06,
      "loss": 0.0803,
      "step": 2702
    },
    {
      "epoch": 0.13958171959721147,
      "grad_norm": 9.974725723266602,
      "learning_rate": 9.583298726110371e-06,
      "loss": 0.0679,
      "step": 2703
    },
    {
      "epoch": 0.13963335915311129,
      "grad_norm": 10.098517417907715,
      "learning_rate": 9.582971191922985e-06,
      "loss": 0.1093,
      "step": 2704
    },
    {
      "epoch": 0.1396849987090111,
      "grad_norm": 4.342756271362305,
      "learning_rate": 9.582643534664142e-06,
      "loss": 0.0896,
      "step": 2705
    },
    {
      "epoch": 0.13973663826491092,
      "grad_norm": 6.094022274017334,
      "learning_rate": 9.582315754342643e-06,
      "loss": 0.1178,
      "step": 2706
    },
    {
      "epoch": 0.13978827782081074,
      "grad_norm": 12.23869800567627,
      "learning_rate": 9.581987850967288e-06,
      "loss": 0.113,
      "step": 2707
    },
    {
      "epoch": 0.13983991737671056,
      "grad_norm": 6.489724159240723,
      "learning_rate": 9.581659824546887e-06,
      "loss": 0.0957,
      "step": 2708
    },
    {
      "epoch": 0.13989155693261038,
      "grad_norm": 4.348581314086914,
      "learning_rate": 9.581331675090244e-06,
      "loss": 0.0703,
      "step": 2709
    },
    {
      "epoch": 0.1399431964885102,
      "grad_norm": 13.68412971496582,
      "learning_rate": 9.581003402606173e-06,
      "loss": 0.0755,
      "step": 2710
    },
    {
      "epoch": 0.13999483604441001,
      "grad_norm": 11.455692291259766,
      "learning_rate": 9.58067500710349e-06,
      "loss": 0.1889,
      "step": 2711
    },
    {
      "epoch": 0.14004647560030983,
      "grad_norm": 3.640138864517212,
      "learning_rate": 9.580346488591013e-06,
      "loss": 0.0724,
      "step": 2712
    },
    {
      "epoch": 0.14009811515620965,
      "grad_norm": 10.03262996673584,
      "learning_rate": 9.580017847077563e-06,
      "loss": 0.1406,
      "step": 2713
    },
    {
      "epoch": 0.14014975471210947,
      "grad_norm": 3.2535057067871094,
      "learning_rate": 9.579689082571969e-06,
      "loss": 0.0509,
      "step": 2714
    },
    {
      "epoch": 0.1402013942680093,
      "grad_norm": 6.467337608337402,
      "learning_rate": 9.579360195083056e-06,
      "loss": 0.0615,
      "step": 2715
    },
    {
      "epoch": 0.1402530338239091,
      "grad_norm": 9.859918594360352,
      "learning_rate": 9.579031184619658e-06,
      "loss": 0.0976,
      "step": 2716
    },
    {
      "epoch": 0.14030467337980893,
      "grad_norm": 4.095170021057129,
      "learning_rate": 9.57870205119061e-06,
      "loss": 0.051,
      "step": 2717
    },
    {
      "epoch": 0.14035631293570874,
      "grad_norm": 5.040444850921631,
      "learning_rate": 9.57837279480475e-06,
      "loss": 0.0418,
      "step": 2718
    },
    {
      "epoch": 0.14040795249160856,
      "grad_norm": 4.350162506103516,
      "learning_rate": 9.578043415470919e-06,
      "loss": 0.0435,
      "step": 2719
    },
    {
      "epoch": 0.14045959204750838,
      "grad_norm": 10.992706298828125,
      "learning_rate": 9.577713913197964e-06,
      "loss": 0.1564,
      "step": 2720
    },
    {
      "epoch": 0.1405112316034082,
      "grad_norm": 2.8581385612487793,
      "learning_rate": 9.577384287994733e-06,
      "loss": 0.057,
      "step": 2721
    },
    {
      "epoch": 0.14056287115930802,
      "grad_norm": 2.8083252906799316,
      "learning_rate": 9.577054539870077e-06,
      "loss": 0.0589,
      "step": 2722
    },
    {
      "epoch": 0.14061451071520784,
      "grad_norm": 6.1773905754089355,
      "learning_rate": 9.576724668832851e-06,
      "loss": 0.0763,
      "step": 2723
    },
    {
      "epoch": 0.14066615027110768,
      "grad_norm": 5.963743209838867,
      "learning_rate": 9.576394674891914e-06,
      "loss": 0.0543,
      "step": 2724
    },
    {
      "epoch": 0.1407177898270075,
      "grad_norm": 34.17905044555664,
      "learning_rate": 9.576064558056129e-06,
      "loss": 0.1823,
      "step": 2725
    },
    {
      "epoch": 0.14076942938290732,
      "grad_norm": 3.414379596710205,
      "learning_rate": 9.575734318334358e-06,
      "loss": 0.0798,
      "step": 2726
    },
    {
      "epoch": 0.14082106893880714,
      "grad_norm": 5.638921737670898,
      "learning_rate": 9.575403955735472e-06,
      "loss": 0.0792,
      "step": 2727
    },
    {
      "epoch": 0.14087270849470696,
      "grad_norm": 5.304111003875732,
      "learning_rate": 9.57507347026834e-06,
      "loss": 0.0616,
      "step": 2728
    },
    {
      "epoch": 0.14092434805060677,
      "grad_norm": 2.022301197052002,
      "learning_rate": 9.57474286194184e-06,
      "loss": 0.0342,
      "step": 2729
    },
    {
      "epoch": 0.1409759876065066,
      "grad_norm": 10.740569114685059,
      "learning_rate": 9.574412130764847e-06,
      "loss": 0.1028,
      "step": 2730
    },
    {
      "epoch": 0.1410276271624064,
      "grad_norm": 8.645272254943848,
      "learning_rate": 9.574081276746245e-06,
      "loss": 0.1116,
      "step": 2731
    },
    {
      "epoch": 0.14107926671830623,
      "grad_norm": 10.8047456741333,
      "learning_rate": 9.57375029989492e-06,
      "loss": 0.1742,
      "step": 2732
    },
    {
      "epoch": 0.14113090627420605,
      "grad_norm": 9.021689414978027,
      "learning_rate": 9.573419200219753e-06,
      "loss": 0.1072,
      "step": 2733
    },
    {
      "epoch": 0.14118254583010587,
      "grad_norm": 8.452527046203613,
      "learning_rate": 9.573087977729644e-06,
      "loss": 0.092,
      "step": 2734
    },
    {
      "epoch": 0.14123418538600568,
      "grad_norm": 6.991357326507568,
      "learning_rate": 9.572756632433483e-06,
      "loss": 0.0913,
      "step": 2735
    },
    {
      "epoch": 0.1412858249419055,
      "grad_norm": 2.651998996734619,
      "learning_rate": 9.572425164340169e-06,
      "loss": 0.0504,
      "step": 2736
    },
    {
      "epoch": 0.14133746449780532,
      "grad_norm": 11.11224365234375,
      "learning_rate": 9.572093573458604e-06,
      "loss": 0.1315,
      "step": 2737
    },
    {
      "epoch": 0.14138910405370514,
      "grad_norm": 11.356287956237793,
      "learning_rate": 9.571761859797688e-06,
      "loss": 0.245,
      "step": 2738
    },
    {
      "epoch": 0.14144074360960496,
      "grad_norm": 2.4880402088165283,
      "learning_rate": 9.571430023366336e-06,
      "loss": 0.0674,
      "step": 2739
    },
    {
      "epoch": 0.14149238316550478,
      "grad_norm": 5.4740071296691895,
      "learning_rate": 9.571098064173456e-06,
      "loss": 0.1214,
      "step": 2740
    },
    {
      "epoch": 0.1415440227214046,
      "grad_norm": 4.602292060852051,
      "learning_rate": 9.570765982227962e-06,
      "loss": 0.1095,
      "step": 2741
    },
    {
      "epoch": 0.1415956622773044,
      "grad_norm": 8.203034400939941,
      "learning_rate": 9.57043377753877e-06,
      "loss": 0.1288,
      "step": 2742
    },
    {
      "epoch": 0.14164730183320423,
      "grad_norm": 5.9269256591796875,
      "learning_rate": 9.570101450114803e-06,
      "loss": 0.0917,
      "step": 2743
    },
    {
      "epoch": 0.14169894138910405,
      "grad_norm": 4.584962368011475,
      "learning_rate": 9.569768999964988e-06,
      "loss": 0.0715,
      "step": 2744
    },
    {
      "epoch": 0.14175058094500387,
      "grad_norm": 8.321074485778809,
      "learning_rate": 9.569436427098248e-06,
      "loss": 0.0969,
      "step": 2745
    },
    {
      "epoch": 0.1418022205009037,
      "grad_norm": 3.682482957839966,
      "learning_rate": 9.569103731523518e-06,
      "loss": 0.064,
      "step": 2746
    },
    {
      "epoch": 0.1418538600568035,
      "grad_norm": 9.05919075012207,
      "learning_rate": 9.568770913249728e-06,
      "loss": 0.0781,
      "step": 2747
    },
    {
      "epoch": 0.14190549961270332,
      "grad_norm": 13.100460052490234,
      "learning_rate": 9.568437972285818e-06,
      "loss": 0.1257,
      "step": 2748
    },
    {
      "epoch": 0.14195713916860314,
      "grad_norm": 6.812422752380371,
      "learning_rate": 9.568104908640727e-06,
      "loss": 0.1342,
      "step": 2749
    },
    {
      "epoch": 0.14200877872450296,
      "grad_norm": 3.6911559104919434,
      "learning_rate": 9.567771722323403e-06,
      "loss": 0.0633,
      "step": 2750
    },
    {
      "epoch": 0.14206041828040278,
      "grad_norm": 6.8298659324646,
      "learning_rate": 9.567438413342791e-06,
      "loss": 0.1034,
      "step": 2751
    },
    {
      "epoch": 0.1421120578363026,
      "grad_norm": 8.382583618164062,
      "learning_rate": 9.567104981707841e-06,
      "loss": 0.1667,
      "step": 2752
    },
    {
      "epoch": 0.14216369739220241,
      "grad_norm": 4.163645267486572,
      "learning_rate": 9.566771427427508e-06,
      "loss": 0.0892,
      "step": 2753
    },
    {
      "epoch": 0.14221533694810223,
      "grad_norm": 7.2361578941345215,
      "learning_rate": 9.56643775051075e-06,
      "loss": 0.123,
      "step": 2754
    },
    {
      "epoch": 0.14226697650400205,
      "grad_norm": 13.128121376037598,
      "learning_rate": 9.566103950966525e-06,
      "loss": 0.1045,
      "step": 2755
    },
    {
      "epoch": 0.1423186160599019,
      "grad_norm": 2.258683681488037,
      "learning_rate": 9.5657700288038e-06,
      "loss": 0.0569,
      "step": 2756
    },
    {
      "epoch": 0.14237025561580172,
      "grad_norm": 2.947089195251465,
      "learning_rate": 9.565435984031538e-06,
      "loss": 0.0805,
      "step": 2757
    },
    {
      "epoch": 0.14242189517170153,
      "grad_norm": 3.5409963130950928,
      "learning_rate": 9.565101816658715e-06,
      "loss": 0.0584,
      "step": 2758
    },
    {
      "epoch": 0.14247353472760135,
      "grad_norm": 6.887526988983154,
      "learning_rate": 9.5647675266943e-06,
      "loss": 0.098,
      "step": 2759
    },
    {
      "epoch": 0.14252517428350117,
      "grad_norm": 3.411245584487915,
      "learning_rate": 9.564433114147273e-06,
      "loss": 0.0641,
      "step": 2760
    },
    {
      "epoch": 0.142576813839401,
      "grad_norm": 8.928794860839844,
      "learning_rate": 9.564098579026613e-06,
      "loss": 0.0618,
      "step": 2761
    },
    {
      "epoch": 0.1426284533953008,
      "grad_norm": 10.217972755432129,
      "learning_rate": 9.563763921341305e-06,
      "loss": 0.1372,
      "step": 2762
    },
    {
      "epoch": 0.14268009295120063,
      "grad_norm": 6.340932369232178,
      "learning_rate": 9.563429141100334e-06,
      "loss": 0.0853,
      "step": 2763
    },
    {
      "epoch": 0.14273173250710045,
      "grad_norm": 6.841894626617432,
      "learning_rate": 9.563094238312691e-06,
      "loss": 0.127,
      "step": 2764
    },
    {
      "epoch": 0.14278337206300026,
      "grad_norm": 4.677881240844727,
      "learning_rate": 9.562759212987369e-06,
      "loss": 0.0711,
      "step": 2765
    },
    {
      "epoch": 0.14283501161890008,
      "grad_norm": 14.046902656555176,
      "learning_rate": 9.562424065133367e-06,
      "loss": 0.1072,
      "step": 2766
    },
    {
      "epoch": 0.1428866511747999,
      "grad_norm": 2.747973680496216,
      "learning_rate": 9.562088794759681e-06,
      "loss": 0.046,
      "step": 2767
    },
    {
      "epoch": 0.14293829073069972,
      "grad_norm": 4.089222431182861,
      "learning_rate": 9.561753401875319e-06,
      "loss": 0.0266,
      "step": 2768
    },
    {
      "epoch": 0.14298993028659954,
      "grad_norm": 3.287097454071045,
      "learning_rate": 9.561417886489283e-06,
      "loss": 0.0386,
      "step": 2769
    },
    {
      "epoch": 0.14304156984249936,
      "grad_norm": 22.879194259643555,
      "learning_rate": 9.561082248610589e-06,
      "loss": 0.0724,
      "step": 2770
    },
    {
      "epoch": 0.14309320939839917,
      "grad_norm": 9.033236503601074,
      "learning_rate": 9.560746488248245e-06,
      "loss": 0.0799,
      "step": 2771
    },
    {
      "epoch": 0.143144848954299,
      "grad_norm": 5.653759479522705,
      "learning_rate": 9.560410605411268e-06,
      "loss": 0.1072,
      "step": 2772
    },
    {
      "epoch": 0.1431964885101988,
      "grad_norm": 15.814759254455566,
      "learning_rate": 9.56007460010868e-06,
      "loss": 0.1674,
      "step": 2773
    },
    {
      "epoch": 0.14324812806609863,
      "grad_norm": 7.942234516143799,
      "learning_rate": 9.559738472349504e-06,
      "loss": 0.1057,
      "step": 2774
    },
    {
      "epoch": 0.14329976762199845,
      "grad_norm": 4.81248664855957,
      "learning_rate": 9.559402222142763e-06,
      "loss": 0.0338,
      "step": 2775
    },
    {
      "epoch": 0.14335140717789827,
      "grad_norm": 8.138212203979492,
      "learning_rate": 9.559065849497492e-06,
      "loss": 0.0723,
      "step": 2776
    },
    {
      "epoch": 0.14340304673379808,
      "grad_norm": 13.203131675720215,
      "learning_rate": 9.55872935442272e-06,
      "loss": 0.1723,
      "step": 2777
    },
    {
      "epoch": 0.1434546862896979,
      "grad_norm": 13.939764022827148,
      "learning_rate": 9.558392736927485e-06,
      "loss": 0.1102,
      "step": 2778
    },
    {
      "epoch": 0.14350632584559772,
      "grad_norm": 13.974457740783691,
      "learning_rate": 9.558055997020825e-06,
      "loss": 0.0761,
      "step": 2779
    },
    {
      "epoch": 0.14355796540149754,
      "grad_norm": 36.95858383178711,
      "learning_rate": 9.557719134711785e-06,
      "loss": 0.1123,
      "step": 2780
    },
    {
      "epoch": 0.14360960495739736,
      "grad_norm": 5.646840572357178,
      "learning_rate": 9.557382150009408e-06,
      "loss": 0.0848,
      "step": 2781
    },
    {
      "epoch": 0.14366124451329718,
      "grad_norm": 16.199865341186523,
      "learning_rate": 9.557045042922747e-06,
      "loss": 0.1412,
      "step": 2782
    },
    {
      "epoch": 0.143712884069197,
      "grad_norm": 6.103509426116943,
      "learning_rate": 9.556707813460854e-06,
      "loss": 0.1044,
      "step": 2783
    },
    {
      "epoch": 0.1437645236250968,
      "grad_norm": 5.35120964050293,
      "learning_rate": 9.556370461632784e-06,
      "loss": 0.0512,
      "step": 2784
    },
    {
      "epoch": 0.14381616318099663,
      "grad_norm": 10.286111831665039,
      "learning_rate": 9.556032987447595e-06,
      "loss": 0.1726,
      "step": 2785
    },
    {
      "epoch": 0.14386780273689645,
      "grad_norm": 11.755876541137695,
      "learning_rate": 9.555695390914352e-06,
      "loss": 0.1674,
      "step": 2786
    },
    {
      "epoch": 0.1439194422927963,
      "grad_norm": 10.093710899353027,
      "learning_rate": 9.555357672042118e-06,
      "loss": 0.0996,
      "step": 2787
    },
    {
      "epoch": 0.14397108184869611,
      "grad_norm": 11.383150100708008,
      "learning_rate": 9.555019830839967e-06,
      "loss": 0.0964,
      "step": 2788
    },
    {
      "epoch": 0.14402272140459593,
      "grad_norm": 3.4376680850982666,
      "learning_rate": 9.554681867316965e-06,
      "loss": 0.1121,
      "step": 2789
    },
    {
      "epoch": 0.14407436096049575,
      "grad_norm": 3.6353635787963867,
      "learning_rate": 9.554343781482193e-06,
      "loss": 0.0536,
      "step": 2790
    },
    {
      "epoch": 0.14412600051639557,
      "grad_norm": 3.535234212875366,
      "learning_rate": 9.554005573344727e-06,
      "loss": 0.0731,
      "step": 2791
    },
    {
      "epoch": 0.1441776400722954,
      "grad_norm": 7.766324996948242,
      "learning_rate": 9.553667242913651e-06,
      "loss": 0.1136,
      "step": 2792
    },
    {
      "epoch": 0.1442292796281952,
      "grad_norm": 5.857366561889648,
      "learning_rate": 9.553328790198048e-06,
      "loss": 0.1009,
      "step": 2793
    },
    {
      "epoch": 0.14428091918409502,
      "grad_norm": 11.219318389892578,
      "learning_rate": 9.552990215207012e-06,
      "loss": 0.0918,
      "step": 2794
    },
    {
      "epoch": 0.14433255873999484,
      "grad_norm": 6.198333740234375,
      "learning_rate": 9.55265151794963e-06,
      "loss": 0.1124,
      "step": 2795
    },
    {
      "epoch": 0.14438419829589466,
      "grad_norm": 5.028686046600342,
      "learning_rate": 9.552312698434997e-06,
      "loss": 0.0619,
      "step": 2796
    },
    {
      "epoch": 0.14443583785179448,
      "grad_norm": 5.557506084442139,
      "learning_rate": 9.551973756672217e-06,
      "loss": 0.0719,
      "step": 2797
    },
    {
      "epoch": 0.1444874774076943,
      "grad_norm": 4.551172256469727,
      "learning_rate": 9.551634692670388e-06,
      "loss": 0.0795,
      "step": 2798
    },
    {
      "epoch": 0.14453911696359412,
      "grad_norm": 2.990403652191162,
      "learning_rate": 9.551295506438616e-06,
      "loss": 0.0528,
      "step": 2799
    },
    {
      "epoch": 0.14459075651949393,
      "grad_norm": 3.1044397354125977,
      "learning_rate": 9.55095619798601e-06,
      "loss": 0.0579,
      "step": 2800
    },
    {
      "epoch": 0.14464239607539375,
      "grad_norm": 3.807352304458618,
      "learning_rate": 9.55061676732168e-06,
      "loss": 0.0651,
      "step": 2801
    },
    {
      "epoch": 0.14469403563129357,
      "grad_norm": 4.778645038604736,
      "learning_rate": 9.550277214454744e-06,
      "loss": 0.0843,
      "step": 2802
    },
    {
      "epoch": 0.1447456751871934,
      "grad_norm": 16.946800231933594,
      "learning_rate": 9.549937539394317e-06,
      "loss": 0.1012,
      "step": 2803
    },
    {
      "epoch": 0.1447973147430932,
      "grad_norm": 6.812874794006348,
      "learning_rate": 9.549597742149524e-06,
      "loss": 0.0689,
      "step": 2804
    },
    {
      "epoch": 0.14484895429899303,
      "grad_norm": 13.054122924804688,
      "learning_rate": 9.549257822729488e-06,
      "loss": 0.0976,
      "step": 2805
    },
    {
      "epoch": 0.14490059385489285,
      "grad_norm": 4.337809085845947,
      "learning_rate": 9.54891778114334e-06,
      "loss": 0.0613,
      "step": 2806
    },
    {
      "epoch": 0.14495223341079266,
      "grad_norm": 7.148487567901611,
      "learning_rate": 9.548577617400205e-06,
      "loss": 0.0963,
      "step": 2807
    },
    {
      "epoch": 0.14500387296669248,
      "grad_norm": 3.809019088745117,
      "learning_rate": 9.548237331509225e-06,
      "loss": 0.0805,
      "step": 2808
    },
    {
      "epoch": 0.1450555125225923,
      "grad_norm": 3.22751784324646,
      "learning_rate": 9.547896923479534e-06,
      "loss": 0.0612,
      "step": 2809
    },
    {
      "epoch": 0.14510715207849212,
      "grad_norm": 6.1567559242248535,
      "learning_rate": 9.547556393320274e-06,
      "loss": 0.0745,
      "step": 2810
    },
    {
      "epoch": 0.14515879163439194,
      "grad_norm": 3.9791784286499023,
      "learning_rate": 9.54721574104059e-06,
      "loss": 0.0609,
      "step": 2811
    },
    {
      "epoch": 0.14521043119029176,
      "grad_norm": 7.597418785095215,
      "learning_rate": 9.546874966649631e-06,
      "loss": 0.0723,
      "step": 2812
    },
    {
      "epoch": 0.14526207074619157,
      "grad_norm": 6.338715076446533,
      "learning_rate": 9.546534070156546e-06,
      "loss": 0.1462,
      "step": 2813
    },
    {
      "epoch": 0.1453137103020914,
      "grad_norm": 14.66997241973877,
      "learning_rate": 9.54619305157049e-06,
      "loss": 0.0919,
      "step": 2814
    },
    {
      "epoch": 0.1453653498579912,
      "grad_norm": 8.410183906555176,
      "learning_rate": 9.545851910900626e-06,
      "loss": 0.0726,
      "step": 2815
    },
    {
      "epoch": 0.14541698941389103,
      "grad_norm": 13.036178588867188,
      "learning_rate": 9.545510648156106e-06,
      "loss": 0.1127,
      "step": 2816
    },
    {
      "epoch": 0.14546862896979085,
      "grad_norm": 15.55511474609375,
      "learning_rate": 9.545169263346099e-06,
      "loss": 0.1864,
      "step": 2817
    },
    {
      "epoch": 0.14552026852569067,
      "grad_norm": 13.307757377624512,
      "learning_rate": 9.544827756479773e-06,
      "loss": 0.1469,
      "step": 2818
    },
    {
      "epoch": 0.1455719080815905,
      "grad_norm": 4.862357139587402,
      "learning_rate": 9.544486127566297e-06,
      "loss": 0.073,
      "step": 2819
    },
    {
      "epoch": 0.14562354763749033,
      "grad_norm": 6.9930524826049805,
      "learning_rate": 9.544144376614848e-06,
      "loss": 0.0893,
      "step": 2820
    },
    {
      "epoch": 0.14567518719339015,
      "grad_norm": 9.250120162963867,
      "learning_rate": 9.5438025036346e-06,
      "loss": 0.0864,
      "step": 2821
    },
    {
      "epoch": 0.14572682674928997,
      "grad_norm": 9.282038688659668,
      "learning_rate": 9.543460508634735e-06,
      "loss": 0.0974,
      "step": 2822
    },
    {
      "epoch": 0.14577846630518979,
      "grad_norm": 17.73465347290039,
      "learning_rate": 9.543118391624439e-06,
      "loss": 0.1858,
      "step": 2823
    },
    {
      "epoch": 0.1458301058610896,
      "grad_norm": 3.961688756942749,
      "learning_rate": 9.542776152612896e-06,
      "loss": 0.0871,
      "step": 2824
    },
    {
      "epoch": 0.14588174541698942,
      "grad_norm": 5.138996124267578,
      "learning_rate": 9.542433791609299e-06,
      "loss": 0.1142,
      "step": 2825
    },
    {
      "epoch": 0.14593338497288924,
      "grad_norm": 3.1296770572662354,
      "learning_rate": 9.54209130862284e-06,
      "loss": 0.0526,
      "step": 2826
    },
    {
      "epoch": 0.14598502452878906,
      "grad_norm": 9.636285781860352,
      "learning_rate": 9.541748703662716e-06,
      "loss": 0.1213,
      "step": 2827
    },
    {
      "epoch": 0.14603666408468888,
      "grad_norm": 6.376885414123535,
      "learning_rate": 9.54140597673813e-06,
      "loss": 0.0646,
      "step": 2828
    },
    {
      "epoch": 0.1460883036405887,
      "grad_norm": 7.362363815307617,
      "learning_rate": 9.541063127858281e-06,
      "loss": 0.116,
      "step": 2829
    },
    {
      "epoch": 0.14613994319648851,
      "grad_norm": 5.468184471130371,
      "learning_rate": 9.540720157032381e-06,
      "loss": 0.0665,
      "step": 2830
    },
    {
      "epoch": 0.14619158275238833,
      "grad_norm": 4.90239143371582,
      "learning_rate": 9.540377064269638e-06,
      "loss": 0.0828,
      "step": 2831
    },
    {
      "epoch": 0.14624322230828815,
      "grad_norm": 5.704601287841797,
      "learning_rate": 9.540033849579264e-06,
      "loss": 0.0681,
      "step": 2832
    },
    {
      "epoch": 0.14629486186418797,
      "grad_norm": 5.093058109283447,
      "learning_rate": 9.539690512970478e-06,
      "loss": 0.0796,
      "step": 2833
    },
    {
      "epoch": 0.1463465014200878,
      "grad_norm": 10.782691955566406,
      "learning_rate": 9.5393470544525e-06,
      "loss": 0.1255,
      "step": 2834
    },
    {
      "epoch": 0.1463981409759876,
      "grad_norm": 19.839176177978516,
      "learning_rate": 9.53900347403455e-06,
      "loss": 0.1936,
      "step": 2835
    },
    {
      "epoch": 0.14644978053188742,
      "grad_norm": 3.1376426219940186,
      "learning_rate": 9.538659771725857e-06,
      "loss": 0.062,
      "step": 2836
    },
    {
      "epoch": 0.14650142008778724,
      "grad_norm": 3.9937989711761475,
      "learning_rate": 9.538315947535652e-06,
      "loss": 0.0534,
      "step": 2837
    },
    {
      "epoch": 0.14655305964368706,
      "grad_norm": 4.839507579803467,
      "learning_rate": 9.537972001473166e-06,
      "loss": 0.0909,
      "step": 2838
    },
    {
      "epoch": 0.14660469919958688,
      "grad_norm": 5.820572376251221,
      "learning_rate": 9.537627933547636e-06,
      "loss": 0.099,
      "step": 2839
    },
    {
      "epoch": 0.1466563387554867,
      "grad_norm": 3.7324202060699463,
      "learning_rate": 9.537283743768302e-06,
      "loss": 0.0525,
      "step": 2840
    },
    {
      "epoch": 0.14670797831138652,
      "grad_norm": 3.568112373352051,
      "learning_rate": 9.536939432144406e-06,
      "loss": 0.0502,
      "step": 2841
    },
    {
      "epoch": 0.14675961786728633,
      "grad_norm": 17.634368896484375,
      "learning_rate": 9.536594998685194e-06,
      "loss": 0.0705,
      "step": 2842
    },
    {
      "epoch": 0.14681125742318615,
      "grad_norm": 8.774928092956543,
      "learning_rate": 9.536250443399917e-06,
      "loss": 0.101,
      "step": 2843
    },
    {
      "epoch": 0.14686289697908597,
      "grad_norm": 10.351092338562012,
      "learning_rate": 9.535905766297826e-06,
      "loss": 0.112,
      "step": 2844
    },
    {
      "epoch": 0.1469145365349858,
      "grad_norm": 10.659852981567383,
      "learning_rate": 9.53556096738818e-06,
      "loss": 0.099,
      "step": 2845
    },
    {
      "epoch": 0.1469661760908856,
      "grad_norm": 17.77236557006836,
      "learning_rate": 9.535216046680233e-06,
      "loss": 0.1585,
      "step": 2846
    },
    {
      "epoch": 0.14701781564678543,
      "grad_norm": 2.1235053539276123,
      "learning_rate": 9.534871004183253e-06,
      "loss": 0.0248,
      "step": 2847
    },
    {
      "epoch": 0.14706945520268525,
      "grad_norm": 3.9817001819610596,
      "learning_rate": 9.534525839906503e-06,
      "loss": 0.0555,
      "step": 2848
    },
    {
      "epoch": 0.14712109475858506,
      "grad_norm": 6.502513885498047,
      "learning_rate": 9.534180553859253e-06,
      "loss": 0.1214,
      "step": 2849
    },
    {
      "epoch": 0.14717273431448488,
      "grad_norm": 3.7722575664520264,
      "learning_rate": 9.533835146050773e-06,
      "loss": 0.0547,
      "step": 2850
    },
    {
      "epoch": 0.14722437387038473,
      "grad_norm": 4.775270938873291,
      "learning_rate": 9.533489616490341e-06,
      "loss": 0.102,
      "step": 2851
    },
    {
      "epoch": 0.14727601342628455,
      "grad_norm": 5.373650550842285,
      "learning_rate": 9.533143965187238e-06,
      "loss": 0.0641,
      "step": 2852
    },
    {
      "epoch": 0.14732765298218437,
      "grad_norm": 6.1205620765686035,
      "learning_rate": 9.532798192150741e-06,
      "loss": 0.1162,
      "step": 2853
    },
    {
      "epoch": 0.14737929253808418,
      "grad_norm": 10.936079978942871,
      "learning_rate": 9.532452297390138e-06,
      "loss": 0.043,
      "step": 2854
    },
    {
      "epoch": 0.147430932093984,
      "grad_norm": 5.020003795623779,
      "learning_rate": 9.532106280914717e-06,
      "loss": 0.0713,
      "step": 2855
    },
    {
      "epoch": 0.14748257164988382,
      "grad_norm": 10.784828186035156,
      "learning_rate": 9.53176014273377e-06,
      "loss": 0.0544,
      "step": 2856
    },
    {
      "epoch": 0.14753421120578364,
      "grad_norm": 5.386527061462402,
      "learning_rate": 9.531413882856595e-06,
      "loss": 0.0564,
      "step": 2857
    },
    {
      "epoch": 0.14758585076168346,
      "grad_norm": 3.641597032546997,
      "learning_rate": 9.531067501292488e-06,
      "loss": 0.0436,
      "step": 2858
    },
    {
      "epoch": 0.14763749031758328,
      "grad_norm": 3.484102725982666,
      "learning_rate": 9.53072099805075e-06,
      "loss": 0.0658,
      "step": 2859
    },
    {
      "epoch": 0.1476891298734831,
      "grad_norm": 4.189530849456787,
      "learning_rate": 9.530374373140687e-06,
      "loss": 0.0581,
      "step": 2860
    },
    {
      "epoch": 0.1477407694293829,
      "grad_norm": 7.0191121101379395,
      "learning_rate": 9.530027626571607e-06,
      "loss": 0.0603,
      "step": 2861
    },
    {
      "epoch": 0.14779240898528273,
      "grad_norm": 7.430602073669434,
      "learning_rate": 9.529680758352823e-06,
      "loss": 0.0878,
      "step": 2862
    },
    {
      "epoch": 0.14784404854118255,
      "grad_norm": 5.20871114730835,
      "learning_rate": 9.529333768493647e-06,
      "loss": 0.0906,
      "step": 2863
    },
    {
      "epoch": 0.14789568809708237,
      "grad_norm": 10.291753768920898,
      "learning_rate": 9.5289866570034e-06,
      "loss": 0.08,
      "step": 2864
    },
    {
      "epoch": 0.14794732765298219,
      "grad_norm": 3.5165905952453613,
      "learning_rate": 9.528639423891403e-06,
      "loss": 0.0777,
      "step": 2865
    },
    {
      "epoch": 0.147998967208882,
      "grad_norm": 3.918562412261963,
      "learning_rate": 9.528292069166977e-06,
      "loss": 0.0683,
      "step": 2866
    },
    {
      "epoch": 0.14805060676478182,
      "grad_norm": 23.403976440429688,
      "learning_rate": 9.527944592839455e-06,
      "loss": 0.109,
      "step": 2867
    },
    {
      "epoch": 0.14810224632068164,
      "grad_norm": 11.01201343536377,
      "learning_rate": 9.527596994918165e-06,
      "loss": 0.07,
      "step": 2868
    },
    {
      "epoch": 0.14815388587658146,
      "grad_norm": 2.8205385208129883,
      "learning_rate": 9.527249275412443e-06,
      "loss": 0.047,
      "step": 2869
    },
    {
      "epoch": 0.14820552543248128,
      "grad_norm": 25.34023094177246,
      "learning_rate": 9.526901434331625e-06,
      "loss": 0.1391,
      "step": 2870
    },
    {
      "epoch": 0.1482571649883811,
      "grad_norm": 5.962536334991455,
      "learning_rate": 9.526553471685053e-06,
      "loss": 0.0949,
      "step": 2871
    },
    {
      "epoch": 0.14830880454428091,
      "grad_norm": 11.867892265319824,
      "learning_rate": 9.526205387482072e-06,
      "loss": 0.1031,
      "step": 2872
    },
    {
      "epoch": 0.14836044410018073,
      "grad_norm": 1.335253357887268,
      "learning_rate": 9.525857181732027e-06,
      "loss": 0.0182,
      "step": 2873
    },
    {
      "epoch": 0.14841208365608055,
      "grad_norm": 4.954525470733643,
      "learning_rate": 9.525508854444271e-06,
      "loss": 0.113,
      "step": 2874
    },
    {
      "epoch": 0.14846372321198037,
      "grad_norm": 5.726619243621826,
      "learning_rate": 9.525160405628158e-06,
      "loss": 0.0586,
      "step": 2875
    },
    {
      "epoch": 0.1485153627678802,
      "grad_norm": 9.771459579467773,
      "learning_rate": 9.524811835293044e-06,
      "loss": 0.1023,
      "step": 2876
    },
    {
      "epoch": 0.14856700232378,
      "grad_norm": 7.436723709106445,
      "learning_rate": 9.524463143448291e-06,
      "loss": 0.1016,
      "step": 2877
    },
    {
      "epoch": 0.14861864187967982,
      "grad_norm": 3.8362934589385986,
      "learning_rate": 9.52411433010326e-06,
      "loss": 0.0593,
      "step": 2878
    },
    {
      "epoch": 0.14867028143557964,
      "grad_norm": 14.969642639160156,
      "learning_rate": 9.523765395267321e-06,
      "loss": 0.0737,
      "step": 2879
    },
    {
      "epoch": 0.14872192099147946,
      "grad_norm": 7.702666759490967,
      "learning_rate": 9.523416338949842e-06,
      "loss": 0.0776,
      "step": 2880
    },
    {
      "epoch": 0.14877356054737928,
      "grad_norm": 9.440598487854004,
      "learning_rate": 9.5230671611602e-06,
      "loss": 0.176,
      "step": 2881
    },
    {
      "epoch": 0.1488252001032791,
      "grad_norm": 6.37684440612793,
      "learning_rate": 9.522717861907768e-06,
      "loss": 0.0487,
      "step": 2882
    },
    {
      "epoch": 0.14887683965917894,
      "grad_norm": 11.36831283569336,
      "learning_rate": 9.52236844120193e-06,
      "loss": 0.168,
      "step": 2883
    },
    {
      "epoch": 0.14892847921507876,
      "grad_norm": 18.566843032836914,
      "learning_rate": 9.522018899052065e-06,
      "loss": 0.0812,
      "step": 2884
    },
    {
      "epoch": 0.14898011877097858,
      "grad_norm": 10.588960647583008,
      "learning_rate": 9.521669235467564e-06,
      "loss": 0.103,
      "step": 2885
    },
    {
      "epoch": 0.1490317583268784,
      "grad_norm": 15.05551528930664,
      "learning_rate": 9.521319450457812e-06,
      "loss": 0.0863,
      "step": 2886
    },
    {
      "epoch": 0.14908339788277822,
      "grad_norm": 9.148137092590332,
      "learning_rate": 9.520969544032206e-06,
      "loss": 0.0582,
      "step": 2887
    },
    {
      "epoch": 0.14913503743867804,
      "grad_norm": 7.32126522064209,
      "learning_rate": 9.520619516200142e-06,
      "loss": 0.1081,
      "step": 2888
    },
    {
      "epoch": 0.14918667699457785,
      "grad_norm": 3.8545079231262207,
      "learning_rate": 9.520269366971018e-06,
      "loss": 0.0733,
      "step": 2889
    },
    {
      "epoch": 0.14923831655047767,
      "grad_norm": 9.750493049621582,
      "learning_rate": 9.519919096354238e-06,
      "loss": 0.0736,
      "step": 2890
    },
    {
      "epoch": 0.1492899561063775,
      "grad_norm": 2.8459837436676025,
      "learning_rate": 9.51956870435921e-06,
      "loss": 0.0401,
      "step": 2891
    },
    {
      "epoch": 0.1493415956622773,
      "grad_norm": 4.189122200012207,
      "learning_rate": 9.51921819099534e-06,
      "loss": 0.039,
      "step": 2892
    },
    {
      "epoch": 0.14939323521817713,
      "grad_norm": 6.391958236694336,
      "learning_rate": 9.518867556272042e-06,
      "loss": 0.1054,
      "step": 2893
    },
    {
      "epoch": 0.14944487477407695,
      "grad_norm": 5.744683265686035,
      "learning_rate": 9.518516800198733e-06,
      "loss": 0.0788,
      "step": 2894
    },
    {
      "epoch": 0.14949651432997677,
      "grad_norm": 16.64226722717285,
      "learning_rate": 9.51816592278483e-06,
      "loss": 0.1278,
      "step": 2895
    },
    {
      "epoch": 0.14954815388587658,
      "grad_norm": 4.181710243225098,
      "learning_rate": 9.517814924039759e-06,
      "loss": 0.0707,
      "step": 2896
    },
    {
      "epoch": 0.1495997934417764,
      "grad_norm": 12.217262268066406,
      "learning_rate": 9.517463803972942e-06,
      "loss": 0.1283,
      "step": 2897
    },
    {
      "epoch": 0.14965143299767622,
      "grad_norm": 5.470862865447998,
      "learning_rate": 9.51711256259381e-06,
      "loss": 0.0806,
      "step": 2898
    },
    {
      "epoch": 0.14970307255357604,
      "grad_norm": 4.513925075531006,
      "learning_rate": 9.516761199911795e-06,
      "loss": 0.0576,
      "step": 2899
    },
    {
      "epoch": 0.14975471210947586,
      "grad_norm": 12.947700500488281,
      "learning_rate": 9.516409715936334e-06,
      "loss": 0.2033,
      "step": 2900
    },
    {
      "epoch": 0.14980635166537568,
      "grad_norm": 4.229698181152344,
      "learning_rate": 9.516058110676862e-06,
      "loss": 0.0675,
      "step": 2901
    },
    {
      "epoch": 0.1498579912212755,
      "grad_norm": 32.20352554321289,
      "learning_rate": 9.515706384142824e-06,
      "loss": 0.0738,
      "step": 2902
    },
    {
      "epoch": 0.1499096307771753,
      "grad_norm": 2.5054070949554443,
      "learning_rate": 9.515354536343666e-06,
      "loss": 0.0283,
      "step": 2903
    },
    {
      "epoch": 0.14996127033307513,
      "grad_norm": 7.098586082458496,
      "learning_rate": 9.515002567288834e-06,
      "loss": 0.0756,
      "step": 2904
    },
    {
      "epoch": 0.15001290988897495,
      "grad_norm": 7.216961860656738,
      "learning_rate": 9.514650476987782e-06,
      "loss": 0.0833,
      "step": 2905
    },
    {
      "epoch": 0.15006454944487477,
      "grad_norm": 8.491555213928223,
      "learning_rate": 9.514298265449962e-06,
      "loss": 0.0545,
      "step": 2906
    },
    {
      "epoch": 0.15011618900077459,
      "grad_norm": 12.74265193939209,
      "learning_rate": 9.513945932684834e-06,
      "loss": 0.1108,
      "step": 2907
    },
    {
      "epoch": 0.1501678285566744,
      "grad_norm": 9.53114128112793,
      "learning_rate": 9.513593478701862e-06,
      "loss": 0.1115,
      "step": 2908
    },
    {
      "epoch": 0.15021946811257422,
      "grad_norm": 3.0505995750427246,
      "learning_rate": 9.513240903510507e-06,
      "loss": 0.018,
      "step": 2909
    },
    {
      "epoch": 0.15027110766847404,
      "grad_norm": 5.6447649002075195,
      "learning_rate": 9.51288820712024e-06,
      "loss": 0.0581,
      "step": 2910
    },
    {
      "epoch": 0.15032274722437386,
      "grad_norm": 4.515944957733154,
      "learning_rate": 9.512535389540532e-06,
      "loss": 0.0597,
      "step": 2911
    },
    {
      "epoch": 0.15037438678027368,
      "grad_norm": 6.516438007354736,
      "learning_rate": 9.512182450780856e-06,
      "loss": 0.0999,
      "step": 2912
    },
    {
      "epoch": 0.1504260263361735,
      "grad_norm": 5.513379096984863,
      "learning_rate": 9.511829390850689e-06,
      "loss": 0.0328,
      "step": 2913
    },
    {
      "epoch": 0.15047766589207331,
      "grad_norm": 25.414867401123047,
      "learning_rate": 9.511476209759515e-06,
      "loss": 0.1968,
      "step": 2914
    },
    {
      "epoch": 0.15052930544797316,
      "grad_norm": 5.4566826820373535,
      "learning_rate": 9.511122907516817e-06,
      "loss": 0.0951,
      "step": 2915
    },
    {
      "epoch": 0.15058094500387298,
      "grad_norm": 11.338152885437012,
      "learning_rate": 9.510769484132082e-06,
      "loss": 0.0934,
      "step": 2916
    },
    {
      "epoch": 0.1506325845597728,
      "grad_norm": 5.18950891494751,
      "learning_rate": 9.510415939614803e-06,
      "loss": 0.0855,
      "step": 2917
    },
    {
      "epoch": 0.15068422411567262,
      "grad_norm": 11.25289249420166,
      "learning_rate": 9.51006227397447e-06,
      "loss": 0.1463,
      "step": 2918
    },
    {
      "epoch": 0.15073586367157243,
      "grad_norm": 31.70257568359375,
      "learning_rate": 9.509708487220585e-06,
      "loss": 0.1752,
      "step": 2919
    },
    {
      "epoch": 0.15078750322747225,
      "grad_norm": 7.127709865570068,
      "learning_rate": 9.509354579362647e-06,
      "loss": 0.1324,
      "step": 2920
    },
    {
      "epoch": 0.15083914278337207,
      "grad_norm": 5.232102870941162,
      "learning_rate": 9.509000550410158e-06,
      "loss": 0.1187,
      "step": 2921
    },
    {
      "epoch": 0.1508907823392719,
      "grad_norm": 12.175374984741211,
      "learning_rate": 9.508646400372625e-06,
      "loss": 0.1491,
      "step": 2922
    },
    {
      "epoch": 0.1509424218951717,
      "grad_norm": 3.623230457305908,
      "learning_rate": 9.508292129259562e-06,
      "loss": 0.0643,
      "step": 2923
    },
    {
      "epoch": 0.15099406145107153,
      "grad_norm": 10.91407299041748,
      "learning_rate": 9.507937737080482e-06,
      "loss": 0.1093,
      "step": 2924
    },
    {
      "epoch": 0.15104570100697134,
      "grad_norm": 15.316548347473145,
      "learning_rate": 9.507583223844898e-06,
      "loss": 0.1486,
      "step": 2925
    },
    {
      "epoch": 0.15109734056287116,
      "grad_norm": 3.3910341262817383,
      "learning_rate": 9.507228589562334e-06,
      "loss": 0.0746,
      "step": 2926
    },
    {
      "epoch": 0.15114898011877098,
      "grad_norm": 27.1737117767334,
      "learning_rate": 9.506873834242311e-06,
      "loss": 0.0686,
      "step": 2927
    },
    {
      "epoch": 0.1512006196746708,
      "grad_norm": 5.5698161125183105,
      "learning_rate": 9.506518957894356e-06,
      "loss": 0.0696,
      "step": 2928
    },
    {
      "epoch": 0.15125225923057062,
      "grad_norm": 15.69558048248291,
      "learning_rate": 9.506163960528e-06,
      "loss": 0.1418,
      "step": 2929
    },
    {
      "epoch": 0.15130389878647044,
      "grad_norm": 5.113656997680664,
      "learning_rate": 9.505808842152775e-06,
      "loss": 0.1016,
      "step": 2930
    },
    {
      "epoch": 0.15135553834237025,
      "grad_norm": 4.767978668212891,
      "learning_rate": 9.50545360277822e-06,
      "loss": 0.0877,
      "step": 2931
    },
    {
      "epoch": 0.15140717789827007,
      "grad_norm": 11.210095405578613,
      "learning_rate": 9.50509824241387e-06,
      "loss": 0.1143,
      "step": 2932
    },
    {
      "epoch": 0.1514588174541699,
      "grad_norm": 10.53370189666748,
      "learning_rate": 9.504742761069271e-06,
      "loss": 0.0841,
      "step": 2933
    },
    {
      "epoch": 0.1515104570100697,
      "grad_norm": 3.722010850906372,
      "learning_rate": 9.504387158753972e-06,
      "loss": 0.062,
      "step": 2934
    },
    {
      "epoch": 0.15156209656596953,
      "grad_norm": 7.923773765563965,
      "learning_rate": 9.504031435477515e-06,
      "loss": 0.0744,
      "step": 2935
    },
    {
      "epoch": 0.15161373612186935,
      "grad_norm": 5.753010272979736,
      "learning_rate": 9.503675591249458e-06,
      "loss": 0.1184,
      "step": 2936
    },
    {
      "epoch": 0.15166537567776917,
      "grad_norm": 13.741536140441895,
      "learning_rate": 9.503319626079354e-06,
      "loss": 0.1504,
      "step": 2937
    },
    {
      "epoch": 0.15171701523366898,
      "grad_norm": 8.027464866638184,
      "learning_rate": 9.502963539976766e-06,
      "loss": 0.09,
      "step": 2938
    },
    {
      "epoch": 0.1517686547895688,
      "grad_norm": 13.223830223083496,
      "learning_rate": 9.502607332951252e-06,
      "loss": 0.1001,
      "step": 2939
    },
    {
      "epoch": 0.15182029434546862,
      "grad_norm": 8.438231468200684,
      "learning_rate": 9.50225100501238e-06,
      "loss": 0.077,
      "step": 2940
    },
    {
      "epoch": 0.15187193390136844,
      "grad_norm": 6.6798224449157715,
      "learning_rate": 9.501894556169719e-06,
      "loss": 0.0872,
      "step": 2941
    },
    {
      "epoch": 0.15192357345726826,
      "grad_norm": 7.259324073791504,
      "learning_rate": 9.501537986432839e-06,
      "loss": 0.0429,
      "step": 2942
    },
    {
      "epoch": 0.15197521301316808,
      "grad_norm": 5.617965221405029,
      "learning_rate": 9.501181295811318e-06,
      "loss": 0.0749,
      "step": 2943
    },
    {
      "epoch": 0.1520268525690679,
      "grad_norm": 2.678666353225708,
      "learning_rate": 9.500824484314733e-06,
      "loss": 0.0694,
      "step": 2944
    },
    {
      "epoch": 0.1520784921249677,
      "grad_norm": 7.851320266723633,
      "learning_rate": 9.500467551952667e-06,
      "loss": 0.1036,
      "step": 2945
    },
    {
      "epoch": 0.15213013168086753,
      "grad_norm": 4.179657936096191,
      "learning_rate": 9.500110498734703e-06,
      "loss": 0.0443,
      "step": 2946
    },
    {
      "epoch": 0.15218177123676738,
      "grad_norm": 5.004298686981201,
      "learning_rate": 9.49975332467043e-06,
      "loss": 0.0457,
      "step": 2947
    },
    {
      "epoch": 0.1522334107926672,
      "grad_norm": 2.961735248565674,
      "learning_rate": 9.499396029769444e-06,
      "loss": 0.0666,
      "step": 2948
    },
    {
      "epoch": 0.152285050348567,
      "grad_norm": 4.927459239959717,
      "learning_rate": 9.499038614041333e-06,
      "loss": 0.0757,
      "step": 2949
    },
    {
      "epoch": 0.15233668990446683,
      "grad_norm": 3.1067988872528076,
      "learning_rate": 9.498681077495699e-06,
      "loss": 0.0441,
      "step": 2950
    },
    {
      "epoch": 0.15238832946036665,
      "grad_norm": 5.797266006469727,
      "learning_rate": 9.498323420142141e-06,
      "loss": 0.0633,
      "step": 2951
    },
    {
      "epoch": 0.15243996901626647,
      "grad_norm": 9.398457527160645,
      "learning_rate": 9.497965641990268e-06,
      "loss": 0.0871,
      "step": 2952
    },
    {
      "epoch": 0.1524916085721663,
      "grad_norm": 51.187477111816406,
      "learning_rate": 9.497607743049681e-06,
      "loss": 0.06,
      "step": 2953
    },
    {
      "epoch": 0.1525432481280661,
      "grad_norm": 6.9555816650390625,
      "learning_rate": 9.497249723329998e-06,
      "loss": 0.0738,
      "step": 2954
    },
    {
      "epoch": 0.15259488768396592,
      "grad_norm": 5.354311466217041,
      "learning_rate": 9.496891582840829e-06,
      "loss": 0.0991,
      "step": 2955
    },
    {
      "epoch": 0.15264652723986574,
      "grad_norm": 4.775583744049072,
      "learning_rate": 9.496533321591793e-06,
      "loss": 0.0831,
      "step": 2956
    },
    {
      "epoch": 0.15269816679576556,
      "grad_norm": 3.501828908920288,
      "learning_rate": 9.496174939592511e-06,
      "loss": 0.0918,
      "step": 2957
    },
    {
      "epoch": 0.15274980635166538,
      "grad_norm": 6.158728122711182,
      "learning_rate": 9.495816436852604e-06,
      "loss": 0.1118,
      "step": 2958
    },
    {
      "epoch": 0.1528014459075652,
      "grad_norm": 3.1483259201049805,
      "learning_rate": 9.495457813381705e-06,
      "loss": 0.0309,
      "step": 2959
    },
    {
      "epoch": 0.15285308546346502,
      "grad_norm": 2.598698616027832,
      "learning_rate": 9.49509906918944e-06,
      "loss": 0.0586,
      "step": 2960
    },
    {
      "epoch": 0.15290472501936483,
      "grad_norm": 6.458876132965088,
      "learning_rate": 9.494740204285443e-06,
      "loss": 0.0973,
      "step": 2961
    },
    {
      "epoch": 0.15295636457526465,
      "grad_norm": 9.1163911819458,
      "learning_rate": 9.494381218679353e-06,
      "loss": 0.0646,
      "step": 2962
    },
    {
      "epoch": 0.15300800413116447,
      "grad_norm": 3.5057809352874756,
      "learning_rate": 9.494022112380809e-06,
      "loss": 0.0623,
      "step": 2963
    },
    {
      "epoch": 0.1530596436870643,
      "grad_norm": 2.978951930999756,
      "learning_rate": 9.493662885399453e-06,
      "loss": 0.0558,
      "step": 2964
    },
    {
      "epoch": 0.1531112832429641,
      "grad_norm": 9.007461547851562,
      "learning_rate": 9.493303537744936e-06,
      "loss": 0.1365,
      "step": 2965
    },
    {
      "epoch": 0.15316292279886393,
      "grad_norm": 4.506552696228027,
      "learning_rate": 9.492944069426904e-06,
      "loss": 0.0397,
      "step": 2966
    },
    {
      "epoch": 0.15321456235476374,
      "grad_norm": 5.649576663970947,
      "learning_rate": 9.492584480455012e-06,
      "loss": 0.0699,
      "step": 2967
    },
    {
      "epoch": 0.15326620191066356,
      "grad_norm": 15.133033752441406,
      "learning_rate": 9.492224770838916e-06,
      "loss": 0.0531,
      "step": 2968
    },
    {
      "epoch": 0.15331784146656338,
      "grad_norm": 10.573235511779785,
      "learning_rate": 9.491864940588275e-06,
      "loss": 0.1286,
      "step": 2969
    },
    {
      "epoch": 0.1533694810224632,
      "grad_norm": 8.70798397064209,
      "learning_rate": 9.491504989712752e-06,
      "loss": 0.1008,
      "step": 2970
    },
    {
      "epoch": 0.15342112057836302,
      "grad_norm": 8.43976879119873,
      "learning_rate": 9.491144918222014e-06,
      "loss": 0.0764,
      "step": 2971
    },
    {
      "epoch": 0.15347276013426284,
      "grad_norm": 4.494546890258789,
      "learning_rate": 9.490784726125732e-06,
      "loss": 0.1082,
      "step": 2972
    },
    {
      "epoch": 0.15352439969016266,
      "grad_norm": 7.129053592681885,
      "learning_rate": 9.490424413433575e-06,
      "loss": 0.0817,
      "step": 2973
    },
    {
      "epoch": 0.15357603924606247,
      "grad_norm": 4.224681854248047,
      "learning_rate": 9.49006398015522e-06,
      "loss": 0.0462,
      "step": 2974
    },
    {
      "epoch": 0.1536276788019623,
      "grad_norm": 3.3131799697875977,
      "learning_rate": 9.489703426300347e-06,
      "loss": 0.0325,
      "step": 2975
    },
    {
      "epoch": 0.1536793183578621,
      "grad_norm": 23.086641311645508,
      "learning_rate": 9.489342751878639e-06,
      "loss": 0.1081,
      "step": 2976
    },
    {
      "epoch": 0.15373095791376193,
      "grad_norm": 3.594961166381836,
      "learning_rate": 9.48898195689978e-06,
      "loss": 0.0526,
      "step": 2977
    },
    {
      "epoch": 0.15378259746966175,
      "grad_norm": 33.348655700683594,
      "learning_rate": 9.48862104137346e-06,
      "loss": 0.185,
      "step": 2978
    },
    {
      "epoch": 0.1538342370255616,
      "grad_norm": 11.247432708740234,
      "learning_rate": 9.48826000530937e-06,
      "loss": 0.1032,
      "step": 2979
    },
    {
      "epoch": 0.1538858765814614,
      "grad_norm": 1.7697830200195312,
      "learning_rate": 9.487898848717205e-06,
      "loss": 0.041,
      "step": 2980
    },
    {
      "epoch": 0.15393751613736123,
      "grad_norm": 4.513772964477539,
      "learning_rate": 9.487537571606664e-06,
      "loss": 0.0662,
      "step": 2981
    },
    {
      "epoch": 0.15398915569326105,
      "grad_norm": 7.9495320320129395,
      "learning_rate": 9.487176173987451e-06,
      "loss": 0.0828,
      "step": 2982
    },
    {
      "epoch": 0.15404079524916087,
      "grad_norm": 4.433658599853516,
      "learning_rate": 9.486814655869267e-06,
      "loss": 0.0784,
      "step": 2983
    },
    {
      "epoch": 0.15409243480506069,
      "grad_norm": 9.723039627075195,
      "learning_rate": 9.486453017261823e-06,
      "loss": 0.0983,
      "step": 2984
    },
    {
      "epoch": 0.1541440743609605,
      "grad_norm": 5.703024864196777,
      "learning_rate": 9.48609125817483e-06,
      "loss": 0.0878,
      "step": 2985
    },
    {
      "epoch": 0.15419571391686032,
      "grad_norm": 17.01174545288086,
      "learning_rate": 9.485729378618e-06,
      "loss": 0.1154,
      "step": 2986
    },
    {
      "epoch": 0.15424735347276014,
      "grad_norm": 3.8854730129241943,
      "learning_rate": 9.485367378601056e-06,
      "loss": 0.0296,
      "step": 2987
    },
    {
      "epoch": 0.15429899302865996,
      "grad_norm": 14.12228012084961,
      "learning_rate": 9.485005258133718e-06,
      "loss": 0.3235,
      "step": 2988
    },
    {
      "epoch": 0.15435063258455978,
      "grad_norm": 3.7578883171081543,
      "learning_rate": 9.484643017225706e-06,
      "loss": 0.0365,
      "step": 2989
    },
    {
      "epoch": 0.1544022721404596,
      "grad_norm": 9.72506332397461,
      "learning_rate": 9.484280655886751e-06,
      "loss": 0.0845,
      "step": 2990
    },
    {
      "epoch": 0.1544539116963594,
      "grad_norm": 8.696615219116211,
      "learning_rate": 9.483918174126586e-06,
      "loss": 0.0909,
      "step": 2991
    },
    {
      "epoch": 0.15450555125225923,
      "grad_norm": 3.1880526542663574,
      "learning_rate": 9.483555571954939e-06,
      "loss": 0.0484,
      "step": 2992
    },
    {
      "epoch": 0.15455719080815905,
      "grad_norm": 9.610145568847656,
      "learning_rate": 9.483192849381554e-06,
      "loss": 0.2222,
      "step": 2993
    },
    {
      "epoch": 0.15460883036405887,
      "grad_norm": 15.559586524963379,
      "learning_rate": 9.482830006416167e-06,
      "loss": 0.0997,
      "step": 2994
    },
    {
      "epoch": 0.1546604699199587,
      "grad_norm": 6.325245380401611,
      "learning_rate": 9.482467043068521e-06,
      "loss": 0.0675,
      "step": 2995
    },
    {
      "epoch": 0.1547121094758585,
      "grad_norm": 8.314116477966309,
      "learning_rate": 9.482103959348368e-06,
      "loss": 0.1806,
      "step": 2996
    },
    {
      "epoch": 0.15476374903175832,
      "grad_norm": 4.389742851257324,
      "learning_rate": 9.481740755265456e-06,
      "loss": 0.1171,
      "step": 2997
    },
    {
      "epoch": 0.15481538858765814,
      "grad_norm": 3.7257165908813477,
      "learning_rate": 9.481377430829537e-06,
      "loss": 0.0638,
      "step": 2998
    },
    {
      "epoch": 0.15486702814355796,
      "grad_norm": 2.710763692855835,
      "learning_rate": 9.48101398605037e-06,
      "loss": 0.0671,
      "step": 2999
    },
    {
      "epoch": 0.15491866769945778,
      "grad_norm": 4.5937042236328125,
      "learning_rate": 9.480650420937714e-06,
      "loss": 0.0917,
      "step": 3000
    },
    {
      "epoch": 0.1549703072553576,
      "grad_norm": 2.7518441677093506,
      "learning_rate": 9.48028673550133e-06,
      "loss": 0.0575,
      "step": 3001
    },
    {
      "epoch": 0.15502194681125742,
      "grad_norm": 11.829773902893066,
      "learning_rate": 9.479922929750989e-06,
      "loss": 0.1309,
      "step": 3002
    },
    {
      "epoch": 0.15507358636715723,
      "grad_norm": 14.313937187194824,
      "learning_rate": 9.479559003696456e-06,
      "loss": 0.1429,
      "step": 3003
    },
    {
      "epoch": 0.15512522592305705,
      "grad_norm": 2.728914737701416,
      "learning_rate": 9.479194957347509e-06,
      "loss": 0.0896,
      "step": 3004
    },
    {
      "epoch": 0.15517686547895687,
      "grad_norm": 4.974431037902832,
      "learning_rate": 9.478830790713918e-06,
      "loss": 0.0561,
      "step": 3005
    },
    {
      "epoch": 0.1552285050348567,
      "grad_norm": 7.751481056213379,
      "learning_rate": 9.478466503805467e-06,
      "loss": 0.1108,
      "step": 3006
    },
    {
      "epoch": 0.1552801445907565,
      "grad_norm": 3.6141421794891357,
      "learning_rate": 9.478102096631936e-06,
      "loss": 0.045,
      "step": 3007
    },
    {
      "epoch": 0.15533178414665633,
      "grad_norm": 9.769404411315918,
      "learning_rate": 9.477737569203114e-06,
      "loss": 0.1233,
      "step": 3008
    },
    {
      "epoch": 0.15538342370255614,
      "grad_norm": 9.318175315856934,
      "learning_rate": 9.477372921528787e-06,
      "loss": 0.128,
      "step": 3009
    },
    {
      "epoch": 0.155435063258456,
      "grad_norm": 8.375364303588867,
      "learning_rate": 9.477008153618749e-06,
      "loss": 0.0658,
      "step": 3010
    },
    {
      "epoch": 0.1554867028143558,
      "grad_norm": 7.628714084625244,
      "learning_rate": 9.476643265482795e-06,
      "loss": 0.1032,
      "step": 3011
    },
    {
      "epoch": 0.15553834237025563,
      "grad_norm": 6.719176769256592,
      "learning_rate": 9.476278257130721e-06,
      "loss": 0.0475,
      "step": 3012
    },
    {
      "epoch": 0.15558998192615545,
      "grad_norm": 3.9425811767578125,
      "learning_rate": 9.475913128572335e-06,
      "loss": 0.0574,
      "step": 3013
    },
    {
      "epoch": 0.15564162148205526,
      "grad_norm": 3.964029550552368,
      "learning_rate": 9.475547879817435e-06,
      "loss": 0.0411,
      "step": 3014
    },
    {
      "epoch": 0.15569326103795508,
      "grad_norm": 11.245575904846191,
      "learning_rate": 9.475182510875836e-06,
      "loss": 0.1074,
      "step": 3015
    },
    {
      "epoch": 0.1557449005938549,
      "grad_norm": 14.57929801940918,
      "learning_rate": 9.474817021757345e-06,
      "loss": 0.1134,
      "step": 3016
    },
    {
      "epoch": 0.15579654014975472,
      "grad_norm": 4.006103992462158,
      "learning_rate": 9.474451412471779e-06,
      "loss": 0.0449,
      "step": 3017
    },
    {
      "epoch": 0.15584817970565454,
      "grad_norm": 3.1006438732147217,
      "learning_rate": 9.474085683028958e-06,
      "loss": 0.0578,
      "step": 3018
    },
    {
      "epoch": 0.15589981926155436,
      "grad_norm": 9.907613754272461,
      "learning_rate": 9.473719833438698e-06,
      "loss": 0.0898,
      "step": 3019
    },
    {
      "epoch": 0.15595145881745417,
      "grad_norm": 3.7393033504486084,
      "learning_rate": 9.473353863710827e-06,
      "loss": 0.0409,
      "step": 3020
    },
    {
      "epoch": 0.156003098373354,
      "grad_norm": 11.656477928161621,
      "learning_rate": 9.472987773855174e-06,
      "loss": 0.0782,
      "step": 3021
    },
    {
      "epoch": 0.1560547379292538,
      "grad_norm": 8.14793586730957,
      "learning_rate": 9.472621563881568e-06,
      "loss": 0.0579,
      "step": 3022
    },
    {
      "epoch": 0.15610637748515363,
      "grad_norm": 13.548062324523926,
      "learning_rate": 9.472255233799843e-06,
      "loss": 0.1577,
      "step": 3023
    },
    {
      "epoch": 0.15615801704105345,
      "grad_norm": 7.3303751945495605,
      "learning_rate": 9.471888783619836e-06,
      "loss": 0.1047,
      "step": 3024
    },
    {
      "epoch": 0.15620965659695327,
      "grad_norm": 6.1075944900512695,
      "learning_rate": 9.47152221335139e-06,
      "loss": 0.0759,
      "step": 3025
    },
    {
      "epoch": 0.15626129615285309,
      "grad_norm": 15.993444442749023,
      "learning_rate": 9.471155523004348e-06,
      "loss": 0.1422,
      "step": 3026
    },
    {
      "epoch": 0.1563129357087529,
      "grad_norm": 5.492521286010742,
      "learning_rate": 9.470788712588557e-06,
      "loss": 0.0995,
      "step": 3027
    },
    {
      "epoch": 0.15636457526465272,
      "grad_norm": 7.080912113189697,
      "learning_rate": 9.470421782113866e-06,
      "loss": 0.0725,
      "step": 3028
    },
    {
      "epoch": 0.15641621482055254,
      "grad_norm": 12.287102699279785,
      "learning_rate": 9.47005473159013e-06,
      "loss": 0.163,
      "step": 3029
    },
    {
      "epoch": 0.15646785437645236,
      "grad_norm": 13.575054168701172,
      "learning_rate": 9.469687561027205e-06,
      "loss": 0.1174,
      "step": 3030
    },
    {
      "epoch": 0.15651949393235218,
      "grad_norm": 6.344703197479248,
      "learning_rate": 9.469320270434952e-06,
      "loss": 0.0925,
      "step": 3031
    },
    {
      "epoch": 0.156571133488252,
      "grad_norm": 5.593818187713623,
      "learning_rate": 9.468952859823235e-06,
      "loss": 0.094,
      "step": 3032
    },
    {
      "epoch": 0.15662277304415181,
      "grad_norm": 2.008765935897827,
      "learning_rate": 9.468585329201918e-06,
      "loss": 0.0445,
      "step": 3033
    },
    {
      "epoch": 0.15667441260005163,
      "grad_norm": 13.163809776306152,
      "learning_rate": 9.468217678580873e-06,
      "loss": 0.146,
      "step": 3034
    },
    {
      "epoch": 0.15672605215595145,
      "grad_norm": 2.8898391723632812,
      "learning_rate": 9.46784990796997e-06,
      "loss": 0.0514,
      "step": 3035
    },
    {
      "epoch": 0.15677769171185127,
      "grad_norm": 4.0135416984558105,
      "learning_rate": 9.467482017379088e-06,
      "loss": 0.0632,
      "step": 3036
    },
    {
      "epoch": 0.1568293312677511,
      "grad_norm": 2.3029448986053467,
      "learning_rate": 9.467114006818107e-06,
      "loss": 0.0777,
      "step": 3037
    },
    {
      "epoch": 0.1568809708236509,
      "grad_norm": 3.4010777473449707,
      "learning_rate": 9.466745876296906e-06,
      "loss": 0.0525,
      "step": 3038
    },
    {
      "epoch": 0.15693261037955072,
      "grad_norm": 4.335797309875488,
      "learning_rate": 9.466377625825374e-06,
      "loss": 0.0508,
      "step": 3039
    },
    {
      "epoch": 0.15698424993545054,
      "grad_norm": 12.536372184753418,
      "learning_rate": 9.466009255413397e-06,
      "loss": 0.1211,
      "step": 3040
    },
    {
      "epoch": 0.15703588949135036,
      "grad_norm": 3.6251931190490723,
      "learning_rate": 9.465640765070872e-06,
      "loss": 0.0825,
      "step": 3041
    },
    {
      "epoch": 0.1570875290472502,
      "grad_norm": 3.5328269004821777,
      "learning_rate": 9.46527215480769e-06,
      "loss": 0.0574,
      "step": 3042
    },
    {
      "epoch": 0.15713916860315003,
      "grad_norm": 20.51573944091797,
      "learning_rate": 9.464903424633749e-06,
      "loss": 0.2276,
      "step": 3043
    },
    {
      "epoch": 0.15719080815904984,
      "grad_norm": 10.49366569519043,
      "learning_rate": 9.464534574558954e-06,
      "loss": 0.0744,
      "step": 3044
    },
    {
      "epoch": 0.15724244771494966,
      "grad_norm": 2.005645751953125,
      "learning_rate": 9.464165604593211e-06,
      "loss": 0.0415,
      "step": 3045
    },
    {
      "epoch": 0.15729408727084948,
      "grad_norm": 4.398406505584717,
      "learning_rate": 9.463796514746426e-06,
      "loss": 0.0718,
      "step": 3046
    },
    {
      "epoch": 0.1573457268267493,
      "grad_norm": 4.564924240112305,
      "learning_rate": 9.463427305028511e-06,
      "loss": 0.0481,
      "step": 3047
    },
    {
      "epoch": 0.15739736638264912,
      "grad_norm": 6.693701267242432,
      "learning_rate": 9.463057975449381e-06,
      "loss": 0.0314,
      "step": 3048
    },
    {
      "epoch": 0.15744900593854894,
      "grad_norm": 3.730207681655884,
      "learning_rate": 9.462688526018955e-06,
      "loss": 0.0517,
      "step": 3049
    },
    {
      "epoch": 0.15750064549444875,
      "grad_norm": 5.738395690917969,
      "learning_rate": 9.462318956747151e-06,
      "loss": 0.0818,
      "step": 3050
    },
    {
      "epoch": 0.15755228505034857,
      "grad_norm": 5.634483337402344,
      "learning_rate": 9.461949267643898e-06,
      "loss": 0.0781,
      "step": 3051
    },
    {
      "epoch": 0.1576039246062484,
      "grad_norm": 1.3573781251907349,
      "learning_rate": 9.461579458719119e-06,
      "loss": 0.029,
      "step": 3052
    },
    {
      "epoch": 0.1576555641621482,
      "grad_norm": 4.1521148681640625,
      "learning_rate": 9.461209529982748e-06,
      "loss": 0.0474,
      "step": 3053
    },
    {
      "epoch": 0.15770720371804803,
      "grad_norm": 4.405172348022461,
      "learning_rate": 9.460839481444717e-06,
      "loss": 0.0855,
      "step": 3054
    },
    {
      "epoch": 0.15775884327394785,
      "grad_norm": 9.205048561096191,
      "learning_rate": 9.460469313114966e-06,
      "loss": 0.1191,
      "step": 3055
    },
    {
      "epoch": 0.15781048282984766,
      "grad_norm": 12.805017471313477,
      "learning_rate": 9.460099025003435e-06,
      "loss": 0.0786,
      "step": 3056
    },
    {
      "epoch": 0.15786212238574748,
      "grad_norm": 9.736449241638184,
      "learning_rate": 9.459728617120065e-06,
      "loss": 0.1072,
      "step": 3057
    },
    {
      "epoch": 0.1579137619416473,
      "grad_norm": 4.938130855560303,
      "learning_rate": 9.459358089474806e-06,
      "loss": 0.1264,
      "step": 3058
    },
    {
      "epoch": 0.15796540149754712,
      "grad_norm": 16.86064338684082,
      "learning_rate": 9.458987442077606e-06,
      "loss": 0.1269,
      "step": 3059
    },
    {
      "epoch": 0.15801704105344694,
      "grad_norm": 4.568706035614014,
      "learning_rate": 9.45861667493842e-06,
      "loss": 0.0498,
      "step": 3060
    },
    {
      "epoch": 0.15806868060934676,
      "grad_norm": 6.73158073425293,
      "learning_rate": 9.458245788067204e-06,
      "loss": 0.0726,
      "step": 3061
    },
    {
      "epoch": 0.15812032016524658,
      "grad_norm": 18.25006675720215,
      "learning_rate": 9.457874781473917e-06,
      "loss": 0.1213,
      "step": 3062
    },
    {
      "epoch": 0.1581719597211464,
      "grad_norm": 5.325740337371826,
      "learning_rate": 9.457503655168525e-06,
      "loss": 0.1066,
      "step": 3063
    },
    {
      "epoch": 0.1582235992770462,
      "grad_norm": 3.2614312171936035,
      "learning_rate": 9.45713240916099e-06,
      "loss": 0.0724,
      "step": 3064
    },
    {
      "epoch": 0.15827523883294603,
      "grad_norm": 4.344670295715332,
      "learning_rate": 9.456761043461283e-06,
      "loss": 0.1277,
      "step": 3065
    },
    {
      "epoch": 0.15832687838884585,
      "grad_norm": 18.15375328063965,
      "learning_rate": 9.456389558079378e-06,
      "loss": 0.1714,
      "step": 3066
    },
    {
      "epoch": 0.15837851794474567,
      "grad_norm": 8.526477813720703,
      "learning_rate": 9.456017953025252e-06,
      "loss": 0.0971,
      "step": 3067
    },
    {
      "epoch": 0.15843015750064549,
      "grad_norm": 2.972050666809082,
      "learning_rate": 9.455646228308878e-06,
      "loss": 0.0746,
      "step": 3068
    },
    {
      "epoch": 0.1584817970565453,
      "grad_norm": 2.614562749862671,
      "learning_rate": 9.455274383940248e-06,
      "loss": 0.0622,
      "step": 3069
    },
    {
      "epoch": 0.15853343661244512,
      "grad_norm": 3.5333995819091797,
      "learning_rate": 9.454902419929339e-06,
      "loss": 0.0883,
      "step": 3070
    },
    {
      "epoch": 0.15858507616834494,
      "grad_norm": 6.655433177947998,
      "learning_rate": 9.454530336286144e-06,
      "loss": 0.0808,
      "step": 3071
    },
    {
      "epoch": 0.15863671572424476,
      "grad_norm": 4.4809770584106445,
      "learning_rate": 9.454158133020651e-06,
      "loss": 0.0821,
      "step": 3072
    },
    {
      "epoch": 0.15868835528014458,
      "grad_norm": 2.7372069358825684,
      "learning_rate": 9.45378581014286e-06,
      "loss": 0.0471,
      "step": 3073
    },
    {
      "epoch": 0.15873999483604442,
      "grad_norm": 11.677654266357422,
      "learning_rate": 9.453413367662768e-06,
      "loss": 0.2772,
      "step": 3074
    },
    {
      "epoch": 0.15879163439194424,
      "grad_norm": 15.276185989379883,
      "learning_rate": 9.453040805590375e-06,
      "loss": 0.2798,
      "step": 3075
    },
    {
      "epoch": 0.15884327394784406,
      "grad_norm": 4.857885837554932,
      "learning_rate": 9.452668123935687e-06,
      "loss": 0.0922,
      "step": 3076
    },
    {
      "epoch": 0.15889491350374388,
      "grad_norm": 2.91960072517395,
      "learning_rate": 9.452295322708711e-06,
      "loss": 0.0842,
      "step": 3077
    },
    {
      "epoch": 0.1589465530596437,
      "grad_norm": 12.193114280700684,
      "learning_rate": 9.45192240191946e-06,
      "loss": 0.0846,
      "step": 3078
    },
    {
      "epoch": 0.15899819261554352,
      "grad_norm": 5.485538959503174,
      "learning_rate": 9.451549361577948e-06,
      "loss": 0.0815,
      "step": 3079
    },
    {
      "epoch": 0.15904983217144333,
      "grad_norm": 3.5766873359680176,
      "learning_rate": 9.45117620169419e-06,
      "loss": 0.0791,
      "step": 3080
    },
    {
      "epoch": 0.15910147172734315,
      "grad_norm": 10.137812614440918,
      "learning_rate": 9.450802922278211e-06,
      "loss": 0.16,
      "step": 3081
    },
    {
      "epoch": 0.15915311128324297,
      "grad_norm": 4.140166759490967,
      "learning_rate": 9.450429523340033e-06,
      "loss": 0.1391,
      "step": 3082
    },
    {
      "epoch": 0.1592047508391428,
      "grad_norm": 10.243961334228516,
      "learning_rate": 9.450056004889683e-06,
      "loss": 0.1425,
      "step": 3083
    },
    {
      "epoch": 0.1592563903950426,
      "grad_norm": 8.074336051940918,
      "learning_rate": 9.449682366937191e-06,
      "loss": 0.1102,
      "step": 3084
    },
    {
      "epoch": 0.15930802995094243,
      "grad_norm": 9.488996505737305,
      "learning_rate": 9.449308609492594e-06,
      "loss": 0.0992,
      "step": 3085
    },
    {
      "epoch": 0.15935966950684224,
      "grad_norm": 8.816608428955078,
      "learning_rate": 9.448934732565924e-06,
      "loss": 0.055,
      "step": 3086
    },
    {
      "epoch": 0.15941130906274206,
      "grad_norm": 2.4900753498077393,
      "learning_rate": 9.448560736167224e-06,
      "loss": 0.0698,
      "step": 3087
    },
    {
      "epoch": 0.15946294861864188,
      "grad_norm": 1.3064405918121338,
      "learning_rate": 9.448186620306536e-06,
      "loss": 0.0297,
      "step": 3088
    },
    {
      "epoch": 0.1595145881745417,
      "grad_norm": 6.516858100891113,
      "learning_rate": 9.447812384993909e-06,
      "loss": 0.0761,
      "step": 3089
    },
    {
      "epoch": 0.15956622773044152,
      "grad_norm": 15.171470642089844,
      "learning_rate": 9.44743803023939e-06,
      "loss": 0.1074,
      "step": 3090
    },
    {
      "epoch": 0.15961786728634134,
      "grad_norm": 2.19924259185791,
      "learning_rate": 9.447063556053035e-06,
      "loss": 0.03,
      "step": 3091
    },
    {
      "epoch": 0.15966950684224115,
      "grad_norm": 8.294928550720215,
      "learning_rate": 9.446688962444897e-06,
      "loss": 0.0864,
      "step": 3092
    },
    {
      "epoch": 0.15972114639814097,
      "grad_norm": 11.811365127563477,
      "learning_rate": 9.446314249425038e-06,
      "loss": 0.069,
      "step": 3093
    },
    {
      "epoch": 0.1597727859540408,
      "grad_norm": 11.488304138183594,
      "learning_rate": 9.445939417003518e-06,
      "loss": 0.1652,
      "step": 3094
    },
    {
      "epoch": 0.1598244255099406,
      "grad_norm": 4.391025066375732,
      "learning_rate": 9.445564465190405e-06,
      "loss": 0.0479,
      "step": 3095
    },
    {
      "epoch": 0.15987606506584043,
      "grad_norm": 5.708139419555664,
      "learning_rate": 9.445189393995765e-06,
      "loss": 0.0745,
      "step": 3096
    },
    {
      "epoch": 0.15992770462174025,
      "grad_norm": 4.486438274383545,
      "learning_rate": 9.444814203429673e-06,
      "loss": 0.0363,
      "step": 3097
    },
    {
      "epoch": 0.15997934417764006,
      "grad_norm": 3.825712203979492,
      "learning_rate": 9.444438893502204e-06,
      "loss": 0.0416,
      "step": 3098
    },
    {
      "epoch": 0.16003098373353988,
      "grad_norm": 5.942570686340332,
      "learning_rate": 9.444063464223436e-06,
      "loss": 0.0692,
      "step": 3099
    },
    {
      "epoch": 0.1600826232894397,
      "grad_norm": 11.431912422180176,
      "learning_rate": 9.443687915603453e-06,
      "loss": 0.1859,
      "step": 3100
    },
    {
      "epoch": 0.16013426284533952,
      "grad_norm": 4.256719589233398,
      "learning_rate": 9.443312247652335e-06,
      "loss": 0.0794,
      "step": 3101
    },
    {
      "epoch": 0.16018590240123934,
      "grad_norm": 17.79850196838379,
      "learning_rate": 9.442936460380175e-06,
      "loss": 0.2356,
      "step": 3102
    },
    {
      "epoch": 0.16023754195713916,
      "grad_norm": 6.013779640197754,
      "learning_rate": 9.44256055379706e-06,
      "loss": 0.0951,
      "step": 3103
    },
    {
      "epoch": 0.16028918151303898,
      "grad_norm": 4.402412414550781,
      "learning_rate": 9.44218452791309e-06,
      "loss": 0.0692,
      "step": 3104
    },
    {
      "epoch": 0.1603408210689388,
      "grad_norm": 8.502350807189941,
      "learning_rate": 9.44180838273836e-06,
      "loss": 0.043,
      "step": 3105
    },
    {
      "epoch": 0.16039246062483864,
      "grad_norm": 9.5706148147583,
      "learning_rate": 9.441432118282969e-06,
      "loss": 0.1582,
      "step": 3106
    },
    {
      "epoch": 0.16044410018073846,
      "grad_norm": 17.98662567138672,
      "learning_rate": 9.441055734557025e-06,
      "loss": 0.1376,
      "step": 3107
    },
    {
      "epoch": 0.16049573973663828,
      "grad_norm": 11.069497108459473,
      "learning_rate": 9.440679231570632e-06,
      "loss": 0.0719,
      "step": 3108
    },
    {
      "epoch": 0.1605473792925381,
      "grad_norm": 4.473464012145996,
      "learning_rate": 9.440302609333905e-06,
      "loss": 0.0446,
      "step": 3109
    },
    {
      "epoch": 0.1605990188484379,
      "grad_norm": 5.370408058166504,
      "learning_rate": 9.439925867856953e-06,
      "loss": 0.0959,
      "step": 3110
    },
    {
      "epoch": 0.16065065840433773,
      "grad_norm": 5.322748184204102,
      "learning_rate": 9.439549007149897e-06,
      "loss": 0.0623,
      "step": 3111
    },
    {
      "epoch": 0.16070229796023755,
      "grad_norm": 2.9757096767425537,
      "learning_rate": 9.439172027222856e-06,
      "loss": 0.1029,
      "step": 3112
    },
    {
      "epoch": 0.16075393751613737,
      "grad_norm": 7.47032356262207,
      "learning_rate": 9.438794928085952e-06,
      "loss": 0.0789,
      "step": 3113
    },
    {
      "epoch": 0.1608055770720372,
      "grad_norm": 8.74948787689209,
      "learning_rate": 9.438417709749312e-06,
      "loss": 0.1289,
      "step": 3114
    },
    {
      "epoch": 0.160857216627937,
      "grad_norm": 4.6937432289123535,
      "learning_rate": 9.438040372223067e-06,
      "loss": 0.1205,
      "step": 3115
    },
    {
      "epoch": 0.16090885618383682,
      "grad_norm": 3.2799530029296875,
      "learning_rate": 9.437662915517352e-06,
      "loss": 0.0617,
      "step": 3116
    },
    {
      "epoch": 0.16096049573973664,
      "grad_norm": 2.482149600982666,
      "learning_rate": 9.437285339642296e-06,
      "loss": 0.0695,
      "step": 3117
    },
    {
      "epoch": 0.16101213529563646,
      "grad_norm": 7.284106731414795,
      "learning_rate": 9.436907644608046e-06,
      "loss": 0.0934,
      "step": 3118
    },
    {
      "epoch": 0.16106377485153628,
      "grad_norm": 5.459811687469482,
      "learning_rate": 9.436529830424743e-06,
      "loss": 0.0603,
      "step": 3119
    },
    {
      "epoch": 0.1611154144074361,
      "grad_norm": 3.153998374938965,
      "learning_rate": 9.43615189710253e-06,
      "loss": 0.0484,
      "step": 3120
    },
    {
      "epoch": 0.16116705396333592,
      "grad_norm": 4.743870258331299,
      "learning_rate": 9.43577384465156e-06,
      "loss": 0.0717,
      "step": 3121
    },
    {
      "epoch": 0.16121869351923573,
      "grad_norm": 6.654898166656494,
      "learning_rate": 9.43539567308198e-06,
      "loss": 0.0837,
      "step": 3122
    },
    {
      "epoch": 0.16127033307513555,
      "grad_norm": 3.2057275772094727,
      "learning_rate": 9.435017382403951e-06,
      "loss": 0.0557,
      "step": 3123
    },
    {
      "epoch": 0.16132197263103537,
      "grad_norm": 8.043968200683594,
      "learning_rate": 9.434638972627628e-06,
      "loss": 0.1336,
      "step": 3124
    },
    {
      "epoch": 0.1613736121869352,
      "grad_norm": 4.0014967918396,
      "learning_rate": 9.434260443763173e-06,
      "loss": 0.0973,
      "step": 3125
    },
    {
      "epoch": 0.161425251742835,
      "grad_norm": 5.006853103637695,
      "learning_rate": 9.433881795820753e-06,
      "loss": 0.0772,
      "step": 3126
    },
    {
      "epoch": 0.16147689129873483,
      "grad_norm": 16.28964614868164,
      "learning_rate": 9.433503028810535e-06,
      "loss": 0.1695,
      "step": 3127
    },
    {
      "epoch": 0.16152853085463464,
      "grad_norm": 3.1501283645629883,
      "learning_rate": 9.43312414274269e-06,
      "loss": 0.0831,
      "step": 3128
    },
    {
      "epoch": 0.16158017041053446,
      "grad_norm": 4.571203708648682,
      "learning_rate": 9.432745137627396e-06,
      "loss": 0.0672,
      "step": 3129
    },
    {
      "epoch": 0.16163180996643428,
      "grad_norm": 6.140068054199219,
      "learning_rate": 9.432366013474825e-06,
      "loss": 0.069,
      "step": 3130
    },
    {
      "epoch": 0.1616834495223341,
      "grad_norm": 4.209128379821777,
      "learning_rate": 9.431986770295164e-06,
      "loss": 0.0423,
      "step": 3131
    },
    {
      "epoch": 0.16173508907823392,
      "grad_norm": 11.748620986938477,
      "learning_rate": 9.431607408098592e-06,
      "loss": 0.1026,
      "step": 3132
    },
    {
      "epoch": 0.16178672863413374,
      "grad_norm": 6.542262077331543,
      "learning_rate": 9.4312279268953e-06,
      "loss": 0.0856,
      "step": 3133
    },
    {
      "epoch": 0.16183836819003355,
      "grad_norm": 4.649244785308838,
      "learning_rate": 9.430848326695479e-06,
      "loss": 0.0852,
      "step": 3134
    },
    {
      "epoch": 0.16189000774593337,
      "grad_norm": 9.42912769317627,
      "learning_rate": 9.430468607509319e-06,
      "loss": 0.1017,
      "step": 3135
    },
    {
      "epoch": 0.1619416473018332,
      "grad_norm": 9.624009132385254,
      "learning_rate": 9.430088769347021e-06,
      "loss": 0.1411,
      "step": 3136
    },
    {
      "epoch": 0.161993286857733,
      "grad_norm": 3.5556728839874268,
      "learning_rate": 9.429708812218783e-06,
      "loss": 0.0625,
      "step": 3137
    },
    {
      "epoch": 0.16204492641363286,
      "grad_norm": 4.129656791687012,
      "learning_rate": 9.42932873613481e-06,
      "loss": 0.0804,
      "step": 3138
    },
    {
      "epoch": 0.16209656596953267,
      "grad_norm": 3.3046345710754395,
      "learning_rate": 9.428948541105304e-06,
      "loss": 0.0584,
      "step": 3139
    },
    {
      "epoch": 0.1621482055254325,
      "grad_norm": 3.014556646347046,
      "learning_rate": 9.42856822714048e-06,
      "loss": 0.0466,
      "step": 3140
    },
    {
      "epoch": 0.1621998450813323,
      "grad_norm": 2.352619171142578,
      "learning_rate": 9.42818779425055e-06,
      "loss": 0.0588,
      "step": 3141
    },
    {
      "epoch": 0.16225148463723213,
      "grad_norm": 2.455746650695801,
      "learning_rate": 9.42780724244573e-06,
      "loss": 0.0484,
      "step": 3142
    },
    {
      "epoch": 0.16230312419313195,
      "grad_norm": 3.324453592300415,
      "learning_rate": 9.427426571736238e-06,
      "loss": 0.0555,
      "step": 3143
    },
    {
      "epoch": 0.16235476374903177,
      "grad_norm": 8.421226501464844,
      "learning_rate": 9.427045782132296e-06,
      "loss": 0.0702,
      "step": 3144
    },
    {
      "epoch": 0.16240640330493158,
      "grad_norm": 6.847921371459961,
      "learning_rate": 9.426664873644133e-06,
      "loss": 0.0577,
      "step": 3145
    },
    {
      "epoch": 0.1624580428608314,
      "grad_norm": 1.8430030345916748,
      "learning_rate": 9.426283846281975e-06,
      "loss": 0.0321,
      "step": 3146
    },
    {
      "epoch": 0.16250968241673122,
      "grad_norm": 6.632254600524902,
      "learning_rate": 9.425902700056056e-06,
      "loss": 0.0987,
      "step": 3147
    },
    {
      "epoch": 0.16256132197263104,
      "grad_norm": 15.570819854736328,
      "learning_rate": 9.42552143497661e-06,
      "loss": 0.1937,
      "step": 3148
    },
    {
      "epoch": 0.16261296152853086,
      "grad_norm": 8.131285667419434,
      "learning_rate": 9.425140051053876e-06,
      "loss": 0.0967,
      "step": 3149
    },
    {
      "epoch": 0.16266460108443068,
      "grad_norm": 7.036497116088867,
      "learning_rate": 9.424758548298095e-06,
      "loss": 0.054,
      "step": 3150
    },
    {
      "epoch": 0.1627162406403305,
      "grad_norm": 13.540681838989258,
      "learning_rate": 9.424376926719513e-06,
      "loss": 0.0864,
      "step": 3151
    },
    {
      "epoch": 0.1627678801962303,
      "grad_norm": 7.8065900802612305,
      "learning_rate": 9.423995186328378e-06,
      "loss": 0.1253,
      "step": 3152
    },
    {
      "epoch": 0.16281951975213013,
      "grad_norm": 6.210328102111816,
      "learning_rate": 9.423613327134942e-06,
      "loss": 0.0849,
      "step": 3153
    },
    {
      "epoch": 0.16287115930802995,
      "grad_norm": 29.78059959411621,
      "learning_rate": 9.423231349149457e-06,
      "loss": 0.0839,
      "step": 3154
    },
    {
      "epoch": 0.16292279886392977,
      "grad_norm": 3.730834722518921,
      "learning_rate": 9.422849252382184e-06,
      "loss": 0.0446,
      "step": 3155
    },
    {
      "epoch": 0.1629744384198296,
      "grad_norm": 8.693326950073242,
      "learning_rate": 9.422467036843382e-06,
      "loss": 0.0996,
      "step": 3156
    },
    {
      "epoch": 0.1630260779757294,
      "grad_norm": 3.766159772872925,
      "learning_rate": 9.422084702543313e-06,
      "loss": 0.0669,
      "step": 3157
    },
    {
      "epoch": 0.16307771753162922,
      "grad_norm": 13.095810890197754,
      "learning_rate": 9.421702249492249e-06,
      "loss": 0.1638,
      "step": 3158
    },
    {
      "epoch": 0.16312935708752904,
      "grad_norm": 3.714505910873413,
      "learning_rate": 9.421319677700456e-06,
      "loss": 0.0722,
      "step": 3159
    },
    {
      "epoch": 0.16318099664342886,
      "grad_norm": 10.337386131286621,
      "learning_rate": 9.42093698717821e-06,
      "loss": 0.1582,
      "step": 3160
    },
    {
      "epoch": 0.16323263619932868,
      "grad_norm": 7.434863567352295,
      "learning_rate": 9.420554177935786e-06,
      "loss": 0.0977,
      "step": 3161
    },
    {
      "epoch": 0.1632842757552285,
      "grad_norm": 2.924973487854004,
      "learning_rate": 9.420171249983466e-06,
      "loss": 0.0938,
      "step": 3162
    },
    {
      "epoch": 0.16333591531112832,
      "grad_norm": 10.973280906677246,
      "learning_rate": 9.419788203331531e-06,
      "loss": 0.1094,
      "step": 3163
    },
    {
      "epoch": 0.16338755486702813,
      "grad_norm": 10.014345169067383,
      "learning_rate": 9.41940503799027e-06,
      "loss": 0.1046,
      "step": 3164
    },
    {
      "epoch": 0.16343919442292795,
      "grad_norm": 22.000083923339844,
      "learning_rate": 9.419021753969971e-06,
      "loss": 0.1636,
      "step": 3165
    },
    {
      "epoch": 0.16349083397882777,
      "grad_norm": 6.673109531402588,
      "learning_rate": 9.418638351280927e-06,
      "loss": 0.1609,
      "step": 3166
    },
    {
      "epoch": 0.1635424735347276,
      "grad_norm": 4.028415679931641,
      "learning_rate": 9.418254829933432e-06,
      "loss": 0.0336,
      "step": 3167
    },
    {
      "epoch": 0.1635941130906274,
      "grad_norm": 12.133637428283691,
      "learning_rate": 9.417871189937791e-06,
      "loss": 0.1379,
      "step": 3168
    },
    {
      "epoch": 0.16364575264652723,
      "grad_norm": 4.150125503540039,
      "learning_rate": 9.4174874313043e-06,
      "loss": 0.1051,
      "step": 3169
    },
    {
      "epoch": 0.16369739220242707,
      "grad_norm": 20.007564544677734,
      "learning_rate": 9.417103554043268e-06,
      "loss": 0.1543,
      "step": 3170
    },
    {
      "epoch": 0.1637490317583269,
      "grad_norm": 4.342329978942871,
      "learning_rate": 9.416719558165001e-06,
      "loss": 0.0807,
      "step": 3171
    },
    {
      "epoch": 0.1638006713142267,
      "grad_norm": 8.452492713928223,
      "learning_rate": 9.41633544367981e-06,
      "loss": 0.0706,
      "step": 3172
    },
    {
      "epoch": 0.16385231087012653,
      "grad_norm": 5.653522968292236,
      "learning_rate": 9.415951210598017e-06,
      "loss": 0.1008,
      "step": 3173
    },
    {
      "epoch": 0.16390395042602635,
      "grad_norm": 4.448588848114014,
      "learning_rate": 9.415566858929934e-06,
      "loss": 0.065,
      "step": 3174
    },
    {
      "epoch": 0.16395558998192616,
      "grad_norm": 5.633265018463135,
      "learning_rate": 9.415182388685884e-06,
      "loss": 0.0926,
      "step": 3175
    },
    {
      "epoch": 0.16400722953782598,
      "grad_norm": 1.8980250358581543,
      "learning_rate": 9.41479779987619e-06,
      "loss": 0.0566,
      "step": 3176
    },
    {
      "epoch": 0.1640588690937258,
      "grad_norm": 5.167078018188477,
      "learning_rate": 9.414413092511183e-06,
      "loss": 0.0847,
      "step": 3177
    },
    {
      "epoch": 0.16411050864962562,
      "grad_norm": 8.154457092285156,
      "learning_rate": 9.414028266601192e-06,
      "loss": 0.1346,
      "step": 3178
    },
    {
      "epoch": 0.16416214820552544,
      "grad_norm": 4.294833183288574,
      "learning_rate": 9.413643322156552e-06,
      "loss": 0.0973,
      "step": 3179
    },
    {
      "epoch": 0.16421378776142526,
      "grad_norm": 6.166207313537598,
      "learning_rate": 9.413258259187598e-06,
      "loss": 0.0822,
      "step": 3180
    },
    {
      "epoch": 0.16426542731732507,
      "grad_norm": 2.2882304191589355,
      "learning_rate": 9.412873077704674e-06,
      "loss": 0.0407,
      "step": 3181
    },
    {
      "epoch": 0.1643170668732249,
      "grad_norm": 14.58292007446289,
      "learning_rate": 9.412487777718123e-06,
      "loss": 0.0966,
      "step": 3182
    },
    {
      "epoch": 0.1643687064291247,
      "grad_norm": 7.963947772979736,
      "learning_rate": 9.412102359238289e-06,
      "loss": 0.1495,
      "step": 3183
    },
    {
      "epoch": 0.16442034598502453,
      "grad_norm": 3.645371675491333,
      "learning_rate": 9.411716822275523e-06,
      "loss": 0.0516,
      "step": 3184
    },
    {
      "epoch": 0.16447198554092435,
      "grad_norm": 9.786040306091309,
      "learning_rate": 9.41133116684018e-06,
      "loss": 0.1677,
      "step": 3185
    },
    {
      "epoch": 0.16452362509682417,
      "grad_norm": 19.157485961914062,
      "learning_rate": 9.410945392942616e-06,
      "loss": 0.1155,
      "step": 3186
    },
    {
      "epoch": 0.16457526465272398,
      "grad_norm": 7.7990031242370605,
      "learning_rate": 9.410559500593191e-06,
      "loss": 0.1155,
      "step": 3187
    },
    {
      "epoch": 0.1646269042086238,
      "grad_norm": 10.993341445922852,
      "learning_rate": 9.410173489802266e-06,
      "loss": 0.0909,
      "step": 3188
    },
    {
      "epoch": 0.16467854376452362,
      "grad_norm": 16.763957977294922,
      "learning_rate": 9.409787360580208e-06,
      "loss": 0.2866,
      "step": 3189
    },
    {
      "epoch": 0.16473018332042344,
      "grad_norm": 5.879546165466309,
      "learning_rate": 9.409401112937386e-06,
      "loss": 0.0742,
      "step": 3190
    },
    {
      "epoch": 0.16478182287632326,
      "grad_norm": 17.324600219726562,
      "learning_rate": 9.409014746884174e-06,
      "loss": 0.1697,
      "step": 3191
    },
    {
      "epoch": 0.16483346243222308,
      "grad_norm": 2.6133029460906982,
      "learning_rate": 9.408628262430943e-06,
      "loss": 0.0527,
      "step": 3192
    },
    {
      "epoch": 0.1648851019881229,
      "grad_norm": 7.1047821044921875,
      "learning_rate": 9.408241659588076e-06,
      "loss": 0.1156,
      "step": 3193
    },
    {
      "epoch": 0.1649367415440227,
      "grad_norm": 8.290193557739258,
      "learning_rate": 9.407854938365953e-06,
      "loss": 0.1193,
      "step": 3194
    },
    {
      "epoch": 0.16498838109992253,
      "grad_norm": 6.519601821899414,
      "learning_rate": 9.40746809877496e-06,
      "loss": 0.0625,
      "step": 3195
    },
    {
      "epoch": 0.16504002065582235,
      "grad_norm": 13.721436500549316,
      "learning_rate": 9.407081140825485e-06,
      "loss": 0.1932,
      "step": 3196
    },
    {
      "epoch": 0.16509166021172217,
      "grad_norm": 17.17200469970703,
      "learning_rate": 9.40669406452792e-06,
      "loss": 0.1452,
      "step": 3197
    },
    {
      "epoch": 0.165143299767622,
      "grad_norm": 10.291814804077148,
      "learning_rate": 9.406306869892657e-06,
      "loss": 0.2169,
      "step": 3198
    },
    {
      "epoch": 0.1651949393235218,
      "grad_norm": 3.9577572345733643,
      "learning_rate": 9.405919556930095e-06,
      "loss": 0.0683,
      "step": 3199
    },
    {
      "epoch": 0.16524657887942162,
      "grad_norm": 3.6242575645446777,
      "learning_rate": 9.405532125650635e-06,
      "loss": 0.0841,
      "step": 3200
    },
    {
      "epoch": 0.16529821843532147,
      "grad_norm": 10.795676231384277,
      "learning_rate": 9.405144576064684e-06,
      "loss": 0.0836,
      "step": 3201
    },
    {
      "epoch": 0.1653498579912213,
      "grad_norm": 7.245362281799316,
      "learning_rate": 9.404756908182646e-06,
      "loss": 0.1013,
      "step": 3202
    },
    {
      "epoch": 0.1654014975471211,
      "grad_norm": 3.2853362560272217,
      "learning_rate": 9.404369122014931e-06,
      "loss": 0.0569,
      "step": 3203
    },
    {
      "epoch": 0.16545313710302093,
      "grad_norm": 7.1163811683654785,
      "learning_rate": 9.403981217571954e-06,
      "loss": 0.0873,
      "step": 3204
    },
    {
      "epoch": 0.16550477665892074,
      "grad_norm": 7.864687442779541,
      "learning_rate": 9.403593194864133e-06,
      "loss": 0.0938,
      "step": 3205
    },
    {
      "epoch": 0.16555641621482056,
      "grad_norm": 6.698964595794678,
      "learning_rate": 9.403205053901886e-06,
      "loss": 0.1054,
      "step": 3206
    },
    {
      "epoch": 0.16560805577072038,
      "grad_norm": 12.474883079528809,
      "learning_rate": 9.402816794695635e-06,
      "loss": 0.1343,
      "step": 3207
    },
    {
      "epoch": 0.1656596953266202,
      "grad_norm": 6.394275188446045,
      "learning_rate": 9.40242841725581e-06,
      "loss": 0.0512,
      "step": 3208
    },
    {
      "epoch": 0.16571133488252002,
      "grad_norm": 8.150120735168457,
      "learning_rate": 9.402039921592839e-06,
      "loss": 0.0795,
      "step": 3209
    },
    {
      "epoch": 0.16576297443841984,
      "grad_norm": 4.112457752227783,
      "learning_rate": 9.401651307717154e-06,
      "loss": 0.0507,
      "step": 3210
    },
    {
      "epoch": 0.16581461399431965,
      "grad_norm": 4.848061561584473,
      "learning_rate": 9.401262575639192e-06,
      "loss": 0.0687,
      "step": 3211
    },
    {
      "epoch": 0.16586625355021947,
      "grad_norm": 4.004753112792969,
      "learning_rate": 9.40087372536939e-06,
      "loss": 0.0581,
      "step": 3212
    },
    {
      "epoch": 0.1659178931061193,
      "grad_norm": 5.312376976013184,
      "learning_rate": 9.400484756918192e-06,
      "loss": 0.0646,
      "step": 3213
    },
    {
      "epoch": 0.1659695326620191,
      "grad_norm": 4.2090911865234375,
      "learning_rate": 9.400095670296043e-06,
      "loss": 0.0579,
      "step": 3214
    },
    {
      "epoch": 0.16602117221791893,
      "grad_norm": 7.4397783279418945,
      "learning_rate": 9.399706465513391e-06,
      "loss": 0.1136,
      "step": 3215
    },
    {
      "epoch": 0.16607281177381875,
      "grad_norm": 13.144679069519043,
      "learning_rate": 9.399317142580688e-06,
      "loss": 0.0961,
      "step": 3216
    },
    {
      "epoch": 0.16612445132971856,
      "grad_norm": 10.376190185546875,
      "learning_rate": 9.398927701508389e-06,
      "loss": 0.085,
      "step": 3217
    },
    {
      "epoch": 0.16617609088561838,
      "grad_norm": 4.380043983459473,
      "learning_rate": 9.398538142306953e-06,
      "loss": 0.0588,
      "step": 3218
    },
    {
      "epoch": 0.1662277304415182,
      "grad_norm": 10.319921493530273,
      "learning_rate": 9.398148464986841e-06,
      "loss": 0.0731,
      "step": 3219
    },
    {
      "epoch": 0.16627936999741802,
      "grad_norm": 7.816030502319336,
      "learning_rate": 9.397758669558515e-06,
      "loss": 0.09,
      "step": 3220
    },
    {
      "epoch": 0.16633100955331784,
      "grad_norm": 3.4367733001708984,
      "learning_rate": 9.397368756032445e-06,
      "loss": 0.0664,
      "step": 3221
    },
    {
      "epoch": 0.16638264910921766,
      "grad_norm": 6.828112602233887,
      "learning_rate": 9.396978724419104e-06,
      "loss": 0.0941,
      "step": 3222
    },
    {
      "epoch": 0.16643428866511747,
      "grad_norm": 25.85384178161621,
      "learning_rate": 9.39658857472896e-06,
      "loss": 0.1413,
      "step": 3223
    },
    {
      "epoch": 0.1664859282210173,
      "grad_norm": 4.597131729125977,
      "learning_rate": 9.396198306972493e-06,
      "loss": 0.0751,
      "step": 3224
    },
    {
      "epoch": 0.1665375677769171,
      "grad_norm": 5.331416130065918,
      "learning_rate": 9.395807921160185e-06,
      "loss": 0.0866,
      "step": 3225
    },
    {
      "epoch": 0.16658920733281693,
      "grad_norm": 8.205095291137695,
      "learning_rate": 9.395417417302518e-06,
      "loss": 0.0799,
      "step": 3226
    },
    {
      "epoch": 0.16664084688871675,
      "grad_norm": 4.963799476623535,
      "learning_rate": 9.395026795409979e-06,
      "loss": 0.0966,
      "step": 3227
    },
    {
      "epoch": 0.16669248644461657,
      "grad_norm": 8.042144775390625,
      "learning_rate": 9.394636055493057e-06,
      "loss": 0.0818,
      "step": 3228
    },
    {
      "epoch": 0.16674412600051638,
      "grad_norm": 10.107346534729004,
      "learning_rate": 9.394245197562245e-06,
      "loss": 0.0749,
      "step": 3229
    },
    {
      "epoch": 0.1667957655564162,
      "grad_norm": 5.022590637207031,
      "learning_rate": 9.393854221628038e-06,
      "loss": 0.0673,
      "step": 3230
    },
    {
      "epoch": 0.16684740511231602,
      "grad_norm": 2.516004800796509,
      "learning_rate": 9.39346312770094e-06,
      "loss": 0.0583,
      "step": 3231
    },
    {
      "epoch": 0.16689904466821584,
      "grad_norm": 2.417232036590576,
      "learning_rate": 9.393071915791447e-06,
      "loss": 0.046,
      "step": 3232
    },
    {
      "epoch": 0.1669506842241157,
      "grad_norm": 12.836045265197754,
      "learning_rate": 9.39268058591007e-06,
      "loss": 0.1319,
      "step": 3233
    },
    {
      "epoch": 0.1670023237800155,
      "grad_norm": 12.002911567687988,
      "learning_rate": 9.392289138067315e-06,
      "loss": 0.1407,
      "step": 3234
    },
    {
      "epoch": 0.16705396333591532,
      "grad_norm": 6.827391624450684,
      "learning_rate": 9.391897572273695e-06,
      "loss": 0.068,
      "step": 3235
    },
    {
      "epoch": 0.16710560289181514,
      "grad_norm": 12.803428649902344,
      "learning_rate": 9.391505888539725e-06,
      "loss": 0.1185,
      "step": 3236
    },
    {
      "epoch": 0.16715724244771496,
      "grad_norm": 9.68801498413086,
      "learning_rate": 9.391114086875923e-06,
      "loss": 0.0691,
      "step": 3237
    },
    {
      "epoch": 0.16720888200361478,
      "grad_norm": 10.606830596923828,
      "learning_rate": 9.39072216729281e-06,
      "loss": 0.072,
      "step": 3238
    },
    {
      "epoch": 0.1672605215595146,
      "grad_norm": 5.270451068878174,
      "learning_rate": 9.39033012980091e-06,
      "loss": 0.0634,
      "step": 3239
    },
    {
      "epoch": 0.16731216111541442,
      "grad_norm": 24.505212783813477,
      "learning_rate": 9.389937974410754e-06,
      "loss": 0.0992,
      "step": 3240
    },
    {
      "epoch": 0.16736380067131423,
      "grad_norm": 6.441568374633789,
      "learning_rate": 9.389545701132872e-06,
      "loss": 0.1058,
      "step": 3241
    },
    {
      "epoch": 0.16741544022721405,
      "grad_norm": 2.96127986907959,
      "learning_rate": 9.389153309977795e-06,
      "loss": 0.0827,
      "step": 3242
    },
    {
      "epoch": 0.16746707978311387,
      "grad_norm": 4.0107855796813965,
      "learning_rate": 9.388760800956062e-06,
      "loss": 0.0486,
      "step": 3243
    },
    {
      "epoch": 0.1675187193390137,
      "grad_norm": 8.963541984558105,
      "learning_rate": 9.388368174078216e-06,
      "loss": 0.1095,
      "step": 3244
    },
    {
      "epoch": 0.1675703588949135,
      "grad_norm": 9.474458694458008,
      "learning_rate": 9.387975429354798e-06,
      "loss": 0.1688,
      "step": 3245
    },
    {
      "epoch": 0.16762199845081333,
      "grad_norm": 5.226538181304932,
      "learning_rate": 9.387582566796354e-06,
      "loss": 0.0975,
      "step": 3246
    },
    {
      "epoch": 0.16767363800671314,
      "grad_norm": 3.3522257804870605,
      "learning_rate": 9.387189586413437e-06,
      "loss": 0.0567,
      "step": 3247
    },
    {
      "epoch": 0.16772527756261296,
      "grad_norm": 3.444113254547119,
      "learning_rate": 9.386796488216598e-06,
      "loss": 0.0606,
      "step": 3248
    },
    {
      "epoch": 0.16777691711851278,
      "grad_norm": 8.080880165100098,
      "learning_rate": 9.386403272216393e-06,
      "loss": 0.0605,
      "step": 3249
    },
    {
      "epoch": 0.1678285566744126,
      "grad_norm": 9.071261405944824,
      "learning_rate": 9.386009938423382e-06,
      "loss": 0.0795,
      "step": 3250
    },
    {
      "epoch": 0.16788019623031242,
      "grad_norm": 3.9676196575164795,
      "learning_rate": 9.385616486848128e-06,
      "loss": 0.0934,
      "step": 3251
    },
    {
      "epoch": 0.16793183578621224,
      "grad_norm": 2.203986883163452,
      "learning_rate": 9.385222917501197e-06,
      "loss": 0.0274,
      "step": 3252
    },
    {
      "epoch": 0.16798347534211205,
      "grad_norm": 6.780379295349121,
      "learning_rate": 9.384829230393158e-06,
      "loss": 0.0747,
      "step": 3253
    },
    {
      "epoch": 0.16803511489801187,
      "grad_norm": 9.4841947555542,
      "learning_rate": 9.384435425534582e-06,
      "loss": 0.1037,
      "step": 3254
    },
    {
      "epoch": 0.1680867544539117,
      "grad_norm": 5.408870697021484,
      "learning_rate": 9.384041502936044e-06,
      "loss": 0.0367,
      "step": 3255
    },
    {
      "epoch": 0.1681383940098115,
      "grad_norm": 6.180603981018066,
      "learning_rate": 9.383647462608123e-06,
      "loss": 0.0875,
      "step": 3256
    },
    {
      "epoch": 0.16819003356571133,
      "grad_norm": 10.069533348083496,
      "learning_rate": 9.383253304561401e-06,
      "loss": 0.1184,
      "step": 3257
    },
    {
      "epoch": 0.16824167312161115,
      "grad_norm": 11.104110717773438,
      "learning_rate": 9.382859028806463e-06,
      "loss": 0.1106,
      "step": 3258
    },
    {
      "epoch": 0.16829331267751096,
      "grad_norm": 6.124701499938965,
      "learning_rate": 9.382464635353898e-06,
      "loss": 0.0733,
      "step": 3259
    },
    {
      "epoch": 0.16834495223341078,
      "grad_norm": 4.153106689453125,
      "learning_rate": 9.382070124214296e-06,
      "loss": 0.0823,
      "step": 3260
    },
    {
      "epoch": 0.1683965917893106,
      "grad_norm": 8.181729316711426,
      "learning_rate": 9.381675495398248e-06,
      "loss": 0.1152,
      "step": 3261
    },
    {
      "epoch": 0.16844823134521042,
      "grad_norm": 2.3755249977111816,
      "learning_rate": 9.381280748916355e-06,
      "loss": 0.0556,
      "step": 3262
    },
    {
      "epoch": 0.16849987090111024,
      "grad_norm": 4.079611301422119,
      "learning_rate": 9.380885884779218e-06,
      "loss": 0.0534,
      "step": 3263
    },
    {
      "epoch": 0.16855151045701006,
      "grad_norm": 11.080970764160156,
      "learning_rate": 9.38049090299744e-06,
      "loss": 0.09,
      "step": 3264
    },
    {
      "epoch": 0.1686031500129099,
      "grad_norm": 4.984097003936768,
      "learning_rate": 9.380095803581626e-06,
      "loss": 0.055,
      "step": 3265
    },
    {
      "epoch": 0.16865478956880972,
      "grad_norm": 3.8397254943847656,
      "learning_rate": 9.379700586542387e-06,
      "loss": 0.0626,
      "step": 3266
    },
    {
      "epoch": 0.16870642912470954,
      "grad_norm": 12.429468154907227,
      "learning_rate": 9.379305251890336e-06,
      "loss": 0.0738,
      "step": 3267
    },
    {
      "epoch": 0.16875806868060936,
      "grad_norm": 5.027521133422852,
      "learning_rate": 9.378909799636093e-06,
      "loss": 0.0753,
      "step": 3268
    },
    {
      "epoch": 0.16880970823650918,
      "grad_norm": 2.48911452293396,
      "learning_rate": 9.378514229790272e-06,
      "loss": 0.0411,
      "step": 3269
    },
    {
      "epoch": 0.168861347792409,
      "grad_norm": 2.5688722133636475,
      "learning_rate": 9.378118542363497e-06,
      "loss": 0.0437,
      "step": 3270
    },
    {
      "epoch": 0.1689129873483088,
      "grad_norm": 1.6965843439102173,
      "learning_rate": 9.377722737366397e-06,
      "loss": 0.034,
      "step": 3271
    },
    {
      "epoch": 0.16896462690420863,
      "grad_norm": 9.210189819335938,
      "learning_rate": 9.3773268148096e-06,
      "loss": 0.0873,
      "step": 3272
    },
    {
      "epoch": 0.16901626646010845,
      "grad_norm": 2.7514517307281494,
      "learning_rate": 9.376930774703734e-06,
      "loss": 0.0566,
      "step": 3273
    },
    {
      "epoch": 0.16906790601600827,
      "grad_norm": 4.1798014640808105,
      "learning_rate": 9.37653461705944e-06,
      "loss": 0.0593,
      "step": 3274
    },
    {
      "epoch": 0.1691195455719081,
      "grad_norm": 6.553795337677002,
      "learning_rate": 9.376138341887352e-06,
      "loss": 0.0831,
      "step": 3275
    },
    {
      "epoch": 0.1691711851278079,
      "grad_norm": 5.061028003692627,
      "learning_rate": 9.375741949198114e-06,
      "loss": 0.0644,
      "step": 3276
    },
    {
      "epoch": 0.16922282468370772,
      "grad_norm": 6.586737155914307,
      "learning_rate": 9.37534543900237e-06,
      "loss": 0.0431,
      "step": 3277
    },
    {
      "epoch": 0.16927446423960754,
      "grad_norm": 5.791024684906006,
      "learning_rate": 9.374948811310768e-06,
      "loss": 0.0846,
      "step": 3278
    },
    {
      "epoch": 0.16932610379550736,
      "grad_norm": 3.576448440551758,
      "learning_rate": 9.37455206613396e-06,
      "loss": 0.1008,
      "step": 3279
    },
    {
      "epoch": 0.16937774335140718,
      "grad_norm": 4.101036548614502,
      "learning_rate": 9.374155203482597e-06,
      "loss": 0.0686,
      "step": 3280
    },
    {
      "epoch": 0.169429382907307,
      "grad_norm": 2.8180501461029053,
      "learning_rate": 9.373758223367342e-06,
      "loss": 0.0488,
      "step": 3281
    },
    {
      "epoch": 0.16948102246320682,
      "grad_norm": 3.002063274383545,
      "learning_rate": 9.37336112579885e-06,
      "loss": 0.034,
      "step": 3282
    },
    {
      "epoch": 0.16953266201910663,
      "grad_norm": 2.865342855453491,
      "learning_rate": 9.372963910787786e-06,
      "loss": 0.0382,
      "step": 3283
    },
    {
      "epoch": 0.16958430157500645,
      "grad_norm": 6.432090759277344,
      "learning_rate": 9.372566578344819e-06,
      "loss": 0.0611,
      "step": 3284
    },
    {
      "epoch": 0.16963594113090627,
      "grad_norm": 19.40105438232422,
      "learning_rate": 9.372169128480619e-06,
      "loss": 0.2487,
      "step": 3285
    },
    {
      "epoch": 0.1696875806868061,
      "grad_norm": 4.729001998901367,
      "learning_rate": 9.371771561205858e-06,
      "loss": 0.0535,
      "step": 3286
    },
    {
      "epoch": 0.1697392202427059,
      "grad_norm": 3.7741305828094482,
      "learning_rate": 9.371373876531211e-06,
      "loss": 0.0505,
      "step": 3287
    },
    {
      "epoch": 0.16979085979860573,
      "grad_norm": 10.966132164001465,
      "learning_rate": 9.370976074467359e-06,
      "loss": 0.1424,
      "step": 3288
    },
    {
      "epoch": 0.16984249935450554,
      "grad_norm": 5.808376789093018,
      "learning_rate": 9.370578155024983e-06,
      "loss": 0.0873,
      "step": 3289
    },
    {
      "epoch": 0.16989413891040536,
      "grad_norm": 6.545229434967041,
      "learning_rate": 9.370180118214772e-06,
      "loss": 0.081,
      "step": 3290
    },
    {
      "epoch": 0.16994577846630518,
      "grad_norm": 5.866652488708496,
      "learning_rate": 9.36978196404741e-06,
      "loss": 0.0633,
      "step": 3291
    },
    {
      "epoch": 0.169997418022205,
      "grad_norm": 1.734807014465332,
      "learning_rate": 9.369383692533596e-06,
      "loss": 0.03,
      "step": 3292
    },
    {
      "epoch": 0.17004905757810482,
      "grad_norm": 9.025823593139648,
      "learning_rate": 9.368985303684019e-06,
      "loss": 0.0747,
      "step": 3293
    },
    {
      "epoch": 0.17010069713400464,
      "grad_norm": 2.886167526245117,
      "learning_rate": 9.368586797509381e-06,
      "loss": 0.0466,
      "step": 3294
    },
    {
      "epoch": 0.17015233668990445,
      "grad_norm": 4.575875759124756,
      "learning_rate": 9.368188174020381e-06,
      "loss": 0.0502,
      "step": 3295
    },
    {
      "epoch": 0.17020397624580427,
      "grad_norm": 8.02669620513916,
      "learning_rate": 9.367789433227724e-06,
      "loss": 0.0711,
      "step": 3296
    },
    {
      "epoch": 0.17025561580170412,
      "grad_norm": 3.876352071762085,
      "learning_rate": 9.367390575142119e-06,
      "loss": 0.0444,
      "step": 3297
    },
    {
      "epoch": 0.17030725535760394,
      "grad_norm": 11.598467826843262,
      "learning_rate": 9.366991599774278e-06,
      "loss": 0.2006,
      "step": 3298
    },
    {
      "epoch": 0.17035889491350376,
      "grad_norm": 6.483092784881592,
      "learning_rate": 9.366592507134913e-06,
      "loss": 0.0574,
      "step": 3299
    },
    {
      "epoch": 0.17041053446940357,
      "grad_norm": 6.753883361816406,
      "learning_rate": 9.36619329723474e-06,
      "loss": 0.0745,
      "step": 3300
    },
    {
      "epoch": 0.1704621740253034,
      "grad_norm": 3.922961950302124,
      "learning_rate": 9.365793970084485e-06,
      "loss": 0.0581,
      "step": 3301
    },
    {
      "epoch": 0.1705138135812032,
      "grad_norm": 20.56624412536621,
      "learning_rate": 9.365394525694865e-06,
      "loss": 0.1365,
      "step": 3302
    },
    {
      "epoch": 0.17056545313710303,
      "grad_norm": 4.060327529907227,
      "learning_rate": 9.36499496407661e-06,
      "loss": 0.0861,
      "step": 3303
    },
    {
      "epoch": 0.17061709269300285,
      "grad_norm": 9.744983673095703,
      "learning_rate": 9.36459528524045e-06,
      "loss": 0.0793,
      "step": 3304
    },
    {
      "epoch": 0.17066873224890267,
      "grad_norm": 1.9611408710479736,
      "learning_rate": 9.364195489197118e-06,
      "loss": 0.0335,
      "step": 3305
    },
    {
      "epoch": 0.17072037180480248,
      "grad_norm": 5.8993988037109375,
      "learning_rate": 9.363795575957349e-06,
      "loss": 0.0823,
      "step": 3306
    },
    {
      "epoch": 0.1707720113607023,
      "grad_norm": 10.57348346710205,
      "learning_rate": 9.363395545531884e-06,
      "loss": 0.112,
      "step": 3307
    },
    {
      "epoch": 0.17082365091660212,
      "grad_norm": 5.937774181365967,
      "learning_rate": 9.362995397931462e-06,
      "loss": 0.0907,
      "step": 3308
    },
    {
      "epoch": 0.17087529047250194,
      "grad_norm": 5.850191593170166,
      "learning_rate": 9.362595133166832e-06,
      "loss": 0.0768,
      "step": 3309
    },
    {
      "epoch": 0.17092693002840176,
      "grad_norm": 4.004758358001709,
      "learning_rate": 9.362194751248744e-06,
      "loss": 0.0588,
      "step": 3310
    },
    {
      "epoch": 0.17097856958430158,
      "grad_norm": 4.900980472564697,
      "learning_rate": 9.361794252187945e-06,
      "loss": 0.0589,
      "step": 3311
    },
    {
      "epoch": 0.1710302091402014,
      "grad_norm": 10.694425582885742,
      "learning_rate": 9.361393635995192e-06,
      "loss": 0.1016,
      "step": 3312
    },
    {
      "epoch": 0.1710818486961012,
      "grad_norm": 3.0462965965270996,
      "learning_rate": 9.360992902681247e-06,
      "loss": 0.0569,
      "step": 3313
    },
    {
      "epoch": 0.17113348825200103,
      "grad_norm": 6.560970783233643,
      "learning_rate": 9.360592052256865e-06,
      "loss": 0.0686,
      "step": 3314
    },
    {
      "epoch": 0.17118512780790085,
      "grad_norm": 3.2249066829681396,
      "learning_rate": 9.360191084732815e-06,
      "loss": 0.078,
      "step": 3315
    },
    {
      "epoch": 0.17123676736380067,
      "grad_norm": 3.0083961486816406,
      "learning_rate": 9.359790000119864e-06,
      "loss": 0.0586,
      "step": 3316
    },
    {
      "epoch": 0.1712884069197005,
      "grad_norm": 6.160336971282959,
      "learning_rate": 9.359388798428782e-06,
      "loss": 0.0681,
      "step": 3317
    },
    {
      "epoch": 0.1713400464756003,
      "grad_norm": 7.039358615875244,
      "learning_rate": 9.35898747967034e-06,
      "loss": 0.0964,
      "step": 3318
    },
    {
      "epoch": 0.17139168603150012,
      "grad_norm": 10.167146682739258,
      "learning_rate": 9.358586043855321e-06,
      "loss": 0.1543,
      "step": 3319
    },
    {
      "epoch": 0.17144332558739994,
      "grad_norm": 4.026955604553223,
      "learning_rate": 9.3581844909945e-06,
      "loss": 0.0506,
      "step": 3320
    },
    {
      "epoch": 0.17149496514329976,
      "grad_norm": 7.466655731201172,
      "learning_rate": 9.357782821098665e-06,
      "loss": 0.0846,
      "step": 3321
    },
    {
      "epoch": 0.17154660469919958,
      "grad_norm": 5.6201491355896,
      "learning_rate": 9.357381034178597e-06,
      "loss": 0.1136,
      "step": 3322
    },
    {
      "epoch": 0.1715982442550994,
      "grad_norm": 2.7121129035949707,
      "learning_rate": 9.356979130245091e-06,
      "loss": 0.0704,
      "step": 3323
    },
    {
      "epoch": 0.17164988381099922,
      "grad_norm": 2.7126290798187256,
      "learning_rate": 9.356577109308936e-06,
      "loss": 0.0606,
      "step": 3324
    },
    {
      "epoch": 0.17170152336689903,
      "grad_norm": 21.394573211669922,
      "learning_rate": 9.356174971380929e-06,
      "loss": 0.3166,
      "step": 3325
    },
    {
      "epoch": 0.17175316292279885,
      "grad_norm": 3.5956225395202637,
      "learning_rate": 9.35577271647187e-06,
      "loss": 0.0862,
      "step": 3326
    },
    {
      "epoch": 0.17180480247869867,
      "grad_norm": 9.923892974853516,
      "learning_rate": 9.35537034459256e-06,
      "loss": 0.11,
      "step": 3327
    },
    {
      "epoch": 0.1718564420345985,
      "grad_norm": 5.168987274169922,
      "learning_rate": 9.354967855753804e-06,
      "loss": 0.0699,
      "step": 3328
    },
    {
      "epoch": 0.17190808159049834,
      "grad_norm": 12.442151069641113,
      "learning_rate": 9.354565249966414e-06,
      "loss": 0.1482,
      "step": 3329
    },
    {
      "epoch": 0.17195972114639815,
      "grad_norm": 7.467346668243408,
      "learning_rate": 9.354162527241195e-06,
      "loss": 0.0898,
      "step": 3330
    },
    {
      "epoch": 0.17201136070229797,
      "grad_norm": 5.384405612945557,
      "learning_rate": 9.353759687588968e-06,
      "loss": 0.054,
      "step": 3331
    },
    {
      "epoch": 0.1720630002581978,
      "grad_norm": 11.773791313171387,
      "learning_rate": 9.353356731020549e-06,
      "loss": 0.1318,
      "step": 3332
    },
    {
      "epoch": 0.1721146398140976,
      "grad_norm": 6.160914421081543,
      "learning_rate": 9.352953657546756e-06,
      "loss": 0.0713,
      "step": 3333
    },
    {
      "epoch": 0.17216627936999743,
      "grad_norm": 2.6562111377716064,
      "learning_rate": 9.352550467178416e-06,
      "loss": 0.0488,
      "step": 3334
    },
    {
      "epoch": 0.17221791892589725,
      "grad_norm": 3.4505112171173096,
      "learning_rate": 9.352147159926356e-06,
      "loss": 0.0412,
      "step": 3335
    },
    {
      "epoch": 0.17226955848179706,
      "grad_norm": 2.7850592136383057,
      "learning_rate": 9.351743735801409e-06,
      "loss": 0.058,
      "step": 3336
    },
    {
      "epoch": 0.17232119803769688,
      "grad_norm": 8.373186111450195,
      "learning_rate": 9.351340194814404e-06,
      "loss": 0.0633,
      "step": 3337
    },
    {
      "epoch": 0.1723728375935967,
      "grad_norm": 4.366161346435547,
      "learning_rate": 9.35093653697618e-06,
      "loss": 0.0349,
      "step": 3338
    },
    {
      "epoch": 0.17242447714949652,
      "grad_norm": 9.143266677856445,
      "learning_rate": 9.350532762297578e-06,
      "loss": 0.0859,
      "step": 3339
    },
    {
      "epoch": 0.17247611670539634,
      "grad_norm": 6.464964389801025,
      "learning_rate": 9.350128870789437e-06,
      "loss": 0.0378,
      "step": 3340
    },
    {
      "epoch": 0.17252775626129616,
      "grad_norm": 8.075690269470215,
      "learning_rate": 9.349724862462606e-06,
      "loss": 0.1371,
      "step": 3341
    },
    {
      "epoch": 0.17257939581719597,
      "grad_norm": 7.891760349273682,
      "learning_rate": 9.349320737327935e-06,
      "loss": 0.0784,
      "step": 3342
    },
    {
      "epoch": 0.1726310353730958,
      "grad_norm": 8.91414737701416,
      "learning_rate": 9.348916495396274e-06,
      "loss": 0.1095,
      "step": 3343
    },
    {
      "epoch": 0.1726826749289956,
      "grad_norm": 3.579514741897583,
      "learning_rate": 9.348512136678481e-06,
      "loss": 0.0359,
      "step": 3344
    },
    {
      "epoch": 0.17273431448489543,
      "grad_norm": 15.027436256408691,
      "learning_rate": 9.348107661185415e-06,
      "loss": 0.1463,
      "step": 3345
    },
    {
      "epoch": 0.17278595404079525,
      "grad_norm": 3.920659303665161,
      "learning_rate": 9.347703068927933e-06,
      "loss": 0.0661,
      "step": 3346
    },
    {
      "epoch": 0.17283759359669507,
      "grad_norm": 21.50909423828125,
      "learning_rate": 9.347298359916905e-06,
      "loss": 0.221,
      "step": 3347
    },
    {
      "epoch": 0.17288923315259488,
      "grad_norm": 2.9050002098083496,
      "learning_rate": 9.3468935341632e-06,
      "loss": 0.0459,
      "step": 3348
    },
    {
      "epoch": 0.1729408727084947,
      "grad_norm": 14.970332145690918,
      "learning_rate": 9.346488591677682e-06,
      "loss": 0.0682,
      "step": 3349
    },
    {
      "epoch": 0.17299251226439452,
      "grad_norm": 11.131439208984375,
      "learning_rate": 9.346083532471233e-06,
      "loss": 0.1008,
      "step": 3350
    },
    {
      "epoch": 0.17304415182029434,
      "grad_norm": 12.217215538024902,
      "learning_rate": 9.345678356554727e-06,
      "loss": 0.1412,
      "step": 3351
    },
    {
      "epoch": 0.17309579137619416,
      "grad_norm": 8.293888092041016,
      "learning_rate": 9.345273063939044e-06,
      "loss": 0.0988,
      "step": 3352
    },
    {
      "epoch": 0.17314743093209398,
      "grad_norm": 7.187489986419678,
      "learning_rate": 9.34486765463507e-06,
      "loss": 0.1015,
      "step": 3353
    },
    {
      "epoch": 0.1731990704879938,
      "grad_norm": 5.084132671356201,
      "learning_rate": 9.34446212865369e-06,
      "loss": 0.0632,
      "step": 3354
    },
    {
      "epoch": 0.1732507100438936,
      "grad_norm": 2.781637668609619,
      "learning_rate": 9.344056486005796e-06,
      "loss": 0.0401,
      "step": 3355
    },
    {
      "epoch": 0.17330234959979343,
      "grad_norm": 11.919472694396973,
      "learning_rate": 9.34365072670228e-06,
      "loss": 0.1935,
      "step": 3356
    },
    {
      "epoch": 0.17335398915569325,
      "grad_norm": 3.6540801525115967,
      "learning_rate": 9.343244850754037e-06,
      "loss": 0.0507,
      "step": 3357
    },
    {
      "epoch": 0.17340562871159307,
      "grad_norm": 13.396814346313477,
      "learning_rate": 9.342838858171968e-06,
      "loss": 0.1411,
      "step": 3358
    },
    {
      "epoch": 0.1734572682674929,
      "grad_norm": 14.219197273254395,
      "learning_rate": 9.342432748966976e-06,
      "loss": 0.1397,
      "step": 3359
    },
    {
      "epoch": 0.1735089078233927,
      "grad_norm": 7.0265116691589355,
      "learning_rate": 9.342026523149965e-06,
      "loss": 0.1097,
      "step": 3360
    },
    {
      "epoch": 0.17356054737929255,
      "grad_norm": 5.535713195800781,
      "learning_rate": 9.341620180731845e-06,
      "loss": 0.0945,
      "step": 3361
    },
    {
      "epoch": 0.17361218693519237,
      "grad_norm": 9.048115730285645,
      "learning_rate": 9.341213721723528e-06,
      "loss": 0.0664,
      "step": 3362
    },
    {
      "epoch": 0.1736638264910922,
      "grad_norm": 13.20899486541748,
      "learning_rate": 9.34080714613593e-06,
      "loss": 0.1899,
      "step": 3363
    },
    {
      "epoch": 0.173715466046992,
      "grad_norm": 2.1622912883758545,
      "learning_rate": 9.340400453979966e-06,
      "loss": 0.0585,
      "step": 3364
    },
    {
      "epoch": 0.17376710560289182,
      "grad_norm": 4.3001813888549805,
      "learning_rate": 9.339993645266558e-06,
      "loss": 0.086,
      "step": 3365
    },
    {
      "epoch": 0.17381874515879164,
      "grad_norm": 9.12658977508545,
      "learning_rate": 9.339586720006634e-06,
      "loss": 0.0729,
      "step": 3366
    },
    {
      "epoch": 0.17387038471469146,
      "grad_norm": 3.289332151412964,
      "learning_rate": 9.339179678211118e-06,
      "loss": 0.0693,
      "step": 3367
    },
    {
      "epoch": 0.17392202427059128,
      "grad_norm": 7.900903701782227,
      "learning_rate": 9.338772519890943e-06,
      "loss": 0.1136,
      "step": 3368
    },
    {
      "epoch": 0.1739736638264911,
      "grad_norm": 5.130281448364258,
      "learning_rate": 9.338365245057045e-06,
      "loss": 0.0568,
      "step": 3369
    },
    {
      "epoch": 0.17402530338239092,
      "grad_norm": 3.160665988922119,
      "learning_rate": 9.337957853720354e-06,
      "loss": 0.0569,
      "step": 3370
    },
    {
      "epoch": 0.17407694293829074,
      "grad_norm": 6.257293224334717,
      "learning_rate": 9.337550345891815e-06,
      "loss": 0.0846,
      "step": 3371
    },
    {
      "epoch": 0.17412858249419055,
      "grad_norm": 16.88035011291504,
      "learning_rate": 9.337142721582371e-06,
      "loss": 0.1075,
      "step": 3372
    },
    {
      "epoch": 0.17418022205009037,
      "grad_norm": 14.129944801330566,
      "learning_rate": 9.336734980802966e-06,
      "loss": 0.1699,
      "step": 3373
    },
    {
      "epoch": 0.1742318616059902,
      "grad_norm": 14.705167770385742,
      "learning_rate": 9.336327123564553e-06,
      "loss": 0.1299,
      "step": 3374
    },
    {
      "epoch": 0.17428350116189,
      "grad_norm": 13.78802490234375,
      "learning_rate": 9.335919149878084e-06,
      "loss": 0.0985,
      "step": 3375
    },
    {
      "epoch": 0.17433514071778983,
      "grad_norm": 8.125528335571289,
      "learning_rate": 9.33551105975451e-06,
      "loss": 0.1158,
      "step": 3376
    },
    {
      "epoch": 0.17438678027368965,
      "grad_norm": 2.0162835121154785,
      "learning_rate": 9.335102853204795e-06,
      "loss": 0.0145,
      "step": 3377
    },
    {
      "epoch": 0.17443841982958946,
      "grad_norm": 2.5344607830047607,
      "learning_rate": 9.3346945302399e-06,
      "loss": 0.0364,
      "step": 3378
    },
    {
      "epoch": 0.17449005938548928,
      "grad_norm": 6.7923264503479,
      "learning_rate": 9.33428609087079e-06,
      "loss": 0.0632,
      "step": 3379
    },
    {
      "epoch": 0.1745416989413891,
      "grad_norm": 17.35548210144043,
      "learning_rate": 9.333877535108432e-06,
      "loss": 0.1227,
      "step": 3380
    },
    {
      "epoch": 0.17459333849728892,
      "grad_norm": 6.541072845458984,
      "learning_rate": 9.3334688629638e-06,
      "loss": 0.108,
      "step": 3381
    },
    {
      "epoch": 0.17464497805318874,
      "grad_norm": 4.672735214233398,
      "learning_rate": 9.333060074447865e-06,
      "loss": 0.0534,
      "step": 3382
    },
    {
      "epoch": 0.17469661760908856,
      "grad_norm": 10.267979621887207,
      "learning_rate": 9.332651169571606e-06,
      "loss": 0.1709,
      "step": 3383
    },
    {
      "epoch": 0.17474825716498837,
      "grad_norm": 11.401556968688965,
      "learning_rate": 9.332242148346005e-06,
      "loss": 0.0647,
      "step": 3384
    },
    {
      "epoch": 0.1747998967208882,
      "grad_norm": 7.57285737991333,
      "learning_rate": 9.331833010782045e-06,
      "loss": 0.1025,
      "step": 3385
    },
    {
      "epoch": 0.174851536276788,
      "grad_norm": 4.338389873504639,
      "learning_rate": 9.331423756890715e-06,
      "loss": 0.0382,
      "step": 3386
    },
    {
      "epoch": 0.17490317583268783,
      "grad_norm": 6.578561305999756,
      "learning_rate": 9.331014386683e-06,
      "loss": 0.0767,
      "step": 3387
    },
    {
      "epoch": 0.17495481538858765,
      "grad_norm": 5.378432273864746,
      "learning_rate": 9.330604900169897e-06,
      "loss": 0.0666,
      "step": 3388
    },
    {
      "epoch": 0.17500645494448747,
      "grad_norm": 8.700605392456055,
      "learning_rate": 9.330195297362402e-06,
      "loss": 0.0688,
      "step": 3389
    },
    {
      "epoch": 0.17505809450038728,
      "grad_norm": 7.1835126876831055,
      "learning_rate": 9.329785578271516e-06,
      "loss": 0.0481,
      "step": 3390
    },
    {
      "epoch": 0.1751097340562871,
      "grad_norm": 6.610920429229736,
      "learning_rate": 9.32937574290824e-06,
      "loss": 0.0516,
      "step": 3391
    },
    {
      "epoch": 0.17516137361218692,
      "grad_norm": 3.6584742069244385,
      "learning_rate": 9.328965791283579e-06,
      "loss": 0.055,
      "step": 3392
    },
    {
      "epoch": 0.17521301316808677,
      "grad_norm": 6.1204938888549805,
      "learning_rate": 9.328555723408542e-06,
      "loss": 0.0635,
      "step": 3393
    },
    {
      "epoch": 0.17526465272398659,
      "grad_norm": 3.699693202972412,
      "learning_rate": 9.328145539294142e-06,
      "loss": 0.0413,
      "step": 3394
    },
    {
      "epoch": 0.1753162922798864,
      "grad_norm": 9.161226272583008,
      "learning_rate": 9.327735238951394e-06,
      "loss": 0.0643,
      "step": 3395
    },
    {
      "epoch": 0.17536793183578622,
      "grad_norm": 5.871004104614258,
      "learning_rate": 9.327324822391314e-06,
      "loss": 0.0532,
      "step": 3396
    },
    {
      "epoch": 0.17541957139168604,
      "grad_norm": 7.361422538757324,
      "learning_rate": 9.326914289624929e-06,
      "loss": 0.0753,
      "step": 3397
    },
    {
      "epoch": 0.17547121094758586,
      "grad_norm": 4.2078328132629395,
      "learning_rate": 9.326503640663257e-06,
      "loss": 0.0412,
      "step": 3398
    },
    {
      "epoch": 0.17552285050348568,
      "grad_norm": 4.510331153869629,
      "learning_rate": 9.32609287551733e-06,
      "loss": 0.0704,
      "step": 3399
    },
    {
      "epoch": 0.1755744900593855,
      "grad_norm": 3.4027669429779053,
      "learning_rate": 9.325681994198177e-06,
      "loss": 0.04,
      "step": 3400
    },
    {
      "epoch": 0.17562612961528531,
      "grad_norm": 20.030467987060547,
      "learning_rate": 9.32527099671683e-06,
      "loss": 0.0977,
      "step": 3401
    },
    {
      "epoch": 0.17567776917118513,
      "grad_norm": 21.618345260620117,
      "learning_rate": 9.32485988308433e-06,
      "loss": 0.2403,
      "step": 3402
    },
    {
      "epoch": 0.17572940872708495,
      "grad_norm": 7.255788803100586,
      "learning_rate": 9.324448653311713e-06,
      "loss": 0.0612,
      "step": 3403
    },
    {
      "epoch": 0.17578104828298477,
      "grad_norm": 3.5693273544311523,
      "learning_rate": 9.324037307410026e-06,
      "loss": 0.0947,
      "step": 3404
    },
    {
      "epoch": 0.1758326878388846,
      "grad_norm": 26.724443435668945,
      "learning_rate": 9.323625845390313e-06,
      "loss": 0.0976,
      "step": 3405
    },
    {
      "epoch": 0.1758843273947844,
      "grad_norm": 3.4819483757019043,
      "learning_rate": 9.323214267263624e-06,
      "loss": 0.0611,
      "step": 3406
    },
    {
      "epoch": 0.17593596695068422,
      "grad_norm": 9.416197776794434,
      "learning_rate": 9.322802573041011e-06,
      "loss": 0.0532,
      "step": 3407
    },
    {
      "epoch": 0.17598760650658404,
      "grad_norm": 8.171070098876953,
      "learning_rate": 9.322390762733531e-06,
      "loss": 0.0839,
      "step": 3408
    },
    {
      "epoch": 0.17603924606248386,
      "grad_norm": 7.785849094390869,
      "learning_rate": 9.321978836352241e-06,
      "loss": 0.1125,
      "step": 3409
    },
    {
      "epoch": 0.17609088561838368,
      "grad_norm": 16.01744842529297,
      "learning_rate": 9.321566793908204e-06,
      "loss": 0.175,
      "step": 3410
    },
    {
      "epoch": 0.1761425251742835,
      "grad_norm": 3.1782820224761963,
      "learning_rate": 9.321154635412486e-06,
      "loss": 0.0619,
      "step": 3411
    },
    {
      "epoch": 0.17619416473018332,
      "grad_norm": 12.083925247192383,
      "learning_rate": 9.320742360876153e-06,
      "loss": 0.1042,
      "step": 3412
    },
    {
      "epoch": 0.17624580428608314,
      "grad_norm": 5.558353900909424,
      "learning_rate": 9.320329970310277e-06,
      "loss": 0.0925,
      "step": 3413
    },
    {
      "epoch": 0.17629744384198295,
      "grad_norm": 6.747273921966553,
      "learning_rate": 9.319917463725932e-06,
      "loss": 0.0683,
      "step": 3414
    },
    {
      "epoch": 0.17634908339788277,
      "grad_norm": 3.41403865814209,
      "learning_rate": 9.319504841134197e-06,
      "loss": 0.036,
      "step": 3415
    },
    {
      "epoch": 0.1764007229537826,
      "grad_norm": 8.239133834838867,
      "learning_rate": 9.319092102546152e-06,
      "loss": 0.0757,
      "step": 3416
    },
    {
      "epoch": 0.1764523625096824,
      "grad_norm": 14.428853988647461,
      "learning_rate": 9.31867924797288e-06,
      "loss": 0.1208,
      "step": 3417
    },
    {
      "epoch": 0.17650400206558223,
      "grad_norm": 4.498054504394531,
      "learning_rate": 9.318266277425468e-06,
      "loss": 0.0553,
      "step": 3418
    },
    {
      "epoch": 0.17655564162148205,
      "grad_norm": 2.7267065048217773,
      "learning_rate": 9.317853190915006e-06,
      "loss": 0.0499,
      "step": 3419
    },
    {
      "epoch": 0.17660728117738186,
      "grad_norm": 3.9768600463867188,
      "learning_rate": 9.317439988452589e-06,
      "loss": 0.0776,
      "step": 3420
    },
    {
      "epoch": 0.17665892073328168,
      "grad_norm": 5.927595138549805,
      "learning_rate": 9.31702667004931e-06,
      "loss": 0.0726,
      "step": 3421
    },
    {
      "epoch": 0.1767105602891815,
      "grad_norm": 3.9643895626068115,
      "learning_rate": 9.31661323571627e-06,
      "loss": 0.0616,
      "step": 3422
    },
    {
      "epoch": 0.17676219984508132,
      "grad_norm": 5.612739086151123,
      "learning_rate": 9.316199685464572e-06,
      "loss": 0.1039,
      "step": 3423
    },
    {
      "epoch": 0.17681383940098117,
      "grad_norm": 2.161508560180664,
      "learning_rate": 9.315786019305317e-06,
      "loss": 0.0546,
      "step": 3424
    },
    {
      "epoch": 0.17686547895688098,
      "grad_norm": 6.825375080108643,
      "learning_rate": 9.31537223724962e-06,
      "loss": 0.0698,
      "step": 3425
    },
    {
      "epoch": 0.1769171185127808,
      "grad_norm": 3.899353504180908,
      "learning_rate": 9.314958339308591e-06,
      "loss": 0.108,
      "step": 3426
    },
    {
      "epoch": 0.17696875806868062,
      "grad_norm": 3.952183485031128,
      "learning_rate": 9.314544325493343e-06,
      "loss": 0.0508,
      "step": 3427
    },
    {
      "epoch": 0.17702039762458044,
      "grad_norm": 6.788403034210205,
      "learning_rate": 9.314130195814993e-06,
      "loss": 0.025,
      "step": 3428
    },
    {
      "epoch": 0.17707203718048026,
      "grad_norm": 5.049844264984131,
      "learning_rate": 9.313715950284666e-06,
      "loss": 0.0671,
      "step": 3429
    },
    {
      "epoch": 0.17712367673638008,
      "grad_norm": 3.3039791584014893,
      "learning_rate": 9.31330158891348e-06,
      "loss": 0.0344,
      "step": 3430
    },
    {
      "epoch": 0.1771753162922799,
      "grad_norm": 4.3340630531311035,
      "learning_rate": 9.312887111712571e-06,
      "loss": 0.0616,
      "step": 3431
    },
    {
      "epoch": 0.1772269558481797,
      "grad_norm": 3.96637225151062,
      "learning_rate": 9.312472518693063e-06,
      "loss": 0.0493,
      "step": 3432
    },
    {
      "epoch": 0.17727859540407953,
      "grad_norm": 6.012303352355957,
      "learning_rate": 9.312057809866091e-06,
      "loss": 0.0782,
      "step": 3433
    },
    {
      "epoch": 0.17733023495997935,
      "grad_norm": 7.280198574066162,
      "learning_rate": 9.311642985242792e-06,
      "loss": 0.1455,
      "step": 3434
    },
    {
      "epoch": 0.17738187451587917,
      "grad_norm": 8.980958938598633,
      "learning_rate": 9.311228044834306e-06,
      "loss": 0.1257,
      "step": 3435
    },
    {
      "epoch": 0.17743351407177899,
      "grad_norm": 9.875679016113281,
      "learning_rate": 9.310812988651774e-06,
      "loss": 0.138,
      "step": 3436
    },
    {
      "epoch": 0.1774851536276788,
      "grad_norm": 8.549158096313477,
      "learning_rate": 9.310397816706345e-06,
      "loss": 0.1234,
      "step": 3437
    },
    {
      "epoch": 0.17753679318357862,
      "grad_norm": 4.683456897735596,
      "learning_rate": 9.309982529009165e-06,
      "loss": 0.0816,
      "step": 3438
    },
    {
      "epoch": 0.17758843273947844,
      "grad_norm": 9.446216583251953,
      "learning_rate": 9.309567125571389e-06,
      "loss": 0.0752,
      "step": 3439
    },
    {
      "epoch": 0.17764007229537826,
      "grad_norm": 10.556962013244629,
      "learning_rate": 9.30915160640417e-06,
      "loss": 0.0595,
      "step": 3440
    },
    {
      "epoch": 0.17769171185127808,
      "grad_norm": 4.684728145599365,
      "learning_rate": 9.308735971518668e-06,
      "loss": 0.1062,
      "step": 3441
    },
    {
      "epoch": 0.1777433514071779,
      "grad_norm": 10.33504867553711,
      "learning_rate": 9.308320220926042e-06,
      "loss": 0.0932,
      "step": 3442
    },
    {
      "epoch": 0.17779499096307771,
      "grad_norm": 13.335216522216797,
      "learning_rate": 9.30790435463746e-06,
      "loss": 0.1222,
      "step": 3443
    },
    {
      "epoch": 0.17784663051897753,
      "grad_norm": 7.441825866699219,
      "learning_rate": 9.307488372664087e-06,
      "loss": 0.0979,
      "step": 3444
    },
    {
      "epoch": 0.17789827007487735,
      "grad_norm": 10.85234546661377,
      "learning_rate": 9.307072275017096e-06,
      "loss": 0.1097,
      "step": 3445
    },
    {
      "epoch": 0.17794990963077717,
      "grad_norm": 4.898049831390381,
      "learning_rate": 9.30665606170766e-06,
      "loss": 0.0506,
      "step": 3446
    },
    {
      "epoch": 0.178001549186677,
      "grad_norm": 10.403362274169922,
      "learning_rate": 9.306239732746953e-06,
      "loss": 0.1367,
      "step": 3447
    },
    {
      "epoch": 0.1780531887425768,
      "grad_norm": 7.355813503265381,
      "learning_rate": 9.305823288146161e-06,
      "loss": 0.0667,
      "step": 3448
    },
    {
      "epoch": 0.17810482829847663,
      "grad_norm": 1.6307944059371948,
      "learning_rate": 9.305406727916463e-06,
      "loss": 0.0389,
      "step": 3449
    },
    {
      "epoch": 0.17815646785437644,
      "grad_norm": 2.491417407989502,
      "learning_rate": 9.304990052069046e-06,
      "loss": 0.0515,
      "step": 3450
    },
    {
      "epoch": 0.17820810741027626,
      "grad_norm": 10.781699180603027,
      "learning_rate": 9.3045732606151e-06,
      "loss": 0.0522,
      "step": 3451
    },
    {
      "epoch": 0.17825974696617608,
      "grad_norm": 1.4547792673110962,
      "learning_rate": 9.304156353565819e-06,
      "loss": 0.0283,
      "step": 3452
    },
    {
      "epoch": 0.1783113865220759,
      "grad_norm": 2.7505712509155273,
      "learning_rate": 9.303739330932395e-06,
      "loss": 0.0369,
      "step": 3453
    },
    {
      "epoch": 0.17836302607797572,
      "grad_norm": 4.123477935791016,
      "learning_rate": 9.30332219272603e-06,
      "loss": 0.0374,
      "step": 3454
    },
    {
      "epoch": 0.17841466563387554,
      "grad_norm": 8.227118492126465,
      "learning_rate": 9.302904938957924e-06,
      "loss": 0.0727,
      "step": 3455
    },
    {
      "epoch": 0.17846630518977538,
      "grad_norm": 4.6827921867370605,
      "learning_rate": 9.302487569639285e-06,
      "loss": 0.0521,
      "step": 3456
    },
    {
      "epoch": 0.1785179447456752,
      "grad_norm": 6.127108097076416,
      "learning_rate": 9.302070084781316e-06,
      "loss": 0.0625,
      "step": 3457
    },
    {
      "epoch": 0.17856958430157502,
      "grad_norm": 8.703375816345215,
      "learning_rate": 9.301652484395232e-06,
      "loss": 0.0601,
      "step": 3458
    },
    {
      "epoch": 0.17862122385747484,
      "grad_norm": 11.605361938476562,
      "learning_rate": 9.301234768492247e-06,
      "loss": 0.0978,
      "step": 3459
    },
    {
      "epoch": 0.17867286341337466,
      "grad_norm": 14.220112800598145,
      "learning_rate": 9.300816937083576e-06,
      "loss": 0.1376,
      "step": 3460
    },
    {
      "epoch": 0.17872450296927447,
      "grad_norm": 17.37699317932129,
      "learning_rate": 9.300398990180442e-06,
      "loss": 0.0318,
      "step": 3461
    },
    {
      "epoch": 0.1787761425251743,
      "grad_norm": 4.780977725982666,
      "learning_rate": 9.299980927794067e-06,
      "loss": 0.0498,
      "step": 3462
    },
    {
      "epoch": 0.1788277820810741,
      "grad_norm": 10.492838859558105,
      "learning_rate": 9.29956274993568e-06,
      "loss": 0.1337,
      "step": 3463
    },
    {
      "epoch": 0.17887942163697393,
      "grad_norm": 11.502586364746094,
      "learning_rate": 9.29914445661651e-06,
      "loss": 0.113,
      "step": 3464
    },
    {
      "epoch": 0.17893106119287375,
      "grad_norm": 6.389218330383301,
      "learning_rate": 9.298726047847786e-06,
      "loss": 0.0388,
      "step": 3465
    },
    {
      "epoch": 0.17898270074877357,
      "grad_norm": 3.6324868202209473,
      "learning_rate": 9.298307523640748e-06,
      "loss": 0.0456,
      "step": 3466
    },
    {
      "epoch": 0.17903434030467338,
      "grad_norm": 5.635067462921143,
      "learning_rate": 9.297888884006633e-06,
      "loss": 0.0793,
      "step": 3467
    },
    {
      "epoch": 0.1790859798605732,
      "grad_norm": 22.78287696838379,
      "learning_rate": 9.297470128956685e-06,
      "loss": 0.0864,
      "step": 3468
    },
    {
      "epoch": 0.17913761941647302,
      "grad_norm": 4.908595561981201,
      "learning_rate": 9.29705125850215e-06,
      "loss": 0.0627,
      "step": 3469
    },
    {
      "epoch": 0.17918925897237284,
      "grad_norm": 8.422218322753906,
      "learning_rate": 9.296632272654274e-06,
      "loss": 0.1129,
      "step": 3470
    },
    {
      "epoch": 0.17924089852827266,
      "grad_norm": 15.06112003326416,
      "learning_rate": 9.296213171424308e-06,
      "loss": 0.1112,
      "step": 3471
    },
    {
      "epoch": 0.17929253808417248,
      "grad_norm": 16.229887008666992,
      "learning_rate": 9.295793954823512e-06,
      "loss": 0.1342,
      "step": 3472
    },
    {
      "epoch": 0.1793441776400723,
      "grad_norm": 6.28723669052124,
      "learning_rate": 9.295374622863135e-06,
      "loss": 0.0781,
      "step": 3473
    },
    {
      "epoch": 0.1793958171959721,
      "grad_norm": 4.05713415145874,
      "learning_rate": 9.294955175554443e-06,
      "loss": 0.0893,
      "step": 3474
    },
    {
      "epoch": 0.17944745675187193,
      "grad_norm": 5.060600757598877,
      "learning_rate": 9.294535612908702e-06,
      "loss": 0.0636,
      "step": 3475
    },
    {
      "epoch": 0.17949909630777175,
      "grad_norm": 2.242914915084839,
      "learning_rate": 9.294115934937173e-06,
      "loss": 0.0415,
      "step": 3476
    },
    {
      "epoch": 0.17955073586367157,
      "grad_norm": 3.1175599098205566,
      "learning_rate": 9.293696141651132e-06,
      "loss": 0.0436,
      "step": 3477
    },
    {
      "epoch": 0.1796023754195714,
      "grad_norm": 2.965027332305908,
      "learning_rate": 9.293276233061848e-06,
      "loss": 0.0398,
      "step": 3478
    },
    {
      "epoch": 0.1796540149754712,
      "grad_norm": 3.399204730987549,
      "learning_rate": 9.2928562091806e-06,
      "loss": 0.0635,
      "step": 3479
    },
    {
      "epoch": 0.17970565453137102,
      "grad_norm": 4.800698280334473,
      "learning_rate": 9.292436070018665e-06,
      "loss": 0.0551,
      "step": 3480
    },
    {
      "epoch": 0.17975729408727084,
      "grad_norm": 4.070358753204346,
      "learning_rate": 9.292015815587326e-06,
      "loss": 0.0658,
      "step": 3481
    },
    {
      "epoch": 0.17980893364317066,
      "grad_norm": 8.00658130645752,
      "learning_rate": 9.29159544589787e-06,
      "loss": 0.0498,
      "step": 3482
    },
    {
      "epoch": 0.17986057319907048,
      "grad_norm": 5.652313709259033,
      "learning_rate": 9.291174960961585e-06,
      "loss": 0.1072,
      "step": 3483
    },
    {
      "epoch": 0.1799122127549703,
      "grad_norm": 5.28953218460083,
      "learning_rate": 9.290754360789762e-06,
      "loss": 0.0778,
      "step": 3484
    },
    {
      "epoch": 0.17996385231087011,
      "grad_norm": 5.229704856872559,
      "learning_rate": 9.290333645393696e-06,
      "loss": 0.0698,
      "step": 3485
    },
    {
      "epoch": 0.18001549186676993,
      "grad_norm": 8.93969440460205,
      "learning_rate": 9.289912814784685e-06,
      "loss": 0.1026,
      "step": 3486
    },
    {
      "epoch": 0.18006713142266975,
      "grad_norm": 3.598127603530884,
      "learning_rate": 9.28949186897403e-06,
      "loss": 0.0713,
      "step": 3487
    },
    {
      "epoch": 0.1801187709785696,
      "grad_norm": 2.288740873336792,
      "learning_rate": 9.289070807973036e-06,
      "loss": 0.0463,
      "step": 3488
    },
    {
      "epoch": 0.18017041053446942,
      "grad_norm": 8.083279609680176,
      "learning_rate": 9.28864963179301e-06,
      "loss": 0.0902,
      "step": 3489
    },
    {
      "epoch": 0.18022205009036923,
      "grad_norm": 19.89106559753418,
      "learning_rate": 9.28822834044526e-06,
      "loss": 0.1103,
      "step": 3490
    },
    {
      "epoch": 0.18027368964626905,
      "grad_norm": 9.868852615356445,
      "learning_rate": 9.287806933941103e-06,
      "loss": 0.0826,
      "step": 3491
    },
    {
      "epoch": 0.18032532920216887,
      "grad_norm": 9.373196601867676,
      "learning_rate": 9.287385412291853e-06,
      "loss": 0.0609,
      "step": 3492
    },
    {
      "epoch": 0.1803769687580687,
      "grad_norm": 2.498621940612793,
      "learning_rate": 9.286963775508831e-06,
      "loss": 0.0712,
      "step": 3493
    },
    {
      "epoch": 0.1804286083139685,
      "grad_norm": 6.910489559173584,
      "learning_rate": 9.286542023603357e-06,
      "loss": 0.1117,
      "step": 3494
    },
    {
      "epoch": 0.18048024786986833,
      "grad_norm": 3.1932806968688965,
      "learning_rate": 9.286120156586762e-06,
      "loss": 0.027,
      "step": 3495
    },
    {
      "epoch": 0.18053188742576814,
      "grad_norm": 14.75586986541748,
      "learning_rate": 9.285698174470369e-06,
      "loss": 0.1246,
      "step": 3496
    },
    {
      "epoch": 0.18058352698166796,
      "grad_norm": 6.316192150115967,
      "learning_rate": 9.285276077265512e-06,
      "loss": 0.0665,
      "step": 3497
    },
    {
      "epoch": 0.18063516653756778,
      "grad_norm": 2.4650793075561523,
      "learning_rate": 9.284853864983528e-06,
      "loss": 0.0498,
      "step": 3498
    },
    {
      "epoch": 0.1806868060934676,
      "grad_norm": 4.956244468688965,
      "learning_rate": 9.284431537635754e-06,
      "loss": 0.0595,
      "step": 3499
    },
    {
      "epoch": 0.18073844564936742,
      "grad_norm": 22.014957427978516,
      "learning_rate": 9.28400909523353e-06,
      "loss": 0.2557,
      "step": 3500
    },
    {
      "epoch": 0.18079008520526724,
      "grad_norm": 2.9285430908203125,
      "learning_rate": 9.2835865377882e-06,
      "loss": 0.0382,
      "step": 3501
    },
    {
      "epoch": 0.18084172476116706,
      "grad_norm": 5.411451816558838,
      "learning_rate": 9.283163865311115e-06,
      "loss": 0.0927,
      "step": 3502
    },
    {
      "epoch": 0.18089336431706687,
      "grad_norm": 17.15008544921875,
      "learning_rate": 9.282741077813622e-06,
      "loss": 0.0852,
      "step": 3503
    },
    {
      "epoch": 0.1809450038729667,
      "grad_norm": 3.629364252090454,
      "learning_rate": 9.282318175307075e-06,
      "loss": 0.0359,
      "step": 3504
    },
    {
      "epoch": 0.1809966434288665,
      "grad_norm": 3.777277946472168,
      "learning_rate": 9.281895157802832e-06,
      "loss": 0.06,
      "step": 3505
    },
    {
      "epoch": 0.18104828298476633,
      "grad_norm": 6.2034525871276855,
      "learning_rate": 9.28147202531225e-06,
      "loss": 0.1287,
      "step": 3506
    },
    {
      "epoch": 0.18109992254066615,
      "grad_norm": 7.418347358703613,
      "learning_rate": 9.281048777846696e-06,
      "loss": 0.0729,
      "step": 3507
    },
    {
      "epoch": 0.18115156209656597,
      "grad_norm": 10.191624641418457,
      "learning_rate": 9.280625415417533e-06,
      "loss": 0.1021,
      "step": 3508
    },
    {
      "epoch": 0.18120320165246578,
      "grad_norm": 5.356970310211182,
      "learning_rate": 9.280201938036132e-06,
      "loss": 0.0989,
      "step": 3509
    },
    {
      "epoch": 0.1812548412083656,
      "grad_norm": 9.907130241394043,
      "learning_rate": 9.279778345713862e-06,
      "loss": 0.1279,
      "step": 3510
    },
    {
      "epoch": 0.18130648076426542,
      "grad_norm": 3.1734347343444824,
      "learning_rate": 9.2793546384621e-06,
      "loss": 0.0477,
      "step": 3511
    },
    {
      "epoch": 0.18135812032016524,
      "grad_norm": 1.6068748235702515,
      "learning_rate": 9.278930816292225e-06,
      "loss": 0.0384,
      "step": 3512
    },
    {
      "epoch": 0.18140975987606506,
      "grad_norm": 3.2450337409973145,
      "learning_rate": 9.278506879215616e-06,
      "loss": 0.0547,
      "step": 3513
    },
    {
      "epoch": 0.18146139943196488,
      "grad_norm": 23.511335372924805,
      "learning_rate": 9.278082827243662e-06,
      "loss": 0.1571,
      "step": 3514
    },
    {
      "epoch": 0.1815130389878647,
      "grad_norm": 11.353089332580566,
      "learning_rate": 9.277658660387744e-06,
      "loss": 0.201,
      "step": 3515
    },
    {
      "epoch": 0.1815646785437645,
      "grad_norm": 2.8511152267456055,
      "learning_rate": 9.277234378659258e-06,
      "loss": 0.0501,
      "step": 3516
    },
    {
      "epoch": 0.18161631809966433,
      "grad_norm": 3.6543538570404053,
      "learning_rate": 9.276809982069596e-06,
      "loss": 0.093,
      "step": 3517
    },
    {
      "epoch": 0.18166795765556415,
      "grad_norm": 2.613847494125366,
      "learning_rate": 9.276385470630156e-06,
      "loss": 0.0443,
      "step": 3518
    },
    {
      "epoch": 0.18171959721146397,
      "grad_norm": 6.311770439147949,
      "learning_rate": 9.275960844352337e-06,
      "loss": 0.0736,
      "step": 3519
    },
    {
      "epoch": 0.18177123676736381,
      "grad_norm": 6.363191604614258,
      "learning_rate": 9.275536103247537e-06,
      "loss": 0.0962,
      "step": 3520
    },
    {
      "epoch": 0.18182287632326363,
      "grad_norm": 12.196622848510742,
      "learning_rate": 9.275111247327172e-06,
      "loss": 0.0875,
      "step": 3521
    },
    {
      "epoch": 0.18187451587916345,
      "grad_norm": 6.6168036460876465,
      "learning_rate": 9.274686276602642e-06,
      "loss": 0.0775,
      "step": 3522
    },
    {
      "epoch": 0.18192615543506327,
      "grad_norm": 3.9448766708374023,
      "learning_rate": 9.274261191085363e-06,
      "loss": 0.0528,
      "step": 3523
    },
    {
      "epoch": 0.1819777949909631,
      "grad_norm": 2.4265658855438232,
      "learning_rate": 9.27383599078675e-06,
      "loss": 0.0335,
      "step": 3524
    },
    {
      "epoch": 0.1820294345468629,
      "grad_norm": 7.277225971221924,
      "learning_rate": 9.273410675718224e-06,
      "loss": 0.0532,
      "step": 3525
    },
    {
      "epoch": 0.18208107410276272,
      "grad_norm": 14.54513931274414,
      "learning_rate": 9.272985245891201e-06,
      "loss": 0.1881,
      "step": 3526
    },
    {
      "epoch": 0.18213271365866254,
      "grad_norm": 5.267324447631836,
      "learning_rate": 9.272559701317112e-06,
      "loss": 0.077,
      "step": 3527
    },
    {
      "epoch": 0.18218435321456236,
      "grad_norm": 5.413528919219971,
      "learning_rate": 9.272134042007376e-06,
      "loss": 0.0799,
      "step": 3528
    },
    {
      "epoch": 0.18223599277046218,
      "grad_norm": 9.278605461120605,
      "learning_rate": 9.271708267973433e-06,
      "loss": 0.0866,
      "step": 3529
    },
    {
      "epoch": 0.182287632326362,
      "grad_norm": 4.7513861656188965,
      "learning_rate": 9.271282379226711e-06,
      "loss": 0.0802,
      "step": 3530
    },
    {
      "epoch": 0.18233927188226182,
      "grad_norm": 3.678894519805908,
      "learning_rate": 9.270856375778648e-06,
      "loss": 0.0953,
      "step": 3531
    },
    {
      "epoch": 0.18239091143816163,
      "grad_norm": 4.739148139953613,
      "learning_rate": 9.270430257640684e-06,
      "loss": 0.0871,
      "step": 3532
    },
    {
      "epoch": 0.18244255099406145,
      "grad_norm": 5.783642292022705,
      "learning_rate": 9.270004024824264e-06,
      "loss": 0.1121,
      "step": 3533
    },
    {
      "epoch": 0.18249419054996127,
      "grad_norm": 12.143136978149414,
      "learning_rate": 9.26957767734083e-06,
      "loss": 0.1519,
      "step": 3534
    },
    {
      "epoch": 0.1825458301058611,
      "grad_norm": 7.855746746063232,
      "learning_rate": 9.269151215201834e-06,
      "loss": 0.1006,
      "step": 3535
    },
    {
      "epoch": 0.1825974696617609,
      "grad_norm": 3.9764463901519775,
      "learning_rate": 9.26872463841873e-06,
      "loss": 0.0731,
      "step": 3536
    },
    {
      "epoch": 0.18264910921766073,
      "grad_norm": 2.860513687133789,
      "learning_rate": 9.268297947002968e-06,
      "loss": 0.0616,
      "step": 3537
    },
    {
      "epoch": 0.18270074877356055,
      "grad_norm": 4.737343788146973,
      "learning_rate": 9.267871140966012e-06,
      "loss": 0.0667,
      "step": 3538
    },
    {
      "epoch": 0.18275238832946036,
      "grad_norm": 1.8835021257400513,
      "learning_rate": 9.26744422031932e-06,
      "loss": 0.0567,
      "step": 3539
    },
    {
      "epoch": 0.18280402788536018,
      "grad_norm": 4.931168556213379,
      "learning_rate": 9.267017185074357e-06,
      "loss": 0.0975,
      "step": 3540
    },
    {
      "epoch": 0.18285566744126,
      "grad_norm": 3.4784531593322754,
      "learning_rate": 9.266590035242593e-06,
      "loss": 0.0508,
      "step": 3541
    },
    {
      "epoch": 0.18290730699715982,
      "grad_norm": 6.6915483474731445,
      "learning_rate": 9.266162770835494e-06,
      "loss": 0.0729,
      "step": 3542
    },
    {
      "epoch": 0.18295894655305964,
      "grad_norm": 9.199048042297363,
      "learning_rate": 9.265735391864539e-06,
      "loss": 0.1027,
      "step": 3543
    },
    {
      "epoch": 0.18301058610895946,
      "grad_norm": 10.530640602111816,
      "learning_rate": 9.265307898341202e-06,
      "loss": 0.1325,
      "step": 3544
    },
    {
      "epoch": 0.18306222566485927,
      "grad_norm": 4.533668518066406,
      "learning_rate": 9.264880290276963e-06,
      "loss": 0.0535,
      "step": 3545
    },
    {
      "epoch": 0.1831138652207591,
      "grad_norm": 2.6721079349517822,
      "learning_rate": 9.264452567683307e-06,
      "loss": 0.047,
      "step": 3546
    },
    {
      "epoch": 0.1831655047766589,
      "grad_norm": 7.600574016571045,
      "learning_rate": 9.264024730571716e-06,
      "loss": 0.0609,
      "step": 3547
    },
    {
      "epoch": 0.18321714433255873,
      "grad_norm": 5.476165294647217,
      "learning_rate": 9.263596778953683e-06,
      "loss": 0.0808,
      "step": 3548
    },
    {
      "epoch": 0.18326878388845855,
      "grad_norm": 6.256567478179932,
      "learning_rate": 9.2631687128407e-06,
      "loss": 0.0997,
      "step": 3549
    },
    {
      "epoch": 0.18332042344435837,
      "grad_norm": 4.596685409545898,
      "learning_rate": 9.262740532244262e-06,
      "loss": 0.0679,
      "step": 3550
    },
    {
      "epoch": 0.18337206300025818,
      "grad_norm": 5.443218231201172,
      "learning_rate": 9.262312237175865e-06,
      "loss": 0.0474,
      "step": 3551
    },
    {
      "epoch": 0.18342370255615803,
      "grad_norm": 11.869162559509277,
      "learning_rate": 9.261883827647011e-06,
      "loss": 0.0778,
      "step": 3552
    },
    {
      "epoch": 0.18347534211205785,
      "grad_norm": 2.82771372795105,
      "learning_rate": 9.261455303669209e-06,
      "loss": 0.0495,
      "step": 3553
    },
    {
      "epoch": 0.18352698166795767,
      "grad_norm": 12.022967338562012,
      "learning_rate": 9.26102666525396e-06,
      "loss": 0.1434,
      "step": 3554
    },
    {
      "epoch": 0.18357862122385749,
      "grad_norm": 3.2618677616119385,
      "learning_rate": 9.26059791241278e-06,
      "loss": 0.0189,
      "step": 3555
    },
    {
      "epoch": 0.1836302607797573,
      "grad_norm": 5.643316745758057,
      "learning_rate": 9.260169045157181e-06,
      "loss": 0.0639,
      "step": 3556
    },
    {
      "epoch": 0.18368190033565712,
      "grad_norm": 3.95703387260437,
      "learning_rate": 9.259740063498679e-06,
      "loss": 0.0849,
      "step": 3557
    },
    {
      "epoch": 0.18373353989155694,
      "grad_norm": 4.1639018058776855,
      "learning_rate": 9.259310967448796e-06,
      "loss": 0.0717,
      "step": 3558
    },
    {
      "epoch": 0.18378517944745676,
      "grad_norm": 5.79049825668335,
      "learning_rate": 9.25888175701905e-06,
      "loss": 0.0805,
      "step": 3559
    },
    {
      "epoch": 0.18383681900335658,
      "grad_norm": 6.860091686248779,
      "learning_rate": 9.258452432220974e-06,
      "loss": 0.0429,
      "step": 3560
    },
    {
      "epoch": 0.1838884585592564,
      "grad_norm": 2.916564702987671,
      "learning_rate": 9.258022993066093e-06,
      "loss": 0.0549,
      "step": 3561
    },
    {
      "epoch": 0.18394009811515621,
      "grad_norm": 16.33740234375,
      "learning_rate": 9.257593439565939e-06,
      "loss": 0.129,
      "step": 3562
    },
    {
      "epoch": 0.18399173767105603,
      "grad_norm": 7.774988174438477,
      "learning_rate": 9.257163771732049e-06,
      "loss": 0.0684,
      "step": 3563
    },
    {
      "epoch": 0.18404337722695585,
      "grad_norm": 9.568811416625977,
      "learning_rate": 9.25673398957596e-06,
      "loss": 0.139,
      "step": 3564
    },
    {
      "epoch": 0.18409501678285567,
      "grad_norm": 3.0474565029144287,
      "learning_rate": 9.256304093109215e-06,
      "loss": 0.0349,
      "step": 3565
    },
    {
      "epoch": 0.1841466563387555,
      "grad_norm": 1.2550104856491089,
      "learning_rate": 9.255874082343355e-06,
      "loss": 0.0161,
      "step": 3566
    },
    {
      "epoch": 0.1841982958946553,
      "grad_norm": 7.98370885848999,
      "learning_rate": 9.255443957289932e-06,
      "loss": 0.0476,
      "step": 3567
    },
    {
      "epoch": 0.18424993545055512,
      "grad_norm": 6.694052219390869,
      "learning_rate": 9.255013717960496e-06,
      "loss": 0.0745,
      "step": 3568
    },
    {
      "epoch": 0.18430157500645494,
      "grad_norm": 18.58270263671875,
      "learning_rate": 9.254583364366596e-06,
      "loss": 0.0856,
      "step": 3569
    },
    {
      "epoch": 0.18435321456235476,
      "grad_norm": 8.954675674438477,
      "learning_rate": 9.254152896519792e-06,
      "loss": 0.1212,
      "step": 3570
    },
    {
      "epoch": 0.18440485411825458,
      "grad_norm": 4.458946228027344,
      "learning_rate": 9.253722314431645e-06,
      "loss": 0.0768,
      "step": 3571
    },
    {
      "epoch": 0.1844564936741544,
      "grad_norm": 7.977340221405029,
      "learning_rate": 9.253291618113715e-06,
      "loss": 0.1297,
      "step": 3572
    },
    {
      "epoch": 0.18450813323005422,
      "grad_norm": 11.472846984863281,
      "learning_rate": 9.252860807577573e-06,
      "loss": 0.0797,
      "step": 3573
    },
    {
      "epoch": 0.18455977278595403,
      "grad_norm": 21.60821533203125,
      "learning_rate": 9.252429882834781e-06,
      "loss": 0.1928,
      "step": 3574
    },
    {
      "epoch": 0.18461141234185385,
      "grad_norm": 5.188042163848877,
      "learning_rate": 9.251998843896916e-06,
      "loss": 0.1013,
      "step": 3575
    },
    {
      "epoch": 0.18466305189775367,
      "grad_norm": 6.804305553436279,
      "learning_rate": 9.251567690775554e-06,
      "loss": 0.0927,
      "step": 3576
    },
    {
      "epoch": 0.1847146914536535,
      "grad_norm": 4.990199089050293,
      "learning_rate": 9.251136423482267e-06,
      "loss": 0.0517,
      "step": 3577
    },
    {
      "epoch": 0.1847663310095533,
      "grad_norm": 2.3779075145721436,
      "learning_rate": 9.250705042028643e-06,
      "loss": 0.0458,
      "step": 3578
    },
    {
      "epoch": 0.18481797056545313,
      "grad_norm": 6.3540215492248535,
      "learning_rate": 9.250273546426265e-06,
      "loss": 0.0875,
      "step": 3579
    },
    {
      "epoch": 0.18486961012135295,
      "grad_norm": 11.968188285827637,
      "learning_rate": 9.249841936686717e-06,
      "loss": 0.2605,
      "step": 3580
    },
    {
      "epoch": 0.18492124967725276,
      "grad_norm": 6.745997428894043,
      "learning_rate": 9.249410212821592e-06,
      "loss": 0.0887,
      "step": 3581
    },
    {
      "epoch": 0.18497288923315258,
      "grad_norm": 5.586888313293457,
      "learning_rate": 9.248978374842483e-06,
      "loss": 0.0563,
      "step": 3582
    },
    {
      "epoch": 0.1850245287890524,
      "grad_norm": 17.421133041381836,
      "learning_rate": 9.248546422760987e-06,
      "loss": 0.1536,
      "step": 3583
    },
    {
      "epoch": 0.18507616834495225,
      "grad_norm": 5.875061988830566,
      "learning_rate": 9.248114356588703e-06,
      "loss": 0.0746,
      "step": 3584
    },
    {
      "epoch": 0.18512780790085206,
      "grad_norm": 5.4751386642456055,
      "learning_rate": 9.247682176337234e-06,
      "loss": 0.0504,
      "step": 3585
    },
    {
      "epoch": 0.18517944745675188,
      "grad_norm": 4.01616096496582,
      "learning_rate": 9.247249882018187e-06,
      "loss": 0.0838,
      "step": 3586
    },
    {
      "epoch": 0.1852310870126517,
      "grad_norm": 4.534946918487549,
      "learning_rate": 9.24681747364317e-06,
      "loss": 0.1115,
      "step": 3587
    },
    {
      "epoch": 0.18528272656855152,
      "grad_norm": 8.048922538757324,
      "learning_rate": 9.246384951223793e-06,
      "loss": 0.1028,
      "step": 3588
    },
    {
      "epoch": 0.18533436612445134,
      "grad_norm": 6.963881015777588,
      "learning_rate": 9.245952314771673e-06,
      "loss": 0.0639,
      "step": 3589
    },
    {
      "epoch": 0.18538600568035116,
      "grad_norm": 14.671655654907227,
      "learning_rate": 9.24551956429843e-06,
      "loss": 0.1508,
      "step": 3590
    },
    {
      "epoch": 0.18543764523625098,
      "grad_norm": 2.7297542095184326,
      "learning_rate": 9.245086699815682e-06,
      "loss": 0.069,
      "step": 3591
    },
    {
      "epoch": 0.1854892847921508,
      "grad_norm": 4.674763202667236,
      "learning_rate": 9.244653721335053e-06,
      "loss": 0.0657,
      "step": 3592
    },
    {
      "epoch": 0.1855409243480506,
      "grad_norm": 4.452726364135742,
      "learning_rate": 9.244220628868172e-06,
      "loss": 0.0593,
      "step": 3593
    },
    {
      "epoch": 0.18559256390395043,
      "grad_norm": 4.819806098937988,
      "learning_rate": 9.243787422426668e-06,
      "loss": 0.0461,
      "step": 3594
    },
    {
      "epoch": 0.18564420345985025,
      "grad_norm": 8.690642356872559,
      "learning_rate": 9.243354102022175e-06,
      "loss": 0.0651,
      "step": 3595
    },
    {
      "epoch": 0.18569584301575007,
      "grad_norm": 2.8550095558166504,
      "learning_rate": 9.24292066766633e-06,
      "loss": 0.0692,
      "step": 3596
    },
    {
      "epoch": 0.18574748257164989,
      "grad_norm": 8.692975997924805,
      "learning_rate": 9.242487119370774e-06,
      "loss": 0.0886,
      "step": 3597
    },
    {
      "epoch": 0.1857991221275497,
      "grad_norm": 11.035941123962402,
      "learning_rate": 9.242053457147145e-06,
      "loss": 0.1076,
      "step": 3598
    },
    {
      "epoch": 0.18585076168344952,
      "grad_norm": 4.457103252410889,
      "learning_rate": 9.24161968100709e-06,
      "loss": 0.0871,
      "step": 3599
    },
    {
      "epoch": 0.18590240123934934,
      "grad_norm": 6.588874816894531,
      "learning_rate": 9.24118579096226e-06,
      "loss": 0.0827,
      "step": 3600
    },
    {
      "epoch": 0.18595404079524916,
      "grad_norm": 3.7680070400238037,
      "learning_rate": 9.240751787024306e-06,
      "loss": 0.0361,
      "step": 3601
    },
    {
      "epoch": 0.18600568035114898,
      "grad_norm": 9.654056549072266,
      "learning_rate": 9.240317669204882e-06,
      "loss": 0.0757,
      "step": 3602
    },
    {
      "epoch": 0.1860573199070488,
      "grad_norm": 5.692248344421387,
      "learning_rate": 9.239883437515644e-06,
      "loss": 0.1009,
      "step": 3603
    },
    {
      "epoch": 0.18610895946294861,
      "grad_norm": 5.699707984924316,
      "learning_rate": 9.239449091968258e-06,
      "loss": 0.0777,
      "step": 3604
    },
    {
      "epoch": 0.18616059901884843,
      "grad_norm": 3.5188441276550293,
      "learning_rate": 9.239014632574384e-06,
      "loss": 0.0659,
      "step": 3605
    },
    {
      "epoch": 0.18621223857474825,
      "grad_norm": 13.870341300964355,
      "learning_rate": 9.238580059345688e-06,
      "loss": 0.1301,
      "step": 3606
    },
    {
      "epoch": 0.18626387813064807,
      "grad_norm": 9.453812599182129,
      "learning_rate": 9.238145372293844e-06,
      "loss": 0.0886,
      "step": 3607
    },
    {
      "epoch": 0.1863155176865479,
      "grad_norm": 10.571586608886719,
      "learning_rate": 9.23771057143052e-06,
      "loss": 0.153,
      "step": 3608
    },
    {
      "epoch": 0.1863671572424477,
      "grad_norm": 6.611946105957031,
      "learning_rate": 9.237275656767398e-06,
      "loss": 0.0537,
      "step": 3609
    },
    {
      "epoch": 0.18641879679834752,
      "grad_norm": 4.269173622131348,
      "learning_rate": 9.236840628316154e-06,
      "loss": 0.0538,
      "step": 3610
    },
    {
      "epoch": 0.18647043635424734,
      "grad_norm": 2.7936012744903564,
      "learning_rate": 9.236405486088472e-06,
      "loss": 0.0458,
      "step": 3611
    },
    {
      "epoch": 0.18652207591014716,
      "grad_norm": 7.670300483703613,
      "learning_rate": 9.235970230096035e-06,
      "loss": 0.0644,
      "step": 3612
    },
    {
      "epoch": 0.18657371546604698,
      "grad_norm": 6.219019412994385,
      "learning_rate": 9.23553486035053e-06,
      "loss": 0.0978,
      "step": 3613
    },
    {
      "epoch": 0.1866253550219468,
      "grad_norm": 2.3890557289123535,
      "learning_rate": 9.235099376863654e-06,
      "loss": 0.042,
      "step": 3614
    },
    {
      "epoch": 0.18667699457784662,
      "grad_norm": 3.6970598697662354,
      "learning_rate": 9.234663779647096e-06,
      "loss": 0.0587,
      "step": 3615
    },
    {
      "epoch": 0.18672863413374646,
      "grad_norm": 4.07744026184082,
      "learning_rate": 9.234228068712558e-06,
      "loss": 0.0777,
      "step": 3616
    },
    {
      "epoch": 0.18678027368964628,
      "grad_norm": 7.287055015563965,
      "learning_rate": 9.233792244071737e-06,
      "loss": 0.0428,
      "step": 3617
    },
    {
      "epoch": 0.1868319132455461,
      "grad_norm": 4.364561080932617,
      "learning_rate": 9.233356305736339e-06,
      "loss": 0.075,
      "step": 3618
    },
    {
      "epoch": 0.18688355280144592,
      "grad_norm": 4.606378078460693,
      "learning_rate": 9.232920253718071e-06,
      "loss": 0.0606,
      "step": 3619
    },
    {
      "epoch": 0.18693519235734574,
      "grad_norm": 4.029018402099609,
      "learning_rate": 9.23248408802864e-06,
      "loss": 0.0517,
      "step": 3620
    },
    {
      "epoch": 0.18698683191324555,
      "grad_norm": 7.547173976898193,
      "learning_rate": 9.232047808679762e-06,
      "loss": 0.0896,
      "step": 3621
    },
    {
      "epoch": 0.18703847146914537,
      "grad_norm": 45.29130554199219,
      "learning_rate": 9.231611415683152e-06,
      "loss": 0.2594,
      "step": 3622
    },
    {
      "epoch": 0.1870901110250452,
      "grad_norm": 13.676462173461914,
      "learning_rate": 9.231174909050526e-06,
      "loss": 0.1468,
      "step": 3623
    },
    {
      "epoch": 0.187141750580945,
      "grad_norm": 2.5848002433776855,
      "learning_rate": 9.23073828879361e-06,
      "loss": 0.0574,
      "step": 3624
    },
    {
      "epoch": 0.18719339013684483,
      "grad_norm": 11.274673461914062,
      "learning_rate": 9.230301554924128e-06,
      "loss": 0.0918,
      "step": 3625
    },
    {
      "epoch": 0.18724502969274465,
      "grad_norm": 2.1697943210601807,
      "learning_rate": 9.229864707453806e-06,
      "loss": 0.0308,
      "step": 3626
    },
    {
      "epoch": 0.18729666924864447,
      "grad_norm": 5.676640033721924,
      "learning_rate": 9.229427746394377e-06,
      "loss": 0.0484,
      "step": 3627
    },
    {
      "epoch": 0.18734830880454428,
      "grad_norm": 5.175059795379639,
      "learning_rate": 9.228990671757577e-06,
      "loss": 0.0823,
      "step": 3628
    },
    {
      "epoch": 0.1873999483604441,
      "grad_norm": 8.441752433776855,
      "learning_rate": 9.228553483555138e-06,
      "loss": 0.1295,
      "step": 3629
    },
    {
      "epoch": 0.18745158791634392,
      "grad_norm": 7.67883825302124,
      "learning_rate": 9.228116181798805e-06,
      "loss": 0.1189,
      "step": 3630
    },
    {
      "epoch": 0.18750322747224374,
      "grad_norm": 7.1501688957214355,
      "learning_rate": 9.227678766500319e-06,
      "loss": 0.0782,
      "step": 3631
    },
    {
      "epoch": 0.18755486702814356,
      "grad_norm": 4.232075214385986,
      "learning_rate": 9.227241237671427e-06,
      "loss": 0.0731,
      "step": 3632
    },
    {
      "epoch": 0.18760650658404338,
      "grad_norm": 1.9458743333816528,
      "learning_rate": 9.226803595323879e-06,
      "loss": 0.0566,
      "step": 3633
    },
    {
      "epoch": 0.1876581461399432,
      "grad_norm": 2.452826976776123,
      "learning_rate": 9.226365839469425e-06,
      "loss": 0.0556,
      "step": 3634
    },
    {
      "epoch": 0.187709785695843,
      "grad_norm": 6.152133464813232,
      "learning_rate": 9.225927970119826e-06,
      "loss": 0.087,
      "step": 3635
    },
    {
      "epoch": 0.18776142525174283,
      "grad_norm": 5.374125003814697,
      "learning_rate": 9.225489987286834e-06,
      "loss": 0.0897,
      "step": 3636
    },
    {
      "epoch": 0.18781306480764265,
      "grad_norm": 15.380433082580566,
      "learning_rate": 9.225051890982216e-06,
      "loss": 0.178,
      "step": 3637
    },
    {
      "epoch": 0.18786470436354247,
      "grad_norm": 3.432814121246338,
      "learning_rate": 9.224613681217732e-06,
      "loss": 0.0593,
      "step": 3638
    },
    {
      "epoch": 0.18791634391944229,
      "grad_norm": 5.7297844886779785,
      "learning_rate": 9.224175358005154e-06,
      "loss": 0.0941,
      "step": 3639
    },
    {
      "epoch": 0.1879679834753421,
      "grad_norm": 8.974932670593262,
      "learning_rate": 9.22373692135625e-06,
      "loss": 0.1189,
      "step": 3640
    },
    {
      "epoch": 0.18801962303124192,
      "grad_norm": 4.080178260803223,
      "learning_rate": 9.223298371282794e-06,
      "loss": 0.0862,
      "step": 3641
    },
    {
      "epoch": 0.18807126258714174,
      "grad_norm": 5.58134126663208,
      "learning_rate": 9.222859707796566e-06,
      "loss": 0.062,
      "step": 3642
    },
    {
      "epoch": 0.18812290214304156,
      "grad_norm": 2.3943912982940674,
      "learning_rate": 9.222420930909341e-06,
      "loss": 0.0261,
      "step": 3643
    },
    {
      "epoch": 0.18817454169894138,
      "grad_norm": 12.372780799865723,
      "learning_rate": 9.221982040632906e-06,
      "loss": 0.1479,
      "step": 3644
    },
    {
      "epoch": 0.1882261812548412,
      "grad_norm": 5.53975248336792,
      "learning_rate": 9.221543036979043e-06,
      "loss": 0.0723,
      "step": 3645
    },
    {
      "epoch": 0.18827782081074101,
      "grad_norm": 3.613938093185425,
      "learning_rate": 9.221103919959547e-06,
      "loss": 0.0575,
      "step": 3646
    },
    {
      "epoch": 0.18832946036664086,
      "grad_norm": 3.4855213165283203,
      "learning_rate": 9.220664689586203e-06,
      "loss": 0.0766,
      "step": 3647
    },
    {
      "epoch": 0.18838109992254068,
      "grad_norm": 3.210663318634033,
      "learning_rate": 9.220225345870812e-06,
      "loss": 0.072,
      "step": 3648
    },
    {
      "epoch": 0.1884327394784405,
      "grad_norm": 5.987829208374023,
      "learning_rate": 9.219785888825168e-06,
      "loss": 0.0981,
      "step": 3649
    },
    {
      "epoch": 0.18848437903434032,
      "grad_norm": 6.500736236572266,
      "learning_rate": 9.219346318461075e-06,
      "loss": 0.0972,
      "step": 3650
    },
    {
      "epoch": 0.18853601859024013,
      "grad_norm": 4.506022930145264,
      "learning_rate": 9.218906634790335e-06,
      "loss": 0.0438,
      "step": 3651
    },
    {
      "epoch": 0.18858765814613995,
      "grad_norm": 3.3871071338653564,
      "learning_rate": 9.218466837824757e-06,
      "loss": 0.0555,
      "step": 3652
    },
    {
      "epoch": 0.18863929770203977,
      "grad_norm": 2.620213270187378,
      "learning_rate": 9.218026927576153e-06,
      "loss": 0.0542,
      "step": 3653
    },
    {
      "epoch": 0.1886909372579396,
      "grad_norm": 5.076262950897217,
      "learning_rate": 9.217586904056333e-06,
      "loss": 0.0462,
      "step": 3654
    },
    {
      "epoch": 0.1887425768138394,
      "grad_norm": 5.546689510345459,
      "learning_rate": 9.217146767277114e-06,
      "loss": 0.066,
      "step": 3655
    },
    {
      "epoch": 0.18879421636973923,
      "grad_norm": 7.118585109710693,
      "learning_rate": 9.216706517250316e-06,
      "loss": 0.0956,
      "step": 3656
    },
    {
      "epoch": 0.18884585592563904,
      "grad_norm": 3.6293580532073975,
      "learning_rate": 9.216266153987763e-06,
      "loss": 0.0576,
      "step": 3657
    },
    {
      "epoch": 0.18889749548153886,
      "grad_norm": 7.7277960777282715,
      "learning_rate": 9.215825677501277e-06,
      "loss": 0.068,
      "step": 3658
    },
    {
      "epoch": 0.18894913503743868,
      "grad_norm": 4.1642656326293945,
      "learning_rate": 9.215385087802692e-06,
      "loss": 0.0494,
      "step": 3659
    },
    {
      "epoch": 0.1890007745933385,
      "grad_norm": 4.631000518798828,
      "learning_rate": 9.214944384903834e-06,
      "loss": 0.1039,
      "step": 3660
    },
    {
      "epoch": 0.18905241414923832,
      "grad_norm": 3.689683437347412,
      "learning_rate": 9.214503568816542e-06,
      "loss": 0.057,
      "step": 3661
    },
    {
      "epoch": 0.18910405370513814,
      "grad_norm": 6.726681232452393,
      "learning_rate": 9.21406263955265e-06,
      "loss": 0.1371,
      "step": 3662
    },
    {
      "epoch": 0.18915569326103795,
      "grad_norm": 6.816111087799072,
      "learning_rate": 9.213621597124001e-06,
      "loss": 0.087,
      "step": 3663
    },
    {
      "epoch": 0.18920733281693777,
      "grad_norm": 4.530293941497803,
      "learning_rate": 9.213180441542438e-06,
      "loss": 0.0441,
      "step": 3664
    },
    {
      "epoch": 0.1892589723728376,
      "grad_norm": 8.705872535705566,
      "learning_rate": 9.21273917281981e-06,
      "loss": 0.2012,
      "step": 3665
    },
    {
      "epoch": 0.1893106119287374,
      "grad_norm": 2.9374170303344727,
      "learning_rate": 9.212297790967964e-06,
      "loss": 0.0463,
      "step": 3666
    },
    {
      "epoch": 0.18936225148463723,
      "grad_norm": 8.181923866271973,
      "learning_rate": 9.211856295998754e-06,
      "loss": 0.0433,
      "step": 3667
    },
    {
      "epoch": 0.18941389104053705,
      "grad_norm": 14.652826309204102,
      "learning_rate": 9.211414687924033e-06,
      "loss": 0.1081,
      "step": 3668
    },
    {
      "epoch": 0.18946553059643687,
      "grad_norm": 7.60006046295166,
      "learning_rate": 9.210972966755668e-06,
      "loss": 0.0878,
      "step": 3669
    },
    {
      "epoch": 0.18951717015233668,
      "grad_norm": 2.177217721939087,
      "learning_rate": 9.210531132505512e-06,
      "loss": 0.0612,
      "step": 3670
    },
    {
      "epoch": 0.1895688097082365,
      "grad_norm": 5.8869242668151855,
      "learning_rate": 9.210089185185436e-06,
      "loss": 0.0992,
      "step": 3671
    },
    {
      "epoch": 0.18962044926413632,
      "grad_norm": 4.839385032653809,
      "learning_rate": 9.209647124807304e-06,
      "loss": 0.0583,
      "step": 3672
    },
    {
      "epoch": 0.18967208882003614,
      "grad_norm": 4.110686302185059,
      "learning_rate": 9.20920495138299e-06,
      "loss": 0.0452,
      "step": 3673
    },
    {
      "epoch": 0.18972372837593596,
      "grad_norm": 5.293500900268555,
      "learning_rate": 9.208762664924369e-06,
      "loss": 0.0745,
      "step": 3674
    },
    {
      "epoch": 0.18977536793183578,
      "grad_norm": 3.2891793251037598,
      "learning_rate": 9.208320265443315e-06,
      "loss": 0.0885,
      "step": 3675
    },
    {
      "epoch": 0.1898270074877356,
      "grad_norm": 2.431340456008911,
      "learning_rate": 9.20787775295171e-06,
      "loss": 0.0545,
      "step": 3676
    },
    {
      "epoch": 0.1898786470436354,
      "grad_norm": 3.3659353256225586,
      "learning_rate": 9.207435127461437e-06,
      "loss": 0.0645,
      "step": 3677
    },
    {
      "epoch": 0.18993028659953523,
      "grad_norm": 2.855677604675293,
      "learning_rate": 9.206992388984382e-06,
      "loss": 0.0675,
      "step": 3678
    },
    {
      "epoch": 0.18998192615543508,
      "grad_norm": 10.910414695739746,
      "learning_rate": 9.206549537532434e-06,
      "loss": 0.1074,
      "step": 3679
    },
    {
      "epoch": 0.1900335657113349,
      "grad_norm": 8.822367668151855,
      "learning_rate": 9.206106573117486e-06,
      "loss": 0.1082,
      "step": 3680
    },
    {
      "epoch": 0.1900852052672347,
      "grad_norm": 7.014044761657715,
      "learning_rate": 9.205663495751434e-06,
      "loss": 0.1022,
      "step": 3681
    },
    {
      "epoch": 0.19013684482313453,
      "grad_norm": 3.8826568126678467,
      "learning_rate": 9.205220305446177e-06,
      "loss": 0.0653,
      "step": 3682
    },
    {
      "epoch": 0.19018848437903435,
      "grad_norm": 9.36783504486084,
      "learning_rate": 9.204777002213613e-06,
      "loss": 0.1153,
      "step": 3683
    },
    {
      "epoch": 0.19024012393493417,
      "grad_norm": 3.9999947547912598,
      "learning_rate": 9.20433358606565e-06,
      "loss": 0.0654,
      "step": 3684
    },
    {
      "epoch": 0.190291763490834,
      "grad_norm": 6.583707809448242,
      "learning_rate": 9.203890057014194e-06,
      "loss": 0.0466,
      "step": 3685
    },
    {
      "epoch": 0.1903434030467338,
      "grad_norm": 2.822321653366089,
      "learning_rate": 9.203446415071155e-06,
      "loss": 0.0535,
      "step": 3686
    },
    {
      "epoch": 0.19039504260263362,
      "grad_norm": 8.535906791687012,
      "learning_rate": 9.20300266024845e-06,
      "loss": 0.0693,
      "step": 3687
    },
    {
      "epoch": 0.19044668215853344,
      "grad_norm": 4.820272922515869,
      "learning_rate": 9.202558792557992e-06,
      "loss": 0.1,
      "step": 3688
    },
    {
      "epoch": 0.19049832171443326,
      "grad_norm": 3.029616355895996,
      "learning_rate": 9.2021148120117e-06,
      "loss": 0.0687,
      "step": 3689
    },
    {
      "epoch": 0.19054996127033308,
      "grad_norm": 12.787368774414062,
      "learning_rate": 9.2016707186215e-06,
      "loss": 0.0597,
      "step": 3690
    },
    {
      "epoch": 0.1906016008262329,
      "grad_norm": 4.890477180480957,
      "learning_rate": 9.201226512399316e-06,
      "loss": 0.058,
      "step": 3691
    },
    {
      "epoch": 0.19065324038213272,
      "grad_norm": 12.701774597167969,
      "learning_rate": 9.200782193357077e-06,
      "loss": 0.0954,
      "step": 3692
    },
    {
      "epoch": 0.19070487993803253,
      "grad_norm": 4.375362396240234,
      "learning_rate": 9.200337761506714e-06,
      "loss": 0.1041,
      "step": 3693
    },
    {
      "epoch": 0.19075651949393235,
      "grad_norm": 19.97855567932129,
      "learning_rate": 9.199893216860163e-06,
      "loss": 0.1152,
      "step": 3694
    },
    {
      "epoch": 0.19080815904983217,
      "grad_norm": 22.86229705810547,
      "learning_rate": 9.199448559429361e-06,
      "loss": 0.162,
      "step": 3695
    },
    {
      "epoch": 0.190859798605732,
      "grad_norm": 3.5216710567474365,
      "learning_rate": 9.199003789226248e-06,
      "loss": 0.0545,
      "step": 3696
    },
    {
      "epoch": 0.1909114381616318,
      "grad_norm": 7.875607967376709,
      "learning_rate": 9.198558906262772e-06,
      "loss": 0.1472,
      "step": 3697
    },
    {
      "epoch": 0.19096307771753163,
      "grad_norm": 2.7743546962738037,
      "learning_rate": 9.198113910550874e-06,
      "loss": 0.0437,
      "step": 3698
    },
    {
      "epoch": 0.19101471727343144,
      "grad_norm": 3.602705478668213,
      "learning_rate": 9.197668802102509e-06,
      "loss": 0.0744,
      "step": 3699
    },
    {
      "epoch": 0.19106635682933126,
      "grad_norm": 8.660967826843262,
      "learning_rate": 9.197223580929629e-06,
      "loss": 0.1667,
      "step": 3700
    },
    {
      "epoch": 0.19111799638523108,
      "grad_norm": 3.7910611629486084,
      "learning_rate": 9.196778247044186e-06,
      "loss": 0.0762,
      "step": 3701
    },
    {
      "epoch": 0.1911696359411309,
      "grad_norm": 56.579811096191406,
      "learning_rate": 9.196332800458144e-06,
      "loss": 0.2082,
      "step": 3702
    },
    {
      "epoch": 0.19122127549703072,
      "grad_norm": 3.0996832847595215,
      "learning_rate": 9.19588724118346e-06,
      "loss": 0.0392,
      "step": 3703
    },
    {
      "epoch": 0.19127291505293054,
      "grad_norm": 9.339536666870117,
      "learning_rate": 9.195441569232106e-06,
      "loss": 0.0512,
      "step": 3704
    },
    {
      "epoch": 0.19132455460883036,
      "grad_norm": 11.756747245788574,
      "learning_rate": 9.194995784616044e-06,
      "loss": 0.1902,
      "step": 3705
    },
    {
      "epoch": 0.19137619416473017,
      "grad_norm": 15.27363109588623,
      "learning_rate": 9.194549887347247e-06,
      "loss": 0.0987,
      "step": 3706
    },
    {
      "epoch": 0.19142783372063,
      "grad_norm": 16.022512435913086,
      "learning_rate": 9.19410387743769e-06,
      "loss": 0.1622,
      "step": 3707
    },
    {
      "epoch": 0.1914794732765298,
      "grad_norm": 2.9597160816192627,
      "learning_rate": 9.19365775489935e-06,
      "loss": 0.0444,
      "step": 3708
    },
    {
      "epoch": 0.19153111283242963,
      "grad_norm": 7.218082427978516,
      "learning_rate": 9.193211519744208e-06,
      "loss": 0.1461,
      "step": 3709
    },
    {
      "epoch": 0.19158275238832945,
      "grad_norm": 2.8515632152557373,
      "learning_rate": 9.192765171984243e-06,
      "loss": 0.0321,
      "step": 3710
    },
    {
      "epoch": 0.1916343919442293,
      "grad_norm": 4.992444038391113,
      "learning_rate": 9.192318711631447e-06,
      "loss": 0.0705,
      "step": 3711
    },
    {
      "epoch": 0.1916860315001291,
      "grad_norm": 4.354465961456299,
      "learning_rate": 9.191872138697806e-06,
      "loss": 0.1052,
      "step": 3712
    },
    {
      "epoch": 0.19173767105602893,
      "grad_norm": 6.618939399719238,
      "learning_rate": 9.191425453195314e-06,
      "loss": 0.1022,
      "step": 3713
    },
    {
      "epoch": 0.19178931061192875,
      "grad_norm": 3.6504180431365967,
      "learning_rate": 9.190978655135963e-06,
      "loss": 0.0986,
      "step": 3714
    },
    {
      "epoch": 0.19184095016782857,
      "grad_norm": 4.250824451446533,
      "learning_rate": 9.190531744531755e-06,
      "loss": 0.067,
      "step": 3715
    },
    {
      "epoch": 0.19189258972372839,
      "grad_norm": 3.7986409664154053,
      "learning_rate": 9.19008472139469e-06,
      "loss": 0.0817,
      "step": 3716
    },
    {
      "epoch": 0.1919442292796282,
      "grad_norm": 2.6082422733306885,
      "learning_rate": 9.189637585736772e-06,
      "loss": 0.0699,
      "step": 3717
    },
    {
      "epoch": 0.19199586883552802,
      "grad_norm": 3.598485231399536,
      "learning_rate": 9.189190337570008e-06,
      "loss": 0.1003,
      "step": 3718
    },
    {
      "epoch": 0.19204750839142784,
      "grad_norm": 5.128357887268066,
      "learning_rate": 9.18874297690641e-06,
      "loss": 0.0562,
      "step": 3719
    },
    {
      "epoch": 0.19209914794732766,
      "grad_norm": 2.1277174949645996,
      "learning_rate": 9.188295503757991e-06,
      "loss": 0.0439,
      "step": 3720
    },
    {
      "epoch": 0.19215078750322748,
      "grad_norm": 5.965864658355713,
      "learning_rate": 9.187847918136768e-06,
      "loss": 0.0479,
      "step": 3721
    },
    {
      "epoch": 0.1922024270591273,
      "grad_norm": 2.8699464797973633,
      "learning_rate": 9.187400220054758e-06,
      "loss": 0.0235,
      "step": 3722
    },
    {
      "epoch": 0.1922540666150271,
      "grad_norm": 6.5985026359558105,
      "learning_rate": 9.186952409523986e-06,
      "loss": 0.0673,
      "step": 3723
    },
    {
      "epoch": 0.19230570617092693,
      "grad_norm": 9.646446228027344,
      "learning_rate": 9.186504486556477e-06,
      "loss": 0.133,
      "step": 3724
    },
    {
      "epoch": 0.19235734572682675,
      "grad_norm": 5.067551136016846,
      "learning_rate": 9.186056451164259e-06,
      "loss": 0.0716,
      "step": 3725
    },
    {
      "epoch": 0.19240898528272657,
      "grad_norm": 9.44299030303955,
      "learning_rate": 9.185608303359364e-06,
      "loss": 0.1269,
      "step": 3726
    },
    {
      "epoch": 0.1924606248386264,
      "grad_norm": 21.616151809692383,
      "learning_rate": 9.185160043153825e-06,
      "loss": 0.281,
      "step": 3727
    },
    {
      "epoch": 0.1925122643945262,
      "grad_norm": 8.879705429077148,
      "learning_rate": 9.184711670559683e-06,
      "loss": 0.1403,
      "step": 3728
    },
    {
      "epoch": 0.19256390395042602,
      "grad_norm": 12.25217056274414,
      "learning_rate": 9.184263185588976e-06,
      "loss": 0.1612,
      "step": 3729
    },
    {
      "epoch": 0.19261554350632584,
      "grad_norm": 8.019648551940918,
      "learning_rate": 9.183814588253748e-06,
      "loss": 0.1201,
      "step": 3730
    },
    {
      "epoch": 0.19266718306222566,
      "grad_norm": 7.040307998657227,
      "learning_rate": 9.183365878566046e-06,
      "loss": 0.0832,
      "step": 3731
    },
    {
      "epoch": 0.19271882261812548,
      "grad_norm": 4.066503047943115,
      "learning_rate": 9.18291705653792e-06,
      "loss": 0.1116,
      "step": 3732
    },
    {
      "epoch": 0.1927704621740253,
      "grad_norm": 3.803865909576416,
      "learning_rate": 9.182468122181424e-06,
      "loss": 0.047,
      "step": 3733
    },
    {
      "epoch": 0.19282210172992512,
      "grad_norm": 6.011661529541016,
      "learning_rate": 9.18201907550861e-06,
      "loss": 0.0609,
      "step": 3734
    },
    {
      "epoch": 0.19287374128582493,
      "grad_norm": 9.701309204101562,
      "learning_rate": 9.18156991653154e-06,
      "loss": 0.096,
      "step": 3735
    },
    {
      "epoch": 0.19292538084172475,
      "grad_norm": 3.205371618270874,
      "learning_rate": 9.181120645262273e-06,
      "loss": 0.0416,
      "step": 3736
    },
    {
      "epoch": 0.19297702039762457,
      "grad_norm": 6.101954936981201,
      "learning_rate": 9.180671261712876e-06,
      "loss": 0.0789,
      "step": 3737
    },
    {
      "epoch": 0.1930286599535244,
      "grad_norm": 8.833907127380371,
      "learning_rate": 9.180221765895419e-06,
      "loss": 0.1033,
      "step": 3738
    },
    {
      "epoch": 0.1930802995094242,
      "grad_norm": 5.009389877319336,
      "learning_rate": 9.179772157821967e-06,
      "loss": 0.0707,
      "step": 3739
    },
    {
      "epoch": 0.19313193906532403,
      "grad_norm": 10.060247421264648,
      "learning_rate": 9.179322437504596e-06,
      "loss": 0.0746,
      "step": 3740
    },
    {
      "epoch": 0.19318357862122384,
      "grad_norm": 4.088791847229004,
      "learning_rate": 9.178872604955386e-06,
      "loss": 0.0965,
      "step": 3741
    },
    {
      "epoch": 0.19323521817712366,
      "grad_norm": 4.557896137237549,
      "learning_rate": 9.178422660186412e-06,
      "loss": 0.0592,
      "step": 3742
    },
    {
      "epoch": 0.1932868577330235,
      "grad_norm": 7.1430487632751465,
      "learning_rate": 9.17797260320976e-06,
      "loss": 0.0807,
      "step": 3743
    },
    {
      "epoch": 0.19333849728892333,
      "grad_norm": 4.49185848236084,
      "learning_rate": 9.177522434037517e-06,
      "loss": 0.0593,
      "step": 3744
    },
    {
      "epoch": 0.19339013684482315,
      "grad_norm": 10.020188331604004,
      "learning_rate": 9.177072152681769e-06,
      "loss": 0.1065,
      "step": 3745
    },
    {
      "epoch": 0.19344177640072296,
      "grad_norm": 6.470309257507324,
      "learning_rate": 9.176621759154607e-06,
      "loss": 0.0541,
      "step": 3746
    },
    {
      "epoch": 0.19349341595662278,
      "grad_norm": 31.564455032348633,
      "learning_rate": 9.17617125346813e-06,
      "loss": 0.1194,
      "step": 3747
    },
    {
      "epoch": 0.1935450555125226,
      "grad_norm": 12.491499900817871,
      "learning_rate": 9.175720635634432e-06,
      "loss": 0.0851,
      "step": 3748
    },
    {
      "epoch": 0.19359669506842242,
      "grad_norm": 3.4681975841522217,
      "learning_rate": 9.175269905665618e-06,
      "loss": 0.0532,
      "step": 3749
    },
    {
      "epoch": 0.19364833462432224,
      "grad_norm": 12.389108657836914,
      "learning_rate": 9.17481906357379e-06,
      "loss": 0.0918,
      "step": 3750
    },
    {
      "epoch": 0.19369997418022206,
      "grad_norm": 5.28337287902832,
      "learning_rate": 9.174368109371053e-06,
      "loss": 0.05,
      "step": 3751
    },
    {
      "epoch": 0.19375161373612187,
      "grad_norm": 9.034316062927246,
      "learning_rate": 9.173917043069516e-06,
      "loss": 0.0683,
      "step": 3752
    },
    {
      "epoch": 0.1938032532920217,
      "grad_norm": 6.401576042175293,
      "learning_rate": 9.173465864681298e-06,
      "loss": 0.086,
      "step": 3753
    },
    {
      "epoch": 0.1938548928479215,
      "grad_norm": 8.983280181884766,
      "learning_rate": 9.17301457421851e-06,
      "loss": 0.0963,
      "step": 3754
    },
    {
      "epoch": 0.19390653240382133,
      "grad_norm": 5.869935035705566,
      "learning_rate": 9.172563171693271e-06,
      "loss": 0.063,
      "step": 3755
    },
    {
      "epoch": 0.19395817195972115,
      "grad_norm": 10.534401893615723,
      "learning_rate": 9.172111657117708e-06,
      "loss": 0.0839,
      "step": 3756
    },
    {
      "epoch": 0.19400981151562097,
      "grad_norm": 5.065035343170166,
      "learning_rate": 9.17166003050394e-06,
      "loss": 0.0675,
      "step": 3757
    },
    {
      "epoch": 0.19406145107152079,
      "grad_norm": 6.559551239013672,
      "learning_rate": 9.171208291864097e-06,
      "loss": 0.0962,
      "step": 3758
    },
    {
      "epoch": 0.1941130906274206,
      "grad_norm": 10.85229778289795,
      "learning_rate": 9.170756441210313e-06,
      "loss": 0.0795,
      "step": 3759
    },
    {
      "epoch": 0.19416473018332042,
      "grad_norm": 2.3365724086761475,
      "learning_rate": 9.170304478554716e-06,
      "loss": 0.0479,
      "step": 3760
    },
    {
      "epoch": 0.19421636973922024,
      "grad_norm": 7.8524627685546875,
      "learning_rate": 9.169852403909449e-06,
      "loss": 0.0764,
      "step": 3761
    },
    {
      "epoch": 0.19426800929512006,
      "grad_norm": 9.424506187438965,
      "learning_rate": 9.169400217286649e-06,
      "loss": 0.1112,
      "step": 3762
    },
    {
      "epoch": 0.19431964885101988,
      "grad_norm": 3.6887876987457275,
      "learning_rate": 9.16894791869846e-06,
      "loss": 0.1171,
      "step": 3763
    },
    {
      "epoch": 0.1943712884069197,
      "grad_norm": 6.881702899932861,
      "learning_rate": 9.168495508157027e-06,
      "loss": 0.116,
      "step": 3764
    },
    {
      "epoch": 0.19442292796281951,
      "grad_norm": 7.6071624755859375,
      "learning_rate": 9.1680429856745e-06,
      "loss": 0.1168,
      "step": 3765
    },
    {
      "epoch": 0.19447456751871933,
      "grad_norm": 5.846518516540527,
      "learning_rate": 9.167590351263031e-06,
      "loss": 0.0807,
      "step": 3766
    },
    {
      "epoch": 0.19452620707461915,
      "grad_norm": 12.987567901611328,
      "learning_rate": 9.167137604934775e-06,
      "loss": 0.1464,
      "step": 3767
    },
    {
      "epoch": 0.19457784663051897,
      "grad_norm": 10.291667938232422,
      "learning_rate": 9.166684746701889e-06,
      "loss": 0.0627,
      "step": 3768
    },
    {
      "epoch": 0.1946294861864188,
      "grad_norm": 12.099553108215332,
      "learning_rate": 9.166231776576537e-06,
      "loss": 0.1586,
      "step": 3769
    },
    {
      "epoch": 0.1946811257423186,
      "grad_norm": 7.589510440826416,
      "learning_rate": 9.165778694570879e-06,
      "loss": 0.0962,
      "step": 3770
    },
    {
      "epoch": 0.19473276529821842,
      "grad_norm": 7.043632984161377,
      "learning_rate": 9.165325500697086e-06,
      "loss": 0.0916,
      "step": 3771
    },
    {
      "epoch": 0.19478440485411824,
      "grad_norm": 3.1600615978240967,
      "learning_rate": 9.164872194967324e-06,
      "loss": 0.0621,
      "step": 3772
    },
    {
      "epoch": 0.19483604441001806,
      "grad_norm": 2.2324585914611816,
      "learning_rate": 9.164418777393772e-06,
      "loss": 0.0367,
      "step": 3773
    },
    {
      "epoch": 0.19488768396591788,
      "grad_norm": 6.863480091094971,
      "learning_rate": 9.1639652479886e-06,
      "loss": 0.0474,
      "step": 3774
    },
    {
      "epoch": 0.19493932352181773,
      "grad_norm": 8.518043518066406,
      "learning_rate": 9.163511606763991e-06,
      "loss": 0.0986,
      "step": 3775
    },
    {
      "epoch": 0.19499096307771754,
      "grad_norm": 3.267582416534424,
      "learning_rate": 9.163057853732125e-06,
      "loss": 0.0425,
      "step": 3776
    },
    {
      "epoch": 0.19504260263361736,
      "grad_norm": 5.8659257888793945,
      "learning_rate": 9.162603988905188e-06,
      "loss": 0.081,
      "step": 3777
    },
    {
      "epoch": 0.19509424218951718,
      "grad_norm": 6.462281227111816,
      "learning_rate": 9.162150012295368e-06,
      "loss": 0.1074,
      "step": 3778
    },
    {
      "epoch": 0.195145881745417,
      "grad_norm": 9.11938190460205,
      "learning_rate": 9.161695923914857e-06,
      "loss": 0.0759,
      "step": 3779
    },
    {
      "epoch": 0.19519752130131682,
      "grad_norm": 6.960662841796875,
      "learning_rate": 9.161241723775847e-06,
      "loss": 0.0404,
      "step": 3780
    },
    {
      "epoch": 0.19524916085721664,
      "grad_norm": 5.64369535446167,
      "learning_rate": 9.160787411890536e-06,
      "loss": 0.0463,
      "step": 3781
    },
    {
      "epoch": 0.19530080041311645,
      "grad_norm": 14.178531646728516,
      "learning_rate": 9.160332988271128e-06,
      "loss": 0.1379,
      "step": 3782
    },
    {
      "epoch": 0.19535243996901627,
      "grad_norm": 4.2106781005859375,
      "learning_rate": 9.15987845292982e-06,
      "loss": 0.0952,
      "step": 3783
    },
    {
      "epoch": 0.1954040795249161,
      "grad_norm": 4.404030799865723,
      "learning_rate": 9.159423805878822e-06,
      "loss": 0.0651,
      "step": 3784
    },
    {
      "epoch": 0.1954557190808159,
      "grad_norm": 2.486466407775879,
      "learning_rate": 9.15896904713034e-06,
      "loss": 0.0258,
      "step": 3785
    },
    {
      "epoch": 0.19550735863671573,
      "grad_norm": 3.9838385581970215,
      "learning_rate": 9.15851417669659e-06,
      "loss": 0.0542,
      "step": 3786
    },
    {
      "epoch": 0.19555899819261555,
      "grad_norm": 4.819949626922607,
      "learning_rate": 9.158059194589784e-06,
      "loss": 0.0248,
      "step": 3787
    },
    {
      "epoch": 0.19561063774851536,
      "grad_norm": 14.713141441345215,
      "learning_rate": 9.157604100822141e-06,
      "loss": 0.1749,
      "step": 3788
    },
    {
      "epoch": 0.19566227730441518,
      "grad_norm": 5.607892990112305,
      "learning_rate": 9.157148895405885e-06,
      "loss": 0.0605,
      "step": 3789
    },
    {
      "epoch": 0.195713916860315,
      "grad_norm": 10.896349906921387,
      "learning_rate": 9.156693578353235e-06,
      "loss": 0.1058,
      "step": 3790
    },
    {
      "epoch": 0.19576555641621482,
      "grad_norm": 7.381856918334961,
      "learning_rate": 9.156238149676422e-06,
      "loss": 0.1187,
      "step": 3791
    },
    {
      "epoch": 0.19581719597211464,
      "grad_norm": 4.707984447479248,
      "learning_rate": 9.155782609387673e-06,
      "loss": 0.0968,
      "step": 3792
    },
    {
      "epoch": 0.19586883552801446,
      "grad_norm": 6.500934600830078,
      "learning_rate": 9.155326957499225e-06,
      "loss": 0.0692,
      "step": 3793
    },
    {
      "epoch": 0.19592047508391428,
      "grad_norm": 14.765262603759766,
      "learning_rate": 9.154871194023312e-06,
      "loss": 0.2126,
      "step": 3794
    },
    {
      "epoch": 0.1959721146398141,
      "grad_norm": 6.5264153480529785,
      "learning_rate": 9.154415318972172e-06,
      "loss": 0.1246,
      "step": 3795
    },
    {
      "epoch": 0.1960237541957139,
      "grad_norm": 3.5329675674438477,
      "learning_rate": 9.153959332358048e-06,
      "loss": 0.0628,
      "step": 3796
    },
    {
      "epoch": 0.19607539375161373,
      "grad_norm": 3.84702730178833,
      "learning_rate": 9.153503234193186e-06,
      "loss": 0.0605,
      "step": 3797
    },
    {
      "epoch": 0.19612703330751355,
      "grad_norm": 7.155388355255127,
      "learning_rate": 9.153047024489832e-06,
      "loss": 0.0755,
      "step": 3798
    },
    {
      "epoch": 0.19617867286341337,
      "grad_norm": 4.211790084838867,
      "learning_rate": 9.15259070326024e-06,
      "loss": 0.0573,
      "step": 3799
    },
    {
      "epoch": 0.19623031241931319,
      "grad_norm": 3.4142329692840576,
      "learning_rate": 9.152134270516663e-06,
      "loss": 0.0414,
      "step": 3800
    },
    {
      "epoch": 0.196281951975213,
      "grad_norm": 8.811169624328613,
      "learning_rate": 9.151677726271358e-06,
      "loss": 0.1228,
      "step": 3801
    },
    {
      "epoch": 0.19633359153111282,
      "grad_norm": 2.8395912647247314,
      "learning_rate": 9.151221070536583e-06,
      "loss": 0.0568,
      "step": 3802
    },
    {
      "epoch": 0.19638523108701264,
      "grad_norm": 2.60565447807312,
      "learning_rate": 9.150764303324605e-06,
      "loss": 0.0379,
      "step": 3803
    },
    {
      "epoch": 0.19643687064291246,
      "grad_norm": 5.92073392868042,
      "learning_rate": 9.150307424647685e-06,
      "loss": 0.0984,
      "step": 3804
    },
    {
      "epoch": 0.19648851019881228,
      "grad_norm": 9.616219520568848,
      "learning_rate": 9.149850434518097e-06,
      "loss": 0.1248,
      "step": 3805
    },
    {
      "epoch": 0.1965401497547121,
      "grad_norm": 9.254823684692383,
      "learning_rate": 9.14939333294811e-06,
      "loss": 0.0848,
      "step": 3806
    },
    {
      "epoch": 0.19659178931061194,
      "grad_norm": 2.691452741622925,
      "learning_rate": 9.14893611995e-06,
      "loss": 0.0488,
      "step": 3807
    },
    {
      "epoch": 0.19664342886651176,
      "grad_norm": 3.7457847595214844,
      "learning_rate": 9.148478795536045e-06,
      "loss": 0.0545,
      "step": 3808
    },
    {
      "epoch": 0.19669506842241158,
      "grad_norm": 24.79654312133789,
      "learning_rate": 9.148021359718527e-06,
      "loss": 0.2437,
      "step": 3809
    },
    {
      "epoch": 0.1967467079783114,
      "grad_norm": 4.2821455001831055,
      "learning_rate": 9.147563812509728e-06,
      "loss": 0.0606,
      "step": 3810
    },
    {
      "epoch": 0.19679834753421122,
      "grad_norm": 4.057889461517334,
      "learning_rate": 9.147106153921937e-06,
      "loss": 0.0683,
      "step": 3811
    },
    {
      "epoch": 0.19684998709011103,
      "grad_norm": 3.611520528793335,
      "learning_rate": 9.146648383967443e-06,
      "loss": 0.0513,
      "step": 3812
    },
    {
      "epoch": 0.19690162664601085,
      "grad_norm": 9.623891830444336,
      "learning_rate": 9.146190502658538e-06,
      "loss": 0.0847,
      "step": 3813
    },
    {
      "epoch": 0.19695326620191067,
      "grad_norm": 8.228231430053711,
      "learning_rate": 9.145732510007522e-06,
      "loss": 0.0767,
      "step": 3814
    },
    {
      "epoch": 0.1970049057578105,
      "grad_norm": 1.7140183448791504,
      "learning_rate": 9.145274406026688e-06,
      "loss": 0.0561,
      "step": 3815
    },
    {
      "epoch": 0.1970565453137103,
      "grad_norm": 3.20184588432312,
      "learning_rate": 9.144816190728341e-06,
      "loss": 0.0584,
      "step": 3816
    },
    {
      "epoch": 0.19710818486961013,
      "grad_norm": 4.314933776855469,
      "learning_rate": 9.144357864124789e-06,
      "loss": 0.0818,
      "step": 3817
    },
    {
      "epoch": 0.19715982442550994,
      "grad_norm": 4.933061122894287,
      "learning_rate": 9.143899426228334e-06,
      "loss": 0.1006,
      "step": 3818
    },
    {
      "epoch": 0.19721146398140976,
      "grad_norm": 7.64523458480835,
      "learning_rate": 9.14344087705129e-06,
      "loss": 0.1161,
      "step": 3819
    },
    {
      "epoch": 0.19726310353730958,
      "grad_norm": 7.040878772735596,
      "learning_rate": 9.142982216605972e-06,
      "loss": 0.0662,
      "step": 3820
    },
    {
      "epoch": 0.1973147430932094,
      "grad_norm": 3.565131902694702,
      "learning_rate": 9.142523444904695e-06,
      "loss": 0.0653,
      "step": 3821
    },
    {
      "epoch": 0.19736638264910922,
      "grad_norm": 4.283853530883789,
      "learning_rate": 9.142064561959779e-06,
      "loss": 0.1362,
      "step": 3822
    },
    {
      "epoch": 0.19741802220500904,
      "grad_norm": 2.4596426486968994,
      "learning_rate": 9.141605567783548e-06,
      "loss": 0.0418,
      "step": 3823
    },
    {
      "epoch": 0.19746966176090885,
      "grad_norm": 2.7393667697906494,
      "learning_rate": 9.141146462388327e-06,
      "loss": 0.0782,
      "step": 3824
    },
    {
      "epoch": 0.19752130131680867,
      "grad_norm": 2.9497082233428955,
      "learning_rate": 9.140687245786445e-06,
      "loss": 0.0476,
      "step": 3825
    },
    {
      "epoch": 0.1975729408727085,
      "grad_norm": 5.090112209320068,
      "learning_rate": 9.140227917990234e-06,
      "loss": 0.0687,
      "step": 3826
    },
    {
      "epoch": 0.1976245804286083,
      "grad_norm": 2.3165743350982666,
      "learning_rate": 9.13976847901203e-06,
      "loss": 0.0648,
      "step": 3827
    },
    {
      "epoch": 0.19767621998450813,
      "grad_norm": 3.274582862854004,
      "learning_rate": 9.139308928864168e-06,
      "loss": 0.0751,
      "step": 3828
    },
    {
      "epoch": 0.19772785954040795,
      "grad_norm": 14.06678295135498,
      "learning_rate": 9.13884926755899e-06,
      "loss": 0.1507,
      "step": 3829
    },
    {
      "epoch": 0.19777949909630776,
      "grad_norm": 3.9830100536346436,
      "learning_rate": 9.138389495108842e-06,
      "loss": 0.0647,
      "step": 3830
    },
    {
      "epoch": 0.19783113865220758,
      "grad_norm": 2.4001662731170654,
      "learning_rate": 9.137929611526069e-06,
      "loss": 0.037,
      "step": 3831
    },
    {
      "epoch": 0.1978827782081074,
      "grad_norm": 4.8545050621032715,
      "learning_rate": 9.137469616823018e-06,
      "loss": 0.0733,
      "step": 3832
    },
    {
      "epoch": 0.19793441776400722,
      "grad_norm": 5.005770206451416,
      "learning_rate": 9.137009511012047e-06,
      "loss": 0.0895,
      "step": 3833
    },
    {
      "epoch": 0.19798605731990704,
      "grad_norm": 10.0040283203125,
      "learning_rate": 9.136549294105508e-06,
      "loss": 0.109,
      "step": 3834
    },
    {
      "epoch": 0.19803769687580686,
      "grad_norm": 2.692869186401367,
      "learning_rate": 9.13608896611576e-06,
      "loss": 0.0496,
      "step": 3835
    },
    {
      "epoch": 0.19808933643170668,
      "grad_norm": 3.9636781215667725,
      "learning_rate": 9.135628527055166e-06,
      "loss": 0.0629,
      "step": 3836
    },
    {
      "epoch": 0.1981409759876065,
      "grad_norm": 5.136909484863281,
      "learning_rate": 9.135167976936092e-06,
      "loss": 0.1053,
      "step": 3837
    },
    {
      "epoch": 0.19819261554350634,
      "grad_norm": 5.178792476654053,
      "learning_rate": 9.134707315770903e-06,
      "loss": 0.0673,
      "step": 3838
    },
    {
      "epoch": 0.19824425509940616,
      "grad_norm": 3.86588716506958,
      "learning_rate": 9.13424654357197e-06,
      "loss": 0.0717,
      "step": 3839
    },
    {
      "epoch": 0.19829589465530598,
      "grad_norm": 8.20169734954834,
      "learning_rate": 9.133785660351666e-06,
      "loss": 0.1224,
      "step": 3840
    },
    {
      "epoch": 0.1983475342112058,
      "grad_norm": 4.777948379516602,
      "learning_rate": 9.13332466612237e-06,
      "loss": 0.0982,
      "step": 3841
    },
    {
      "epoch": 0.1983991737671056,
      "grad_norm": 2.5266690254211426,
      "learning_rate": 9.132863560896458e-06,
      "loss": 0.059,
      "step": 3842
    },
    {
      "epoch": 0.19845081332300543,
      "grad_norm": 10.037764549255371,
      "learning_rate": 9.132402344686317e-06,
      "loss": 0.1584,
      "step": 3843
    },
    {
      "epoch": 0.19850245287890525,
      "grad_norm": 5.775354385375977,
      "learning_rate": 9.131941017504329e-06,
      "loss": 0.0962,
      "step": 3844
    },
    {
      "epoch": 0.19855409243480507,
      "grad_norm": 5.427224159240723,
      "learning_rate": 9.131479579362885e-06,
      "loss": 0.0859,
      "step": 3845
    },
    {
      "epoch": 0.1986057319907049,
      "grad_norm": 6.3256425857543945,
      "learning_rate": 9.131018030274376e-06,
      "loss": 0.0676,
      "step": 3846
    },
    {
      "epoch": 0.1986573715466047,
      "grad_norm": 6.0661516189575195,
      "learning_rate": 9.130556370251195e-06,
      "loss": 0.1327,
      "step": 3847
    },
    {
      "epoch": 0.19870901110250452,
      "grad_norm": 11.265608787536621,
      "learning_rate": 9.130094599305739e-06,
      "loss": 0.1267,
      "step": 3848
    },
    {
      "epoch": 0.19876065065840434,
      "grad_norm": 10.202543258666992,
      "learning_rate": 9.129632717450412e-06,
      "loss": 0.107,
      "step": 3849
    },
    {
      "epoch": 0.19881229021430416,
      "grad_norm": 6.563423156738281,
      "learning_rate": 9.129170724697614e-06,
      "loss": 0.1443,
      "step": 3850
    },
    {
      "epoch": 0.19886392977020398,
      "grad_norm": 3.8841657638549805,
      "learning_rate": 9.12870862105975e-06,
      "loss": 0.0989,
      "step": 3851
    },
    {
      "epoch": 0.1989155693261038,
      "grad_norm": 3.733614683151245,
      "learning_rate": 9.128246406549234e-06,
      "loss": 0.0653,
      "step": 3852
    },
    {
      "epoch": 0.19896720888200362,
      "grad_norm": 4.251763343811035,
      "learning_rate": 9.127784081178476e-06,
      "loss": 0.0781,
      "step": 3853
    },
    {
      "epoch": 0.19901884843790343,
      "grad_norm": 3.7028677463531494,
      "learning_rate": 9.127321644959894e-06,
      "loss": 0.071,
      "step": 3854
    },
    {
      "epoch": 0.19907048799380325,
      "grad_norm": 7.201657772064209,
      "learning_rate": 9.126859097905901e-06,
      "loss": 0.0709,
      "step": 3855
    },
    {
      "epoch": 0.19912212754970307,
      "grad_norm": 3.2106635570526123,
      "learning_rate": 9.126396440028922e-06,
      "loss": 0.0557,
      "step": 3856
    },
    {
      "epoch": 0.1991737671056029,
      "grad_norm": 7.325685977935791,
      "learning_rate": 9.125933671341379e-06,
      "loss": 0.2096,
      "step": 3857
    },
    {
      "epoch": 0.1992254066615027,
      "grad_norm": 4.437594890594482,
      "learning_rate": 9.125470791855702e-06,
      "loss": 0.0608,
      "step": 3858
    },
    {
      "epoch": 0.19927704621740253,
      "grad_norm": 2.842357635498047,
      "learning_rate": 9.12500780158432e-06,
      "loss": 0.1018,
      "step": 3859
    },
    {
      "epoch": 0.19932868577330234,
      "grad_norm": 7.190378189086914,
      "learning_rate": 9.124544700539664e-06,
      "loss": 0.1105,
      "step": 3860
    },
    {
      "epoch": 0.19938032532920216,
      "grad_norm": 4.396107196807861,
      "learning_rate": 9.124081488734173e-06,
      "loss": 0.0881,
      "step": 3861
    },
    {
      "epoch": 0.19943196488510198,
      "grad_norm": 10.650915145874023,
      "learning_rate": 9.123618166180286e-06,
      "loss": 0.11,
      "step": 3862
    },
    {
      "epoch": 0.1994836044410018,
      "grad_norm": 6.877256393432617,
      "learning_rate": 9.123154732890443e-06,
      "loss": 0.1063,
      "step": 3863
    },
    {
      "epoch": 0.19953524399690162,
      "grad_norm": 2.6463892459869385,
      "learning_rate": 9.12269118887709e-06,
      "loss": 0.0525,
      "step": 3864
    },
    {
      "epoch": 0.19958688355280144,
      "grad_norm": 5.48792028427124,
      "learning_rate": 9.12222753415268e-06,
      "loss": 0.1106,
      "step": 3865
    },
    {
      "epoch": 0.19963852310870125,
      "grad_norm": 3.4898788928985596,
      "learning_rate": 9.121763768729654e-06,
      "loss": 0.0737,
      "step": 3866
    },
    {
      "epoch": 0.19969016266460107,
      "grad_norm": 2.716735363006592,
      "learning_rate": 9.121299892620475e-06,
      "loss": 0.0677,
      "step": 3867
    },
    {
      "epoch": 0.1997418022205009,
      "grad_norm": 5.35215425491333,
      "learning_rate": 9.120835905837593e-06,
      "loss": 0.0961,
      "step": 3868
    },
    {
      "epoch": 0.1997934417764007,
      "grad_norm": 4.874456405639648,
      "learning_rate": 9.120371808393473e-06,
      "loss": 0.0754,
      "step": 3869
    },
    {
      "epoch": 0.19984508133230056,
      "grad_norm": 10.116751670837402,
      "learning_rate": 9.119907600300577e-06,
      "loss": 0.0995,
      "step": 3870
    },
    {
      "epoch": 0.19989672088820037,
      "grad_norm": 9.970123291015625,
      "learning_rate": 9.11944328157137e-06,
      "loss": 0.1923,
      "step": 3871
    },
    {
      "epoch": 0.1999483604441002,
      "grad_norm": 3.505636692047119,
      "learning_rate": 9.11897885221832e-06,
      "loss": 0.086,
      "step": 3872
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.049424171447754,
      "learning_rate": 9.1185143122539e-06,
      "loss": 0.0706,
      "step": 3873
    },
    {
      "epoch": 0.20005163955589983,
      "grad_norm": 7.462953090667725,
      "learning_rate": 9.118049661690584e-06,
      "loss": 0.114,
      "step": 3874
    },
    {
      "epoch": 0.20005163955589983,
      "eval_loss": 0.08347094804048538,
      "eval_runtime": 34.1906,
      "eval_samples_per_second": 45.773,
      "eval_steps_per_second": 5.733,
      "step": 3874
    },
    {
      "epoch": 0.20010327911179965,
      "grad_norm": 4.219438076019287,
      "learning_rate": 9.117584900540852e-06,
      "loss": 0.0902,
      "step": 3875
    },
    {
      "epoch": 0.20015491866769947,
      "grad_norm": 4.803033828735352,
      "learning_rate": 9.117120028817182e-06,
      "loss": 0.0558,
      "step": 3876
    },
    {
      "epoch": 0.20020655822359928,
      "grad_norm": 20.905906677246094,
      "learning_rate": 9.116655046532059e-06,
      "loss": 0.1573,
      "step": 3877
    },
    {
      "epoch": 0.2002581977794991,
      "grad_norm": 5.40611457824707,
      "learning_rate": 9.116189953697969e-06,
      "loss": 0.0602,
      "step": 3878
    },
    {
      "epoch": 0.20030983733539892,
      "grad_norm": 2.961488962173462,
      "learning_rate": 9.115724750327405e-06,
      "loss": 0.0545,
      "step": 3879
    },
    {
      "epoch": 0.20036147689129874,
      "grad_norm": 14.403563499450684,
      "learning_rate": 9.115259436432852e-06,
      "loss": 0.2147,
      "step": 3880
    },
    {
      "epoch": 0.20041311644719856,
      "grad_norm": 9.647571563720703,
      "learning_rate": 9.114794012026812e-06,
      "loss": 0.0535,
      "step": 3881
    },
    {
      "epoch": 0.20046475600309838,
      "grad_norm": 15.315519332885742,
      "learning_rate": 9.114328477121784e-06,
      "loss": 0.1612,
      "step": 3882
    },
    {
      "epoch": 0.2005163955589982,
      "grad_norm": 4.697983741760254,
      "learning_rate": 9.113862831730264e-06,
      "loss": 0.074,
      "step": 3883
    },
    {
      "epoch": 0.200568035114898,
      "grad_norm": 7.183439254760742,
      "learning_rate": 9.113397075864763e-06,
      "loss": 0.1065,
      "step": 3884
    },
    {
      "epoch": 0.20061967467079783,
      "grad_norm": 3.567199230194092,
      "learning_rate": 9.112931209537783e-06,
      "loss": 0.0625,
      "step": 3885
    },
    {
      "epoch": 0.20067131422669765,
      "grad_norm": 8.0267915725708,
      "learning_rate": 9.112465232761837e-06,
      "loss": 0.0985,
      "step": 3886
    },
    {
      "epoch": 0.20072295378259747,
      "grad_norm": 4.955127239227295,
      "learning_rate": 9.111999145549438e-06,
      "loss": 0.0531,
      "step": 3887
    },
    {
      "epoch": 0.2007745933384973,
      "grad_norm": 7.996601581573486,
      "learning_rate": 9.111532947913102e-06,
      "loss": 0.0898,
      "step": 3888
    },
    {
      "epoch": 0.2008262328943971,
      "grad_norm": 6.773326396942139,
      "learning_rate": 9.11106663986535e-06,
      "loss": 0.0861,
      "step": 3889
    },
    {
      "epoch": 0.20087787245029692,
      "grad_norm": 11.823284149169922,
      "learning_rate": 9.110600221418702e-06,
      "loss": 0.0792,
      "step": 3890
    },
    {
      "epoch": 0.20092951200619674,
      "grad_norm": 6.400703430175781,
      "learning_rate": 9.110133692585686e-06,
      "loss": 0.0941,
      "step": 3891
    },
    {
      "epoch": 0.20098115156209656,
      "grad_norm": 5.84575891494751,
      "learning_rate": 9.109667053378825e-06,
      "loss": 0.049,
      "step": 3892
    },
    {
      "epoch": 0.20103279111799638,
      "grad_norm": 2.559417486190796,
      "learning_rate": 9.109200303810656e-06,
      "loss": 0.0555,
      "step": 3893
    },
    {
      "epoch": 0.2010844306738962,
      "grad_norm": 6.854331970214844,
      "learning_rate": 9.10873344389371e-06,
      "loss": 0.084,
      "step": 3894
    },
    {
      "epoch": 0.20113607022979602,
      "grad_norm": 11.634614944458008,
      "learning_rate": 9.108266473640525e-06,
      "loss": 0.103,
      "step": 3895
    },
    {
      "epoch": 0.20118770978569583,
      "grad_norm": 2.4311747550964355,
      "learning_rate": 9.10779939306364e-06,
      "loss": 0.0361,
      "step": 3896
    },
    {
      "epoch": 0.20123934934159565,
      "grad_norm": 8.147890090942383,
      "learning_rate": 9.1073322021756e-06,
      "loss": 0.1283,
      "step": 3897
    },
    {
      "epoch": 0.20129098889749547,
      "grad_norm": 3.6209726333618164,
      "learning_rate": 9.106864900988948e-06,
      "loss": 0.0521,
      "step": 3898
    },
    {
      "epoch": 0.2013426284533953,
      "grad_norm": 3.7764313220977783,
      "learning_rate": 9.106397489516237e-06,
      "loss": 0.0397,
      "step": 3899
    },
    {
      "epoch": 0.2013942680092951,
      "grad_norm": 10.588961601257324,
      "learning_rate": 9.105929967770016e-06,
      "loss": 0.0632,
      "step": 3900
    },
    {
      "epoch": 0.20144590756519493,
      "grad_norm": 1.8026103973388672,
      "learning_rate": 9.105462335762841e-06,
      "loss": 0.0177,
      "step": 3901
    },
    {
      "epoch": 0.20149754712109477,
      "grad_norm": 9.739554405212402,
      "learning_rate": 9.104994593507267e-06,
      "loss": 0.0941,
      "step": 3902
    },
    {
      "epoch": 0.2015491866769946,
      "grad_norm": 8.868182182312012,
      "learning_rate": 9.10452674101586e-06,
      "loss": 0.065,
      "step": 3903
    },
    {
      "epoch": 0.2016008262328944,
      "grad_norm": 24.559839248657227,
      "learning_rate": 9.104058778301179e-06,
      "loss": 0.1449,
      "step": 3904
    },
    {
      "epoch": 0.20165246578879423,
      "grad_norm": 4.3440093994140625,
      "learning_rate": 9.103590705375794e-06,
      "loss": 0.0431,
      "step": 3905
    },
    {
      "epoch": 0.20170410534469405,
      "grad_norm": 8.4173583984375,
      "learning_rate": 9.103122522252271e-06,
      "loss": 0.0349,
      "step": 3906
    },
    {
      "epoch": 0.20175574490059386,
      "grad_norm": 9.686970710754395,
      "learning_rate": 9.102654228943186e-06,
      "loss": 0.1267,
      "step": 3907
    },
    {
      "epoch": 0.20180738445649368,
      "grad_norm": 18.93001937866211,
      "learning_rate": 9.102185825461113e-06,
      "loss": 0.0805,
      "step": 3908
    },
    {
      "epoch": 0.2018590240123935,
      "grad_norm": 7.026850700378418,
      "learning_rate": 9.101717311818631e-06,
      "loss": 0.0936,
      "step": 3909
    },
    {
      "epoch": 0.20191066356829332,
      "grad_norm": 8.623259544372559,
      "learning_rate": 9.101248688028324e-06,
      "loss": 0.1727,
      "step": 3910
    },
    {
      "epoch": 0.20196230312419314,
      "grad_norm": 5.801090717315674,
      "learning_rate": 9.100779954102772e-06,
      "loss": 0.0383,
      "step": 3911
    },
    {
      "epoch": 0.20201394268009296,
      "grad_norm": 6.58200740814209,
      "learning_rate": 9.100311110054566e-06,
      "loss": 0.0561,
      "step": 3912
    },
    {
      "epoch": 0.20206558223599277,
      "grad_norm": 37.6772346496582,
      "learning_rate": 9.099842155896292e-06,
      "loss": 0.1002,
      "step": 3913
    },
    {
      "epoch": 0.2021172217918926,
      "grad_norm": 2.1907777786254883,
      "learning_rate": 9.099373091640546e-06,
      "loss": 0.0349,
      "step": 3914
    },
    {
      "epoch": 0.2021688613477924,
      "grad_norm": 5.517948627471924,
      "learning_rate": 9.098903917299928e-06,
      "loss": 0.0757,
      "step": 3915
    },
    {
      "epoch": 0.20222050090369223,
      "grad_norm": 3.989839792251587,
      "learning_rate": 9.09843463288703e-06,
      "loss": 0.052,
      "step": 3916
    },
    {
      "epoch": 0.20227214045959205,
      "grad_norm": 6.0631818771362305,
      "learning_rate": 9.09796523841446e-06,
      "loss": 0.1342,
      "step": 3917
    },
    {
      "epoch": 0.20232378001549187,
      "grad_norm": 3.718484878540039,
      "learning_rate": 9.09749573389482e-06,
      "loss": 0.0321,
      "step": 3918
    },
    {
      "epoch": 0.20237541957139168,
      "grad_norm": 4.032725811004639,
      "learning_rate": 9.097026119340719e-06,
      "loss": 0.0683,
      "step": 3919
    },
    {
      "epoch": 0.2024270591272915,
      "grad_norm": 9.898628234863281,
      "learning_rate": 9.096556394764769e-06,
      "loss": 0.1041,
      "step": 3920
    },
    {
      "epoch": 0.20247869868319132,
      "grad_norm": 4.259200096130371,
      "learning_rate": 9.096086560179582e-06,
      "loss": 0.0605,
      "step": 3921
    },
    {
      "epoch": 0.20253033823909114,
      "grad_norm": 4.074065685272217,
      "learning_rate": 9.095616615597776e-06,
      "loss": 0.039,
      "step": 3922
    },
    {
      "epoch": 0.20258197779499096,
      "grad_norm": 11.68131160736084,
      "learning_rate": 9.09514656103197e-06,
      "loss": 0.1133,
      "step": 3923
    },
    {
      "epoch": 0.20263361735089078,
      "grad_norm": 2.7946231365203857,
      "learning_rate": 9.09467639649479e-06,
      "loss": 0.0595,
      "step": 3924
    },
    {
      "epoch": 0.2026852569067906,
      "grad_norm": 12.559158325195312,
      "learning_rate": 9.094206121998862e-06,
      "loss": 0.1069,
      "step": 3925
    },
    {
      "epoch": 0.2027368964626904,
      "grad_norm": 7.774507522583008,
      "learning_rate": 9.093735737556808e-06,
      "loss": 0.0798,
      "step": 3926
    },
    {
      "epoch": 0.20278853601859023,
      "grad_norm": 8.570439338684082,
      "learning_rate": 9.093265243181268e-06,
      "loss": 0.0671,
      "step": 3927
    },
    {
      "epoch": 0.20284017557449005,
      "grad_norm": 17.547626495361328,
      "learning_rate": 9.09279463888487e-06,
      "loss": 0.227,
      "step": 3928
    },
    {
      "epoch": 0.20289181513038987,
      "grad_norm": 2.847590446472168,
      "learning_rate": 9.092323924680257e-06,
      "loss": 0.0409,
      "step": 3929
    },
    {
      "epoch": 0.2029434546862897,
      "grad_norm": 2.3174867630004883,
      "learning_rate": 9.091853100580067e-06,
      "loss": 0.0553,
      "step": 3930
    },
    {
      "epoch": 0.2029950942421895,
      "grad_norm": 3.172971248626709,
      "learning_rate": 9.091382166596942e-06,
      "loss": 0.1,
      "step": 3931
    },
    {
      "epoch": 0.20304673379808932,
      "grad_norm": 2.7981112003326416,
      "learning_rate": 9.09091112274353e-06,
      "loss": 0.0536,
      "step": 3932
    },
    {
      "epoch": 0.20309837335398914,
      "grad_norm": 4.486863613128662,
      "learning_rate": 9.090439969032484e-06,
      "loss": 0.0701,
      "step": 3933
    },
    {
      "epoch": 0.203150012909889,
      "grad_norm": 7.037099361419678,
      "learning_rate": 9.089968705476449e-06,
      "loss": 0.0985,
      "step": 3934
    },
    {
      "epoch": 0.2032016524657888,
      "grad_norm": 24.914844512939453,
      "learning_rate": 9.089497332088086e-06,
      "loss": 0.2825,
      "step": 3935
    },
    {
      "epoch": 0.20325329202168863,
      "grad_norm": 10.528411865234375,
      "learning_rate": 9.089025848880052e-06,
      "loss": 0.1304,
      "step": 3936
    },
    {
      "epoch": 0.20330493157758844,
      "grad_norm": 11.362388610839844,
      "learning_rate": 9.08855425586501e-06,
      "loss": 0.1241,
      "step": 3937
    },
    {
      "epoch": 0.20335657113348826,
      "grad_norm": 2.8221287727355957,
      "learning_rate": 9.088082553055619e-06,
      "loss": 0.0473,
      "step": 3938
    },
    {
      "epoch": 0.20340821068938808,
      "grad_norm": 9.075616836547852,
      "learning_rate": 9.087610740464552e-06,
      "loss": 0.1124,
      "step": 3939
    },
    {
      "epoch": 0.2034598502452879,
      "grad_norm": 3.2360880374908447,
      "learning_rate": 9.087138818104475e-06,
      "loss": 0.0875,
      "step": 3940
    },
    {
      "epoch": 0.20351148980118772,
      "grad_norm": 3.080658197402954,
      "learning_rate": 9.086666785988063e-06,
      "loss": 0.0792,
      "step": 3941
    },
    {
      "epoch": 0.20356312935708754,
      "grad_norm": 4.294335842132568,
      "learning_rate": 9.086194644127992e-06,
      "loss": 0.068,
      "step": 3942
    },
    {
      "epoch": 0.20361476891298735,
      "grad_norm": 4.405970573425293,
      "learning_rate": 9.085722392536938e-06,
      "loss": 0.0651,
      "step": 3943
    },
    {
      "epoch": 0.20366640846888717,
      "grad_norm": 3.738111734390259,
      "learning_rate": 9.085250031227588e-06,
      "loss": 0.0781,
      "step": 3944
    },
    {
      "epoch": 0.203718048024787,
      "grad_norm": 4.456640720367432,
      "learning_rate": 9.084777560212623e-06,
      "loss": 0.0873,
      "step": 3945
    },
    {
      "epoch": 0.2037696875806868,
      "grad_norm": 12.278152465820312,
      "learning_rate": 9.084304979504732e-06,
      "loss": 0.1032,
      "step": 3946
    },
    {
      "epoch": 0.20382132713658663,
      "grad_norm": 8.699244499206543,
      "learning_rate": 9.083832289116606e-06,
      "loss": 0.1268,
      "step": 3947
    },
    {
      "epoch": 0.20387296669248645,
      "grad_norm": 3.120864152908325,
      "learning_rate": 9.083359489060936e-06,
      "loss": 0.0322,
      "step": 3948
    },
    {
      "epoch": 0.20392460624838626,
      "grad_norm": 3.752922773361206,
      "learning_rate": 9.082886579350423e-06,
      "loss": 0.0763,
      "step": 3949
    },
    {
      "epoch": 0.20397624580428608,
      "grad_norm": 6.6780595779418945,
      "learning_rate": 9.082413559997764e-06,
      "loss": 0.1222,
      "step": 3950
    },
    {
      "epoch": 0.2040278853601859,
      "grad_norm": 5.766803741455078,
      "learning_rate": 9.081940431015663e-06,
      "loss": 0.0837,
      "step": 3951
    },
    {
      "epoch": 0.20407952491608572,
      "grad_norm": 3.613201856613159,
      "learning_rate": 9.081467192416823e-06,
      "loss": 0.0798,
      "step": 3952
    },
    {
      "epoch": 0.20413116447198554,
      "grad_norm": 6.839271068572998,
      "learning_rate": 9.080993844213955e-06,
      "loss": 0.0665,
      "step": 3953
    },
    {
      "epoch": 0.20418280402788536,
      "grad_norm": 5.227346897125244,
      "learning_rate": 9.080520386419766e-06,
      "loss": 0.0455,
      "step": 3954
    },
    {
      "epoch": 0.20423444358378517,
      "grad_norm": 3.075085163116455,
      "learning_rate": 9.080046819046977e-06,
      "loss": 0.0669,
      "step": 3955
    },
    {
      "epoch": 0.204286083139685,
      "grad_norm": 4.393484592437744,
      "learning_rate": 9.0795731421083e-06,
      "loss": 0.0837,
      "step": 3956
    },
    {
      "epoch": 0.2043377226955848,
      "grad_norm": 3.4746668338775635,
      "learning_rate": 9.079099355616456e-06,
      "loss": 0.0622,
      "step": 3957
    },
    {
      "epoch": 0.20438936225148463,
      "grad_norm": 15.9212646484375,
      "learning_rate": 9.07862545958417e-06,
      "loss": 0.1254,
      "step": 3958
    },
    {
      "epoch": 0.20444100180738445,
      "grad_norm": 5.059668064117432,
      "learning_rate": 9.078151454024166e-06,
      "loss": 0.0518,
      "step": 3959
    },
    {
      "epoch": 0.20449264136328427,
      "grad_norm": 3.830806016921997,
      "learning_rate": 9.077677338949175e-06,
      "loss": 0.0643,
      "step": 3960
    },
    {
      "epoch": 0.20454428091918408,
      "grad_norm": 3.6856260299682617,
      "learning_rate": 9.077203114371926e-06,
      "loss": 0.0474,
      "step": 3961
    },
    {
      "epoch": 0.2045959204750839,
      "grad_norm": 2.594212293624878,
      "learning_rate": 9.076728780305158e-06,
      "loss": 0.0618,
      "step": 3962
    },
    {
      "epoch": 0.20464756003098372,
      "grad_norm": 11.77745246887207,
      "learning_rate": 9.076254336761603e-06,
      "loss": 0.1228,
      "step": 3963
    },
    {
      "epoch": 0.20469919958688354,
      "grad_norm": 2.724634885787964,
      "learning_rate": 9.075779783754009e-06,
      "loss": 0.0568,
      "step": 3964
    },
    {
      "epoch": 0.20475083914278336,
      "grad_norm": 7.978205680847168,
      "learning_rate": 9.075305121295113e-06,
      "loss": 0.0681,
      "step": 3965
    },
    {
      "epoch": 0.2048024786986832,
      "grad_norm": 3.093806505203247,
      "learning_rate": 9.074830349397666e-06,
      "loss": 0.0586,
      "step": 3966
    },
    {
      "epoch": 0.20485411825458302,
      "grad_norm": 4.616779327392578,
      "learning_rate": 9.074355468074416e-06,
      "loss": 0.0349,
      "step": 3967
    },
    {
      "epoch": 0.20490575781048284,
      "grad_norm": 9.98720932006836,
      "learning_rate": 9.073880477338114e-06,
      "loss": 0.168,
      "step": 3968
    },
    {
      "epoch": 0.20495739736638266,
      "grad_norm": 6.075533390045166,
      "learning_rate": 9.073405377201519e-06,
      "loss": 0.0601,
      "step": 3969
    },
    {
      "epoch": 0.20500903692228248,
      "grad_norm": 11.39665699005127,
      "learning_rate": 9.072930167677385e-06,
      "loss": 0.2099,
      "step": 3970
    },
    {
      "epoch": 0.2050606764781823,
      "grad_norm": 4.887941360473633,
      "learning_rate": 9.072454848778478e-06,
      "loss": 0.0681,
      "step": 3971
    },
    {
      "epoch": 0.20511231603408211,
      "grad_norm": 2.9001059532165527,
      "learning_rate": 9.071979420517558e-06,
      "loss": 0.0435,
      "step": 3972
    },
    {
      "epoch": 0.20516395558998193,
      "grad_norm": 5.5125627517700195,
      "learning_rate": 9.071503882907396e-06,
      "loss": 0.0851,
      "step": 3973
    },
    {
      "epoch": 0.20521559514588175,
      "grad_norm": 4.522228240966797,
      "learning_rate": 9.07102823596076e-06,
      "loss": 0.0935,
      "step": 3974
    },
    {
      "epoch": 0.20526723470178157,
      "grad_norm": 10.672245979309082,
      "learning_rate": 9.070552479690421e-06,
      "loss": 0.1069,
      "step": 3975
    },
    {
      "epoch": 0.2053188742576814,
      "grad_norm": 2.974382162094116,
      "learning_rate": 9.070076614109159e-06,
      "loss": 0.0185,
      "step": 3976
    },
    {
      "epoch": 0.2053705138135812,
      "grad_norm": 5.233972549438477,
      "learning_rate": 9.069600639229752e-06,
      "loss": 0.0551,
      "step": 3977
    },
    {
      "epoch": 0.20542215336948103,
      "grad_norm": 8.973264694213867,
      "learning_rate": 9.069124555064978e-06,
      "loss": 0.0819,
      "step": 3978
    },
    {
      "epoch": 0.20547379292538084,
      "grad_norm": 3.645474910736084,
      "learning_rate": 9.068648361627628e-06,
      "loss": 0.0776,
      "step": 3979
    },
    {
      "epoch": 0.20552543248128066,
      "grad_norm": 9.975993156433105,
      "learning_rate": 9.068172058930485e-06,
      "loss": 0.1269,
      "step": 3980
    },
    {
      "epoch": 0.20557707203718048,
      "grad_norm": 5.540329456329346,
      "learning_rate": 9.067695646986341e-06,
      "loss": 0.0557,
      "step": 3981
    },
    {
      "epoch": 0.2056287115930803,
      "grad_norm": 10.52869987487793,
      "learning_rate": 9.067219125807992e-06,
      "loss": 0.1589,
      "step": 3982
    },
    {
      "epoch": 0.20568035114898012,
      "grad_norm": 6.331674098968506,
      "learning_rate": 9.066742495408231e-06,
      "loss": 0.0766,
      "step": 3983
    },
    {
      "epoch": 0.20573199070487994,
      "grad_norm": 6.140235424041748,
      "learning_rate": 9.06626575579986e-06,
      "loss": 0.0708,
      "step": 3984
    },
    {
      "epoch": 0.20578363026077975,
      "grad_norm": 3.2201220989227295,
      "learning_rate": 9.065788906995678e-06,
      "loss": 0.0427,
      "step": 3985
    },
    {
      "epoch": 0.20583526981667957,
      "grad_norm": 5.187060832977295,
      "learning_rate": 9.065311949008494e-06,
      "loss": 0.0666,
      "step": 3986
    },
    {
      "epoch": 0.2058869093725794,
      "grad_norm": 10.204495429992676,
      "learning_rate": 9.064834881851114e-06,
      "loss": 0.0966,
      "step": 3987
    },
    {
      "epoch": 0.2059385489284792,
      "grad_norm": 2.177490234375,
      "learning_rate": 9.06435770553635e-06,
      "loss": 0.062,
      "step": 3988
    },
    {
      "epoch": 0.20599018848437903,
      "grad_norm": 3.3154141902923584,
      "learning_rate": 9.063880420077018e-06,
      "loss": 0.0971,
      "step": 3989
    },
    {
      "epoch": 0.20604182804027885,
      "grad_norm": 4.436887741088867,
      "learning_rate": 9.063403025485932e-06,
      "loss": 0.0966,
      "step": 3990
    },
    {
      "epoch": 0.20609346759617866,
      "grad_norm": 5.7286295890808105,
      "learning_rate": 9.062925521775913e-06,
      "loss": 0.0861,
      "step": 3991
    },
    {
      "epoch": 0.20614510715207848,
      "grad_norm": 4.290014266967773,
      "learning_rate": 9.062447908959785e-06,
      "loss": 0.0997,
      "step": 3992
    },
    {
      "epoch": 0.2061967467079783,
      "grad_norm": 4.1341633796691895,
      "learning_rate": 9.061970187050372e-06,
      "loss": 0.0496,
      "step": 3993
    },
    {
      "epoch": 0.20624838626387812,
      "grad_norm": 5.251471519470215,
      "learning_rate": 9.061492356060505e-06,
      "loss": 0.0919,
      "step": 3994
    },
    {
      "epoch": 0.20630002581977794,
      "grad_norm": 4.815141201019287,
      "learning_rate": 9.061014416003014e-06,
      "loss": 0.0679,
      "step": 3995
    },
    {
      "epoch": 0.20635166537567776,
      "grad_norm": 3.4982964992523193,
      "learning_rate": 9.060536366890733e-06,
      "loss": 0.066,
      "step": 3996
    },
    {
      "epoch": 0.20640330493157757,
      "grad_norm": 2.3778457641601562,
      "learning_rate": 9.060058208736503e-06,
      "loss": 0.0338,
      "step": 3997
    },
    {
      "epoch": 0.20645494448747742,
      "grad_norm": 5.809787750244141,
      "learning_rate": 9.05957994155316e-06,
      "loss": 0.0895,
      "step": 3998
    },
    {
      "epoch": 0.20650658404337724,
      "grad_norm": 3.4143006801605225,
      "learning_rate": 9.059101565353552e-06,
      "loss": 0.0583,
      "step": 3999
    },
    {
      "epoch": 0.20655822359927706,
      "grad_norm": 10.63387393951416,
      "learning_rate": 9.058623080150521e-06,
      "loss": 0.1028,
      "step": 4000
    },
    {
      "epoch": 0.20660986315517688,
      "grad_norm": 18.944480895996094,
      "learning_rate": 9.05814448595692e-06,
      "loss": 0.1086,
      "step": 4001
    },
    {
      "epoch": 0.2066615027110767,
      "grad_norm": 11.526955604553223,
      "learning_rate": 9.057665782785597e-06,
      "loss": 0.2143,
      "step": 4002
    },
    {
      "epoch": 0.2067131422669765,
      "grad_norm": 12.632859230041504,
      "learning_rate": 9.057186970649412e-06,
      "loss": 0.0899,
      "step": 4003
    },
    {
      "epoch": 0.20676478182287633,
      "grad_norm": 9.473052978515625,
      "learning_rate": 9.056708049561219e-06,
      "loss": 0.1019,
      "step": 4004
    },
    {
      "epoch": 0.20681642137877615,
      "grad_norm": 4.273702621459961,
      "learning_rate": 9.056229019533882e-06,
      "loss": 0.0834,
      "step": 4005
    },
    {
      "epoch": 0.20686806093467597,
      "grad_norm": 7.8446221351623535,
      "learning_rate": 9.055749880580263e-06,
      "loss": 0.1373,
      "step": 4006
    },
    {
      "epoch": 0.2069197004905758,
      "grad_norm": 14.136872291564941,
      "learning_rate": 9.05527063271323e-06,
      "loss": 0.1801,
      "step": 4007
    },
    {
      "epoch": 0.2069713400464756,
      "grad_norm": 5.625674724578857,
      "learning_rate": 9.054791275945649e-06,
      "loss": 0.0664,
      "step": 4008
    },
    {
      "epoch": 0.20702297960237542,
      "grad_norm": 2.4002654552459717,
      "learning_rate": 9.054311810290399e-06,
      "loss": 0.0574,
      "step": 4009
    },
    {
      "epoch": 0.20707461915827524,
      "grad_norm": 4.16328239440918,
      "learning_rate": 9.05383223576035e-06,
      "loss": 0.0845,
      "step": 4010
    },
    {
      "epoch": 0.20712625871417506,
      "grad_norm": 8.727642059326172,
      "learning_rate": 9.053352552368387e-06,
      "loss": 0.1209,
      "step": 4011
    },
    {
      "epoch": 0.20717789827007488,
      "grad_norm": 2.378227472305298,
      "learning_rate": 9.052872760127385e-06,
      "loss": 0.0347,
      "step": 4012
    },
    {
      "epoch": 0.2072295378259747,
      "grad_norm": 6.031287670135498,
      "learning_rate": 9.05239285905023e-06,
      "loss": 0.1203,
      "step": 4013
    },
    {
      "epoch": 0.20728117738187452,
      "grad_norm": 5.658989429473877,
      "learning_rate": 9.051912849149811e-06,
      "loss": 0.0993,
      "step": 4014
    },
    {
      "epoch": 0.20733281693777433,
      "grad_norm": 2.752939462661743,
      "learning_rate": 9.051432730439017e-06,
      "loss": 0.0513,
      "step": 4015
    },
    {
      "epoch": 0.20738445649367415,
      "grad_norm": 12.55160903930664,
      "learning_rate": 9.050952502930741e-06,
      "loss": 0.0999,
      "step": 4016
    },
    {
      "epoch": 0.20743609604957397,
      "grad_norm": 22.846773147583008,
      "learning_rate": 9.05047216663788e-06,
      "loss": 0.2784,
      "step": 4017
    },
    {
      "epoch": 0.2074877356054738,
      "grad_norm": 4.902595520019531,
      "learning_rate": 9.049991721573334e-06,
      "loss": 0.0762,
      "step": 4018
    },
    {
      "epoch": 0.2075393751613736,
      "grad_norm": 16.415908813476562,
      "learning_rate": 9.049511167750002e-06,
      "loss": 0.1605,
      "step": 4019
    },
    {
      "epoch": 0.20759101471727343,
      "grad_norm": 5.2961578369140625,
      "learning_rate": 9.04903050518079e-06,
      "loss": 0.0921,
      "step": 4020
    },
    {
      "epoch": 0.20764265427317324,
      "grad_norm": 3.165499210357666,
      "learning_rate": 9.048549733878606e-06,
      "loss": 0.0796,
      "step": 4021
    },
    {
      "epoch": 0.20769429382907306,
      "grad_norm": 2.638352155685425,
      "learning_rate": 9.048068853856362e-06,
      "loss": 0.0709,
      "step": 4022
    },
    {
      "epoch": 0.20774593338497288,
      "grad_norm": 5.913818836212158,
      "learning_rate": 9.04758786512697e-06,
      "loss": 0.091,
      "step": 4023
    },
    {
      "epoch": 0.2077975729408727,
      "grad_norm": 3.514751434326172,
      "learning_rate": 9.047106767703345e-06,
      "loss": 0.0611,
      "step": 4024
    },
    {
      "epoch": 0.20784921249677252,
      "grad_norm": 2.936540126800537,
      "learning_rate": 9.04662556159841e-06,
      "loss": 0.07,
      "step": 4025
    },
    {
      "epoch": 0.20790085205267234,
      "grad_norm": 3.259289503097534,
      "learning_rate": 9.046144246825084e-06,
      "loss": 0.0688,
      "step": 4026
    },
    {
      "epoch": 0.20795249160857215,
      "grad_norm": 2.8948488235473633,
      "learning_rate": 9.045662823396295e-06,
      "loss": 0.0514,
      "step": 4027
    },
    {
      "epoch": 0.20800413116447197,
      "grad_norm": 7.927031993865967,
      "learning_rate": 9.045181291324969e-06,
      "loss": 0.0864,
      "step": 4028
    },
    {
      "epoch": 0.2080557707203718,
      "grad_norm": 3.4334590435028076,
      "learning_rate": 9.044699650624039e-06,
      "loss": 0.052,
      "step": 4029
    },
    {
      "epoch": 0.20810741027627164,
      "grad_norm": 5.78835916519165,
      "learning_rate": 9.044217901306438e-06,
      "loss": 0.0777,
      "step": 4030
    },
    {
      "epoch": 0.20815904983217146,
      "grad_norm": 8.200758934020996,
      "learning_rate": 9.043736043385103e-06,
      "loss": 0.1579,
      "step": 4031
    },
    {
      "epoch": 0.20821068938807127,
      "grad_norm": 27.97536849975586,
      "learning_rate": 9.043254076872974e-06,
      "loss": 0.2258,
      "step": 4032
    },
    {
      "epoch": 0.2082623289439711,
      "grad_norm": 5.726352214813232,
      "learning_rate": 9.042772001782993e-06,
      "loss": 0.1208,
      "step": 4033
    },
    {
      "epoch": 0.2083139684998709,
      "grad_norm": 3.1503708362579346,
      "learning_rate": 9.042289818128107e-06,
      "loss": 0.0593,
      "step": 4034
    },
    {
      "epoch": 0.20836560805577073,
      "grad_norm": 2.6368210315704346,
      "learning_rate": 9.041807525921262e-06,
      "loss": 0.0416,
      "step": 4035
    },
    {
      "epoch": 0.20841724761167055,
      "grad_norm": 6.29469108581543,
      "learning_rate": 9.041325125175414e-06,
      "loss": 0.0622,
      "step": 4036
    },
    {
      "epoch": 0.20846888716757037,
      "grad_norm": 2.277918577194214,
      "learning_rate": 9.040842615903514e-06,
      "loss": 0.0402,
      "step": 4037
    },
    {
      "epoch": 0.20852052672347018,
      "grad_norm": 15.42555046081543,
      "learning_rate": 9.04035999811852e-06,
      "loss": 0.0747,
      "step": 4038
    },
    {
      "epoch": 0.20857216627937,
      "grad_norm": 4.48771333694458,
      "learning_rate": 9.039877271833393e-06,
      "loss": 0.0539,
      "step": 4039
    },
    {
      "epoch": 0.20862380583526982,
      "grad_norm": 3.325486898422241,
      "learning_rate": 9.039394437061095e-06,
      "loss": 0.0298,
      "step": 4040
    },
    {
      "epoch": 0.20867544539116964,
      "grad_norm": 6.79789924621582,
      "learning_rate": 9.038911493814593e-06,
      "loss": 0.085,
      "step": 4041
    },
    {
      "epoch": 0.20872708494706946,
      "grad_norm": 7.036989688873291,
      "learning_rate": 9.038428442106854e-06,
      "loss": 0.073,
      "step": 4042
    },
    {
      "epoch": 0.20877872450296928,
      "grad_norm": 6.184421062469482,
      "learning_rate": 9.037945281950854e-06,
      "loss": 0.088,
      "step": 4043
    },
    {
      "epoch": 0.2088303640588691,
      "grad_norm": 6.421226501464844,
      "learning_rate": 9.037462013359565e-06,
      "loss": 0.0906,
      "step": 4044
    },
    {
      "epoch": 0.2088820036147689,
      "grad_norm": 5.882344722747803,
      "learning_rate": 9.036978636345965e-06,
      "loss": 0.0554,
      "step": 4045
    },
    {
      "epoch": 0.20893364317066873,
      "grad_norm": 2.4829132556915283,
      "learning_rate": 9.036495150923035e-06,
      "loss": 0.0396,
      "step": 4046
    },
    {
      "epoch": 0.20898528272656855,
      "grad_norm": 19.340417861938477,
      "learning_rate": 9.036011557103757e-06,
      "loss": 0.1842,
      "step": 4047
    },
    {
      "epoch": 0.20903692228246837,
      "grad_norm": 5.341668605804443,
      "learning_rate": 9.03552785490112e-06,
      "loss": 0.073,
      "step": 4048
    },
    {
      "epoch": 0.2090885618383682,
      "grad_norm": 4.081088066101074,
      "learning_rate": 9.035044044328112e-06,
      "loss": 0.0808,
      "step": 4049
    },
    {
      "epoch": 0.209140201394268,
      "grad_norm": 16.740665435791016,
      "learning_rate": 9.034560125397724e-06,
      "loss": 0.1776,
      "step": 4050
    },
    {
      "epoch": 0.20919184095016782,
      "grad_norm": 12.635180473327637,
      "learning_rate": 9.034076098122954e-06,
      "loss": 0.1811,
      "step": 4051
    },
    {
      "epoch": 0.20924348050606764,
      "grad_norm": 8.417553901672363,
      "learning_rate": 9.0335919625168e-06,
      "loss": 0.064,
      "step": 4052
    },
    {
      "epoch": 0.20929512006196746,
      "grad_norm": 9.60696792602539,
      "learning_rate": 9.03310771859226e-06,
      "loss": 0.132,
      "step": 4053
    },
    {
      "epoch": 0.20934675961786728,
      "grad_norm": 3.611917495727539,
      "learning_rate": 9.032623366362341e-06,
      "loss": 0.0745,
      "step": 4054
    },
    {
      "epoch": 0.2093983991737671,
      "grad_norm": 5.173318862915039,
      "learning_rate": 9.032138905840048e-06,
      "loss": 0.0734,
      "step": 4055
    },
    {
      "epoch": 0.20945003872966692,
      "grad_norm": 4.174980640411377,
      "learning_rate": 9.031654337038392e-06,
      "loss": 0.0797,
      "step": 4056
    },
    {
      "epoch": 0.20950167828556673,
      "grad_norm": 4.114012241363525,
      "learning_rate": 9.031169659970382e-06,
      "loss": 0.0757,
      "step": 4057
    },
    {
      "epoch": 0.20955331784146655,
      "grad_norm": 15.504814147949219,
      "learning_rate": 9.03068487464904e-06,
      "loss": 0.1646,
      "step": 4058
    },
    {
      "epoch": 0.20960495739736637,
      "grad_norm": 4.084572792053223,
      "learning_rate": 9.03019998108738e-06,
      "loss": 0.0435,
      "step": 4059
    },
    {
      "epoch": 0.2096565969532662,
      "grad_norm": 13.379566192626953,
      "learning_rate": 9.029714979298424e-06,
      "loss": 0.1142,
      "step": 4060
    },
    {
      "epoch": 0.20970823650916603,
      "grad_norm": 2.460561990737915,
      "learning_rate": 9.029229869295195e-06,
      "loss": 0.0657,
      "step": 4061
    },
    {
      "epoch": 0.20975987606506585,
      "grad_norm": 6.018482685089111,
      "learning_rate": 9.028744651090724e-06,
      "loss": 0.0505,
      "step": 4062
    },
    {
      "epoch": 0.20981151562096567,
      "grad_norm": 3.735577344894409,
      "learning_rate": 9.028259324698038e-06,
      "loss": 0.0509,
      "step": 4063
    },
    {
      "epoch": 0.2098631551768655,
      "grad_norm": 6.756983280181885,
      "learning_rate": 9.027773890130171e-06,
      "loss": 0.0807,
      "step": 4064
    },
    {
      "epoch": 0.2099147947327653,
      "grad_norm": 9.81044864654541,
      "learning_rate": 9.027288347400158e-06,
      "loss": 0.0716,
      "step": 4065
    },
    {
      "epoch": 0.20996643428866513,
      "grad_norm": 2.281407594680786,
      "learning_rate": 9.02680269652104e-06,
      "loss": 0.0664,
      "step": 4066
    },
    {
      "epoch": 0.21001807384456495,
      "grad_norm": 2.6603682041168213,
      "learning_rate": 9.026316937505857e-06,
      "loss": 0.0407,
      "step": 4067
    },
    {
      "epoch": 0.21006971340046476,
      "grad_norm": 2.9108517169952393,
      "learning_rate": 9.025831070367654e-06,
      "loss": 0.0432,
      "step": 4068
    },
    {
      "epoch": 0.21012135295636458,
      "grad_norm": 8.166243553161621,
      "learning_rate": 9.025345095119477e-06,
      "loss": 0.0589,
      "step": 4069
    },
    {
      "epoch": 0.2101729925122644,
      "grad_norm": 2.6088716983795166,
      "learning_rate": 9.024859011774378e-06,
      "loss": 0.0472,
      "step": 4070
    },
    {
      "epoch": 0.21022463206816422,
      "grad_norm": 4.749986171722412,
      "learning_rate": 9.02437282034541e-06,
      "loss": 0.0471,
      "step": 4071
    },
    {
      "epoch": 0.21027627162406404,
      "grad_norm": 6.04132604598999,
      "learning_rate": 9.02388652084563e-06,
      "loss": 0.08,
      "step": 4072
    },
    {
      "epoch": 0.21032791117996386,
      "grad_norm": 12.665468215942383,
      "learning_rate": 9.023400113288098e-06,
      "loss": 0.1062,
      "step": 4073
    },
    {
      "epoch": 0.21037955073586367,
      "grad_norm": 8.693053245544434,
      "learning_rate": 9.022913597685874e-06,
      "loss": 0.0641,
      "step": 4074
    },
    {
      "epoch": 0.2104311902917635,
      "grad_norm": 4.3802266120910645,
      "learning_rate": 9.02242697405202e-06,
      "loss": 0.0431,
      "step": 4075
    },
    {
      "epoch": 0.2104828298476633,
      "grad_norm": 22.491518020629883,
      "learning_rate": 9.02194024239961e-06,
      "loss": 0.2508,
      "step": 4076
    },
    {
      "epoch": 0.21053446940356313,
      "grad_norm": 5.990903854370117,
      "learning_rate": 9.021453402741713e-06,
      "loss": 0.0603,
      "step": 4077
    },
    {
      "epoch": 0.21058610895946295,
      "grad_norm": 8.935306549072266,
      "learning_rate": 9.0209664550914e-06,
      "loss": 0.1251,
      "step": 4078
    },
    {
      "epoch": 0.21063774851536277,
      "grad_norm": 10.69802474975586,
      "learning_rate": 9.02047939946175e-06,
      "loss": 0.0565,
      "step": 4079
    },
    {
      "epoch": 0.21068938807126258,
      "grad_norm": 3.553208112716675,
      "learning_rate": 9.01999223586584e-06,
      "loss": 0.0519,
      "step": 4080
    },
    {
      "epoch": 0.2107410276271624,
      "grad_norm": 16.326669692993164,
      "learning_rate": 9.019504964316754e-06,
      "loss": 0.1664,
      "step": 4081
    },
    {
      "epoch": 0.21079266718306222,
      "grad_norm": 4.935323238372803,
      "learning_rate": 9.019017584827579e-06,
      "loss": 0.0608,
      "step": 4082
    },
    {
      "epoch": 0.21084430673896204,
      "grad_norm": 6.999495983123779,
      "learning_rate": 9.0185300974114e-06,
      "loss": 0.1352,
      "step": 4083
    },
    {
      "epoch": 0.21089594629486186,
      "grad_norm": 3.85577654838562,
      "learning_rate": 9.01804250208131e-06,
      "loss": 0.0159,
      "step": 4084
    },
    {
      "epoch": 0.21094758585076168,
      "grad_norm": 14.368792533874512,
      "learning_rate": 9.0175547988504e-06,
      "loss": 0.0552,
      "step": 4085
    },
    {
      "epoch": 0.2109992254066615,
      "grad_norm": 4.448890209197998,
      "learning_rate": 9.017066987731771e-06,
      "loss": 0.0673,
      "step": 4086
    },
    {
      "epoch": 0.2110508649625613,
      "grad_norm": 6.989251613616943,
      "learning_rate": 9.01657906873852e-06,
      "loss": 0.097,
      "step": 4087
    },
    {
      "epoch": 0.21110250451846113,
      "grad_norm": 4.280754089355469,
      "learning_rate": 9.016091041883752e-06,
      "loss": 0.0748,
      "step": 4088
    },
    {
      "epoch": 0.21115414407436095,
      "grad_norm": 6.1869611740112305,
      "learning_rate": 9.01560290718057e-06,
      "loss": 0.0766,
      "step": 4089
    },
    {
      "epoch": 0.21120578363026077,
      "grad_norm": 11.338632583618164,
      "learning_rate": 9.015114664642082e-06,
      "loss": 0.0706,
      "step": 4090
    },
    {
      "epoch": 0.2112574231861606,
      "grad_norm": 3.2448062896728516,
      "learning_rate": 9.0146263142814e-06,
      "loss": 0.0555,
      "step": 4091
    },
    {
      "epoch": 0.2113090627420604,
      "grad_norm": 6.725411891937256,
      "learning_rate": 9.014137856111641e-06,
      "loss": 0.0851,
      "step": 4092
    },
    {
      "epoch": 0.21136070229796025,
      "grad_norm": 6.5563249588012695,
      "learning_rate": 9.013649290145919e-06,
      "loss": 0.076,
      "step": 4093
    },
    {
      "epoch": 0.21141234185386007,
      "grad_norm": 18.524765014648438,
      "learning_rate": 9.013160616397354e-06,
      "loss": 0.0421,
      "step": 4094
    },
    {
      "epoch": 0.2114639814097599,
      "grad_norm": 6.94788932800293,
      "learning_rate": 9.012671834879071e-06,
      "loss": 0.0497,
      "step": 4095
    },
    {
      "epoch": 0.2115156209656597,
      "grad_norm": 5.119638919830322,
      "learning_rate": 9.012182945604193e-06,
      "loss": 0.0484,
      "step": 4096
    },
    {
      "epoch": 0.21156726052155952,
      "grad_norm": 8.814032554626465,
      "learning_rate": 9.011693948585851e-06,
      "loss": 0.1302,
      "step": 4097
    },
    {
      "epoch": 0.21161890007745934,
      "grad_norm": 4.150890827178955,
      "learning_rate": 9.011204843837177e-06,
      "loss": 0.1038,
      "step": 4098
    },
    {
      "epoch": 0.21167053963335916,
      "grad_norm": 7.011745452880859,
      "learning_rate": 9.010715631371302e-06,
      "loss": 0.0962,
      "step": 4099
    },
    {
      "epoch": 0.21172217918925898,
      "grad_norm": 8.163985252380371,
      "learning_rate": 9.010226311201367e-06,
      "loss": 0.0807,
      "step": 4100
    },
    {
      "epoch": 0.2117738187451588,
      "grad_norm": 6.881673336029053,
      "learning_rate": 9.009736883340511e-06,
      "loss": 0.0947,
      "step": 4101
    },
    {
      "epoch": 0.21182545830105862,
      "grad_norm": 5.69691801071167,
      "learning_rate": 9.009247347801874e-06,
      "loss": 0.077,
      "step": 4102
    },
    {
      "epoch": 0.21187709785695844,
      "grad_norm": 3.6698360443115234,
      "learning_rate": 9.008757704598609e-06,
      "loss": 0.0413,
      "step": 4103
    },
    {
      "epoch": 0.21192873741285825,
      "grad_norm": 6.217745780944824,
      "learning_rate": 9.008267953743858e-06,
      "loss": 0.063,
      "step": 4104
    },
    {
      "epoch": 0.21198037696875807,
      "grad_norm": 3.0999460220336914,
      "learning_rate": 9.007778095250778e-06,
      "loss": 0.0401,
      "step": 4105
    },
    {
      "epoch": 0.2120320165246579,
      "grad_norm": 2.4368767738342285,
      "learning_rate": 9.007288129132519e-06,
      "loss": 0.0568,
      "step": 4106
    },
    {
      "epoch": 0.2120836560805577,
      "grad_norm": 19.42025375366211,
      "learning_rate": 9.00679805540224e-06,
      "loss": 0.2023,
      "step": 4107
    },
    {
      "epoch": 0.21213529563645753,
      "grad_norm": 13.746637344360352,
      "learning_rate": 9.006307874073107e-06,
      "loss": 0.2568,
      "step": 4108
    },
    {
      "epoch": 0.21218693519235735,
      "grad_norm": 6.20150089263916,
      "learning_rate": 9.005817585158276e-06,
      "loss": 0.0861,
      "step": 4109
    },
    {
      "epoch": 0.21223857474825716,
      "grad_norm": 3.653190851211548,
      "learning_rate": 9.005327188670916e-06,
      "loss": 0.0511,
      "step": 4110
    },
    {
      "epoch": 0.21229021430415698,
      "grad_norm": 11.70158863067627,
      "learning_rate": 9.004836684624196e-06,
      "loss": 0.0724,
      "step": 4111
    },
    {
      "epoch": 0.2123418538600568,
      "grad_norm": 1.9696931838989258,
      "learning_rate": 9.004346073031288e-06,
      "loss": 0.0436,
      "step": 4112
    },
    {
      "epoch": 0.21239349341595662,
      "grad_norm": 10.046014785766602,
      "learning_rate": 9.003855353905367e-06,
      "loss": 0.1164,
      "step": 4113
    },
    {
      "epoch": 0.21244513297185644,
      "grad_norm": 9.24923038482666,
      "learning_rate": 9.003364527259613e-06,
      "loss": 0.1007,
      "step": 4114
    },
    {
      "epoch": 0.21249677252775626,
      "grad_norm": 4.162918567657471,
      "learning_rate": 9.002873593107202e-06,
      "loss": 0.072,
      "step": 4115
    },
    {
      "epoch": 0.21254841208365607,
      "grad_norm": 5.090132713317871,
      "learning_rate": 9.002382551461322e-06,
      "loss": 0.056,
      "step": 4116
    },
    {
      "epoch": 0.2126000516395559,
      "grad_norm": 2.317049026489258,
      "learning_rate": 9.001891402335157e-06,
      "loss": 0.029,
      "step": 4117
    },
    {
      "epoch": 0.2126516911954557,
      "grad_norm": 6.844313144683838,
      "learning_rate": 9.001400145741897e-06,
      "loss": 0.0342,
      "step": 4118
    },
    {
      "epoch": 0.21270333075135553,
      "grad_norm": 15.77582836151123,
      "learning_rate": 9.000908781694734e-06,
      "loss": 0.1008,
      "step": 4119
    },
    {
      "epoch": 0.21275497030725535,
      "grad_norm": 5.921971797943115,
      "learning_rate": 9.000417310206863e-06,
      "loss": 0.0451,
      "step": 4120
    },
    {
      "epoch": 0.21280660986315517,
      "grad_norm": 5.449661731719971,
      "learning_rate": 8.999925731291484e-06,
      "loss": 0.0872,
      "step": 4121
    },
    {
      "epoch": 0.21285824941905498,
      "grad_norm": 5.426797389984131,
      "learning_rate": 8.999434044961795e-06,
      "loss": 0.0635,
      "step": 4122
    },
    {
      "epoch": 0.2129098889749548,
      "grad_norm": 7.848128795623779,
      "learning_rate": 8.998942251231002e-06,
      "loss": 0.1079,
      "step": 4123
    },
    {
      "epoch": 0.21296152853085462,
      "grad_norm": 6.26927375793457,
      "learning_rate": 8.99845035011231e-06,
      "loss": 0.113,
      "step": 4124
    },
    {
      "epoch": 0.21301316808675447,
      "grad_norm": 4.9820027351379395,
      "learning_rate": 8.997958341618929e-06,
      "loss": 0.0578,
      "step": 4125
    },
    {
      "epoch": 0.21306480764265429,
      "grad_norm": 3.5875463485717773,
      "learning_rate": 8.997466225764072e-06,
      "loss": 0.0753,
      "step": 4126
    },
    {
      "epoch": 0.2131164471985541,
      "grad_norm": 12.961711883544922,
      "learning_rate": 8.996974002560956e-06,
      "loss": 0.0763,
      "step": 4127
    },
    {
      "epoch": 0.21316808675445392,
      "grad_norm": 16.614768981933594,
      "learning_rate": 8.996481672022797e-06,
      "loss": 0.104,
      "step": 4128
    },
    {
      "epoch": 0.21321972631035374,
      "grad_norm": 7.693753242492676,
      "learning_rate": 8.995989234162814e-06,
      "loss": 0.0822,
      "step": 4129
    },
    {
      "epoch": 0.21327136586625356,
      "grad_norm": 6.693681716918945,
      "learning_rate": 8.995496688994234e-06,
      "loss": 0.0701,
      "step": 4130
    },
    {
      "epoch": 0.21332300542215338,
      "grad_norm": 16.710338592529297,
      "learning_rate": 8.995004036530283e-06,
      "loss": 0.1066,
      "step": 4131
    },
    {
      "epoch": 0.2133746449780532,
      "grad_norm": 6.787795066833496,
      "learning_rate": 8.994511276784192e-06,
      "loss": 0.0921,
      "step": 4132
    },
    {
      "epoch": 0.21342628453395301,
      "grad_norm": 2.8976056575775146,
      "learning_rate": 8.994018409769193e-06,
      "loss": 0.0398,
      "step": 4133
    },
    {
      "epoch": 0.21347792408985283,
      "grad_norm": 12.073786735534668,
      "learning_rate": 8.99352543549852e-06,
      "loss": 0.1767,
      "step": 4134
    },
    {
      "epoch": 0.21352956364575265,
      "grad_norm": 11.075065612792969,
      "learning_rate": 8.993032353985412e-06,
      "loss": 0.0504,
      "step": 4135
    },
    {
      "epoch": 0.21358120320165247,
      "grad_norm": 6.40057897567749,
      "learning_rate": 8.99253916524311e-06,
      "loss": 0.1023,
      "step": 4136
    },
    {
      "epoch": 0.2136328427575523,
      "grad_norm": 3.193321943283081,
      "learning_rate": 8.99204586928486e-06,
      "loss": 0.0539,
      "step": 4137
    },
    {
      "epoch": 0.2136844823134521,
      "grad_norm": 12.342140197753906,
      "learning_rate": 8.991552466123907e-06,
      "loss": 0.1489,
      "step": 4138
    },
    {
      "epoch": 0.21373612186935192,
      "grad_norm": 9.48413372039795,
      "learning_rate": 8.991058955773503e-06,
      "loss": 0.1008,
      "step": 4139
    },
    {
      "epoch": 0.21378776142525174,
      "grad_norm": 4.628359317779541,
      "learning_rate": 8.990565338246898e-06,
      "loss": 0.067,
      "step": 4140
    },
    {
      "epoch": 0.21383940098115156,
      "grad_norm": 5.930041313171387,
      "learning_rate": 8.99007161355735e-06,
      "loss": 0.1004,
      "step": 4141
    },
    {
      "epoch": 0.21389104053705138,
      "grad_norm": 4.508421421051025,
      "learning_rate": 8.989577781718115e-06,
      "loss": 0.0503,
      "step": 4142
    },
    {
      "epoch": 0.2139426800929512,
      "grad_norm": 4.819838047027588,
      "learning_rate": 8.989083842742456e-06,
      "loss": 0.0985,
      "step": 4143
    },
    {
      "epoch": 0.21399431964885102,
      "grad_norm": 5.4139299392700195,
      "learning_rate": 8.988589796643639e-06,
      "loss": 0.083,
      "step": 4144
    },
    {
      "epoch": 0.21404595920475084,
      "grad_norm": 2.898740768432617,
      "learning_rate": 8.988095643434928e-06,
      "loss": 0.0288,
      "step": 4145
    },
    {
      "epoch": 0.21409759876065065,
      "grad_norm": 9.489869117736816,
      "learning_rate": 8.987601383129597e-06,
      "loss": 0.0917,
      "step": 4146
    },
    {
      "epoch": 0.21414923831655047,
      "grad_norm": 4.347675800323486,
      "learning_rate": 8.987107015740913e-06,
      "loss": 0.0866,
      "step": 4147
    },
    {
      "epoch": 0.2142008778724503,
      "grad_norm": 3.1043853759765625,
      "learning_rate": 8.986612541282155e-06,
      "loss": 0.0663,
      "step": 4148
    },
    {
      "epoch": 0.2142525174283501,
      "grad_norm": 7.010867118835449,
      "learning_rate": 8.986117959766604e-06,
      "loss": 0.0978,
      "step": 4149
    },
    {
      "epoch": 0.21430415698424993,
      "grad_norm": 2.707803964614868,
      "learning_rate": 8.985623271207538e-06,
      "loss": 0.0399,
      "step": 4150
    },
    {
      "epoch": 0.21435579654014975,
      "grad_norm": 3.243004322052002,
      "learning_rate": 8.985128475618244e-06,
      "loss": 0.0978,
      "step": 4151
    },
    {
      "epoch": 0.21440743609604956,
      "grad_norm": 5.993607521057129,
      "learning_rate": 8.984633573012004e-06,
      "loss": 0.0661,
      "step": 4152
    },
    {
      "epoch": 0.21445907565194938,
      "grad_norm": 4.332311630249023,
      "learning_rate": 8.984138563402117e-06,
      "loss": 0.0998,
      "step": 4153
    },
    {
      "epoch": 0.2145107152078492,
      "grad_norm": 7.37567138671875,
      "learning_rate": 8.98364344680187e-06,
      "loss": 0.0855,
      "step": 4154
    },
    {
      "epoch": 0.21456235476374902,
      "grad_norm": 7.22476053237915,
      "learning_rate": 8.983148223224557e-06,
      "loss": 0.0526,
      "step": 4155
    },
    {
      "epoch": 0.21461399431964884,
      "grad_norm": 5.317347526550293,
      "learning_rate": 8.982652892683483e-06,
      "loss": 0.0809,
      "step": 4156
    },
    {
      "epoch": 0.21466563387554868,
      "grad_norm": 10.857172966003418,
      "learning_rate": 8.982157455191944e-06,
      "loss": 0.1664,
      "step": 4157
    },
    {
      "epoch": 0.2147172734314485,
      "grad_norm": 9.93951416015625,
      "learning_rate": 8.981661910763248e-06,
      "loss": 0.2083,
      "step": 4158
    },
    {
      "epoch": 0.21476891298734832,
      "grad_norm": 8.8270845413208,
      "learning_rate": 8.981166259410703e-06,
      "loss": 0.0542,
      "step": 4159
    },
    {
      "epoch": 0.21482055254324814,
      "grad_norm": 11.841545104980469,
      "learning_rate": 8.980670501147616e-06,
      "loss": 0.1022,
      "step": 4160
    },
    {
      "epoch": 0.21487219209914796,
      "grad_norm": 3.8586926460266113,
      "learning_rate": 8.980174635987302e-06,
      "loss": 0.0756,
      "step": 4161
    },
    {
      "epoch": 0.21492383165504778,
      "grad_norm": 12.80797290802002,
      "learning_rate": 8.979678663943077e-06,
      "loss": 0.1085,
      "step": 4162
    },
    {
      "epoch": 0.2149754712109476,
      "grad_norm": 7.855527400970459,
      "learning_rate": 8.979182585028258e-06,
      "loss": 0.0608,
      "step": 4163
    },
    {
      "epoch": 0.2150271107668474,
      "grad_norm": 6.972224712371826,
      "learning_rate": 8.97868639925617e-06,
      "loss": 0.0999,
      "step": 4164
    },
    {
      "epoch": 0.21507875032274723,
      "grad_norm": 3.266423225402832,
      "learning_rate": 8.978190106640136e-06,
      "loss": 0.0759,
      "step": 4165
    },
    {
      "epoch": 0.21513038987864705,
      "grad_norm": 6.733211517333984,
      "learning_rate": 8.977693707193482e-06,
      "loss": 0.0667,
      "step": 4166
    },
    {
      "epoch": 0.21518202943454687,
      "grad_norm": 7.255064487457275,
      "learning_rate": 8.977197200929542e-06,
      "loss": 0.0972,
      "step": 4167
    },
    {
      "epoch": 0.21523366899044669,
      "grad_norm": 3.304685592651367,
      "learning_rate": 8.976700587861645e-06,
      "loss": 0.0559,
      "step": 4168
    },
    {
      "epoch": 0.2152853085463465,
      "grad_norm": 2.5368824005126953,
      "learning_rate": 8.97620386800313e-06,
      "loss": 0.0701,
      "step": 4169
    },
    {
      "epoch": 0.21533694810224632,
      "grad_norm": 1.964590072631836,
      "learning_rate": 8.975707041367334e-06,
      "loss": 0.0254,
      "step": 4170
    },
    {
      "epoch": 0.21538858765814614,
      "grad_norm": 3.66011381149292,
      "learning_rate": 8.975210107967601e-06,
      "loss": 0.0691,
      "step": 4171
    },
    {
      "epoch": 0.21544022721404596,
      "grad_norm": 2.5262746810913086,
      "learning_rate": 8.974713067817274e-06,
      "loss": 0.0366,
      "step": 4172
    },
    {
      "epoch": 0.21549186676994578,
      "grad_norm": 5.448301792144775,
      "learning_rate": 8.974215920929701e-06,
      "loss": 0.0638,
      "step": 4173
    },
    {
      "epoch": 0.2155435063258456,
      "grad_norm": 10.987539291381836,
      "learning_rate": 8.973718667318234e-06,
      "loss": 0.0883,
      "step": 4174
    },
    {
      "epoch": 0.21559514588174541,
      "grad_norm": 15.305109024047852,
      "learning_rate": 8.973221306996224e-06,
      "loss": 0.1415,
      "step": 4175
    },
    {
      "epoch": 0.21564678543764523,
      "grad_norm": 4.372467517852783,
      "learning_rate": 8.972723839977027e-06,
      "loss": 0.0593,
      "step": 4176
    },
    {
      "epoch": 0.21569842499354505,
      "grad_norm": 4.081401348114014,
      "learning_rate": 8.972226266274004e-06,
      "loss": 0.0552,
      "step": 4177
    },
    {
      "epoch": 0.21575006454944487,
      "grad_norm": 19.03792953491211,
      "learning_rate": 8.971728585900514e-06,
      "loss": 0.151,
      "step": 4178
    },
    {
      "epoch": 0.2158017041053447,
      "grad_norm": 6.048220157623291,
      "learning_rate": 8.971230798869927e-06,
      "loss": 0.035,
      "step": 4179
    },
    {
      "epoch": 0.2158533436612445,
      "grad_norm": 11.884683609008789,
      "learning_rate": 8.970732905195604e-06,
      "loss": 0.0827,
      "step": 4180
    },
    {
      "epoch": 0.21590498321714433,
      "grad_norm": 9.68479061126709,
      "learning_rate": 8.970234904890919e-06,
      "loss": 0.0608,
      "step": 4181
    },
    {
      "epoch": 0.21595662277304414,
      "grad_norm": 3.9158968925476074,
      "learning_rate": 8.969736797969248e-06,
      "loss": 0.0778,
      "step": 4182
    },
    {
      "epoch": 0.21600826232894396,
      "grad_norm": 15.734869956970215,
      "learning_rate": 8.969238584443961e-06,
      "loss": 0.143,
      "step": 4183
    },
    {
      "epoch": 0.21605990188484378,
      "grad_norm": 4.553938865661621,
      "learning_rate": 8.968740264328441e-06,
      "loss": 0.0773,
      "step": 4184
    },
    {
      "epoch": 0.2161115414407436,
      "grad_norm": 3.821225643157959,
      "learning_rate": 8.968241837636071e-06,
      "loss": 0.0415,
      "step": 4185
    },
    {
      "epoch": 0.21616318099664342,
      "grad_norm": 7.032502174377441,
      "learning_rate": 8.967743304380232e-06,
      "loss": 0.0832,
      "step": 4186
    },
    {
      "epoch": 0.21621482055254324,
      "grad_norm": 3.906085968017578,
      "learning_rate": 8.967244664574315e-06,
      "loss": 0.094,
      "step": 4187
    },
    {
      "epoch": 0.21626646010844305,
      "grad_norm": 6.428750514984131,
      "learning_rate": 8.966745918231708e-06,
      "loss": 0.1053,
      "step": 4188
    },
    {
      "epoch": 0.2163180996643429,
      "grad_norm": 7.772029876708984,
      "learning_rate": 8.966247065365806e-06,
      "loss": 0.0591,
      "step": 4189
    },
    {
      "epoch": 0.21636973922024272,
      "grad_norm": 8.785489082336426,
      "learning_rate": 8.965748105990004e-06,
      "loss": 0.124,
      "step": 4190
    },
    {
      "epoch": 0.21642137877614254,
      "grad_norm": 7.915868759155273,
      "learning_rate": 8.965249040117702e-06,
      "loss": 0.0796,
      "step": 4191
    },
    {
      "epoch": 0.21647301833204236,
      "grad_norm": 5.005774021148682,
      "learning_rate": 8.964749867762303e-06,
      "loss": 0.0593,
      "step": 4192
    },
    {
      "epoch": 0.21652465788794217,
      "grad_norm": 8.008028984069824,
      "learning_rate": 8.96425058893721e-06,
      "loss": 0.0738,
      "step": 4193
    },
    {
      "epoch": 0.216576297443842,
      "grad_norm": 9.841712951660156,
      "learning_rate": 8.963751203655833e-06,
      "loss": 0.0872,
      "step": 4194
    },
    {
      "epoch": 0.2166279369997418,
      "grad_norm": 18.934646606445312,
      "learning_rate": 8.963251711931577e-06,
      "loss": 0.1062,
      "step": 4195
    },
    {
      "epoch": 0.21667957655564163,
      "grad_norm": 4.9141740798950195,
      "learning_rate": 8.962752113777864e-06,
      "loss": 0.052,
      "step": 4196
    },
    {
      "epoch": 0.21673121611154145,
      "grad_norm": 5.195682525634766,
      "learning_rate": 8.962252409208103e-06,
      "loss": 0.0356,
      "step": 4197
    },
    {
      "epoch": 0.21678285566744127,
      "grad_norm": 7.719667911529541,
      "learning_rate": 8.961752598235713e-06,
      "loss": 0.0765,
      "step": 4198
    },
    {
      "epoch": 0.21683449522334108,
      "grad_norm": 5.66157865524292,
      "learning_rate": 8.961252680874121e-06,
      "loss": 0.0777,
      "step": 4199
    },
    {
      "epoch": 0.2168861347792409,
      "grad_norm": 2.399111270904541,
      "learning_rate": 8.96075265713675e-06,
      "loss": 0.0375,
      "step": 4200
    },
    {
      "epoch": 0.21693777433514072,
      "grad_norm": 5.282463073730469,
      "learning_rate": 8.960252527037027e-06,
      "loss": 0.0866,
      "step": 4201
    },
    {
      "epoch": 0.21698941389104054,
      "grad_norm": 3.6646623611450195,
      "learning_rate": 8.959752290588382e-06,
      "loss": 0.0436,
      "step": 4202
    },
    {
      "epoch": 0.21704105344694036,
      "grad_norm": 13.520332336425781,
      "learning_rate": 8.959251947804249e-06,
      "loss": 0.1636,
      "step": 4203
    },
    {
      "epoch": 0.21709269300284018,
      "grad_norm": 4.803678512573242,
      "learning_rate": 8.958751498698063e-06,
      "loss": 0.0461,
      "step": 4204
    },
    {
      "epoch": 0.21714433255874,
      "grad_norm": 10.972677230834961,
      "learning_rate": 8.958250943283266e-06,
      "loss": 0.0936,
      "step": 4205
    },
    {
      "epoch": 0.2171959721146398,
      "grad_norm": 3.611102342605591,
      "learning_rate": 8.957750281573297e-06,
      "loss": 0.0526,
      "step": 4206
    },
    {
      "epoch": 0.21724761167053963,
      "grad_norm": 22.016817092895508,
      "learning_rate": 8.957249513581602e-06,
      "loss": 0.1049,
      "step": 4207
    },
    {
      "epoch": 0.21729925122643945,
      "grad_norm": 14.233765602111816,
      "learning_rate": 8.95674863932163e-06,
      "loss": 0.2272,
      "step": 4208
    },
    {
      "epoch": 0.21735089078233927,
      "grad_norm": 2.3183536529541016,
      "learning_rate": 8.95624765880683e-06,
      "loss": 0.0449,
      "step": 4209
    },
    {
      "epoch": 0.2174025303382391,
      "grad_norm": 13.121808052062988,
      "learning_rate": 8.955746572050654e-06,
      "loss": 0.1425,
      "step": 4210
    },
    {
      "epoch": 0.2174541698941389,
      "grad_norm": 6.39617395401001,
      "learning_rate": 8.955245379066561e-06,
      "loss": 0.0947,
      "step": 4211
    },
    {
      "epoch": 0.21750580945003872,
      "grad_norm": 5.7157487869262695,
      "learning_rate": 8.95474407986801e-06,
      "loss": 0.0358,
      "step": 4212
    },
    {
      "epoch": 0.21755744900593854,
      "grad_norm": 3.4182446002960205,
      "learning_rate": 8.954242674468461e-06,
      "loss": 0.0436,
      "step": 4213
    },
    {
      "epoch": 0.21760908856183836,
      "grad_norm": 11.113301277160645,
      "learning_rate": 8.953741162881377e-06,
      "loss": 0.1392,
      "step": 4214
    },
    {
      "epoch": 0.21766072811773818,
      "grad_norm": 5.484668731689453,
      "learning_rate": 8.953239545120231e-06,
      "loss": 0.0626,
      "step": 4215
    },
    {
      "epoch": 0.217712367673638,
      "grad_norm": 5.510799407958984,
      "learning_rate": 8.952737821198491e-06,
      "loss": 0.0659,
      "step": 4216
    },
    {
      "epoch": 0.21776400722953781,
      "grad_norm": 7.515837669372559,
      "learning_rate": 8.95223599112963e-06,
      "loss": 0.1086,
      "step": 4217
    },
    {
      "epoch": 0.21781564678543763,
      "grad_norm": 4.167096138000488,
      "learning_rate": 8.951734054927121e-06,
      "loss": 0.0317,
      "step": 4218
    },
    {
      "epoch": 0.21786728634133745,
      "grad_norm": 2.3422248363494873,
      "learning_rate": 8.951232012604449e-06,
      "loss": 0.0464,
      "step": 4219
    },
    {
      "epoch": 0.21791892589723727,
      "grad_norm": 3.6844089031219482,
      "learning_rate": 8.950729864175093e-06,
      "loss": 0.0359,
      "step": 4220
    },
    {
      "epoch": 0.21797056545313712,
      "grad_norm": 4.006019592285156,
      "learning_rate": 8.950227609652538e-06,
      "loss": 0.0716,
      "step": 4221
    },
    {
      "epoch": 0.21802220500903693,
      "grad_norm": 2.4462344646453857,
      "learning_rate": 8.949725249050272e-06,
      "loss": 0.0421,
      "step": 4222
    },
    {
      "epoch": 0.21807384456493675,
      "grad_norm": 13.485722541809082,
      "learning_rate": 8.949222782381783e-06,
      "loss": 0.192,
      "step": 4223
    },
    {
      "epoch": 0.21812548412083657,
      "grad_norm": 2.098677158355713,
      "learning_rate": 8.948720209660566e-06,
      "loss": 0.0347,
      "step": 4224
    },
    {
      "epoch": 0.2181771236767364,
      "grad_norm": 4.295846462249756,
      "learning_rate": 8.948217530900119e-06,
      "loss": 0.0609,
      "step": 4225
    },
    {
      "epoch": 0.2182287632326362,
      "grad_norm": 3.0048069953918457,
      "learning_rate": 8.94771474611394e-06,
      "loss": 0.0409,
      "step": 4226
    },
    {
      "epoch": 0.21828040278853603,
      "grad_norm": 3.3977489471435547,
      "learning_rate": 8.947211855315528e-06,
      "loss": 0.0391,
      "step": 4227
    },
    {
      "epoch": 0.21833204234443584,
      "grad_norm": 3.0376312732696533,
      "learning_rate": 8.94670885851839e-06,
      "loss": 0.0448,
      "step": 4228
    },
    {
      "epoch": 0.21838368190033566,
      "grad_norm": 25.091703414916992,
      "learning_rate": 8.946205755736032e-06,
      "loss": 0.0821,
      "step": 4229
    },
    {
      "epoch": 0.21843532145623548,
      "grad_norm": 4.941691875457764,
      "learning_rate": 8.94570254698197e-06,
      "loss": 0.0623,
      "step": 4230
    },
    {
      "epoch": 0.2184869610121353,
      "grad_norm": 6.231695175170898,
      "learning_rate": 8.94519923226971e-06,
      "loss": 0.0621,
      "step": 4231
    },
    {
      "epoch": 0.21853860056803512,
      "grad_norm": 18.49629020690918,
      "learning_rate": 8.944695811612771e-06,
      "loss": 0.3442,
      "step": 4232
    },
    {
      "epoch": 0.21859024012393494,
      "grad_norm": 24.014419555664062,
      "learning_rate": 8.94419228502467e-06,
      "loss": 0.1486,
      "step": 4233
    },
    {
      "epoch": 0.21864187967983476,
      "grad_norm": 5.690499782562256,
      "learning_rate": 8.943688652518933e-06,
      "loss": 0.0975,
      "step": 4234
    },
    {
      "epoch": 0.21869351923573457,
      "grad_norm": 12.0725679397583,
      "learning_rate": 8.94318491410908e-06,
      "loss": 0.1366,
      "step": 4235
    },
    {
      "epoch": 0.2187451587916344,
      "grad_norm": 9.128961563110352,
      "learning_rate": 8.942681069808643e-06,
      "loss": 0.0607,
      "step": 4236
    },
    {
      "epoch": 0.2187967983475342,
      "grad_norm": 3.753079414367676,
      "learning_rate": 8.942177119631148e-06,
      "loss": 0.0321,
      "step": 4237
    },
    {
      "epoch": 0.21884843790343403,
      "grad_norm": 2.5191304683685303,
      "learning_rate": 8.94167306359013e-06,
      "loss": 0.0319,
      "step": 4238
    },
    {
      "epoch": 0.21890007745933385,
      "grad_norm": 7.117066383361816,
      "learning_rate": 8.941168901699126e-06,
      "loss": 0.0418,
      "step": 4239
    },
    {
      "epoch": 0.21895171701523367,
      "grad_norm": 4.689333915710449,
      "learning_rate": 8.940664633971674e-06,
      "loss": 0.064,
      "step": 4240
    },
    {
      "epoch": 0.21900335657113348,
      "grad_norm": 14.85407543182373,
      "learning_rate": 8.940160260421315e-06,
      "loss": 0.1276,
      "step": 4241
    },
    {
      "epoch": 0.2190549961270333,
      "grad_norm": 6.745484828948975,
      "learning_rate": 8.939655781061592e-06,
      "loss": 0.1384,
      "step": 4242
    },
    {
      "epoch": 0.21910663568293312,
      "grad_norm": 32.358848571777344,
      "learning_rate": 8.939151195906056e-06,
      "loss": 0.1142,
      "step": 4243
    },
    {
      "epoch": 0.21915827523883294,
      "grad_norm": 5.078873634338379,
      "learning_rate": 8.938646504968253e-06,
      "loss": 0.0674,
      "step": 4244
    },
    {
      "epoch": 0.21920991479473276,
      "grad_norm": 13.51007080078125,
      "learning_rate": 8.938141708261741e-06,
      "loss": 0.1396,
      "step": 4245
    },
    {
      "epoch": 0.21926155435063258,
      "grad_norm": 5.762360095977783,
      "learning_rate": 8.937636805800072e-06,
      "loss": 0.0753,
      "step": 4246
    },
    {
      "epoch": 0.2193131939065324,
      "grad_norm": 3.1422243118286133,
      "learning_rate": 8.937131797596807e-06,
      "loss": 0.0568,
      "step": 4247
    },
    {
      "epoch": 0.2193648334624322,
      "grad_norm": 9.415493965148926,
      "learning_rate": 8.936626683665504e-06,
      "loss": 0.0992,
      "step": 4248
    },
    {
      "epoch": 0.21941647301833203,
      "grad_norm": 2.780885696411133,
      "learning_rate": 8.93612146401973e-06,
      "loss": 0.0645,
      "step": 4249
    },
    {
      "epoch": 0.21946811257423185,
      "grad_norm": 5.790852069854736,
      "learning_rate": 8.935616138673054e-06,
      "loss": 0.1042,
      "step": 4250
    },
    {
      "epoch": 0.21951975213013167,
      "grad_norm": 11.316829681396484,
      "learning_rate": 8.935110707639043e-06,
      "loss": 0.0811,
      "step": 4251
    },
    {
      "epoch": 0.2195713916860315,
      "grad_norm": 8.980566024780273,
      "learning_rate": 8.934605170931272e-06,
      "loss": 0.1164,
      "step": 4252
    },
    {
      "epoch": 0.21962303124193133,
      "grad_norm": 11.286039352416992,
      "learning_rate": 8.934099528563313e-06,
      "loss": 0.1298,
      "step": 4253
    },
    {
      "epoch": 0.21967467079783115,
      "grad_norm": 3.9465935230255127,
      "learning_rate": 8.93359378054875e-06,
      "loss": 0.0367,
      "step": 4254
    },
    {
      "epoch": 0.21972631035373097,
      "grad_norm": 5.015744209289551,
      "learning_rate": 8.93308792690116e-06,
      "loss": 0.0662,
      "step": 4255
    },
    {
      "epoch": 0.2197779499096308,
      "grad_norm": 6.647285461425781,
      "learning_rate": 8.932581967634128e-06,
      "loss": 0.0709,
      "step": 4256
    },
    {
      "epoch": 0.2198295894655306,
      "grad_norm": 7.5725626945495605,
      "learning_rate": 8.932075902761244e-06,
      "loss": 0.1077,
      "step": 4257
    },
    {
      "epoch": 0.21988122902143042,
      "grad_norm": 7.868321895599365,
      "learning_rate": 8.931569732296095e-06,
      "loss": 0.0862,
      "step": 4258
    },
    {
      "epoch": 0.21993286857733024,
      "grad_norm": 6.625125885009766,
      "learning_rate": 8.931063456252273e-06,
      "loss": 0.0908,
      "step": 4259
    },
    {
      "epoch": 0.21998450813323006,
      "grad_norm": 3.117518424987793,
      "learning_rate": 8.930557074643375e-06,
      "loss": 0.0327,
      "step": 4260
    },
    {
      "epoch": 0.22003614768912988,
      "grad_norm": 8.234017372131348,
      "learning_rate": 8.930050587483002e-06,
      "loss": 0.1654,
      "step": 4261
    },
    {
      "epoch": 0.2200877872450297,
      "grad_norm": 16.620071411132812,
      "learning_rate": 8.929543994784752e-06,
      "loss": 0.1258,
      "step": 4262
    },
    {
      "epoch": 0.22013942680092952,
      "grad_norm": 5.179795742034912,
      "learning_rate": 8.929037296562228e-06,
      "loss": 0.0705,
      "step": 4263
    },
    {
      "epoch": 0.22019106635682933,
      "grad_norm": 21.5516300201416,
      "learning_rate": 8.92853049282904e-06,
      "loss": 0.1037,
      "step": 4264
    },
    {
      "epoch": 0.22024270591272915,
      "grad_norm": 4.773989200592041,
      "learning_rate": 8.928023583598797e-06,
      "loss": 0.0538,
      "step": 4265
    },
    {
      "epoch": 0.22029434546862897,
      "grad_norm": 2.491473913192749,
      "learning_rate": 8.927516568885111e-06,
      "loss": 0.0558,
      "step": 4266
    },
    {
      "epoch": 0.2203459850245288,
      "grad_norm": 3.440002679824829,
      "learning_rate": 8.927009448701596e-06,
      "loss": 0.0557,
      "step": 4267
    },
    {
      "epoch": 0.2203976245804286,
      "grad_norm": 5.3842315673828125,
      "learning_rate": 8.926502223061873e-06,
      "loss": 0.0561,
      "step": 4268
    },
    {
      "epoch": 0.22044926413632843,
      "grad_norm": 3.6447012424468994,
      "learning_rate": 8.92599489197956e-06,
      "loss": 0.0365,
      "step": 4269
    },
    {
      "epoch": 0.22050090369222825,
      "grad_norm": 9.794671058654785,
      "learning_rate": 8.925487455468285e-06,
      "loss": 0.072,
      "step": 4270
    },
    {
      "epoch": 0.22055254324812806,
      "grad_norm": 4.400841236114502,
      "learning_rate": 8.924979913541672e-06,
      "loss": 0.0632,
      "step": 4271
    },
    {
      "epoch": 0.22060418280402788,
      "grad_norm": 5.803325653076172,
      "learning_rate": 8.92447226621335e-06,
      "loss": 0.0292,
      "step": 4272
    },
    {
      "epoch": 0.2206558223599277,
      "grad_norm": 20.586626052856445,
      "learning_rate": 8.923964513496952e-06,
      "loss": 0.0793,
      "step": 4273
    },
    {
      "epoch": 0.22070746191582752,
      "grad_norm": 6.442362308502197,
      "learning_rate": 8.923456655406115e-06,
      "loss": 0.079,
      "step": 4274
    },
    {
      "epoch": 0.22075910147172734,
      "grad_norm": 6.484548091888428,
      "learning_rate": 8.922948691954474e-06,
      "loss": 0.046,
      "step": 4275
    },
    {
      "epoch": 0.22081074102762716,
      "grad_norm": 4.928504467010498,
      "learning_rate": 8.922440623155674e-06,
      "loss": 0.0474,
      "step": 4276
    },
    {
      "epoch": 0.22086238058352697,
      "grad_norm": 4.005149841308594,
      "learning_rate": 8.921932449023354e-06,
      "loss": 0.0655,
      "step": 4277
    },
    {
      "epoch": 0.2209140201394268,
      "grad_norm": 4.654189586639404,
      "learning_rate": 8.921424169571164e-06,
      "loss": 0.0423,
      "step": 4278
    },
    {
      "epoch": 0.2209656596953266,
      "grad_norm": 5.12437629699707,
      "learning_rate": 8.920915784812752e-06,
      "loss": 0.0562,
      "step": 4279
    },
    {
      "epoch": 0.22101729925122643,
      "grad_norm": 6.297632694244385,
      "learning_rate": 8.920407294761771e-06,
      "loss": 0.0383,
      "step": 4280
    },
    {
      "epoch": 0.22106893880712625,
      "grad_norm": 5.657761096954346,
      "learning_rate": 8.919898699431875e-06,
      "loss": 0.061,
      "step": 4281
    },
    {
      "epoch": 0.22112057836302607,
      "grad_norm": 6.75119161605835,
      "learning_rate": 8.919389998836721e-06,
      "loss": 0.0852,
      "step": 4282
    },
    {
      "epoch": 0.22117221791892588,
      "grad_norm": 6.537166595458984,
      "learning_rate": 8.918881192989972e-06,
      "loss": 0.0676,
      "step": 4283
    },
    {
      "epoch": 0.22122385747482573,
      "grad_norm": 8.867945671081543,
      "learning_rate": 8.918372281905292e-06,
      "loss": 0.1375,
      "step": 4284
    },
    {
      "epoch": 0.22127549703072555,
      "grad_norm": 5.676558494567871,
      "learning_rate": 8.917863265596344e-06,
      "loss": 0.0762,
      "step": 4285
    },
    {
      "epoch": 0.22132713658662537,
      "grad_norm": 9.466896057128906,
      "learning_rate": 8.917354144076799e-06,
      "loss": 0.1013,
      "step": 4286
    },
    {
      "epoch": 0.22137877614252519,
      "grad_norm": 4.528562545776367,
      "learning_rate": 8.91684491736033e-06,
      "loss": 0.098,
      "step": 4287
    },
    {
      "epoch": 0.221430415698425,
      "grad_norm": 9.255322456359863,
      "learning_rate": 8.91633558546061e-06,
      "loss": 0.0448,
      "step": 4288
    },
    {
      "epoch": 0.22148205525432482,
      "grad_norm": 15.587035179138184,
      "learning_rate": 8.915826148391317e-06,
      "loss": 0.1094,
      "step": 4289
    },
    {
      "epoch": 0.22153369481022464,
      "grad_norm": 6.648265838623047,
      "learning_rate": 8.915316606166133e-06,
      "loss": 0.074,
      "step": 4290
    },
    {
      "epoch": 0.22158533436612446,
      "grad_norm": 6.432137966156006,
      "learning_rate": 8.91480695879874e-06,
      "loss": 0.0889,
      "step": 4291
    },
    {
      "epoch": 0.22163697392202428,
      "grad_norm": 3.662863254547119,
      "learning_rate": 8.914297206302823e-06,
      "loss": 0.0678,
      "step": 4292
    },
    {
      "epoch": 0.2216886134779241,
      "grad_norm": 9.765856742858887,
      "learning_rate": 8.913787348692074e-06,
      "loss": 0.1097,
      "step": 4293
    },
    {
      "epoch": 0.22174025303382391,
      "grad_norm": 2.770808219909668,
      "learning_rate": 8.913277385980182e-06,
      "loss": 0.0444,
      "step": 4294
    },
    {
      "epoch": 0.22179189258972373,
      "grad_norm": 5.024147987365723,
      "learning_rate": 8.912767318180842e-06,
      "loss": 0.0866,
      "step": 4295
    },
    {
      "epoch": 0.22184353214562355,
      "grad_norm": 3.4031121730804443,
      "learning_rate": 8.912257145307753e-06,
      "loss": 0.0478,
      "step": 4296
    },
    {
      "epoch": 0.22189517170152337,
      "grad_norm": 14.946114540100098,
      "learning_rate": 8.911746867374613e-06,
      "loss": 0.1736,
      "step": 4297
    },
    {
      "epoch": 0.2219468112574232,
      "grad_norm": 8.63025951385498,
      "learning_rate": 8.911236484395127e-06,
      "loss": 0.0913,
      "step": 4298
    },
    {
      "epoch": 0.221998450813323,
      "grad_norm": 5.063940048217773,
      "learning_rate": 8.910725996383e-06,
      "loss": 0.0717,
      "step": 4299
    },
    {
      "epoch": 0.22205009036922282,
      "grad_norm": 2.3252861499786377,
      "learning_rate": 8.910215403351942e-06,
      "loss": 0.0418,
      "step": 4300
    },
    {
      "epoch": 0.22210172992512264,
      "grad_norm": 11.501228332519531,
      "learning_rate": 8.909704705315663e-06,
      "loss": 0.1198,
      "step": 4301
    },
    {
      "epoch": 0.22215336948102246,
      "grad_norm": 4.430087566375732,
      "learning_rate": 8.909193902287876e-06,
      "loss": 0.0581,
      "step": 4302
    },
    {
      "epoch": 0.22220500903692228,
      "grad_norm": 3.3044726848602295,
      "learning_rate": 8.9086829942823e-06,
      "loss": 0.0736,
      "step": 4303
    },
    {
      "epoch": 0.2222566485928221,
      "grad_norm": 1.7288986444473267,
      "learning_rate": 8.908171981312655e-06,
      "loss": 0.0628,
      "step": 4304
    },
    {
      "epoch": 0.22230828814872192,
      "grad_norm": 4.104494094848633,
      "learning_rate": 8.907660863392664e-06,
      "loss": 0.0586,
      "step": 4305
    },
    {
      "epoch": 0.22235992770462173,
      "grad_norm": 3.31105899810791,
      "learning_rate": 8.907149640536053e-06,
      "loss": 0.0556,
      "step": 4306
    },
    {
      "epoch": 0.22241156726052155,
      "grad_norm": 10.197897911071777,
      "learning_rate": 8.906638312756547e-06,
      "loss": 0.0957,
      "step": 4307
    },
    {
      "epoch": 0.22246320681642137,
      "grad_norm": 2.6290435791015625,
      "learning_rate": 8.90612688006788e-06,
      "loss": 0.0643,
      "step": 4308
    },
    {
      "epoch": 0.2225148463723212,
      "grad_norm": 13.251558303833008,
      "learning_rate": 8.905615342483787e-06,
      "loss": 0.1649,
      "step": 4309
    },
    {
      "epoch": 0.222566485928221,
      "grad_norm": 9.64191722869873,
      "learning_rate": 8.905103700018003e-06,
      "loss": 0.1112,
      "step": 4310
    },
    {
      "epoch": 0.22261812548412083,
      "grad_norm": 2.9687106609344482,
      "learning_rate": 8.90459195268427e-06,
      "loss": 0.049,
      "step": 4311
    },
    {
      "epoch": 0.22266976504002065,
      "grad_norm": 3.5765111446380615,
      "learning_rate": 8.904080100496328e-06,
      "loss": 0.0539,
      "step": 4312
    },
    {
      "epoch": 0.22272140459592046,
      "grad_norm": 7.493796348571777,
      "learning_rate": 8.903568143467923e-06,
      "loss": 0.1252,
      "step": 4313
    },
    {
      "epoch": 0.22277304415182028,
      "grad_norm": 9.888468742370605,
      "learning_rate": 8.903056081612803e-06,
      "loss": 0.077,
      "step": 4314
    },
    {
      "epoch": 0.2228246837077201,
      "grad_norm": 8.658734321594238,
      "learning_rate": 8.902543914944718e-06,
      "loss": 0.0974,
      "step": 4315
    },
    {
      "epoch": 0.22287632326361995,
      "grad_norm": 4.7951860427856445,
      "learning_rate": 8.902031643477424e-06,
      "loss": 0.0956,
      "step": 4316
    },
    {
      "epoch": 0.22292796281951976,
      "grad_norm": 6.432164192199707,
      "learning_rate": 8.901519267224676e-06,
      "loss": 0.0778,
      "step": 4317
    },
    {
      "epoch": 0.22297960237541958,
      "grad_norm": 8.868595123291016,
      "learning_rate": 8.901006786200235e-06,
      "loss": 0.0725,
      "step": 4318
    },
    {
      "epoch": 0.2230312419313194,
      "grad_norm": 13.305741310119629,
      "learning_rate": 8.900494200417862e-06,
      "loss": 0.1092,
      "step": 4319
    },
    {
      "epoch": 0.22308288148721922,
      "grad_norm": 6.2510175704956055,
      "learning_rate": 8.899981509891323e-06,
      "loss": 0.0777,
      "step": 4320
    },
    {
      "epoch": 0.22313452104311904,
      "grad_norm": 3.349524736404419,
      "learning_rate": 8.899468714634384e-06,
      "loss": 0.0597,
      "step": 4321
    },
    {
      "epoch": 0.22318616059901886,
      "grad_norm": 9.346650123596191,
      "learning_rate": 8.898955814660816e-06,
      "loss": 0.1129,
      "step": 4322
    },
    {
      "epoch": 0.22323780015491868,
      "grad_norm": 7.828361988067627,
      "learning_rate": 8.898442809984393e-06,
      "loss": 0.0754,
      "step": 4323
    },
    {
      "epoch": 0.2232894397108185,
      "grad_norm": 4.134401321411133,
      "learning_rate": 8.897929700618892e-06,
      "loss": 0.0728,
      "step": 4324
    },
    {
      "epoch": 0.2233410792667183,
      "grad_norm": 4.064324378967285,
      "learning_rate": 8.897416486578093e-06,
      "loss": 0.0706,
      "step": 4325
    },
    {
      "epoch": 0.22339271882261813,
      "grad_norm": 5.615513324737549,
      "learning_rate": 8.896903167875775e-06,
      "loss": 0.1024,
      "step": 4326
    },
    {
      "epoch": 0.22344435837851795,
      "grad_norm": 5.083674430847168,
      "learning_rate": 8.896389744525725e-06,
      "loss": 0.0907,
      "step": 4327
    },
    {
      "epoch": 0.22349599793441777,
      "grad_norm": 6.7639875411987305,
      "learning_rate": 8.895876216541728e-06,
      "loss": 0.0799,
      "step": 4328
    },
    {
      "epoch": 0.22354763749031759,
      "grad_norm": 3.1931402683258057,
      "learning_rate": 8.895362583937577e-06,
      "loss": 0.0936,
      "step": 4329
    },
    {
      "epoch": 0.2235992770462174,
      "grad_norm": 15.696414947509766,
      "learning_rate": 8.894848846727063e-06,
      "loss": 0.175,
      "step": 4330
    },
    {
      "epoch": 0.22365091660211722,
      "grad_norm": 3.4618732929229736,
      "learning_rate": 8.894335004923983e-06,
      "loss": 0.0664,
      "step": 4331
    },
    {
      "epoch": 0.22370255615801704,
      "grad_norm": 7.1565046310424805,
      "learning_rate": 8.893821058542135e-06,
      "loss": 0.0578,
      "step": 4332
    },
    {
      "epoch": 0.22375419571391686,
      "grad_norm": 10.382990837097168,
      "learning_rate": 8.893307007595323e-06,
      "loss": 0.1052,
      "step": 4333
    },
    {
      "epoch": 0.22380583526981668,
      "grad_norm": 10.136570930480957,
      "learning_rate": 8.89279285209735e-06,
      "loss": 0.134,
      "step": 4334
    },
    {
      "epoch": 0.2238574748257165,
      "grad_norm": 3.025066614151001,
      "learning_rate": 8.89227859206202e-06,
      "loss": 0.0776,
      "step": 4335
    },
    {
      "epoch": 0.22390911438161631,
      "grad_norm": 2.8154385089874268,
      "learning_rate": 8.891764227503148e-06,
      "loss": 0.0425,
      "step": 4336
    },
    {
      "epoch": 0.22396075393751613,
      "grad_norm": 3.507953643798828,
      "learning_rate": 8.891249758434544e-06,
      "loss": 0.0449,
      "step": 4337
    },
    {
      "epoch": 0.22401239349341595,
      "grad_norm": 4.877304553985596,
      "learning_rate": 8.890735184870025e-06,
      "loss": 0.1051,
      "step": 4338
    },
    {
      "epoch": 0.22406403304931577,
      "grad_norm": 14.142964363098145,
      "learning_rate": 8.890220506823409e-06,
      "loss": 0.0909,
      "step": 4339
    },
    {
      "epoch": 0.2241156726052156,
      "grad_norm": 7.925802230834961,
      "learning_rate": 8.889705724308515e-06,
      "loss": 0.0926,
      "step": 4340
    },
    {
      "epoch": 0.2241673121611154,
      "grad_norm": 4.70009708404541,
      "learning_rate": 8.889190837339167e-06,
      "loss": 0.0328,
      "step": 4341
    },
    {
      "epoch": 0.22421895171701522,
      "grad_norm": 3.1443686485290527,
      "learning_rate": 8.888675845929197e-06,
      "loss": 0.0257,
      "step": 4342
    },
    {
      "epoch": 0.22427059127291504,
      "grad_norm": 9.905572891235352,
      "learning_rate": 8.88816075009243e-06,
      "loss": 0.1695,
      "step": 4343
    },
    {
      "epoch": 0.22432223082881486,
      "grad_norm": 3.0356802940368652,
      "learning_rate": 8.887645549842699e-06,
      "loss": 0.045,
      "step": 4344
    },
    {
      "epoch": 0.22437387038471468,
      "grad_norm": 13.830561637878418,
      "learning_rate": 8.88713024519384e-06,
      "loss": 0.2201,
      "step": 4345
    },
    {
      "epoch": 0.2244255099406145,
      "grad_norm": 7.72384786605835,
      "learning_rate": 8.886614836159691e-06,
      "loss": 0.0969,
      "step": 4346
    },
    {
      "epoch": 0.22447714949651432,
      "grad_norm": 2.9625582695007324,
      "learning_rate": 8.886099322754093e-06,
      "loss": 0.0496,
      "step": 4347
    },
    {
      "epoch": 0.22452878905241416,
      "grad_norm": 3.771989107131958,
      "learning_rate": 8.88558370499089e-06,
      "loss": 0.0741,
      "step": 4348
    },
    {
      "epoch": 0.22458042860831398,
      "grad_norm": 3.3387224674224854,
      "learning_rate": 8.885067982883925e-06,
      "loss": 0.0596,
      "step": 4349
    },
    {
      "epoch": 0.2246320681642138,
      "grad_norm": 2.633430004119873,
      "learning_rate": 8.884552156447053e-06,
      "loss": 0.028,
      "step": 4350
    },
    {
      "epoch": 0.22468370772011362,
      "grad_norm": 7.230668067932129,
      "learning_rate": 8.884036225694121e-06,
      "loss": 0.0808,
      "step": 4351
    },
    {
      "epoch": 0.22473534727601344,
      "grad_norm": 17.166866302490234,
      "learning_rate": 8.883520190638988e-06,
      "loss": 0.2135,
      "step": 4352
    },
    {
      "epoch": 0.22478698683191325,
      "grad_norm": 9.698424339294434,
      "learning_rate": 8.883004051295508e-06,
      "loss": 0.1199,
      "step": 4353
    },
    {
      "epoch": 0.22483862638781307,
      "grad_norm": 4.628873348236084,
      "learning_rate": 8.882487807677545e-06,
      "loss": 0.0729,
      "step": 4354
    },
    {
      "epoch": 0.2248902659437129,
      "grad_norm": 4.432403564453125,
      "learning_rate": 8.881971459798957e-06,
      "loss": 0.0555,
      "step": 4355
    },
    {
      "epoch": 0.2249419054996127,
      "grad_norm": 3.794567108154297,
      "learning_rate": 8.881455007673615e-06,
      "loss": 0.073,
      "step": 4356
    },
    {
      "epoch": 0.22499354505551253,
      "grad_norm": 9.116499900817871,
      "learning_rate": 8.880938451315386e-06,
      "loss": 0.1079,
      "step": 4357
    },
    {
      "epoch": 0.22504518461141235,
      "grad_norm": 2.1348729133605957,
      "learning_rate": 8.880421790738141e-06,
      "loss": 0.059,
      "step": 4358
    },
    {
      "epoch": 0.22509682416731217,
      "grad_norm": 3.675055503845215,
      "learning_rate": 8.879905025955756e-06,
      "loss": 0.068,
      "step": 4359
    },
    {
      "epoch": 0.22514846372321198,
      "grad_norm": 6.253688335418701,
      "learning_rate": 8.879388156982107e-06,
      "loss": 0.0309,
      "step": 4360
    },
    {
      "epoch": 0.2252001032791118,
      "grad_norm": 19.56639862060547,
      "learning_rate": 8.878871183831074e-06,
      "loss": 0.2029,
      "step": 4361
    },
    {
      "epoch": 0.22525174283501162,
      "grad_norm": 2.756819009780884,
      "learning_rate": 8.878354106516542e-06,
      "loss": 0.047,
      "step": 4362
    },
    {
      "epoch": 0.22530338239091144,
      "grad_norm": 3.376875162124634,
      "learning_rate": 8.877836925052395e-06,
      "loss": 0.0542,
      "step": 4363
    },
    {
      "epoch": 0.22535502194681126,
      "grad_norm": 13.8948392868042,
      "learning_rate": 8.87731963945252e-06,
      "loss": 0.1909,
      "step": 4364
    },
    {
      "epoch": 0.22540666150271108,
      "grad_norm": 3.866466522216797,
      "learning_rate": 8.87680224973081e-06,
      "loss": 0.0625,
      "step": 4365
    },
    {
      "epoch": 0.2254583010586109,
      "grad_norm": 2.1950788497924805,
      "learning_rate": 8.876284755901158e-06,
      "loss": 0.0618,
      "step": 4366
    },
    {
      "epoch": 0.2255099406145107,
      "grad_norm": 3.2266016006469727,
      "learning_rate": 8.875767157977462e-06,
      "loss": 0.0866,
      "step": 4367
    },
    {
      "epoch": 0.22556158017041053,
      "grad_norm": 2.871610641479492,
      "learning_rate": 8.875249455973623e-06,
      "loss": 0.046,
      "step": 4368
    },
    {
      "epoch": 0.22561321972631035,
      "grad_norm": 2.9560580253601074,
      "learning_rate": 8.874731649903538e-06,
      "loss": 0.038,
      "step": 4369
    },
    {
      "epoch": 0.22566485928221017,
      "grad_norm": 5.751621723175049,
      "learning_rate": 8.874213739781116e-06,
      "loss": 0.069,
      "step": 4370
    },
    {
      "epoch": 0.22571649883810999,
      "grad_norm": 2.9042158126831055,
      "learning_rate": 8.873695725620269e-06,
      "loss": 0.0593,
      "step": 4371
    },
    {
      "epoch": 0.2257681383940098,
      "grad_norm": 6.587323188781738,
      "learning_rate": 8.8731776074349e-06,
      "loss": 0.0959,
      "step": 4372
    },
    {
      "epoch": 0.22581977794990962,
      "grad_norm": 2.293996810913086,
      "learning_rate": 8.872659385238926e-06,
      "loss": 0.0314,
      "step": 4373
    },
    {
      "epoch": 0.22587141750580944,
      "grad_norm": 5.491858959197998,
      "learning_rate": 8.872141059046263e-06,
      "loss": 0.1131,
      "step": 4374
    },
    {
      "epoch": 0.22592305706170926,
      "grad_norm": 2.181553363800049,
      "learning_rate": 8.871622628870831e-06,
      "loss": 0.0324,
      "step": 4375
    },
    {
      "epoch": 0.22597469661760908,
      "grad_norm": 3.1559858322143555,
      "learning_rate": 8.871104094726552e-06,
      "loss": 0.0581,
      "step": 4376
    },
    {
      "epoch": 0.2260263361735089,
      "grad_norm": 33.09617614746094,
      "learning_rate": 8.87058545662735e-06,
      "loss": 0.1736,
      "step": 4377
    },
    {
      "epoch": 0.22607797572940871,
      "grad_norm": 5.953604698181152,
      "learning_rate": 8.870066714587154e-06,
      "loss": 0.114,
      "step": 4378
    },
    {
      "epoch": 0.22612961528530853,
      "grad_norm": 18.435489654541016,
      "learning_rate": 8.869547868619893e-06,
      "loss": 0.1491,
      "step": 4379
    },
    {
      "epoch": 0.22618125484120838,
      "grad_norm": 3.562070369720459,
      "learning_rate": 8.8690289187395e-06,
      "loss": 0.0606,
      "step": 4380
    },
    {
      "epoch": 0.2262328943971082,
      "grad_norm": 8.271159172058105,
      "learning_rate": 8.86850986495991e-06,
      "loss": 0.0717,
      "step": 4381
    },
    {
      "epoch": 0.22628453395300802,
      "grad_norm": 2.9308230876922607,
      "learning_rate": 8.867990707295064e-06,
      "loss": 0.0433,
      "step": 4382
    },
    {
      "epoch": 0.22633617350890783,
      "grad_norm": 7.449619293212891,
      "learning_rate": 8.867471445758902e-06,
      "loss": 0.1003,
      "step": 4383
    },
    {
      "epoch": 0.22638781306480765,
      "grad_norm": 3.9367477893829346,
      "learning_rate": 8.86695208036537e-06,
      "loss": 0.0745,
      "step": 4384
    },
    {
      "epoch": 0.22643945262070747,
      "grad_norm": 2.20621657371521,
      "learning_rate": 8.866432611128413e-06,
      "loss": 0.0173,
      "step": 4385
    },
    {
      "epoch": 0.2264910921766073,
      "grad_norm": 2.9404096603393555,
      "learning_rate": 8.865913038061981e-06,
      "loss": 0.0442,
      "step": 4386
    },
    {
      "epoch": 0.2265427317325071,
      "grad_norm": 2.457505702972412,
      "learning_rate": 8.865393361180028e-06,
      "loss": 0.0416,
      "step": 4387
    },
    {
      "epoch": 0.22659437128840693,
      "grad_norm": 3.247570276260376,
      "learning_rate": 8.864873580496509e-06,
      "loss": 0.0541,
      "step": 4388
    },
    {
      "epoch": 0.22664601084430674,
      "grad_norm": 11.25959587097168,
      "learning_rate": 8.864353696025381e-06,
      "loss": 0.1058,
      "step": 4389
    },
    {
      "epoch": 0.22669765040020656,
      "grad_norm": 16.64886474609375,
      "learning_rate": 8.863833707780607e-06,
      "loss": 0.1366,
      "step": 4390
    },
    {
      "epoch": 0.22674928995610638,
      "grad_norm": 3.6492042541503906,
      "learning_rate": 8.863313615776148e-06,
      "loss": 0.0212,
      "step": 4391
    },
    {
      "epoch": 0.2268009295120062,
      "grad_norm": 5.782066345214844,
      "learning_rate": 8.862793420025975e-06,
      "loss": 0.0862,
      "step": 4392
    },
    {
      "epoch": 0.22685256906790602,
      "grad_norm": 6.8252739906311035,
      "learning_rate": 8.862273120544053e-06,
      "loss": 0.0727,
      "step": 4393
    },
    {
      "epoch": 0.22690420862380584,
      "grad_norm": 6.0261921882629395,
      "learning_rate": 8.861752717344359e-06,
      "loss": 0.0986,
      "step": 4394
    },
    {
      "epoch": 0.22695584817970565,
      "grad_norm": 3.60929274559021,
      "learning_rate": 8.861232210440862e-06,
      "loss": 0.0358,
      "step": 4395
    },
    {
      "epoch": 0.22700748773560547,
      "grad_norm": 4.189229965209961,
      "learning_rate": 8.860711599847542e-06,
      "loss": 0.0764,
      "step": 4396
    },
    {
      "epoch": 0.2270591272915053,
      "grad_norm": 1.622934103012085,
      "learning_rate": 8.86019088557838e-06,
      "loss": 0.015,
      "step": 4397
    },
    {
      "epoch": 0.2271107668474051,
      "grad_norm": 13.89324951171875,
      "learning_rate": 8.859670067647362e-06,
      "loss": 0.1413,
      "step": 4398
    },
    {
      "epoch": 0.22716240640330493,
      "grad_norm": 3.888741970062256,
      "learning_rate": 8.85914914606847e-06,
      "loss": 0.0435,
      "step": 4399
    },
    {
      "epoch": 0.22721404595920475,
      "grad_norm": 3.5640697479248047,
      "learning_rate": 8.858628120855696e-06,
      "loss": 0.0605,
      "step": 4400
    },
    {
      "epoch": 0.22726568551510457,
      "grad_norm": 4.798946380615234,
      "learning_rate": 8.858106992023028e-06,
      "loss": 0.0406,
      "step": 4401
    },
    {
      "epoch": 0.22731732507100438,
      "grad_norm": 2.8282198905944824,
      "learning_rate": 8.857585759584463e-06,
      "loss": 0.0633,
      "step": 4402
    },
    {
      "epoch": 0.2273689646269042,
      "grad_norm": 7.946663856506348,
      "learning_rate": 8.857064423553999e-06,
      "loss": 0.0877,
      "step": 4403
    },
    {
      "epoch": 0.22742060418280402,
      "grad_norm": 10.306281089782715,
      "learning_rate": 8.856542983945634e-06,
      "loss": 0.1004,
      "step": 4404
    },
    {
      "epoch": 0.22747224373870384,
      "grad_norm": 3.8545823097229004,
      "learning_rate": 8.856021440773371e-06,
      "loss": 0.0507,
      "step": 4405
    },
    {
      "epoch": 0.22752388329460366,
      "grad_norm": 6.612061500549316,
      "learning_rate": 8.855499794051217e-06,
      "loss": 0.0581,
      "step": 4406
    },
    {
      "epoch": 0.22757552285050348,
      "grad_norm": 2.0618855953216553,
      "learning_rate": 8.85497804379318e-06,
      "loss": 0.0281,
      "step": 4407
    },
    {
      "epoch": 0.2276271624064033,
      "grad_norm": 26.346689224243164,
      "learning_rate": 8.854456190013267e-06,
      "loss": 0.3081,
      "step": 4408
    },
    {
      "epoch": 0.2276788019623031,
      "grad_norm": 2.0307037830352783,
      "learning_rate": 8.853934232725499e-06,
      "loss": 0.0201,
      "step": 4409
    },
    {
      "epoch": 0.22773044151820293,
      "grad_norm": 6.791439056396484,
      "learning_rate": 8.853412171943887e-06,
      "loss": 0.0455,
      "step": 4410
    },
    {
      "epoch": 0.22778208107410275,
      "grad_norm": 6.339528560638428,
      "learning_rate": 8.852890007682454e-06,
      "loss": 0.0914,
      "step": 4411
    },
    {
      "epoch": 0.2278337206300026,
      "grad_norm": 15.681663513183594,
      "learning_rate": 8.85236773995522e-06,
      "loss": 0.134,
      "step": 4412
    },
    {
      "epoch": 0.2278853601859024,
      "grad_norm": 5.941494941711426,
      "learning_rate": 8.851845368776209e-06,
      "loss": 0.0653,
      "step": 4413
    },
    {
      "epoch": 0.22793699974180223,
      "grad_norm": 5.299298286437988,
      "learning_rate": 8.851322894159453e-06,
      "loss": 0.0933,
      "step": 4414
    },
    {
      "epoch": 0.22798863929770205,
      "grad_norm": 3.661539077758789,
      "learning_rate": 8.85080031611898e-06,
      "loss": 0.084,
      "step": 4415
    },
    {
      "epoch": 0.22804027885360187,
      "grad_norm": 2.8020517826080322,
      "learning_rate": 8.850277634668822e-06,
      "loss": 0.0513,
      "step": 4416
    },
    {
      "epoch": 0.2280919184095017,
      "grad_norm": 5.009568691253662,
      "learning_rate": 8.849754849823018e-06,
      "loss": 0.0407,
      "step": 4417
    },
    {
      "epoch": 0.2281435579654015,
      "grad_norm": 11.197550773620605,
      "learning_rate": 8.849231961595605e-06,
      "loss": 0.1214,
      "step": 4418
    },
    {
      "epoch": 0.22819519752130132,
      "grad_norm": 12.61378288269043,
      "learning_rate": 8.848708970000626e-06,
      "loss": 0.0807,
      "step": 4419
    },
    {
      "epoch": 0.22824683707720114,
      "grad_norm": 3.3649075031280518,
      "learning_rate": 8.848185875052125e-06,
      "loss": 0.0234,
      "step": 4420
    },
    {
      "epoch": 0.22829847663310096,
      "grad_norm": 3.404799461364746,
      "learning_rate": 8.847662676764149e-06,
      "loss": 0.053,
      "step": 4421
    },
    {
      "epoch": 0.22835011618900078,
      "grad_norm": 3.123945951461792,
      "learning_rate": 8.847139375150745e-06,
      "loss": 0.0361,
      "step": 4422
    },
    {
      "epoch": 0.2284017557449006,
      "grad_norm": 11.488557815551758,
      "learning_rate": 8.846615970225972e-06,
      "loss": 0.1318,
      "step": 4423
    },
    {
      "epoch": 0.22845339530080042,
      "grad_norm": 3.3181796073913574,
      "learning_rate": 8.84609246200388e-06,
      "loss": 0.0636,
      "step": 4424
    },
    {
      "epoch": 0.22850503485670023,
      "grad_norm": 23.23625946044922,
      "learning_rate": 8.845568850498531e-06,
      "loss": 0.3676,
      "step": 4425
    },
    {
      "epoch": 0.22855667441260005,
      "grad_norm": 17.64366912841797,
      "learning_rate": 8.845045135723984e-06,
      "loss": 0.089,
      "step": 4426
    },
    {
      "epoch": 0.22860831396849987,
      "grad_norm": 3.091524124145508,
      "learning_rate": 8.844521317694303e-06,
      "loss": 0.0383,
      "step": 4427
    },
    {
      "epoch": 0.2286599535243997,
      "grad_norm": 5.147474765777588,
      "learning_rate": 8.843997396423558e-06,
      "loss": 0.0497,
      "step": 4428
    },
    {
      "epoch": 0.2287115930802995,
      "grad_norm": 14.249748229980469,
      "learning_rate": 8.843473371925813e-06,
      "loss": 0.1317,
      "step": 4429
    },
    {
      "epoch": 0.22876323263619933,
      "grad_norm": 9.343255043029785,
      "learning_rate": 8.842949244215144e-06,
      "loss": 0.0864,
      "step": 4430
    },
    {
      "epoch": 0.22881487219209914,
      "grad_norm": 8.153406143188477,
      "learning_rate": 8.842425013305624e-06,
      "loss": 0.181,
      "step": 4431
    },
    {
      "epoch": 0.22886651174799896,
      "grad_norm": 15.095611572265625,
      "learning_rate": 8.841900679211332e-06,
      "loss": 0.1067,
      "step": 4432
    },
    {
      "epoch": 0.22891815130389878,
      "grad_norm": 6.638213157653809,
      "learning_rate": 8.841376241946346e-06,
      "loss": 0.0414,
      "step": 4433
    },
    {
      "epoch": 0.2289697908597986,
      "grad_norm": 3.681314468383789,
      "learning_rate": 8.840851701524755e-06,
      "loss": 0.0722,
      "step": 4434
    },
    {
      "epoch": 0.22902143041569842,
      "grad_norm": 4.7404327392578125,
      "learning_rate": 8.84032705796064e-06,
      "loss": 0.0909,
      "step": 4435
    },
    {
      "epoch": 0.22907306997159824,
      "grad_norm": 5.077275276184082,
      "learning_rate": 8.839802311268091e-06,
      "loss": 0.0575,
      "step": 4436
    },
    {
      "epoch": 0.22912470952749805,
      "grad_norm": 5.33388090133667,
      "learning_rate": 8.839277461461199e-06,
      "loss": 0.0666,
      "step": 4437
    },
    {
      "epoch": 0.22917634908339787,
      "grad_norm": 10.468631744384766,
      "learning_rate": 8.83875250855406e-06,
      "loss": 0.1374,
      "step": 4438
    },
    {
      "epoch": 0.2292279886392977,
      "grad_norm": 4.663488388061523,
      "learning_rate": 8.83822745256077e-06,
      "loss": 0.0701,
      "step": 4439
    },
    {
      "epoch": 0.2292796281951975,
      "grad_norm": 3.5895419120788574,
      "learning_rate": 8.837702293495428e-06,
      "loss": 0.0925,
      "step": 4440
    },
    {
      "epoch": 0.22933126775109733,
      "grad_norm": 5.340178966522217,
      "learning_rate": 8.83717703137214e-06,
      "loss": 0.0507,
      "step": 4441
    },
    {
      "epoch": 0.22938290730699715,
      "grad_norm": 4.372637748718262,
      "learning_rate": 8.836651666205007e-06,
      "loss": 0.0527,
      "step": 4442
    },
    {
      "epoch": 0.22943454686289697,
      "grad_norm": 4.319395542144775,
      "learning_rate": 8.836126198008142e-06,
      "loss": 0.0759,
      "step": 4443
    },
    {
      "epoch": 0.2294861864187968,
      "grad_norm": 1.796149492263794,
      "learning_rate": 8.83560062679565e-06,
      "loss": 0.0511,
      "step": 4444
    },
    {
      "epoch": 0.22953782597469663,
      "grad_norm": 3.761497735977173,
      "learning_rate": 8.83507495258165e-06,
      "loss": 0.0437,
      "step": 4445
    },
    {
      "epoch": 0.22958946553059645,
      "grad_norm": 6.227187156677246,
      "learning_rate": 8.834549175380258e-06,
      "loss": 0.0797,
      "step": 4446
    },
    {
      "epoch": 0.22964110508649627,
      "grad_norm": 5.360013961791992,
      "learning_rate": 8.83402329520559e-06,
      "loss": 0.0767,
      "step": 4447
    },
    {
      "epoch": 0.22969274464239609,
      "grad_norm": 6.828617572784424,
      "learning_rate": 8.83349731207177e-06,
      "loss": 0.0387,
      "step": 4448
    },
    {
      "epoch": 0.2297443841982959,
      "grad_norm": 2.44543194770813,
      "learning_rate": 8.832971225992924e-06,
      "loss": 0.0329,
      "step": 4449
    },
    {
      "epoch": 0.22979602375419572,
      "grad_norm": 4.22143030166626,
      "learning_rate": 8.832445036983176e-06,
      "loss": 0.0579,
      "step": 4450
    },
    {
      "epoch": 0.22984766331009554,
      "grad_norm": 3.338233232498169,
      "learning_rate": 8.831918745056662e-06,
      "loss": 0.0311,
      "step": 4451
    },
    {
      "epoch": 0.22989930286599536,
      "grad_norm": 6.749471664428711,
      "learning_rate": 8.83139235022751e-06,
      "loss": 0.1388,
      "step": 4452
    },
    {
      "epoch": 0.22995094242189518,
      "grad_norm": 2.8631279468536377,
      "learning_rate": 8.830865852509857e-06,
      "loss": 0.0303,
      "step": 4453
    },
    {
      "epoch": 0.230002581977795,
      "grad_norm": 3.8607733249664307,
      "learning_rate": 8.830339251917842e-06,
      "loss": 0.0639,
      "step": 4454
    },
    {
      "epoch": 0.2300542215336948,
      "grad_norm": 4.447676658630371,
      "learning_rate": 8.829812548465606e-06,
      "loss": 0.0732,
      "step": 4455
    },
    {
      "epoch": 0.23010586108959463,
      "grad_norm": 4.049728870391846,
      "learning_rate": 8.829285742167295e-06,
      "loss": 0.0408,
      "step": 4456
    },
    {
      "epoch": 0.23015750064549445,
      "grad_norm": 4.971036911010742,
      "learning_rate": 8.828758833037054e-06,
      "loss": 0.058,
      "step": 4457
    },
    {
      "epoch": 0.23020914020139427,
      "grad_norm": 19.465356826782227,
      "learning_rate": 8.828231821089032e-06,
      "loss": 0.2606,
      "step": 4458
    },
    {
      "epoch": 0.2302607797572941,
      "grad_norm": 5.608879089355469,
      "learning_rate": 8.827704706337386e-06,
      "loss": 0.0654,
      "step": 4459
    },
    {
      "epoch": 0.2303124193131939,
      "grad_norm": 6.960256099700928,
      "learning_rate": 8.827177488796264e-06,
      "loss": 0.0747,
      "step": 4460
    },
    {
      "epoch": 0.23036405886909372,
      "grad_norm": 13.977187156677246,
      "learning_rate": 8.82665016847983e-06,
      "loss": 0.1928,
      "step": 4461
    },
    {
      "epoch": 0.23041569842499354,
      "grad_norm": 6.031872749328613,
      "learning_rate": 8.826122745402241e-06,
      "loss": 0.0826,
      "step": 4462
    },
    {
      "epoch": 0.23046733798089336,
      "grad_norm": 4.887537956237793,
      "learning_rate": 8.825595219577662e-06,
      "loss": 0.0771,
      "step": 4463
    },
    {
      "epoch": 0.23051897753679318,
      "grad_norm": 4.853940010070801,
      "learning_rate": 8.82506759102026e-06,
      "loss": 0.0638,
      "step": 4464
    },
    {
      "epoch": 0.230570617092693,
      "grad_norm": 4.9403510093688965,
      "learning_rate": 8.824539859744203e-06,
      "loss": 0.0815,
      "step": 4465
    },
    {
      "epoch": 0.23062225664859282,
      "grad_norm": 1.7047418355941772,
      "learning_rate": 8.824012025763662e-06,
      "loss": 0.0186,
      "step": 4466
    },
    {
      "epoch": 0.23067389620449263,
      "grad_norm": 2.324059009552002,
      "learning_rate": 8.823484089092813e-06,
      "loss": 0.0258,
      "step": 4467
    },
    {
      "epoch": 0.23072553576039245,
      "grad_norm": 7.141522407531738,
      "learning_rate": 8.822956049745833e-06,
      "loss": 0.0878,
      "step": 4468
    },
    {
      "epoch": 0.23077717531629227,
      "grad_norm": 5.917729377746582,
      "learning_rate": 8.822427907736901e-06,
      "loss": 0.0948,
      "step": 4469
    },
    {
      "epoch": 0.2308288148721921,
      "grad_norm": 11.477703094482422,
      "learning_rate": 8.8218996630802e-06,
      "loss": 0.1898,
      "step": 4470
    },
    {
      "epoch": 0.2308804544280919,
      "grad_norm": 3.5335986614227295,
      "learning_rate": 8.821371315789914e-06,
      "loss": 0.0509,
      "step": 4471
    },
    {
      "epoch": 0.23093209398399173,
      "grad_norm": 6.835648059844971,
      "learning_rate": 8.820842865880237e-06,
      "loss": 0.061,
      "step": 4472
    },
    {
      "epoch": 0.23098373353989154,
      "grad_norm": 8.146089553833008,
      "learning_rate": 8.820314313365354e-06,
      "loss": 0.097,
      "step": 4473
    },
    {
      "epoch": 0.23103537309579136,
      "grad_norm": 13.453790664672852,
      "learning_rate": 8.819785658259462e-06,
      "loss": 0.1362,
      "step": 4474
    },
    {
      "epoch": 0.2310870126516912,
      "grad_norm": 4.439918518066406,
      "learning_rate": 8.819256900576756e-06,
      "loss": 0.0651,
      "step": 4475
    },
    {
      "epoch": 0.23113865220759103,
      "grad_norm": 7.011351108551025,
      "learning_rate": 8.818728040331434e-06,
      "loss": 0.0492,
      "step": 4476
    },
    {
      "epoch": 0.23119029176349085,
      "grad_norm": 11.853463172912598,
      "learning_rate": 8.818199077537701e-06,
      "loss": 0.1288,
      "step": 4477
    },
    {
      "epoch": 0.23124193131939066,
      "grad_norm": 3.0100934505462646,
      "learning_rate": 8.817670012209763e-06,
      "loss": 0.0297,
      "step": 4478
    },
    {
      "epoch": 0.23129357087529048,
      "grad_norm": 1.9048655033111572,
      "learning_rate": 8.817140844361822e-06,
      "loss": 0.0435,
      "step": 4479
    },
    {
      "epoch": 0.2313452104311903,
      "grad_norm": 7.1518378257751465,
      "learning_rate": 8.816611574008094e-06,
      "loss": 0.0436,
      "step": 4480
    },
    {
      "epoch": 0.23139684998709012,
      "grad_norm": 1.6074661016464233,
      "learning_rate": 8.816082201162787e-06,
      "loss": 0.015,
      "step": 4481
    },
    {
      "epoch": 0.23144848954298994,
      "grad_norm": 5.60258150100708,
      "learning_rate": 8.815552725840121e-06,
      "loss": 0.064,
      "step": 4482
    },
    {
      "epoch": 0.23150012909888976,
      "grad_norm": 3.054370641708374,
      "learning_rate": 8.815023148054314e-06,
      "loss": 0.0462,
      "step": 4483
    },
    {
      "epoch": 0.23155176865478957,
      "grad_norm": 6.388960838317871,
      "learning_rate": 8.814493467819584e-06,
      "loss": 0.1002,
      "step": 4484
    },
    {
      "epoch": 0.2316034082106894,
      "grad_norm": 10.04162311553955,
      "learning_rate": 8.813963685150158e-06,
      "loss": 0.0547,
      "step": 4485
    },
    {
      "epoch": 0.2316550477665892,
      "grad_norm": 6.378443241119385,
      "learning_rate": 8.813433800060262e-06,
      "loss": 0.1031,
      "step": 4486
    },
    {
      "epoch": 0.23170668732248903,
      "grad_norm": 5.52916145324707,
      "learning_rate": 8.812903812564127e-06,
      "loss": 0.0661,
      "step": 4487
    },
    {
      "epoch": 0.23175832687838885,
      "grad_norm": 3.706855058670044,
      "learning_rate": 8.812373722675985e-06,
      "loss": 0.0649,
      "step": 4488
    },
    {
      "epoch": 0.23180996643428867,
      "grad_norm": 2.4684829711914062,
      "learning_rate": 8.811843530410067e-06,
      "loss": 0.0324,
      "step": 4489
    },
    {
      "epoch": 0.23186160599018849,
      "grad_norm": 25.975814819335938,
      "learning_rate": 8.811313235780617e-06,
      "loss": 0.0786,
      "step": 4490
    },
    {
      "epoch": 0.2319132455460883,
      "grad_norm": 13.1110200881958,
      "learning_rate": 8.81078283880187e-06,
      "loss": 0.1994,
      "step": 4491
    },
    {
      "epoch": 0.23196488510198812,
      "grad_norm": 2.669548749923706,
      "learning_rate": 8.810252339488074e-06,
      "loss": 0.0187,
      "step": 4492
    },
    {
      "epoch": 0.23201652465788794,
      "grad_norm": 3.9194066524505615,
      "learning_rate": 8.809721737853472e-06,
      "loss": 0.0668,
      "step": 4493
    },
    {
      "epoch": 0.23206816421378776,
      "grad_norm": 10.332257270812988,
      "learning_rate": 8.809191033912315e-06,
      "loss": 0.0991,
      "step": 4494
    },
    {
      "epoch": 0.23211980376968758,
      "grad_norm": 6.595093727111816,
      "learning_rate": 8.808660227678851e-06,
      "loss": 0.0453,
      "step": 4495
    },
    {
      "epoch": 0.2321714433255874,
      "grad_norm": 8.308650016784668,
      "learning_rate": 8.808129319167338e-06,
      "loss": 0.0883,
      "step": 4496
    },
    {
      "epoch": 0.23222308288148721,
      "grad_norm": 2.081465721130371,
      "learning_rate": 8.807598308392031e-06,
      "loss": 0.0402,
      "step": 4497
    },
    {
      "epoch": 0.23227472243738703,
      "grad_norm": 5.167535305023193,
      "learning_rate": 8.807067195367191e-06,
      "loss": 0.1678,
      "step": 4498
    },
    {
      "epoch": 0.23232636199328685,
      "grad_norm": 5.38110876083374,
      "learning_rate": 8.80653598010708e-06,
      "loss": 0.0831,
      "step": 4499
    },
    {
      "epoch": 0.23237800154918667,
      "grad_norm": 7.837052345275879,
      "learning_rate": 8.806004662625964e-06,
      "loss": 0.037,
      "step": 4500
    },
    {
      "epoch": 0.2324296411050865,
      "grad_norm": 21.673925399780273,
      "learning_rate": 8.80547324293811e-06,
      "loss": 0.0889,
      "step": 4501
    },
    {
      "epoch": 0.2324812806609863,
      "grad_norm": 3.310927629470825,
      "learning_rate": 8.804941721057789e-06,
      "loss": 0.0345,
      "step": 4502
    },
    {
      "epoch": 0.23253292021688612,
      "grad_norm": 6.310761451721191,
      "learning_rate": 8.804410096999275e-06,
      "loss": 0.0568,
      "step": 4503
    },
    {
      "epoch": 0.23258455977278594,
      "grad_norm": 3.216312885284424,
      "learning_rate": 8.803878370776843e-06,
      "loss": 0.0602,
      "step": 4504
    },
    {
      "epoch": 0.23263619932868576,
      "grad_norm": 6.152970790863037,
      "learning_rate": 8.803346542404773e-06,
      "loss": 0.0952,
      "step": 4505
    },
    {
      "epoch": 0.23268783888458558,
      "grad_norm": 8.529510498046875,
      "learning_rate": 8.802814611897348e-06,
      "loss": 0.087,
      "step": 4506
    },
    {
      "epoch": 0.23273947844048543,
      "grad_norm": 12.392685890197754,
      "learning_rate": 8.80228257926885e-06,
      "loss": 0.1173,
      "step": 4507
    },
    {
      "epoch": 0.23279111799638524,
      "grad_norm": 5.173950672149658,
      "learning_rate": 8.80175044453357e-06,
      "loss": 0.0666,
      "step": 4508
    },
    {
      "epoch": 0.23284275755228506,
      "grad_norm": 8.362259864807129,
      "learning_rate": 8.801218207705793e-06,
      "loss": 0.1083,
      "step": 4509
    },
    {
      "epoch": 0.23289439710818488,
      "grad_norm": 7.668168544769287,
      "learning_rate": 8.800685868799815e-06,
      "loss": 0.0728,
      "step": 4510
    },
    {
      "epoch": 0.2329460366640847,
      "grad_norm": 10.983780860900879,
      "learning_rate": 8.80015342782993e-06,
      "loss": 0.0666,
      "step": 4511
    },
    {
      "epoch": 0.23299767621998452,
      "grad_norm": 9.175789833068848,
      "learning_rate": 8.799620884810438e-06,
      "loss": 0.0863,
      "step": 4512
    },
    {
      "epoch": 0.23304931577588434,
      "grad_norm": 2.1260030269622803,
      "learning_rate": 8.799088239755638e-06,
      "loss": 0.0582,
      "step": 4513
    },
    {
      "epoch": 0.23310095533178415,
      "grad_norm": 6.027707576751709,
      "learning_rate": 8.798555492679835e-06,
      "loss": 0.0698,
      "step": 4514
    },
    {
      "epoch": 0.23315259488768397,
      "grad_norm": 2.318974494934082,
      "learning_rate": 8.798022643597336e-06,
      "loss": 0.0468,
      "step": 4515
    },
    {
      "epoch": 0.2332042344435838,
      "grad_norm": 3.649440288543701,
      "learning_rate": 8.797489692522447e-06,
      "loss": 0.0626,
      "step": 4516
    },
    {
      "epoch": 0.2332558739994836,
      "grad_norm": 7.901430606842041,
      "learning_rate": 8.796956639469482e-06,
      "loss": 0.0805,
      "step": 4517
    },
    {
      "epoch": 0.23330751355538343,
      "grad_norm": 5.351517200469971,
      "learning_rate": 8.796423484452756e-06,
      "loss": 0.0663,
      "step": 4518
    },
    {
      "epoch": 0.23335915311128325,
      "grad_norm": 6.775315284729004,
      "learning_rate": 8.795890227486588e-06,
      "loss": 0.0866,
      "step": 4519
    },
    {
      "epoch": 0.23341079266718306,
      "grad_norm": 6.782743453979492,
      "learning_rate": 8.795356868585294e-06,
      "loss": 0.0909,
      "step": 4520
    },
    {
      "epoch": 0.23346243222308288,
      "grad_norm": 5.435931205749512,
      "learning_rate": 8.7948234077632e-06,
      "loss": 0.0833,
      "step": 4521
    },
    {
      "epoch": 0.2335140717789827,
      "grad_norm": 3.2945303916931152,
      "learning_rate": 8.794289845034632e-06,
      "loss": 0.0623,
      "step": 4522
    },
    {
      "epoch": 0.23356571133488252,
      "grad_norm": 2.275622844696045,
      "learning_rate": 8.793756180413915e-06,
      "loss": 0.0374,
      "step": 4523
    },
    {
      "epoch": 0.23361735089078234,
      "grad_norm": 22.898801803588867,
      "learning_rate": 8.793222413915383e-06,
      "loss": 0.0946,
      "step": 4524
    },
    {
      "epoch": 0.23366899044668216,
      "grad_norm": 2.776442527770996,
      "learning_rate": 8.792688545553369e-06,
      "loss": 0.0539,
      "step": 4525
    },
    {
      "epoch": 0.23372063000258197,
      "grad_norm": 6.028848171234131,
      "learning_rate": 8.79215457534221e-06,
      "loss": 0.0809,
      "step": 4526
    },
    {
      "epoch": 0.2337722695584818,
      "grad_norm": 5.379674911499023,
      "learning_rate": 8.791620503296242e-06,
      "loss": 0.0755,
      "step": 4527
    },
    {
      "epoch": 0.2338239091143816,
      "grad_norm": 3.7757301330566406,
      "learning_rate": 8.791086329429812e-06,
      "loss": 0.06,
      "step": 4528
    },
    {
      "epoch": 0.23387554867028143,
      "grad_norm": 1.9272511005401611,
      "learning_rate": 8.790552053757262e-06,
      "loss": 0.0471,
      "step": 4529
    },
    {
      "epoch": 0.23392718822618125,
      "grad_norm": 5.515166759490967,
      "learning_rate": 8.790017676292939e-06,
      "loss": 0.0491,
      "step": 4530
    },
    {
      "epoch": 0.23397882778208107,
      "grad_norm": 7.593656539916992,
      "learning_rate": 8.789483197051195e-06,
      "loss": 0.0759,
      "step": 4531
    },
    {
      "epoch": 0.23403046733798089,
      "grad_norm": 7.141937255859375,
      "learning_rate": 8.788948616046383e-06,
      "loss": 0.1015,
      "step": 4532
    },
    {
      "epoch": 0.2340821068938807,
      "grad_norm": 5.600403308868408,
      "learning_rate": 8.788413933292856e-06,
      "loss": 0.0884,
      "step": 4533
    },
    {
      "epoch": 0.23413374644978052,
      "grad_norm": 13.107383728027344,
      "learning_rate": 8.787879148804975e-06,
      "loss": 0.0605,
      "step": 4534
    },
    {
      "epoch": 0.23418538600568034,
      "grad_norm": 5.880209922790527,
      "learning_rate": 8.7873442625971e-06,
      "loss": 0.1084,
      "step": 4535
    },
    {
      "epoch": 0.23423702556158016,
      "grad_norm": 4.432179927825928,
      "learning_rate": 8.786809274683596e-06,
      "loss": 0.0505,
      "step": 4536
    },
    {
      "epoch": 0.23428866511747998,
      "grad_norm": 6.726681232452393,
      "learning_rate": 8.786274185078826e-06,
      "loss": 0.1083,
      "step": 4537
    },
    {
      "epoch": 0.2343403046733798,
      "grad_norm": 3.821552276611328,
      "learning_rate": 8.785738993797165e-06,
      "loss": 0.0391,
      "step": 4538
    },
    {
      "epoch": 0.23439194422927964,
      "grad_norm": 16.601985931396484,
      "learning_rate": 8.785203700852982e-06,
      "loss": 0.2149,
      "step": 4539
    },
    {
      "epoch": 0.23444358378517946,
      "grad_norm": 13.739940643310547,
      "learning_rate": 8.784668306260652e-06,
      "loss": 0.0836,
      "step": 4540
    },
    {
      "epoch": 0.23449522334107928,
      "grad_norm": 25.30282211303711,
      "learning_rate": 8.784132810034552e-06,
      "loss": 0.224,
      "step": 4541
    },
    {
      "epoch": 0.2345468628969791,
      "grad_norm": 3.988621950149536,
      "learning_rate": 8.783597212189063e-06,
      "loss": 0.0433,
      "step": 4542
    },
    {
      "epoch": 0.23459850245287892,
      "grad_norm": 6.1954345703125,
      "learning_rate": 8.783061512738566e-06,
      "loss": 0.066,
      "step": 4543
    },
    {
      "epoch": 0.23465014200877873,
      "grad_norm": 9.880227088928223,
      "learning_rate": 8.782525711697451e-06,
      "loss": 0.1146,
      "step": 4544
    },
    {
      "epoch": 0.23470178156467855,
      "grad_norm": 6.748126983642578,
      "learning_rate": 8.781989809080103e-06,
      "loss": 0.0673,
      "step": 4545
    },
    {
      "epoch": 0.23475342112057837,
      "grad_norm": 6.540605545043945,
      "learning_rate": 8.781453804900914e-06,
      "loss": 0.0849,
      "step": 4546
    },
    {
      "epoch": 0.2348050606764782,
      "grad_norm": 4.353790760040283,
      "learning_rate": 8.780917699174279e-06,
      "loss": 0.0709,
      "step": 4547
    },
    {
      "epoch": 0.234856700232378,
      "grad_norm": 5.137007713317871,
      "learning_rate": 8.780381491914592e-06,
      "loss": 0.0366,
      "step": 4548
    },
    {
      "epoch": 0.23490833978827783,
      "grad_norm": 2.8694300651550293,
      "learning_rate": 8.779845183136255e-06,
      "loss": 0.0468,
      "step": 4549
    },
    {
      "epoch": 0.23495997934417764,
      "grad_norm": 1.9862596988677979,
      "learning_rate": 8.779308772853669e-06,
      "loss": 0.0588,
      "step": 4550
    },
    {
      "epoch": 0.23501161890007746,
      "grad_norm": 2.600240468978882,
      "learning_rate": 8.778772261081238e-06,
      "loss": 0.0695,
      "step": 4551
    },
    {
      "epoch": 0.23506325845597728,
      "grad_norm": 2.7516040802001953,
      "learning_rate": 8.77823564783337e-06,
      "loss": 0.0556,
      "step": 4552
    },
    {
      "epoch": 0.2351148980118771,
      "grad_norm": 8.375221252441406,
      "learning_rate": 8.777698933124477e-06,
      "loss": 0.0901,
      "step": 4553
    },
    {
      "epoch": 0.23516653756777692,
      "grad_norm": 2.352897882461548,
      "learning_rate": 8.77716211696897e-06,
      "loss": 0.0213,
      "step": 4554
    },
    {
      "epoch": 0.23521817712367674,
      "grad_norm": 12.199763298034668,
      "learning_rate": 8.776625199381266e-06,
      "loss": 0.1039,
      "step": 4555
    },
    {
      "epoch": 0.23526981667957655,
      "grad_norm": 4.695825099945068,
      "learning_rate": 8.77608818037578e-06,
      "loss": 0.0444,
      "step": 4556
    },
    {
      "epoch": 0.23532145623547637,
      "grad_norm": 9.169820785522461,
      "learning_rate": 8.77555105996694e-06,
      "loss": 0.075,
      "step": 4557
    },
    {
      "epoch": 0.2353730957913762,
      "grad_norm": 7.623353958129883,
      "learning_rate": 8.775013838169165e-06,
      "loss": 0.1027,
      "step": 4558
    },
    {
      "epoch": 0.235424735347276,
      "grad_norm": 4.142362594604492,
      "learning_rate": 8.774476514996879e-06,
      "loss": 0.0658,
      "step": 4559
    },
    {
      "epoch": 0.23547637490317583,
      "grad_norm": 5.894446849822998,
      "learning_rate": 8.773939090464517e-06,
      "loss": 0.0556,
      "step": 4560
    },
    {
      "epoch": 0.23552801445907565,
      "grad_norm": 5.363933563232422,
      "learning_rate": 8.773401564586508e-06,
      "loss": 0.0448,
      "step": 4561
    },
    {
      "epoch": 0.23557965401497546,
      "grad_norm": 4.024179458618164,
      "learning_rate": 8.772863937377289e-06,
      "loss": 0.0423,
      "step": 4562
    },
    {
      "epoch": 0.23563129357087528,
      "grad_norm": 6.949131011962891,
      "learning_rate": 8.772326208851294e-06,
      "loss": 0.06,
      "step": 4563
    },
    {
      "epoch": 0.2356829331267751,
      "grad_norm": 4.263096332550049,
      "learning_rate": 8.771788379022966e-06,
      "loss": 0.0477,
      "step": 4564
    },
    {
      "epoch": 0.23573457268267492,
      "grad_norm": 3.1503684520721436,
      "learning_rate": 8.771250447906747e-06,
      "loss": 0.0353,
      "step": 4565
    },
    {
      "epoch": 0.23578621223857474,
      "grad_norm": 4.189324378967285,
      "learning_rate": 8.770712415517084e-06,
      "loss": 0.0558,
      "step": 4566
    },
    {
      "epoch": 0.23583785179447456,
      "grad_norm": 7.294341564178467,
      "learning_rate": 8.770174281868423e-06,
      "loss": 0.1274,
      "step": 4567
    },
    {
      "epoch": 0.23588949135037438,
      "grad_norm": 19.09000587463379,
      "learning_rate": 8.769636046975216e-06,
      "loss": 0.1926,
      "step": 4568
    },
    {
      "epoch": 0.2359411309062742,
      "grad_norm": 7.569113731384277,
      "learning_rate": 8.769097710851915e-06,
      "loss": 0.0989,
      "step": 4569
    },
    {
      "epoch": 0.235992770462174,
      "grad_norm": 10.565645217895508,
      "learning_rate": 8.76855927351298e-06,
      "loss": 0.1447,
      "step": 4570
    },
    {
      "epoch": 0.23604441001807386,
      "grad_norm": 13.275025367736816,
      "learning_rate": 8.768020734972871e-06,
      "loss": 0.126,
      "step": 4571
    },
    {
      "epoch": 0.23609604957397368,
      "grad_norm": 7.844635009765625,
      "learning_rate": 8.767482095246046e-06,
      "loss": 0.1258,
      "step": 4572
    },
    {
      "epoch": 0.2361476891298735,
      "grad_norm": 4.001683235168457,
      "learning_rate": 8.76694335434697e-06,
      "loss": 0.0684,
      "step": 4573
    },
    {
      "epoch": 0.2361993286857733,
      "grad_norm": 2.8710551261901855,
      "learning_rate": 8.766404512290111e-06,
      "loss": 0.0605,
      "step": 4574
    },
    {
      "epoch": 0.23625096824167313,
      "grad_norm": 2.624141216278076,
      "learning_rate": 8.765865569089942e-06,
      "loss": 0.0631,
      "step": 4575
    },
    {
      "epoch": 0.23630260779757295,
      "grad_norm": 6.701535224914551,
      "learning_rate": 8.765326524760934e-06,
      "loss": 0.0562,
      "step": 4576
    },
    {
      "epoch": 0.23635424735347277,
      "grad_norm": 2.6511611938476562,
      "learning_rate": 8.764787379317562e-06,
      "loss": 0.0307,
      "step": 4577
    },
    {
      "epoch": 0.2364058869093726,
      "grad_norm": 2.89941668510437,
      "learning_rate": 8.764248132774305e-06,
      "loss": 0.0534,
      "step": 4578
    },
    {
      "epoch": 0.2364575264652724,
      "grad_norm": 2.7652266025543213,
      "learning_rate": 8.763708785145641e-06,
      "loss": 0.0562,
      "step": 4579
    },
    {
      "epoch": 0.23650916602117222,
      "grad_norm": 3.772684097290039,
      "learning_rate": 8.763169336446059e-06,
      "loss": 0.0824,
      "step": 4580
    },
    {
      "epoch": 0.23656080557707204,
      "grad_norm": 6.315340042114258,
      "learning_rate": 8.76262978669004e-06,
      "loss": 0.0847,
      "step": 4581
    },
    {
      "epoch": 0.23661244513297186,
      "grad_norm": 4.336952209472656,
      "learning_rate": 8.762090135892075e-06,
      "loss": 0.066,
      "step": 4582
    },
    {
      "epoch": 0.23666408468887168,
      "grad_norm": 17.369050979614258,
      "learning_rate": 8.761550384066659e-06,
      "loss": 0.1941,
      "step": 4583
    },
    {
      "epoch": 0.2367157242447715,
      "grad_norm": 1.9066553115844727,
      "learning_rate": 8.761010531228282e-06,
      "loss": 0.0361,
      "step": 4584
    },
    {
      "epoch": 0.23676736380067132,
      "grad_norm": 8.877835273742676,
      "learning_rate": 8.760470577391445e-06,
      "loss": 0.0722,
      "step": 4585
    },
    {
      "epoch": 0.23681900335657113,
      "grad_norm": 4.106984615325928,
      "learning_rate": 8.759930522570645e-06,
      "loss": 0.0819,
      "step": 4586
    },
    {
      "epoch": 0.23687064291247095,
      "grad_norm": 12.473093032836914,
      "learning_rate": 8.759390366780386e-06,
      "loss": 0.11,
      "step": 4587
    },
    {
      "epoch": 0.23692228246837077,
      "grad_norm": 4.899022102355957,
      "learning_rate": 8.758850110035174e-06,
      "loss": 0.0335,
      "step": 4588
    },
    {
      "epoch": 0.2369739220242706,
      "grad_norm": 4.997825622558594,
      "learning_rate": 8.758309752349516e-06,
      "loss": 0.0706,
      "step": 4589
    },
    {
      "epoch": 0.2370255615801704,
      "grad_norm": 3.9410312175750732,
      "learning_rate": 8.757769293737922e-06,
      "loss": 0.0772,
      "step": 4590
    },
    {
      "epoch": 0.23707720113607023,
      "grad_norm": 4.141500473022461,
      "learning_rate": 8.757228734214908e-06,
      "loss": 0.0688,
      "step": 4591
    },
    {
      "epoch": 0.23712884069197004,
      "grad_norm": 7.838853359222412,
      "learning_rate": 8.756688073794988e-06,
      "loss": 0.1027,
      "step": 4592
    },
    {
      "epoch": 0.23718048024786986,
      "grad_norm": 4.396302700042725,
      "learning_rate": 8.756147312492681e-06,
      "loss": 0.0623,
      "step": 4593
    },
    {
      "epoch": 0.23723211980376968,
      "grad_norm": 6.1339497566223145,
      "learning_rate": 8.755606450322511e-06,
      "loss": 0.0957,
      "step": 4594
    },
    {
      "epoch": 0.2372837593596695,
      "grad_norm": 14.68421459197998,
      "learning_rate": 8.755065487299001e-06,
      "loss": 0.3312,
      "step": 4595
    },
    {
      "epoch": 0.23733539891556932,
      "grad_norm": 5.798393249511719,
      "learning_rate": 8.754524423436676e-06,
      "loss": 0.0768,
      "step": 4596
    },
    {
      "epoch": 0.23738703847146914,
      "grad_norm": 4.585227012634277,
      "learning_rate": 8.75398325875007e-06,
      "loss": 0.0478,
      "step": 4597
    },
    {
      "epoch": 0.23743867802736895,
      "grad_norm": 3.075993537902832,
      "learning_rate": 8.753441993253711e-06,
      "loss": 0.0395,
      "step": 4598
    },
    {
      "epoch": 0.23749031758326877,
      "grad_norm": 5.903932094573975,
      "learning_rate": 8.752900626962138e-06,
      "loss": 0.0847,
      "step": 4599
    },
    {
      "epoch": 0.2375419571391686,
      "grad_norm": 25.840730667114258,
      "learning_rate": 8.752359159889887e-06,
      "loss": 0.114,
      "step": 4600
    },
    {
      "epoch": 0.2375935966950684,
      "grad_norm": 9.323272705078125,
      "learning_rate": 8.751817592051496e-06,
      "loss": 0.0962,
      "step": 4601
    },
    {
      "epoch": 0.23764523625096823,
      "grad_norm": 3.5724172592163086,
      "learning_rate": 8.751275923461515e-06,
      "loss": 0.1113,
      "step": 4602
    },
    {
      "epoch": 0.23769687580686807,
      "grad_norm": 2.196716785430908,
      "learning_rate": 8.750734154134484e-06,
      "loss": 0.0268,
      "step": 4603
    },
    {
      "epoch": 0.2377485153627679,
      "grad_norm": 3.8282010555267334,
      "learning_rate": 8.750192284084953e-06,
      "loss": 0.0867,
      "step": 4604
    },
    {
      "epoch": 0.2378001549186677,
      "grad_norm": 8.241241455078125,
      "learning_rate": 8.749650313327477e-06,
      "loss": 0.1196,
      "step": 4605
    },
    {
      "epoch": 0.23785179447456753,
      "grad_norm": 8.441153526306152,
      "learning_rate": 8.749108241876605e-06,
      "loss": 0.089,
      "step": 4606
    },
    {
      "epoch": 0.23790343403046735,
      "grad_norm": 3.849796772003174,
      "learning_rate": 8.748566069746898e-06,
      "loss": 0.0558,
      "step": 4607
    },
    {
      "epoch": 0.23795507358636717,
      "grad_norm": 4.626833915710449,
      "learning_rate": 8.748023796952913e-06,
      "loss": 0.0832,
      "step": 4608
    },
    {
      "epoch": 0.23800671314226698,
      "grad_norm": 1.5602232217788696,
      "learning_rate": 8.747481423509211e-06,
      "loss": 0.0172,
      "step": 4609
    },
    {
      "epoch": 0.2380583526981668,
      "grad_norm": 3.954362392425537,
      "learning_rate": 8.746938949430362e-06,
      "loss": 0.0302,
      "step": 4610
    },
    {
      "epoch": 0.23810999225406662,
      "grad_norm": 2.1922659873962402,
      "learning_rate": 8.746396374730929e-06,
      "loss": 0.0456,
      "step": 4611
    },
    {
      "epoch": 0.23816163180996644,
      "grad_norm": 4.411335468292236,
      "learning_rate": 8.745853699425484e-06,
      "loss": 0.0818,
      "step": 4612
    },
    {
      "epoch": 0.23821327136586626,
      "grad_norm": 3.5186870098114014,
      "learning_rate": 8.745310923528601e-06,
      "loss": 0.0571,
      "step": 4613
    },
    {
      "epoch": 0.23826491092176608,
      "grad_norm": 8.27772045135498,
      "learning_rate": 8.744768047054853e-06,
      "loss": 0.093,
      "step": 4614
    },
    {
      "epoch": 0.2383165504776659,
      "grad_norm": 19.251501083374023,
      "learning_rate": 8.74422507001882e-06,
      "loss": 0.136,
      "step": 4615
    },
    {
      "epoch": 0.2383681900335657,
      "grad_norm": 13.677136421203613,
      "learning_rate": 8.743681992435084e-06,
      "loss": 0.1306,
      "step": 4616
    },
    {
      "epoch": 0.23841982958946553,
      "grad_norm": 2.8466107845306396,
      "learning_rate": 8.743138814318227e-06,
      "loss": 0.0432,
      "step": 4617
    },
    {
      "epoch": 0.23847146914536535,
      "grad_norm": 15.10321044921875,
      "learning_rate": 8.742595535682838e-06,
      "loss": 0.0975,
      "step": 4618
    },
    {
      "epoch": 0.23852310870126517,
      "grad_norm": 2.324892282485962,
      "learning_rate": 8.742052156543504e-06,
      "loss": 0.0279,
      "step": 4619
    },
    {
      "epoch": 0.238574748257165,
      "grad_norm": 9.658395767211914,
      "learning_rate": 8.741508676914819e-06,
      "loss": 0.0707,
      "step": 4620
    },
    {
      "epoch": 0.2386263878130648,
      "grad_norm": 6.794988632202148,
      "learning_rate": 8.740965096811376e-06,
      "loss": 0.0691,
      "step": 4621
    },
    {
      "epoch": 0.23867802736896462,
      "grad_norm": 6.242916107177734,
      "learning_rate": 8.740421416247773e-06,
      "loss": 0.0784,
      "step": 4622
    },
    {
      "epoch": 0.23872966692486444,
      "grad_norm": 8.892073631286621,
      "learning_rate": 8.73987763523861e-06,
      "loss": 0.0449,
      "step": 4623
    },
    {
      "epoch": 0.23878130648076426,
      "grad_norm": 3.781538248062134,
      "learning_rate": 8.739333753798489e-06,
      "loss": 0.0672,
      "step": 4624
    },
    {
      "epoch": 0.23883294603666408,
      "grad_norm": 1.4241951704025269,
      "learning_rate": 8.738789771942016e-06,
      "loss": 0.0197,
      "step": 4625
    },
    {
      "epoch": 0.2388845855925639,
      "grad_norm": 8.178296089172363,
      "learning_rate": 8.7382456896838e-06,
      "loss": 0.054,
      "step": 4626
    },
    {
      "epoch": 0.23893622514846372,
      "grad_norm": 7.334721088409424,
      "learning_rate": 8.73770150703845e-06,
      "loss": 0.0838,
      "step": 4627
    },
    {
      "epoch": 0.23898786470436353,
      "grad_norm": 3.459764242172241,
      "learning_rate": 8.737157224020581e-06,
      "loss": 0.0329,
      "step": 4628
    },
    {
      "epoch": 0.23903950426026335,
      "grad_norm": 21.269153594970703,
      "learning_rate": 8.736612840644808e-06,
      "loss": 0.1038,
      "step": 4629
    },
    {
      "epoch": 0.23909114381616317,
      "grad_norm": 16.324710845947266,
      "learning_rate": 8.73606835692575e-06,
      "loss": 0.2641,
      "step": 4630
    },
    {
      "epoch": 0.239142783372063,
      "grad_norm": 7.141110897064209,
      "learning_rate": 8.735523772878033e-06,
      "loss": 0.0532,
      "step": 4631
    },
    {
      "epoch": 0.2391944229279628,
      "grad_norm": 18.4490909576416,
      "learning_rate": 8.734979088516274e-06,
      "loss": 0.1094,
      "step": 4632
    },
    {
      "epoch": 0.23924606248386263,
      "grad_norm": 6.013301372528076,
      "learning_rate": 8.734434303855105e-06,
      "loss": 0.0598,
      "step": 4633
    },
    {
      "epoch": 0.23929770203976244,
      "grad_norm": 8.769685745239258,
      "learning_rate": 8.73388941890915e-06,
      "loss": 0.2196,
      "step": 4634
    },
    {
      "epoch": 0.2393493415956623,
      "grad_norm": 5.477766513824463,
      "learning_rate": 8.733344433693051e-06,
      "loss": 0.0923,
      "step": 4635
    },
    {
      "epoch": 0.2394009811515621,
      "grad_norm": 10.113762855529785,
      "learning_rate": 8.732799348221438e-06,
      "loss": 0.1058,
      "step": 4636
    },
    {
      "epoch": 0.23945262070746193,
      "grad_norm": 4.996968746185303,
      "learning_rate": 8.732254162508945e-06,
      "loss": 0.0465,
      "step": 4637
    },
    {
      "epoch": 0.23950426026336175,
      "grad_norm": 6.854598045349121,
      "learning_rate": 8.731708876570218e-06,
      "loss": 0.0952,
      "step": 4638
    },
    {
      "epoch": 0.23955589981926156,
      "grad_norm": 3.0439565181732178,
      "learning_rate": 8.731163490419895e-06,
      "loss": 0.0943,
      "step": 4639
    },
    {
      "epoch": 0.23960753937516138,
      "grad_norm": 9.546829223632812,
      "learning_rate": 8.73061800407263e-06,
      "loss": 0.0802,
      "step": 4640
    },
    {
      "epoch": 0.2396591789310612,
      "grad_norm": 4.386401653289795,
      "learning_rate": 8.730072417543063e-06,
      "loss": 0.0717,
      "step": 4641
    },
    {
      "epoch": 0.23971081848696102,
      "grad_norm": 6.69783353805542,
      "learning_rate": 8.729526730845849e-06,
      "loss": 0.0865,
      "step": 4642
    },
    {
      "epoch": 0.23976245804286084,
      "grad_norm": 12.462156295776367,
      "learning_rate": 8.728980943995642e-06,
      "loss": 0.1475,
      "step": 4643
    },
    {
      "epoch": 0.23981409759876066,
      "grad_norm": 5.238275051116943,
      "learning_rate": 8.728435057007098e-06,
      "loss": 0.08,
      "step": 4644
    },
    {
      "epoch": 0.23986573715466047,
      "grad_norm": 3.2126150131225586,
      "learning_rate": 8.727889069894877e-06,
      "loss": 0.0819,
      "step": 4645
    },
    {
      "epoch": 0.2399173767105603,
      "grad_norm": 2.0848939418792725,
      "learning_rate": 8.72734298267364e-06,
      "loss": 0.0712,
      "step": 4646
    },
    {
      "epoch": 0.2399690162664601,
      "grad_norm": 6.416651248931885,
      "learning_rate": 8.726796795358052e-06,
      "loss": 0.0918,
      "step": 4647
    },
    {
      "epoch": 0.24002065582235993,
      "grad_norm": 3.0166749954223633,
      "learning_rate": 8.726250507962782e-06,
      "loss": 0.0642,
      "step": 4648
    },
    {
      "epoch": 0.24007229537825975,
      "grad_norm": 7.10011625289917,
      "learning_rate": 8.725704120502496e-06,
      "loss": 0.0798,
      "step": 4649
    },
    {
      "epoch": 0.24012393493415957,
      "grad_norm": 2.303504467010498,
      "learning_rate": 8.72515763299187e-06,
      "loss": 0.0408,
      "step": 4650
    },
    {
      "epoch": 0.24017557449005938,
      "grad_norm": 3.90132737159729,
      "learning_rate": 8.724611045445579e-06,
      "loss": 0.0883,
      "step": 4651
    },
    {
      "epoch": 0.2402272140459592,
      "grad_norm": 2.781651496887207,
      "learning_rate": 8.724064357878301e-06,
      "loss": 0.0793,
      "step": 4652
    },
    {
      "epoch": 0.24027885360185902,
      "grad_norm": 3.4603865146636963,
      "learning_rate": 8.723517570304717e-06,
      "loss": 0.0671,
      "step": 4653
    },
    {
      "epoch": 0.24033049315775884,
      "grad_norm": 7.531458854675293,
      "learning_rate": 8.722970682739508e-06,
      "loss": 0.1111,
      "step": 4654
    },
    {
      "epoch": 0.24038213271365866,
      "grad_norm": 16.37373924255371,
      "learning_rate": 8.722423695197364e-06,
      "loss": 0.1673,
      "step": 4655
    },
    {
      "epoch": 0.24043377226955848,
      "grad_norm": 4.846179008483887,
      "learning_rate": 8.721876607692971e-06,
      "loss": 0.0797,
      "step": 4656
    },
    {
      "epoch": 0.2404854118254583,
      "grad_norm": 5.074355602264404,
      "learning_rate": 8.721329420241023e-06,
      "loss": 0.0811,
      "step": 4657
    },
    {
      "epoch": 0.2405370513813581,
      "grad_norm": 7.552001953125,
      "learning_rate": 8.720782132856212e-06,
      "loss": 0.1521,
      "step": 4658
    },
    {
      "epoch": 0.24058869093725793,
      "grad_norm": 6.043913841247559,
      "learning_rate": 8.720234745553236e-06,
      "loss": 0.0888,
      "step": 4659
    },
    {
      "epoch": 0.24064033049315775,
      "grad_norm": 16.130117416381836,
      "learning_rate": 8.719687258346794e-06,
      "loss": 0.2025,
      "step": 4660
    },
    {
      "epoch": 0.24069197004905757,
      "grad_norm": 4.188620090484619,
      "learning_rate": 8.719139671251587e-06,
      "loss": 0.0484,
      "step": 4661
    },
    {
      "epoch": 0.2407436096049574,
      "grad_norm": 8.899707794189453,
      "learning_rate": 8.718591984282322e-06,
      "loss": 0.0764,
      "step": 4662
    },
    {
      "epoch": 0.2407952491608572,
      "grad_norm": 16.410795211791992,
      "learning_rate": 8.718044197453708e-06,
      "loss": 0.0937,
      "step": 4663
    },
    {
      "epoch": 0.24084688871675702,
      "grad_norm": 9.30457592010498,
      "learning_rate": 8.717496310780454e-06,
      "loss": 0.0659,
      "step": 4664
    },
    {
      "epoch": 0.24089852827265684,
      "grad_norm": 3.916309118270874,
      "learning_rate": 8.71694832427727e-06,
      "loss": 0.0491,
      "step": 4665
    },
    {
      "epoch": 0.24095016782855666,
      "grad_norm": 3.9257659912109375,
      "learning_rate": 8.716400237958873e-06,
      "loss": 0.0659,
      "step": 4666
    },
    {
      "epoch": 0.2410018073844565,
      "grad_norm": 7.167754173278809,
      "learning_rate": 8.715852051839984e-06,
      "loss": 0.1011,
      "step": 4667
    },
    {
      "epoch": 0.24105344694035633,
      "grad_norm": 4.50932502746582,
      "learning_rate": 8.715303765935324e-06,
      "loss": 0.0673,
      "step": 4668
    },
    {
      "epoch": 0.24110508649625614,
      "grad_norm": 4.851873874664307,
      "learning_rate": 8.714755380259613e-06,
      "loss": 0.0797,
      "step": 4669
    },
    {
      "epoch": 0.24115672605215596,
      "grad_norm": 7.5805344581604,
      "learning_rate": 8.71420689482758e-06,
      "loss": 0.093,
      "step": 4670
    },
    {
      "epoch": 0.24120836560805578,
      "grad_norm": 4.790761947631836,
      "learning_rate": 8.713658309653954e-06,
      "loss": 0.0658,
      "step": 4671
    },
    {
      "epoch": 0.2412600051639556,
      "grad_norm": 2.75260853767395,
      "learning_rate": 8.713109624753466e-06,
      "loss": 0.0396,
      "step": 4672
    },
    {
      "epoch": 0.24131164471985542,
      "grad_norm": 2.7308990955352783,
      "learning_rate": 8.712560840140851e-06,
      "loss": 0.0669,
      "step": 4673
    },
    {
      "epoch": 0.24136328427575524,
      "grad_norm": 1.9432991743087769,
      "learning_rate": 8.712011955830845e-06,
      "loss": 0.0316,
      "step": 4674
    },
    {
      "epoch": 0.24141492383165505,
      "grad_norm": 4.607336044311523,
      "learning_rate": 8.711462971838189e-06,
      "loss": 0.1457,
      "step": 4675
    },
    {
      "epoch": 0.24146656338755487,
      "grad_norm": 2.5355968475341797,
      "learning_rate": 8.710913888177627e-06,
      "loss": 0.0399,
      "step": 4676
    },
    {
      "epoch": 0.2415182029434547,
      "grad_norm": 6.291322231292725,
      "learning_rate": 8.7103647048639e-06,
      "loss": 0.0976,
      "step": 4677
    },
    {
      "epoch": 0.2415698424993545,
      "grad_norm": 6.267587184906006,
      "learning_rate": 8.70981542191176e-06,
      "loss": 0.0745,
      "step": 4678
    },
    {
      "epoch": 0.24162148205525433,
      "grad_norm": 3.768808603286743,
      "learning_rate": 8.709266039335953e-06,
      "loss": 0.0644,
      "step": 4679
    },
    {
      "epoch": 0.24167312161115415,
      "grad_norm": 7.8736572265625,
      "learning_rate": 8.708716557151236e-06,
      "loss": 0.138,
      "step": 4680
    },
    {
      "epoch": 0.24172476116705396,
      "grad_norm": 16.30666732788086,
      "learning_rate": 8.708166975372363e-06,
      "loss": 0.149,
      "step": 4681
    },
    {
      "epoch": 0.24177640072295378,
      "grad_norm": 4.638820648193359,
      "learning_rate": 8.707617294014095e-06,
      "loss": 0.0675,
      "step": 4682
    },
    {
      "epoch": 0.2418280402788536,
      "grad_norm": 5.628675937652588,
      "learning_rate": 8.70706751309119e-06,
      "loss": 0.1153,
      "step": 4683
    },
    {
      "epoch": 0.24187967983475342,
      "grad_norm": 5.640071392059326,
      "learning_rate": 8.706517632618414e-06,
      "loss": 0.0518,
      "step": 4684
    },
    {
      "epoch": 0.24193131939065324,
      "grad_norm": 8.1807222366333,
      "learning_rate": 8.705967652610531e-06,
      "loss": 0.1326,
      "step": 4685
    },
    {
      "epoch": 0.24198295894655306,
      "grad_norm": 14.243943214416504,
      "learning_rate": 8.705417573082314e-06,
      "loss": 0.0805,
      "step": 4686
    },
    {
      "epoch": 0.24203459850245287,
      "grad_norm": 15.027592658996582,
      "learning_rate": 8.704867394048531e-06,
      "loss": 0.083,
      "step": 4687
    },
    {
      "epoch": 0.2420862380583527,
      "grad_norm": 2.1063969135284424,
      "learning_rate": 8.704317115523956e-06,
      "loss": 0.0524,
      "step": 4688
    },
    {
      "epoch": 0.2421378776142525,
      "grad_norm": 10.454330444335938,
      "learning_rate": 8.703766737523371e-06,
      "loss": 0.1344,
      "step": 4689
    },
    {
      "epoch": 0.24218951717015233,
      "grad_norm": 4.9640913009643555,
      "learning_rate": 8.703216260061555e-06,
      "loss": 0.067,
      "step": 4690
    },
    {
      "epoch": 0.24224115672605215,
      "grad_norm": 11.479437828063965,
      "learning_rate": 8.702665683153287e-06,
      "loss": 0.1545,
      "step": 4691
    },
    {
      "epoch": 0.24229279628195197,
      "grad_norm": 1.824518084526062,
      "learning_rate": 8.702115006813354e-06,
      "loss": 0.0324,
      "step": 4692
    },
    {
      "epoch": 0.24234443583785178,
      "grad_norm": 2.871427297592163,
      "learning_rate": 8.701564231056546e-06,
      "loss": 0.0411,
      "step": 4693
    },
    {
      "epoch": 0.2423960753937516,
      "grad_norm": 3.3787240982055664,
      "learning_rate": 8.70101335589765e-06,
      "loss": 0.069,
      "step": 4694
    },
    {
      "epoch": 0.24244771494965142,
      "grad_norm": 1.9849826097488403,
      "learning_rate": 8.70046238135146e-06,
      "loss": 0.0385,
      "step": 4695
    },
    {
      "epoch": 0.24249935450555124,
      "grad_norm": 3.346064567565918,
      "learning_rate": 8.699911307432773e-06,
      "loss": 0.0372,
      "step": 4696
    },
    {
      "epoch": 0.24255099406145106,
      "grad_norm": 12.709136962890625,
      "learning_rate": 8.699360134156389e-06,
      "loss": 0.1517,
      "step": 4697
    },
    {
      "epoch": 0.2426026336173509,
      "grad_norm": 7.116477966308594,
      "learning_rate": 8.698808861537108e-06,
      "loss": 0.0888,
      "step": 4698
    },
    {
      "epoch": 0.24265427317325072,
      "grad_norm": 8.66405963897705,
      "learning_rate": 8.698257489589733e-06,
      "loss": 0.0501,
      "step": 4699
    },
    {
      "epoch": 0.24270591272915054,
      "grad_norm": 3.808051824569702,
      "learning_rate": 8.69770601832907e-06,
      "loss": 0.051,
      "step": 4700
    },
    {
      "epoch": 0.24275755228505036,
      "grad_norm": 3.6344165802001953,
      "learning_rate": 8.697154447769932e-06,
      "loss": 0.0562,
      "step": 4701
    },
    {
      "epoch": 0.24280919184095018,
      "grad_norm": 8.376076698303223,
      "learning_rate": 8.696602777927125e-06,
      "loss": 0.0999,
      "step": 4702
    },
    {
      "epoch": 0.24286083139685,
      "grad_norm": 10.078277587890625,
      "learning_rate": 8.696051008815467e-06,
      "loss": 0.1734,
      "step": 4703
    },
    {
      "epoch": 0.24291247095274981,
      "grad_norm": 6.266360282897949,
      "learning_rate": 8.695499140449776e-06,
      "loss": 0.0759,
      "step": 4704
    },
    {
      "epoch": 0.24296411050864963,
      "grad_norm": 5.970333576202393,
      "learning_rate": 8.694947172844873e-06,
      "loss": 0.0755,
      "step": 4705
    },
    {
      "epoch": 0.24301575006454945,
      "grad_norm": 5.286907196044922,
      "learning_rate": 8.694395106015576e-06,
      "loss": 0.0702,
      "step": 4706
    },
    {
      "epoch": 0.24306738962044927,
      "grad_norm": 3.614173173904419,
      "learning_rate": 8.693842939976715e-06,
      "loss": 0.0663,
      "step": 4707
    },
    {
      "epoch": 0.2431190291763491,
      "grad_norm": 8.586983680725098,
      "learning_rate": 8.693290674743115e-06,
      "loss": 0.0741,
      "step": 4708
    },
    {
      "epoch": 0.2431706687322489,
      "grad_norm": 4.092669486999512,
      "learning_rate": 8.692738310329607e-06,
      "loss": 0.0511,
      "step": 4709
    },
    {
      "epoch": 0.24322230828814873,
      "grad_norm": 9.068339347839355,
      "learning_rate": 8.692185846751024e-06,
      "loss": 0.0326,
      "step": 4710
    },
    {
      "epoch": 0.24327394784404854,
      "grad_norm": 8.051661491394043,
      "learning_rate": 8.691633284022204e-06,
      "loss": 0.1771,
      "step": 4711
    },
    {
      "epoch": 0.24332558739994836,
      "grad_norm": 3.174755334854126,
      "learning_rate": 8.691080622157984e-06,
      "loss": 0.0244,
      "step": 4712
    },
    {
      "epoch": 0.24337722695584818,
      "grad_norm": 4.366893291473389,
      "learning_rate": 8.690527861173203e-06,
      "loss": 0.0742,
      "step": 4713
    },
    {
      "epoch": 0.243428866511748,
      "grad_norm": 12.376099586486816,
      "learning_rate": 8.68997500108271e-06,
      "loss": 0.0649,
      "step": 4714
    },
    {
      "epoch": 0.24348050606764782,
      "grad_norm": 6.6246466636657715,
      "learning_rate": 8.689422041901348e-06,
      "loss": 0.0748,
      "step": 4715
    },
    {
      "epoch": 0.24353214562354764,
      "grad_norm": 7.656357765197754,
      "learning_rate": 8.688868983643966e-06,
      "loss": 0.0873,
      "step": 4716
    },
    {
      "epoch": 0.24358378517944745,
      "grad_norm": 3.347585439682007,
      "learning_rate": 8.688315826325416e-06,
      "loss": 0.072,
      "step": 4717
    },
    {
      "epoch": 0.24363542473534727,
      "grad_norm": 10.559614181518555,
      "learning_rate": 8.687762569960555e-06,
      "loss": 0.1193,
      "step": 4718
    },
    {
      "epoch": 0.2436870642912471,
      "grad_norm": 8.443565368652344,
      "learning_rate": 8.687209214564238e-06,
      "loss": 0.0992,
      "step": 4719
    },
    {
      "epoch": 0.2437387038471469,
      "grad_norm": 2.522111654281616,
      "learning_rate": 8.686655760151326e-06,
      "loss": 0.0377,
      "step": 4720
    },
    {
      "epoch": 0.24379034340304673,
      "grad_norm": 4.2093186378479,
      "learning_rate": 8.68610220673668e-06,
      "loss": 0.0858,
      "step": 4721
    },
    {
      "epoch": 0.24384198295894655,
      "grad_norm": 12.618629455566406,
      "learning_rate": 8.685548554335166e-06,
      "loss": 0.0924,
      "step": 4722
    },
    {
      "epoch": 0.24389362251484636,
      "grad_norm": 3.8802106380462646,
      "learning_rate": 8.684994802961653e-06,
      "loss": 0.0478,
      "step": 4723
    },
    {
      "epoch": 0.24394526207074618,
      "grad_norm": 10.496194839477539,
      "learning_rate": 8.684440952631009e-06,
      "loss": 0.1109,
      "step": 4724
    },
    {
      "epoch": 0.243996901626646,
      "grad_norm": 3.178284168243408,
      "learning_rate": 8.683887003358109e-06,
      "loss": 0.0582,
      "step": 4725
    },
    {
      "epoch": 0.24404854118254582,
      "grad_norm": 3.0664660930633545,
      "learning_rate": 8.683332955157829e-06,
      "loss": 0.0631,
      "step": 4726
    },
    {
      "epoch": 0.24410018073844564,
      "grad_norm": 6.567568302154541,
      "learning_rate": 8.682778808045045e-06,
      "loss": 0.0149,
      "step": 4727
    },
    {
      "epoch": 0.24415182029434546,
      "grad_norm": 10.78017520904541,
      "learning_rate": 8.68222456203464e-06,
      "loss": 0.0864,
      "step": 4728
    },
    {
      "epoch": 0.24420345985024527,
      "grad_norm": 8.5984525680542,
      "learning_rate": 8.681670217141499e-06,
      "loss": 0.1018,
      "step": 4729
    },
    {
      "epoch": 0.24425509940614512,
      "grad_norm": 3.0981996059417725,
      "learning_rate": 8.681115773380507e-06,
      "loss": 0.0503,
      "step": 4730
    },
    {
      "epoch": 0.24430673896204494,
      "grad_norm": 6.592211723327637,
      "learning_rate": 8.680561230766552e-06,
      "loss": 0.0984,
      "step": 4731
    },
    {
      "epoch": 0.24435837851794476,
      "grad_norm": 3.4043099880218506,
      "learning_rate": 8.680006589314526e-06,
      "loss": 0.0259,
      "step": 4732
    },
    {
      "epoch": 0.24441001807384458,
      "grad_norm": 8.495909690856934,
      "learning_rate": 8.679451849039326e-06,
      "loss": 0.0468,
      "step": 4733
    },
    {
      "epoch": 0.2444616576297444,
      "grad_norm": 4.645732879638672,
      "learning_rate": 8.678897009955847e-06,
      "loss": 0.0595,
      "step": 4734
    },
    {
      "epoch": 0.2445132971856442,
      "grad_norm": 3.320819139480591,
      "learning_rate": 8.67834207207899e-06,
      "loss": 0.0552,
      "step": 4735
    },
    {
      "epoch": 0.24456493674154403,
      "grad_norm": 5.816509246826172,
      "learning_rate": 8.677787035423653e-06,
      "loss": 0.1059,
      "step": 4736
    },
    {
      "epoch": 0.24461657629744385,
      "grad_norm": 5.752998352050781,
      "learning_rate": 8.677231900004747e-06,
      "loss": 0.072,
      "step": 4737
    },
    {
      "epoch": 0.24466821585334367,
      "grad_norm": 4.457581996917725,
      "learning_rate": 8.676676665837176e-06,
      "loss": 0.0604,
      "step": 4738
    },
    {
      "epoch": 0.2447198554092435,
      "grad_norm": 8.03306770324707,
      "learning_rate": 8.676121332935852e-06,
      "loss": 0.0635,
      "step": 4739
    },
    {
      "epoch": 0.2447714949651433,
      "grad_norm": 9.37848949432373,
      "learning_rate": 8.675565901315685e-06,
      "loss": 0.0923,
      "step": 4740
    },
    {
      "epoch": 0.24482313452104312,
      "grad_norm": 8.764623641967773,
      "learning_rate": 8.675010370991594e-06,
      "loss": 0.0935,
      "step": 4741
    },
    {
      "epoch": 0.24487477407694294,
      "grad_norm": 4.185247421264648,
      "learning_rate": 8.674454741978497e-06,
      "loss": 0.0395,
      "step": 4742
    },
    {
      "epoch": 0.24492641363284276,
      "grad_norm": 26.111486434936523,
      "learning_rate": 8.673899014291313e-06,
      "loss": 0.1295,
      "step": 4743
    },
    {
      "epoch": 0.24497805318874258,
      "grad_norm": 12.992602348327637,
      "learning_rate": 8.673343187944967e-06,
      "loss": 0.0918,
      "step": 4744
    },
    {
      "epoch": 0.2450296927446424,
      "grad_norm": 4.549017429351807,
      "learning_rate": 8.672787262954384e-06,
      "loss": 0.0785,
      "step": 4745
    },
    {
      "epoch": 0.24508133230054222,
      "grad_norm": 5.971802711486816,
      "learning_rate": 8.672231239334495e-06,
      "loss": 0.1073,
      "step": 4746
    },
    {
      "epoch": 0.24513297185644203,
      "grad_norm": 3.4546892642974854,
      "learning_rate": 8.671675117100228e-06,
      "loss": 0.0428,
      "step": 4747
    },
    {
      "epoch": 0.24518461141234185,
      "grad_norm": 3.6347453594207764,
      "learning_rate": 8.671118896266519e-06,
      "loss": 0.0286,
      "step": 4748
    },
    {
      "epoch": 0.24523625096824167,
      "grad_norm": 3.5900650024414062,
      "learning_rate": 8.670562576848306e-06,
      "loss": 0.0805,
      "step": 4749
    },
    {
      "epoch": 0.2452878905241415,
      "grad_norm": 2.693758487701416,
      "learning_rate": 8.67000615886053e-06,
      "loss": 0.0442,
      "step": 4750
    },
    {
      "epoch": 0.2453395300800413,
      "grad_norm": 6.8292131423950195,
      "learning_rate": 8.669449642318126e-06,
      "loss": 0.0809,
      "step": 4751
    },
    {
      "epoch": 0.24539116963594113,
      "grad_norm": 6.490137100219727,
      "learning_rate": 8.668893027236046e-06,
      "loss": 0.1549,
      "step": 4752
    },
    {
      "epoch": 0.24544280919184094,
      "grad_norm": 6.543389797210693,
      "learning_rate": 8.668336313629234e-06,
      "loss": 0.0481,
      "step": 4753
    },
    {
      "epoch": 0.24549444874774076,
      "grad_norm": 7.546401500701904,
      "learning_rate": 8.66777950151264e-06,
      "loss": 0.0953,
      "step": 4754
    },
    {
      "epoch": 0.24554608830364058,
      "grad_norm": 13.672233581542969,
      "learning_rate": 8.667222590901218e-06,
      "loss": 0.1501,
      "step": 4755
    },
    {
      "epoch": 0.2455977278595404,
      "grad_norm": 6.067549705505371,
      "learning_rate": 8.666665581809924e-06,
      "loss": 0.0882,
      "step": 4756
    },
    {
      "epoch": 0.24564936741544022,
      "grad_norm": 6.153543949127197,
      "learning_rate": 8.666108474253712e-06,
      "loss": 0.0852,
      "step": 4757
    },
    {
      "epoch": 0.24570100697134004,
      "grad_norm": 4.4776387214660645,
      "learning_rate": 8.665551268247547e-06,
      "loss": 0.0611,
      "step": 4758
    },
    {
      "epoch": 0.24575264652723985,
      "grad_norm": 7.153636455535889,
      "learning_rate": 8.66499396380639e-06,
      "loss": 0.0661,
      "step": 4759
    },
    {
      "epoch": 0.24580428608313967,
      "grad_norm": 3.004150152206421,
      "learning_rate": 8.664436560945208e-06,
      "loss": 0.0508,
      "step": 4760
    },
    {
      "epoch": 0.2458559256390395,
      "grad_norm": 4.079917907714844,
      "learning_rate": 8.663879059678968e-06,
      "loss": 0.0621,
      "step": 4761
    },
    {
      "epoch": 0.24590756519493934,
      "grad_norm": 3.478641986846924,
      "learning_rate": 8.663321460022643e-06,
      "loss": 0.0495,
      "step": 4762
    },
    {
      "epoch": 0.24595920475083916,
      "grad_norm": 10.193923950195312,
      "learning_rate": 8.662763761991203e-06,
      "loss": 0.0818,
      "step": 4763
    },
    {
      "epoch": 0.24601084430673897,
      "grad_norm": 7.566965579986572,
      "learning_rate": 8.66220596559963e-06,
      "loss": 0.0399,
      "step": 4764
    },
    {
      "epoch": 0.2460624838626388,
      "grad_norm": 2.35920786857605,
      "learning_rate": 8.6616480708629e-06,
      "loss": 0.0531,
      "step": 4765
    },
    {
      "epoch": 0.2461141234185386,
      "grad_norm": 3.800414562225342,
      "learning_rate": 8.661090077795997e-06,
      "loss": 0.0674,
      "step": 4766
    },
    {
      "epoch": 0.24616576297443843,
      "grad_norm": 5.004693031311035,
      "learning_rate": 8.660531986413902e-06,
      "loss": 0.051,
      "step": 4767
    },
    {
      "epoch": 0.24621740253033825,
      "grad_norm": 3.967254400253296,
      "learning_rate": 8.659973796731604e-06,
      "loss": 0.0731,
      "step": 4768
    },
    {
      "epoch": 0.24626904208623807,
      "grad_norm": 8.079118728637695,
      "learning_rate": 8.65941550876409e-06,
      "loss": 0.0614,
      "step": 4769
    },
    {
      "epoch": 0.24632068164213788,
      "grad_norm": 3.60455322265625,
      "learning_rate": 8.658857122526358e-06,
      "loss": 0.0749,
      "step": 4770
    },
    {
      "epoch": 0.2463723211980377,
      "grad_norm": 6.57417631149292,
      "learning_rate": 8.658298638033398e-06,
      "loss": 0.1076,
      "step": 4771
    },
    {
      "epoch": 0.24642396075393752,
      "grad_norm": 8.205937385559082,
      "learning_rate": 8.65774005530021e-06,
      "loss": 0.0931,
      "step": 4772
    },
    {
      "epoch": 0.24647560030983734,
      "grad_norm": 7.8176703453063965,
      "learning_rate": 8.657181374341792e-06,
      "loss": 0.0488,
      "step": 4773
    },
    {
      "epoch": 0.24652723986573716,
      "grad_norm": 3.033121347427368,
      "learning_rate": 8.65662259517315e-06,
      "loss": 0.0477,
      "step": 4774
    },
    {
      "epoch": 0.24657887942163698,
      "grad_norm": 6.195152759552002,
      "learning_rate": 8.656063717809286e-06,
      "loss": 0.0773,
      "step": 4775
    },
    {
      "epoch": 0.2466305189775368,
      "grad_norm": 2.6454691886901855,
      "learning_rate": 8.65550474226521e-06,
      "loss": 0.0508,
      "step": 4776
    },
    {
      "epoch": 0.2466821585334366,
      "grad_norm": 7.2091383934021,
      "learning_rate": 8.654945668555935e-06,
      "loss": 0.1773,
      "step": 4777
    },
    {
      "epoch": 0.24673379808933643,
      "grad_norm": 2.2588207721710205,
      "learning_rate": 8.65438649669647e-06,
      "loss": 0.038,
      "step": 4778
    },
    {
      "epoch": 0.24678543764523625,
      "grad_norm": 2.657752752304077,
      "learning_rate": 8.653827226701835e-06,
      "loss": 0.0514,
      "step": 4779
    },
    {
      "epoch": 0.24683707720113607,
      "grad_norm": 7.582340717315674,
      "learning_rate": 8.653267858587045e-06,
      "loss": 0.1807,
      "step": 4780
    },
    {
      "epoch": 0.2468887167570359,
      "grad_norm": 4.2055344581604,
      "learning_rate": 8.652708392367121e-06,
      "loss": 0.0522,
      "step": 4781
    },
    {
      "epoch": 0.2469403563129357,
      "grad_norm": 5.477573871612549,
      "learning_rate": 8.652148828057093e-06,
      "loss": 0.1109,
      "step": 4782
    },
    {
      "epoch": 0.24699199586883552,
      "grad_norm": 5.759164333343506,
      "learning_rate": 8.651589165671982e-06,
      "loss": 0.0568,
      "step": 4783
    },
    {
      "epoch": 0.24704363542473534,
      "grad_norm": 7.30668830871582,
      "learning_rate": 8.651029405226818e-06,
      "loss": 0.064,
      "step": 4784
    },
    {
      "epoch": 0.24709527498063516,
      "grad_norm": 2.254218339920044,
      "learning_rate": 8.650469546736635e-06,
      "loss": 0.0548,
      "step": 4785
    },
    {
      "epoch": 0.24714691453653498,
      "grad_norm": 14.780954360961914,
      "learning_rate": 8.649909590216467e-06,
      "loss": 0.0853,
      "step": 4786
    },
    {
      "epoch": 0.2471985540924348,
      "grad_norm": 18.827924728393555,
      "learning_rate": 8.649349535681348e-06,
      "loss": 0.1298,
      "step": 4787
    },
    {
      "epoch": 0.24725019364833462,
      "grad_norm": 12.483717918395996,
      "learning_rate": 8.648789383146322e-06,
      "loss": 0.2057,
      "step": 4788
    },
    {
      "epoch": 0.24730183320423443,
      "grad_norm": 2.8623428344726562,
      "learning_rate": 8.648229132626429e-06,
      "loss": 0.0385,
      "step": 4789
    },
    {
      "epoch": 0.24735347276013425,
      "grad_norm": 12.157931327819824,
      "learning_rate": 8.647668784136713e-06,
      "loss": 0.1204,
      "step": 4790
    },
    {
      "epoch": 0.24740511231603407,
      "grad_norm": 5.92317533493042,
      "learning_rate": 8.647108337692225e-06,
      "loss": 0.0611,
      "step": 4791
    },
    {
      "epoch": 0.2474567518719339,
      "grad_norm": 2.616621971130371,
      "learning_rate": 8.646547793308012e-06,
      "loss": 0.0247,
      "step": 4792
    },
    {
      "epoch": 0.2475083914278337,
      "grad_norm": 5.77543306350708,
      "learning_rate": 8.64598715099913e-06,
      "loss": 0.1271,
      "step": 4793
    },
    {
      "epoch": 0.24756003098373355,
      "grad_norm": 5.001983642578125,
      "learning_rate": 8.64542641078063e-06,
      "loss": 0.0722,
      "step": 4794
    },
    {
      "epoch": 0.24761167053963337,
      "grad_norm": 4.562959671020508,
      "learning_rate": 8.644865572667575e-06,
      "loss": 0.0555,
      "step": 4795
    },
    {
      "epoch": 0.2476633100955332,
      "grad_norm": 23.0467472076416,
      "learning_rate": 8.644304636675022e-06,
      "loss": 0.2581,
      "step": 4796
    },
    {
      "epoch": 0.247714949651433,
      "grad_norm": 2.5222764015197754,
      "learning_rate": 8.643743602818038e-06,
      "loss": 0.0498,
      "step": 4797
    },
    {
      "epoch": 0.24776658920733283,
      "grad_norm": 3.9587559700012207,
      "learning_rate": 8.643182471111687e-06,
      "loss": 0.0862,
      "step": 4798
    },
    {
      "epoch": 0.24781822876323265,
      "grad_norm": 4.701454162597656,
      "learning_rate": 8.642621241571037e-06,
      "loss": 0.0737,
      "step": 4799
    },
    {
      "epoch": 0.24786986831913246,
      "grad_norm": 5.025585174560547,
      "learning_rate": 8.64205991421116e-06,
      "loss": 0.0439,
      "step": 4800
    },
    {
      "epoch": 0.24792150787503228,
      "grad_norm": 2.559691905975342,
      "learning_rate": 8.641498489047132e-06,
      "loss": 0.0555,
      "step": 4801
    },
    {
      "epoch": 0.2479731474309321,
      "grad_norm": 6.177893161773682,
      "learning_rate": 8.640936966094025e-06,
      "loss": 0.0847,
      "step": 4802
    },
    {
      "epoch": 0.24802478698683192,
      "grad_norm": 7.23362398147583,
      "learning_rate": 8.640375345366921e-06,
      "loss": 0.0835,
      "step": 4803
    },
    {
      "epoch": 0.24807642654273174,
      "grad_norm": 3.9316635131835938,
      "learning_rate": 8.639813626880902e-06,
      "loss": 0.0556,
      "step": 4804
    },
    {
      "epoch": 0.24812806609863156,
      "grad_norm": 12.826696395874023,
      "learning_rate": 8.639251810651054e-06,
      "loss": 0.074,
      "step": 4805
    },
    {
      "epoch": 0.24817970565453137,
      "grad_norm": 4.9510416984558105,
      "learning_rate": 8.638689896692462e-06,
      "loss": 0.1006,
      "step": 4806
    },
    {
      "epoch": 0.2482313452104312,
      "grad_norm": 1.437386393547058,
      "learning_rate": 8.638127885020213e-06,
      "loss": 0.023,
      "step": 4807
    },
    {
      "epoch": 0.248282984766331,
      "grad_norm": 5.579485893249512,
      "learning_rate": 8.637565775649404e-06,
      "loss": 0.0477,
      "step": 4808
    },
    {
      "epoch": 0.24833462432223083,
      "grad_norm": 3.584398031234741,
      "learning_rate": 8.637003568595129e-06,
      "loss": 0.0619,
      "step": 4809
    },
    {
      "epoch": 0.24838626387813065,
      "grad_norm": 3.13767671585083,
      "learning_rate": 8.63644126387248e-06,
      "loss": 0.0532,
      "step": 4810
    },
    {
      "epoch": 0.24843790343403047,
      "grad_norm": 10.098499298095703,
      "learning_rate": 8.635878861496566e-06,
      "loss": 0.1442,
      "step": 4811
    },
    {
      "epoch": 0.24848954298993028,
      "grad_norm": 3.9022769927978516,
      "learning_rate": 8.635316361482487e-06,
      "loss": 0.0595,
      "step": 4812
    },
    {
      "epoch": 0.2485411825458301,
      "grad_norm": 16.210433959960938,
      "learning_rate": 8.634753763845344e-06,
      "loss": 0.1578,
      "step": 4813
    },
    {
      "epoch": 0.24859282210172992,
      "grad_norm": 3.8776185512542725,
      "learning_rate": 8.634191068600248e-06,
      "loss": 0.0275,
      "step": 4814
    },
    {
      "epoch": 0.24864446165762974,
      "grad_norm": 9.41016960144043,
      "learning_rate": 8.633628275762311e-06,
      "loss": 0.0868,
      "step": 4815
    },
    {
      "epoch": 0.24869610121352956,
      "grad_norm": 2.0926029682159424,
      "learning_rate": 8.633065385346645e-06,
      "loss": 0.0166,
      "step": 4816
    },
    {
      "epoch": 0.24874774076942938,
      "grad_norm": 4.3425164222717285,
      "learning_rate": 8.632502397368363e-06,
      "loss": 0.0266,
      "step": 4817
    },
    {
      "epoch": 0.2487993803253292,
      "grad_norm": 4.1209397315979,
      "learning_rate": 8.63193931184259e-06,
      "loss": 0.0685,
      "step": 4818
    },
    {
      "epoch": 0.248851019881229,
      "grad_norm": 4.2912678718566895,
      "learning_rate": 8.63137612878444e-06,
      "loss": 0.0716,
      "step": 4819
    },
    {
      "epoch": 0.24890265943712883,
      "grad_norm": 3.9423985481262207,
      "learning_rate": 8.630812848209042e-06,
      "loss": 0.0338,
      "step": 4820
    },
    {
      "epoch": 0.24895429899302865,
      "grad_norm": 8.877421379089355,
      "learning_rate": 8.630249470131522e-06,
      "loss": 0.067,
      "step": 4821
    },
    {
      "epoch": 0.24900593854892847,
      "grad_norm": 19.841524124145508,
      "learning_rate": 8.629685994567007e-06,
      "loss": 0.1866,
      "step": 4822
    },
    {
      "epoch": 0.2490575781048283,
      "grad_norm": 3.367079257965088,
      "learning_rate": 8.629122421530627e-06,
      "loss": 0.0669,
      "step": 4823
    },
    {
      "epoch": 0.2491092176607281,
      "grad_norm": 4.258636474609375,
      "learning_rate": 8.628558751037522e-06,
      "loss": 0.0427,
      "step": 4824
    },
    {
      "epoch": 0.24916085721662792,
      "grad_norm": 44.628379821777344,
      "learning_rate": 8.627994983102823e-06,
      "loss": 0.1949,
      "step": 4825
    },
    {
      "epoch": 0.24921249677252777,
      "grad_norm": 11.592528343200684,
      "learning_rate": 8.627431117741672e-06,
      "loss": 0.0892,
      "step": 4826
    },
    {
      "epoch": 0.2492641363284276,
      "grad_norm": 5.86536979675293,
      "learning_rate": 8.626867154969211e-06,
      "loss": 0.0728,
      "step": 4827
    },
    {
      "epoch": 0.2493157758843274,
      "grad_norm": 7.149857997894287,
      "learning_rate": 8.626303094800585e-06,
      "loss": 0.0632,
      "step": 4828
    },
    {
      "epoch": 0.24936741544022722,
      "grad_norm": 16.71067237854004,
      "learning_rate": 8.625738937250942e-06,
      "loss": 0.1706,
      "step": 4829
    },
    {
      "epoch": 0.24941905499612704,
      "grad_norm": 3.5469417572021484,
      "learning_rate": 8.62517468233543e-06,
      "loss": 0.0705,
      "step": 4830
    },
    {
      "epoch": 0.24947069455202686,
      "grad_norm": 10.739893913269043,
      "learning_rate": 8.624610330069201e-06,
      "loss": 0.1102,
      "step": 4831
    },
    {
      "epoch": 0.24952233410792668,
      "grad_norm": 4.276482582092285,
      "learning_rate": 8.624045880467413e-06,
      "loss": 0.0627,
      "step": 4832
    },
    {
      "epoch": 0.2495739736638265,
      "grad_norm": 6.86613655090332,
      "learning_rate": 8.623481333545223e-06,
      "loss": 0.0927,
      "step": 4833
    },
    {
      "epoch": 0.24962561321972632,
      "grad_norm": 14.071330070495605,
      "learning_rate": 8.622916689317788e-06,
      "loss": 0.1486,
      "step": 4834
    },
    {
      "epoch": 0.24967725277562614,
      "grad_norm": 5.686269283294678,
      "learning_rate": 8.622351947800276e-06,
      "loss": 0.0696,
      "step": 4835
    },
    {
      "epoch": 0.24972889233152595,
      "grad_norm": 7.421027660369873,
      "learning_rate": 8.621787109007849e-06,
      "loss": 0.1675,
      "step": 4836
    },
    {
      "epoch": 0.24978053188742577,
      "grad_norm": 6.514510154724121,
      "learning_rate": 8.621222172955676e-06,
      "loss": 0.0902,
      "step": 4837
    },
    {
      "epoch": 0.2498321714433256,
      "grad_norm": 9.687719345092773,
      "learning_rate": 8.62065713965893e-06,
      "loss": 0.0438,
      "step": 4838
    },
    {
      "epoch": 0.2498838109992254,
      "grad_norm": 2.126923084259033,
      "learning_rate": 8.620092009132783e-06,
      "loss": 0.0292,
      "step": 4839
    },
    {
      "epoch": 0.24993545055512523,
      "grad_norm": 5.452966213226318,
      "learning_rate": 8.61952678139241e-06,
      "loss": 0.0604,
      "step": 4840
    },
    {
      "epoch": 0.24998709011102505,
      "grad_norm": 9.189090728759766,
      "learning_rate": 8.618961456452992e-06,
      "loss": 0.0637,
      "step": 4841
    },
    {
      "epoch": 0.25003872966692486,
      "grad_norm": 2.8846607208251953,
      "learning_rate": 8.618396034329706e-06,
      "loss": 0.0401,
      "step": 4842
    },
    {
      "epoch": 0.2500903692228247,
      "grad_norm": 5.707699775695801,
      "learning_rate": 8.617830515037741e-06,
      "loss": 0.0827,
      "step": 4843
    },
    {
      "epoch": 0.2501420087787245,
      "grad_norm": 3.6670706272125244,
      "learning_rate": 8.617264898592281e-06,
      "loss": 0.0472,
      "step": 4844
    },
    {
      "epoch": 0.2501936483346243,
      "grad_norm": 4.897774696350098,
      "learning_rate": 8.616699185008515e-06,
      "loss": 0.0904,
      "step": 4845
    },
    {
      "epoch": 0.25024528789052414,
      "grad_norm": 2.71683406829834,
      "learning_rate": 8.616133374301634e-06,
      "loss": 0.0515,
      "step": 4846
    },
    {
      "epoch": 0.25029692744642396,
      "grad_norm": 7.350612640380859,
      "learning_rate": 8.615567466486835e-06,
      "loss": 0.0761,
      "step": 4847
    },
    {
      "epoch": 0.2503485670023238,
      "grad_norm": 3.4015071392059326,
      "learning_rate": 8.615001461579313e-06,
      "loss": 0.0506,
      "step": 4848
    },
    {
      "epoch": 0.2504002065582236,
      "grad_norm": 7.473184108734131,
      "learning_rate": 8.614435359594266e-06,
      "loss": 0.1208,
      "step": 4849
    },
    {
      "epoch": 0.2504518461141234,
      "grad_norm": 4.588456153869629,
      "learning_rate": 8.6138691605469e-06,
      "loss": 0.0697,
      "step": 4850
    },
    {
      "epoch": 0.25050348567002323,
      "grad_norm": 10.284258842468262,
      "learning_rate": 8.613302864452417e-06,
      "loss": 0.0866,
      "step": 4851
    },
    {
      "epoch": 0.25055512522592305,
      "grad_norm": 17.9669189453125,
      "learning_rate": 8.612736471326025e-06,
      "loss": 0.1529,
      "step": 4852
    },
    {
      "epoch": 0.25060676478182287,
      "grad_norm": 2.0508530139923096,
      "learning_rate": 8.612169981182934e-06,
      "loss": 0.0342,
      "step": 4853
    },
    {
      "epoch": 0.2506584043377227,
      "grad_norm": 3.023881673812866,
      "learning_rate": 8.611603394038357e-06,
      "loss": 0.0291,
      "step": 4854
    },
    {
      "epoch": 0.2507100438936225,
      "grad_norm": 8.40319538116455,
      "learning_rate": 8.611036709907508e-06,
      "loss": 0.1032,
      "step": 4855
    },
    {
      "epoch": 0.2507616834495223,
      "grad_norm": 10.627766609191895,
      "learning_rate": 8.610469928805606e-06,
      "loss": 0.0735,
      "step": 4856
    },
    {
      "epoch": 0.25081332300542214,
      "grad_norm": 8.914466857910156,
      "learning_rate": 8.609903050747872e-06,
      "loss": 0.0663,
      "step": 4857
    },
    {
      "epoch": 0.25086496256132196,
      "grad_norm": 17.452972412109375,
      "learning_rate": 8.609336075749527e-06,
      "loss": 0.1012,
      "step": 4858
    },
    {
      "epoch": 0.2509166021172218,
      "grad_norm": 7.436403751373291,
      "learning_rate": 8.608769003825797e-06,
      "loss": 0.0466,
      "step": 4859
    },
    {
      "epoch": 0.2509682416731216,
      "grad_norm": 5.6604766845703125,
      "learning_rate": 8.60820183499191e-06,
      "loss": 0.0981,
      "step": 4860
    },
    {
      "epoch": 0.2510198812290214,
      "grad_norm": 6.0858869552612305,
      "learning_rate": 8.6076345692631e-06,
      "loss": 0.0424,
      "step": 4861
    },
    {
      "epoch": 0.25107152078492123,
      "grad_norm": 8.201815605163574,
      "learning_rate": 8.607067206654596e-06,
      "loss": 0.1058,
      "step": 4862
    },
    {
      "epoch": 0.25112316034082105,
      "grad_norm": 12.160849571228027,
      "learning_rate": 8.606499747181637e-06,
      "loss": 0.12,
      "step": 4863
    },
    {
      "epoch": 0.25117479989672087,
      "grad_norm": 3.9931628704071045,
      "learning_rate": 8.60593219085946e-06,
      "loss": 0.0791,
      "step": 4864
    },
    {
      "epoch": 0.2512264394526207,
      "grad_norm": 2.5860159397125244,
      "learning_rate": 8.605364537703307e-06,
      "loss": 0.0186,
      "step": 4865
    },
    {
      "epoch": 0.2512780790085205,
      "grad_norm": 3.594438076019287,
      "learning_rate": 8.604796787728422e-06,
      "loss": 0.0868,
      "step": 4866
    },
    {
      "epoch": 0.2513297185644203,
      "grad_norm": 10.195670127868652,
      "learning_rate": 8.604228940950051e-06,
      "loss": 0.0972,
      "step": 4867
    },
    {
      "epoch": 0.25138135812032014,
      "grad_norm": 2.8022360801696777,
      "learning_rate": 8.603660997383441e-06,
      "loss": 0.0507,
      "step": 4868
    },
    {
      "epoch": 0.25143299767621996,
      "grad_norm": 8.39094066619873,
      "learning_rate": 8.603092957043845e-06,
      "loss": 0.0706,
      "step": 4869
    },
    {
      "epoch": 0.2514846372321198,
      "grad_norm": 2.35170316696167,
      "learning_rate": 8.602524819946521e-06,
      "loss": 0.0299,
      "step": 4870
    },
    {
      "epoch": 0.2515362767880196,
      "grad_norm": 5.305037021636963,
      "learning_rate": 8.601956586106721e-06,
      "loss": 0.0647,
      "step": 4871
    },
    {
      "epoch": 0.2515879163439194,
      "grad_norm": 5.106627464294434,
      "learning_rate": 8.601388255539705e-06,
      "loss": 0.0733,
      "step": 4872
    },
    {
      "epoch": 0.25163955589981923,
      "grad_norm": 17.91129493713379,
      "learning_rate": 8.600819828260738e-06,
      "loss": 0.102,
      "step": 4873
    },
    {
      "epoch": 0.2516911954557191,
      "grad_norm": 7.9244465827941895,
      "learning_rate": 8.60025130428508e-06,
      "loss": 0.1498,
      "step": 4874
    },
    {
      "epoch": 0.2517428350116189,
      "grad_norm": 4.510180473327637,
      "learning_rate": 8.599682683628002e-06,
      "loss": 0.0557,
      "step": 4875
    },
    {
      "epoch": 0.25179447456751874,
      "grad_norm": 8.201774597167969,
      "learning_rate": 8.599113966304772e-06,
      "loss": 0.0879,
      "step": 4876
    },
    {
      "epoch": 0.25184611412341856,
      "grad_norm": 2.844893455505371,
      "learning_rate": 8.598545152330663e-06,
      "loss": 0.0499,
      "step": 4877
    },
    {
      "epoch": 0.2518977536793184,
      "grad_norm": 16.702083587646484,
      "learning_rate": 8.59797624172095e-06,
      "loss": 0.1299,
      "step": 4878
    },
    {
      "epoch": 0.2519493932352182,
      "grad_norm": 5.293068885803223,
      "learning_rate": 8.597407234490911e-06,
      "loss": 0.0883,
      "step": 4879
    },
    {
      "epoch": 0.252001032791118,
      "grad_norm": 7.661333084106445,
      "learning_rate": 8.596838130655824e-06,
      "loss": 0.0617,
      "step": 4880
    },
    {
      "epoch": 0.25205267234701784,
      "grad_norm": 11.280652046203613,
      "learning_rate": 8.596268930230975e-06,
      "loss": 0.1328,
      "step": 4881
    },
    {
      "epoch": 0.25210431190291765,
      "grad_norm": 2.206164836883545,
      "learning_rate": 8.595699633231647e-06,
      "loss": 0.0417,
      "step": 4882
    },
    {
      "epoch": 0.2521559514588175,
      "grad_norm": 6.845259189605713,
      "learning_rate": 8.595130239673128e-06,
      "loss": 0.0624,
      "step": 4883
    },
    {
      "epoch": 0.2522075910147173,
      "grad_norm": 3.9853248596191406,
      "learning_rate": 8.594560749570708e-06,
      "loss": 0.0484,
      "step": 4884
    },
    {
      "epoch": 0.2522592305706171,
      "grad_norm": 2.305265188217163,
      "learning_rate": 8.593991162939684e-06,
      "loss": 0.0657,
      "step": 4885
    },
    {
      "epoch": 0.25231087012651693,
      "grad_norm": 1.3896057605743408,
      "learning_rate": 8.593421479795347e-06,
      "loss": 0.0262,
      "step": 4886
    },
    {
      "epoch": 0.25236250968241675,
      "grad_norm": 1.9277623891830444,
      "learning_rate": 8.592851700152999e-06,
      "loss": 0.0386,
      "step": 4887
    },
    {
      "epoch": 0.25241414923831657,
      "grad_norm": 4.801656246185303,
      "learning_rate": 8.592281824027938e-06,
      "loss": 0.0381,
      "step": 4888
    },
    {
      "epoch": 0.2524657887942164,
      "grad_norm": 3.0294346809387207,
      "learning_rate": 8.59171185143547e-06,
      "loss": 0.0623,
      "step": 4889
    },
    {
      "epoch": 0.2525174283501162,
      "grad_norm": 14.927995681762695,
      "learning_rate": 8.591141782390896e-06,
      "loss": 0.2375,
      "step": 4890
    },
    {
      "epoch": 0.252569067906016,
      "grad_norm": 26.440750122070312,
      "learning_rate": 8.590571616909533e-06,
      "loss": 0.2469,
      "step": 4891
    },
    {
      "epoch": 0.25262070746191584,
      "grad_norm": 5.018368244171143,
      "learning_rate": 8.590001355006685e-06,
      "loss": 0.0678,
      "step": 4892
    },
    {
      "epoch": 0.25267234701781566,
      "grad_norm": 3.2342796325683594,
      "learning_rate": 8.58943099669767e-06,
      "loss": 0.0622,
      "step": 4893
    },
    {
      "epoch": 0.2527239865737155,
      "grad_norm": 3.89726185798645,
      "learning_rate": 8.588860541997801e-06,
      "loss": 0.0669,
      "step": 4894
    },
    {
      "epoch": 0.2527756261296153,
      "grad_norm": 34.543861389160156,
      "learning_rate": 8.588289990922402e-06,
      "loss": 0.2419,
      "step": 4895
    },
    {
      "epoch": 0.2528272656855151,
      "grad_norm": 4.479204177856445,
      "learning_rate": 8.58771934348679e-06,
      "loss": 0.0942,
      "step": 4896
    },
    {
      "epoch": 0.25287890524141493,
      "grad_norm": 8.36497974395752,
      "learning_rate": 8.58714859970629e-06,
      "loss": 0.1587,
      "step": 4897
    },
    {
      "epoch": 0.25293054479731475,
      "grad_norm": 3.156309127807617,
      "learning_rate": 8.58657775959623e-06,
      "loss": 0.0384,
      "step": 4898
    },
    {
      "epoch": 0.25298218435321457,
      "grad_norm": 4.308324337005615,
      "learning_rate": 8.586006823171941e-06,
      "loss": 0.0953,
      "step": 4899
    },
    {
      "epoch": 0.2530338239091144,
      "grad_norm": 33.510650634765625,
      "learning_rate": 8.585435790448753e-06,
      "loss": 0.1451,
      "step": 4900
    },
    {
      "epoch": 0.2530854634650142,
      "grad_norm": 3.1319053173065186,
      "learning_rate": 8.584864661441999e-06,
      "loss": 0.0424,
      "step": 4901
    },
    {
      "epoch": 0.253137103020914,
      "grad_norm": 6.774385929107666,
      "learning_rate": 8.584293436167017e-06,
      "loss": 0.0864,
      "step": 4902
    },
    {
      "epoch": 0.25318874257681384,
      "grad_norm": 11.143296241760254,
      "learning_rate": 8.58372211463915e-06,
      "loss": 0.0769,
      "step": 4903
    },
    {
      "epoch": 0.25324038213271366,
      "grad_norm": 5.01397705078125,
      "learning_rate": 8.583150696873736e-06,
      "loss": 0.0725,
      "step": 4904
    },
    {
      "epoch": 0.2532920216886135,
      "grad_norm": 5.100067138671875,
      "learning_rate": 8.582579182886122e-06,
      "loss": 0.1201,
      "step": 4905
    },
    {
      "epoch": 0.2533436612445133,
      "grad_norm": 5.032685279846191,
      "learning_rate": 8.582007572691655e-06,
      "loss": 0.0404,
      "step": 4906
    },
    {
      "epoch": 0.2533953008004131,
      "grad_norm": 4.095651149749756,
      "learning_rate": 8.581435866305686e-06,
      "loss": 0.0827,
      "step": 4907
    },
    {
      "epoch": 0.25344694035631293,
      "grad_norm": 6.99715518951416,
      "learning_rate": 8.580864063743565e-06,
      "loss": 0.0435,
      "step": 4908
    },
    {
      "epoch": 0.25349857991221275,
      "grad_norm": 4.761717796325684,
      "learning_rate": 8.580292165020652e-06,
      "loss": 0.0649,
      "step": 4909
    },
    {
      "epoch": 0.25355021946811257,
      "grad_norm": 3.2779765129089355,
      "learning_rate": 8.579720170152299e-06,
      "loss": 0.0526,
      "step": 4910
    },
    {
      "epoch": 0.2536018590240124,
      "grad_norm": 2.3428053855895996,
      "learning_rate": 8.579148079153872e-06,
      "loss": 0.0205,
      "step": 4911
    },
    {
      "epoch": 0.2536534985799122,
      "grad_norm": 5.42629337310791,
      "learning_rate": 8.57857589204073e-06,
      "loss": 0.0462,
      "step": 4912
    },
    {
      "epoch": 0.253705138135812,
      "grad_norm": 4.444467067718506,
      "learning_rate": 8.57800360882824e-06,
      "loss": 0.0994,
      "step": 4913
    },
    {
      "epoch": 0.25375677769171184,
      "grad_norm": 6.293278694152832,
      "learning_rate": 8.57743122953177e-06,
      "loss": 0.0954,
      "step": 4914
    },
    {
      "epoch": 0.25380841724761166,
      "grad_norm": 6.3740620613098145,
      "learning_rate": 8.57685875416669e-06,
      "loss": 0.1179,
      "step": 4915
    },
    {
      "epoch": 0.2538600568035115,
      "grad_norm": 3.9889326095581055,
      "learning_rate": 8.576286182748374e-06,
      "loss": 0.0563,
      "step": 4916
    },
    {
      "epoch": 0.2539116963594113,
      "grad_norm": 11.239313125610352,
      "learning_rate": 8.5757135152922e-06,
      "loss": 0.1142,
      "step": 4917
    },
    {
      "epoch": 0.2539633359153111,
      "grad_norm": 3.6513750553131104,
      "learning_rate": 8.57514075181354e-06,
      "loss": 0.063,
      "step": 4918
    },
    {
      "epoch": 0.25401497547121094,
      "grad_norm": 15.520234107971191,
      "learning_rate": 8.574567892327782e-06,
      "loss": 0.1215,
      "step": 4919
    },
    {
      "epoch": 0.25406661502711075,
      "grad_norm": 3.499572992324829,
      "learning_rate": 8.573994936850307e-06,
      "loss": 0.0533,
      "step": 4920
    },
    {
      "epoch": 0.25411825458301057,
      "grad_norm": 7.456197738647461,
      "learning_rate": 8.573421885396502e-06,
      "loss": 0.1088,
      "step": 4921
    },
    {
      "epoch": 0.2541698941389104,
      "grad_norm": 5.885707855224609,
      "learning_rate": 8.572848737981753e-06,
      "loss": 0.1042,
      "step": 4922
    },
    {
      "epoch": 0.2542215336948102,
      "grad_norm": 50.54495620727539,
      "learning_rate": 8.572275494621453e-06,
      "loss": 0.172,
      "step": 4923
    },
    {
      "epoch": 0.25427317325071,
      "grad_norm": 4.6091628074646,
      "learning_rate": 8.571702155331e-06,
      "loss": 0.0532,
      "step": 4924
    },
    {
      "epoch": 0.25432481280660985,
      "grad_norm": 4.977388858795166,
      "learning_rate": 8.571128720125784e-06,
      "loss": 0.0547,
      "step": 4925
    },
    {
      "epoch": 0.25437645236250966,
      "grad_norm": 5.548500061035156,
      "learning_rate": 8.570555189021205e-06,
      "loss": 0.0603,
      "step": 4926
    },
    {
      "epoch": 0.2544280919184095,
      "grad_norm": 3.8712570667266846,
      "learning_rate": 8.56998156203267e-06,
      "loss": 0.0846,
      "step": 4927
    },
    {
      "epoch": 0.2544797314743093,
      "grad_norm": 3.1265006065368652,
      "learning_rate": 8.569407839175577e-06,
      "loss": 0.0628,
      "step": 4928
    },
    {
      "epoch": 0.2545313710302091,
      "grad_norm": 14.079891204833984,
      "learning_rate": 8.568834020465336e-06,
      "loss": 0.1198,
      "step": 4929
    },
    {
      "epoch": 0.25458301058610894,
      "grad_norm": 5.509148120880127,
      "learning_rate": 8.568260105917356e-06,
      "loss": 0.0761,
      "step": 4930
    },
    {
      "epoch": 0.25463465014200876,
      "grad_norm": 4.600715160369873,
      "learning_rate": 8.567686095547049e-06,
      "loss": 0.1285,
      "step": 4931
    },
    {
      "epoch": 0.2546862896979086,
      "grad_norm": 10.953070640563965,
      "learning_rate": 8.567111989369829e-06,
      "loss": 0.1376,
      "step": 4932
    },
    {
      "epoch": 0.2547379292538084,
      "grad_norm": 9.36451244354248,
      "learning_rate": 8.566537787401112e-06,
      "loss": 0.1374,
      "step": 4933
    },
    {
      "epoch": 0.2547895688097082,
      "grad_norm": 7.448079586029053,
      "learning_rate": 8.56596348965632e-06,
      "loss": 0.0526,
      "step": 4934
    },
    {
      "epoch": 0.25484120836560803,
      "grad_norm": 3.1743452548980713,
      "learning_rate": 8.565389096150876e-06,
      "loss": 0.0566,
      "step": 4935
    },
    {
      "epoch": 0.25489284792150785,
      "grad_norm": 2.1836419105529785,
      "learning_rate": 8.564814606900202e-06,
      "loss": 0.0309,
      "step": 4936
    },
    {
      "epoch": 0.2549444874774077,
      "grad_norm": 4.474606990814209,
      "learning_rate": 8.564240021919725e-06,
      "loss": 0.0824,
      "step": 4937
    },
    {
      "epoch": 0.25499612703330754,
      "grad_norm": 1.5103785991668701,
      "learning_rate": 8.563665341224878e-06,
      "loss": 0.0238,
      "step": 4938
    },
    {
      "epoch": 0.25504776658920736,
      "grad_norm": 4.1246209144592285,
      "learning_rate": 8.563090564831092e-06,
      "loss": 0.058,
      "step": 4939
    },
    {
      "epoch": 0.2550994061451072,
      "grad_norm": 5.9932780265808105,
      "learning_rate": 8.562515692753801e-06,
      "loss": 0.0787,
      "step": 4940
    },
    {
      "epoch": 0.255151045701007,
      "grad_norm": 6.381645202636719,
      "learning_rate": 8.561940725008444e-06,
      "loss": 0.094,
      "step": 4941
    },
    {
      "epoch": 0.2552026852569068,
      "grad_norm": 7.456196308135986,
      "learning_rate": 8.56136566161046e-06,
      "loss": 0.1032,
      "step": 4942
    },
    {
      "epoch": 0.25525432481280663,
      "grad_norm": 2.1694536209106445,
      "learning_rate": 8.560790502575294e-06,
      "loss": 0.0196,
      "step": 4943
    },
    {
      "epoch": 0.25530596436870645,
      "grad_norm": 2.1801211833953857,
      "learning_rate": 8.56021524791839e-06,
      "loss": 0.027,
      "step": 4944
    },
    {
      "epoch": 0.25535760392460627,
      "grad_norm": 8.169354438781738,
      "learning_rate": 8.559639897655196e-06,
      "loss": 0.1144,
      "step": 4945
    },
    {
      "epoch": 0.2554092434805061,
      "grad_norm": 8.387413024902344,
      "learning_rate": 8.559064451801161e-06,
      "loss": 0.0489,
      "step": 4946
    },
    {
      "epoch": 0.2554608830364059,
      "grad_norm": 4.486747741699219,
      "learning_rate": 8.55848891037174e-06,
      "loss": 0.04,
      "step": 4947
    },
    {
      "epoch": 0.2555125225923057,
      "grad_norm": 2.790721893310547,
      "learning_rate": 8.55791327338239e-06,
      "loss": 0.0493,
      "step": 4948
    },
    {
      "epoch": 0.25556416214820554,
      "grad_norm": 10.86604118347168,
      "learning_rate": 8.557337540848565e-06,
      "loss": 0.1702,
      "step": 4949
    },
    {
      "epoch": 0.25561580170410536,
      "grad_norm": 15.707700729370117,
      "learning_rate": 8.556761712785729e-06,
      "loss": 0.1021,
      "step": 4950
    },
    {
      "epoch": 0.2556674412600052,
      "grad_norm": 5.8267130851745605,
      "learning_rate": 8.556185789209345e-06,
      "loss": 0.119,
      "step": 4951
    },
    {
      "epoch": 0.255719080815905,
      "grad_norm": 4.735403060913086,
      "learning_rate": 8.555609770134878e-06,
      "loss": 0.0964,
      "step": 4952
    },
    {
      "epoch": 0.2557707203718048,
      "grad_norm": 5.280795574188232,
      "learning_rate": 8.555033655577796e-06,
      "loss": 0.0839,
      "step": 4953
    },
    {
      "epoch": 0.25582235992770463,
      "grad_norm": 2.627835512161255,
      "learning_rate": 8.554457445553572e-06,
      "loss": 0.0594,
      "step": 4954
    },
    {
      "epoch": 0.25587399948360445,
      "grad_norm": 4.016244888305664,
      "learning_rate": 8.553881140077678e-06,
      "loss": 0.0291,
      "step": 4955
    },
    {
      "epoch": 0.25592563903950427,
      "grad_norm": 3.6305344104766846,
      "learning_rate": 8.55330473916559e-06,
      "loss": 0.0486,
      "step": 4956
    },
    {
      "epoch": 0.2559772785954041,
      "grad_norm": 7.496952056884766,
      "learning_rate": 8.55272824283279e-06,
      "loss": 0.1048,
      "step": 4957
    },
    {
      "epoch": 0.2560289181513039,
      "grad_norm": 9.013606071472168,
      "learning_rate": 8.552151651094752e-06,
      "loss": 0.0972,
      "step": 4958
    },
    {
      "epoch": 0.2560805577072037,
      "grad_norm": 2.704904556274414,
      "learning_rate": 8.551574963966967e-06,
      "loss": 0.0509,
      "step": 4959
    },
    {
      "epoch": 0.25613219726310354,
      "grad_norm": 4.510021209716797,
      "learning_rate": 8.55099818146492e-06,
      "loss": 0.0809,
      "step": 4960
    },
    {
      "epoch": 0.25618383681900336,
      "grad_norm": 4.367525577545166,
      "learning_rate": 8.550421303604096e-06,
      "loss": 0.0912,
      "step": 4961
    },
    {
      "epoch": 0.2562354763749032,
      "grad_norm": 8.296046257019043,
      "learning_rate": 8.54984433039999e-06,
      "loss": 0.1141,
      "step": 4962
    },
    {
      "epoch": 0.256287115930803,
      "grad_norm": 2.3304309844970703,
      "learning_rate": 8.549267261868097e-06,
      "loss": 0.0466,
      "step": 4963
    },
    {
      "epoch": 0.2563387554867028,
      "grad_norm": 12.9039945602417,
      "learning_rate": 8.54869009802391e-06,
      "loss": 0.3358,
      "step": 4964
    },
    {
      "epoch": 0.25639039504260264,
      "grad_norm": 6.404733657836914,
      "learning_rate": 8.548112838882932e-06,
      "loss": 0.0486,
      "step": 4965
    },
    {
      "epoch": 0.25644203459850246,
      "grad_norm": 4.7223310470581055,
      "learning_rate": 8.547535484460662e-06,
      "loss": 0.0557,
      "step": 4966
    },
    {
      "epoch": 0.2564936741544023,
      "grad_norm": 3.1312782764434814,
      "learning_rate": 8.546958034772606e-06,
      "loss": 0.0605,
      "step": 4967
    },
    {
      "epoch": 0.2565453137103021,
      "grad_norm": 8.619488716125488,
      "learning_rate": 8.54638048983427e-06,
      "loss": 0.0481,
      "step": 4968
    },
    {
      "epoch": 0.2565969532662019,
      "grad_norm": 6.6218414306640625,
      "learning_rate": 8.545802849661163e-06,
      "loss": 0.1157,
      "step": 4969
    },
    {
      "epoch": 0.25664859282210173,
      "grad_norm": 6.480082988739014,
      "learning_rate": 8.545225114268798e-06,
      "loss": 0.0606,
      "step": 4970
    },
    {
      "epoch": 0.25670023237800155,
      "grad_norm": 9.922913551330566,
      "learning_rate": 8.544647283672687e-06,
      "loss": 0.1012,
      "step": 4971
    },
    {
      "epoch": 0.25675187193390137,
      "grad_norm": 7.482956409454346,
      "learning_rate": 8.544069357888351e-06,
      "loss": 0.1248,
      "step": 4972
    },
    {
      "epoch": 0.2568035114898012,
      "grad_norm": 9.298571586608887,
      "learning_rate": 8.543491336931308e-06,
      "loss": 0.1964,
      "step": 4973
    },
    {
      "epoch": 0.256855151045701,
      "grad_norm": 6.949913024902344,
      "learning_rate": 8.542913220817081e-06,
      "loss": 0.0739,
      "step": 4974
    },
    {
      "epoch": 0.2569067906016008,
      "grad_norm": 2.077924966812134,
      "learning_rate": 8.542335009561193e-06,
      "loss": 0.0569,
      "step": 4975
    },
    {
      "epoch": 0.25695843015750064,
      "grad_norm": 3.0160152912139893,
      "learning_rate": 8.541756703179173e-06,
      "loss": 0.0683,
      "step": 4976
    },
    {
      "epoch": 0.25701006971340046,
      "grad_norm": 3.3023104667663574,
      "learning_rate": 8.54117830168655e-06,
      "loss": 0.0613,
      "step": 4977
    },
    {
      "epoch": 0.2570617092693003,
      "grad_norm": 2.826622486114502,
      "learning_rate": 8.540599805098855e-06,
      "loss": 0.0465,
      "step": 4978
    },
    {
      "epoch": 0.2571133488252001,
      "grad_norm": 3.860184907913208,
      "learning_rate": 8.540021213431627e-06,
      "loss": 0.105,
      "step": 4979
    },
    {
      "epoch": 0.2571649883810999,
      "grad_norm": 3.358428716659546,
      "learning_rate": 8.539442526700399e-06,
      "loss": 0.059,
      "step": 4980
    },
    {
      "epoch": 0.25721662793699973,
      "grad_norm": 12.830957412719727,
      "learning_rate": 8.538863744920714e-06,
      "loss": 0.1118,
      "step": 4981
    },
    {
      "epoch": 0.25726826749289955,
      "grad_norm": 2.827753782272339,
      "learning_rate": 8.538284868108113e-06,
      "loss": 0.0655,
      "step": 4982
    },
    {
      "epoch": 0.25731990704879937,
      "grad_norm": 4.731797695159912,
      "learning_rate": 8.537705896278143e-06,
      "loss": 0.0451,
      "step": 4983
    },
    {
      "epoch": 0.2573715466046992,
      "grad_norm": 1.9170796871185303,
      "learning_rate": 8.53712682944635e-06,
      "loss": 0.0336,
      "step": 4984
    },
    {
      "epoch": 0.257423186160599,
      "grad_norm": 12.062782287597656,
      "learning_rate": 8.536547667628286e-06,
      "loss": 0.1067,
      "step": 4985
    },
    {
      "epoch": 0.2574748257164988,
      "grad_norm": 6.083703994750977,
      "learning_rate": 8.535968410839504e-06,
      "loss": 0.0985,
      "step": 4986
    },
    {
      "epoch": 0.25752646527239864,
      "grad_norm": 13.270843505859375,
      "learning_rate": 8.535389059095555e-06,
      "loss": 0.2253,
      "step": 4987
    },
    {
      "epoch": 0.25757810482829846,
      "grad_norm": 7.753177642822266,
      "learning_rate": 8.534809612412003e-06,
      "loss": 0.1032,
      "step": 4988
    },
    {
      "epoch": 0.2576297443841983,
      "grad_norm": 2.4098258018493652,
      "learning_rate": 8.534230070804406e-06,
      "loss": 0.06,
      "step": 4989
    },
    {
      "epoch": 0.2576813839400981,
      "grad_norm": 16.559844970703125,
      "learning_rate": 8.533650434288325e-06,
      "loss": 0.1654,
      "step": 4990
    },
    {
      "epoch": 0.2577330234959979,
      "grad_norm": 4.178810119628906,
      "learning_rate": 8.53307070287933e-06,
      "loss": 0.0776,
      "step": 4991
    },
    {
      "epoch": 0.25778466305189773,
      "grad_norm": 2.790566921234131,
      "learning_rate": 8.532490876592984e-06,
      "loss": 0.047,
      "step": 4992
    },
    {
      "epoch": 0.25783630260779755,
      "grad_norm": 11.403155326843262,
      "learning_rate": 8.531910955444861e-06,
      "loss": 0.081,
      "step": 4993
    },
    {
      "epoch": 0.25788794216369737,
      "grad_norm": 5.156162738800049,
      "learning_rate": 8.531330939450535e-06,
      "loss": 0.0404,
      "step": 4994
    },
    {
      "epoch": 0.2579395817195972,
      "grad_norm": 5.486363410949707,
      "learning_rate": 8.530750828625578e-06,
      "loss": 0.0894,
      "step": 4995
    },
    {
      "epoch": 0.257991221275497,
      "grad_norm": 3.992835283279419,
      "learning_rate": 8.530170622985572e-06,
      "loss": 0.0745,
      "step": 4996
    },
    {
      "epoch": 0.2580428608313968,
      "grad_norm": 4.065525531768799,
      "learning_rate": 8.529590322546097e-06,
      "loss": 0.0506,
      "step": 4997
    },
    {
      "epoch": 0.25809450038729664,
      "grad_norm": 2.9051456451416016,
      "learning_rate": 8.529009927322735e-06,
      "loss": 0.0355,
      "step": 4998
    },
    {
      "epoch": 0.25814613994319646,
      "grad_norm": 12.692060470581055,
      "learning_rate": 8.528429437331073e-06,
      "loss": 0.1186,
      "step": 4999
    },
    {
      "epoch": 0.2581977794990963,
      "grad_norm": 6.0951128005981445,
      "learning_rate": 8.527848852586701e-06,
      "loss": 0.1094,
      "step": 5000
    },
    {
      "epoch": 0.25824941905499615,
      "grad_norm": 10.92998218536377,
      "learning_rate": 8.52726817310521e-06,
      "loss": 0.1031,
      "step": 5001
    },
    {
      "epoch": 0.258301058610896,
      "grad_norm": 6.770282745361328,
      "learning_rate": 8.526687398902188e-06,
      "loss": 0.0789,
      "step": 5002
    },
    {
      "epoch": 0.2583526981667958,
      "grad_norm": 4.149222373962402,
      "learning_rate": 8.526106529993237e-06,
      "loss": 0.1137,
      "step": 5003
    },
    {
      "epoch": 0.2584043377226956,
      "grad_norm": 10.165730476379395,
      "learning_rate": 8.525525566393956e-06,
      "loss": 0.0993,
      "step": 5004
    },
    {
      "epoch": 0.25845597727859543,
      "grad_norm": 2.663483142852783,
      "learning_rate": 8.524944508119941e-06,
      "loss": 0.0378,
      "step": 5005
    },
    {
      "epoch": 0.25850761683449525,
      "grad_norm": 3.9462192058563232,
      "learning_rate": 8.524363355186801e-06,
      "loss": 0.0618,
      "step": 5006
    },
    {
      "epoch": 0.25855925639039506,
      "grad_norm": 12.653244018554688,
      "learning_rate": 8.52378210761014e-06,
      "loss": 0.0689,
      "step": 5007
    },
    {
      "epoch": 0.2586108959462949,
      "grad_norm": 4.532001495361328,
      "learning_rate": 8.523200765405568e-06,
      "loss": 0.053,
      "step": 5008
    },
    {
      "epoch": 0.2586625355021947,
      "grad_norm": 4.634668350219727,
      "learning_rate": 8.522619328588696e-06,
      "loss": 0.0752,
      "step": 5009
    },
    {
      "epoch": 0.2587141750580945,
      "grad_norm": 12.610798835754395,
      "learning_rate": 8.522037797175135e-06,
      "loss": 0.1395,
      "step": 5010
    },
    {
      "epoch": 0.25876581461399434,
      "grad_norm": 5.2624382972717285,
      "learning_rate": 8.521456171180506e-06,
      "loss": 0.0994,
      "step": 5011
    },
    {
      "epoch": 0.25881745416989416,
      "grad_norm": 1.8757822513580322,
      "learning_rate": 8.520874450620427e-06,
      "loss": 0.0351,
      "step": 5012
    },
    {
      "epoch": 0.258869093725794,
      "grad_norm": 7.27897834777832,
      "learning_rate": 8.520292635510518e-06,
      "loss": 0.0874,
      "step": 5013
    },
    {
      "epoch": 0.2589207332816938,
      "grad_norm": 4.644496917724609,
      "learning_rate": 8.519710725866404e-06,
      "loss": 0.0489,
      "step": 5014
    },
    {
      "epoch": 0.2589723728375936,
      "grad_norm": 3.185856580734253,
      "learning_rate": 8.51912872170371e-06,
      "loss": 0.0732,
      "step": 5015
    },
    {
      "epoch": 0.25902401239349343,
      "grad_norm": 2.318995237350464,
      "learning_rate": 8.518546623038068e-06,
      "loss": 0.0304,
      "step": 5016
    },
    {
      "epoch": 0.25907565194939325,
      "grad_norm": 4.705417156219482,
      "learning_rate": 8.517964429885108e-06,
      "loss": 0.0856,
      "step": 5017
    },
    {
      "epoch": 0.25912729150529307,
      "grad_norm": 12.818804740905762,
      "learning_rate": 8.517382142260464e-06,
      "loss": 0.1107,
      "step": 5018
    },
    {
      "epoch": 0.2591789310611929,
      "grad_norm": 2.9974541664123535,
      "learning_rate": 8.516799760179775e-06,
      "loss": 0.0706,
      "step": 5019
    },
    {
      "epoch": 0.2592305706170927,
      "grad_norm": 2.7348287105560303,
      "learning_rate": 8.516217283658677e-06,
      "loss": 0.037,
      "step": 5020
    },
    {
      "epoch": 0.2592822101729925,
      "grad_norm": 3.0213823318481445,
      "learning_rate": 8.515634712712815e-06,
      "loss": 0.0774,
      "step": 5021
    },
    {
      "epoch": 0.25933384972889234,
      "grad_norm": 2.1300132274627686,
      "learning_rate": 8.515052047357832e-06,
      "loss": 0.0244,
      "step": 5022
    },
    {
      "epoch": 0.25938548928479216,
      "grad_norm": 2.5505688190460205,
      "learning_rate": 8.514469287609373e-06,
      "loss": 0.0588,
      "step": 5023
    },
    {
      "epoch": 0.259437128840692,
      "grad_norm": 14.31572437286377,
      "learning_rate": 8.513886433483092e-06,
      "loss": 0.0813,
      "step": 5024
    },
    {
      "epoch": 0.2594887683965918,
      "grad_norm": 26.597923278808594,
      "learning_rate": 8.513303484994637e-06,
      "loss": 0.1576,
      "step": 5025
    },
    {
      "epoch": 0.2595404079524916,
      "grad_norm": 6.197727203369141,
      "learning_rate": 8.512720442159664e-06,
      "loss": 0.0907,
      "step": 5026
    },
    {
      "epoch": 0.25959204750839143,
      "grad_norm": 12.145596504211426,
      "learning_rate": 8.51213730499383e-06,
      "loss": 0.1562,
      "step": 5027
    },
    {
      "epoch": 0.25964368706429125,
      "grad_norm": 7.614645957946777,
      "learning_rate": 8.511554073512796e-06,
      "loss": 0.0874,
      "step": 5028
    },
    {
      "epoch": 0.25969532662019107,
      "grad_norm": 8.244816780090332,
      "learning_rate": 8.51097074773222e-06,
      "loss": 0.0893,
      "step": 5029
    },
    {
      "epoch": 0.2597469661760909,
      "grad_norm": 12.172764778137207,
      "learning_rate": 8.510387327667772e-06,
      "loss": 0.1092,
      "step": 5030
    },
    {
      "epoch": 0.2597986057319907,
      "grad_norm": 4.796215534210205,
      "learning_rate": 8.509803813335116e-06,
      "loss": 0.0809,
      "step": 5031
    },
    {
      "epoch": 0.2598502452878905,
      "grad_norm": 4.483304500579834,
      "learning_rate": 8.509220204749925e-06,
      "loss": 0.0449,
      "step": 5032
    },
    {
      "epoch": 0.25990188484379034,
      "grad_norm": 5.70201301574707,
      "learning_rate": 8.508636501927866e-06,
      "loss": 0.0776,
      "step": 5033
    },
    {
      "epoch": 0.25995352439969016,
      "grad_norm": 5.116962432861328,
      "learning_rate": 8.508052704884616e-06,
      "loss": 0.0594,
      "step": 5034
    },
    {
      "epoch": 0.26000516395559,
      "grad_norm": 3.2307958602905273,
      "learning_rate": 8.507468813635853e-06,
      "loss": 0.0669,
      "step": 5035
    },
    {
      "epoch": 0.2600568035114898,
      "grad_norm": 7.407689094543457,
      "learning_rate": 8.506884828197256e-06,
      "loss": 0.0882,
      "step": 5036
    },
    {
      "epoch": 0.2601084430673896,
      "grad_norm": 6.586919784545898,
      "learning_rate": 8.50630074858451e-06,
      "loss": 0.1191,
      "step": 5037
    },
    {
      "epoch": 0.26016008262328943,
      "grad_norm": 4.896877288818359,
      "learning_rate": 8.505716574813298e-06,
      "loss": 0.0759,
      "step": 5038
    },
    {
      "epoch": 0.26021172217918925,
      "grad_norm": 2.0328562259674072,
      "learning_rate": 8.505132306899307e-06,
      "loss": 0.0367,
      "step": 5039
    },
    {
      "epoch": 0.26026336173508907,
      "grad_norm": 6.204102516174316,
      "learning_rate": 8.504547944858227e-06,
      "loss": 0.143,
      "step": 5040
    },
    {
      "epoch": 0.2603150012909889,
      "grad_norm": 5.068661212921143,
      "learning_rate": 8.503963488705753e-06,
      "loss": 0.0872,
      "step": 5041
    },
    {
      "epoch": 0.2603666408468887,
      "grad_norm": 6.262572288513184,
      "learning_rate": 8.503378938457574e-06,
      "loss": 0.0754,
      "step": 5042
    },
    {
      "epoch": 0.2604182804027885,
      "grad_norm": 22.129871368408203,
      "learning_rate": 8.502794294129395e-06,
      "loss": 0.186,
      "step": 5043
    },
    {
      "epoch": 0.26046991995868835,
      "grad_norm": 2.2882401943206787,
      "learning_rate": 8.502209555736912e-06,
      "loss": 0.0541,
      "step": 5044
    },
    {
      "epoch": 0.26052155951458816,
      "grad_norm": 2.590294122695923,
      "learning_rate": 8.501624723295827e-06,
      "loss": 0.0652,
      "step": 5045
    },
    {
      "epoch": 0.260573199070488,
      "grad_norm": 1.8562939167022705,
      "learning_rate": 8.501039796821846e-06,
      "loss": 0.0361,
      "step": 5046
    },
    {
      "epoch": 0.2606248386263878,
      "grad_norm": 2.0372157096862793,
      "learning_rate": 8.500454776330679e-06,
      "loss": 0.0481,
      "step": 5047
    },
    {
      "epoch": 0.2606764781822876,
      "grad_norm": 5.555243015289307,
      "learning_rate": 8.499869661838031e-06,
      "loss": 0.0325,
      "step": 5048
    },
    {
      "epoch": 0.26072811773818744,
      "grad_norm": 6.435550212860107,
      "learning_rate": 8.499284453359623e-06,
      "loss": 0.1001,
      "step": 5049
    },
    {
      "epoch": 0.26077975729408726,
      "grad_norm": 11.964665412902832,
      "learning_rate": 8.49869915091116e-06,
      "loss": 0.0752,
      "step": 5050
    },
    {
      "epoch": 0.2608313968499871,
      "grad_norm": 2.386927604675293,
      "learning_rate": 8.498113754508369e-06,
      "loss": 0.0149,
      "step": 5051
    },
    {
      "epoch": 0.2608830364058869,
      "grad_norm": 3.92686128616333,
      "learning_rate": 8.497528264166963e-06,
      "loss": 0.0378,
      "step": 5052
    },
    {
      "epoch": 0.2609346759617867,
      "grad_norm": 9.72169017791748,
      "learning_rate": 8.496942679902669e-06,
      "loss": 0.1096,
      "step": 5053
    },
    {
      "epoch": 0.26098631551768653,
      "grad_norm": 7.84089994430542,
      "learning_rate": 8.496357001731213e-06,
      "loss": 0.1022,
      "step": 5054
    },
    {
      "epoch": 0.26103795507358635,
      "grad_norm": 3.642744541168213,
      "learning_rate": 8.495771229668318e-06,
      "loss": 0.0545,
      "step": 5055
    },
    {
      "epoch": 0.26108959462948617,
      "grad_norm": 12.628592491149902,
      "learning_rate": 8.495185363729719e-06,
      "loss": 0.1163,
      "step": 5056
    },
    {
      "epoch": 0.261141234185386,
      "grad_norm": 3.732515811920166,
      "learning_rate": 8.494599403931148e-06,
      "loss": 0.0392,
      "step": 5057
    },
    {
      "epoch": 0.2611928737412858,
      "grad_norm": 7.582117557525635,
      "learning_rate": 8.49401335028834e-06,
      "loss": 0.0424,
      "step": 5058
    },
    {
      "epoch": 0.2612445132971856,
      "grad_norm": 1.9089809656143188,
      "learning_rate": 8.493427202817031e-06,
      "loss": 0.0326,
      "step": 5059
    },
    {
      "epoch": 0.26129615285308544,
      "grad_norm": 5.649701118469238,
      "learning_rate": 8.492840961532964e-06,
      "loss": 0.0579,
      "step": 5060
    },
    {
      "epoch": 0.26134779240898526,
      "grad_norm": 4.327955722808838,
      "learning_rate": 8.492254626451882e-06,
      "loss": 0.0481,
      "step": 5061
    },
    {
      "epoch": 0.2613994319648851,
      "grad_norm": 2.562366485595703,
      "learning_rate": 8.491668197589528e-06,
      "loss": 0.0212,
      "step": 5062
    },
    {
      "epoch": 0.2614510715207849,
      "grad_norm": 5.986599922180176,
      "learning_rate": 8.491081674961654e-06,
      "loss": 0.0891,
      "step": 5063
    },
    {
      "epoch": 0.2615027110766847,
      "grad_norm": 10.384500503540039,
      "learning_rate": 8.490495058584005e-06,
      "loss": 0.138,
      "step": 5064
    },
    {
      "epoch": 0.2615543506325846,
      "grad_norm": 4.815372467041016,
      "learning_rate": 8.48990834847234e-06,
      "loss": 0.0611,
      "step": 5065
    },
    {
      "epoch": 0.2616059901884844,
      "grad_norm": 13.010285377502441,
      "learning_rate": 8.48932154464241e-06,
      "loss": 0.1852,
      "step": 5066
    },
    {
      "epoch": 0.2616576297443842,
      "grad_norm": 3.637410879135132,
      "learning_rate": 8.488734647109977e-06,
      "loss": 0.0741,
      "step": 5067
    },
    {
      "epoch": 0.26170926930028404,
      "grad_norm": 2.8369905948638916,
      "learning_rate": 8.488147655890797e-06,
      "loss": 0.0306,
      "step": 5068
    },
    {
      "epoch": 0.26176090885618386,
      "grad_norm": 3.2426352500915527,
      "learning_rate": 8.487560571000636e-06,
      "loss": 0.0404,
      "step": 5069
    },
    {
      "epoch": 0.2618125484120837,
      "grad_norm": 6.136146068572998,
      "learning_rate": 8.48697339245526e-06,
      "loss": 0.1054,
      "step": 5070
    },
    {
      "epoch": 0.2618641879679835,
      "grad_norm": 4.248814105987549,
      "learning_rate": 8.486386120270438e-06,
      "loss": 0.0949,
      "step": 5071
    },
    {
      "epoch": 0.2619158275238833,
      "grad_norm": 4.442965030670166,
      "learning_rate": 8.485798754461935e-06,
      "loss": 0.0606,
      "step": 5072
    },
    {
      "epoch": 0.26196746707978313,
      "grad_norm": 10.128104209899902,
      "learning_rate": 8.485211295045531e-06,
      "loss": 0.1303,
      "step": 5073
    },
    {
      "epoch": 0.26201910663568295,
      "grad_norm": 7.294447898864746,
      "learning_rate": 8.484623742036998e-06,
      "loss": 0.0991,
      "step": 5074
    },
    {
      "epoch": 0.26207074619158277,
      "grad_norm": 2.5690884590148926,
      "learning_rate": 8.484036095452115e-06,
      "loss": 0.0441,
      "step": 5075
    },
    {
      "epoch": 0.2621223857474826,
      "grad_norm": 5.570414066314697,
      "learning_rate": 8.483448355306662e-06,
      "loss": 0.0759,
      "step": 5076
    },
    {
      "epoch": 0.2621740253033824,
      "grad_norm": 6.136719226837158,
      "learning_rate": 8.482860521616423e-06,
      "loss": 0.147,
      "step": 5077
    },
    {
      "epoch": 0.2622256648592822,
      "grad_norm": 5.159173488616943,
      "learning_rate": 8.482272594397185e-06,
      "loss": 0.0728,
      "step": 5078
    },
    {
      "epoch": 0.26227730441518204,
      "grad_norm": 9.781227111816406,
      "learning_rate": 8.481684573664734e-06,
      "loss": 0.132,
      "step": 5079
    },
    {
      "epoch": 0.26232894397108186,
      "grad_norm": 16.312341690063477,
      "learning_rate": 8.481096459434862e-06,
      "loss": 0.1624,
      "step": 5080
    },
    {
      "epoch": 0.2623805835269817,
      "grad_norm": 18.42970085144043,
      "learning_rate": 8.48050825172336e-06,
      "loss": 0.181,
      "step": 5081
    },
    {
      "epoch": 0.2624322230828815,
      "grad_norm": 2.9658124446868896,
      "learning_rate": 8.479919950546028e-06,
      "loss": 0.0803,
      "step": 5082
    },
    {
      "epoch": 0.2624838626387813,
      "grad_norm": 5.961873531341553,
      "learning_rate": 8.479331555918662e-06,
      "loss": 0.07,
      "step": 5083
    },
    {
      "epoch": 0.26253550219468114,
      "grad_norm": 2.666877031326294,
      "learning_rate": 8.478743067857061e-06,
      "loss": 0.0611,
      "step": 5084
    },
    {
      "epoch": 0.26258714175058095,
      "grad_norm": 2.9161150455474854,
      "learning_rate": 8.478154486377032e-06,
      "loss": 0.0497,
      "step": 5085
    },
    {
      "epoch": 0.2626387813064808,
      "grad_norm": 1.700762152671814,
      "learning_rate": 8.477565811494376e-06,
      "loss": 0.0566,
      "step": 5086
    },
    {
      "epoch": 0.2626904208623806,
      "grad_norm": 2.5307202339172363,
      "learning_rate": 8.476977043224907e-06,
      "loss": 0.0389,
      "step": 5087
    },
    {
      "epoch": 0.2627420604182804,
      "grad_norm": 3.405068874359131,
      "learning_rate": 8.476388181584432e-06,
      "loss": 0.061,
      "step": 5088
    },
    {
      "epoch": 0.26279369997418023,
      "grad_norm": 3.4054055213928223,
      "learning_rate": 8.475799226588766e-06,
      "loss": 0.0502,
      "step": 5089
    },
    {
      "epoch": 0.26284533953008005,
      "grad_norm": 5.947010517120361,
      "learning_rate": 8.475210178253724e-06,
      "loss": 0.093,
      "step": 5090
    },
    {
      "epoch": 0.26289697908597986,
      "grad_norm": 13.127108573913574,
      "learning_rate": 8.474621036595123e-06,
      "loss": 0.2163,
      "step": 5091
    },
    {
      "epoch": 0.2629486186418797,
      "grad_norm": 4.647229194641113,
      "learning_rate": 8.474031801628787e-06,
      "loss": 0.0756,
      "step": 5092
    },
    {
      "epoch": 0.2630002581977795,
      "grad_norm": 3.202439308166504,
      "learning_rate": 8.473442473370537e-06,
      "loss": 0.0708,
      "step": 5093
    },
    {
      "epoch": 0.2630518977536793,
      "grad_norm": 14.714278221130371,
      "learning_rate": 8.4728530518362e-06,
      "loss": 0.1737,
      "step": 5094
    },
    {
      "epoch": 0.26310353730957914,
      "grad_norm": 8.454996109008789,
      "learning_rate": 8.472263537041603e-06,
      "loss": 0.1056,
      "step": 5095
    },
    {
      "epoch": 0.26315517686547896,
      "grad_norm": 6.830498218536377,
      "learning_rate": 8.47167392900258e-06,
      "loss": 0.0423,
      "step": 5096
    },
    {
      "epoch": 0.2632068164213788,
      "grad_norm": 5.768155097961426,
      "learning_rate": 8.471084227734961e-06,
      "loss": 0.0744,
      "step": 5097
    },
    {
      "epoch": 0.2632584559772786,
      "grad_norm": 7.1830925941467285,
      "learning_rate": 8.470494433254584e-06,
      "loss": 0.0971,
      "step": 5098
    },
    {
      "epoch": 0.2633100955331784,
      "grad_norm": 7.517816066741943,
      "learning_rate": 8.469904545577285e-06,
      "loss": 0.0817,
      "step": 5099
    },
    {
      "epoch": 0.26336173508907823,
      "grad_norm": 5.644775390625,
      "learning_rate": 8.469314564718908e-06,
      "loss": 0.0698,
      "step": 5100
    },
    {
      "epoch": 0.26341337464497805,
      "grad_norm": 21.758424758911133,
      "learning_rate": 8.468724490695293e-06,
      "loss": 0.1283,
      "step": 5101
    },
    {
      "epoch": 0.26346501420087787,
      "grad_norm": 27.568527221679688,
      "learning_rate": 8.468134323522289e-06,
      "loss": 0.0455,
      "step": 5102
    },
    {
      "epoch": 0.2635166537567777,
      "grad_norm": 4.801733493804932,
      "learning_rate": 8.467544063215741e-06,
      "loss": 0.088,
      "step": 5103
    },
    {
      "epoch": 0.2635682933126775,
      "grad_norm": 3.684379816055298,
      "learning_rate": 8.466953709791504e-06,
      "loss": 0.0479,
      "step": 5104
    },
    {
      "epoch": 0.2636199328685773,
      "grad_norm": 7.454929828643799,
      "learning_rate": 8.466363263265429e-06,
      "loss": 0.0758,
      "step": 5105
    },
    {
      "epoch": 0.26367157242447714,
      "grad_norm": 3.8421270847320557,
      "learning_rate": 8.465772723653372e-06,
      "loss": 0.0803,
      "step": 5106
    },
    {
      "epoch": 0.26372321198037696,
      "grad_norm": 3.1991899013519287,
      "learning_rate": 8.465182090971193e-06,
      "loss": 0.0551,
      "step": 5107
    },
    {
      "epoch": 0.2637748515362768,
      "grad_norm": 9.128251075744629,
      "learning_rate": 8.46459136523475e-06,
      "loss": 0.1025,
      "step": 5108
    },
    {
      "epoch": 0.2638264910921766,
      "grad_norm": 4.715816974639893,
      "learning_rate": 8.464000546459908e-06,
      "loss": 0.0789,
      "step": 5109
    },
    {
      "epoch": 0.2638781306480764,
      "grad_norm": 2.8164920806884766,
      "learning_rate": 8.463409634662533e-06,
      "loss": 0.0452,
      "step": 5110
    },
    {
      "epoch": 0.26392977020397623,
      "grad_norm": 3.439389944076538,
      "learning_rate": 8.462818629858494e-06,
      "loss": 0.0497,
      "step": 5111
    },
    {
      "epoch": 0.26398140975987605,
      "grad_norm": 10.57083511352539,
      "learning_rate": 8.46222753206366e-06,
      "loss": 0.0929,
      "step": 5112
    },
    {
      "epoch": 0.26403304931577587,
      "grad_norm": 13.943062782287598,
      "learning_rate": 8.461636341293907e-06,
      "loss": 0.1367,
      "step": 5113
    },
    {
      "epoch": 0.2640846888716757,
      "grad_norm": 1.6487072706222534,
      "learning_rate": 8.461045057565108e-06,
      "loss": 0.0206,
      "step": 5114
    },
    {
      "epoch": 0.2641363284275755,
      "grad_norm": 4.340873718261719,
      "learning_rate": 8.460453680893143e-06,
      "loss": 0.0938,
      "step": 5115
    },
    {
      "epoch": 0.2641879679834753,
      "grad_norm": 4.771779537200928,
      "learning_rate": 8.459862211293895e-06,
      "loss": 0.0652,
      "step": 5116
    },
    {
      "epoch": 0.26423960753937514,
      "grad_norm": 3.0112950801849365,
      "learning_rate": 8.459270648783241e-06,
      "loss": 0.0589,
      "step": 5117
    },
    {
      "epoch": 0.26429124709527496,
      "grad_norm": 5.439959526062012,
      "learning_rate": 8.458678993377073e-06,
      "loss": 0.0717,
      "step": 5118
    },
    {
      "epoch": 0.2643428866511748,
      "grad_norm": 7.369544506072998,
      "learning_rate": 8.458087245091278e-06,
      "loss": 0.086,
      "step": 5119
    },
    {
      "epoch": 0.2643945262070746,
      "grad_norm": 6.852931976318359,
      "learning_rate": 8.457495403941745e-06,
      "loss": 0.0968,
      "step": 5120
    },
    {
      "epoch": 0.2644461657629744,
      "grad_norm": 4.595730304718018,
      "learning_rate": 8.456903469944367e-06,
      "loss": 0.0475,
      "step": 5121
    },
    {
      "epoch": 0.26449780531887424,
      "grad_norm": 7.68597936630249,
      "learning_rate": 8.456311443115043e-06,
      "loss": 0.0915,
      "step": 5122
    },
    {
      "epoch": 0.26454944487477405,
      "grad_norm": 8.308457374572754,
      "learning_rate": 8.455719323469668e-06,
      "loss": 0.1253,
      "step": 5123
    },
    {
      "epoch": 0.26460108443067387,
      "grad_norm": 6.661538600921631,
      "learning_rate": 8.455127111024148e-06,
      "loss": 0.0711,
      "step": 5124
    },
    {
      "epoch": 0.2646527239865737,
      "grad_norm": 2.2525217533111572,
      "learning_rate": 8.45453480579438e-06,
      "loss": 0.0478,
      "step": 5125
    },
    {
      "epoch": 0.2647043635424735,
      "grad_norm": 10.036046028137207,
      "learning_rate": 8.453942407796274e-06,
      "loss": 0.0733,
      "step": 5126
    },
    {
      "epoch": 0.2647560030983733,
      "grad_norm": 2.814260721206665,
      "learning_rate": 8.453349917045736e-06,
      "loss": 0.0621,
      "step": 5127
    },
    {
      "epoch": 0.2648076426542732,
      "grad_norm": 2.387774705886841,
      "learning_rate": 8.452757333558677e-06,
      "loss": 0.0418,
      "step": 5128
    },
    {
      "epoch": 0.264859282210173,
      "grad_norm": 8.715319633483887,
      "learning_rate": 8.452164657351011e-06,
      "loss": 0.0861,
      "step": 5129
    },
    {
      "epoch": 0.26491092176607284,
      "grad_norm": 9.031669616699219,
      "learning_rate": 8.451571888438655e-06,
      "loss": 0.0853,
      "step": 5130
    },
    {
      "epoch": 0.26496256132197266,
      "grad_norm": 2.991791248321533,
      "learning_rate": 8.450979026837524e-06,
      "loss": 0.0727,
      "step": 5131
    },
    {
      "epoch": 0.2650142008778725,
      "grad_norm": 9.088384628295898,
      "learning_rate": 8.450386072563544e-06,
      "loss": 0.084,
      "step": 5132
    },
    {
      "epoch": 0.2650658404337723,
      "grad_norm": 2.476287841796875,
      "learning_rate": 8.449793025632634e-06,
      "loss": 0.0606,
      "step": 5133
    },
    {
      "epoch": 0.2651174799896721,
      "grad_norm": 3.7258737087249756,
      "learning_rate": 8.449199886060718e-06,
      "loss": 0.055,
      "step": 5134
    },
    {
      "epoch": 0.26516911954557193,
      "grad_norm": 6.7602033615112305,
      "learning_rate": 8.44860665386373e-06,
      "loss": 0.0526,
      "step": 5135
    },
    {
      "epoch": 0.26522075910147175,
      "grad_norm": 10.227161407470703,
      "learning_rate": 8.448013329057594e-06,
      "loss": 0.0608,
      "step": 5136
    },
    {
      "epoch": 0.26527239865737157,
      "grad_norm": 15.059618949890137,
      "learning_rate": 8.44741991165825e-06,
      "loss": 0.1266,
      "step": 5137
    },
    {
      "epoch": 0.2653240382132714,
      "grad_norm": 4.858183860778809,
      "learning_rate": 8.446826401681629e-06,
      "loss": 0.0578,
      "step": 5138
    },
    {
      "epoch": 0.2653756777691712,
      "grad_norm": 7.352145195007324,
      "learning_rate": 8.446232799143671e-06,
      "loss": 0.0695,
      "step": 5139
    },
    {
      "epoch": 0.265427317325071,
      "grad_norm": 5.329650402069092,
      "learning_rate": 8.445639104060318e-06,
      "loss": 0.0981,
      "step": 5140
    },
    {
      "epoch": 0.26547895688097084,
      "grad_norm": 5.181488990783691,
      "learning_rate": 8.445045316447507e-06,
      "loss": 0.0648,
      "step": 5141
    },
    {
      "epoch": 0.26553059643687066,
      "grad_norm": 2.867332696914673,
      "learning_rate": 8.444451436321192e-06,
      "loss": 0.0463,
      "step": 5142
    },
    {
      "epoch": 0.2655822359927705,
      "grad_norm": 2.9588916301727295,
      "learning_rate": 8.443857463697314e-06,
      "loss": 0.0524,
      "step": 5143
    },
    {
      "epoch": 0.2656338755486703,
      "grad_norm": 8.741055488586426,
      "learning_rate": 8.44326339859183e-06,
      "loss": 0.1123,
      "step": 5144
    },
    {
      "epoch": 0.2656855151045701,
      "grad_norm": 3.285071849822998,
      "learning_rate": 8.442669241020687e-06,
      "loss": 0.0837,
      "step": 5145
    },
    {
      "epoch": 0.26573715466046993,
      "grad_norm": 3.6790106296539307,
      "learning_rate": 8.442074990999846e-06,
      "loss": 0.0484,
      "step": 5146
    },
    {
      "epoch": 0.26578879421636975,
      "grad_norm": 6.579230308532715,
      "learning_rate": 8.441480648545261e-06,
      "loss": 0.0694,
      "step": 5147
    },
    {
      "epoch": 0.26584043377226957,
      "grad_norm": 6.031303882598877,
      "learning_rate": 8.440886213672895e-06,
      "loss": 0.0457,
      "step": 5148
    },
    {
      "epoch": 0.2658920733281694,
      "grad_norm": 2.5559029579162598,
      "learning_rate": 8.440291686398707e-06,
      "loss": 0.0427,
      "step": 5149
    },
    {
      "epoch": 0.2659437128840692,
      "grad_norm": 3.2429728507995605,
      "learning_rate": 8.439697066738668e-06,
      "loss": 0.0533,
      "step": 5150
    },
    {
      "epoch": 0.265995352439969,
      "grad_norm": 12.281577110290527,
      "learning_rate": 8.439102354708742e-06,
      "loss": 0.0867,
      "step": 5151
    },
    {
      "epoch": 0.26604699199586884,
      "grad_norm": 14.346183776855469,
      "learning_rate": 8.438507550324902e-06,
      "loss": 0.1134,
      "step": 5152
    },
    {
      "epoch": 0.26609863155176866,
      "grad_norm": 5.814205169677734,
      "learning_rate": 8.437912653603118e-06,
      "loss": 0.0436,
      "step": 5153
    },
    {
      "epoch": 0.2661502711076685,
      "grad_norm": 4.896270275115967,
      "learning_rate": 8.437317664559368e-06,
      "loss": 0.052,
      "step": 5154
    },
    {
      "epoch": 0.2662019106635683,
      "grad_norm": 7.871638774871826,
      "learning_rate": 8.43672258320963e-06,
      "loss": 0.035,
      "step": 5155
    },
    {
      "epoch": 0.2662535502194681,
      "grad_norm": 3.323550224304199,
      "learning_rate": 8.43612740956988e-06,
      "loss": 0.0462,
      "step": 5156
    },
    {
      "epoch": 0.26630518977536793,
      "grad_norm": 3.371290683746338,
      "learning_rate": 8.435532143656109e-06,
      "loss": 0.0369,
      "step": 5157
    },
    {
      "epoch": 0.26635682933126775,
      "grad_norm": 5.767765522003174,
      "learning_rate": 8.434936785484296e-06,
      "loss": 0.0676,
      "step": 5158
    },
    {
      "epoch": 0.26640846888716757,
      "grad_norm": 3.7330806255340576,
      "learning_rate": 8.434341335070429e-06,
      "loss": 0.095,
      "step": 5159
    },
    {
      "epoch": 0.2664601084430674,
      "grad_norm": 8.097787857055664,
      "learning_rate": 8.433745792430498e-06,
      "loss": 0.1544,
      "step": 5160
    },
    {
      "epoch": 0.2665117479989672,
      "grad_norm": 3.36576247215271,
      "learning_rate": 8.433150157580499e-06,
      "loss": 0.0546,
      "step": 5161
    },
    {
      "epoch": 0.266563387554867,
      "grad_norm": 4.040228366851807,
      "learning_rate": 8.432554430536427e-06,
      "loss": 0.0591,
      "step": 5162
    },
    {
      "epoch": 0.26661502711076684,
      "grad_norm": 1.7852603197097778,
      "learning_rate": 8.431958611314278e-06,
      "loss": 0.0351,
      "step": 5163
    },
    {
      "epoch": 0.26666666666666666,
      "grad_norm": 3.9390573501586914,
      "learning_rate": 8.43136269993005e-06,
      "loss": 0.0771,
      "step": 5164
    },
    {
      "epoch": 0.2667183062225665,
      "grad_norm": 8.77720832824707,
      "learning_rate": 8.43076669639975e-06,
      "loss": 0.2037,
      "step": 5165
    },
    {
      "epoch": 0.2667699457784663,
      "grad_norm": 5.265783309936523,
      "learning_rate": 8.43017060073938e-06,
      "loss": 0.0721,
      "step": 5166
    },
    {
      "epoch": 0.2668215853343661,
      "grad_norm": 2.6988401412963867,
      "learning_rate": 8.429574412964949e-06,
      "loss": 0.0302,
      "step": 5167
    },
    {
      "epoch": 0.26687322489026594,
      "grad_norm": 7.685461044311523,
      "learning_rate": 8.428978133092467e-06,
      "loss": 0.1197,
      "step": 5168
    },
    {
      "epoch": 0.26692486444616575,
      "grad_norm": 13.052532196044922,
      "learning_rate": 8.428381761137947e-06,
      "loss": 0.119,
      "step": 5169
    },
    {
      "epoch": 0.2669765040020656,
      "grad_norm": 4.743468761444092,
      "learning_rate": 8.427785297117403e-06,
      "loss": 0.0966,
      "step": 5170
    },
    {
      "epoch": 0.2670281435579654,
      "grad_norm": 7.750429630279541,
      "learning_rate": 8.427188741046852e-06,
      "loss": 0.0869,
      "step": 5171
    },
    {
      "epoch": 0.2670797831138652,
      "grad_norm": 6.8650803565979,
      "learning_rate": 8.426592092942316e-06,
      "loss": 0.1104,
      "step": 5172
    },
    {
      "epoch": 0.26713142266976503,
      "grad_norm": 3.9241433143615723,
      "learning_rate": 8.425995352819816e-06,
      "loss": 0.0771,
      "step": 5173
    },
    {
      "epoch": 0.26718306222566485,
      "grad_norm": 4.123037338256836,
      "learning_rate": 8.425398520695376e-06,
      "loss": 0.054,
      "step": 5174
    },
    {
      "epoch": 0.26723470178156467,
      "grad_norm": 5.307165145874023,
      "learning_rate": 8.424801596585028e-06,
      "loss": 0.098,
      "step": 5175
    },
    {
      "epoch": 0.2672863413374645,
      "grad_norm": 5.112392902374268,
      "learning_rate": 8.424204580504795e-06,
      "loss": 0.0598,
      "step": 5176
    },
    {
      "epoch": 0.2673379808933643,
      "grad_norm": 3.318742513656616,
      "learning_rate": 8.423607472470713e-06,
      "loss": 0.0504,
      "step": 5177
    },
    {
      "epoch": 0.2673896204492641,
      "grad_norm": 6.784242153167725,
      "learning_rate": 8.423010272498818e-06,
      "loss": 0.1112,
      "step": 5178
    },
    {
      "epoch": 0.26744126000516394,
      "grad_norm": 6.075645923614502,
      "learning_rate": 8.422412980605146e-06,
      "loss": 0.0645,
      "step": 5179
    },
    {
      "epoch": 0.26749289956106376,
      "grad_norm": 4.731207847595215,
      "learning_rate": 8.421815596805736e-06,
      "loss": 0.0673,
      "step": 5180
    },
    {
      "epoch": 0.2675445391169636,
      "grad_norm": 3.7043824195861816,
      "learning_rate": 8.42121812111663e-06,
      "loss": 0.0563,
      "step": 5181
    },
    {
      "epoch": 0.2675961786728634,
      "grad_norm": 7.105748653411865,
      "learning_rate": 8.420620553553873e-06,
      "loss": 0.0758,
      "step": 5182
    },
    {
      "epoch": 0.2676478182287632,
      "grad_norm": 2.1034014225006104,
      "learning_rate": 8.420022894133513e-06,
      "loss": 0.0352,
      "step": 5183
    },
    {
      "epoch": 0.26769945778466303,
      "grad_norm": 3.1852574348449707,
      "learning_rate": 8.4194251428716e-06,
      "loss": 0.0412,
      "step": 5184
    },
    {
      "epoch": 0.26775109734056285,
      "grad_norm": 25.91417121887207,
      "learning_rate": 8.418827299784183e-06,
      "loss": 0.099,
      "step": 5185
    },
    {
      "epoch": 0.26780273689646267,
      "grad_norm": 5.176849842071533,
      "learning_rate": 8.41822936488732e-06,
      "loss": 0.0856,
      "step": 5186
    },
    {
      "epoch": 0.2678543764523625,
      "grad_norm": 6.4081807136535645,
      "learning_rate": 8.417631338197068e-06,
      "loss": 0.1161,
      "step": 5187
    },
    {
      "epoch": 0.2679060160082623,
      "grad_norm": 5.279532432556152,
      "learning_rate": 8.417033219729482e-06,
      "loss": 0.0811,
      "step": 5188
    },
    {
      "epoch": 0.2679576555641621,
      "grad_norm": 3.071892499923706,
      "learning_rate": 8.416435009500628e-06,
      "loss": 0.0564,
      "step": 5189
    },
    {
      "epoch": 0.26800929512006194,
      "grad_norm": 5.173732280731201,
      "learning_rate": 8.415836707526572e-06,
      "loss": 0.0636,
      "step": 5190
    },
    {
      "epoch": 0.26806093467596176,
      "grad_norm": 2.6634905338287354,
      "learning_rate": 8.415238313823375e-06,
      "loss": 0.0473,
      "step": 5191
    },
    {
      "epoch": 0.26811257423186163,
      "grad_norm": 11.125566482543945,
      "learning_rate": 8.414639828407112e-06,
      "loss": 0.0464,
      "step": 5192
    },
    {
      "epoch": 0.26816421378776145,
      "grad_norm": 4.3530707359313965,
      "learning_rate": 8.41404125129385e-06,
      "loss": 0.0918,
      "step": 5193
    },
    {
      "epoch": 0.26821585334366127,
      "grad_norm": 2.6038174629211426,
      "learning_rate": 8.413442582499667e-06,
      "loss": 0.0612,
      "step": 5194
    },
    {
      "epoch": 0.2682674928995611,
      "grad_norm": 9.382564544677734,
      "learning_rate": 8.412843822040636e-06,
      "loss": 0.1582,
      "step": 5195
    },
    {
      "epoch": 0.2683191324554609,
      "grad_norm": 3.4344112873077393,
      "learning_rate": 8.412244969932841e-06,
      "loss": 0.0629,
      "step": 5196
    },
    {
      "epoch": 0.2683707720113607,
      "grad_norm": 2.4453811645507812,
      "learning_rate": 8.411646026192361e-06,
      "loss": 0.0395,
      "step": 5197
    },
    {
      "epoch": 0.26842241156726054,
      "grad_norm": 2.934434175491333,
      "learning_rate": 8.41104699083528e-06,
      "loss": 0.0512,
      "step": 5198
    },
    {
      "epoch": 0.26847405112316036,
      "grad_norm": 2.5582149028778076,
      "learning_rate": 8.410447863877683e-06,
      "loss": 0.0157,
      "step": 5199
    },
    {
      "epoch": 0.2685256906790602,
      "grad_norm": 13.390679359436035,
      "learning_rate": 8.409848645335661e-06,
      "loss": 0.2012,
      "step": 5200
    },
    {
      "epoch": 0.26857733023496,
      "grad_norm": 9.531688690185547,
      "learning_rate": 8.409249335225306e-06,
      "loss": 0.0804,
      "step": 5201
    },
    {
      "epoch": 0.2686289697908598,
      "grad_norm": 4.582295894622803,
      "learning_rate": 8.408649933562711e-06,
      "loss": 0.0583,
      "step": 5202
    },
    {
      "epoch": 0.26868060934675964,
      "grad_norm": 4.5309224128723145,
      "learning_rate": 8.408050440363972e-06,
      "loss": 0.0604,
      "step": 5203
    },
    {
      "epoch": 0.26873224890265945,
      "grad_norm": 9.055119514465332,
      "learning_rate": 8.407450855645189e-06,
      "loss": 0.0929,
      "step": 5204
    },
    {
      "epoch": 0.2687838884585593,
      "grad_norm": 4.159726619720459,
      "learning_rate": 8.406851179422461e-06,
      "loss": 0.0715,
      "step": 5205
    },
    {
      "epoch": 0.2688355280144591,
      "grad_norm": 2.395515203475952,
      "learning_rate": 8.406251411711894e-06,
      "loss": 0.0509,
      "step": 5206
    },
    {
      "epoch": 0.2688871675703589,
      "grad_norm": 12.237931251525879,
      "learning_rate": 8.405651552529594e-06,
      "loss": 0.1246,
      "step": 5207
    },
    {
      "epoch": 0.2689388071262587,
      "grad_norm": 8.289705276489258,
      "learning_rate": 8.40505160189167e-06,
      "loss": 0.1435,
      "step": 5208
    },
    {
      "epoch": 0.26899044668215855,
      "grad_norm": 5.521714687347412,
      "learning_rate": 8.404451559814229e-06,
      "loss": 0.0782,
      "step": 5209
    },
    {
      "epoch": 0.26904208623805836,
      "grad_norm": 2.172443151473999,
      "learning_rate": 8.40385142631339e-06,
      "loss": 0.0294,
      "step": 5210
    },
    {
      "epoch": 0.2690937257939582,
      "grad_norm": 3.432382583618164,
      "learning_rate": 8.403251201405267e-06,
      "loss": 0.0499,
      "step": 5211
    },
    {
      "epoch": 0.269145365349858,
      "grad_norm": 4.712459564208984,
      "learning_rate": 8.402650885105979e-06,
      "loss": 0.0797,
      "step": 5212
    },
    {
      "epoch": 0.2691970049057578,
      "grad_norm": 8.634231567382812,
      "learning_rate": 8.402050477431644e-06,
      "loss": 0.118,
      "step": 5213
    },
    {
      "epoch": 0.26924864446165764,
      "grad_norm": 2.801955461502075,
      "learning_rate": 8.40144997839839e-06,
      "loss": 0.0648,
      "step": 5214
    },
    {
      "epoch": 0.26930028401755746,
      "grad_norm": 1.926292896270752,
      "learning_rate": 8.400849388022339e-06,
      "loss": 0.0302,
      "step": 5215
    },
    {
      "epoch": 0.2693519235734573,
      "grad_norm": 3.2445244789123535,
      "learning_rate": 8.400248706319622e-06,
      "loss": 0.0526,
      "step": 5216
    },
    {
      "epoch": 0.2694035631293571,
      "grad_norm": 3.7423059940338135,
      "learning_rate": 8.399647933306367e-06,
      "loss": 0.0715,
      "step": 5217
    },
    {
      "epoch": 0.2694552026852569,
      "grad_norm": 6.542614459991455,
      "learning_rate": 8.399047068998712e-06,
      "loss": 0.0922,
      "step": 5218
    },
    {
      "epoch": 0.26950684224115673,
      "grad_norm": 3.5540342330932617,
      "learning_rate": 8.398446113412785e-06,
      "loss": 0.0592,
      "step": 5219
    },
    {
      "epoch": 0.26955848179705655,
      "grad_norm": 13.446687698364258,
      "learning_rate": 8.397845066564732e-06,
      "loss": 0.2235,
      "step": 5220
    },
    {
      "epoch": 0.26961012135295637,
      "grad_norm": 2.4153621196746826,
      "learning_rate": 8.397243928470689e-06,
      "loss": 0.0418,
      "step": 5221
    },
    {
      "epoch": 0.2696617609088562,
      "grad_norm": 2.66829252243042,
      "learning_rate": 8.3966426991468e-06,
      "loss": 0.0628,
      "step": 5222
    },
    {
      "epoch": 0.269713400464756,
      "grad_norm": 9.077445030212402,
      "learning_rate": 8.39604137860921e-06,
      "loss": 0.0657,
      "step": 5223
    },
    {
      "epoch": 0.2697650400206558,
      "grad_norm": 9.977373123168945,
      "learning_rate": 8.395439966874068e-06,
      "loss": 0.1509,
      "step": 5224
    },
    {
      "epoch": 0.26981667957655564,
      "grad_norm": 4.083282947540283,
      "learning_rate": 8.394838463957525e-06,
      "loss": 0.0709,
      "step": 5225
    },
    {
      "epoch": 0.26986831913245546,
      "grad_norm": 4.702123165130615,
      "learning_rate": 8.394236869875732e-06,
      "loss": 0.0942,
      "step": 5226
    },
    {
      "epoch": 0.2699199586883553,
      "grad_norm": 5.938112735748291,
      "learning_rate": 8.393635184644844e-06,
      "loss": 0.0475,
      "step": 5227
    },
    {
      "epoch": 0.2699715982442551,
      "grad_norm": 13.854580879211426,
      "learning_rate": 8.39303340828102e-06,
      "loss": 0.1619,
      "step": 5228
    },
    {
      "epoch": 0.2700232378001549,
      "grad_norm": 8.313126564025879,
      "learning_rate": 8.392431540800419e-06,
      "loss": 0.1417,
      "step": 5229
    },
    {
      "epoch": 0.27007487735605473,
      "grad_norm": 4.122748374938965,
      "learning_rate": 8.391829582219207e-06,
      "loss": 0.0803,
      "step": 5230
    },
    {
      "epoch": 0.27012651691195455,
      "grad_norm": 4.654393672943115,
      "learning_rate": 8.391227532553543e-06,
      "loss": 0.0624,
      "step": 5231
    },
    {
      "epoch": 0.27017815646785437,
      "grad_norm": 5.711096286773682,
      "learning_rate": 8.3906253918196e-06,
      "loss": 0.0663,
      "step": 5232
    },
    {
      "epoch": 0.2702297960237542,
      "grad_norm": 3.349104642868042,
      "learning_rate": 8.390023160033545e-06,
      "loss": 0.0481,
      "step": 5233
    },
    {
      "epoch": 0.270281435579654,
      "grad_norm": 3.9131669998168945,
      "learning_rate": 8.389420837211552e-06,
      "loss": 0.0859,
      "step": 5234
    },
    {
      "epoch": 0.2703330751355538,
      "grad_norm": 5.7426838874816895,
      "learning_rate": 8.388818423369794e-06,
      "loss": 0.1124,
      "step": 5235
    },
    {
      "epoch": 0.27038471469145364,
      "grad_norm": 2.168358087539673,
      "learning_rate": 8.388215918524449e-06,
      "loss": 0.0372,
      "step": 5236
    },
    {
      "epoch": 0.27043635424735346,
      "grad_norm": 4.642940044403076,
      "learning_rate": 8.387613322691699e-06,
      "loss": 0.0588,
      "step": 5237
    },
    {
      "epoch": 0.2704879938032533,
      "grad_norm": 10.649658203125,
      "learning_rate": 8.387010635887722e-06,
      "loss": 0.1073,
      "step": 5238
    },
    {
      "epoch": 0.2705396333591531,
      "grad_norm": 4.967543601989746,
      "learning_rate": 8.386407858128707e-06,
      "loss": 0.0781,
      "step": 5239
    },
    {
      "epoch": 0.2705912729150529,
      "grad_norm": 1.6355516910552979,
      "learning_rate": 8.385804989430837e-06,
      "loss": 0.0368,
      "step": 5240
    },
    {
      "epoch": 0.27064291247095273,
      "grad_norm": 8.49625015258789,
      "learning_rate": 8.385202029810302e-06,
      "loss": 0.1117,
      "step": 5241
    },
    {
      "epoch": 0.27069455202685255,
      "grad_norm": 4.387425899505615,
      "learning_rate": 8.384598979283299e-06,
      "loss": 0.1199,
      "step": 5242
    },
    {
      "epoch": 0.27074619158275237,
      "grad_norm": 4.599014759063721,
      "learning_rate": 8.383995837866014e-06,
      "loss": 0.105,
      "step": 5243
    },
    {
      "epoch": 0.2707978311386522,
      "grad_norm": 4.854527950286865,
      "learning_rate": 8.383392605574652e-06,
      "loss": 0.061,
      "step": 5244
    },
    {
      "epoch": 0.270849470694552,
      "grad_norm": 1.741725206375122,
      "learning_rate": 8.382789282425408e-06,
      "loss": 0.0419,
      "step": 5245
    },
    {
      "epoch": 0.2709011102504518,
      "grad_norm": 4.45339298248291,
      "learning_rate": 8.382185868434483e-06,
      "loss": 0.1021,
      "step": 5246
    },
    {
      "epoch": 0.27095274980635164,
      "grad_norm": 6.06882381439209,
      "learning_rate": 8.381582363618083e-06,
      "loss": 0.0762,
      "step": 5247
    },
    {
      "epoch": 0.27100438936225146,
      "grad_norm": 2.2861745357513428,
      "learning_rate": 8.380978767992413e-06,
      "loss": 0.0565,
      "step": 5248
    },
    {
      "epoch": 0.2710560289181513,
      "grad_norm": 4.77047872543335,
      "learning_rate": 8.380375081573684e-06,
      "loss": 0.0698,
      "step": 5249
    },
    {
      "epoch": 0.2711076684740511,
      "grad_norm": 19.03074073791504,
      "learning_rate": 8.379771304378108e-06,
      "loss": 0.086,
      "step": 5250
    },
    {
      "epoch": 0.2711593080299509,
      "grad_norm": 5.674633979797363,
      "learning_rate": 8.379167436421894e-06,
      "loss": 0.0645,
      "step": 5251
    },
    {
      "epoch": 0.27121094758585074,
      "grad_norm": 3.6537320613861084,
      "learning_rate": 8.378563477721265e-06,
      "loss": 0.0669,
      "step": 5252
    },
    {
      "epoch": 0.27126258714175056,
      "grad_norm": 3.408252239227295,
      "learning_rate": 8.377959428292433e-06,
      "loss": 0.0659,
      "step": 5253
    },
    {
      "epoch": 0.2713142266976504,
      "grad_norm": 5.786510944366455,
      "learning_rate": 8.377355288151626e-06,
      "loss": 0.0674,
      "step": 5254
    },
    {
      "epoch": 0.2713658662535502,
      "grad_norm": 1.9770227670669556,
      "learning_rate": 8.37675105731506e-06,
      "loss": 0.0342,
      "step": 5255
    },
    {
      "epoch": 0.27141750580945007,
      "grad_norm": 13.44058609008789,
      "learning_rate": 8.37614673579897e-06,
      "loss": 0.2557,
      "step": 5256
    },
    {
      "epoch": 0.2714691453653499,
      "grad_norm": 3.5300729274749756,
      "learning_rate": 8.375542323619576e-06,
      "loss": 0.0492,
      "step": 5257
    },
    {
      "epoch": 0.2715207849212497,
      "grad_norm": 6.61871337890625,
      "learning_rate": 8.374937820793115e-06,
      "loss": 0.0731,
      "step": 5258
    },
    {
      "epoch": 0.2715724244771495,
      "grad_norm": 6.368987083435059,
      "learning_rate": 8.374333227335815e-06,
      "loss": 0.0927,
      "step": 5259
    },
    {
      "epoch": 0.27162406403304934,
      "grad_norm": 4.433865070343018,
      "learning_rate": 8.373728543263918e-06,
      "loss": 0.0411,
      "step": 5260
    },
    {
      "epoch": 0.27167570358894916,
      "grad_norm": 1.8532723188400269,
      "learning_rate": 8.373123768593658e-06,
      "loss": 0.0396,
      "step": 5261
    },
    {
      "epoch": 0.271727343144849,
      "grad_norm": 10.835604667663574,
      "learning_rate": 8.372518903341276e-06,
      "loss": 0.197,
      "step": 5262
    },
    {
      "epoch": 0.2717789827007488,
      "grad_norm": 9.213435173034668,
      "learning_rate": 8.371913947523016e-06,
      "loss": 0.1206,
      "step": 5263
    },
    {
      "epoch": 0.2718306222566486,
      "grad_norm": 4.500839710235596,
      "learning_rate": 8.371308901155125e-06,
      "loss": 0.0766,
      "step": 5264
    },
    {
      "epoch": 0.27188226181254843,
      "grad_norm": 6.850279808044434,
      "learning_rate": 8.370703764253846e-06,
      "loss": 0.1005,
      "step": 5265
    },
    {
      "epoch": 0.27193390136844825,
      "grad_norm": 2.1417479515075684,
      "learning_rate": 8.370098536835436e-06,
      "loss": 0.0472,
      "step": 5266
    },
    {
      "epoch": 0.27198554092434807,
      "grad_norm": 6.701316833496094,
      "learning_rate": 8.369493218916142e-06,
      "loss": 0.0716,
      "step": 5267
    },
    {
      "epoch": 0.2720371804802479,
      "grad_norm": 9.396927833557129,
      "learning_rate": 8.368887810512225e-06,
      "loss": 0.0902,
      "step": 5268
    },
    {
      "epoch": 0.2720888200361477,
      "grad_norm": 4.073146343231201,
      "learning_rate": 8.368282311639937e-06,
      "loss": 0.0599,
      "step": 5269
    },
    {
      "epoch": 0.2721404595920475,
      "grad_norm": 4.0724992752075195,
      "learning_rate": 8.36767672231554e-06,
      "loss": 0.0786,
      "step": 5270
    },
    {
      "epoch": 0.27219209914794734,
      "grad_norm": 3.8739254474639893,
      "learning_rate": 8.367071042555298e-06,
      "loss": 0.0616,
      "step": 5271
    },
    {
      "epoch": 0.27224373870384716,
      "grad_norm": 3.546881675720215,
      "learning_rate": 8.366465272375477e-06,
      "loss": 0.0675,
      "step": 5272
    },
    {
      "epoch": 0.272295378259747,
      "grad_norm": 1.8111298084259033,
      "learning_rate": 8.36585941179234e-06,
      "loss": 0.0246,
      "step": 5273
    },
    {
      "epoch": 0.2723470178156468,
      "grad_norm": 5.484699726104736,
      "learning_rate": 8.36525346082216e-06,
      "loss": 0.068,
      "step": 5274
    },
    {
      "epoch": 0.2723986573715466,
      "grad_norm": 4.514432907104492,
      "learning_rate": 8.364647419481208e-06,
      "loss": 0.0417,
      "step": 5275
    },
    {
      "epoch": 0.27245029692744643,
      "grad_norm": 5.84868860244751,
      "learning_rate": 8.36404128778576e-06,
      "loss": 0.062,
      "step": 5276
    },
    {
      "epoch": 0.27250193648334625,
      "grad_norm": 17.38832664489746,
      "learning_rate": 8.363435065752092e-06,
      "loss": 0.2364,
      "step": 5277
    },
    {
      "epoch": 0.27255357603924607,
      "grad_norm": 12.739180564880371,
      "learning_rate": 8.362828753396484e-06,
      "loss": 0.246,
      "step": 5278
    },
    {
      "epoch": 0.2726052155951459,
      "grad_norm": 3.730088949203491,
      "learning_rate": 8.362222350735218e-06,
      "loss": 0.05,
      "step": 5279
    },
    {
      "epoch": 0.2726568551510457,
      "grad_norm": 6.041232109069824,
      "learning_rate": 8.36161585778458e-06,
      "loss": 0.0797,
      "step": 5280
    },
    {
      "epoch": 0.2727084947069455,
      "grad_norm": 6.577685356140137,
      "learning_rate": 8.361009274560852e-06,
      "loss": 0.0792,
      "step": 5281
    },
    {
      "epoch": 0.27276013426284534,
      "grad_norm": 4.0498504638671875,
      "learning_rate": 8.360402601080327e-06,
      "loss": 0.0527,
      "step": 5282
    },
    {
      "epoch": 0.27281177381874516,
      "grad_norm": 12.110209465026855,
      "learning_rate": 8.359795837359296e-06,
      "loss": 0.093,
      "step": 5283
    },
    {
      "epoch": 0.272863413374645,
      "grad_norm": 7.433352947235107,
      "learning_rate": 8.359188983414054e-06,
      "loss": 0.1081,
      "step": 5284
    },
    {
      "epoch": 0.2729150529305448,
      "grad_norm": 8.448890686035156,
      "learning_rate": 8.358582039260896e-06,
      "loss": 0.136,
      "step": 5285
    },
    {
      "epoch": 0.2729666924864446,
      "grad_norm": 10.54899787902832,
      "learning_rate": 8.357975004916122e-06,
      "loss": 0.1315,
      "step": 5286
    },
    {
      "epoch": 0.27301833204234444,
      "grad_norm": 16.251659393310547,
      "learning_rate": 8.357367880396032e-06,
      "loss": 0.1558,
      "step": 5287
    },
    {
      "epoch": 0.27306997159824425,
      "grad_norm": 16.00926971435547,
      "learning_rate": 8.35676066571693e-06,
      "loss": 0.091,
      "step": 5288
    },
    {
      "epoch": 0.2731216111541441,
      "grad_norm": 3.511786937713623,
      "learning_rate": 8.356153360895124e-06,
      "loss": 0.0459,
      "step": 5289
    },
    {
      "epoch": 0.2731732507100439,
      "grad_norm": 6.473163604736328,
      "learning_rate": 8.355545965946921e-06,
      "loss": 0.0581,
      "step": 5290
    },
    {
      "epoch": 0.2732248902659437,
      "grad_norm": 6.701935768127441,
      "learning_rate": 8.354938480888632e-06,
      "loss": 0.071,
      "step": 5291
    },
    {
      "epoch": 0.27327652982184353,
      "grad_norm": 4.0836687088012695,
      "learning_rate": 8.35433090573657e-06,
      "loss": 0.0962,
      "step": 5292
    },
    {
      "epoch": 0.27332816937774335,
      "grad_norm": 2.9527251720428467,
      "learning_rate": 8.353723240507052e-06,
      "loss": 0.0583,
      "step": 5293
    },
    {
      "epoch": 0.27337980893364316,
      "grad_norm": 3.341049909591675,
      "learning_rate": 8.353115485216397e-06,
      "loss": 0.069,
      "step": 5294
    },
    {
      "epoch": 0.273431448489543,
      "grad_norm": 4.741557598114014,
      "learning_rate": 8.352507639880924e-06,
      "loss": 0.1029,
      "step": 5295
    },
    {
      "epoch": 0.2734830880454428,
      "grad_norm": 11.710344314575195,
      "learning_rate": 8.351899704516956e-06,
      "loss": 0.1497,
      "step": 5296
    },
    {
      "epoch": 0.2735347276013426,
      "grad_norm": 3.120055675506592,
      "learning_rate": 8.351291679140822e-06,
      "loss": 0.0582,
      "step": 5297
    },
    {
      "epoch": 0.27358636715724244,
      "grad_norm": 5.037420749664307,
      "learning_rate": 8.350683563768846e-06,
      "loss": 0.0676,
      "step": 5298
    },
    {
      "epoch": 0.27363800671314226,
      "grad_norm": 6.900123596191406,
      "learning_rate": 8.350075358417358e-06,
      "loss": 0.0795,
      "step": 5299
    },
    {
      "epoch": 0.2736896462690421,
      "grad_norm": 6.782236099243164,
      "learning_rate": 8.349467063102694e-06,
      "loss": 0.0901,
      "step": 5300
    },
    {
      "epoch": 0.2737412858249419,
      "grad_norm": 4.3543381690979,
      "learning_rate": 8.34885867784119e-06,
      "loss": 0.0422,
      "step": 5301
    },
    {
      "epoch": 0.2737929253808417,
      "grad_norm": 3.4915804862976074,
      "learning_rate": 8.348250202649178e-06,
      "loss": 0.0852,
      "step": 5302
    },
    {
      "epoch": 0.27384456493674153,
      "grad_norm": 8.723493576049805,
      "learning_rate": 8.347641637543002e-06,
      "loss": 0.1162,
      "step": 5303
    },
    {
      "epoch": 0.27389620449264135,
      "grad_norm": 2.5976834297180176,
      "learning_rate": 8.347032982539002e-06,
      "loss": 0.0626,
      "step": 5304
    },
    {
      "epoch": 0.27394784404854117,
      "grad_norm": 2.612579822540283,
      "learning_rate": 8.346424237653527e-06,
      "loss": 0.0561,
      "step": 5305
    },
    {
      "epoch": 0.273999483604441,
      "grad_norm": 1.86219322681427,
      "learning_rate": 8.345815402902922e-06,
      "loss": 0.0328,
      "step": 5306
    },
    {
      "epoch": 0.2740511231603408,
      "grad_norm": 4.654117584228516,
      "learning_rate": 8.345206478303536e-06,
      "loss": 0.0332,
      "step": 5307
    },
    {
      "epoch": 0.2741027627162406,
      "grad_norm": 3.583331823348999,
      "learning_rate": 8.344597463871722e-06,
      "loss": 0.0556,
      "step": 5308
    },
    {
      "epoch": 0.27415440227214044,
      "grad_norm": 3.7086455821990967,
      "learning_rate": 8.343988359623835e-06,
      "loss": 0.0495,
      "step": 5309
    },
    {
      "epoch": 0.27420604182804026,
      "grad_norm": 4.865461349487305,
      "learning_rate": 8.343379165576232e-06,
      "loss": 0.074,
      "step": 5310
    },
    {
      "epoch": 0.2742576813839401,
      "grad_norm": 2.523778200149536,
      "learning_rate": 8.342769881745268e-06,
      "loss": 0.0127,
      "step": 5311
    },
    {
      "epoch": 0.2743093209398399,
      "grad_norm": 10.88981819152832,
      "learning_rate": 8.34216050814731e-06,
      "loss": 0.0948,
      "step": 5312
    },
    {
      "epoch": 0.2743609604957397,
      "grad_norm": 12.228546142578125,
      "learning_rate": 8.341551044798721e-06,
      "loss": 0.0719,
      "step": 5313
    },
    {
      "epoch": 0.27441260005163953,
      "grad_norm": 12.687535285949707,
      "learning_rate": 8.340941491715867e-06,
      "loss": 0.1834,
      "step": 5314
    },
    {
      "epoch": 0.27446423960753935,
      "grad_norm": 3.3016507625579834,
      "learning_rate": 8.340331848915116e-06,
      "loss": 0.045,
      "step": 5315
    },
    {
      "epoch": 0.27451587916343917,
      "grad_norm": 8.75401496887207,
      "learning_rate": 8.339722116412843e-06,
      "loss": 0.0755,
      "step": 5316
    },
    {
      "epoch": 0.274567518719339,
      "grad_norm": 6.8543572425842285,
      "learning_rate": 8.339112294225416e-06,
      "loss": 0.1309,
      "step": 5317
    },
    {
      "epoch": 0.2746191582752388,
      "grad_norm": 6.7941365242004395,
      "learning_rate": 8.338502382369216e-06,
      "loss": 0.1356,
      "step": 5318
    },
    {
      "epoch": 0.2746707978311386,
      "grad_norm": 6.1730875968933105,
      "learning_rate": 8.33789238086062e-06,
      "loss": 0.0976,
      "step": 5319
    },
    {
      "epoch": 0.2747224373870385,
      "grad_norm": 7.3824310302734375,
      "learning_rate": 8.337282289716008e-06,
      "loss": 0.0496,
      "step": 5320
    },
    {
      "epoch": 0.2747740769429383,
      "grad_norm": 1.6574790477752686,
      "learning_rate": 8.336672108951767e-06,
      "loss": 0.0283,
      "step": 5321
    },
    {
      "epoch": 0.27482571649883814,
      "grad_norm": 2.0509817600250244,
      "learning_rate": 8.336061838584278e-06,
      "loss": 0.0415,
      "step": 5322
    },
    {
      "epoch": 0.27487735605473795,
      "grad_norm": 4.507053375244141,
      "learning_rate": 8.335451478629934e-06,
      "loss": 0.1211,
      "step": 5323
    },
    {
      "epoch": 0.27492899561063777,
      "grad_norm": 8.127909660339355,
      "learning_rate": 8.33484102910512e-06,
      "loss": 0.125,
      "step": 5324
    },
    {
      "epoch": 0.2749806351665376,
      "grad_norm": 10.085957527160645,
      "learning_rate": 8.334230490026235e-06,
      "loss": 0.0899,
      "step": 5325
    },
    {
      "epoch": 0.2750322747224374,
      "grad_norm": 5.35077428817749,
      "learning_rate": 8.33361986140967e-06,
      "loss": 0.0719,
      "step": 5326
    },
    {
      "epoch": 0.2750839142783372,
      "grad_norm": 2.789827585220337,
      "learning_rate": 8.333009143271826e-06,
      "loss": 0.0433,
      "step": 5327
    },
    {
      "epoch": 0.27513555383423705,
      "grad_norm": 4.979292392730713,
      "learning_rate": 8.332398335629102e-06,
      "loss": 0.1001,
      "step": 5328
    },
    {
      "epoch": 0.27518719339013686,
      "grad_norm": 9.435296058654785,
      "learning_rate": 8.3317874384979e-06,
      "loss": 0.0835,
      "step": 5329
    },
    {
      "epoch": 0.2752388329460367,
      "grad_norm": 3.518482208251953,
      "learning_rate": 8.331176451894625e-06,
      "loss": 0.051,
      "step": 5330
    },
    {
      "epoch": 0.2752904725019365,
      "grad_norm": 4.5433669090271,
      "learning_rate": 8.330565375835685e-06,
      "loss": 0.082,
      "step": 5331
    },
    {
      "epoch": 0.2753421120578363,
      "grad_norm": 7.559693336486816,
      "learning_rate": 8.329954210337492e-06,
      "loss": 0.0649,
      "step": 5332
    },
    {
      "epoch": 0.27539375161373614,
      "grad_norm": 4.570307731628418,
      "learning_rate": 8.329342955416454e-06,
      "loss": 0.091,
      "step": 5333
    },
    {
      "epoch": 0.27544539116963596,
      "grad_norm": 1.8506700992584229,
      "learning_rate": 8.32873161108899e-06,
      "loss": 0.0327,
      "step": 5334
    },
    {
      "epoch": 0.2754970307255358,
      "grad_norm": 2.683126926422119,
      "learning_rate": 8.328120177371512e-06,
      "loss": 0.0599,
      "step": 5335
    },
    {
      "epoch": 0.2755486702814356,
      "grad_norm": 4.7194013595581055,
      "learning_rate": 8.327508654280446e-06,
      "loss": 0.0644,
      "step": 5336
    },
    {
      "epoch": 0.2756003098373354,
      "grad_norm": 6.771997451782227,
      "learning_rate": 8.32689704183221e-06,
      "loss": 0.1605,
      "step": 5337
    },
    {
      "epoch": 0.27565194939323523,
      "grad_norm": 4.728031158447266,
      "learning_rate": 8.326285340043227e-06,
      "loss": 0.0522,
      "step": 5338
    },
    {
      "epoch": 0.27570358894913505,
      "grad_norm": 2.966099262237549,
      "learning_rate": 8.325673548929927e-06,
      "loss": 0.0508,
      "step": 5339
    },
    {
      "epoch": 0.27575522850503487,
      "grad_norm": 7.779635906219482,
      "learning_rate": 8.325061668508735e-06,
      "loss": 0.0929,
      "step": 5340
    },
    {
      "epoch": 0.2758068680609347,
      "grad_norm": 4.410200595855713,
      "learning_rate": 8.324449698796088e-06,
      "loss": 0.067,
      "step": 5341
    },
    {
      "epoch": 0.2758585076168345,
      "grad_norm": 8.91673755645752,
      "learning_rate": 8.323837639808416e-06,
      "loss": 0.0841,
      "step": 5342
    },
    {
      "epoch": 0.2759101471727343,
      "grad_norm": 5.62050724029541,
      "learning_rate": 8.323225491562158e-06,
      "loss": 0.0444,
      "step": 5343
    },
    {
      "epoch": 0.27596178672863414,
      "grad_norm": 5.489266395568848,
      "learning_rate": 8.322613254073746e-06,
      "loss": 0.1002,
      "step": 5344
    },
    {
      "epoch": 0.27601342628453396,
      "grad_norm": 5.037083625793457,
      "learning_rate": 8.322000927359629e-06,
      "loss": 0.1001,
      "step": 5345
    },
    {
      "epoch": 0.2760650658404338,
      "grad_norm": 5.234340190887451,
      "learning_rate": 8.321388511436247e-06,
      "loss": 0.0576,
      "step": 5346
    },
    {
      "epoch": 0.2761167053963336,
      "grad_norm": 15.915533065795898,
      "learning_rate": 8.320776006320046e-06,
      "loss": 0.1595,
      "step": 5347
    },
    {
      "epoch": 0.2761683449522334,
      "grad_norm": 3.5547170639038086,
      "learning_rate": 8.320163412027475e-06,
      "loss": 0.0527,
      "step": 5348
    },
    {
      "epoch": 0.27621998450813323,
      "grad_norm": 6.600675582885742,
      "learning_rate": 8.319550728574984e-06,
      "loss": 0.0997,
      "step": 5349
    },
    {
      "epoch": 0.27627162406403305,
      "grad_norm": 3.7687957286834717,
      "learning_rate": 8.318937955979025e-06,
      "loss": 0.0334,
      "step": 5350
    },
    {
      "epoch": 0.27632326361993287,
      "grad_norm": 1.7848032712936401,
      "learning_rate": 8.318325094256057e-06,
      "loss": 0.0363,
      "step": 5351
    },
    {
      "epoch": 0.2763749031758327,
      "grad_norm": 2.714686393737793,
      "learning_rate": 8.317712143422534e-06,
      "loss": 0.0472,
      "step": 5352
    },
    {
      "epoch": 0.2764265427317325,
      "grad_norm": 7.9519195556640625,
      "learning_rate": 8.317099103494918e-06,
      "loss": 0.1014,
      "step": 5353
    },
    {
      "epoch": 0.2764781822876323,
      "grad_norm": 5.699878692626953,
      "learning_rate": 8.316485974489672e-06,
      "loss": 0.053,
      "step": 5354
    },
    {
      "epoch": 0.27652982184353214,
      "grad_norm": 3.7851502895355225,
      "learning_rate": 8.315872756423259e-06,
      "loss": 0.0511,
      "step": 5355
    },
    {
      "epoch": 0.27658146139943196,
      "grad_norm": 6.1524128913879395,
      "learning_rate": 8.315259449312148e-06,
      "loss": 0.0828,
      "step": 5356
    },
    {
      "epoch": 0.2766331009553318,
      "grad_norm": 3.2079687118530273,
      "learning_rate": 8.314646053172807e-06,
      "loss": 0.0484,
      "step": 5357
    },
    {
      "epoch": 0.2766847405112316,
      "grad_norm": 13.130579948425293,
      "learning_rate": 8.314032568021712e-06,
      "loss": 0.1026,
      "step": 5358
    },
    {
      "epoch": 0.2767363800671314,
      "grad_norm": 3.5293569564819336,
      "learning_rate": 8.313418993875336e-06,
      "loss": 0.0431,
      "step": 5359
    },
    {
      "epoch": 0.27678801962303123,
      "grad_norm": 2.259690999984741,
      "learning_rate": 8.312805330750155e-06,
      "loss": 0.0525,
      "step": 5360
    },
    {
      "epoch": 0.27683965917893105,
      "grad_norm": 10.949966430664062,
      "learning_rate": 8.312191578662647e-06,
      "loss": 0.201,
      "step": 5361
    },
    {
      "epoch": 0.27689129873483087,
      "grad_norm": 5.082334518432617,
      "learning_rate": 8.311577737629297e-06,
      "loss": 0.0658,
      "step": 5362
    },
    {
      "epoch": 0.2769429382907307,
      "grad_norm": 23.684986114501953,
      "learning_rate": 8.310963807666586e-06,
      "loss": 0.1903,
      "step": 5363
    },
    {
      "epoch": 0.2769945778466305,
      "grad_norm": 3.084543466567993,
      "learning_rate": 8.310349788791002e-06,
      "loss": 0.0398,
      "step": 5364
    },
    {
      "epoch": 0.2770462174025303,
      "grad_norm": 3.0863759517669678,
      "learning_rate": 8.309735681019035e-06,
      "loss": 0.0668,
      "step": 5365
    },
    {
      "epoch": 0.27709785695843014,
      "grad_norm": 5.356058120727539,
      "learning_rate": 8.309121484367175e-06,
      "loss": 0.0787,
      "step": 5366
    },
    {
      "epoch": 0.27714949651432996,
      "grad_norm": 3.0063729286193848,
      "learning_rate": 8.308507198851915e-06,
      "loss": 0.0528,
      "step": 5367
    },
    {
      "epoch": 0.2772011360702298,
      "grad_norm": 8.737452507019043,
      "learning_rate": 8.307892824489754e-06,
      "loss": 0.0908,
      "step": 5368
    },
    {
      "epoch": 0.2772527756261296,
      "grad_norm": 2.3514914512634277,
      "learning_rate": 8.307278361297186e-06,
      "loss": 0.0423,
      "step": 5369
    },
    {
      "epoch": 0.2773044151820294,
      "grad_norm": 3.1885738372802734,
      "learning_rate": 8.306663809290716e-06,
      "loss": 0.0556,
      "step": 5370
    },
    {
      "epoch": 0.27735605473792924,
      "grad_norm": 1.4533216953277588,
      "learning_rate": 8.306049168486844e-06,
      "loss": 0.0312,
      "step": 5371
    },
    {
      "epoch": 0.27740769429382905,
      "grad_norm": 4.279135704040527,
      "learning_rate": 8.305434438902078e-06,
      "loss": 0.0644,
      "step": 5372
    },
    {
      "epoch": 0.2774593338497289,
      "grad_norm": 14.500287055969238,
      "learning_rate": 8.304819620552926e-06,
      "loss": 0.0993,
      "step": 5373
    },
    {
      "epoch": 0.2775109734056287,
      "grad_norm": 5.5429277420043945,
      "learning_rate": 8.304204713455897e-06,
      "loss": 0.0944,
      "step": 5374
    },
    {
      "epoch": 0.2775626129615285,
      "grad_norm": 6.532596111297607,
      "learning_rate": 8.303589717627503e-06,
      "loss": 0.0807,
      "step": 5375
    },
    {
      "epoch": 0.27761425251742833,
      "grad_norm": 6.939639568328857,
      "learning_rate": 8.302974633084262e-06,
      "loss": 0.0576,
      "step": 5376
    },
    {
      "epoch": 0.27766589207332815,
      "grad_norm": 5.615077972412109,
      "learning_rate": 8.302359459842688e-06,
      "loss": 0.0867,
      "step": 5377
    },
    {
      "epoch": 0.27771753162922796,
      "grad_norm": 5.012855052947998,
      "learning_rate": 8.301744197919305e-06,
      "loss": 0.0619,
      "step": 5378
    },
    {
      "epoch": 0.2777691711851278,
      "grad_norm": 2.4595906734466553,
      "learning_rate": 8.301128847330631e-06,
      "loss": 0.0373,
      "step": 5379
    },
    {
      "epoch": 0.2778208107410276,
      "grad_norm": 17.350566864013672,
      "learning_rate": 8.300513408093196e-06,
      "loss": 0.1156,
      "step": 5380
    },
    {
      "epoch": 0.2778724502969274,
      "grad_norm": 6.633977890014648,
      "learning_rate": 8.29989788022352e-06,
      "loss": 0.0742,
      "step": 5381
    },
    {
      "epoch": 0.27792408985282724,
      "grad_norm": 5.547778606414795,
      "learning_rate": 8.29928226373814e-06,
      "loss": 0.0559,
      "step": 5382
    },
    {
      "epoch": 0.2779757294087271,
      "grad_norm": 17.881118774414062,
      "learning_rate": 8.298666558653582e-06,
      "loss": 0.2042,
      "step": 5383
    },
    {
      "epoch": 0.27802736896462693,
      "grad_norm": 7.436159133911133,
      "learning_rate": 8.29805076498638e-06,
      "loss": 0.0531,
      "step": 5384
    },
    {
      "epoch": 0.27807900852052675,
      "grad_norm": 7.983704090118408,
      "learning_rate": 8.297434882753076e-06,
      "loss": 0.102,
      "step": 5385
    },
    {
      "epoch": 0.27813064807642657,
      "grad_norm": 3.1750147342681885,
      "learning_rate": 8.296818911970206e-06,
      "loss": 0.0731,
      "step": 5386
    },
    {
      "epoch": 0.2781822876323264,
      "grad_norm": 7.6977434158325195,
      "learning_rate": 8.296202852654311e-06,
      "loss": 0.0501,
      "step": 5387
    },
    {
      "epoch": 0.2782339271882262,
      "grad_norm": 4.815088748931885,
      "learning_rate": 8.295586704821934e-06,
      "loss": 0.0884,
      "step": 5388
    },
    {
      "epoch": 0.278285566744126,
      "grad_norm": 17.26820945739746,
      "learning_rate": 8.29497046848962e-06,
      "loss": 0.1937,
      "step": 5389
    },
    {
      "epoch": 0.27833720630002584,
      "grad_norm": 9.779212951660156,
      "learning_rate": 8.294354143673922e-06,
      "loss": 0.0883,
      "step": 5390
    },
    {
      "epoch": 0.27838884585592566,
      "grad_norm": 7.926348686218262,
      "learning_rate": 8.293737730391387e-06,
      "loss": 0.0569,
      "step": 5391
    },
    {
      "epoch": 0.2784404854118255,
      "grad_norm": 2.909449815750122,
      "learning_rate": 8.293121228658571e-06,
      "loss": 0.0471,
      "step": 5392
    },
    {
      "epoch": 0.2784921249677253,
      "grad_norm": 4.12155294418335,
      "learning_rate": 8.292504638492028e-06,
      "loss": 0.0255,
      "step": 5393
    },
    {
      "epoch": 0.2785437645236251,
      "grad_norm": 8.982019424438477,
      "learning_rate": 8.291887959908314e-06,
      "loss": 0.1048,
      "step": 5394
    },
    {
      "epoch": 0.27859540407952493,
      "grad_norm": 3.0076920986175537,
      "learning_rate": 8.29127119292399e-06,
      "loss": 0.0197,
      "step": 5395
    },
    {
      "epoch": 0.27864704363542475,
      "grad_norm": 6.521435737609863,
      "learning_rate": 8.290654337555624e-06,
      "loss": 0.1323,
      "step": 5396
    },
    {
      "epoch": 0.27869868319132457,
      "grad_norm": 2.987152338027954,
      "learning_rate": 8.290037393819774e-06,
      "loss": 0.0384,
      "step": 5397
    },
    {
      "epoch": 0.2787503227472244,
      "grad_norm": 2.7592051029205322,
      "learning_rate": 8.289420361733013e-06,
      "loss": 0.0236,
      "step": 5398
    },
    {
      "epoch": 0.2788019623031242,
      "grad_norm": 8.507075309753418,
      "learning_rate": 8.288803241311907e-06,
      "loss": 0.1221,
      "step": 5399
    },
    {
      "epoch": 0.278853601859024,
      "grad_norm": 7.513215065002441,
      "learning_rate": 8.28818603257303e-06,
      "loss": 0.091,
      "step": 5400
    },
    {
      "epoch": 0.27890524141492384,
      "grad_norm": 2.3372209072113037,
      "learning_rate": 8.287568735532955e-06,
      "loss": 0.0608,
      "step": 5401
    },
    {
      "epoch": 0.27895688097082366,
      "grad_norm": 3.8993663787841797,
      "learning_rate": 8.28695135020826e-06,
      "loss": 0.0804,
      "step": 5402
    },
    {
      "epoch": 0.2790085205267235,
      "grad_norm": 8.375905990600586,
      "learning_rate": 8.286333876615525e-06,
      "loss": 0.1285,
      "step": 5403
    },
    {
      "epoch": 0.2790601600826233,
      "grad_norm": 10.663785934448242,
      "learning_rate": 8.285716314771333e-06,
      "loss": 0.108,
      "step": 5404
    },
    {
      "epoch": 0.2791117996385231,
      "grad_norm": 4.222553730010986,
      "learning_rate": 8.285098664692264e-06,
      "loss": 0.0522,
      "step": 5405
    },
    {
      "epoch": 0.27916343919442294,
      "grad_norm": 16.09261703491211,
      "learning_rate": 8.284480926394907e-06,
      "loss": 0.1415,
      "step": 5406
    },
    {
      "epoch": 0.27921507875032275,
      "grad_norm": 11.278653144836426,
      "learning_rate": 8.28386309989585e-06,
      "loss": 0.1697,
      "step": 5407
    },
    {
      "epoch": 0.27926671830622257,
      "grad_norm": 7.306586265563965,
      "learning_rate": 8.283245185211684e-06,
      "loss": 0.1,
      "step": 5408
    },
    {
      "epoch": 0.2793183578621224,
      "grad_norm": 2.9414148330688477,
      "learning_rate": 8.282627182359004e-06,
      "loss": 0.0644,
      "step": 5409
    },
    {
      "epoch": 0.2793699974180222,
      "grad_norm": 11.629561424255371,
      "learning_rate": 8.282009091354403e-06,
      "loss": 0.1083,
      "step": 5410
    },
    {
      "epoch": 0.279421636973922,
      "grad_norm": 3.0493080615997314,
      "learning_rate": 8.281390912214485e-06,
      "loss": 0.0659,
      "step": 5411
    },
    {
      "epoch": 0.27947327652982185,
      "grad_norm": 3.690478801727295,
      "learning_rate": 8.280772644955843e-06,
      "loss": 0.0399,
      "step": 5412
    },
    {
      "epoch": 0.27952491608572166,
      "grad_norm": 15.379682540893555,
      "learning_rate": 8.280154289595086e-06,
      "loss": 0.1433,
      "step": 5413
    },
    {
      "epoch": 0.2795765556416215,
      "grad_norm": 5.530844688415527,
      "learning_rate": 8.279535846148816e-06,
      "loss": 0.0615,
      "step": 5414
    },
    {
      "epoch": 0.2796281951975213,
      "grad_norm": 2.7243685722351074,
      "learning_rate": 8.278917314633644e-06,
      "loss": 0.0678,
      "step": 5415
    },
    {
      "epoch": 0.2796798347534211,
      "grad_norm": 7.651076793670654,
      "learning_rate": 8.278298695066174e-06,
      "loss": 0.0902,
      "step": 5416
    },
    {
      "epoch": 0.27973147430932094,
      "grad_norm": 6.140198230743408,
      "learning_rate": 8.277679987463025e-06,
      "loss": 0.106,
      "step": 5417
    },
    {
      "epoch": 0.27978311386522076,
      "grad_norm": 5.814450263977051,
      "learning_rate": 8.27706119184081e-06,
      "loss": 0.0556,
      "step": 5418
    },
    {
      "epoch": 0.2798347534211206,
      "grad_norm": 3.0949692726135254,
      "learning_rate": 8.276442308216144e-06,
      "loss": 0.0671,
      "step": 5419
    },
    {
      "epoch": 0.2798863929770204,
      "grad_norm": 12.32674503326416,
      "learning_rate": 8.27582333660565e-06,
      "loss": 0.1402,
      "step": 5420
    },
    {
      "epoch": 0.2799380325329202,
      "grad_norm": 3.4582581520080566,
      "learning_rate": 8.275204277025945e-06,
      "loss": 0.0542,
      "step": 5421
    },
    {
      "epoch": 0.27998967208882003,
      "grad_norm": 2.2164347171783447,
      "learning_rate": 8.274585129493658e-06,
      "loss": 0.0394,
      "step": 5422
    },
    {
      "epoch": 0.28004131164471985,
      "grad_norm": 2.58024525642395,
      "learning_rate": 8.273965894025414e-06,
      "loss": 0.0455,
      "step": 5423
    },
    {
      "epoch": 0.28009295120061967,
      "grad_norm": 7.085788726806641,
      "learning_rate": 8.27334657063784e-06,
      "loss": 0.0603,
      "step": 5424
    },
    {
      "epoch": 0.2801445907565195,
      "grad_norm": 3.716841459274292,
      "learning_rate": 8.272727159347572e-06,
      "loss": 0.0629,
      "step": 5425
    },
    {
      "epoch": 0.2801962303124193,
      "grad_norm": 3.546212673187256,
      "learning_rate": 8.272107660171237e-06,
      "loss": 0.0509,
      "step": 5426
    },
    {
      "epoch": 0.2802478698683191,
      "grad_norm": 2.2338359355926514,
      "learning_rate": 8.271488073125478e-06,
      "loss": 0.0281,
      "step": 5427
    },
    {
      "epoch": 0.28029950942421894,
      "grad_norm": 4.6103291511535645,
      "learning_rate": 8.27086839822693e-06,
      "loss": 0.0431,
      "step": 5428
    },
    {
      "epoch": 0.28035114898011876,
      "grad_norm": 6.463133335113525,
      "learning_rate": 8.270248635492232e-06,
      "loss": 0.0798,
      "step": 5429
    },
    {
      "epoch": 0.2804027885360186,
      "grad_norm": 4.400264739990234,
      "learning_rate": 8.269628784938032e-06,
      "loss": 0.0564,
      "step": 5430
    },
    {
      "epoch": 0.2804544280919184,
      "grad_norm": 3.208181619644165,
      "learning_rate": 8.269008846580972e-06,
      "loss": 0.0606,
      "step": 5431
    },
    {
      "epoch": 0.2805060676478182,
      "grad_norm": 4.370327949523926,
      "learning_rate": 8.268388820437701e-06,
      "loss": 0.0656,
      "step": 5432
    },
    {
      "epoch": 0.28055770720371803,
      "grad_norm": 4.082813262939453,
      "learning_rate": 8.267768706524869e-06,
      "loss": 0.0525,
      "step": 5433
    },
    {
      "epoch": 0.28060934675961785,
      "grad_norm": 4.516117572784424,
      "learning_rate": 8.267148504859127e-06,
      "loss": 0.0838,
      "step": 5434
    },
    {
      "epoch": 0.28066098631551767,
      "grad_norm": 12.132347106933594,
      "learning_rate": 8.266528215457133e-06,
      "loss": 0.074,
      "step": 5435
    },
    {
      "epoch": 0.2807126258714175,
      "grad_norm": 6.091349124908447,
      "learning_rate": 8.265907838335543e-06,
      "loss": 0.1148,
      "step": 5436
    },
    {
      "epoch": 0.2807642654273173,
      "grad_norm": 7.139803886413574,
      "learning_rate": 8.265287373511014e-06,
      "loss": 0.0923,
      "step": 5437
    },
    {
      "epoch": 0.2808159049832171,
      "grad_norm": 5.338864803314209,
      "learning_rate": 8.264666821000214e-06,
      "loss": 0.0538,
      "step": 5438
    },
    {
      "epoch": 0.28086754453911694,
      "grad_norm": 2.6059505939483643,
      "learning_rate": 8.264046180819801e-06,
      "loss": 0.0291,
      "step": 5439
    },
    {
      "epoch": 0.28091918409501676,
      "grad_norm": 6.963951110839844,
      "learning_rate": 8.263425452986445e-06,
      "loss": 0.0641,
      "step": 5440
    },
    {
      "epoch": 0.2809708236509166,
      "grad_norm": 7.149350166320801,
      "learning_rate": 8.262804637516816e-06,
      "loss": 0.0975,
      "step": 5441
    },
    {
      "epoch": 0.2810224632068164,
      "grad_norm": 4.507737636566162,
      "learning_rate": 8.262183734427583e-06,
      "loss": 0.0731,
      "step": 5442
    },
    {
      "epoch": 0.2810741027627162,
      "grad_norm": 5.603969097137451,
      "learning_rate": 8.261562743735421e-06,
      "loss": 0.0727,
      "step": 5443
    },
    {
      "epoch": 0.28112574231861603,
      "grad_norm": 24.413646697998047,
      "learning_rate": 8.260941665457006e-06,
      "loss": 0.232,
      "step": 5444
    },
    {
      "epoch": 0.28117738187451585,
      "grad_norm": 3.6812996864318848,
      "learning_rate": 8.260320499609015e-06,
      "loss": 0.0808,
      "step": 5445
    },
    {
      "epoch": 0.28122902143041567,
      "grad_norm": 12.621649742126465,
      "learning_rate": 8.259699246208135e-06,
      "loss": 0.1048,
      "step": 5446
    },
    {
      "epoch": 0.28128066098631554,
      "grad_norm": 5.353607177734375,
      "learning_rate": 8.259077905271039e-06,
      "loss": 0.046,
      "step": 5447
    },
    {
      "epoch": 0.28133230054221536,
      "grad_norm": 2.2685399055480957,
      "learning_rate": 8.258456476814421e-06,
      "loss": 0.0483,
      "step": 5448
    },
    {
      "epoch": 0.2813839400981152,
      "grad_norm": 14.794947624206543,
      "learning_rate": 8.257834960854965e-06,
      "loss": 0.0754,
      "step": 5449
    },
    {
      "epoch": 0.281435579654015,
      "grad_norm": 18.402442932128906,
      "learning_rate": 8.257213357409362e-06,
      "loss": 0.0814,
      "step": 5450
    },
    {
      "epoch": 0.2814872192099148,
      "grad_norm": 2.35758638381958,
      "learning_rate": 8.256591666494304e-06,
      "loss": 0.0525,
      "step": 5451
    },
    {
      "epoch": 0.28153885876581464,
      "grad_norm": 1.508847713470459,
      "learning_rate": 8.255969888126487e-06,
      "loss": 0.0225,
      "step": 5452
    },
    {
      "epoch": 0.28159049832171446,
      "grad_norm": 2.6706013679504395,
      "learning_rate": 8.25534802232261e-06,
      "loss": 0.069,
      "step": 5453
    },
    {
      "epoch": 0.2816421378776143,
      "grad_norm": 7.100154399871826,
      "learning_rate": 8.254726069099367e-06,
      "loss": 0.0826,
      "step": 5454
    },
    {
      "epoch": 0.2816937774335141,
      "grad_norm": 10.644585609436035,
      "learning_rate": 8.254104028473467e-06,
      "loss": 0.1004,
      "step": 5455
    },
    {
      "epoch": 0.2817454169894139,
      "grad_norm": 7.823297023773193,
      "learning_rate": 8.253481900461607e-06,
      "loss": 0.0771,
      "step": 5456
    },
    {
      "epoch": 0.28179705654531373,
      "grad_norm": 5.354516983032227,
      "learning_rate": 8.252859685080502e-06,
      "loss": 0.0825,
      "step": 5457
    },
    {
      "epoch": 0.28184869610121355,
      "grad_norm": 4.883264541625977,
      "learning_rate": 8.252237382346852e-06,
      "loss": 0.0727,
      "step": 5458
    },
    {
      "epoch": 0.28190033565711337,
      "grad_norm": 3.9913318157196045,
      "learning_rate": 8.251614992277375e-06,
      "loss": 0.053,
      "step": 5459
    },
    {
      "epoch": 0.2819519752130132,
      "grad_norm": 4.655886173248291,
      "learning_rate": 8.250992514888783e-06,
      "loss": 0.06,
      "step": 5460
    },
    {
      "epoch": 0.282003614768913,
      "grad_norm": 3.9629578590393066,
      "learning_rate": 8.250369950197793e-06,
      "loss": 0.0508,
      "step": 5461
    },
    {
      "epoch": 0.2820552543248128,
      "grad_norm": 6.014085292816162,
      "learning_rate": 8.24974729822112e-06,
      "loss": 0.0869,
      "step": 5462
    },
    {
      "epoch": 0.28210689388071264,
      "grad_norm": 12.0154390335083,
      "learning_rate": 8.249124558975486e-06,
      "loss": 0.1637,
      "step": 5463
    },
    {
      "epoch": 0.28215853343661246,
      "grad_norm": 20.13524627685547,
      "learning_rate": 8.248501732477616e-06,
      "loss": 0.0869,
      "step": 5464
    },
    {
      "epoch": 0.2822101729925123,
      "grad_norm": 2.6235430240631104,
      "learning_rate": 8.247878818744235e-06,
      "loss": 0.0573,
      "step": 5465
    },
    {
      "epoch": 0.2822618125484121,
      "grad_norm": 2.2138473987579346,
      "learning_rate": 8.247255817792069e-06,
      "loss": 0.0582,
      "step": 5466
    },
    {
      "epoch": 0.2823134521043119,
      "grad_norm": 4.603752136230469,
      "learning_rate": 8.24663272963785e-06,
      "loss": 0.0677,
      "step": 5467
    },
    {
      "epoch": 0.28236509166021173,
      "grad_norm": 2.882450580596924,
      "learning_rate": 8.246009554298308e-06,
      "loss": 0.0438,
      "step": 5468
    },
    {
      "epoch": 0.28241673121611155,
      "grad_norm": 11.128565788269043,
      "learning_rate": 8.245386291790181e-06,
      "loss": 0.0529,
      "step": 5469
    },
    {
      "epoch": 0.28246837077201137,
      "grad_norm": 20.156518936157227,
      "learning_rate": 8.244762942130206e-06,
      "loss": 0.1583,
      "step": 5470
    },
    {
      "epoch": 0.2825200103279112,
      "grad_norm": 7.024374008178711,
      "learning_rate": 8.24413950533512e-06,
      "loss": 0.0684,
      "step": 5471
    },
    {
      "epoch": 0.282571649883811,
      "grad_norm": 3.2017533779144287,
      "learning_rate": 8.243515981421664e-06,
      "loss": 0.0643,
      "step": 5472
    },
    {
      "epoch": 0.2826232894397108,
      "grad_norm": 3.948683738708496,
      "learning_rate": 8.242892370406586e-06,
      "loss": 0.0467,
      "step": 5473
    },
    {
      "epoch": 0.28267492899561064,
      "grad_norm": 7.482657432556152,
      "learning_rate": 8.24226867230663e-06,
      "loss": 0.04,
      "step": 5474
    },
    {
      "epoch": 0.28272656855151046,
      "grad_norm": 8.741303443908691,
      "learning_rate": 8.241644887138546e-06,
      "loss": 0.0934,
      "step": 5475
    },
    {
      "epoch": 0.2827782081074103,
      "grad_norm": 2.5836904048919678,
      "learning_rate": 8.241021014919085e-06,
      "loss": 0.0479,
      "step": 5476
    },
    {
      "epoch": 0.2828298476633101,
      "grad_norm": 2.495795488357544,
      "learning_rate": 8.240397055664998e-06,
      "loss": 0.0244,
      "step": 5477
    },
    {
      "epoch": 0.2828814872192099,
      "grad_norm": 6.023441314697266,
      "learning_rate": 8.239773009393045e-06,
      "loss": 0.0807,
      "step": 5478
    },
    {
      "epoch": 0.28293312677510973,
      "grad_norm": 8.650731086730957,
      "learning_rate": 8.23914887611998e-06,
      "loss": 0.056,
      "step": 5479
    },
    {
      "epoch": 0.28298476633100955,
      "grad_norm": 3.6665892601013184,
      "learning_rate": 8.238524655862566e-06,
      "loss": 0.0576,
      "step": 5480
    },
    {
      "epoch": 0.28303640588690937,
      "grad_norm": 5.657196044921875,
      "learning_rate": 8.237900348637566e-06,
      "loss": 0.0802,
      "step": 5481
    },
    {
      "epoch": 0.2830880454428092,
      "grad_norm": 5.24649715423584,
      "learning_rate": 8.237275954461746e-06,
      "loss": 0.0673,
      "step": 5482
    },
    {
      "epoch": 0.283139684998709,
      "grad_norm": 5.439725399017334,
      "learning_rate": 8.236651473351872e-06,
      "loss": 0.0528,
      "step": 5483
    },
    {
      "epoch": 0.2831913245546088,
      "grad_norm": 4.968580722808838,
      "learning_rate": 8.236026905324713e-06,
      "loss": 0.0822,
      "step": 5484
    },
    {
      "epoch": 0.28324296411050864,
      "grad_norm": 3.6807823181152344,
      "learning_rate": 8.235402250397042e-06,
      "loss": 0.0915,
      "step": 5485
    },
    {
      "epoch": 0.28329460366640846,
      "grad_norm": 3.6524815559387207,
      "learning_rate": 8.234777508585634e-06,
      "loss": 0.052,
      "step": 5486
    },
    {
      "epoch": 0.2833462432223083,
      "grad_norm": 3.2731688022613525,
      "learning_rate": 8.234152679907265e-06,
      "loss": 0.045,
      "step": 5487
    },
    {
      "epoch": 0.2833978827782081,
      "grad_norm": 4.068142414093018,
      "learning_rate": 8.233527764378717e-06,
      "loss": 0.0871,
      "step": 5488
    },
    {
      "epoch": 0.2834495223341079,
      "grad_norm": 1.6342138051986694,
      "learning_rate": 8.232902762016768e-06,
      "loss": 0.0168,
      "step": 5489
    },
    {
      "epoch": 0.28350116189000774,
      "grad_norm": 2.4523568153381348,
      "learning_rate": 8.232277672838204e-06,
      "loss": 0.0414,
      "step": 5490
    },
    {
      "epoch": 0.28355280144590755,
      "grad_norm": 8.15127944946289,
      "learning_rate": 8.23165249685981e-06,
      "loss": 0.0498,
      "step": 5491
    },
    {
      "epoch": 0.2836044410018074,
      "grad_norm": 36.8995246887207,
      "learning_rate": 8.231027234098374e-06,
      "loss": 0.0967,
      "step": 5492
    },
    {
      "epoch": 0.2836560805577072,
      "grad_norm": 5.398308277130127,
      "learning_rate": 8.230401884570689e-06,
      "loss": 0.0732,
      "step": 5493
    },
    {
      "epoch": 0.283707720113607,
      "grad_norm": 3.7226269245147705,
      "learning_rate": 8.229776448293547e-06,
      "loss": 0.067,
      "step": 5494
    },
    {
      "epoch": 0.2837593596695068,
      "grad_norm": 10.27381420135498,
      "learning_rate": 8.229150925283744e-06,
      "loss": 0.0591,
      "step": 5495
    },
    {
      "epoch": 0.28381099922540665,
      "grad_norm": 6.170619964599609,
      "learning_rate": 8.228525315558075e-06,
      "loss": 0.09,
      "step": 5496
    },
    {
      "epoch": 0.28386263878130646,
      "grad_norm": 3.782647132873535,
      "learning_rate": 8.227899619133346e-06,
      "loss": 0.0446,
      "step": 5497
    },
    {
      "epoch": 0.2839142783372063,
      "grad_norm": 7.04803466796875,
      "learning_rate": 8.227273836026353e-06,
      "loss": 0.1008,
      "step": 5498
    },
    {
      "epoch": 0.2839659178931061,
      "grad_norm": 2.0497753620147705,
      "learning_rate": 8.226647966253906e-06,
      "loss": 0.036,
      "step": 5499
    },
    {
      "epoch": 0.2840175574490059,
      "grad_norm": 5.734914302825928,
      "learning_rate": 8.226022009832808e-06,
      "loss": 0.1173,
      "step": 5500
    },
    {
      "epoch": 0.28406919700490574,
      "grad_norm": 2.271738290786743,
      "learning_rate": 8.225395966779873e-06,
      "loss": 0.0285,
      "step": 5501
    },
    {
      "epoch": 0.28412083656080556,
      "grad_norm": 3.636528253555298,
      "learning_rate": 8.224769837111908e-06,
      "loss": 0.0648,
      "step": 5502
    },
    {
      "epoch": 0.2841724761167054,
      "grad_norm": 4.067110061645508,
      "learning_rate": 8.22414362084573e-06,
      "loss": 0.0661,
      "step": 5503
    },
    {
      "epoch": 0.2842241156726052,
      "grad_norm": 2.7937445640563965,
      "learning_rate": 8.223517317998156e-06,
      "loss": 0.0479,
      "step": 5504
    },
    {
      "epoch": 0.284275755228505,
      "grad_norm": 14.256572723388672,
      "learning_rate": 8.222890928586003e-06,
      "loss": 0.1211,
      "step": 5505
    },
    {
      "epoch": 0.28432739478440483,
      "grad_norm": 7.342391490936279,
      "learning_rate": 8.222264452626093e-06,
      "loss": 0.0958,
      "step": 5506
    },
    {
      "epoch": 0.28437903434030465,
      "grad_norm": 11.023009300231934,
      "learning_rate": 8.22163789013525e-06,
      "loss": 0.0694,
      "step": 5507
    },
    {
      "epoch": 0.28443067389620447,
      "grad_norm": 15.450300216674805,
      "learning_rate": 8.221011241130297e-06,
      "loss": 0.2505,
      "step": 5508
    },
    {
      "epoch": 0.2844823134521043,
      "grad_norm": 5.938093662261963,
      "learning_rate": 8.220384505628066e-06,
      "loss": 0.1129,
      "step": 5509
    },
    {
      "epoch": 0.2845339530080041,
      "grad_norm": 10.062484741210938,
      "learning_rate": 8.219757683645384e-06,
      "loss": 0.0374,
      "step": 5510
    },
    {
      "epoch": 0.284585592563904,
      "grad_norm": 3.873624324798584,
      "learning_rate": 8.219130775199084e-06,
      "loss": 0.0368,
      "step": 5511
    },
    {
      "epoch": 0.2846372321198038,
      "grad_norm": 3.6034693717956543,
      "learning_rate": 8.218503780306005e-06,
      "loss": 0.0786,
      "step": 5512
    },
    {
      "epoch": 0.2846888716757036,
      "grad_norm": 5.7940826416015625,
      "learning_rate": 8.217876698982981e-06,
      "loss": 0.0499,
      "step": 5513
    },
    {
      "epoch": 0.28474051123160343,
      "grad_norm": 3.4534270763397217,
      "learning_rate": 8.217249531246851e-06,
      "loss": 0.0539,
      "step": 5514
    },
    {
      "epoch": 0.28479215078750325,
      "grad_norm": 11.424901008605957,
      "learning_rate": 8.21662227711446e-06,
      "loss": 0.0883,
      "step": 5515
    },
    {
      "epoch": 0.28484379034340307,
      "grad_norm": 12.07349681854248,
      "learning_rate": 8.215994936602649e-06,
      "loss": 0.1083,
      "step": 5516
    },
    {
      "epoch": 0.2848954298993029,
      "grad_norm": 2.549276828765869,
      "learning_rate": 8.215367509728267e-06,
      "loss": 0.04,
      "step": 5517
    },
    {
      "epoch": 0.2849470694552027,
      "grad_norm": 6.099863529205322,
      "learning_rate": 8.214739996508161e-06,
      "loss": 0.0694,
      "step": 5518
    },
    {
      "epoch": 0.2849987090111025,
      "grad_norm": 3.639091968536377,
      "learning_rate": 8.214112396959186e-06,
      "loss": 0.0501,
      "step": 5519
    },
    {
      "epoch": 0.28505034856700234,
      "grad_norm": 10.071221351623535,
      "learning_rate": 8.213484711098193e-06,
      "loss": 0.1303,
      "step": 5520
    },
    {
      "epoch": 0.28510198812290216,
      "grad_norm": 12.08144474029541,
      "learning_rate": 8.212856938942036e-06,
      "loss": 0.1287,
      "step": 5521
    },
    {
      "epoch": 0.285153627678802,
      "grad_norm": 9.756168365478516,
      "learning_rate": 8.212229080507575e-06,
      "loss": 0.0897,
      "step": 5522
    },
    {
      "epoch": 0.2852052672347018,
      "grad_norm": 14.519383430480957,
      "learning_rate": 8.211601135811673e-06,
      "loss": 0.0795,
      "step": 5523
    },
    {
      "epoch": 0.2852569067906016,
      "grad_norm": 2.7754158973693848,
      "learning_rate": 8.210973104871189e-06,
      "loss": 0.0502,
      "step": 5524
    },
    {
      "epoch": 0.28530854634650143,
      "grad_norm": 2.748950242996216,
      "learning_rate": 8.210344987702992e-06,
      "loss": 0.0375,
      "step": 5525
    },
    {
      "epoch": 0.28536018590240125,
      "grad_norm": 12.411674499511719,
      "learning_rate": 8.209716784323946e-06,
      "loss": 0.3104,
      "step": 5526
    },
    {
      "epoch": 0.28541182545830107,
      "grad_norm": 2.6238350868225098,
      "learning_rate": 8.209088494750923e-06,
      "loss": 0.0497,
      "step": 5527
    },
    {
      "epoch": 0.2854634650142009,
      "grad_norm": 13.610759735107422,
      "learning_rate": 8.208460119000794e-06,
      "loss": 0.1054,
      "step": 5528
    },
    {
      "epoch": 0.2855151045701007,
      "grad_norm": 7.001957416534424,
      "learning_rate": 8.207831657090433e-06,
      "loss": 0.0786,
      "step": 5529
    },
    {
      "epoch": 0.2855667441260005,
      "grad_norm": 3.7297515869140625,
      "learning_rate": 8.207203109036718e-06,
      "loss": 0.0482,
      "step": 5530
    },
    {
      "epoch": 0.28561838368190035,
      "grad_norm": 7.224002361297607,
      "learning_rate": 8.206574474856527e-06,
      "loss": 0.0753,
      "step": 5531
    },
    {
      "epoch": 0.28567002323780016,
      "grad_norm": 23.12602996826172,
      "learning_rate": 8.205945754566745e-06,
      "loss": 0.2197,
      "step": 5532
    },
    {
      "epoch": 0.2857216627937,
      "grad_norm": 4.3874006271362305,
      "learning_rate": 8.205316948184251e-06,
      "loss": 0.0504,
      "step": 5533
    },
    {
      "epoch": 0.2857733023495998,
      "grad_norm": 19.522846221923828,
      "learning_rate": 8.204688055725933e-06,
      "loss": 0.3147,
      "step": 5534
    },
    {
      "epoch": 0.2858249419054996,
      "grad_norm": 4.742977142333984,
      "learning_rate": 8.204059077208679e-06,
      "loss": 0.0643,
      "step": 5535
    },
    {
      "epoch": 0.28587658146139944,
      "grad_norm": 5.42938232421875,
      "learning_rate": 8.203430012649379e-06,
      "loss": 0.0832,
      "step": 5536
    },
    {
      "epoch": 0.28592822101729926,
      "grad_norm": 5.134372234344482,
      "learning_rate": 8.202800862064926e-06,
      "loss": 0.0707,
      "step": 5537
    },
    {
      "epoch": 0.2859798605731991,
      "grad_norm": 6.073047161102295,
      "learning_rate": 8.202171625472218e-06,
      "loss": 0.0641,
      "step": 5538
    },
    {
      "epoch": 0.2860315001290989,
      "grad_norm": 4.581185340881348,
      "learning_rate": 8.201542302888151e-06,
      "loss": 0.0656,
      "step": 5539
    },
    {
      "epoch": 0.2860831396849987,
      "grad_norm": 15.574216842651367,
      "learning_rate": 8.200912894329623e-06,
      "loss": 0.2227,
      "step": 5540
    },
    {
      "epoch": 0.28613477924089853,
      "grad_norm": 9.973549842834473,
      "learning_rate": 8.200283399813538e-06,
      "loss": 0.0924,
      "step": 5541
    },
    {
      "epoch": 0.28618641879679835,
      "grad_norm": 3.35988712310791,
      "learning_rate": 8.199653819356798e-06,
      "loss": 0.046,
      "step": 5542
    },
    {
      "epoch": 0.28623805835269817,
      "grad_norm": 4.110933780670166,
      "learning_rate": 8.199024152976317e-06,
      "loss": 0.0739,
      "step": 5543
    },
    {
      "epoch": 0.286289697908598,
      "grad_norm": 5.333912372589111,
      "learning_rate": 8.198394400688995e-06,
      "loss": 0.0754,
      "step": 5544
    },
    {
      "epoch": 0.2863413374644978,
      "grad_norm": 4.588953971862793,
      "learning_rate": 8.197764562511747e-06,
      "loss": 0.0617,
      "step": 5545
    },
    {
      "epoch": 0.2863929770203976,
      "grad_norm": 7.364378929138184,
      "learning_rate": 8.197134638461488e-06,
      "loss": 0.0695,
      "step": 5546
    },
    {
      "epoch": 0.28644461657629744,
      "grad_norm": 1.744658350944519,
      "learning_rate": 8.196504628555133e-06,
      "loss": 0.0495,
      "step": 5547
    },
    {
      "epoch": 0.28649625613219726,
      "grad_norm": 13.833060264587402,
      "learning_rate": 8.195874532809602e-06,
      "loss": 0.2927,
      "step": 5548
    },
    {
      "epoch": 0.2865478956880971,
      "grad_norm": 3.803219795227051,
      "learning_rate": 8.195244351241813e-06,
      "loss": 0.0694,
      "step": 5549
    },
    {
      "epoch": 0.2865995352439969,
      "grad_norm": 3.3248181343078613,
      "learning_rate": 8.19461408386869e-06,
      "loss": 0.0411,
      "step": 5550
    },
    {
      "epoch": 0.2866511747998967,
      "grad_norm": 9.985294342041016,
      "learning_rate": 8.193983730707158e-06,
      "loss": 0.0587,
      "step": 5551
    },
    {
      "epoch": 0.28670281435579653,
      "grad_norm": 3.983675003051758,
      "learning_rate": 8.193353291774146e-06,
      "loss": 0.0758,
      "step": 5552
    },
    {
      "epoch": 0.28675445391169635,
      "grad_norm": 5.975755214691162,
      "learning_rate": 8.192722767086582e-06,
      "loss": 0.0824,
      "step": 5553
    },
    {
      "epoch": 0.28680609346759617,
      "grad_norm": 6.818210124969482,
      "learning_rate": 8.192092156661399e-06,
      "loss": 0.0844,
      "step": 5554
    },
    {
      "epoch": 0.286857733023496,
      "grad_norm": 4.2603302001953125,
      "learning_rate": 8.191461460515531e-06,
      "loss": 0.1207,
      "step": 5555
    },
    {
      "epoch": 0.2869093725793958,
      "grad_norm": 7.778621673583984,
      "learning_rate": 8.190830678665915e-06,
      "loss": 0.0856,
      "step": 5556
    },
    {
      "epoch": 0.2869610121352956,
      "grad_norm": 7.034149646759033,
      "learning_rate": 8.19019981112949e-06,
      "loss": 0.0961,
      "step": 5557
    },
    {
      "epoch": 0.28701265169119544,
      "grad_norm": 6.748521327972412,
      "learning_rate": 8.189568857923198e-06,
      "loss": 0.1456,
      "step": 5558
    },
    {
      "epoch": 0.28706429124709526,
      "grad_norm": 5.142654895782471,
      "learning_rate": 8.188937819063982e-06,
      "loss": 0.0841,
      "step": 5559
    },
    {
      "epoch": 0.2871159308029951,
      "grad_norm": 7.295263290405273,
      "learning_rate": 8.18830669456879e-06,
      "loss": 0.1258,
      "step": 5560
    },
    {
      "epoch": 0.2871675703588949,
      "grad_norm": 5.4516706466674805,
      "learning_rate": 8.187675484454566e-06,
      "loss": 0.0504,
      "step": 5561
    },
    {
      "epoch": 0.2872192099147947,
      "grad_norm": 5.905914783477783,
      "learning_rate": 8.187044188738265e-06,
      "loss": 0.0462,
      "step": 5562
    },
    {
      "epoch": 0.28727084947069453,
      "grad_norm": 9.58675479888916,
      "learning_rate": 8.186412807436837e-06,
      "loss": 0.159,
      "step": 5563
    },
    {
      "epoch": 0.28732248902659435,
      "grad_norm": 4.55053186416626,
      "learning_rate": 8.185781340567235e-06,
      "loss": 0.0657,
      "step": 5564
    },
    {
      "epoch": 0.28737412858249417,
      "grad_norm": 4.472623825073242,
      "learning_rate": 8.185149788146425e-06,
      "loss": 0.0789,
      "step": 5565
    },
    {
      "epoch": 0.287425768138394,
      "grad_norm": 6.841032028198242,
      "learning_rate": 8.184518150191358e-06,
      "loss": 0.1189,
      "step": 5566
    },
    {
      "epoch": 0.2874774076942938,
      "grad_norm": 2.819380521774292,
      "learning_rate": 8.183886426718998e-06,
      "loss": 0.0387,
      "step": 5567
    },
    {
      "epoch": 0.2875290472501936,
      "grad_norm": 7.092250823974609,
      "learning_rate": 8.183254617746313e-06,
      "loss": 0.1391,
      "step": 5568
    },
    {
      "epoch": 0.28758068680609344,
      "grad_norm": 3.879366159439087,
      "learning_rate": 8.182622723290267e-06,
      "loss": 0.0919,
      "step": 5569
    },
    {
      "epoch": 0.28763232636199326,
      "grad_norm": 3.9656729698181152,
      "learning_rate": 8.181990743367829e-06,
      "loss": 0.0761,
      "step": 5570
    },
    {
      "epoch": 0.2876839659178931,
      "grad_norm": 2.9729039669036865,
      "learning_rate": 8.18135867799597e-06,
      "loss": 0.0542,
      "step": 5571
    },
    {
      "epoch": 0.2877356054737929,
      "grad_norm": 10.655341148376465,
      "learning_rate": 8.180726527191665e-06,
      "loss": 0.1514,
      "step": 5572
    },
    {
      "epoch": 0.2877872450296927,
      "grad_norm": 2.476091146469116,
      "learning_rate": 8.180094290971889e-06,
      "loss": 0.04,
      "step": 5573
    },
    {
      "epoch": 0.2878388845855926,
      "grad_norm": 2.523513078689575,
      "learning_rate": 8.179461969353619e-06,
      "loss": 0.0578,
      "step": 5574
    },
    {
      "epoch": 0.2878905241414924,
      "grad_norm": 2.466587543487549,
      "learning_rate": 8.178829562353837e-06,
      "loss": 0.0298,
      "step": 5575
    },
    {
      "epoch": 0.28794216369739223,
      "grad_norm": 3.107210159301758,
      "learning_rate": 8.178197069989524e-06,
      "loss": 0.0617,
      "step": 5576
    },
    {
      "epoch": 0.28799380325329205,
      "grad_norm": 2.5362024307250977,
      "learning_rate": 8.177564492277666e-06,
      "loss": 0.051,
      "step": 5577
    },
    {
      "epoch": 0.28804544280919187,
      "grad_norm": 3.7555410861968994,
      "learning_rate": 8.176931829235251e-06,
      "loss": 0.0598,
      "step": 5578
    },
    {
      "epoch": 0.2880970823650917,
      "grad_norm": 5.6147236824035645,
      "learning_rate": 8.176299080879267e-06,
      "loss": 0.0688,
      "step": 5579
    },
    {
      "epoch": 0.2881487219209915,
      "grad_norm": 7.797418117523193,
      "learning_rate": 8.175666247226709e-06,
      "loss": 0.0511,
      "step": 5580
    },
    {
      "epoch": 0.2882003614768913,
      "grad_norm": 10.035651206970215,
      "learning_rate": 8.175033328294567e-06,
      "loss": 0.114,
      "step": 5581
    },
    {
      "epoch": 0.28825200103279114,
      "grad_norm": 4.232930660247803,
      "learning_rate": 8.17440032409984e-06,
      "loss": 0.1124,
      "step": 5582
    },
    {
      "epoch": 0.28830364058869096,
      "grad_norm": 2.8712563514709473,
      "learning_rate": 8.173767234659527e-06,
      "loss": 0.0463,
      "step": 5583
    },
    {
      "epoch": 0.2883552801445908,
      "grad_norm": 6.38267707824707,
      "learning_rate": 8.173134059990627e-06,
      "loss": 0.1002,
      "step": 5584
    },
    {
      "epoch": 0.2884069197004906,
      "grad_norm": 2.3695805072784424,
      "learning_rate": 8.172500800110144e-06,
      "loss": 0.0514,
      "step": 5585
    },
    {
      "epoch": 0.2884585592563904,
      "grad_norm": 6.559054851531982,
      "learning_rate": 8.171867455035085e-06,
      "loss": 0.0703,
      "step": 5586
    },
    {
      "epoch": 0.28851019881229023,
      "grad_norm": 8.052544593811035,
      "learning_rate": 8.171234024782458e-06,
      "loss": 0.0843,
      "step": 5587
    },
    {
      "epoch": 0.28856183836819005,
      "grad_norm": 3.6549384593963623,
      "learning_rate": 8.170600509369271e-06,
      "loss": 0.0536,
      "step": 5588
    },
    {
      "epoch": 0.28861347792408987,
      "grad_norm": 11.670805931091309,
      "learning_rate": 8.169966908812539e-06,
      "loss": 0.1029,
      "step": 5589
    },
    {
      "epoch": 0.2886651174799897,
      "grad_norm": 17.803234100341797,
      "learning_rate": 8.169333223129273e-06,
      "loss": 0.1131,
      "step": 5590
    },
    {
      "epoch": 0.2887167570358895,
      "grad_norm": 3.8422484397888184,
      "learning_rate": 8.168699452336494e-06,
      "loss": 0.0747,
      "step": 5591
    },
    {
      "epoch": 0.2887683965917893,
      "grad_norm": 4.024356842041016,
      "learning_rate": 8.16806559645122e-06,
      "loss": 0.0716,
      "step": 5592
    },
    {
      "epoch": 0.28882003614768914,
      "grad_norm": 7.309370994567871,
      "learning_rate": 8.167431655490472e-06,
      "loss": 0.1135,
      "step": 5593
    },
    {
      "epoch": 0.28887167570358896,
      "grad_norm": 8.320544242858887,
      "learning_rate": 8.166797629471274e-06,
      "loss": 0.0856,
      "step": 5594
    },
    {
      "epoch": 0.2889233152594888,
      "grad_norm": 5.466094970703125,
      "learning_rate": 8.166163518410652e-06,
      "loss": 0.0805,
      "step": 5595
    },
    {
      "epoch": 0.2889749548153886,
      "grad_norm": 3.051929235458374,
      "learning_rate": 8.165529322325638e-06,
      "loss": 0.0541,
      "step": 5596
    },
    {
      "epoch": 0.2890265943712884,
      "grad_norm": 2.429999351501465,
      "learning_rate": 8.164895041233256e-06,
      "loss": 0.0369,
      "step": 5597
    },
    {
      "epoch": 0.28907823392718823,
      "grad_norm": 6.609143257141113,
      "learning_rate": 8.164260675150545e-06,
      "loss": 0.0621,
      "step": 5598
    },
    {
      "epoch": 0.28912987348308805,
      "grad_norm": 12.144331932067871,
      "learning_rate": 8.163626224094537e-06,
      "loss": 0.0803,
      "step": 5599
    },
    {
      "epoch": 0.28918151303898787,
      "grad_norm": 2.48622465133667,
      "learning_rate": 8.162991688082271e-06,
      "loss": 0.0482,
      "step": 5600
    },
    {
      "epoch": 0.2892331525948877,
      "grad_norm": 16.498384475708008,
      "learning_rate": 8.162357067130785e-06,
      "loss": 0.1201,
      "step": 5601
    },
    {
      "epoch": 0.2892847921507875,
      "grad_norm": 1.8103721141815186,
      "learning_rate": 8.161722361257125e-06,
      "loss": 0.0284,
      "step": 5602
    },
    {
      "epoch": 0.2893364317066873,
      "grad_norm": 3.057323932647705,
      "learning_rate": 8.161087570478331e-06,
      "loss": 0.0582,
      "step": 5603
    },
    {
      "epoch": 0.28938807126258714,
      "grad_norm": 6.164488792419434,
      "learning_rate": 8.160452694811456e-06,
      "loss": 0.0759,
      "step": 5604
    },
    {
      "epoch": 0.28943971081848696,
      "grad_norm": 12.947260856628418,
      "learning_rate": 8.15981773427354e-06,
      "loss": 0.1472,
      "step": 5605
    },
    {
      "epoch": 0.2894913503743868,
      "grad_norm": 6.836422920227051,
      "learning_rate": 8.159182688881642e-06,
      "loss": 0.0643,
      "step": 5606
    },
    {
      "epoch": 0.2895429899302866,
      "grad_norm": 9.671298027038574,
      "learning_rate": 8.158547558652813e-06,
      "loss": 0.0662,
      "step": 5607
    },
    {
      "epoch": 0.2895946294861864,
      "grad_norm": 2.3874969482421875,
      "learning_rate": 8.157912343604107e-06,
      "loss": 0.0552,
      "step": 5608
    },
    {
      "epoch": 0.28964626904208624,
      "grad_norm": 1.9221370220184326,
      "learning_rate": 8.157277043752583e-06,
      "loss": 0.0439,
      "step": 5609
    },
    {
      "epoch": 0.28969790859798605,
      "grad_norm": 8.102151870727539,
      "learning_rate": 8.156641659115303e-06,
      "loss": 0.0936,
      "step": 5610
    },
    {
      "epoch": 0.28974954815388587,
      "grad_norm": 5.505202293395996,
      "learning_rate": 8.15600618970933e-06,
      "loss": 0.1102,
      "step": 5611
    },
    {
      "epoch": 0.2898011877097857,
      "grad_norm": 7.267271518707275,
      "learning_rate": 8.155370635551725e-06,
      "loss": 0.0802,
      "step": 5612
    },
    {
      "epoch": 0.2898528272656855,
      "grad_norm": 7.894371032714844,
      "learning_rate": 8.15473499665956e-06,
      "loss": 0.1104,
      "step": 5613
    },
    {
      "epoch": 0.2899044668215853,
      "grad_norm": 8.668791770935059,
      "learning_rate": 8.154099273049901e-06,
      "loss": 0.1356,
      "step": 5614
    },
    {
      "epoch": 0.28995610637748515,
      "grad_norm": 9.013887405395508,
      "learning_rate": 8.153463464739822e-06,
      "loss": 0.0843,
      "step": 5615
    },
    {
      "epoch": 0.29000774593338496,
      "grad_norm": 10.99766731262207,
      "learning_rate": 8.152827571746397e-06,
      "loss": 0.161,
      "step": 5616
    },
    {
      "epoch": 0.2900593854892848,
      "grad_norm": 4.582876682281494,
      "learning_rate": 8.1521915940867e-06,
      "loss": 0.0523,
      "step": 5617
    },
    {
      "epoch": 0.2901110250451846,
      "grad_norm": 3.4082775115966797,
      "learning_rate": 8.15155553177781e-06,
      "loss": 0.0324,
      "step": 5618
    },
    {
      "epoch": 0.2901626646010844,
      "grad_norm": 3.6385514736175537,
      "learning_rate": 8.150919384836811e-06,
      "loss": 0.0697,
      "step": 5619
    },
    {
      "epoch": 0.29021430415698424,
      "grad_norm": 7.828831195831299,
      "learning_rate": 8.150283153280785e-06,
      "loss": 0.0755,
      "step": 5620
    },
    {
      "epoch": 0.29026594371288406,
      "grad_norm": 4.6265482902526855,
      "learning_rate": 8.149646837126814e-06,
      "loss": 0.0817,
      "step": 5621
    },
    {
      "epoch": 0.2903175832687839,
      "grad_norm": 3.3941643238067627,
      "learning_rate": 8.14901043639199e-06,
      "loss": 0.0505,
      "step": 5622
    },
    {
      "epoch": 0.2903692228246837,
      "grad_norm": 3.4161581993103027,
      "learning_rate": 8.148373951093399e-06,
      "loss": 0.0447,
      "step": 5623
    },
    {
      "epoch": 0.2904208623805835,
      "grad_norm": 2.1058743000030518,
      "learning_rate": 8.147737381248138e-06,
      "loss": 0.0674,
      "step": 5624
    },
    {
      "epoch": 0.29047250193648333,
      "grad_norm": 7.052057266235352,
      "learning_rate": 8.147100726873296e-06,
      "loss": 0.0895,
      "step": 5625
    },
    {
      "epoch": 0.29052414149238315,
      "grad_norm": 6.90155029296875,
      "learning_rate": 8.146463987985975e-06,
      "loss": 0.057,
      "step": 5626
    },
    {
      "epoch": 0.29057578104828297,
      "grad_norm": 3.1240508556365967,
      "learning_rate": 8.14582716460327e-06,
      "loss": 0.0675,
      "step": 5627
    },
    {
      "epoch": 0.2906274206041828,
      "grad_norm": 2.553377151489258,
      "learning_rate": 8.145190256742285e-06,
      "loss": 0.0414,
      "step": 5628
    },
    {
      "epoch": 0.2906790601600826,
      "grad_norm": 11.7262544631958,
      "learning_rate": 8.144553264420121e-06,
      "loss": 0.1061,
      "step": 5629
    },
    {
      "epoch": 0.2907306997159824,
      "grad_norm": 2.5890188217163086,
      "learning_rate": 8.143916187653887e-06,
      "loss": 0.0479,
      "step": 5630
    },
    {
      "epoch": 0.29078233927188224,
      "grad_norm": 6.4668121337890625,
      "learning_rate": 8.143279026460689e-06,
      "loss": 0.0752,
      "step": 5631
    },
    {
      "epoch": 0.29083397882778206,
      "grad_norm": 2.991042137145996,
      "learning_rate": 8.142641780857639e-06,
      "loss": 0.0617,
      "step": 5632
    },
    {
      "epoch": 0.2908856183836819,
      "grad_norm": 3.8972702026367188,
      "learning_rate": 8.142004450861847e-06,
      "loss": 0.0804,
      "step": 5633
    },
    {
      "epoch": 0.2909372579395817,
      "grad_norm": 4.355978012084961,
      "learning_rate": 8.141367036490428e-06,
      "loss": 0.072,
      "step": 5634
    },
    {
      "epoch": 0.2909888974954815,
      "grad_norm": 6.435050010681152,
      "learning_rate": 8.140729537760504e-06,
      "loss": 0.0691,
      "step": 5635
    },
    {
      "epoch": 0.29104053705138133,
      "grad_norm": 5.447753429412842,
      "learning_rate": 8.140091954689189e-06,
      "loss": 0.0957,
      "step": 5636
    },
    {
      "epoch": 0.29109217660728115,
      "grad_norm": 6.459087371826172,
      "learning_rate": 8.139454287293606e-06,
      "loss": 0.1296,
      "step": 5637
    },
    {
      "epoch": 0.291143816163181,
      "grad_norm": 3.1703450679779053,
      "learning_rate": 8.13881653559088e-06,
      "loss": 0.055,
      "step": 5638
    },
    {
      "epoch": 0.29119545571908084,
      "grad_norm": 9.36401653289795,
      "learning_rate": 8.13817869959814e-06,
      "loss": 0.0856,
      "step": 5639
    },
    {
      "epoch": 0.29124709527498066,
      "grad_norm": 6.022827625274658,
      "learning_rate": 8.13754077933251e-06,
      "loss": 0.0576,
      "step": 5640
    },
    {
      "epoch": 0.2912987348308805,
      "grad_norm": 7.740255832672119,
      "learning_rate": 8.136902774811119e-06,
      "loss": 0.0678,
      "step": 5641
    },
    {
      "epoch": 0.2913503743867803,
      "grad_norm": 7.864124298095703,
      "learning_rate": 8.136264686051105e-06,
      "loss": 0.0728,
      "step": 5642
    },
    {
      "epoch": 0.2914020139426801,
      "grad_norm": 5.992852210998535,
      "learning_rate": 8.135626513069603e-06,
      "loss": 0.0447,
      "step": 5643
    },
    {
      "epoch": 0.29145365349857993,
      "grad_norm": 11.134273529052734,
      "learning_rate": 8.134988255883747e-06,
      "loss": 0.0702,
      "step": 5644
    },
    {
      "epoch": 0.29150529305447975,
      "grad_norm": 7.236660003662109,
      "learning_rate": 8.13434991451068e-06,
      "loss": 0.1283,
      "step": 5645
    },
    {
      "epoch": 0.29155693261037957,
      "grad_norm": 6.761788845062256,
      "learning_rate": 8.133711488967543e-06,
      "loss": 0.1091,
      "step": 5646
    },
    {
      "epoch": 0.2916085721662794,
      "grad_norm": 3.5197556018829346,
      "learning_rate": 8.133072979271479e-06,
      "loss": 0.0679,
      "step": 5647
    },
    {
      "epoch": 0.2916602117221792,
      "grad_norm": 7.43234395980835,
      "learning_rate": 8.132434385439637e-06,
      "loss": 0.0451,
      "step": 5648
    },
    {
      "epoch": 0.291711851278079,
      "grad_norm": 13.984541893005371,
      "learning_rate": 8.131795707489165e-06,
      "loss": 0.1871,
      "step": 5649
    },
    {
      "epoch": 0.29176349083397884,
      "grad_norm": 3.0464701652526855,
      "learning_rate": 8.131156945437212e-06,
      "loss": 0.0531,
      "step": 5650
    },
    {
      "epoch": 0.29181513038987866,
      "grad_norm": 2.5517048835754395,
      "learning_rate": 8.130518099300934e-06,
      "loss": 0.0473,
      "step": 5651
    },
    {
      "epoch": 0.2918667699457785,
      "grad_norm": 2.580277442932129,
      "learning_rate": 8.129879169097486e-06,
      "loss": 0.0355,
      "step": 5652
    },
    {
      "epoch": 0.2919184095016783,
      "grad_norm": 6.9486470222473145,
      "learning_rate": 8.129240154844026e-06,
      "loss": 0.0852,
      "step": 5653
    },
    {
      "epoch": 0.2919700490575781,
      "grad_norm": 5.930320739746094,
      "learning_rate": 8.128601056557713e-06,
      "loss": 0.0577,
      "step": 5654
    },
    {
      "epoch": 0.29202168861347794,
      "grad_norm": 3.2149405479431152,
      "learning_rate": 8.12796187425571e-06,
      "loss": 0.0519,
      "step": 5655
    },
    {
      "epoch": 0.29207332816937775,
      "grad_norm": 6.31633186340332,
      "learning_rate": 8.127322607955182e-06,
      "loss": 0.067,
      "step": 5656
    },
    {
      "epoch": 0.2921249677252776,
      "grad_norm": 10.23898696899414,
      "learning_rate": 8.126683257673296e-06,
      "loss": 0.1158,
      "step": 5657
    },
    {
      "epoch": 0.2921766072811774,
      "grad_norm": 3.766401529312134,
      "learning_rate": 8.12604382342722e-06,
      "loss": 0.0735,
      "step": 5658
    },
    {
      "epoch": 0.2922282468370772,
      "grad_norm": 7.597470283508301,
      "learning_rate": 8.125404305234128e-06,
      "loss": 0.0769,
      "step": 5659
    },
    {
      "epoch": 0.29227988639297703,
      "grad_norm": 9.480586051940918,
      "learning_rate": 8.124764703111193e-06,
      "loss": 0.0816,
      "step": 5660
    },
    {
      "epoch": 0.29233152594887685,
      "grad_norm": 3.6318721771240234,
      "learning_rate": 8.124125017075585e-06,
      "loss": 0.0571,
      "step": 5661
    },
    {
      "epoch": 0.29238316550477667,
      "grad_norm": 8.388504981994629,
      "learning_rate": 8.123485247144492e-06,
      "loss": 0.0569,
      "step": 5662
    },
    {
      "epoch": 0.2924348050606765,
      "grad_norm": 4.2364373207092285,
      "learning_rate": 8.122845393335086e-06,
      "loss": 0.0423,
      "step": 5663
    },
    {
      "epoch": 0.2924864446165763,
      "grad_norm": 2.2955267429351807,
      "learning_rate": 8.122205455664556e-06,
      "loss": 0.0204,
      "step": 5664
    },
    {
      "epoch": 0.2925380841724761,
      "grad_norm": 4.355836868286133,
      "learning_rate": 8.121565434150082e-06,
      "loss": 0.0482,
      "step": 5665
    },
    {
      "epoch": 0.29258972372837594,
      "grad_norm": 3.7635557651519775,
      "learning_rate": 8.120925328808855e-06,
      "loss": 0.0614,
      "step": 5666
    },
    {
      "epoch": 0.29264136328427576,
      "grad_norm": 4.631556034088135,
      "learning_rate": 8.120285139658062e-06,
      "loss": 0.0753,
      "step": 5667
    },
    {
      "epoch": 0.2926930028401756,
      "grad_norm": 3.5269105434417725,
      "learning_rate": 8.119644866714896e-06,
      "loss": 0.0711,
      "step": 5668
    },
    {
      "epoch": 0.2927446423960754,
      "grad_norm": 29.106082916259766,
      "learning_rate": 8.11900450999655e-06,
      "loss": 0.0871,
      "step": 5669
    },
    {
      "epoch": 0.2927962819519752,
      "grad_norm": 7.013868808746338,
      "learning_rate": 8.118364069520219e-06,
      "loss": 0.1071,
      "step": 5670
    },
    {
      "epoch": 0.29284792150787503,
      "grad_norm": 7.5179314613342285,
      "learning_rate": 8.117723545303104e-06,
      "loss": 0.0465,
      "step": 5671
    },
    {
      "epoch": 0.29289956106377485,
      "grad_norm": 5.989478588104248,
      "learning_rate": 8.117082937362404e-06,
      "loss": 0.0594,
      "step": 5672
    },
    {
      "epoch": 0.29295120061967467,
      "grad_norm": 3.296719551086426,
      "learning_rate": 8.116442245715324e-06,
      "loss": 0.0563,
      "step": 5673
    },
    {
      "epoch": 0.2930028401755745,
      "grad_norm": 4.529387474060059,
      "learning_rate": 8.115801470379068e-06,
      "loss": 0.0684,
      "step": 5674
    },
    {
      "epoch": 0.2930544797314743,
      "grad_norm": 22.191499710083008,
      "learning_rate": 8.115160611370843e-06,
      "loss": 0.2304,
      "step": 5675
    },
    {
      "epoch": 0.2931061192873741,
      "grad_norm": 11.087569236755371,
      "learning_rate": 8.114519668707857e-06,
      "loss": 0.0457,
      "step": 5676
    },
    {
      "epoch": 0.29315775884327394,
      "grad_norm": 7.066823482513428,
      "learning_rate": 8.113878642407324e-06,
      "loss": 0.0838,
      "step": 5677
    },
    {
      "epoch": 0.29320939839917376,
      "grad_norm": 9.648849487304688,
      "learning_rate": 8.11323753248646e-06,
      "loss": 0.1559,
      "step": 5678
    },
    {
      "epoch": 0.2932610379550736,
      "grad_norm": 5.203167915344238,
      "learning_rate": 8.11259633896248e-06,
      "loss": 0.0877,
      "step": 5679
    },
    {
      "epoch": 0.2933126775109734,
      "grad_norm": 11.095516204833984,
      "learning_rate": 8.1119550618526e-06,
      "loss": 0.1039,
      "step": 5680
    },
    {
      "epoch": 0.2933643170668732,
      "grad_norm": 2.540308713912964,
      "learning_rate": 8.111313701174043e-06,
      "loss": 0.0396,
      "step": 5681
    },
    {
      "epoch": 0.29341595662277303,
      "grad_norm": 3.034919261932373,
      "learning_rate": 8.110672256944033e-06,
      "loss": 0.0583,
      "step": 5682
    },
    {
      "epoch": 0.29346759617867285,
      "grad_norm": 3.020627737045288,
      "learning_rate": 8.110030729179792e-06,
      "loss": 0.0759,
      "step": 5683
    },
    {
      "epoch": 0.29351923573457267,
      "grad_norm": 8.10546875,
      "learning_rate": 8.109389117898554e-06,
      "loss": 0.1255,
      "step": 5684
    },
    {
      "epoch": 0.2935708752904725,
      "grad_norm": 8.06692123413086,
      "learning_rate": 8.108747423117544e-06,
      "loss": 0.1059,
      "step": 5685
    },
    {
      "epoch": 0.2936225148463723,
      "grad_norm": 5.8836565017700195,
      "learning_rate": 8.108105644853994e-06,
      "loss": 0.1095,
      "step": 5686
    },
    {
      "epoch": 0.2936741544022721,
      "grad_norm": 1.7429578304290771,
      "learning_rate": 8.107463783125138e-06,
      "loss": 0.0349,
      "step": 5687
    },
    {
      "epoch": 0.29372579395817194,
      "grad_norm": 4.0940093994140625,
      "learning_rate": 8.106821837948218e-06,
      "loss": 0.0562,
      "step": 5688
    },
    {
      "epoch": 0.29377743351407176,
      "grad_norm": 8.17209529876709,
      "learning_rate": 8.106179809340467e-06,
      "loss": 0.0663,
      "step": 5689
    },
    {
      "epoch": 0.2938290730699716,
      "grad_norm": 4.356247901916504,
      "learning_rate": 8.105537697319129e-06,
      "loss": 0.0645,
      "step": 5690
    },
    {
      "epoch": 0.2938807126258714,
      "grad_norm": 5.338048934936523,
      "learning_rate": 8.104895501901444e-06,
      "loss": 0.0769,
      "step": 5691
    },
    {
      "epoch": 0.2939323521817712,
      "grad_norm": 7.1543660163879395,
      "learning_rate": 8.104253223104661e-06,
      "loss": 0.1469,
      "step": 5692
    },
    {
      "epoch": 0.29398399173767104,
      "grad_norm": 3.1634345054626465,
      "learning_rate": 8.103610860946028e-06,
      "loss": 0.0518,
      "step": 5693
    },
    {
      "epoch": 0.29403563129357085,
      "grad_norm": 2.37532639503479,
      "learning_rate": 8.102968415442792e-06,
      "loss": 0.032,
      "step": 5694
    },
    {
      "epoch": 0.29408727084947067,
      "grad_norm": 4.931708335876465,
      "learning_rate": 8.102325886612207e-06,
      "loss": 0.0612,
      "step": 5695
    },
    {
      "epoch": 0.2941389104053705,
      "grad_norm": 8.176325798034668,
      "learning_rate": 8.101683274471529e-06,
      "loss": 0.074,
      "step": 5696
    },
    {
      "epoch": 0.2941905499612703,
      "grad_norm": 14.754752159118652,
      "learning_rate": 8.101040579038013e-06,
      "loss": 0.1547,
      "step": 5697
    },
    {
      "epoch": 0.2942421895171701,
      "grad_norm": 6.853609561920166,
      "learning_rate": 8.100397800328919e-06,
      "loss": 0.1241,
      "step": 5698
    },
    {
      "epoch": 0.29429382907306995,
      "grad_norm": 8.716156005859375,
      "learning_rate": 8.099754938361506e-06,
      "loss": 0.0753,
      "step": 5699
    },
    {
      "epoch": 0.29434546862896976,
      "grad_norm": 3.44761323928833,
      "learning_rate": 8.09911199315304e-06,
      "loss": 0.1304,
      "step": 5700
    },
    {
      "epoch": 0.2943971081848696,
      "grad_norm": 5.726634502410889,
      "learning_rate": 8.098468964720785e-06,
      "loss": 0.0947,
      "step": 5701
    },
    {
      "epoch": 0.29444874774076946,
      "grad_norm": 10.716598510742188,
      "learning_rate": 8.09782585308201e-06,
      "loss": 0.128,
      "step": 5702
    },
    {
      "epoch": 0.2945003872966693,
      "grad_norm": 3.379020929336548,
      "learning_rate": 8.097182658253984e-06,
      "loss": 0.0633,
      "step": 5703
    },
    {
      "epoch": 0.2945520268525691,
      "grad_norm": 1.9173991680145264,
      "learning_rate": 8.09653938025398e-06,
      "loss": 0.0686,
      "step": 5704
    },
    {
      "epoch": 0.2946036664084689,
      "grad_norm": 1.9108940362930298,
      "learning_rate": 8.095896019099274e-06,
      "loss": 0.0419,
      "step": 5705
    },
    {
      "epoch": 0.29465530596436873,
      "grad_norm": 6.328918933868408,
      "learning_rate": 8.095252574807141e-06,
      "loss": 0.1314,
      "step": 5706
    },
    {
      "epoch": 0.29470694552026855,
      "grad_norm": 3.19047474861145,
      "learning_rate": 8.094609047394863e-06,
      "loss": 0.0654,
      "step": 5707
    },
    {
      "epoch": 0.29475858507616837,
      "grad_norm": 2.433237314224243,
      "learning_rate": 8.093965436879718e-06,
      "loss": 0.0291,
      "step": 5708
    },
    {
      "epoch": 0.2948102246320682,
      "grad_norm": 3.2595760822296143,
      "learning_rate": 8.093321743278987e-06,
      "loss": 0.0651,
      "step": 5709
    },
    {
      "epoch": 0.294861864187968,
      "grad_norm": 5.7248125076293945,
      "learning_rate": 8.092677966609966e-06,
      "loss": 0.0562,
      "step": 5710
    },
    {
      "epoch": 0.2949135037438678,
      "grad_norm": 4.0350847244262695,
      "learning_rate": 8.092034106889931e-06,
      "loss": 0.0578,
      "step": 5711
    },
    {
      "epoch": 0.29496514329976764,
      "grad_norm": 5.7001237869262695,
      "learning_rate": 8.091390164136182e-06,
      "loss": 0.0768,
      "step": 5712
    },
    {
      "epoch": 0.29501678285566746,
      "grad_norm": 5.726407051086426,
      "learning_rate": 8.090746138366003e-06,
      "loss": 0.0456,
      "step": 5713
    },
    {
      "epoch": 0.2950684224115673,
      "grad_norm": 3.332094430923462,
      "learning_rate": 8.090102029596696e-06,
      "loss": 0.055,
      "step": 5714
    },
    {
      "epoch": 0.2951200619674671,
      "grad_norm": 3.6236414909362793,
      "learning_rate": 8.089457837845555e-06,
      "loss": 0.0821,
      "step": 5715
    },
    {
      "epoch": 0.2951717015233669,
      "grad_norm": 6.330197334289551,
      "learning_rate": 8.088813563129879e-06,
      "loss": 0.1266,
      "step": 5716
    },
    {
      "epoch": 0.29522334107926673,
      "grad_norm": 16.225467681884766,
      "learning_rate": 8.088169205466967e-06,
      "loss": 0.1049,
      "step": 5717
    },
    {
      "epoch": 0.29527498063516655,
      "grad_norm": 4.473453521728516,
      "learning_rate": 8.087524764874128e-06,
      "loss": 0.0557,
      "step": 5718
    },
    {
      "epoch": 0.29532662019106637,
      "grad_norm": 4.9692702293396,
      "learning_rate": 8.086880241368665e-06,
      "loss": 0.0632,
      "step": 5719
    },
    {
      "epoch": 0.2953782597469662,
      "grad_norm": 3.6219332218170166,
      "learning_rate": 8.086235634967884e-06,
      "loss": 0.0365,
      "step": 5720
    },
    {
      "epoch": 0.295429899302866,
      "grad_norm": 5.547406196594238,
      "learning_rate": 8.0855909456891e-06,
      "loss": 0.0733,
      "step": 5721
    },
    {
      "epoch": 0.2954815388587658,
      "grad_norm": 5.439542770385742,
      "learning_rate": 8.08494617354962e-06,
      "loss": 0.0823,
      "step": 5722
    },
    {
      "epoch": 0.29553317841466564,
      "grad_norm": 7.64113712310791,
      "learning_rate": 8.084301318566763e-06,
      "loss": 0.1348,
      "step": 5723
    },
    {
      "epoch": 0.29558481797056546,
      "grad_norm": 7.167909145355225,
      "learning_rate": 8.083656380757845e-06,
      "loss": 0.0511,
      "step": 5724
    },
    {
      "epoch": 0.2956364575264653,
      "grad_norm": 2.92012095451355,
      "learning_rate": 8.083011360140185e-06,
      "loss": 0.0495,
      "step": 5725
    },
    {
      "epoch": 0.2956880970823651,
      "grad_norm": 4.617087364196777,
      "learning_rate": 8.082366256731103e-06,
      "loss": 0.085,
      "step": 5726
    },
    {
      "epoch": 0.2957397366382649,
      "grad_norm": 15.523223876953125,
      "learning_rate": 8.081721070547926e-06,
      "loss": 0.1072,
      "step": 5727
    },
    {
      "epoch": 0.29579137619416473,
      "grad_norm": 5.068345069885254,
      "learning_rate": 8.081075801607976e-06,
      "loss": 0.0794,
      "step": 5728
    },
    {
      "epoch": 0.29584301575006455,
      "grad_norm": 2.2628114223480225,
      "learning_rate": 8.080430449928584e-06,
      "loss": 0.015,
      "step": 5729
    },
    {
      "epoch": 0.29589465530596437,
      "grad_norm": 8.179887771606445,
      "learning_rate": 8.07978501552708e-06,
      "loss": 0.1083,
      "step": 5730
    },
    {
      "epoch": 0.2959462948618642,
      "grad_norm": 8.214402198791504,
      "learning_rate": 8.079139498420793e-06,
      "loss": 0.1488,
      "step": 5731
    },
    {
      "epoch": 0.295997934417764,
      "grad_norm": 5.164076328277588,
      "learning_rate": 8.078493898627064e-06,
      "loss": 0.0548,
      "step": 5732
    },
    {
      "epoch": 0.2960495739736638,
      "grad_norm": 4.1147284507751465,
      "learning_rate": 8.077848216163224e-06,
      "loss": 0.0876,
      "step": 5733
    },
    {
      "epoch": 0.29610121352956364,
      "grad_norm": 3.9892125129699707,
      "learning_rate": 8.077202451046618e-06,
      "loss": 0.0598,
      "step": 5734
    },
    {
      "epoch": 0.29615285308546346,
      "grad_norm": 8.845123291015625,
      "learning_rate": 8.076556603294581e-06,
      "loss": 0.0725,
      "step": 5735
    },
    {
      "epoch": 0.2962044926413633,
      "grad_norm": 5.38938045501709,
      "learning_rate": 8.07591067292446e-06,
      "loss": 0.129,
      "step": 5736
    },
    {
      "epoch": 0.2962561321972631,
      "grad_norm": 3.7946834564208984,
      "learning_rate": 8.075264659953603e-06,
      "loss": 0.0445,
      "step": 5737
    },
    {
      "epoch": 0.2963077717531629,
      "grad_norm": 2.0440475940704346,
      "learning_rate": 8.074618564399355e-06,
      "loss": 0.036,
      "step": 5738
    },
    {
      "epoch": 0.29635941130906274,
      "grad_norm": 2.7120838165283203,
      "learning_rate": 8.073972386279065e-06,
      "loss": 0.0847,
      "step": 5739
    },
    {
      "epoch": 0.29641105086496256,
      "grad_norm": 3.9127743244171143,
      "learning_rate": 8.073326125610089e-06,
      "loss": 0.0609,
      "step": 5740
    },
    {
      "epoch": 0.2964626904208624,
      "grad_norm": 3.421027421951294,
      "learning_rate": 8.07267978240978e-06,
      "loss": 0.0534,
      "step": 5741
    },
    {
      "epoch": 0.2965143299767622,
      "grad_norm": 4.038944244384766,
      "learning_rate": 8.072033356695496e-06,
      "loss": 0.0552,
      "step": 5742
    },
    {
      "epoch": 0.296565969532662,
      "grad_norm": 4.682583808898926,
      "learning_rate": 8.071386848484596e-06,
      "loss": 0.0763,
      "step": 5743
    },
    {
      "epoch": 0.29661760908856183,
      "grad_norm": 3.400482177734375,
      "learning_rate": 8.07074025779444e-06,
      "loss": 0.0543,
      "step": 5744
    },
    {
      "epoch": 0.29666924864446165,
      "grad_norm": 6.03399133682251,
      "learning_rate": 8.070093584642393e-06,
      "loss": 0.0868,
      "step": 5745
    },
    {
      "epoch": 0.29672088820036147,
      "grad_norm": 8.950239181518555,
      "learning_rate": 8.06944682904582e-06,
      "loss": 0.1072,
      "step": 5746
    },
    {
      "epoch": 0.2967725277562613,
      "grad_norm": 3.538388967514038,
      "learning_rate": 8.068799991022087e-06,
      "loss": 0.0457,
      "step": 5747
    },
    {
      "epoch": 0.2968241673121611,
      "grad_norm": 6.459550857543945,
      "learning_rate": 8.06815307058857e-06,
      "loss": 0.0893,
      "step": 5748
    },
    {
      "epoch": 0.2968758068680609,
      "grad_norm": 16.724849700927734,
      "learning_rate": 8.067506067762636e-06,
      "loss": 0.0644,
      "step": 5749
    },
    {
      "epoch": 0.29692744642396074,
      "grad_norm": 2.4045374393463135,
      "learning_rate": 8.066858982561664e-06,
      "loss": 0.0546,
      "step": 5750
    },
    {
      "epoch": 0.29697908597986056,
      "grad_norm": 13.60914134979248,
      "learning_rate": 8.066211815003024e-06,
      "loss": 0.1633,
      "step": 5751
    },
    {
      "epoch": 0.2970307255357604,
      "grad_norm": 5.76535701751709,
      "learning_rate": 8.065564565104103e-06,
      "loss": 0.0551,
      "step": 5752
    },
    {
      "epoch": 0.2970823650916602,
      "grad_norm": 2.1206393241882324,
      "learning_rate": 8.06491723288228e-06,
      "loss": 0.0239,
      "step": 5753
    },
    {
      "epoch": 0.29713400464756,
      "grad_norm": 4.847890853881836,
      "learning_rate": 8.064269818354934e-06,
      "loss": 0.0728,
      "step": 5754
    },
    {
      "epoch": 0.29718564420345983,
      "grad_norm": 7.802407264709473,
      "learning_rate": 8.063622321539455e-06,
      "loss": 0.1543,
      "step": 5755
    },
    {
      "epoch": 0.29723728375935965,
      "grad_norm": 4.574452877044678,
      "learning_rate": 8.062974742453231e-06,
      "loss": 0.0852,
      "step": 5756
    },
    {
      "epoch": 0.29728892331525947,
      "grad_norm": 14.103424072265625,
      "learning_rate": 8.062327081113652e-06,
      "loss": 0.2608,
      "step": 5757
    },
    {
      "epoch": 0.2973405628711593,
      "grad_norm": 11.43088436126709,
      "learning_rate": 8.061679337538107e-06,
      "loss": 0.1287,
      "step": 5758
    },
    {
      "epoch": 0.2973922024270591,
      "grad_norm": 6.5772294998168945,
      "learning_rate": 8.061031511743994e-06,
      "loss": 0.0669,
      "step": 5759
    },
    {
      "epoch": 0.2974438419829589,
      "grad_norm": 4.094453811645508,
      "learning_rate": 8.06038360374871e-06,
      "loss": 0.0683,
      "step": 5760
    },
    {
      "epoch": 0.29749548153885874,
      "grad_norm": 4.257543563842773,
      "learning_rate": 8.059735613569653e-06,
      "loss": 0.0789,
      "step": 5761
    },
    {
      "epoch": 0.29754712109475856,
      "grad_norm": 2.5423877239227295,
      "learning_rate": 8.05908754122422e-06,
      "loss": 0.0496,
      "step": 5762
    },
    {
      "epoch": 0.2975987606506584,
      "grad_norm": 2.3544790744781494,
      "learning_rate": 8.058439386729823e-06,
      "loss": 0.0372,
      "step": 5763
    },
    {
      "epoch": 0.2976504002065582,
      "grad_norm": 4.513719081878662,
      "learning_rate": 8.05779115010386e-06,
      "loss": 0.0876,
      "step": 5764
    },
    {
      "epoch": 0.29770203976245807,
      "grad_norm": 8.497767448425293,
      "learning_rate": 8.057142831363745e-06,
      "loss": 0.0946,
      "step": 5765
    },
    {
      "epoch": 0.2977536793183579,
      "grad_norm": 3.9537353515625,
      "learning_rate": 8.056494430526883e-06,
      "loss": 0.0955,
      "step": 5766
    },
    {
      "epoch": 0.2978053188742577,
      "grad_norm": 6.215909957885742,
      "learning_rate": 8.055845947610686e-06,
      "loss": 0.1234,
      "step": 5767
    },
    {
      "epoch": 0.2978569584301575,
      "grad_norm": 11.979781150817871,
      "learning_rate": 8.055197382632574e-06,
      "loss": 0.0553,
      "step": 5768
    },
    {
      "epoch": 0.29790859798605734,
      "grad_norm": 2.0501456260681152,
      "learning_rate": 8.054548735609958e-06,
      "loss": 0.0546,
      "step": 5769
    },
    {
      "epoch": 0.29796023754195716,
      "grad_norm": 7.172950267791748,
      "learning_rate": 8.053900006560258e-06,
      "loss": 0.0654,
      "step": 5770
    },
    {
      "epoch": 0.298011877097857,
      "grad_norm": 7.072928428649902,
      "learning_rate": 8.053251195500895e-06,
      "loss": 0.0427,
      "step": 5771
    },
    {
      "epoch": 0.2980635166537568,
      "grad_norm": 2.697611093521118,
      "learning_rate": 8.052602302449296e-06,
      "loss": 0.03,
      "step": 5772
    },
    {
      "epoch": 0.2981151562096566,
      "grad_norm": 3.1674437522888184,
      "learning_rate": 8.051953327422883e-06,
      "loss": 0.0384,
      "step": 5773
    },
    {
      "epoch": 0.29816679576555644,
      "grad_norm": 3.9244658946990967,
      "learning_rate": 8.051304270439081e-06,
      "loss": 0.031,
      "step": 5774
    },
    {
      "epoch": 0.29821843532145625,
      "grad_norm": 2.879307746887207,
      "learning_rate": 8.050655131515325e-06,
      "loss": 0.0597,
      "step": 5775
    },
    {
      "epoch": 0.2982700748773561,
      "grad_norm": 6.826414585113525,
      "learning_rate": 8.050005910669048e-06,
      "loss": 0.0968,
      "step": 5776
    },
    {
      "epoch": 0.2983217144332559,
      "grad_norm": 3.975558280944824,
      "learning_rate": 8.049356607917678e-06,
      "loss": 0.0786,
      "step": 5777
    },
    {
      "epoch": 0.2983733539891557,
      "grad_norm": 6.008717060089111,
      "learning_rate": 8.048707223278653e-06,
      "loss": 0.053,
      "step": 5778
    },
    {
      "epoch": 0.29842499354505553,
      "grad_norm": 9.524571418762207,
      "learning_rate": 8.048057756769416e-06,
      "loss": 0.1974,
      "step": 5779
    },
    {
      "epoch": 0.29847663310095535,
      "grad_norm": 20.70040512084961,
      "learning_rate": 8.047408208407404e-06,
      "loss": 0.1943,
      "step": 5780
    },
    {
      "epoch": 0.29852827265685516,
      "grad_norm": 4.789031505584717,
      "learning_rate": 8.046758578210061e-06,
      "loss": 0.084,
      "step": 5781
    },
    {
      "epoch": 0.298579912212755,
      "grad_norm": 5.602688312530518,
      "learning_rate": 8.046108866194833e-06,
      "loss": 0.0123,
      "step": 5782
    },
    {
      "epoch": 0.2986315517686548,
      "grad_norm": 10.407672882080078,
      "learning_rate": 8.045459072379168e-06,
      "loss": 0.1414,
      "step": 5783
    },
    {
      "epoch": 0.2986831913245546,
      "grad_norm": 3.298405170440674,
      "learning_rate": 8.04480919678051e-06,
      "loss": 0.0204,
      "step": 5784
    },
    {
      "epoch": 0.29873483088045444,
      "grad_norm": 3.4526562690734863,
      "learning_rate": 8.044159239416318e-06,
      "loss": 0.0524,
      "step": 5785
    },
    {
      "epoch": 0.29878647043635426,
      "grad_norm": 5.355932712554932,
      "learning_rate": 8.043509200304042e-06,
      "loss": 0.0346,
      "step": 5786
    },
    {
      "epoch": 0.2988381099922541,
      "grad_norm": 8.403310775756836,
      "learning_rate": 8.04285907946114e-06,
      "loss": 0.1042,
      "step": 5787
    },
    {
      "epoch": 0.2988897495481539,
      "grad_norm": 4.194834232330322,
      "learning_rate": 8.042208876905071e-06,
      "loss": 0.0602,
      "step": 5788
    },
    {
      "epoch": 0.2989413891040537,
      "grad_norm": 6.069411277770996,
      "learning_rate": 8.041558592653293e-06,
      "loss": 0.0685,
      "step": 5789
    },
    {
      "epoch": 0.29899302865995353,
      "grad_norm": 11.9500732421875,
      "learning_rate": 8.040908226723269e-06,
      "loss": 0.0878,
      "step": 5790
    },
    {
      "epoch": 0.29904466821585335,
      "grad_norm": 25.129180908203125,
      "learning_rate": 8.040257779132466e-06,
      "loss": 0.2367,
      "step": 5791
    },
    {
      "epoch": 0.29909630777175317,
      "grad_norm": 8.541399955749512,
      "learning_rate": 8.039607249898351e-06,
      "loss": 0.1864,
      "step": 5792
    },
    {
      "epoch": 0.299147947327653,
      "grad_norm": 4.091602802276611,
      "learning_rate": 8.038956639038393e-06,
      "loss": 0.0342,
      "step": 5793
    },
    {
      "epoch": 0.2991995868835528,
      "grad_norm": 12.694801330566406,
      "learning_rate": 8.038305946570062e-06,
      "loss": 0.1976,
      "step": 5794
    },
    {
      "epoch": 0.2992512264394526,
      "grad_norm": 8.130784034729004,
      "learning_rate": 8.037655172510833e-06,
      "loss": 0.0751,
      "step": 5795
    },
    {
      "epoch": 0.29930286599535244,
      "grad_norm": 5.137282848358154,
      "learning_rate": 8.03700431687818e-06,
      "loss": 0.0803,
      "step": 5796
    },
    {
      "epoch": 0.29935450555125226,
      "grad_norm": 5.30049467086792,
      "learning_rate": 8.036353379689586e-06,
      "loss": 0.0716,
      "step": 5797
    },
    {
      "epoch": 0.2994061451071521,
      "grad_norm": 3.6960103511810303,
      "learning_rate": 8.035702360962527e-06,
      "loss": 0.0478,
      "step": 5798
    },
    {
      "epoch": 0.2994577846630519,
      "grad_norm": 13.419083595275879,
      "learning_rate": 8.035051260714486e-06,
      "loss": 0.1153,
      "step": 5799
    },
    {
      "epoch": 0.2995094242189517,
      "grad_norm": 21.756258010864258,
      "learning_rate": 8.03440007896295e-06,
      "loss": 0.0971,
      "step": 5800
    },
    {
      "epoch": 0.29956106377485153,
      "grad_norm": 1.9797885417938232,
      "learning_rate": 8.033748815725404e-06,
      "loss": 0.0359,
      "step": 5801
    },
    {
      "epoch": 0.29961270333075135,
      "grad_norm": 8.29172134399414,
      "learning_rate": 8.033097471019336e-06,
      "loss": 0.1377,
      "step": 5802
    },
    {
      "epoch": 0.29966434288665117,
      "grad_norm": 3.4410808086395264,
      "learning_rate": 8.03244604486224e-06,
      "loss": 0.087,
      "step": 5803
    },
    {
      "epoch": 0.299715982442551,
      "grad_norm": 11.393887519836426,
      "learning_rate": 8.031794537271606e-06,
      "loss": 0.1048,
      "step": 5804
    },
    {
      "epoch": 0.2997676219984508,
      "grad_norm": 16.368505477905273,
      "learning_rate": 8.031142948264936e-06,
      "loss": 0.1528,
      "step": 5805
    },
    {
      "epoch": 0.2998192615543506,
      "grad_norm": 4.058346748352051,
      "learning_rate": 8.03049127785972e-06,
      "loss": 0.0805,
      "step": 5806
    },
    {
      "epoch": 0.29987090111025044,
      "grad_norm": 2.543104887008667,
      "learning_rate": 8.029839526073463e-06,
      "loss": 0.0586,
      "step": 5807
    },
    {
      "epoch": 0.29992254066615026,
      "grad_norm": 5.338653087615967,
      "learning_rate": 8.029187692923665e-06,
      "loss": 0.1266,
      "step": 5808
    },
    {
      "epoch": 0.2999741802220501,
      "grad_norm": 3.1898560523986816,
      "learning_rate": 8.028535778427832e-06,
      "loss": 0.0641,
      "step": 5809
    },
    {
      "epoch": 0.3000258197779499,
      "grad_norm": 5.7213921546936035,
      "learning_rate": 8.02788378260347e-06,
      "loss": 0.0636,
      "step": 5810
    },
    {
      "epoch": 0.3000774593338497,
      "grad_norm": 4.2361249923706055,
      "learning_rate": 8.027231705468087e-06,
      "loss": 0.1033,
      "step": 5811
    },
    {
      "epoch": 0.3000774593338497,
      "eval_loss": 0.07345248013734818,
      "eval_runtime": 34.1733,
      "eval_samples_per_second": 45.796,
      "eval_steps_per_second": 5.735,
      "step": 5811
    },
    {
      "epoch": 0.30012909888974953,
      "grad_norm": 2.648545503616333,
      "learning_rate": 8.026579547039195e-06,
      "loss": 0.0611,
      "step": 5812
    },
    {
      "epoch": 0.30018073844564935,
      "grad_norm": 3.276815176010132,
      "learning_rate": 8.025927307334305e-06,
      "loss": 0.0697,
      "step": 5813
    },
    {
      "epoch": 0.30023237800154917,
      "grad_norm": 3.602879524230957,
      "learning_rate": 8.025274986370934e-06,
      "loss": 0.0377,
      "step": 5814
    },
    {
      "epoch": 0.300284017557449,
      "grad_norm": 3.616623640060425,
      "learning_rate": 8.0246225841666e-06,
      "loss": 0.048,
      "step": 5815
    },
    {
      "epoch": 0.3003356571133488,
      "grad_norm": 6.744170665740967,
      "learning_rate": 8.023970100738822e-06,
      "loss": 0.058,
      "step": 5816
    },
    {
      "epoch": 0.3003872966692486,
      "grad_norm": 6.697353839874268,
      "learning_rate": 8.023317536105123e-06,
      "loss": 0.073,
      "step": 5817
    },
    {
      "epoch": 0.30043893622514845,
      "grad_norm": 5.738083362579346,
      "learning_rate": 8.022664890283022e-06,
      "loss": 0.1224,
      "step": 5818
    },
    {
      "epoch": 0.30049057578104826,
      "grad_norm": 4.560184001922607,
      "learning_rate": 8.022012163290051e-06,
      "loss": 0.0504,
      "step": 5819
    },
    {
      "epoch": 0.3005422153369481,
      "grad_norm": 2.9817357063293457,
      "learning_rate": 8.021359355143739e-06,
      "loss": 0.0478,
      "step": 5820
    },
    {
      "epoch": 0.3005938548928479,
      "grad_norm": 3.445615291595459,
      "learning_rate": 8.020706465861611e-06,
      "loss": 0.0288,
      "step": 5821
    },
    {
      "epoch": 0.3006454944487477,
      "grad_norm": 13.782726287841797,
      "learning_rate": 8.020053495461204e-06,
      "loss": 0.2314,
      "step": 5822
    },
    {
      "epoch": 0.30069713400464754,
      "grad_norm": 18.62116050720215,
      "learning_rate": 8.01940044396005e-06,
      "loss": 0.0935,
      "step": 5823
    },
    {
      "epoch": 0.30074877356054736,
      "grad_norm": 2.953171730041504,
      "learning_rate": 8.018747311375688e-06,
      "loss": 0.0688,
      "step": 5824
    },
    {
      "epoch": 0.3008004131164472,
      "grad_norm": 2.42928147315979,
      "learning_rate": 8.018094097725658e-06,
      "loss": 0.0241,
      "step": 5825
    },
    {
      "epoch": 0.300852052672347,
      "grad_norm": 3.804157257080078,
      "learning_rate": 8.0174408030275e-06,
      "loss": 0.0629,
      "step": 5826
    },
    {
      "epoch": 0.3009036922282468,
      "grad_norm": 17.846904754638672,
      "learning_rate": 8.016787427298759e-06,
      "loss": 0.1783,
      "step": 5827
    },
    {
      "epoch": 0.30095533178414663,
      "grad_norm": 5.875911712646484,
      "learning_rate": 8.016133970556978e-06,
      "loss": 0.062,
      "step": 5828
    },
    {
      "epoch": 0.3010069713400465,
      "grad_norm": 5.6068806648254395,
      "learning_rate": 8.015480432819708e-06,
      "loss": 0.0686,
      "step": 5829
    },
    {
      "epoch": 0.3010586108959463,
      "grad_norm": 9.763391494750977,
      "learning_rate": 8.014826814104497e-06,
      "loss": 0.1677,
      "step": 5830
    },
    {
      "epoch": 0.30111025045184614,
      "grad_norm": 9.479825973510742,
      "learning_rate": 8.0141731144289e-06,
      "loss": 0.1327,
      "step": 5831
    },
    {
      "epoch": 0.30116189000774596,
      "grad_norm": 2.913019895553589,
      "learning_rate": 8.01351933381047e-06,
      "loss": 0.0522,
      "step": 5832
    },
    {
      "epoch": 0.3012135295636458,
      "grad_norm": 4.11810827255249,
      "learning_rate": 8.01286547226676e-06,
      "loss": 0.068,
      "step": 5833
    },
    {
      "epoch": 0.3012651691195456,
      "grad_norm": 2.1306204795837402,
      "learning_rate": 8.012211529815335e-06,
      "loss": 0.0255,
      "step": 5834
    },
    {
      "epoch": 0.3013168086754454,
      "grad_norm": 3.551718235015869,
      "learning_rate": 8.011557506473753e-06,
      "loss": 0.055,
      "step": 5835
    },
    {
      "epoch": 0.30136844823134523,
      "grad_norm": 17.336198806762695,
      "learning_rate": 8.010903402259578e-06,
      "loss": 0.0724,
      "step": 5836
    },
    {
      "epoch": 0.30142008778724505,
      "grad_norm": 3.5993409156799316,
      "learning_rate": 8.010249217190373e-06,
      "loss": 0.0745,
      "step": 5837
    },
    {
      "epoch": 0.30147172734314487,
      "grad_norm": 4.2292609214782715,
      "learning_rate": 8.009594951283708e-06,
      "loss": 0.0827,
      "step": 5838
    },
    {
      "epoch": 0.3015233668990447,
      "grad_norm": 3.8759219646453857,
      "learning_rate": 8.008940604557154e-06,
      "loss": 0.0358,
      "step": 5839
    },
    {
      "epoch": 0.3015750064549445,
      "grad_norm": 8.897503852844238,
      "learning_rate": 8.00828617702828e-06,
      "loss": 0.1311,
      "step": 5840
    },
    {
      "epoch": 0.3016266460108443,
      "grad_norm": 15.90965461730957,
      "learning_rate": 8.007631668714659e-06,
      "loss": 0.1416,
      "step": 5841
    },
    {
      "epoch": 0.30167828556674414,
      "grad_norm": 4.4109110832214355,
      "learning_rate": 8.006977079633872e-06,
      "loss": 0.0607,
      "step": 5842
    },
    {
      "epoch": 0.30172992512264396,
      "grad_norm": 2.7288689613342285,
      "learning_rate": 8.006322409803495e-06,
      "loss": 0.0468,
      "step": 5843
    },
    {
      "epoch": 0.3017815646785438,
      "grad_norm": 4.3619608879089355,
      "learning_rate": 8.005667659241104e-06,
      "loss": 0.0474,
      "step": 5844
    },
    {
      "epoch": 0.3018332042344436,
      "grad_norm": 1.9650278091430664,
      "learning_rate": 8.00501282796429e-06,
      "loss": 0.0253,
      "step": 5845
    },
    {
      "epoch": 0.3018848437903434,
      "grad_norm": 3.6247379779815674,
      "learning_rate": 8.004357915990633e-06,
      "loss": 0.0883,
      "step": 5846
    },
    {
      "epoch": 0.30193648334624323,
      "grad_norm": 2.5396571159362793,
      "learning_rate": 8.003702923337722e-06,
      "loss": 0.033,
      "step": 5847
    },
    {
      "epoch": 0.30198812290214305,
      "grad_norm": 7.200725078582764,
      "learning_rate": 8.003047850023143e-06,
      "loss": 0.1119,
      "step": 5848
    },
    {
      "epoch": 0.30203976245804287,
      "grad_norm": 3.269681692123413,
      "learning_rate": 8.00239269606449e-06,
      "loss": 0.0453,
      "step": 5849
    },
    {
      "epoch": 0.3020914020139427,
      "grad_norm": 6.143890857696533,
      "learning_rate": 8.001737461479355e-06,
      "loss": 0.0968,
      "step": 5850
    },
    {
      "epoch": 0.3021430415698425,
      "grad_norm": 5.726672649383545,
      "learning_rate": 8.001082146285337e-06,
      "loss": 0.0722,
      "step": 5851
    },
    {
      "epoch": 0.3021946811257423,
      "grad_norm": 5.227757930755615,
      "learning_rate": 8.00042675050003e-06,
      "loss": 0.0262,
      "step": 5852
    },
    {
      "epoch": 0.30224632068164214,
      "grad_norm": 3.2253825664520264,
      "learning_rate": 7.999771274141038e-06,
      "loss": 0.0517,
      "step": 5853
    },
    {
      "epoch": 0.30229796023754196,
      "grad_norm": 6.309185028076172,
      "learning_rate": 7.999115717225958e-06,
      "loss": 0.1264,
      "step": 5854
    },
    {
      "epoch": 0.3023495997934418,
      "grad_norm": 4.103766918182373,
      "learning_rate": 7.998460079772398e-06,
      "loss": 0.0641,
      "step": 5855
    },
    {
      "epoch": 0.3024012393493416,
      "grad_norm": 18.381532669067383,
      "learning_rate": 7.997804361797964e-06,
      "loss": 0.0428,
      "step": 5856
    },
    {
      "epoch": 0.3024528789052414,
      "grad_norm": 6.778690814971924,
      "learning_rate": 7.997148563320266e-06,
      "loss": 0.1217,
      "step": 5857
    },
    {
      "epoch": 0.30250451846114124,
      "grad_norm": 6.939395427703857,
      "learning_rate": 7.99649268435691e-06,
      "loss": 0.0663,
      "step": 5858
    },
    {
      "epoch": 0.30255615801704105,
      "grad_norm": 13.58227252960205,
      "learning_rate": 7.995836724925515e-06,
      "loss": 0.1533,
      "step": 5859
    },
    {
      "epoch": 0.3026077975729409,
      "grad_norm": 4.251084804534912,
      "learning_rate": 7.995180685043693e-06,
      "loss": 0.0543,
      "step": 5860
    },
    {
      "epoch": 0.3026594371288407,
      "grad_norm": 3.5878102779388428,
      "learning_rate": 7.994524564729063e-06,
      "loss": 0.0537,
      "step": 5861
    },
    {
      "epoch": 0.3027110766847405,
      "grad_norm": 6.4769439697265625,
      "learning_rate": 7.993868363999242e-06,
      "loss": 0.0717,
      "step": 5862
    },
    {
      "epoch": 0.30276271624064033,
      "grad_norm": 7.916511535644531,
      "learning_rate": 7.993212082871853e-06,
      "loss": 0.0541,
      "step": 5863
    },
    {
      "epoch": 0.30281435579654015,
      "grad_norm": 7.308625221252441,
      "learning_rate": 7.99255572136452e-06,
      "loss": 0.1062,
      "step": 5864
    },
    {
      "epoch": 0.30286599535243997,
      "grad_norm": 4.454384803771973,
      "learning_rate": 7.99189927949487e-06,
      "loss": 0.0524,
      "step": 5865
    },
    {
      "epoch": 0.3029176349083398,
      "grad_norm": 6.361854076385498,
      "learning_rate": 7.99124275728053e-06,
      "loss": 0.0598,
      "step": 5866
    },
    {
      "epoch": 0.3029692744642396,
      "grad_norm": 5.750885009765625,
      "learning_rate": 7.990586154739131e-06,
      "loss": 0.1136,
      "step": 5867
    },
    {
      "epoch": 0.3030209140201394,
      "grad_norm": 4.576197147369385,
      "learning_rate": 7.989929471888303e-06,
      "loss": 0.0537,
      "step": 5868
    },
    {
      "epoch": 0.30307255357603924,
      "grad_norm": 8.750289916992188,
      "learning_rate": 7.98927270874568e-06,
      "loss": 0.0448,
      "step": 5869
    },
    {
      "epoch": 0.30312419313193906,
      "grad_norm": 4.0901079177856445,
      "learning_rate": 7.988615865328906e-06,
      "loss": 0.0443,
      "step": 5870
    },
    {
      "epoch": 0.3031758326878389,
      "grad_norm": 3.566396951675415,
      "learning_rate": 7.987958941655613e-06,
      "loss": 0.0492,
      "step": 5871
    },
    {
      "epoch": 0.3032274722437387,
      "grad_norm": 2.9716808795928955,
      "learning_rate": 7.987301937743444e-06,
      "loss": 0.0516,
      "step": 5872
    },
    {
      "epoch": 0.3032791117996385,
      "grad_norm": 10.081628799438477,
      "learning_rate": 7.986644853610044e-06,
      "loss": 0.1297,
      "step": 5873
    },
    {
      "epoch": 0.30333075135553833,
      "grad_norm": 9.713566780090332,
      "learning_rate": 7.985987689273053e-06,
      "loss": 0.0634,
      "step": 5874
    },
    {
      "epoch": 0.30338239091143815,
      "grad_norm": 4.3662190437316895,
      "learning_rate": 7.985330444750126e-06,
      "loss": 0.0524,
      "step": 5875
    },
    {
      "epoch": 0.30343403046733797,
      "grad_norm": 4.776942253112793,
      "learning_rate": 7.984673120058908e-06,
      "loss": 0.068,
      "step": 5876
    },
    {
      "epoch": 0.3034856700232378,
      "grad_norm": 15.040531158447266,
      "learning_rate": 7.984015715217051e-06,
      "loss": 0.1033,
      "step": 5877
    },
    {
      "epoch": 0.3035373095791376,
      "grad_norm": 5.087726593017578,
      "learning_rate": 7.983358230242212e-06,
      "loss": 0.1613,
      "step": 5878
    },
    {
      "epoch": 0.3035889491350374,
      "grad_norm": 2.5723583698272705,
      "learning_rate": 7.982700665152043e-06,
      "loss": 0.0554,
      "step": 5879
    },
    {
      "epoch": 0.30364058869093724,
      "grad_norm": 2.2464756965637207,
      "learning_rate": 7.982043019964206e-06,
      "loss": 0.0474,
      "step": 5880
    },
    {
      "epoch": 0.30369222824683706,
      "grad_norm": 4.883141040802002,
      "learning_rate": 7.981385294696357e-06,
      "loss": 0.0594,
      "step": 5881
    },
    {
      "epoch": 0.3037438678027369,
      "grad_norm": 4.585298538208008,
      "learning_rate": 7.980727489366165e-06,
      "loss": 0.0903,
      "step": 5882
    },
    {
      "epoch": 0.3037955073586367,
      "grad_norm": 5.438168525695801,
      "learning_rate": 7.980069603991287e-06,
      "loss": 0.1066,
      "step": 5883
    },
    {
      "epoch": 0.3038471469145365,
      "grad_norm": 4.55398416519165,
      "learning_rate": 7.979411638589398e-06,
      "loss": 0.0786,
      "step": 5884
    },
    {
      "epoch": 0.30389878647043633,
      "grad_norm": 9.09473991394043,
      "learning_rate": 7.978753593178161e-06,
      "loss": 0.0606,
      "step": 5885
    },
    {
      "epoch": 0.30395042602633615,
      "grad_norm": 4.721724510192871,
      "learning_rate": 7.978095467775252e-06,
      "loss": 0.0472,
      "step": 5886
    },
    {
      "epoch": 0.30400206558223597,
      "grad_norm": 12.847050666809082,
      "learning_rate": 7.977437262398337e-06,
      "loss": 0.0671,
      "step": 5887
    },
    {
      "epoch": 0.3040537051381358,
      "grad_norm": 6.622143745422363,
      "learning_rate": 7.976778977065098e-06,
      "loss": 0.0804,
      "step": 5888
    },
    {
      "epoch": 0.3041053446940356,
      "grad_norm": 6.174838066101074,
      "learning_rate": 7.976120611793212e-06,
      "loss": 0.0822,
      "step": 5889
    },
    {
      "epoch": 0.3041569842499354,
      "grad_norm": 5.491322994232178,
      "learning_rate": 7.975462166600357e-06,
      "loss": 0.0522,
      "step": 5890
    },
    {
      "epoch": 0.30420862380583524,
      "grad_norm": 4.222060680389404,
      "learning_rate": 7.974803641504214e-06,
      "loss": 0.0546,
      "step": 5891
    },
    {
      "epoch": 0.30426026336173506,
      "grad_norm": 12.066967010498047,
      "learning_rate": 7.974145036522472e-06,
      "loss": 0.1565,
      "step": 5892
    },
    {
      "epoch": 0.30431190291763494,
      "grad_norm": 2.60688853263855,
      "learning_rate": 7.973486351672812e-06,
      "loss": 0.0483,
      "step": 5893
    },
    {
      "epoch": 0.30436354247353475,
      "grad_norm": 2.4772145748138428,
      "learning_rate": 7.972827586972922e-06,
      "loss": 0.0195,
      "step": 5894
    },
    {
      "epoch": 0.3044151820294346,
      "grad_norm": 5.47965669631958,
      "learning_rate": 7.972168742440498e-06,
      "loss": 0.083,
      "step": 5895
    },
    {
      "epoch": 0.3044668215853344,
      "grad_norm": 4.234801769256592,
      "learning_rate": 7.971509818093227e-06,
      "loss": 0.0618,
      "step": 5896
    },
    {
      "epoch": 0.3045184611412342,
      "grad_norm": 5.941784858703613,
      "learning_rate": 7.970850813948808e-06,
      "loss": 0.0502,
      "step": 5897
    },
    {
      "epoch": 0.304570100697134,
      "grad_norm": 5.760707378387451,
      "learning_rate": 7.970191730024935e-06,
      "loss": 0.0742,
      "step": 5898
    },
    {
      "epoch": 0.30462174025303385,
      "grad_norm": 13.20080280303955,
      "learning_rate": 7.969532566339308e-06,
      "loss": 0.1148,
      "step": 5899
    },
    {
      "epoch": 0.30467337980893366,
      "grad_norm": 5.966351509094238,
      "learning_rate": 7.968873322909631e-06,
      "loss": 0.0905,
      "step": 5900
    },
    {
      "epoch": 0.3047250193648335,
      "grad_norm": 3.0943710803985596,
      "learning_rate": 7.968213999753603e-06,
      "loss": 0.0863,
      "step": 5901
    },
    {
      "epoch": 0.3047766589207333,
      "grad_norm": 4.608373641967773,
      "learning_rate": 7.967554596888931e-06,
      "loss": 0.0726,
      "step": 5902
    },
    {
      "epoch": 0.3048282984766331,
      "grad_norm": 3.3566577434539795,
      "learning_rate": 7.966895114333322e-06,
      "loss": 0.0465,
      "step": 5903
    },
    {
      "epoch": 0.30487993803253294,
      "grad_norm": 5.839597702026367,
      "learning_rate": 7.966235552104489e-06,
      "loss": 0.0464,
      "step": 5904
    },
    {
      "epoch": 0.30493157758843276,
      "grad_norm": 4.212308406829834,
      "learning_rate": 7.96557591022014e-06,
      "loss": 0.0804,
      "step": 5905
    },
    {
      "epoch": 0.3049832171443326,
      "grad_norm": 15.19775104522705,
      "learning_rate": 7.964916188697991e-06,
      "loss": 0.2736,
      "step": 5906
    },
    {
      "epoch": 0.3050348567002324,
      "grad_norm": 9.64355182647705,
      "learning_rate": 7.964256387555758e-06,
      "loss": 0.1877,
      "step": 5907
    },
    {
      "epoch": 0.3050864962561322,
      "grad_norm": 20.938556671142578,
      "learning_rate": 7.963596506811161e-06,
      "loss": 0.1217,
      "step": 5908
    },
    {
      "epoch": 0.30513813581203203,
      "grad_norm": 4.657806873321533,
      "learning_rate": 7.962936546481918e-06,
      "loss": 0.0614,
      "step": 5909
    },
    {
      "epoch": 0.30518977536793185,
      "grad_norm": 5.989072799682617,
      "learning_rate": 7.962276506585751e-06,
      "loss": 0.0857,
      "step": 5910
    },
    {
      "epoch": 0.30524141492383167,
      "grad_norm": 2.4485740661621094,
      "learning_rate": 7.961616387140388e-06,
      "loss": 0.0594,
      "step": 5911
    },
    {
      "epoch": 0.3052930544797315,
      "grad_norm": 2.3174941539764404,
      "learning_rate": 7.960956188163553e-06,
      "loss": 0.0365,
      "step": 5912
    },
    {
      "epoch": 0.3053446940356313,
      "grad_norm": 3.606959581375122,
      "learning_rate": 7.960295909672977e-06,
      "loss": 0.0447,
      "step": 5913
    },
    {
      "epoch": 0.3053963335915311,
      "grad_norm": 4.419679641723633,
      "learning_rate": 7.95963555168639e-06,
      "loss": 0.098,
      "step": 5914
    },
    {
      "epoch": 0.30544797314743094,
      "grad_norm": 11.091986656188965,
      "learning_rate": 7.958975114221525e-06,
      "loss": 0.1856,
      "step": 5915
    },
    {
      "epoch": 0.30549961270333076,
      "grad_norm": 5.0398101806640625,
      "learning_rate": 7.95831459729612e-06,
      "loss": 0.0639,
      "step": 5916
    },
    {
      "epoch": 0.3055512522592306,
      "grad_norm": 7.386066913604736,
      "learning_rate": 7.957654000927909e-06,
      "loss": 0.1217,
      "step": 5917
    },
    {
      "epoch": 0.3056028918151304,
      "grad_norm": 3.3856029510498047,
      "learning_rate": 7.956993325134635e-06,
      "loss": 0.0805,
      "step": 5918
    },
    {
      "epoch": 0.3056545313710302,
      "grad_norm": 15.543991088867188,
      "learning_rate": 7.956332569934036e-06,
      "loss": 0.073,
      "step": 5919
    },
    {
      "epoch": 0.30570617092693003,
      "grad_norm": 1.986241102218628,
      "learning_rate": 7.95567173534386e-06,
      "loss": 0.0475,
      "step": 5920
    },
    {
      "epoch": 0.30575781048282985,
      "grad_norm": 2.5046157836914062,
      "learning_rate": 7.955010821381849e-06,
      "loss": 0.0384,
      "step": 5921
    },
    {
      "epoch": 0.30580945003872967,
      "grad_norm": 7.2662672996521,
      "learning_rate": 7.954349828065755e-06,
      "loss": 0.0963,
      "step": 5922
    },
    {
      "epoch": 0.3058610895946295,
      "grad_norm": 2.8698980808258057,
      "learning_rate": 7.953688755413326e-06,
      "loss": 0.0513,
      "step": 5923
    },
    {
      "epoch": 0.3059127291505293,
      "grad_norm": 11.58342170715332,
      "learning_rate": 7.953027603442317e-06,
      "loss": 0.126,
      "step": 5924
    },
    {
      "epoch": 0.3059643687064291,
      "grad_norm": 3.7281062602996826,
      "learning_rate": 7.952366372170479e-06,
      "loss": 0.0556,
      "step": 5925
    },
    {
      "epoch": 0.30601600826232894,
      "grad_norm": 4.6912431716918945,
      "learning_rate": 7.951705061615573e-06,
      "loss": 0.0611,
      "step": 5926
    },
    {
      "epoch": 0.30606764781822876,
      "grad_norm": 5.46566915512085,
      "learning_rate": 7.951043671795352e-06,
      "loss": 0.0669,
      "step": 5927
    },
    {
      "epoch": 0.3061192873741286,
      "grad_norm": 2.6635124683380127,
      "learning_rate": 7.950382202727585e-06,
      "loss": 0.0631,
      "step": 5928
    },
    {
      "epoch": 0.3061709269300284,
      "grad_norm": 3.432312488555908,
      "learning_rate": 7.949720654430026e-06,
      "loss": 0.0514,
      "step": 5929
    },
    {
      "epoch": 0.3062225664859282,
      "grad_norm": 11.064658164978027,
      "learning_rate": 7.94905902692045e-06,
      "loss": 0.0823,
      "step": 5930
    },
    {
      "epoch": 0.30627420604182803,
      "grad_norm": 5.361155033111572,
      "learning_rate": 7.948397320216616e-06,
      "loss": 0.0642,
      "step": 5931
    },
    {
      "epoch": 0.30632584559772785,
      "grad_norm": 10.332901000976562,
      "learning_rate": 7.947735534336299e-06,
      "loss": 0.0832,
      "step": 5932
    },
    {
      "epoch": 0.30637748515362767,
      "grad_norm": 2.0726318359375,
      "learning_rate": 7.947073669297268e-06,
      "loss": 0.0426,
      "step": 5933
    },
    {
      "epoch": 0.3064291247095275,
      "grad_norm": 14.972683906555176,
      "learning_rate": 7.946411725117295e-06,
      "loss": 0.1564,
      "step": 5934
    },
    {
      "epoch": 0.3064807642654273,
      "grad_norm": 12.186128616333008,
      "learning_rate": 7.945749701814161e-06,
      "loss": 0.1374,
      "step": 5935
    },
    {
      "epoch": 0.3065324038213271,
      "grad_norm": 9.675642013549805,
      "learning_rate": 7.94508759940564e-06,
      "loss": 0.1236,
      "step": 5936
    },
    {
      "epoch": 0.30658404337722694,
      "grad_norm": 18.1549015045166,
      "learning_rate": 7.944425417909515e-06,
      "loss": 0.1724,
      "step": 5937
    },
    {
      "epoch": 0.30663568293312676,
      "grad_norm": 4.902152061462402,
      "learning_rate": 7.943763157343565e-06,
      "loss": 0.1042,
      "step": 5938
    },
    {
      "epoch": 0.3066873224890266,
      "grad_norm": 16.510099411010742,
      "learning_rate": 7.943100817725577e-06,
      "loss": 0.1512,
      "step": 5939
    },
    {
      "epoch": 0.3067389620449264,
      "grad_norm": 6.650203227996826,
      "learning_rate": 7.942438399073338e-06,
      "loss": 0.085,
      "step": 5940
    },
    {
      "epoch": 0.3067906016008262,
      "grad_norm": 5.003183364868164,
      "learning_rate": 7.941775901404633e-06,
      "loss": 0.0851,
      "step": 5941
    },
    {
      "epoch": 0.30684224115672604,
      "grad_norm": 2.9200387001037598,
      "learning_rate": 7.941113324737256e-06,
      "loss": 0.0613,
      "step": 5942
    },
    {
      "epoch": 0.30689388071262586,
      "grad_norm": 2.5521240234375,
      "learning_rate": 7.940450669088999e-06,
      "loss": 0.0558,
      "step": 5943
    },
    {
      "epoch": 0.3069455202685257,
      "grad_norm": 2.6814143657684326,
      "learning_rate": 7.939787934477657e-06,
      "loss": 0.051,
      "step": 5944
    },
    {
      "epoch": 0.3069971598244255,
      "grad_norm": 3.499223232269287,
      "learning_rate": 7.939125120921026e-06,
      "loss": 0.074,
      "step": 5945
    },
    {
      "epoch": 0.3070487993803253,
      "grad_norm": 3.218343734741211,
      "learning_rate": 7.938462228436906e-06,
      "loss": 0.0809,
      "step": 5946
    },
    {
      "epoch": 0.30710043893622513,
      "grad_norm": 13.720624923706055,
      "learning_rate": 7.9377992570431e-06,
      "loss": 0.1683,
      "step": 5947
    },
    {
      "epoch": 0.30715207849212495,
      "grad_norm": 3.0459446907043457,
      "learning_rate": 7.93713620675741e-06,
      "loss": 0.0387,
      "step": 5948
    },
    {
      "epoch": 0.30720371804802477,
      "grad_norm": 6.667078495025635,
      "learning_rate": 7.936473077597643e-06,
      "loss": 0.0673,
      "step": 5949
    },
    {
      "epoch": 0.3072553576039246,
      "grad_norm": 3.303097724914551,
      "learning_rate": 7.935809869581604e-06,
      "loss": 0.0867,
      "step": 5950
    },
    {
      "epoch": 0.3073069971598244,
      "grad_norm": 9.625539779663086,
      "learning_rate": 7.935146582727103e-06,
      "loss": 0.0813,
      "step": 5951
    },
    {
      "epoch": 0.3073586367157242,
      "grad_norm": 3.3208792209625244,
      "learning_rate": 7.934483217051954e-06,
      "loss": 0.03,
      "step": 5952
    },
    {
      "epoch": 0.30741027627162404,
      "grad_norm": 4.3972063064575195,
      "learning_rate": 7.933819772573971e-06,
      "loss": 0.0397,
      "step": 5953
    },
    {
      "epoch": 0.30746191582752386,
      "grad_norm": 2.962280035018921,
      "learning_rate": 7.93315624931097e-06,
      "loss": 0.0416,
      "step": 5954
    },
    {
      "epoch": 0.3075135553834237,
      "grad_norm": 17.847379684448242,
      "learning_rate": 7.932492647280768e-06,
      "loss": 0.1881,
      "step": 5955
    },
    {
      "epoch": 0.3075651949393235,
      "grad_norm": 3.4244983196258545,
      "learning_rate": 7.931828966501185e-06,
      "loss": 0.0715,
      "step": 5956
    },
    {
      "epoch": 0.30761683449522337,
      "grad_norm": 6.070005416870117,
      "learning_rate": 7.931165206990048e-06,
      "loss": 0.0864,
      "step": 5957
    },
    {
      "epoch": 0.3076684740511232,
      "grad_norm": 5.8920488357543945,
      "learning_rate": 7.930501368765176e-06,
      "loss": 0.0662,
      "step": 5958
    },
    {
      "epoch": 0.307720113607023,
      "grad_norm": 4.085729122161865,
      "learning_rate": 7.9298374518444e-06,
      "loss": 0.0671,
      "step": 5959
    },
    {
      "epoch": 0.3077717531629228,
      "grad_norm": 11.124551773071289,
      "learning_rate": 7.929173456245545e-06,
      "loss": 0.1213,
      "step": 5960
    },
    {
      "epoch": 0.30782339271882264,
      "grad_norm": 5.114633083343506,
      "learning_rate": 7.928509381986445e-06,
      "loss": 0.0656,
      "step": 5961
    },
    {
      "epoch": 0.30787503227472246,
      "grad_norm": 3.4618513584136963,
      "learning_rate": 7.927845229084931e-06,
      "loss": 0.0341,
      "step": 5962
    },
    {
      "epoch": 0.3079266718306223,
      "grad_norm": 8.93612003326416,
      "learning_rate": 7.927180997558842e-06,
      "loss": 0.0555,
      "step": 5963
    },
    {
      "epoch": 0.3079783113865221,
      "grad_norm": 2.6735281944274902,
      "learning_rate": 7.926516687426012e-06,
      "loss": 0.0507,
      "step": 5964
    },
    {
      "epoch": 0.3080299509424219,
      "grad_norm": 6.6797943115234375,
      "learning_rate": 7.92585229870428e-06,
      "loss": 0.0853,
      "step": 5965
    },
    {
      "epoch": 0.30808159049832173,
      "grad_norm": 8.076233863830566,
      "learning_rate": 7.925187831411487e-06,
      "loss": 0.1208,
      "step": 5966
    },
    {
      "epoch": 0.30813323005422155,
      "grad_norm": 15.149308204650879,
      "learning_rate": 7.924523285565482e-06,
      "loss": 0.1119,
      "step": 5967
    },
    {
      "epoch": 0.30818486961012137,
      "grad_norm": 5.486983299255371,
      "learning_rate": 7.923858661184105e-06,
      "loss": 0.1031,
      "step": 5968
    },
    {
      "epoch": 0.3082365091660212,
      "grad_norm": 3.0709967613220215,
      "learning_rate": 7.923193958285205e-06,
      "loss": 0.025,
      "step": 5969
    },
    {
      "epoch": 0.308288148721921,
      "grad_norm": 4.659209251403809,
      "learning_rate": 7.922529176886636e-06,
      "loss": 0.062,
      "step": 5970
    },
    {
      "epoch": 0.3083397882778208,
      "grad_norm": 6.056024074554443,
      "learning_rate": 7.921864317006246e-06,
      "loss": 0.0528,
      "step": 5971
    },
    {
      "epoch": 0.30839142783372064,
      "grad_norm": 7.333767890930176,
      "learning_rate": 7.921199378661888e-06,
      "loss": 0.0654,
      "step": 5972
    },
    {
      "epoch": 0.30844306738962046,
      "grad_norm": 7.1972880363464355,
      "learning_rate": 7.920534361871421e-06,
      "loss": 0.1199,
      "step": 5973
    },
    {
      "epoch": 0.3084947069455203,
      "grad_norm": 5.44082498550415,
      "learning_rate": 7.919869266652704e-06,
      "loss": 0.058,
      "step": 5974
    },
    {
      "epoch": 0.3085463465014201,
      "grad_norm": 9.432657241821289,
      "learning_rate": 7.919204093023597e-06,
      "loss": 0.0481,
      "step": 5975
    },
    {
      "epoch": 0.3085979860573199,
      "grad_norm": 8.09091567993164,
      "learning_rate": 7.918538841001962e-06,
      "loss": 0.077,
      "step": 5976
    },
    {
      "epoch": 0.30864962561321974,
      "grad_norm": 14.674487113952637,
      "learning_rate": 7.917873510605662e-06,
      "loss": 0.2482,
      "step": 5977
    },
    {
      "epoch": 0.30870126516911955,
      "grad_norm": 11.194457054138184,
      "learning_rate": 7.917208101852569e-06,
      "loss": 0.1073,
      "step": 5978
    },
    {
      "epoch": 0.3087529047250194,
      "grad_norm": 5.050597667694092,
      "learning_rate": 7.916542614760544e-06,
      "loss": 0.0738,
      "step": 5979
    },
    {
      "epoch": 0.3088045442809192,
      "grad_norm": 8.16178035736084,
      "learning_rate": 7.915877049347467e-06,
      "loss": 0.0952,
      "step": 5980
    },
    {
      "epoch": 0.308856183836819,
      "grad_norm": 4.376867294311523,
      "learning_rate": 7.915211405631205e-06,
      "loss": 0.0303,
      "step": 5981
    },
    {
      "epoch": 0.3089078233927188,
      "grad_norm": 3.1252071857452393,
      "learning_rate": 7.914545683629634e-06,
      "loss": 0.0625,
      "step": 5982
    },
    {
      "epoch": 0.30895946294861865,
      "grad_norm": 5.558148384094238,
      "learning_rate": 7.913879883360633e-06,
      "loss": 0.0654,
      "step": 5983
    },
    {
      "epoch": 0.30901110250451846,
      "grad_norm": 10.245326042175293,
      "learning_rate": 7.913214004842083e-06,
      "loss": 0.1002,
      "step": 5984
    },
    {
      "epoch": 0.3090627420604183,
      "grad_norm": 3.4512436389923096,
      "learning_rate": 7.91254804809186e-06,
      "loss": 0.0509,
      "step": 5985
    },
    {
      "epoch": 0.3091143816163181,
      "grad_norm": 3.3083622455596924,
      "learning_rate": 7.911882013127854e-06,
      "loss": 0.0678,
      "step": 5986
    },
    {
      "epoch": 0.3091660211722179,
      "grad_norm": 10.323721885681152,
      "learning_rate": 7.911215899967945e-06,
      "loss": 0.171,
      "step": 5987
    },
    {
      "epoch": 0.30921766072811774,
      "grad_norm": 12.140131950378418,
      "learning_rate": 7.910549708630026e-06,
      "loss": 0.0995,
      "step": 5988
    },
    {
      "epoch": 0.30926930028401756,
      "grad_norm": 3.103203058242798,
      "learning_rate": 7.909883439131982e-06,
      "loss": 0.0583,
      "step": 5989
    },
    {
      "epoch": 0.3093209398399174,
      "grad_norm": 5.466939926147461,
      "learning_rate": 7.90921709149171e-06,
      "loss": 0.0757,
      "step": 5990
    },
    {
      "epoch": 0.3093725793958172,
      "grad_norm": 3.0791332721710205,
      "learning_rate": 7.908550665727102e-06,
      "loss": 0.0368,
      "step": 5991
    },
    {
      "epoch": 0.309424218951717,
      "grad_norm": 14.761110305786133,
      "learning_rate": 7.907884161856052e-06,
      "loss": 0.2081,
      "step": 5992
    },
    {
      "epoch": 0.30947585850761683,
      "grad_norm": 3.6984570026397705,
      "learning_rate": 7.907217579896461e-06,
      "loss": 0.0568,
      "step": 5993
    },
    {
      "epoch": 0.30952749806351665,
      "grad_norm": 2.291147232055664,
      "learning_rate": 7.906550919866229e-06,
      "loss": 0.0517,
      "step": 5994
    },
    {
      "epoch": 0.30957913761941647,
      "grad_norm": 3.8203508853912354,
      "learning_rate": 7.905884181783258e-06,
      "loss": 0.0555,
      "step": 5995
    },
    {
      "epoch": 0.3096307771753163,
      "grad_norm": 3.84792160987854,
      "learning_rate": 7.905217365665455e-06,
      "loss": 0.1008,
      "step": 5996
    },
    {
      "epoch": 0.3096824167312161,
      "grad_norm": 4.907369613647461,
      "learning_rate": 7.904550471530723e-06,
      "loss": 0.072,
      "step": 5997
    },
    {
      "epoch": 0.3097340562871159,
      "grad_norm": 3.409449815750122,
      "learning_rate": 7.903883499396972e-06,
      "loss": 0.04,
      "step": 5998
    },
    {
      "epoch": 0.30978569584301574,
      "grad_norm": 7.692700386047363,
      "learning_rate": 7.903216449282114e-06,
      "loss": 0.0531,
      "step": 5999
    },
    {
      "epoch": 0.30983733539891556,
      "grad_norm": 3.6969268321990967,
      "learning_rate": 7.902549321204062e-06,
      "loss": 0.0674,
      "step": 6000
    },
    {
      "epoch": 0.3098889749548154,
      "grad_norm": 6.322739124298096,
      "learning_rate": 7.901882115180732e-06,
      "loss": 0.0766,
      "step": 6001
    },
    {
      "epoch": 0.3099406145107152,
      "grad_norm": 8.060162544250488,
      "learning_rate": 7.901214831230038e-06,
      "loss": 0.0598,
      "step": 6002
    },
    {
      "epoch": 0.309992254066615,
      "grad_norm": 26.39920425415039,
      "learning_rate": 7.9005474693699e-06,
      "loss": 0.1955,
      "step": 6003
    },
    {
      "epoch": 0.31004389362251483,
      "grad_norm": 3.4334049224853516,
      "learning_rate": 7.89988002961824e-06,
      "loss": 0.0527,
      "step": 6004
    },
    {
      "epoch": 0.31009553317841465,
      "grad_norm": 6.569737911224365,
      "learning_rate": 7.899212511992985e-06,
      "loss": 0.087,
      "step": 6005
    },
    {
      "epoch": 0.31014717273431447,
      "grad_norm": 1.9123011827468872,
      "learning_rate": 7.898544916512056e-06,
      "loss": 0.0471,
      "step": 6006
    },
    {
      "epoch": 0.3101988122902143,
      "grad_norm": 2.5026161670684814,
      "learning_rate": 7.897877243193383e-06,
      "loss": 0.0383,
      "step": 6007
    },
    {
      "epoch": 0.3102504518461141,
      "grad_norm": 3.642285108566284,
      "learning_rate": 7.897209492054894e-06,
      "loss": 0.0573,
      "step": 6008
    },
    {
      "epoch": 0.3103020914020139,
      "grad_norm": 13.008405685424805,
      "learning_rate": 7.896541663114522e-06,
      "loss": 0.1561,
      "step": 6009
    },
    {
      "epoch": 0.31035373095791374,
      "grad_norm": 3.3139824867248535,
      "learning_rate": 7.8958737563902e-06,
      "loss": 0.0715,
      "step": 6010
    },
    {
      "epoch": 0.31040537051381356,
      "grad_norm": 2.5373451709747314,
      "learning_rate": 7.895205771899865e-06,
      "loss": 0.0402,
      "step": 6011
    },
    {
      "epoch": 0.3104570100697134,
      "grad_norm": 12.233325004577637,
      "learning_rate": 7.894537709661456e-06,
      "loss": 0.1675,
      "step": 6012
    },
    {
      "epoch": 0.3105086496256132,
      "grad_norm": 6.4871649742126465,
      "learning_rate": 7.893869569692911e-06,
      "loss": 0.0565,
      "step": 6013
    },
    {
      "epoch": 0.310560289181513,
      "grad_norm": 6.543755054473877,
      "learning_rate": 7.893201352012173e-06,
      "loss": 0.0542,
      "step": 6014
    },
    {
      "epoch": 0.31061192873741283,
      "grad_norm": 6.123234272003174,
      "learning_rate": 7.892533056637188e-06,
      "loss": 0.0928,
      "step": 6015
    },
    {
      "epoch": 0.31066356829331265,
      "grad_norm": 4.403817176818848,
      "learning_rate": 7.8918646835859e-06,
      "loss": 0.0622,
      "step": 6016
    },
    {
      "epoch": 0.31071520784921247,
      "grad_norm": 4.415512561798096,
      "learning_rate": 7.89119623287626e-06,
      "loss": 0.0465,
      "step": 6017
    },
    {
      "epoch": 0.3107668474051123,
      "grad_norm": 7.297407627105713,
      "learning_rate": 7.890527704526216e-06,
      "loss": 0.0862,
      "step": 6018
    },
    {
      "epoch": 0.3108184869610121,
      "grad_norm": 3.7809925079345703,
      "learning_rate": 7.889859098553722e-06,
      "loss": 0.0532,
      "step": 6019
    },
    {
      "epoch": 0.310870126516912,
      "grad_norm": 12.544290542602539,
      "learning_rate": 7.889190414976733e-06,
      "loss": 0.1154,
      "step": 6020
    },
    {
      "epoch": 0.3109217660728118,
      "grad_norm": 5.867025375366211,
      "learning_rate": 7.888521653813205e-06,
      "loss": 0.0863,
      "step": 6021
    },
    {
      "epoch": 0.3109734056287116,
      "grad_norm": 3.6505653858184814,
      "learning_rate": 7.887852815081099e-06,
      "loss": 0.0583,
      "step": 6022
    },
    {
      "epoch": 0.31102504518461144,
      "grad_norm": 8.231582641601562,
      "learning_rate": 7.887183898798375e-06,
      "loss": 0.0748,
      "step": 6023
    },
    {
      "epoch": 0.31107668474051126,
      "grad_norm": 3.177259683609009,
      "learning_rate": 7.886514904982994e-06,
      "loss": 0.0628,
      "step": 6024
    },
    {
      "epoch": 0.3111283242964111,
      "grad_norm": 3.9352200031280518,
      "learning_rate": 7.885845833652923e-06,
      "loss": 0.0663,
      "step": 6025
    },
    {
      "epoch": 0.3111799638523109,
      "grad_norm": 5.0819501876831055,
      "learning_rate": 7.88517668482613e-06,
      "loss": 0.0886,
      "step": 6026
    },
    {
      "epoch": 0.3112316034082107,
      "grad_norm": 2.7886862754821777,
      "learning_rate": 7.884507458520584e-06,
      "loss": 0.0345,
      "step": 6027
    },
    {
      "epoch": 0.31128324296411053,
      "grad_norm": 7.735263824462891,
      "learning_rate": 7.883838154754256e-06,
      "loss": 0.0709,
      "step": 6028
    },
    {
      "epoch": 0.31133488252001035,
      "grad_norm": 4.680630207061768,
      "learning_rate": 7.88316877354512e-06,
      "loss": 0.0617,
      "step": 6029
    },
    {
      "epoch": 0.31138652207591017,
      "grad_norm": 4.327723503112793,
      "learning_rate": 7.882499314911151e-06,
      "loss": 0.0533,
      "step": 6030
    },
    {
      "epoch": 0.31143816163181,
      "grad_norm": 3.199408531188965,
      "learning_rate": 7.881829778870326e-06,
      "loss": 0.0526,
      "step": 6031
    },
    {
      "epoch": 0.3114898011877098,
      "grad_norm": 5.110429763793945,
      "learning_rate": 7.881160165440626e-06,
      "loss": 0.0805,
      "step": 6032
    },
    {
      "epoch": 0.3115414407436096,
      "grad_norm": 4.767116069793701,
      "learning_rate": 7.880490474640034e-06,
      "loss": 0.0643,
      "step": 6033
    },
    {
      "epoch": 0.31159308029950944,
      "grad_norm": 3.3115079402923584,
      "learning_rate": 7.879820706486532e-06,
      "loss": 0.0352,
      "step": 6034
    },
    {
      "epoch": 0.31164471985540926,
      "grad_norm": 7.459845542907715,
      "learning_rate": 7.879150860998105e-06,
      "loss": 0.0794,
      "step": 6035
    },
    {
      "epoch": 0.3116963594113091,
      "grad_norm": 6.472443580627441,
      "learning_rate": 7.878480938192745e-06,
      "loss": 0.0596,
      "step": 6036
    },
    {
      "epoch": 0.3117479989672089,
      "grad_norm": 11.027530670166016,
      "learning_rate": 7.877810938088439e-06,
      "loss": 0.1237,
      "step": 6037
    },
    {
      "epoch": 0.3117996385231087,
      "grad_norm": 2.0630557537078857,
      "learning_rate": 7.87714086070318e-06,
      "loss": 0.0633,
      "step": 6038
    },
    {
      "epoch": 0.31185127807900853,
      "grad_norm": 10.216978073120117,
      "learning_rate": 7.876470706054962e-06,
      "loss": 0.0812,
      "step": 6039
    },
    {
      "epoch": 0.31190291763490835,
      "grad_norm": 2.214843511581421,
      "learning_rate": 7.875800474161781e-06,
      "loss": 0.0373,
      "step": 6040
    },
    {
      "epoch": 0.31195455719080817,
      "grad_norm": 8.892000198364258,
      "learning_rate": 7.875130165041638e-06,
      "loss": 0.1063,
      "step": 6041
    },
    {
      "epoch": 0.312006196746708,
      "grad_norm": 2.672051191329956,
      "learning_rate": 7.874459778712529e-06,
      "loss": 0.0336,
      "step": 6042
    },
    {
      "epoch": 0.3120578363026078,
      "grad_norm": 2.5742721557617188,
      "learning_rate": 7.873789315192461e-06,
      "loss": 0.0334,
      "step": 6043
    },
    {
      "epoch": 0.3121094758585076,
      "grad_norm": 3.937128782272339,
      "learning_rate": 7.873118774499437e-06,
      "loss": 0.0803,
      "step": 6044
    },
    {
      "epoch": 0.31216111541440744,
      "grad_norm": 7.828178882598877,
      "learning_rate": 7.872448156651465e-06,
      "loss": 0.1104,
      "step": 6045
    },
    {
      "epoch": 0.31221275497030726,
      "grad_norm": 5.032127857208252,
      "learning_rate": 7.87177746166655e-06,
      "loss": 0.0468,
      "step": 6046
    },
    {
      "epoch": 0.3122643945262071,
      "grad_norm": 4.16654634475708,
      "learning_rate": 7.871106689562707e-06,
      "loss": 0.0748,
      "step": 6047
    },
    {
      "epoch": 0.3123160340821069,
      "grad_norm": 3.852125406265259,
      "learning_rate": 7.870435840357946e-06,
      "loss": 0.0437,
      "step": 6048
    },
    {
      "epoch": 0.3123676736380067,
      "grad_norm": 13.995912551879883,
      "learning_rate": 7.869764914070284e-06,
      "loss": 0.1246,
      "step": 6049
    },
    {
      "epoch": 0.31241931319390653,
      "grad_norm": 3.8358829021453857,
      "learning_rate": 7.86909391071774e-06,
      "loss": 0.058,
      "step": 6050
    },
    {
      "epoch": 0.31247095274980635,
      "grad_norm": 23.369258880615234,
      "learning_rate": 7.868422830318327e-06,
      "loss": 0.1634,
      "step": 6051
    },
    {
      "epoch": 0.31252259230570617,
      "grad_norm": 2.737889051437378,
      "learning_rate": 7.867751672890072e-06,
      "loss": 0.0469,
      "step": 6052
    },
    {
      "epoch": 0.312574231861606,
      "grad_norm": 4.713735580444336,
      "learning_rate": 7.867080438450996e-06,
      "loss": 0.0282,
      "step": 6053
    },
    {
      "epoch": 0.3126258714175058,
      "grad_norm": 11.83816909790039,
      "learning_rate": 7.866409127019124e-06,
      "loss": 0.1139,
      "step": 6054
    },
    {
      "epoch": 0.3126775109734056,
      "grad_norm": 8.063480377197266,
      "learning_rate": 7.865737738612484e-06,
      "loss": 0.1314,
      "step": 6055
    },
    {
      "epoch": 0.31272915052930544,
      "grad_norm": 6.575472354888916,
      "learning_rate": 7.865066273249108e-06,
      "loss": 0.0565,
      "step": 6056
    },
    {
      "epoch": 0.31278079008520526,
      "grad_norm": 3.872924327850342,
      "learning_rate": 7.864394730947022e-06,
      "loss": 0.0703,
      "step": 6057
    },
    {
      "epoch": 0.3128324296411051,
      "grad_norm": 7.004853248596191,
      "learning_rate": 7.863723111724264e-06,
      "loss": 0.1183,
      "step": 6058
    },
    {
      "epoch": 0.3128840691970049,
      "grad_norm": 4.4730448722839355,
      "learning_rate": 7.863051415598867e-06,
      "loss": 0.0626,
      "step": 6059
    },
    {
      "epoch": 0.3129357087529047,
      "grad_norm": 4.628721237182617,
      "learning_rate": 7.862379642588874e-06,
      "loss": 0.0745,
      "step": 6060
    },
    {
      "epoch": 0.31298734830880454,
      "grad_norm": 8.90036392211914,
      "learning_rate": 7.86170779271232e-06,
      "loss": 0.1019,
      "step": 6061
    },
    {
      "epoch": 0.31303898786470435,
      "grad_norm": 11.239219665527344,
      "learning_rate": 7.861035865987246e-06,
      "loss": 0.0978,
      "step": 6062
    },
    {
      "epoch": 0.3130906274206042,
      "grad_norm": 1.9264612197875977,
      "learning_rate": 7.860363862431702e-06,
      "loss": 0.0396,
      "step": 6063
    },
    {
      "epoch": 0.313142266976504,
      "grad_norm": 2.4016194343566895,
      "learning_rate": 7.859691782063725e-06,
      "loss": 0.0481,
      "step": 6064
    },
    {
      "epoch": 0.3131939065324038,
      "grad_norm": 5.336464881896973,
      "learning_rate": 7.859019624901371e-06,
      "loss": 0.0363,
      "step": 6065
    },
    {
      "epoch": 0.31324554608830363,
      "grad_norm": 2.7636373043060303,
      "learning_rate": 7.858347390962688e-06,
      "loss": 0.0699,
      "step": 6066
    },
    {
      "epoch": 0.31329718564420345,
      "grad_norm": 8.317276954650879,
      "learning_rate": 7.857675080265726e-06,
      "loss": 0.0972,
      "step": 6067
    },
    {
      "epoch": 0.31334882520010326,
      "grad_norm": 4.63477897644043,
      "learning_rate": 7.857002692828541e-06,
      "loss": 0.0722,
      "step": 6068
    },
    {
      "epoch": 0.3134004647560031,
      "grad_norm": 17.34699058532715,
      "learning_rate": 7.85633022866919e-06,
      "loss": 0.1134,
      "step": 6069
    },
    {
      "epoch": 0.3134521043119029,
      "grad_norm": 5.505843162536621,
      "learning_rate": 7.855657687805731e-06,
      "loss": 0.0791,
      "step": 6070
    },
    {
      "epoch": 0.3135037438678027,
      "grad_norm": 2.4393749237060547,
      "learning_rate": 7.854985070256222e-06,
      "loss": 0.0283,
      "step": 6071
    },
    {
      "epoch": 0.31355538342370254,
      "grad_norm": 3.1470396518707275,
      "learning_rate": 7.85431237603873e-06,
      "loss": 0.0511,
      "step": 6072
    },
    {
      "epoch": 0.31360702297960236,
      "grad_norm": 3.644807815551758,
      "learning_rate": 7.853639605171318e-06,
      "loss": 0.0635,
      "step": 6073
    },
    {
      "epoch": 0.3136586625355022,
      "grad_norm": 5.957269191741943,
      "learning_rate": 7.85296675767205e-06,
      "loss": 0.0884,
      "step": 6074
    },
    {
      "epoch": 0.313710302091402,
      "grad_norm": 3.917658805847168,
      "learning_rate": 7.852293833558994e-06,
      "loss": 0.0656,
      "step": 6075
    },
    {
      "epoch": 0.3137619416473018,
      "grad_norm": 3.077052116394043,
      "learning_rate": 7.851620832850227e-06,
      "loss": 0.0624,
      "step": 6076
    },
    {
      "epoch": 0.31381358120320163,
      "grad_norm": 8.217047691345215,
      "learning_rate": 7.850947755563816e-06,
      "loss": 0.0538,
      "step": 6077
    },
    {
      "epoch": 0.31386522075910145,
      "grad_norm": 3.379843235015869,
      "learning_rate": 7.85027460171784e-06,
      "loss": 0.076,
      "step": 6078
    },
    {
      "epoch": 0.31391686031500127,
      "grad_norm": 1.9615768194198608,
      "learning_rate": 7.849601371330372e-06,
      "loss": 0.0631,
      "step": 6079
    },
    {
      "epoch": 0.3139684998709011,
      "grad_norm": 4.01035213470459,
      "learning_rate": 7.848928064419494e-06,
      "loss": 0.0962,
      "step": 6080
    },
    {
      "epoch": 0.3140201394268009,
      "grad_norm": 3.4734909534454346,
      "learning_rate": 7.848254681003285e-06,
      "loss": 0.0512,
      "step": 6081
    },
    {
      "epoch": 0.3140717789827007,
      "grad_norm": 4.90303373336792,
      "learning_rate": 7.84758122109983e-06,
      "loss": 0.0778,
      "step": 6082
    },
    {
      "epoch": 0.31412341853860054,
      "grad_norm": 4.6185078620910645,
      "learning_rate": 7.846907684727211e-06,
      "loss": 0.0566,
      "step": 6083
    },
    {
      "epoch": 0.3141750580945004,
      "grad_norm": 2.474280834197998,
      "learning_rate": 7.84623407190352e-06,
      "loss": 0.057,
      "step": 6084
    },
    {
      "epoch": 0.31422669765040023,
      "grad_norm": 2.3988819122314453,
      "learning_rate": 7.845560382646841e-06,
      "loss": 0.0521,
      "step": 6085
    },
    {
      "epoch": 0.31427833720630005,
      "grad_norm": 2.6282355785369873,
      "learning_rate": 7.844886616975268e-06,
      "loss": 0.0477,
      "step": 6086
    },
    {
      "epoch": 0.31432997676219987,
      "grad_norm": 1.3107320070266724,
      "learning_rate": 7.844212774906897e-06,
      "loss": 0.0176,
      "step": 6087
    },
    {
      "epoch": 0.3143816163180997,
      "grad_norm": 3.795649766921997,
      "learning_rate": 7.843538856459818e-06,
      "loss": 0.079,
      "step": 6088
    },
    {
      "epoch": 0.3144332558739995,
      "grad_norm": 4.508023738861084,
      "learning_rate": 7.842864861652133e-06,
      "loss": 0.0629,
      "step": 6089
    },
    {
      "epoch": 0.3144848954298993,
      "grad_norm": 5.9609904289245605,
      "learning_rate": 7.842190790501938e-06,
      "loss": 0.0618,
      "step": 6090
    },
    {
      "epoch": 0.31453653498579914,
      "grad_norm": 3.9341509342193604,
      "learning_rate": 7.841516643027339e-06,
      "loss": 0.0662,
      "step": 6091
    },
    {
      "epoch": 0.31458817454169896,
      "grad_norm": 21.220802307128906,
      "learning_rate": 7.840842419246434e-06,
      "loss": 0.0985,
      "step": 6092
    },
    {
      "epoch": 0.3146398140975988,
      "grad_norm": 7.1458282470703125,
      "learning_rate": 7.840168119177331e-06,
      "loss": 0.097,
      "step": 6093
    },
    {
      "epoch": 0.3146914536534986,
      "grad_norm": 7.029621601104736,
      "learning_rate": 7.83949374283814e-06,
      "loss": 0.0961,
      "step": 6094
    },
    {
      "epoch": 0.3147430932093984,
      "grad_norm": 8.045014381408691,
      "learning_rate": 7.838819290246967e-06,
      "loss": 0.1026,
      "step": 6095
    },
    {
      "epoch": 0.31479473276529824,
      "grad_norm": 5.9125518798828125,
      "learning_rate": 7.838144761421927e-06,
      "loss": 0.0844,
      "step": 6096
    },
    {
      "epoch": 0.31484637232119805,
      "grad_norm": 7.244169235229492,
      "learning_rate": 7.837470156381132e-06,
      "loss": 0.1098,
      "step": 6097
    },
    {
      "epoch": 0.31489801187709787,
      "grad_norm": 3.5805773735046387,
      "learning_rate": 7.836795475142699e-06,
      "loss": 0.0744,
      "step": 6098
    },
    {
      "epoch": 0.3149496514329977,
      "grad_norm": 8.611689567565918,
      "learning_rate": 7.836120717724744e-06,
      "loss": 0.0932,
      "step": 6099
    },
    {
      "epoch": 0.3150012909888975,
      "grad_norm": 16.960966110229492,
      "learning_rate": 7.83544588414539e-06,
      "loss": 0.1244,
      "step": 6100
    },
    {
      "epoch": 0.3150529305447973,
      "grad_norm": 2.8485684394836426,
      "learning_rate": 7.834770974422757e-06,
      "loss": 0.04,
      "step": 6101
    },
    {
      "epoch": 0.31510457010069715,
      "grad_norm": 11.265310287475586,
      "learning_rate": 7.83409598857497e-06,
      "loss": 0.1767,
      "step": 6102
    },
    {
      "epoch": 0.31515620965659696,
      "grad_norm": 12.493687629699707,
      "learning_rate": 7.833420926620153e-06,
      "loss": 0.1627,
      "step": 6103
    },
    {
      "epoch": 0.3152078492124968,
      "grad_norm": 4.710011959075928,
      "learning_rate": 7.832745788576436e-06,
      "loss": 0.0784,
      "step": 6104
    },
    {
      "epoch": 0.3152594887683966,
      "grad_norm": 2.4529576301574707,
      "learning_rate": 7.83207057446195e-06,
      "loss": 0.0589,
      "step": 6105
    },
    {
      "epoch": 0.3153111283242964,
      "grad_norm": 3.886751890182495,
      "learning_rate": 7.831395284294825e-06,
      "loss": 0.0467,
      "step": 6106
    },
    {
      "epoch": 0.31536276788019624,
      "grad_norm": 3.283874988555908,
      "learning_rate": 7.830719918093197e-06,
      "loss": 0.0786,
      "step": 6107
    },
    {
      "epoch": 0.31541440743609606,
      "grad_norm": 2.0783753395080566,
      "learning_rate": 7.830044475875202e-06,
      "loss": 0.0538,
      "step": 6108
    },
    {
      "epoch": 0.3154660469919959,
      "grad_norm": 2.5125410556793213,
      "learning_rate": 7.829368957658978e-06,
      "loss": 0.0517,
      "step": 6109
    },
    {
      "epoch": 0.3155176865478957,
      "grad_norm": 3.570662260055542,
      "learning_rate": 7.828693363462665e-06,
      "loss": 0.0477,
      "step": 6110
    },
    {
      "epoch": 0.3155693261037955,
      "grad_norm": 8.05401611328125,
      "learning_rate": 7.828017693304408e-06,
      "loss": 0.0917,
      "step": 6111
    },
    {
      "epoch": 0.31562096565969533,
      "grad_norm": 3.3707423210144043,
      "learning_rate": 7.827341947202346e-06,
      "loss": 0.0894,
      "step": 6112
    },
    {
      "epoch": 0.31567260521559515,
      "grad_norm": 3.510382890701294,
      "learning_rate": 7.826666125174633e-06,
      "loss": 0.0896,
      "step": 6113
    },
    {
      "epoch": 0.31572424477149497,
      "grad_norm": 4.887164115905762,
      "learning_rate": 7.825990227239413e-06,
      "loss": 0.0547,
      "step": 6114
    },
    {
      "epoch": 0.3157758843273948,
      "grad_norm": 3.1626293659210205,
      "learning_rate": 7.825314253414835e-06,
      "loss": 0.0334,
      "step": 6115
    },
    {
      "epoch": 0.3158275238832946,
      "grad_norm": 4.932890892028809,
      "learning_rate": 7.824638203719057e-06,
      "loss": 0.0998,
      "step": 6116
    },
    {
      "epoch": 0.3158791634391944,
      "grad_norm": 2.729999542236328,
      "learning_rate": 7.823962078170229e-06,
      "loss": 0.0434,
      "step": 6117
    },
    {
      "epoch": 0.31593080299509424,
      "grad_norm": 7.20080041885376,
      "learning_rate": 7.823285876786508e-06,
      "loss": 0.1036,
      "step": 6118
    },
    {
      "epoch": 0.31598244255099406,
      "grad_norm": 10.4234619140625,
      "learning_rate": 7.822609599586055e-06,
      "loss": 0.1263,
      "step": 6119
    },
    {
      "epoch": 0.3160340821068939,
      "grad_norm": 8.847350120544434,
      "learning_rate": 7.821933246587033e-06,
      "loss": 0.0476,
      "step": 6120
    },
    {
      "epoch": 0.3160857216627937,
      "grad_norm": 5.184169769287109,
      "learning_rate": 7.821256817807597e-06,
      "loss": 0.0448,
      "step": 6121
    },
    {
      "epoch": 0.3161373612186935,
      "grad_norm": 4.7308173179626465,
      "learning_rate": 7.82058031326592e-06,
      "loss": 0.0457,
      "step": 6122
    },
    {
      "epoch": 0.31618900077459333,
      "grad_norm": 4.904362678527832,
      "learning_rate": 7.819903732980164e-06,
      "loss": 0.0601,
      "step": 6123
    },
    {
      "epoch": 0.31624064033049315,
      "grad_norm": 4.2893829345703125,
      "learning_rate": 7.8192270769685e-06,
      "loss": 0.0765,
      "step": 6124
    },
    {
      "epoch": 0.31629227988639297,
      "grad_norm": 4.5839009284973145,
      "learning_rate": 7.818550345249098e-06,
      "loss": 0.0479,
      "step": 6125
    },
    {
      "epoch": 0.3163439194422928,
      "grad_norm": 7.855040073394775,
      "learning_rate": 7.817873537840132e-06,
      "loss": 0.075,
      "step": 6126
    },
    {
      "epoch": 0.3163955589981926,
      "grad_norm": 5.117353439331055,
      "learning_rate": 7.817196654759776e-06,
      "loss": 0.0401,
      "step": 6127
    },
    {
      "epoch": 0.3164471985540924,
      "grad_norm": 3.3526997566223145,
      "learning_rate": 7.816519696026208e-06,
      "loss": 0.0749,
      "step": 6128
    },
    {
      "epoch": 0.31649883810999224,
      "grad_norm": 6.976875305175781,
      "learning_rate": 7.815842661657607e-06,
      "loss": 0.0759,
      "step": 6129
    },
    {
      "epoch": 0.31655047766589206,
      "grad_norm": 4.695199489593506,
      "learning_rate": 7.815165551672152e-06,
      "loss": 0.0328,
      "step": 6130
    },
    {
      "epoch": 0.3166021172217919,
      "grad_norm": 4.547394275665283,
      "learning_rate": 7.81448836608803e-06,
      "loss": 0.0396,
      "step": 6131
    },
    {
      "epoch": 0.3166537567776917,
      "grad_norm": 5.599379062652588,
      "learning_rate": 7.813811104923423e-06,
      "loss": 0.0603,
      "step": 6132
    },
    {
      "epoch": 0.3167053963335915,
      "grad_norm": 14.977056503295898,
      "learning_rate": 7.81313376819652e-06,
      "loss": 0.0935,
      "step": 6133
    },
    {
      "epoch": 0.31675703588949133,
      "grad_norm": 6.784121036529541,
      "learning_rate": 7.81245635592551e-06,
      "loss": 0.0816,
      "step": 6134
    },
    {
      "epoch": 0.31680867544539115,
      "grad_norm": 2.684401035308838,
      "learning_rate": 7.811778868128584e-06,
      "loss": 0.0397,
      "step": 6135
    },
    {
      "epoch": 0.31686031500129097,
      "grad_norm": 10.72828483581543,
      "learning_rate": 7.811101304823934e-06,
      "loss": 0.1263,
      "step": 6136
    },
    {
      "epoch": 0.3169119545571908,
      "grad_norm": 7.443442344665527,
      "learning_rate": 7.810423666029755e-06,
      "loss": 0.1125,
      "step": 6137
    },
    {
      "epoch": 0.3169635941130906,
      "grad_norm": 4.927013397216797,
      "learning_rate": 7.809745951764248e-06,
      "loss": 0.0781,
      "step": 6138
    },
    {
      "epoch": 0.3170152336689904,
      "grad_norm": 3.1944210529327393,
      "learning_rate": 7.809068162045613e-06,
      "loss": 0.069,
      "step": 6139
    },
    {
      "epoch": 0.31706687322489024,
      "grad_norm": 9.013153076171875,
      "learning_rate": 7.808390296892045e-06,
      "loss": 0.0921,
      "step": 6140
    },
    {
      "epoch": 0.31711851278079006,
      "grad_norm": 21.685440063476562,
      "learning_rate": 7.80771235632175e-06,
      "loss": 0.0685,
      "step": 6141
    },
    {
      "epoch": 0.3171701523366899,
      "grad_norm": 5.585023880004883,
      "learning_rate": 7.807034340352939e-06,
      "loss": 0.061,
      "step": 6142
    },
    {
      "epoch": 0.3172217918925897,
      "grad_norm": 2.040797472000122,
      "learning_rate": 7.806356249003812e-06,
      "loss": 0.02,
      "step": 6143
    },
    {
      "epoch": 0.3172734314484895,
      "grad_norm": 4.36867618560791,
      "learning_rate": 7.805678082292583e-06,
      "loss": 0.0677,
      "step": 6144
    },
    {
      "epoch": 0.31732507100438934,
      "grad_norm": 3.3419508934020996,
      "learning_rate": 7.804999840237462e-06,
      "loss": 0.0452,
      "step": 6145
    },
    {
      "epoch": 0.31737671056028915,
      "grad_norm": 2.1470558643341064,
      "learning_rate": 7.804321522856662e-06,
      "loss": 0.0184,
      "step": 6146
    },
    {
      "epoch": 0.317428350116189,
      "grad_norm": 9.193568229675293,
      "learning_rate": 7.8036431301684e-06,
      "loss": 0.1121,
      "step": 6147
    },
    {
      "epoch": 0.31747998967208885,
      "grad_norm": 2.326460123062134,
      "learning_rate": 7.802964662190893e-06,
      "loss": 0.0412,
      "step": 6148
    },
    {
      "epoch": 0.31753162922798867,
      "grad_norm": 4.685604095458984,
      "learning_rate": 7.802286118942359e-06,
      "loss": 0.0442,
      "step": 6149
    },
    {
      "epoch": 0.3175832687838885,
      "grad_norm": 5.295156955718994,
      "learning_rate": 7.801607500441023e-06,
      "loss": 0.0423,
      "step": 6150
    },
    {
      "epoch": 0.3176349083397883,
      "grad_norm": 9.424604415893555,
      "learning_rate": 7.800928806705107e-06,
      "loss": 0.0926,
      "step": 6151
    },
    {
      "epoch": 0.3176865478956881,
      "grad_norm": 5.451590538024902,
      "learning_rate": 7.800250037752835e-06,
      "loss": 0.0756,
      "step": 6152
    },
    {
      "epoch": 0.31773818745158794,
      "grad_norm": 14.82713794708252,
      "learning_rate": 7.799571193602438e-06,
      "loss": 0.1813,
      "step": 6153
    },
    {
      "epoch": 0.31778982700748776,
      "grad_norm": 4.162665843963623,
      "learning_rate": 7.798892274272142e-06,
      "loss": 0.0328,
      "step": 6154
    },
    {
      "epoch": 0.3178414665633876,
      "grad_norm": 7.186397075653076,
      "learning_rate": 7.798213279780183e-06,
      "loss": 0.0864,
      "step": 6155
    },
    {
      "epoch": 0.3178931061192874,
      "grad_norm": 21.691007614135742,
      "learning_rate": 7.797534210144792e-06,
      "loss": 0.0649,
      "step": 6156
    },
    {
      "epoch": 0.3179447456751872,
      "grad_norm": 4.883171081542969,
      "learning_rate": 7.796855065384206e-06,
      "loss": 0.0454,
      "step": 6157
    },
    {
      "epoch": 0.31799638523108703,
      "grad_norm": 9.668641090393066,
      "learning_rate": 7.79617584551666e-06,
      "loss": 0.1384,
      "step": 6158
    },
    {
      "epoch": 0.31804802478698685,
      "grad_norm": 13.40261173248291,
      "learning_rate": 7.795496550560397e-06,
      "loss": 0.2664,
      "step": 6159
    },
    {
      "epoch": 0.31809966434288667,
      "grad_norm": 7.997302055358887,
      "learning_rate": 7.794817180533658e-06,
      "loss": 0.068,
      "step": 6160
    },
    {
      "epoch": 0.3181513038987865,
      "grad_norm": 5.7743940353393555,
      "learning_rate": 7.794137735454686e-06,
      "loss": 0.0796,
      "step": 6161
    },
    {
      "epoch": 0.3182029434546863,
      "grad_norm": 13.392386436462402,
      "learning_rate": 7.79345821534173e-06,
      "loss": 0.0978,
      "step": 6162
    },
    {
      "epoch": 0.3182545830105861,
      "grad_norm": 7.148099899291992,
      "learning_rate": 7.792778620213032e-06,
      "loss": 0.1663,
      "step": 6163
    },
    {
      "epoch": 0.31830622256648594,
      "grad_norm": 3.742288589477539,
      "learning_rate": 7.792098950086847e-06,
      "loss": 0.0672,
      "step": 6164
    },
    {
      "epoch": 0.31835786212238576,
      "grad_norm": 7.799104690551758,
      "learning_rate": 7.791419204981426e-06,
      "loss": 0.0687,
      "step": 6165
    },
    {
      "epoch": 0.3184095016782856,
      "grad_norm": 5.34528923034668,
      "learning_rate": 7.79073938491502e-06,
      "loss": 0.146,
      "step": 6166
    },
    {
      "epoch": 0.3184611412341854,
      "grad_norm": 7.001514911651611,
      "learning_rate": 7.790059489905887e-06,
      "loss": 0.0877,
      "step": 6167
    },
    {
      "epoch": 0.3185127807900852,
      "grad_norm": 3.150468111038208,
      "learning_rate": 7.789379519972289e-06,
      "loss": 0.0635,
      "step": 6168
    },
    {
      "epoch": 0.31856442034598503,
      "grad_norm": 1.788003921508789,
      "learning_rate": 7.788699475132479e-06,
      "loss": 0.0348,
      "step": 6169
    },
    {
      "epoch": 0.31861605990188485,
      "grad_norm": 2.9109838008880615,
      "learning_rate": 7.78801935540472e-06,
      "loss": 0.0695,
      "step": 6170
    },
    {
      "epoch": 0.31866769945778467,
      "grad_norm": 2.156999111175537,
      "learning_rate": 7.787339160807283e-06,
      "loss": 0.0259,
      "step": 6171
    },
    {
      "epoch": 0.3187193390136845,
      "grad_norm": 13.028997421264648,
      "learning_rate": 7.786658891358425e-06,
      "loss": 0.0635,
      "step": 6172
    },
    {
      "epoch": 0.3187709785695843,
      "grad_norm": 5.176676273345947,
      "learning_rate": 7.78597854707642e-06,
      "loss": 0.0751,
      "step": 6173
    },
    {
      "epoch": 0.3188226181254841,
      "grad_norm": 3.1052093505859375,
      "learning_rate": 7.785298127979535e-06,
      "loss": 0.0472,
      "step": 6174
    },
    {
      "epoch": 0.31887425768138394,
      "grad_norm": 5.750711917877197,
      "learning_rate": 7.784617634086044e-06,
      "loss": 0.1201,
      "step": 6175
    },
    {
      "epoch": 0.31892589723728376,
      "grad_norm": 4.316614627838135,
      "learning_rate": 7.783937065414218e-06,
      "loss": 0.0385,
      "step": 6176
    },
    {
      "epoch": 0.3189775367931836,
      "grad_norm": 5.73007869720459,
      "learning_rate": 7.783256421982336e-06,
      "loss": 0.0701,
      "step": 6177
    },
    {
      "epoch": 0.3190291763490834,
      "grad_norm": 8.10039234161377,
      "learning_rate": 7.782575703808677e-06,
      "loss": 0.0658,
      "step": 6178
    },
    {
      "epoch": 0.3190808159049832,
      "grad_norm": 3.269991874694824,
      "learning_rate": 7.781894910911518e-06,
      "loss": 0.0838,
      "step": 6179
    },
    {
      "epoch": 0.31913245546088304,
      "grad_norm": 4.783228874206543,
      "learning_rate": 7.781214043309143e-06,
      "loss": 0.0663,
      "step": 6180
    },
    {
      "epoch": 0.31918409501678285,
      "grad_norm": 3.3677947521209717,
      "learning_rate": 7.780533101019832e-06,
      "loss": 0.0636,
      "step": 6181
    },
    {
      "epoch": 0.3192357345726827,
      "grad_norm": 3.2336857318878174,
      "learning_rate": 7.779852084061878e-06,
      "loss": 0.0136,
      "step": 6182
    },
    {
      "epoch": 0.3192873741285825,
      "grad_norm": 1.3898268938064575,
      "learning_rate": 7.779170992453562e-06,
      "loss": 0.0159,
      "step": 6183
    },
    {
      "epoch": 0.3193390136844823,
      "grad_norm": 2.1618165969848633,
      "learning_rate": 7.77848982621318e-06,
      "loss": 0.0268,
      "step": 6184
    },
    {
      "epoch": 0.3193906532403821,
      "grad_norm": 3.2335431575775146,
      "learning_rate": 7.777808585359022e-06,
      "loss": 0.0915,
      "step": 6185
    },
    {
      "epoch": 0.31944229279628195,
      "grad_norm": 3.607724666595459,
      "learning_rate": 7.777127269909382e-06,
      "loss": 0.0726,
      "step": 6186
    },
    {
      "epoch": 0.31949393235218176,
      "grad_norm": 7.027190208435059,
      "learning_rate": 7.776445879882556e-06,
      "loss": 0.0907,
      "step": 6187
    },
    {
      "epoch": 0.3195455719080816,
      "grad_norm": 4.462314605712891,
      "learning_rate": 7.77576441529684e-06,
      "loss": 0.0529,
      "step": 6188
    },
    {
      "epoch": 0.3195972114639814,
      "grad_norm": 2.9171271324157715,
      "learning_rate": 7.775082876170537e-06,
      "loss": 0.0595,
      "step": 6189
    },
    {
      "epoch": 0.3196488510198812,
      "grad_norm": 2.3392395973205566,
      "learning_rate": 7.774401262521949e-06,
      "loss": 0.0166,
      "step": 6190
    },
    {
      "epoch": 0.31970049057578104,
      "grad_norm": 3.4368515014648438,
      "learning_rate": 7.773719574369379e-06,
      "loss": 0.0668,
      "step": 6191
    },
    {
      "epoch": 0.31975213013168086,
      "grad_norm": 12.637438774108887,
      "learning_rate": 7.773037811731133e-06,
      "loss": 0.1556,
      "step": 6192
    },
    {
      "epoch": 0.3198037696875807,
      "grad_norm": 2.471923351287842,
      "learning_rate": 7.772355974625519e-06,
      "loss": 0.0344,
      "step": 6193
    },
    {
      "epoch": 0.3198554092434805,
      "grad_norm": 5.516467094421387,
      "learning_rate": 7.771674063070847e-06,
      "loss": 0.069,
      "step": 6194
    },
    {
      "epoch": 0.3199070487993803,
      "grad_norm": 6.925252914428711,
      "learning_rate": 7.770992077085431e-06,
      "loss": 0.0964,
      "step": 6195
    },
    {
      "epoch": 0.31995868835528013,
      "grad_norm": 3.9700348377227783,
      "learning_rate": 7.770310016687584e-06,
      "loss": 0.0716,
      "step": 6196
    },
    {
      "epoch": 0.32001032791117995,
      "grad_norm": 7.173914432525635,
      "learning_rate": 7.76962788189562e-06,
      "loss": 0.1033,
      "step": 6197
    },
    {
      "epoch": 0.32006196746707977,
      "grad_norm": 4.260102272033691,
      "learning_rate": 7.76894567272786e-06,
      "loss": 0.0668,
      "step": 6198
    },
    {
      "epoch": 0.3201136070229796,
      "grad_norm": 2.9716384410858154,
      "learning_rate": 7.768263389202625e-06,
      "loss": 0.0416,
      "step": 6199
    },
    {
      "epoch": 0.3201652465788794,
      "grad_norm": 3.927818775177002,
      "learning_rate": 7.767581031338233e-06,
      "loss": 0.013,
      "step": 6200
    },
    {
      "epoch": 0.3202168861347792,
      "grad_norm": 3.676056146621704,
      "learning_rate": 7.76689859915301e-06,
      "loss": 0.0628,
      "step": 6201
    },
    {
      "epoch": 0.32026852569067904,
      "grad_norm": 3.2567250728607178,
      "learning_rate": 7.766216092665285e-06,
      "loss": 0.0358,
      "step": 6202
    },
    {
      "epoch": 0.32032016524657886,
      "grad_norm": 4.962523937225342,
      "learning_rate": 7.765533511893381e-06,
      "loss": 0.0792,
      "step": 6203
    },
    {
      "epoch": 0.3203718048024787,
      "grad_norm": 3.1882543563842773,
      "learning_rate": 7.764850856855632e-06,
      "loss": 0.044,
      "step": 6204
    },
    {
      "epoch": 0.3204234443583785,
      "grad_norm": 4.577144145965576,
      "learning_rate": 7.764168127570367e-06,
      "loss": 0.0396,
      "step": 6205
    },
    {
      "epoch": 0.3204750839142783,
      "grad_norm": 4.259387493133545,
      "learning_rate": 7.763485324055922e-06,
      "loss": 0.0707,
      "step": 6206
    },
    {
      "epoch": 0.32052672347017813,
      "grad_norm": 3.9986653327941895,
      "learning_rate": 7.762802446330634e-06,
      "loss": 0.0698,
      "step": 6207
    },
    {
      "epoch": 0.32057836302607795,
      "grad_norm": 6.92335319519043,
      "learning_rate": 7.762119494412838e-06,
      "loss": 0.0798,
      "step": 6208
    },
    {
      "epoch": 0.32063000258197777,
      "grad_norm": 2.1520910263061523,
      "learning_rate": 7.761436468320878e-06,
      "loss": 0.018,
      "step": 6209
    },
    {
      "epoch": 0.3206816421378776,
      "grad_norm": 4.305610179901123,
      "learning_rate": 7.760753368073092e-06,
      "loss": 0.0653,
      "step": 6210
    },
    {
      "epoch": 0.32073328169377746,
      "grad_norm": 8.043476104736328,
      "learning_rate": 7.760070193687826e-06,
      "loss": 0.1291,
      "step": 6211
    },
    {
      "epoch": 0.3207849212496773,
      "grad_norm": 10.68597412109375,
      "learning_rate": 7.759386945183428e-06,
      "loss": 0.1264,
      "step": 6212
    },
    {
      "epoch": 0.3208365608055771,
      "grad_norm": 5.198373794555664,
      "learning_rate": 7.758703622578241e-06,
      "loss": 0.0731,
      "step": 6213
    },
    {
      "epoch": 0.3208882003614769,
      "grad_norm": 17.209869384765625,
      "learning_rate": 7.75802022589062e-06,
      "loss": 0.2699,
      "step": 6214
    },
    {
      "epoch": 0.32093983991737673,
      "grad_norm": 4.609441757202148,
      "learning_rate": 7.757336755138913e-06,
      "loss": 0.0708,
      "step": 6215
    },
    {
      "epoch": 0.32099147947327655,
      "grad_norm": 3.1314451694488525,
      "learning_rate": 7.756653210341475e-06,
      "loss": 0.0345,
      "step": 6216
    },
    {
      "epoch": 0.32104311902917637,
      "grad_norm": 3.8408570289611816,
      "learning_rate": 7.755969591516665e-06,
      "loss": 0.0585,
      "step": 6217
    },
    {
      "epoch": 0.3210947585850762,
      "grad_norm": 6.194389820098877,
      "learning_rate": 7.75528589868284e-06,
      "loss": 0.0817,
      "step": 6218
    },
    {
      "epoch": 0.321146398140976,
      "grad_norm": 3.6240954399108887,
      "learning_rate": 7.754602131858355e-06,
      "loss": 0.0807,
      "step": 6219
    },
    {
      "epoch": 0.3211980376968758,
      "grad_norm": 6.563910484313965,
      "learning_rate": 7.75391829106158e-06,
      "loss": 0.108,
      "step": 6220
    },
    {
      "epoch": 0.32124967725277565,
      "grad_norm": 4.935878276824951,
      "learning_rate": 7.753234376310872e-06,
      "loss": 0.0586,
      "step": 6221
    },
    {
      "epoch": 0.32130131680867546,
      "grad_norm": 10.596182823181152,
      "learning_rate": 7.752550387624599e-06,
      "loss": 0.1265,
      "step": 6222
    },
    {
      "epoch": 0.3213529563645753,
      "grad_norm": 1.7216448783874512,
      "learning_rate": 7.75186632502113e-06,
      "loss": 0.0275,
      "step": 6223
    },
    {
      "epoch": 0.3214045959204751,
      "grad_norm": 8.854488372802734,
      "learning_rate": 7.751182188518833e-06,
      "loss": 0.0632,
      "step": 6224
    },
    {
      "epoch": 0.3214562354763749,
      "grad_norm": 3.982954978942871,
      "learning_rate": 7.750497978136082e-06,
      "loss": 0.0766,
      "step": 6225
    },
    {
      "epoch": 0.32150787503227474,
      "grad_norm": 5.888963222503662,
      "learning_rate": 7.749813693891251e-06,
      "loss": 0.0987,
      "step": 6226
    },
    {
      "epoch": 0.32155951458817456,
      "grad_norm": 5.083281993865967,
      "learning_rate": 7.749129335802715e-06,
      "loss": 0.1021,
      "step": 6227
    },
    {
      "epoch": 0.3216111541440744,
      "grad_norm": 7.19161319732666,
      "learning_rate": 7.748444903888849e-06,
      "loss": 0.0593,
      "step": 6228
    },
    {
      "epoch": 0.3216627936999742,
      "grad_norm": 13.8712797164917,
      "learning_rate": 7.747760398168036e-06,
      "loss": 0.1393,
      "step": 6229
    },
    {
      "epoch": 0.321714433255874,
      "grad_norm": 10.644060134887695,
      "learning_rate": 7.747075818658659e-06,
      "loss": 0.1302,
      "step": 6230
    },
    {
      "epoch": 0.32176607281177383,
      "grad_norm": 7.9129958152771,
      "learning_rate": 7.746391165379096e-06,
      "loss": 0.085,
      "step": 6231
    },
    {
      "epoch": 0.32181771236767365,
      "grad_norm": 4.647531509399414,
      "learning_rate": 7.74570643834774e-06,
      "loss": 0.085,
      "step": 6232
    },
    {
      "epoch": 0.32186935192357347,
      "grad_norm": 9.597574234008789,
      "learning_rate": 7.745021637582974e-06,
      "loss": 0.1389,
      "step": 6233
    },
    {
      "epoch": 0.3219209914794733,
      "grad_norm": 2.9272677898406982,
      "learning_rate": 7.74433676310319e-06,
      "loss": 0.0441,
      "step": 6234
    },
    {
      "epoch": 0.3219726310353731,
      "grad_norm": 2.403909206390381,
      "learning_rate": 7.743651814926778e-06,
      "loss": 0.037,
      "step": 6235
    },
    {
      "epoch": 0.3220242705912729,
      "grad_norm": 5.55336332321167,
      "learning_rate": 7.742966793072131e-06,
      "loss": 0.0931,
      "step": 6236
    },
    {
      "epoch": 0.32207591014717274,
      "grad_norm": 2.5557854175567627,
      "learning_rate": 7.742281697557648e-06,
      "loss": 0.0516,
      "step": 6237
    },
    {
      "epoch": 0.32212754970307256,
      "grad_norm": 3.8194494247436523,
      "learning_rate": 7.741596528401722e-06,
      "loss": 0.0951,
      "step": 6238
    },
    {
      "epoch": 0.3221791892589724,
      "grad_norm": 6.458088397979736,
      "learning_rate": 7.740911285622756e-06,
      "loss": 0.0915,
      "step": 6239
    },
    {
      "epoch": 0.3222308288148722,
      "grad_norm": 14.378085136413574,
      "learning_rate": 7.740225969239152e-06,
      "loss": 0.0641,
      "step": 6240
    },
    {
      "epoch": 0.322282468370772,
      "grad_norm": 12.560282707214355,
      "learning_rate": 7.73954057926931e-06,
      "loss": 0.0674,
      "step": 6241
    },
    {
      "epoch": 0.32233410792667183,
      "grad_norm": 18.930070877075195,
      "learning_rate": 7.73885511573164e-06,
      "loss": 0.0995,
      "step": 6242
    },
    {
      "epoch": 0.32238574748257165,
      "grad_norm": 3.71164870262146,
      "learning_rate": 7.738169578644546e-06,
      "loss": 0.0468,
      "step": 6243
    },
    {
      "epoch": 0.32243738703847147,
      "grad_norm": 2.3336379528045654,
      "learning_rate": 7.737483968026439e-06,
      "loss": 0.0478,
      "step": 6244
    },
    {
      "epoch": 0.3224890265943713,
      "grad_norm": 11.321380615234375,
      "learning_rate": 7.73679828389573e-06,
      "loss": 0.0822,
      "step": 6245
    },
    {
      "epoch": 0.3225406661502711,
      "grad_norm": 11.944036483764648,
      "learning_rate": 7.736112526270833e-06,
      "loss": 0.1042,
      "step": 6246
    },
    {
      "epoch": 0.3225923057061709,
      "grad_norm": 5.389405250549316,
      "learning_rate": 7.735426695170162e-06,
      "loss": 0.0922,
      "step": 6247
    },
    {
      "epoch": 0.32264394526207074,
      "grad_norm": 14.123550415039062,
      "learning_rate": 7.734740790612137e-06,
      "loss": 0.1772,
      "step": 6248
    },
    {
      "epoch": 0.32269558481797056,
      "grad_norm": 2.4463672637939453,
      "learning_rate": 7.734054812615173e-06,
      "loss": 0.0473,
      "step": 6249
    },
    {
      "epoch": 0.3227472243738704,
      "grad_norm": 4.267626762390137,
      "learning_rate": 7.733368761197695e-06,
      "loss": 0.0497,
      "step": 6250
    },
    {
      "epoch": 0.3227988639297702,
      "grad_norm": 1.9212276935577393,
      "learning_rate": 7.732682636378124e-06,
      "loss": 0.04,
      "step": 6251
    },
    {
      "epoch": 0.32285050348567,
      "grad_norm": 4.46858549118042,
      "learning_rate": 7.731996438174888e-06,
      "loss": 0.0693,
      "step": 6252
    },
    {
      "epoch": 0.32290214304156983,
      "grad_norm": 5.021968841552734,
      "learning_rate": 7.731310166606411e-06,
      "loss": 0.104,
      "step": 6253
    },
    {
      "epoch": 0.32295378259746965,
      "grad_norm": 3.9584386348724365,
      "learning_rate": 7.730623821691127e-06,
      "loss": 0.0809,
      "step": 6254
    },
    {
      "epoch": 0.32300542215336947,
      "grad_norm": 16.06856918334961,
      "learning_rate": 7.729937403447463e-06,
      "loss": 0.1253,
      "step": 6255
    },
    {
      "epoch": 0.3230570617092693,
      "grad_norm": 5.37358283996582,
      "learning_rate": 7.72925091189385e-06,
      "loss": 0.0543,
      "step": 6256
    },
    {
      "epoch": 0.3231087012651691,
      "grad_norm": 9.575571060180664,
      "learning_rate": 7.728564347048728e-06,
      "loss": 0.1145,
      "step": 6257
    },
    {
      "epoch": 0.3231603408210689,
      "grad_norm": 5.708923816680908,
      "learning_rate": 7.727877708930534e-06,
      "loss": 0.0643,
      "step": 6258
    },
    {
      "epoch": 0.32321198037696874,
      "grad_norm": 3.550898790359497,
      "learning_rate": 7.727190997557702e-06,
      "loss": 0.0318,
      "step": 6259
    },
    {
      "epoch": 0.32326361993286856,
      "grad_norm": 3.7981603145599365,
      "learning_rate": 7.72650421294868e-06,
      "loss": 0.079,
      "step": 6260
    },
    {
      "epoch": 0.3233152594887684,
      "grad_norm": 7.497217655181885,
      "learning_rate": 7.725817355121905e-06,
      "loss": 0.0693,
      "step": 6261
    },
    {
      "epoch": 0.3233668990446682,
      "grad_norm": 4.685092449188232,
      "learning_rate": 7.725130424095824e-06,
      "loss": 0.0527,
      "step": 6262
    },
    {
      "epoch": 0.323418538600568,
      "grad_norm": 2.4289402961730957,
      "learning_rate": 7.724443419888885e-06,
      "loss": 0.0199,
      "step": 6263
    },
    {
      "epoch": 0.32347017815646784,
      "grad_norm": 14.73749828338623,
      "learning_rate": 7.723756342519536e-06,
      "loss": 0.1785,
      "step": 6264
    },
    {
      "epoch": 0.32352181771236765,
      "grad_norm": 7.160696029663086,
      "learning_rate": 7.723069192006227e-06,
      "loss": 0.0586,
      "step": 6265
    },
    {
      "epoch": 0.3235734572682675,
      "grad_norm": 4.3539509773254395,
      "learning_rate": 7.722381968367414e-06,
      "loss": 0.0817,
      "step": 6266
    },
    {
      "epoch": 0.3236250968241673,
      "grad_norm": 2.866434097290039,
      "learning_rate": 7.721694671621548e-06,
      "loss": 0.0978,
      "step": 6267
    },
    {
      "epoch": 0.3236767363800671,
      "grad_norm": 15.878840446472168,
      "learning_rate": 7.721007301787085e-06,
      "loss": 0.0635,
      "step": 6268
    },
    {
      "epoch": 0.3237283759359669,
      "grad_norm": 6.113029479980469,
      "learning_rate": 7.720319858882487e-06,
      "loss": 0.0884,
      "step": 6269
    },
    {
      "epoch": 0.32378001549186675,
      "grad_norm": 6.108445644378662,
      "learning_rate": 7.719632342926214e-06,
      "loss": 0.1023,
      "step": 6270
    },
    {
      "epoch": 0.32383165504776656,
      "grad_norm": 3.2049593925476074,
      "learning_rate": 7.718944753936729e-06,
      "loss": 0.0611,
      "step": 6271
    },
    {
      "epoch": 0.3238832946036664,
      "grad_norm": 19.4457950592041,
      "learning_rate": 7.718257091932493e-06,
      "loss": 0.1266,
      "step": 6272
    },
    {
      "epoch": 0.3239349341595662,
      "grad_norm": 3.465510368347168,
      "learning_rate": 7.717569356931975e-06,
      "loss": 0.0306,
      "step": 6273
    },
    {
      "epoch": 0.323986573715466,
      "grad_norm": 2.867295980453491,
      "learning_rate": 7.716881548953646e-06,
      "loss": 0.0338,
      "step": 6274
    },
    {
      "epoch": 0.3240382132713659,
      "grad_norm": 10.29613971710205,
      "learning_rate": 7.716193668015973e-06,
      "loss": 0.1715,
      "step": 6275
    },
    {
      "epoch": 0.3240898528272657,
      "grad_norm": 3.5860140323638916,
      "learning_rate": 7.71550571413743e-06,
      "loss": 0.0603,
      "step": 6276
    },
    {
      "epoch": 0.32414149238316553,
      "grad_norm": 3.587124824523926,
      "learning_rate": 7.714817687336489e-06,
      "loss": 0.0934,
      "step": 6277
    },
    {
      "epoch": 0.32419313193906535,
      "grad_norm": 10.746783256530762,
      "learning_rate": 7.714129587631628e-06,
      "loss": 0.1189,
      "step": 6278
    },
    {
      "epoch": 0.32424477149496517,
      "grad_norm": 7.120670318603516,
      "learning_rate": 7.713441415041326e-06,
      "loss": 0.0773,
      "step": 6279
    },
    {
      "epoch": 0.324296411050865,
      "grad_norm": 4.271682262420654,
      "learning_rate": 7.712753169584065e-06,
      "loss": 0.091,
      "step": 6280
    },
    {
      "epoch": 0.3243480506067648,
      "grad_norm": 3.634493589401245,
      "learning_rate": 7.712064851278322e-06,
      "loss": 0.0668,
      "step": 6281
    },
    {
      "epoch": 0.3243996901626646,
      "grad_norm": 2.1252450942993164,
      "learning_rate": 7.711376460142586e-06,
      "loss": 0.0225,
      "step": 6282
    },
    {
      "epoch": 0.32445132971856444,
      "grad_norm": 5.397585391998291,
      "learning_rate": 7.710687996195339e-06,
      "loss": 0.0712,
      "step": 6283
    },
    {
      "epoch": 0.32450296927446426,
      "grad_norm": 2.3792612552642822,
      "learning_rate": 7.709999459455073e-06,
      "loss": 0.041,
      "step": 6284
    },
    {
      "epoch": 0.3245546088303641,
      "grad_norm": 2.31990122795105,
      "learning_rate": 7.709310849940274e-06,
      "loss": 0.0596,
      "step": 6285
    },
    {
      "epoch": 0.3246062483862639,
      "grad_norm": 7.616552352905273,
      "learning_rate": 7.708622167669438e-06,
      "loss": 0.1494,
      "step": 6286
    },
    {
      "epoch": 0.3246578879421637,
      "grad_norm": 8.845318794250488,
      "learning_rate": 7.707933412661057e-06,
      "loss": 0.077,
      "step": 6287
    },
    {
      "epoch": 0.32470952749806353,
      "grad_norm": 6.874298572540283,
      "learning_rate": 7.707244584933628e-06,
      "loss": 0.1175,
      "step": 6288
    },
    {
      "epoch": 0.32476116705396335,
      "grad_norm": 5.0303144454956055,
      "learning_rate": 7.706555684505646e-06,
      "loss": 0.0851,
      "step": 6289
    },
    {
      "epoch": 0.32481280660986317,
      "grad_norm": 5.108643054962158,
      "learning_rate": 7.705866711395612e-06,
      "loss": 0.0762,
      "step": 6290
    },
    {
      "epoch": 0.324864446165763,
      "grad_norm": 2.8608405590057373,
      "learning_rate": 7.70517766562203e-06,
      "loss": 0.0409,
      "step": 6291
    },
    {
      "epoch": 0.3249160857216628,
      "grad_norm": 7.57668924331665,
      "learning_rate": 7.7044885472034e-06,
      "loss": 0.1489,
      "step": 6292
    },
    {
      "epoch": 0.3249677252775626,
      "grad_norm": 1.6271967887878418,
      "learning_rate": 7.70379935615823e-06,
      "loss": 0.0348,
      "step": 6293
    },
    {
      "epoch": 0.32501936483346244,
      "grad_norm": 8.992240905761719,
      "learning_rate": 7.703110092505029e-06,
      "loss": 0.0983,
      "step": 6294
    },
    {
      "epoch": 0.32507100438936226,
      "grad_norm": 3.637808084487915,
      "learning_rate": 7.702420756262304e-06,
      "loss": 0.0322,
      "step": 6295
    },
    {
      "epoch": 0.3251226439452621,
      "grad_norm": 25.245906829833984,
      "learning_rate": 7.701731347448567e-06,
      "loss": 0.095,
      "step": 6296
    },
    {
      "epoch": 0.3251742835011619,
      "grad_norm": 15.809964179992676,
      "learning_rate": 7.70104186608233e-06,
      "loss": 0.1435,
      "step": 6297
    },
    {
      "epoch": 0.3252259230570617,
      "grad_norm": 11.35322093963623,
      "learning_rate": 7.700352312182113e-06,
      "loss": 0.0828,
      "step": 6298
    },
    {
      "epoch": 0.32527756261296153,
      "grad_norm": 6.019932270050049,
      "learning_rate": 7.699662685766429e-06,
      "loss": 0.0852,
      "step": 6299
    },
    {
      "epoch": 0.32532920216886135,
      "grad_norm": 15.41086196899414,
      "learning_rate": 7.6989729868538e-06,
      "loss": 0.099,
      "step": 6300
    },
    {
      "epoch": 0.32538084172476117,
      "grad_norm": 1.5235862731933594,
      "learning_rate": 7.698283215462743e-06,
      "loss": 0.0286,
      "step": 6301
    },
    {
      "epoch": 0.325432481280661,
      "grad_norm": 3.386503219604492,
      "learning_rate": 7.697593371611785e-06,
      "loss": 0.0891,
      "step": 6302
    },
    {
      "epoch": 0.3254841208365608,
      "grad_norm": 5.300930500030518,
      "learning_rate": 7.696903455319453e-06,
      "loss": 0.0622,
      "step": 6303
    },
    {
      "epoch": 0.3255357603924606,
      "grad_norm": 4.416952610015869,
      "learning_rate": 7.696213466604267e-06,
      "loss": 0.1,
      "step": 6304
    },
    {
      "epoch": 0.32558739994836045,
      "grad_norm": 8.753764152526855,
      "learning_rate": 7.695523405484764e-06,
      "loss": 0.1345,
      "step": 6305
    },
    {
      "epoch": 0.32563903950426026,
      "grad_norm": 2.9038004875183105,
      "learning_rate": 7.694833271979469e-06,
      "loss": 0.0482,
      "step": 6306
    },
    {
      "epoch": 0.3256906790601601,
      "grad_norm": 7.00599479675293,
      "learning_rate": 7.694143066106917e-06,
      "loss": 0.0771,
      "step": 6307
    },
    {
      "epoch": 0.3257423186160599,
      "grad_norm": 6.028944969177246,
      "learning_rate": 7.693452787885645e-06,
      "loss": 0.0494,
      "step": 6308
    },
    {
      "epoch": 0.3257939581719597,
      "grad_norm": 9.737699508666992,
      "learning_rate": 7.692762437334186e-06,
      "loss": 0.1245,
      "step": 6309
    },
    {
      "epoch": 0.32584559772785954,
      "grad_norm": 18.097898483276367,
      "learning_rate": 7.69207201447108e-06,
      "loss": 0.2236,
      "step": 6310
    },
    {
      "epoch": 0.32589723728375936,
      "grad_norm": 3.3654251098632812,
      "learning_rate": 7.691381519314869e-06,
      "loss": 0.0469,
      "step": 6311
    },
    {
      "epoch": 0.3259488768396592,
      "grad_norm": 6.693331241607666,
      "learning_rate": 7.690690951884096e-06,
      "loss": 0.0861,
      "step": 6312
    },
    {
      "epoch": 0.326000516395559,
      "grad_norm": 2.685509204864502,
      "learning_rate": 7.690000312197302e-06,
      "loss": 0.047,
      "step": 6313
    },
    {
      "epoch": 0.3260521559514588,
      "grad_norm": 3.3876657485961914,
      "learning_rate": 7.689309600273038e-06,
      "loss": 0.048,
      "step": 6314
    },
    {
      "epoch": 0.32610379550735863,
      "grad_norm": 11.964347839355469,
      "learning_rate": 7.688618816129848e-06,
      "loss": 0.1005,
      "step": 6315
    },
    {
      "epoch": 0.32615543506325845,
      "grad_norm": 6.0052361488342285,
      "learning_rate": 7.687927959786287e-06,
      "loss": 0.0551,
      "step": 6316
    },
    {
      "epoch": 0.32620707461915827,
      "grad_norm": 15.735739707946777,
      "learning_rate": 7.687237031260904e-06,
      "loss": 0.1562,
      "step": 6317
    },
    {
      "epoch": 0.3262587141750581,
      "grad_norm": 8.211387634277344,
      "learning_rate": 7.686546030572252e-06,
      "loss": 0.0889,
      "step": 6318
    },
    {
      "epoch": 0.3263103537309579,
      "grad_norm": 2.6012089252471924,
      "learning_rate": 7.685854957738893e-06,
      "loss": 0.0586,
      "step": 6319
    },
    {
      "epoch": 0.3263619932868577,
      "grad_norm": 5.406511306762695,
      "learning_rate": 7.685163812779376e-06,
      "loss": 0.0576,
      "step": 6320
    },
    {
      "epoch": 0.32641363284275754,
      "grad_norm": 4.588871002197266,
      "learning_rate": 7.684472595712272e-06,
      "loss": 0.0747,
      "step": 6321
    },
    {
      "epoch": 0.32646527239865736,
      "grad_norm": 14.184871673583984,
      "learning_rate": 7.683781306556135e-06,
      "loss": 0.1394,
      "step": 6322
    },
    {
      "epoch": 0.3265169119545572,
      "grad_norm": 16.865234375,
      "learning_rate": 7.683089945329533e-06,
      "loss": 0.0833,
      "step": 6323
    },
    {
      "epoch": 0.326568551510457,
      "grad_norm": 2.8143365383148193,
      "learning_rate": 7.682398512051028e-06,
      "loss": 0.0622,
      "step": 6324
    },
    {
      "epoch": 0.3266201910663568,
      "grad_norm": 3.555445909500122,
      "learning_rate": 7.681707006739191e-06,
      "loss": 0.0633,
      "step": 6325
    },
    {
      "epoch": 0.32667183062225663,
      "grad_norm": 6.273993015289307,
      "learning_rate": 7.681015429412591e-06,
      "loss": 0.0539,
      "step": 6326
    },
    {
      "epoch": 0.32672347017815645,
      "grad_norm": 9.071528434753418,
      "learning_rate": 7.6803237800898e-06,
      "loss": 0.0871,
      "step": 6327
    },
    {
      "epoch": 0.32677510973405627,
      "grad_norm": 3.83537220954895,
      "learning_rate": 7.67963205878939e-06,
      "loss": 0.0311,
      "step": 6328
    },
    {
      "epoch": 0.3268267492899561,
      "grad_norm": 2.4023406505584717,
      "learning_rate": 7.67894026552994e-06,
      "loss": 0.0561,
      "step": 6329
    },
    {
      "epoch": 0.3268783888458559,
      "grad_norm": 8.333515167236328,
      "learning_rate": 7.678248400330023e-06,
      "loss": 0.0662,
      "step": 6330
    },
    {
      "epoch": 0.3269300284017557,
      "grad_norm": 2.731076955795288,
      "learning_rate": 7.677556463208221e-06,
      "loss": 0.0538,
      "step": 6331
    },
    {
      "epoch": 0.32698166795765554,
      "grad_norm": 1.9738978147506714,
      "learning_rate": 7.676864454183114e-06,
      "loss": 0.0528,
      "step": 6332
    },
    {
      "epoch": 0.32703330751355536,
      "grad_norm": 3.78475022315979,
      "learning_rate": 7.676172373273287e-06,
      "loss": 0.0568,
      "step": 6333
    },
    {
      "epoch": 0.3270849470694552,
      "grad_norm": 12.856626510620117,
      "learning_rate": 7.675480220497325e-06,
      "loss": 0.139,
      "step": 6334
    },
    {
      "epoch": 0.327136586625355,
      "grad_norm": 2.919257879257202,
      "learning_rate": 7.674787995873813e-06,
      "loss": 0.0725,
      "step": 6335
    },
    {
      "epoch": 0.3271882261812548,
      "grad_norm": 3.0934407711029053,
      "learning_rate": 7.674095699421342e-06,
      "loss": 0.0657,
      "step": 6336
    },
    {
      "epoch": 0.32723986573715463,
      "grad_norm": 9.52999496459961,
      "learning_rate": 7.673403331158503e-06,
      "loss": 0.1241,
      "step": 6337
    },
    {
      "epoch": 0.32729150529305445,
      "grad_norm": 3.205082654953003,
      "learning_rate": 7.67271089110389e-06,
      "loss": 0.0528,
      "step": 6338
    },
    {
      "epoch": 0.3273431448489543,
      "grad_norm": 3.542738676071167,
      "learning_rate": 7.672018379276092e-06,
      "loss": 0.0768,
      "step": 6339
    },
    {
      "epoch": 0.32739478440485414,
      "grad_norm": 12.217191696166992,
      "learning_rate": 7.671325795693713e-06,
      "loss": 0.1361,
      "step": 6340
    },
    {
      "epoch": 0.32744642396075396,
      "grad_norm": 6.725118637084961,
      "learning_rate": 7.670633140375348e-06,
      "loss": 0.1043,
      "step": 6341
    },
    {
      "epoch": 0.3274980635166538,
      "grad_norm": 14.60498046875,
      "learning_rate": 7.669940413339598e-06,
      "loss": 0.152,
      "step": 6342
    },
    {
      "epoch": 0.3275497030725536,
      "grad_norm": 2.9964840412139893,
      "learning_rate": 7.669247614605066e-06,
      "loss": 0.0598,
      "step": 6343
    },
    {
      "epoch": 0.3276013426284534,
      "grad_norm": 6.720771789550781,
      "learning_rate": 7.668554744190354e-06,
      "loss": 0.1017,
      "step": 6344
    },
    {
      "epoch": 0.32765298218435324,
      "grad_norm": 10.284761428833008,
      "learning_rate": 7.667861802114077e-06,
      "loss": 0.1194,
      "step": 6345
    },
    {
      "epoch": 0.32770462174025305,
      "grad_norm": 2.2671642303466797,
      "learning_rate": 7.66716878839483e-06,
      "loss": 0.0452,
      "step": 6346
    },
    {
      "epoch": 0.3277562612961529,
      "grad_norm": 4.537168502807617,
      "learning_rate": 7.666475703051234e-06,
      "loss": 0.0651,
      "step": 6347
    },
    {
      "epoch": 0.3278079008520527,
      "grad_norm": 5.635990142822266,
      "learning_rate": 7.665782546101897e-06,
      "loss": 0.047,
      "step": 6348
    },
    {
      "epoch": 0.3278595404079525,
      "grad_norm": 2.7497663497924805,
      "learning_rate": 7.665089317565432e-06,
      "loss": 0.0566,
      "step": 6349
    },
    {
      "epoch": 0.32791117996385233,
      "grad_norm": 4.489284038543701,
      "learning_rate": 7.664396017460457e-06,
      "loss": 0.0711,
      "step": 6350
    },
    {
      "epoch": 0.32796281951975215,
      "grad_norm": 12.931346893310547,
      "learning_rate": 7.663702645805591e-06,
      "loss": 0.1081,
      "step": 6351
    },
    {
      "epoch": 0.32801445907565197,
      "grad_norm": 9.473583221435547,
      "learning_rate": 7.663009202619452e-06,
      "loss": 0.0895,
      "step": 6352
    },
    {
      "epoch": 0.3280660986315518,
      "grad_norm": 5.7883758544921875,
      "learning_rate": 7.662315687920661e-06,
      "loss": 0.0928,
      "step": 6353
    },
    {
      "epoch": 0.3281177381874516,
      "grad_norm": 3.4080653190612793,
      "learning_rate": 7.661622101727845e-06,
      "loss": 0.1048,
      "step": 6354
    },
    {
      "epoch": 0.3281693777433514,
      "grad_norm": 2.418515920639038,
      "learning_rate": 7.660928444059624e-06,
      "loss": 0.0348,
      "step": 6355
    },
    {
      "epoch": 0.32822101729925124,
      "grad_norm": 3.9716529846191406,
      "learning_rate": 7.660234714934632e-06,
      "loss": 0.0479,
      "step": 6356
    },
    {
      "epoch": 0.32827265685515106,
      "grad_norm": 1.5525935888290405,
      "learning_rate": 7.659540914371493e-06,
      "loss": 0.0241,
      "step": 6357
    },
    {
      "epoch": 0.3283242964110509,
      "grad_norm": 4.930118083953857,
      "learning_rate": 7.658847042388841e-06,
      "loss": 0.064,
      "step": 6358
    },
    {
      "epoch": 0.3283759359669507,
      "grad_norm": 5.7362847328186035,
      "learning_rate": 7.658153099005311e-06,
      "loss": 0.1225,
      "step": 6359
    },
    {
      "epoch": 0.3284275755228505,
      "grad_norm": 4.157914161682129,
      "learning_rate": 7.657459084239534e-06,
      "loss": 0.0558,
      "step": 6360
    },
    {
      "epoch": 0.32847921507875033,
      "grad_norm": 19.848783493041992,
      "learning_rate": 7.656764998110151e-06,
      "loss": 0.1302,
      "step": 6361
    },
    {
      "epoch": 0.32853085463465015,
      "grad_norm": 6.004983901977539,
      "learning_rate": 7.6560708406358e-06,
      "loss": 0.0599,
      "step": 6362
    },
    {
      "epoch": 0.32858249419054997,
      "grad_norm": 5.8066253662109375,
      "learning_rate": 7.655376611835119e-06,
      "loss": 0.0694,
      "step": 6363
    },
    {
      "epoch": 0.3286341337464498,
      "grad_norm": 5.263220310211182,
      "learning_rate": 7.654682311726755e-06,
      "loss": 0.0816,
      "step": 6364
    },
    {
      "epoch": 0.3286857733023496,
      "grad_norm": 4.79127836227417,
      "learning_rate": 7.65398794032935e-06,
      "loss": 0.0629,
      "step": 6365
    },
    {
      "epoch": 0.3287374128582494,
      "grad_norm": 12.991960525512695,
      "learning_rate": 7.653293497661554e-06,
      "loss": 0.0713,
      "step": 6366
    },
    {
      "epoch": 0.32878905241414924,
      "grad_norm": 4.541641712188721,
      "learning_rate": 7.65259898374201e-06,
      "loss": 0.0571,
      "step": 6367
    },
    {
      "epoch": 0.32884069197004906,
      "grad_norm": 4.445070266723633,
      "learning_rate": 7.651904398589374e-06,
      "loss": 0.0326,
      "step": 6368
    },
    {
      "epoch": 0.3288923315259489,
      "grad_norm": 4.573145389556885,
      "learning_rate": 7.651209742222297e-06,
      "loss": 0.0663,
      "step": 6369
    },
    {
      "epoch": 0.3289439710818487,
      "grad_norm": 3.813847780227661,
      "learning_rate": 7.65051501465943e-06,
      "loss": 0.0579,
      "step": 6370
    },
    {
      "epoch": 0.3289956106377485,
      "grad_norm": 3.8059003353118896,
      "learning_rate": 7.649820215919434e-06,
      "loss": 0.0478,
      "step": 6371
    },
    {
      "epoch": 0.32904725019364833,
      "grad_norm": 8.482529640197754,
      "learning_rate": 7.649125346020964e-06,
      "loss": 0.1208,
      "step": 6372
    },
    {
      "epoch": 0.32909888974954815,
      "grad_norm": 20.874223709106445,
      "learning_rate": 7.648430404982683e-06,
      "loss": 0.1191,
      "step": 6373
    },
    {
      "epoch": 0.32915052930544797,
      "grad_norm": 16.920398712158203,
      "learning_rate": 7.64773539282325e-06,
      "loss": 0.1721,
      "step": 6374
    },
    {
      "epoch": 0.3292021688613478,
      "grad_norm": 4.606656551361084,
      "learning_rate": 7.64704030956133e-06,
      "loss": 0.0637,
      "step": 6375
    },
    {
      "epoch": 0.3292538084172476,
      "grad_norm": 6.572342872619629,
      "learning_rate": 7.646345155215589e-06,
      "loss": 0.0829,
      "step": 6376
    },
    {
      "epoch": 0.3293054479731474,
      "grad_norm": 6.383398056030273,
      "learning_rate": 7.645649929804694e-06,
      "loss": 0.0586,
      "step": 6377
    },
    {
      "epoch": 0.32935708752904724,
      "grad_norm": 13.462642669677734,
      "learning_rate": 7.644954633347317e-06,
      "loss": 0.2007,
      "step": 6378
    },
    {
      "epoch": 0.32940872708494706,
      "grad_norm": 3.675668716430664,
      "learning_rate": 7.644259265862126e-06,
      "loss": 0.0526,
      "step": 6379
    },
    {
      "epoch": 0.3294603666408469,
      "grad_norm": 7.303722858428955,
      "learning_rate": 7.6435638273678e-06,
      "loss": 0.1016,
      "step": 6380
    },
    {
      "epoch": 0.3295120061967467,
      "grad_norm": 5.071949481964111,
      "learning_rate": 7.642868317883005e-06,
      "loss": 0.0706,
      "step": 6381
    },
    {
      "epoch": 0.3295636457526465,
      "grad_norm": 3.1226046085357666,
      "learning_rate": 7.64217273742643e-06,
      "loss": 0.0464,
      "step": 6382
    },
    {
      "epoch": 0.32961528530854634,
      "grad_norm": 10.305790901184082,
      "learning_rate": 7.641477086016745e-06,
      "loss": 0.0838,
      "step": 6383
    },
    {
      "epoch": 0.32966692486444615,
      "grad_norm": 6.518255233764648,
      "learning_rate": 7.640781363672633e-06,
      "loss": 0.0626,
      "step": 6384
    },
    {
      "epoch": 0.32971856442034597,
      "grad_norm": 1.880955457687378,
      "learning_rate": 7.640085570412779e-06,
      "loss": 0.0523,
      "step": 6385
    },
    {
      "epoch": 0.3297702039762458,
      "grad_norm": 3.600288152694702,
      "learning_rate": 7.639389706255868e-06,
      "loss": 0.0563,
      "step": 6386
    },
    {
      "epoch": 0.3298218435321456,
      "grad_norm": 4.387271404266357,
      "learning_rate": 7.638693771220586e-06,
      "loss": 0.0515,
      "step": 6387
    },
    {
      "epoch": 0.3298734830880454,
      "grad_norm": 2.6395297050476074,
      "learning_rate": 7.63799776532562e-06,
      "loss": 0.0605,
      "step": 6388
    },
    {
      "epoch": 0.32992512264394525,
      "grad_norm": 3.0039055347442627,
      "learning_rate": 7.637301688589663e-06,
      "loss": 0.0551,
      "step": 6389
    },
    {
      "epoch": 0.32997676219984506,
      "grad_norm": 1.6919814348220825,
      "learning_rate": 7.636605541031405e-06,
      "loss": 0.0426,
      "step": 6390
    },
    {
      "epoch": 0.3300284017557449,
      "grad_norm": 3.0314784049987793,
      "learning_rate": 7.635909322669545e-06,
      "loss": 0.0752,
      "step": 6391
    },
    {
      "epoch": 0.3300800413116447,
      "grad_norm": 4.33677339553833,
      "learning_rate": 7.635213033522776e-06,
      "loss": 0.0575,
      "step": 6392
    },
    {
      "epoch": 0.3301316808675445,
      "grad_norm": 13.812926292419434,
      "learning_rate": 7.634516673609794e-06,
      "loss": 0.0719,
      "step": 6393
    },
    {
      "epoch": 0.33018332042344434,
      "grad_norm": 6.442193984985352,
      "learning_rate": 7.633820242949303e-06,
      "loss": 0.0716,
      "step": 6394
    },
    {
      "epoch": 0.33023495997934416,
      "grad_norm": 6.177915573120117,
      "learning_rate": 7.633123741560003e-06,
      "loss": 0.1124,
      "step": 6395
    },
    {
      "epoch": 0.330286599535244,
      "grad_norm": 6.4527363777160645,
      "learning_rate": 7.632427169460598e-06,
      "loss": 0.036,
      "step": 6396
    },
    {
      "epoch": 0.3303382390911438,
      "grad_norm": 7.90030574798584,
      "learning_rate": 7.631730526669797e-06,
      "loss": 0.1183,
      "step": 6397
    },
    {
      "epoch": 0.3303898786470436,
      "grad_norm": 7.518804550170898,
      "learning_rate": 7.631033813206303e-06,
      "loss": 0.1352,
      "step": 6398
    },
    {
      "epoch": 0.33044151820294343,
      "grad_norm": 4.820240020751953,
      "learning_rate": 7.630337029088827e-06,
      "loss": 0.0364,
      "step": 6399
    },
    {
      "epoch": 0.33049315775884325,
      "grad_norm": 10.92322826385498,
      "learning_rate": 7.629640174336084e-06,
      "loss": 0.0969,
      "step": 6400
    },
    {
      "epoch": 0.33054479731474307,
      "grad_norm": 3.988945960998535,
      "learning_rate": 7.628943248966782e-06,
      "loss": 0.0623,
      "step": 6401
    },
    {
      "epoch": 0.33059643687064294,
      "grad_norm": 9.732184410095215,
      "learning_rate": 7.628246252999639e-06,
      "loss": 0.0876,
      "step": 6402
    },
    {
      "epoch": 0.33064807642654276,
      "grad_norm": 16.0844669342041,
      "learning_rate": 7.627549186453372e-06,
      "loss": 0.0957,
      "step": 6403
    },
    {
      "epoch": 0.3306997159824426,
      "grad_norm": 18.749919891357422,
      "learning_rate": 7.626852049346702e-06,
      "loss": 0.2276,
      "step": 6404
    },
    {
      "epoch": 0.3307513555383424,
      "grad_norm": 16.170007705688477,
      "learning_rate": 7.6261548416983455e-06,
      "loss": 0.1366,
      "step": 6405
    },
    {
      "epoch": 0.3308029950942422,
      "grad_norm": 21.813949584960938,
      "learning_rate": 7.62545756352703e-06,
      "loss": 0.2315,
      "step": 6406
    },
    {
      "epoch": 0.33085463465014203,
      "grad_norm": 4.770065784454346,
      "learning_rate": 7.624760214851477e-06,
      "loss": 0.0948,
      "step": 6407
    },
    {
      "epoch": 0.33090627420604185,
      "grad_norm": 2.5926482677459717,
      "learning_rate": 7.624062795690413e-06,
      "loss": 0.041,
      "step": 6408
    },
    {
      "epoch": 0.33095791376194167,
      "grad_norm": 3.3735427856445312,
      "learning_rate": 7.623365306062569e-06,
      "loss": 0.0401,
      "step": 6409
    },
    {
      "epoch": 0.3310095533178415,
      "grad_norm": 2.940749168395996,
      "learning_rate": 7.622667745986675e-06,
      "loss": 0.0436,
      "step": 6410
    },
    {
      "epoch": 0.3310611928737413,
      "grad_norm": 12.037283897399902,
      "learning_rate": 7.621970115481463e-06,
      "loss": 0.0891,
      "step": 6411
    },
    {
      "epoch": 0.3311128324296411,
      "grad_norm": 4.231298446655273,
      "learning_rate": 7.621272414565665e-06,
      "loss": 0.0788,
      "step": 6412
    },
    {
      "epoch": 0.33116447198554094,
      "grad_norm": 2.2888448238372803,
      "learning_rate": 7.620574643258021e-06,
      "loss": 0.0379,
      "step": 6413
    },
    {
      "epoch": 0.33121611154144076,
      "grad_norm": 2.994813919067383,
      "learning_rate": 7.619876801577264e-06,
      "loss": 0.0507,
      "step": 6414
    },
    {
      "epoch": 0.3312677510973406,
      "grad_norm": 8.063576698303223,
      "learning_rate": 7.619178889542141e-06,
      "loss": 0.0974,
      "step": 6415
    },
    {
      "epoch": 0.3313193906532404,
      "grad_norm": 3.7224972248077393,
      "learning_rate": 7.618480907171386e-06,
      "loss": 0.0922,
      "step": 6416
    },
    {
      "epoch": 0.3313710302091402,
      "grad_norm": 2.865981101989746,
      "learning_rate": 7.617782854483747e-06,
      "loss": 0.0666,
      "step": 6417
    },
    {
      "epoch": 0.33142266976504003,
      "grad_norm": 7.485466957092285,
      "learning_rate": 7.61708473149797e-06,
      "loss": 0.0644,
      "step": 6418
    },
    {
      "epoch": 0.33147430932093985,
      "grad_norm": 3.3750619888305664,
      "learning_rate": 7.616386538232801e-06,
      "loss": 0.0552,
      "step": 6419
    },
    {
      "epoch": 0.33152594887683967,
      "grad_norm": 4.335024833679199,
      "learning_rate": 7.615688274706989e-06,
      "loss": 0.06,
      "step": 6420
    },
    {
      "epoch": 0.3315775884327395,
      "grad_norm": 4.217319488525391,
      "learning_rate": 7.614989940939286e-06,
      "loss": 0.0658,
      "step": 6421
    },
    {
      "epoch": 0.3316292279886393,
      "grad_norm": 16.469684600830078,
      "learning_rate": 7.614291536948446e-06,
      "loss": 0.2866,
      "step": 6422
    },
    {
      "epoch": 0.3316808675445391,
      "grad_norm": 1.8412728309631348,
      "learning_rate": 7.613593062753221e-06,
      "loss": 0.0505,
      "step": 6423
    },
    {
      "epoch": 0.33173250710043894,
      "grad_norm": 4.201260566711426,
      "learning_rate": 7.612894518372371e-06,
      "loss": 0.0458,
      "step": 6424
    },
    {
      "epoch": 0.33178414665633876,
      "grad_norm": 21.787336349487305,
      "learning_rate": 7.612195903824655e-06,
      "loss": 0.1483,
      "step": 6425
    },
    {
      "epoch": 0.3318357862122386,
      "grad_norm": 1.7984896898269653,
      "learning_rate": 7.61149721912883e-06,
      "loss": 0.0384,
      "step": 6426
    },
    {
      "epoch": 0.3318874257681384,
      "grad_norm": 3.281473398208618,
      "learning_rate": 7.610798464303662e-06,
      "loss": 0.06,
      "step": 6427
    },
    {
      "epoch": 0.3319390653240382,
      "grad_norm": 8.338569641113281,
      "learning_rate": 7.610099639367914e-06,
      "loss": 0.0757,
      "step": 6428
    },
    {
      "epoch": 0.33199070487993804,
      "grad_norm": 1.1349068880081177,
      "learning_rate": 7.609400744340354e-06,
      "loss": 0.0194,
      "step": 6429
    },
    {
      "epoch": 0.33204234443583786,
      "grad_norm": 6.41914176940918,
      "learning_rate": 7.608701779239746e-06,
      "loss": 0.0921,
      "step": 6430
    },
    {
      "epoch": 0.3320939839917377,
      "grad_norm": 3.6671547889709473,
      "learning_rate": 7.608002744084864e-06,
      "loss": 0.0824,
      "step": 6431
    },
    {
      "epoch": 0.3321456235476375,
      "grad_norm": 7.944011688232422,
      "learning_rate": 7.6073036388944785e-06,
      "loss": 0.1283,
      "step": 6432
    },
    {
      "epoch": 0.3321972631035373,
      "grad_norm": 5.295161724090576,
      "learning_rate": 7.606604463687364e-06,
      "loss": 0.0477,
      "step": 6433
    },
    {
      "epoch": 0.33224890265943713,
      "grad_norm": 10.66015338897705,
      "learning_rate": 7.605905218482294e-06,
      "loss": 0.1051,
      "step": 6434
    },
    {
      "epoch": 0.33230054221533695,
      "grad_norm": 2.8608062267303467,
      "learning_rate": 7.605205903298049e-06,
      "loss": 0.049,
      "step": 6435
    },
    {
      "epoch": 0.33235218177123677,
      "grad_norm": 11.128961563110352,
      "learning_rate": 7.604506518153407e-06,
      "loss": 0.1133,
      "step": 6436
    },
    {
      "epoch": 0.3324038213271366,
      "grad_norm": 11.828657150268555,
      "learning_rate": 7.603807063067153e-06,
      "loss": 0.0984,
      "step": 6437
    },
    {
      "epoch": 0.3324554608830364,
      "grad_norm": 9.377365112304688,
      "learning_rate": 7.603107538058061e-06,
      "loss": 0.0849,
      "step": 6438
    },
    {
      "epoch": 0.3325071004389362,
      "grad_norm": 2.4846885204315186,
      "learning_rate": 7.6024079431449264e-06,
      "loss": 0.0513,
      "step": 6439
    },
    {
      "epoch": 0.33255873999483604,
      "grad_norm": 2.549971103668213,
      "learning_rate": 7.601708278346529e-06,
      "loss": 0.0576,
      "step": 6440
    },
    {
      "epoch": 0.33261037955073586,
      "grad_norm": 1.743102788925171,
      "learning_rate": 7.601008543681662e-06,
      "loss": 0.039,
      "step": 6441
    },
    {
      "epoch": 0.3326620191066357,
      "grad_norm": 7.712802410125732,
      "learning_rate": 7.600308739169113e-06,
      "loss": 0.1039,
      "step": 6442
    },
    {
      "epoch": 0.3327136586625355,
      "grad_norm": 3.09468412399292,
      "learning_rate": 7.599608864827677e-06,
      "loss": 0.0364,
      "step": 6443
    },
    {
      "epoch": 0.3327652982184353,
      "grad_norm": 2.1847984790802,
      "learning_rate": 7.598908920676149e-06,
      "loss": 0.0418,
      "step": 6444
    },
    {
      "epoch": 0.33281693777433513,
      "grad_norm": 6.926403522491455,
      "learning_rate": 7.598208906733322e-06,
      "loss": 0.0989,
      "step": 6445
    },
    {
      "epoch": 0.33286857733023495,
      "grad_norm": 14.784632682800293,
      "learning_rate": 7.597508823017996e-06,
      "loss": 0.161,
      "step": 6446
    },
    {
      "epoch": 0.33292021688613477,
      "grad_norm": 4.044144153594971,
      "learning_rate": 7.596808669548972e-06,
      "loss": 0.0357,
      "step": 6447
    },
    {
      "epoch": 0.3329718564420346,
      "grad_norm": 3.8955907821655273,
      "learning_rate": 7.596108446345052e-06,
      "loss": 0.0534,
      "step": 6448
    },
    {
      "epoch": 0.3330234959979344,
      "grad_norm": 8.581636428833008,
      "learning_rate": 7.5954081534250365e-06,
      "loss": 0.0676,
      "step": 6449
    },
    {
      "epoch": 0.3330751355538342,
      "grad_norm": 1.6148779392242432,
      "learning_rate": 7.594707790807736e-06,
      "loss": 0.0414,
      "step": 6450
    },
    {
      "epoch": 0.33312677510973404,
      "grad_norm": 5.4704108238220215,
      "learning_rate": 7.594007358511956e-06,
      "loss": 0.0818,
      "step": 6451
    },
    {
      "epoch": 0.33317841466563386,
      "grad_norm": 2.5226705074310303,
      "learning_rate": 7.593306856556504e-06,
      "loss": 0.0393,
      "step": 6452
    },
    {
      "epoch": 0.3332300542215337,
      "grad_norm": 4.279814720153809,
      "learning_rate": 7.592606284960196e-06,
      "loss": 0.0753,
      "step": 6453
    },
    {
      "epoch": 0.3332816937774335,
      "grad_norm": 6.148282051086426,
      "learning_rate": 7.5919056437418394e-06,
      "loss": 0.0707,
      "step": 6454
    },
    {
      "epoch": 0.3333333333333333,
      "grad_norm": 15.721205711364746,
      "learning_rate": 7.591204932920253e-06,
      "loss": 0.0791,
      "step": 6455
    },
    {
      "epoch": 0.33338497288923313,
      "grad_norm": 4.2522807121276855,
      "learning_rate": 7.590504152514253e-06,
      "loss": 0.061,
      "step": 6456
    },
    {
      "epoch": 0.33343661244513295,
      "grad_norm": 5.9102582931518555,
      "learning_rate": 7.58980330254266e-06,
      "loss": 0.0896,
      "step": 6457
    },
    {
      "epoch": 0.33348825200103277,
      "grad_norm": 4.810573101043701,
      "learning_rate": 7.589102383024292e-06,
      "loss": 0.0526,
      "step": 6458
    },
    {
      "epoch": 0.3335398915569326,
      "grad_norm": 6.531050682067871,
      "learning_rate": 7.588401393977971e-06,
      "loss": 0.0622,
      "step": 6459
    },
    {
      "epoch": 0.3335915311128324,
      "grad_norm": 6.328919887542725,
      "learning_rate": 7.587700335422523e-06,
      "loss": 0.1136,
      "step": 6460
    },
    {
      "epoch": 0.3336431706687322,
      "grad_norm": 8.939032554626465,
      "learning_rate": 7.586999207376775e-06,
      "loss": 0.1409,
      "step": 6461
    },
    {
      "epoch": 0.33369481022463204,
      "grad_norm": 6.882841110229492,
      "learning_rate": 7.586298009859555e-06,
      "loss": 0.0811,
      "step": 6462
    },
    {
      "epoch": 0.33374644978053186,
      "grad_norm": 2.1015243530273438,
      "learning_rate": 7.585596742889691e-06,
      "loss": 0.0359,
      "step": 6463
    },
    {
      "epoch": 0.3337980893364317,
      "grad_norm": 1.5804616212844849,
      "learning_rate": 7.584895406486017e-06,
      "loss": 0.0369,
      "step": 6464
    },
    {
      "epoch": 0.3338497288923315,
      "grad_norm": 2.9301915168762207,
      "learning_rate": 7.584194000667364e-06,
      "loss": 0.0702,
      "step": 6465
    },
    {
      "epoch": 0.3339013684482314,
      "grad_norm": 10.6419677734375,
      "learning_rate": 7.583492525452572e-06,
      "loss": 0.0881,
      "step": 6466
    },
    {
      "epoch": 0.3339530080041312,
      "grad_norm": 2.401923179626465,
      "learning_rate": 7.582790980860475e-06,
      "loss": 0.0475,
      "step": 6467
    },
    {
      "epoch": 0.334004647560031,
      "grad_norm": 25.68247413635254,
      "learning_rate": 7.582089366909913e-06,
      "loss": 0.2147,
      "step": 6468
    },
    {
      "epoch": 0.33405628711593083,
      "grad_norm": 4.498342990875244,
      "learning_rate": 7.581387683619728e-06,
      "loss": 0.0429,
      "step": 6469
    },
    {
      "epoch": 0.33410792667183065,
      "grad_norm": 3.0407185554504395,
      "learning_rate": 7.5806859310087626e-06,
      "loss": 0.0451,
      "step": 6470
    },
    {
      "epoch": 0.33415956622773046,
      "grad_norm": 5.459805011749268,
      "learning_rate": 7.579984109095859e-06,
      "loss": 0.0888,
      "step": 6471
    },
    {
      "epoch": 0.3342112057836303,
      "grad_norm": 3.94315767288208,
      "learning_rate": 7.579282217899871e-06,
      "loss": 0.0608,
      "step": 6472
    },
    {
      "epoch": 0.3342628453395301,
      "grad_norm": 2.0844810009002686,
      "learning_rate": 7.57858025743964e-06,
      "loss": 0.0429,
      "step": 6473
    },
    {
      "epoch": 0.3343144848954299,
      "grad_norm": 6.492523193359375,
      "learning_rate": 7.57787822773402e-06,
      "loss": 0.0709,
      "step": 6474
    },
    {
      "epoch": 0.33436612445132974,
      "grad_norm": 2.8461785316467285,
      "learning_rate": 7.577176128801863e-06,
      "loss": 0.0442,
      "step": 6475
    },
    {
      "epoch": 0.33441776400722956,
      "grad_norm": 8.926630020141602,
      "learning_rate": 7.576473960662023e-06,
      "loss": 0.133,
      "step": 6476
    },
    {
      "epoch": 0.3344694035631294,
      "grad_norm": 2.6132400035858154,
      "learning_rate": 7.575771723333356e-06,
      "loss": 0.0467,
      "step": 6477
    },
    {
      "epoch": 0.3345210431190292,
      "grad_norm": 1.584672212600708,
      "learning_rate": 7.575069416834719e-06,
      "loss": 0.0129,
      "step": 6478
    },
    {
      "epoch": 0.334572682674929,
      "grad_norm": 4.947836875915527,
      "learning_rate": 7.574367041184974e-06,
      "loss": 0.0455,
      "step": 6479
    },
    {
      "epoch": 0.33462432223082883,
      "grad_norm": 6.3750410079956055,
      "learning_rate": 7.573664596402981e-06,
      "loss": 0.0625,
      "step": 6480
    },
    {
      "epoch": 0.33467596178672865,
      "grad_norm": 7.397749423980713,
      "learning_rate": 7.5729620825076045e-06,
      "loss": 0.0828,
      "step": 6481
    },
    {
      "epoch": 0.33472760134262847,
      "grad_norm": 6.460799694061279,
      "learning_rate": 7.572259499517708e-06,
      "loss": 0.0919,
      "step": 6482
    },
    {
      "epoch": 0.3347792408985283,
      "grad_norm": 8.405752182006836,
      "learning_rate": 7.571556847452161e-06,
      "loss": 0.0773,
      "step": 6483
    },
    {
      "epoch": 0.3348308804544281,
      "grad_norm": 3.6295413970947266,
      "learning_rate": 7.570854126329832e-06,
      "loss": 0.051,
      "step": 6484
    },
    {
      "epoch": 0.3348825200103279,
      "grad_norm": 6.364694595336914,
      "learning_rate": 7.570151336169589e-06,
      "loss": 0.0639,
      "step": 6485
    },
    {
      "epoch": 0.33493415956622774,
      "grad_norm": 4.179497241973877,
      "learning_rate": 7.569448476990311e-06,
      "loss": 0.0231,
      "step": 6486
    },
    {
      "epoch": 0.33498579912212756,
      "grad_norm": 7.2754340171813965,
      "learning_rate": 7.568745548810865e-06,
      "loss": 0.0935,
      "step": 6487
    },
    {
      "epoch": 0.3350374386780274,
      "grad_norm": 7.79273796081543,
      "learning_rate": 7.568042551650134e-06,
      "loss": 0.1135,
      "step": 6488
    },
    {
      "epoch": 0.3350890782339272,
      "grad_norm": 4.237560272216797,
      "learning_rate": 7.567339485526992e-06,
      "loss": 0.0768,
      "step": 6489
    },
    {
      "epoch": 0.335140717789827,
      "grad_norm": 5.433151721954346,
      "learning_rate": 7.566636350460322e-06,
      "loss": 0.0763,
      "step": 6490
    },
    {
      "epoch": 0.33519235734572683,
      "grad_norm": 22.742429733276367,
      "learning_rate": 7.565933146469006e-06,
      "loss": 0.2109,
      "step": 6491
    },
    {
      "epoch": 0.33524399690162665,
      "grad_norm": 4.72611141204834,
      "learning_rate": 7.5652298735719245e-06,
      "loss": 0.0668,
      "step": 6492
    },
    {
      "epoch": 0.33529563645752647,
      "grad_norm": 7.930948257446289,
      "learning_rate": 7.564526531787968e-06,
      "loss": 0.1032,
      "step": 6493
    },
    {
      "epoch": 0.3353472760134263,
      "grad_norm": 3.854463577270508,
      "learning_rate": 7.563823121136021e-06,
      "loss": 0.0558,
      "step": 6494
    },
    {
      "epoch": 0.3353989155693261,
      "grad_norm": 5.478950500488281,
      "learning_rate": 7.563119641634973e-06,
      "loss": 0.1002,
      "step": 6495
    },
    {
      "epoch": 0.3354505551252259,
      "grad_norm": 3.6564459800720215,
      "learning_rate": 7.562416093303715e-06,
      "loss": 0.0507,
      "step": 6496
    },
    {
      "epoch": 0.33550219468112574,
      "grad_norm": 10.68608570098877,
      "learning_rate": 7.561712476161143e-06,
      "loss": 0.1234,
      "step": 6497
    },
    {
      "epoch": 0.33555383423702556,
      "grad_norm": 3.8497161865234375,
      "learning_rate": 7.5610087902261485e-06,
      "loss": 0.0992,
      "step": 6498
    },
    {
      "epoch": 0.3356054737929254,
      "grad_norm": 12.633810997009277,
      "learning_rate": 7.56030503551763e-06,
      "loss": 0.1097,
      "step": 6499
    },
    {
      "epoch": 0.3356571133488252,
      "grad_norm": 2.7590525150299072,
      "learning_rate": 7.559601212054487e-06,
      "loss": 0.0659,
      "step": 6500
    },
    {
      "epoch": 0.335708752904725,
      "grad_norm": 3.346126079559326,
      "learning_rate": 7.558897319855618e-06,
      "loss": 0.0498,
      "step": 6501
    },
    {
      "epoch": 0.33576039246062483,
      "grad_norm": 4.540650844573975,
      "learning_rate": 7.558193358939927e-06,
      "loss": 0.0725,
      "step": 6502
    },
    {
      "epoch": 0.33581203201652465,
      "grad_norm": 7.167328357696533,
      "learning_rate": 7.557489329326319e-06,
      "loss": 0.0479,
      "step": 6503
    },
    {
      "epoch": 0.33586367157242447,
      "grad_norm": 2.987626314163208,
      "learning_rate": 7.556785231033696e-06,
      "loss": 0.0563,
      "step": 6504
    },
    {
      "epoch": 0.3359153111283243,
      "grad_norm": 2.7083351612091064,
      "learning_rate": 7.5560810640809716e-06,
      "loss": 0.0599,
      "step": 6505
    },
    {
      "epoch": 0.3359669506842241,
      "grad_norm": 7.295248985290527,
      "learning_rate": 7.555376828487051e-06,
      "loss": 0.0862,
      "step": 6506
    },
    {
      "epoch": 0.3360185902401239,
      "grad_norm": 3.4164042472839355,
      "learning_rate": 7.554672524270849e-06,
      "loss": 0.084,
      "step": 6507
    },
    {
      "epoch": 0.33607022979602375,
      "grad_norm": 3.4755020141601562,
      "learning_rate": 7.553968151451276e-06,
      "loss": 0.0933,
      "step": 6508
    },
    {
      "epoch": 0.33612186935192356,
      "grad_norm": 2.0780844688415527,
      "learning_rate": 7.553263710047248e-06,
      "loss": 0.0271,
      "step": 6509
    },
    {
      "epoch": 0.3361735089078234,
      "grad_norm": 1.8886297941207886,
      "learning_rate": 7.5525592000776845e-06,
      "loss": 0.0276,
      "step": 6510
    },
    {
      "epoch": 0.3362251484637232,
      "grad_norm": 2.958725929260254,
      "learning_rate": 7.551854621561501e-06,
      "loss": 0.0657,
      "step": 6511
    },
    {
      "epoch": 0.336276788019623,
      "grad_norm": 2.763615131378174,
      "learning_rate": 7.551149974517622e-06,
      "loss": 0.0633,
      "step": 6512
    },
    {
      "epoch": 0.33632842757552284,
      "grad_norm": 4.628999710083008,
      "learning_rate": 7.550445258964966e-06,
      "loss": 0.0465,
      "step": 6513
    },
    {
      "epoch": 0.33638006713142266,
      "grad_norm": 10.358197212219238,
      "learning_rate": 7.549740474922462e-06,
      "loss": 0.1064,
      "step": 6514
    },
    {
      "epoch": 0.3364317066873225,
      "grad_norm": 3.2341198921203613,
      "learning_rate": 7.549035622409035e-06,
      "loss": 0.0335,
      "step": 6515
    },
    {
      "epoch": 0.3364833462432223,
      "grad_norm": 5.083799839019775,
      "learning_rate": 7.5483307014436095e-06,
      "loss": 0.0568,
      "step": 6516
    },
    {
      "epoch": 0.3365349857991221,
      "grad_norm": 8.383495330810547,
      "learning_rate": 7.54762571204512e-06,
      "loss": 0.0553,
      "step": 6517
    },
    {
      "epoch": 0.33658662535502193,
      "grad_norm": 3.5047976970672607,
      "learning_rate": 7.546920654232495e-06,
      "loss": 0.0568,
      "step": 6518
    },
    {
      "epoch": 0.33663826491092175,
      "grad_norm": 1.9601891040802002,
      "learning_rate": 7.546215528024672e-06,
      "loss": 0.0223,
      "step": 6519
    },
    {
      "epoch": 0.33668990446682157,
      "grad_norm": 4.330197334289551,
      "learning_rate": 7.545510333440582e-06,
      "loss": 0.0483,
      "step": 6520
    },
    {
      "epoch": 0.3367415440227214,
      "grad_norm": 2.0065805912017822,
      "learning_rate": 7.544805070499166e-06,
      "loss": 0.0337,
      "step": 6521
    },
    {
      "epoch": 0.3367931835786212,
      "grad_norm": 15.969279289245605,
      "learning_rate": 7.5440997392193615e-06,
      "loss": 0.0746,
      "step": 6522
    },
    {
      "epoch": 0.336844823134521,
      "grad_norm": 4.385169506072998,
      "learning_rate": 7.543394339620109e-06,
      "loss": 0.0666,
      "step": 6523
    },
    {
      "epoch": 0.33689646269042084,
      "grad_norm": 3.8481996059417725,
      "learning_rate": 7.542688871720354e-06,
      "loss": 0.0444,
      "step": 6524
    },
    {
      "epoch": 0.33694810224632066,
      "grad_norm": 6.756246566772461,
      "learning_rate": 7.541983335539038e-06,
      "loss": 0.0744,
      "step": 6525
    },
    {
      "epoch": 0.3369997418022205,
      "grad_norm": 6.927114009857178,
      "learning_rate": 7.541277731095109e-06,
      "loss": 0.1026,
      "step": 6526
    },
    {
      "epoch": 0.3370513813581203,
      "grad_norm": 26.280738830566406,
      "learning_rate": 7.540572058407515e-06,
      "loss": 0.06,
      "step": 6527
    },
    {
      "epoch": 0.3371030209140201,
      "grad_norm": 14.55402660369873,
      "learning_rate": 7.539866317495208e-06,
      "loss": 0.207,
      "step": 6528
    },
    {
      "epoch": 0.33715466046991993,
      "grad_norm": 8.07740592956543,
      "learning_rate": 7.539160508377137e-06,
      "loss": 0.0927,
      "step": 6529
    },
    {
      "epoch": 0.3372063000258198,
      "grad_norm": 12.623845100402832,
      "learning_rate": 7.538454631072256e-06,
      "loss": 0.1199,
      "step": 6530
    },
    {
      "epoch": 0.3372579395817196,
      "grad_norm": 3.630192995071411,
      "learning_rate": 7.537748685599524e-06,
      "loss": 0.0482,
      "step": 6531
    },
    {
      "epoch": 0.33730957913761944,
      "grad_norm": 7.027617931365967,
      "learning_rate": 7.5370426719778965e-06,
      "loss": 0.079,
      "step": 6532
    },
    {
      "epoch": 0.33736121869351926,
      "grad_norm": 3.945305109024048,
      "learning_rate": 7.5363365902263316e-06,
      "loss": 0.073,
      "step": 6533
    },
    {
      "epoch": 0.3374128582494191,
      "grad_norm": 3.52284574508667,
      "learning_rate": 7.535630440363793e-06,
      "loss": 0.0906,
      "step": 6534
    },
    {
      "epoch": 0.3374644978053189,
      "grad_norm": 4.543690204620361,
      "learning_rate": 7.534924222409241e-06,
      "loss": 0.0539,
      "step": 6535
    },
    {
      "epoch": 0.3375161373612187,
      "grad_norm": 7.351944923400879,
      "learning_rate": 7.534217936381645e-06,
      "loss": 0.0677,
      "step": 6536
    },
    {
      "epoch": 0.33756777691711853,
      "grad_norm": 4.216504096984863,
      "learning_rate": 7.533511582299965e-06,
      "loss": 0.0587,
      "step": 6537
    },
    {
      "epoch": 0.33761941647301835,
      "grad_norm": 9.218238830566406,
      "learning_rate": 7.5328051601831745e-06,
      "loss": 0.1227,
      "step": 6538
    },
    {
      "epoch": 0.33767105602891817,
      "grad_norm": 2.5759122371673584,
      "learning_rate": 7.532098670050243e-06,
      "loss": 0.0351,
      "step": 6539
    },
    {
      "epoch": 0.337722695584818,
      "grad_norm": 3.5889711380004883,
      "learning_rate": 7.531392111920141e-06,
      "loss": 0.0654,
      "step": 6540
    },
    {
      "epoch": 0.3377743351407178,
      "grad_norm": 4.053648471832275,
      "learning_rate": 7.5306854858118425e-06,
      "loss": 0.1072,
      "step": 6541
    },
    {
      "epoch": 0.3378259746966176,
      "grad_norm": 6.284488677978516,
      "learning_rate": 7.529978791744326e-06,
      "loss": 0.0493,
      "step": 6542
    },
    {
      "epoch": 0.33787761425251744,
      "grad_norm": 3.3685669898986816,
      "learning_rate": 7.5292720297365664e-06,
      "loss": 0.0363,
      "step": 6543
    },
    {
      "epoch": 0.33792925380841726,
      "grad_norm": 3.1988840103149414,
      "learning_rate": 7.5285651998075424e-06,
      "loss": 0.0587,
      "step": 6544
    },
    {
      "epoch": 0.3379808933643171,
      "grad_norm": 9.830303192138672,
      "learning_rate": 7.527858301976241e-06,
      "loss": 0.1042,
      "step": 6545
    },
    {
      "epoch": 0.3380325329202169,
      "grad_norm": 3.1860575675964355,
      "learning_rate": 7.527151336261638e-06,
      "loss": 0.0548,
      "step": 6546
    },
    {
      "epoch": 0.3380841724761167,
      "grad_norm": 4.784613609313965,
      "learning_rate": 7.5264443026827226e-06,
      "loss": 0.0553,
      "step": 6547
    },
    {
      "epoch": 0.33813581203201654,
      "grad_norm": 5.601731777191162,
      "learning_rate": 7.525737201258481e-06,
      "loss": 0.0721,
      "step": 6548
    },
    {
      "epoch": 0.33818745158791635,
      "grad_norm": 7.543539047241211,
      "learning_rate": 7.525030032007898e-06,
      "loss": 0.1159,
      "step": 6549
    },
    {
      "epoch": 0.3382390911438162,
      "grad_norm": 3.6886260509490967,
      "learning_rate": 7.524322794949972e-06,
      "loss": 0.0436,
      "step": 6550
    },
    {
      "epoch": 0.338290730699716,
      "grad_norm": 3.673261880874634,
      "learning_rate": 7.5236154901036865e-06,
      "loss": 0.049,
      "step": 6551
    },
    {
      "epoch": 0.3383423702556158,
      "grad_norm": 2.952995777130127,
      "learning_rate": 7.522908117488041e-06,
      "loss": 0.0373,
      "step": 6552
    },
    {
      "epoch": 0.33839400981151563,
      "grad_norm": 2.341343402862549,
      "learning_rate": 7.522200677122029e-06,
      "loss": 0.0375,
      "step": 6553
    },
    {
      "epoch": 0.33844564936741545,
      "grad_norm": 4.709575653076172,
      "learning_rate": 7.521493169024652e-06,
      "loss": 0.047,
      "step": 6554
    },
    {
      "epoch": 0.33849728892331526,
      "grad_norm": 4.002880096435547,
      "learning_rate": 7.520785593214903e-06,
      "loss": 0.0559,
      "step": 6555
    },
    {
      "epoch": 0.3385489284792151,
      "grad_norm": 4.606955051422119,
      "learning_rate": 7.520077949711787e-06,
      "loss": 0.0649,
      "step": 6556
    },
    {
      "epoch": 0.3386005680351149,
      "grad_norm": 6.977701663970947,
      "learning_rate": 7.519370238534307e-06,
      "loss": 0.1049,
      "step": 6557
    },
    {
      "epoch": 0.3386522075910147,
      "grad_norm": 7.24526309967041,
      "learning_rate": 7.518662459701469e-06,
      "loss": 0.0672,
      "step": 6558
    },
    {
      "epoch": 0.33870384714691454,
      "grad_norm": 3.4633710384368896,
      "learning_rate": 7.517954613232278e-06,
      "loss": 0.0376,
      "step": 6559
    },
    {
      "epoch": 0.33875548670281436,
      "grad_norm": 5.630515098571777,
      "learning_rate": 7.517246699145741e-06,
      "loss": 0.0554,
      "step": 6560
    },
    {
      "epoch": 0.3388071262587142,
      "grad_norm": 10.474349975585938,
      "learning_rate": 7.516538717460873e-06,
      "loss": 0.1123,
      "step": 6561
    },
    {
      "epoch": 0.338858765814614,
      "grad_norm": 3.903878927230835,
      "learning_rate": 7.515830668196683e-06,
      "loss": 0.053,
      "step": 6562
    },
    {
      "epoch": 0.3389104053705138,
      "grad_norm": 3.6480743885040283,
      "learning_rate": 7.515122551372185e-06,
      "loss": 0.0445,
      "step": 6563
    },
    {
      "epoch": 0.33896204492641363,
      "grad_norm": 6.084506511688232,
      "learning_rate": 7.514414367006395e-06,
      "loss": 0.033,
      "step": 6564
    },
    {
      "epoch": 0.33901368448231345,
      "grad_norm": 8.510736465454102,
      "learning_rate": 7.513706115118332e-06,
      "loss": 0.1251,
      "step": 6565
    },
    {
      "epoch": 0.33906532403821327,
      "grad_norm": 4.6556477546691895,
      "learning_rate": 7.512997795727014e-06,
      "loss": 0.0363,
      "step": 6566
    },
    {
      "epoch": 0.3391169635941131,
      "grad_norm": 7.690682411193848,
      "learning_rate": 7.512289408851463e-06,
      "loss": 0.0969,
      "step": 6567
    },
    {
      "epoch": 0.3391686031500129,
      "grad_norm": 11.571138381958008,
      "learning_rate": 7.511580954510702e-06,
      "loss": 0.1162,
      "step": 6568
    },
    {
      "epoch": 0.3392202427059127,
      "grad_norm": 5.167801856994629,
      "learning_rate": 7.510872432723756e-06,
      "loss": 0.0301,
      "step": 6569
    },
    {
      "epoch": 0.33927188226181254,
      "grad_norm": 13.308140754699707,
      "learning_rate": 7.51016384350965e-06,
      "loss": 0.1237,
      "step": 6570
    },
    {
      "epoch": 0.33932352181771236,
      "grad_norm": 3.4665310382843018,
      "learning_rate": 7.509455186887416e-06,
      "loss": 0.0615,
      "step": 6571
    },
    {
      "epoch": 0.3393751613736122,
      "grad_norm": 6.370068550109863,
      "learning_rate": 7.508746462876079e-06,
      "loss": 0.1403,
      "step": 6572
    },
    {
      "epoch": 0.339426800929512,
      "grad_norm": 5.706859111785889,
      "learning_rate": 7.508037671494678e-06,
      "loss": 0.0915,
      "step": 6573
    },
    {
      "epoch": 0.3394784404854118,
      "grad_norm": 4.174709320068359,
      "learning_rate": 7.507328812762242e-06,
      "loss": 0.099,
      "step": 6574
    },
    {
      "epoch": 0.33953008004131163,
      "grad_norm": 8.709807395935059,
      "learning_rate": 7.5066198866978075e-06,
      "loss": 0.093,
      "step": 6575
    },
    {
      "epoch": 0.33958171959721145,
      "grad_norm": 2.999373435974121,
      "learning_rate": 7.505910893320413e-06,
      "loss": 0.0493,
      "step": 6576
    },
    {
      "epoch": 0.33963335915311127,
      "grad_norm": 7.767099380493164,
      "learning_rate": 7.505201832649097e-06,
      "loss": 0.0555,
      "step": 6577
    },
    {
      "epoch": 0.3396849987090111,
      "grad_norm": 11.191328048706055,
      "learning_rate": 7.504492704702904e-06,
      "loss": 0.1021,
      "step": 6578
    },
    {
      "epoch": 0.3397366382649109,
      "grad_norm": 4.2576003074646,
      "learning_rate": 7.50378350950087e-06,
      "loss": 0.0787,
      "step": 6579
    },
    {
      "epoch": 0.3397882778208107,
      "grad_norm": 5.725499629974365,
      "learning_rate": 7.5030742470620465e-06,
      "loss": 0.0645,
      "step": 6580
    },
    {
      "epoch": 0.33983991737671054,
      "grad_norm": 3.6119437217712402,
      "learning_rate": 7.5023649174054756e-06,
      "loss": 0.0627,
      "step": 6581
    },
    {
      "epoch": 0.33989155693261036,
      "grad_norm": 5.283507823944092,
      "learning_rate": 7.50165552055021e-06,
      "loss": 0.0827,
      "step": 6582
    },
    {
      "epoch": 0.3399431964885102,
      "grad_norm": 1.6284700632095337,
      "learning_rate": 7.500946056515296e-06,
      "loss": 0.0321,
      "step": 6583
    },
    {
      "epoch": 0.33999483604441,
      "grad_norm": 4.782844543457031,
      "learning_rate": 7.500236525319786e-06,
      "loss": 0.1063,
      "step": 6584
    },
    {
      "epoch": 0.3400464756003098,
      "grad_norm": 2.41679310798645,
      "learning_rate": 7.499526926982737e-06,
      "loss": 0.0421,
      "step": 6585
    },
    {
      "epoch": 0.34009811515620963,
      "grad_norm": 5.595926761627197,
      "learning_rate": 7.498817261523201e-06,
      "loss": 0.0856,
      "step": 6586
    },
    {
      "epoch": 0.34014975471210945,
      "grad_norm": 3.708796739578247,
      "learning_rate": 7.498107528960239e-06,
      "loss": 0.0597,
      "step": 6587
    },
    {
      "epoch": 0.34020139426800927,
      "grad_norm": 2.458237648010254,
      "learning_rate": 7.497397729312906e-06,
      "loss": 0.0347,
      "step": 6588
    },
    {
      "epoch": 0.3402530338239091,
      "grad_norm": 3.569138765335083,
      "learning_rate": 7.496687862600267e-06,
      "loss": 0.0708,
      "step": 6589
    },
    {
      "epoch": 0.3403046733798089,
      "grad_norm": 6.862924575805664,
      "learning_rate": 7.495977928841382e-06,
      "loss": 0.0983,
      "step": 6590
    },
    {
      "epoch": 0.3403563129357087,
      "grad_norm": 4.352903366088867,
      "learning_rate": 7.495267928055315e-06,
      "loss": 0.0628,
      "step": 6591
    },
    {
      "epoch": 0.34040795249160855,
      "grad_norm": 4.98049783706665,
      "learning_rate": 7.494557860261137e-06,
      "loss": 0.0931,
      "step": 6592
    },
    {
      "epoch": 0.34045959204750836,
      "grad_norm": 2.463202953338623,
      "learning_rate": 7.49384772547791e-06,
      "loss": 0.0686,
      "step": 6593
    },
    {
      "epoch": 0.34051123160340824,
      "grad_norm": 7.296339511871338,
      "learning_rate": 7.49313752372471e-06,
      "loss": 0.1212,
      "step": 6594
    },
    {
      "epoch": 0.34056287115930806,
      "grad_norm": 3.2997844219207764,
      "learning_rate": 7.492427255020605e-06,
      "loss": 0.0406,
      "step": 6595
    },
    {
      "epoch": 0.3406145107152079,
      "grad_norm": 7.340175628662109,
      "learning_rate": 7.491716919384668e-06,
      "loss": 0.085,
      "step": 6596
    },
    {
      "epoch": 0.3406661502711077,
      "grad_norm": 5.840261936187744,
      "learning_rate": 7.491006516835978e-06,
      "loss": 0.0866,
      "step": 6597
    },
    {
      "epoch": 0.3407177898270075,
      "grad_norm": 1.935828447341919,
      "learning_rate": 7.49029604739361e-06,
      "loss": 0.0275,
      "step": 6598
    },
    {
      "epoch": 0.34076942938290733,
      "grad_norm": 3.237795829772949,
      "learning_rate": 7.489585511076643e-06,
      "loss": 0.0524,
      "step": 6599
    },
    {
      "epoch": 0.34082106893880715,
      "grad_norm": 3.673651933670044,
      "learning_rate": 7.488874907904158e-06,
      "loss": 0.0448,
      "step": 6600
    },
    {
      "epoch": 0.34087270849470697,
      "grad_norm": 2.721517324447632,
      "learning_rate": 7.488164237895237e-06,
      "loss": 0.0604,
      "step": 6601
    },
    {
      "epoch": 0.3409243480506068,
      "grad_norm": 3.8537721633911133,
      "learning_rate": 7.487453501068967e-06,
      "loss": 0.0444,
      "step": 6602
    },
    {
      "epoch": 0.3409759876065066,
      "grad_norm": 4.589937210083008,
      "learning_rate": 7.486742697444431e-06,
      "loss": 0.0847,
      "step": 6603
    },
    {
      "epoch": 0.3410276271624064,
      "grad_norm": 1.8713793754577637,
      "learning_rate": 7.486031827040717e-06,
      "loss": 0.0205,
      "step": 6604
    },
    {
      "epoch": 0.34107926671830624,
      "grad_norm": 4.435050964355469,
      "learning_rate": 7.485320889876917e-06,
      "loss": 0.0639,
      "step": 6605
    },
    {
      "epoch": 0.34113090627420606,
      "grad_norm": 12.165899276733398,
      "learning_rate": 7.4846098859721216e-06,
      "loss": 0.1491,
      "step": 6606
    },
    {
      "epoch": 0.3411825458301059,
      "grad_norm": 17.11646842956543,
      "learning_rate": 7.483898815345424e-06,
      "loss": 0.1436,
      "step": 6607
    },
    {
      "epoch": 0.3412341853860057,
      "grad_norm": 2.2419753074645996,
      "learning_rate": 7.483187678015919e-06,
      "loss": 0.0316,
      "step": 6608
    },
    {
      "epoch": 0.3412858249419055,
      "grad_norm": 43.87302017211914,
      "learning_rate": 7.4824764740027045e-06,
      "loss": 0.0849,
      "step": 6609
    },
    {
      "epoch": 0.34133746449780533,
      "grad_norm": 3.2271270751953125,
      "learning_rate": 7.481765203324878e-06,
      "loss": 0.0687,
      "step": 6610
    },
    {
      "epoch": 0.34138910405370515,
      "grad_norm": 11.7846097946167,
      "learning_rate": 7.481053866001542e-06,
      "loss": 0.0997,
      "step": 6611
    },
    {
      "epoch": 0.34144074360960497,
      "grad_norm": 3.018048048019409,
      "learning_rate": 7.480342462051795e-06,
      "loss": 0.0634,
      "step": 6612
    },
    {
      "epoch": 0.3414923831655048,
      "grad_norm": 13.641589164733887,
      "learning_rate": 7.479630991494747e-06,
      "loss": 0.1425,
      "step": 6613
    },
    {
      "epoch": 0.3415440227214046,
      "grad_norm": 3.099315881729126,
      "learning_rate": 7.478919454349498e-06,
      "loss": 0.0429,
      "step": 6614
    },
    {
      "epoch": 0.3415956622773044,
      "grad_norm": 3.3054816722869873,
      "learning_rate": 7.478207850635159e-06,
      "loss": 0.0467,
      "step": 6615
    },
    {
      "epoch": 0.34164730183320424,
      "grad_norm": 16.841073989868164,
      "learning_rate": 7.477496180370838e-06,
      "loss": 0.1615,
      "step": 6616
    },
    {
      "epoch": 0.34169894138910406,
      "grad_norm": 2.7741854190826416,
      "learning_rate": 7.476784443575647e-06,
      "loss": 0.0662,
      "step": 6617
    },
    {
      "epoch": 0.3417505809450039,
      "grad_norm": 10.56362533569336,
      "learning_rate": 7.4760726402687e-06,
      "loss": 0.0446,
      "step": 6618
    },
    {
      "epoch": 0.3418022205009037,
      "grad_norm": 4.629213809967041,
      "learning_rate": 7.475360770469109e-06,
      "loss": 0.0684,
      "step": 6619
    },
    {
      "epoch": 0.3418538600568035,
      "grad_norm": 2.4035820960998535,
      "learning_rate": 7.4746488341959936e-06,
      "loss": 0.0339,
      "step": 6620
    },
    {
      "epoch": 0.34190549961270333,
      "grad_norm": 2.7702434062957764,
      "learning_rate": 7.473936831468469e-06,
      "loss": 0.055,
      "step": 6621
    },
    {
      "epoch": 0.34195713916860315,
      "grad_norm": 3.853193998336792,
      "learning_rate": 7.47322476230566e-06,
      "loss": 0.0733,
      "step": 6622
    },
    {
      "epoch": 0.34200877872450297,
      "grad_norm": 10.94140338897705,
      "learning_rate": 7.472512626726684e-06,
      "loss": 0.0918,
      "step": 6623
    },
    {
      "epoch": 0.3420604182804028,
      "grad_norm": 12.222076416015625,
      "learning_rate": 7.471800424750666e-06,
      "loss": 0.0805,
      "step": 6624
    },
    {
      "epoch": 0.3421120578363026,
      "grad_norm": 2.8320229053497314,
      "learning_rate": 7.471088156396735e-06,
      "loss": 0.0487,
      "step": 6625
    },
    {
      "epoch": 0.3421636973922024,
      "grad_norm": 3.2363429069519043,
      "learning_rate": 7.470375821684012e-06,
      "loss": 0.0387,
      "step": 6626
    },
    {
      "epoch": 0.34221533694810224,
      "grad_norm": 5.6428656578063965,
      "learning_rate": 7.469663420631632e-06,
      "loss": 0.052,
      "step": 6627
    },
    {
      "epoch": 0.34226697650400206,
      "grad_norm": 1.7029502391815186,
      "learning_rate": 7.468950953258721e-06,
      "loss": 0.0309,
      "step": 6628
    },
    {
      "epoch": 0.3423186160599019,
      "grad_norm": 3.829223871231079,
      "learning_rate": 7.468238419584416e-06,
      "loss": 0.0595,
      "step": 6629
    },
    {
      "epoch": 0.3423702556158017,
      "grad_norm": 3.056135654449463,
      "learning_rate": 7.467525819627849e-06,
      "loss": 0.0329,
      "step": 6630
    },
    {
      "epoch": 0.3424218951717015,
      "grad_norm": 7.582636833190918,
      "learning_rate": 7.466813153408156e-06,
      "loss": 0.0695,
      "step": 6631
    },
    {
      "epoch": 0.34247353472760134,
      "grad_norm": 1.724338173866272,
      "learning_rate": 7.466100420944476e-06,
      "loss": 0.0149,
      "step": 6632
    },
    {
      "epoch": 0.34252517428350115,
      "grad_norm": 6.250033378601074,
      "learning_rate": 7.465387622255949e-06,
      "loss": 0.106,
      "step": 6633
    },
    {
      "epoch": 0.342576813839401,
      "grad_norm": 8.276140213012695,
      "learning_rate": 7.464674757361715e-06,
      "loss": 0.0953,
      "step": 6634
    },
    {
      "epoch": 0.3426284533953008,
      "grad_norm": 2.6760799884796143,
      "learning_rate": 7.463961826280919e-06,
      "loss": 0.0229,
      "step": 6635
    },
    {
      "epoch": 0.3426800929512006,
      "grad_norm": 15.651623725891113,
      "learning_rate": 7.4632488290327064e-06,
      "loss": 0.0689,
      "step": 6636
    },
    {
      "epoch": 0.34273173250710043,
      "grad_norm": 3.4118492603302,
      "learning_rate": 7.462535765636222e-06,
      "loss": 0.0291,
      "step": 6637
    },
    {
      "epoch": 0.34278337206300025,
      "grad_norm": 4.327075958251953,
      "learning_rate": 7.461822636110615e-06,
      "loss": 0.0306,
      "step": 6638
    },
    {
      "epoch": 0.34283501161890007,
      "grad_norm": 3.2237868309020996,
      "learning_rate": 7.461109440475036e-06,
      "loss": 0.0595,
      "step": 6639
    },
    {
      "epoch": 0.3428866511747999,
      "grad_norm": 24.384502410888672,
      "learning_rate": 7.460396178748639e-06,
      "loss": 0.0626,
      "step": 6640
    },
    {
      "epoch": 0.3429382907306997,
      "grad_norm": 5.64939022064209,
      "learning_rate": 7.459682850950576e-06,
      "loss": 0.117,
      "step": 6641
    },
    {
      "epoch": 0.3429899302865995,
      "grad_norm": 7.262131214141846,
      "learning_rate": 7.458969457100004e-06,
      "loss": 0.0598,
      "step": 6642
    },
    {
      "epoch": 0.34304156984249934,
      "grad_norm": 5.780760288238525,
      "learning_rate": 7.458255997216079e-06,
      "loss": 0.0693,
      "step": 6643
    },
    {
      "epoch": 0.34309320939839916,
      "grad_norm": 3.7280008792877197,
      "learning_rate": 7.457542471317962e-06,
      "loss": 0.0427,
      "step": 6644
    },
    {
      "epoch": 0.343144848954299,
      "grad_norm": 2.748623847961426,
      "learning_rate": 7.456828879424811e-06,
      "loss": 0.0206,
      "step": 6645
    },
    {
      "epoch": 0.3431964885101988,
      "grad_norm": 4.64517879486084,
      "learning_rate": 7.456115221555794e-06,
      "loss": 0.053,
      "step": 6646
    },
    {
      "epoch": 0.3432481280660986,
      "grad_norm": 4.2041850090026855,
      "learning_rate": 7.455401497730072e-06,
      "loss": 0.0618,
      "step": 6647
    },
    {
      "epoch": 0.34329976762199843,
      "grad_norm": 4.34185791015625,
      "learning_rate": 7.454687707966811e-06,
      "loss": 0.0611,
      "step": 6648
    },
    {
      "epoch": 0.34335140717789825,
      "grad_norm": 7.679950714111328,
      "learning_rate": 7.453973852285181e-06,
      "loss": 0.0777,
      "step": 6649
    },
    {
      "epoch": 0.34340304673379807,
      "grad_norm": 14.886635780334473,
      "learning_rate": 7.453259930704351e-06,
      "loss": 0.3317,
      "step": 6650
    },
    {
      "epoch": 0.3434546862896979,
      "grad_norm": 7.9946608543396,
      "learning_rate": 7.4525459432434946e-06,
      "loss": 0.0665,
      "step": 6651
    },
    {
      "epoch": 0.3435063258455977,
      "grad_norm": 5.104147434234619,
      "learning_rate": 7.451831889921781e-06,
      "loss": 0.0545,
      "step": 6652
    },
    {
      "epoch": 0.3435579654014975,
      "grad_norm": 3.3711137771606445,
      "learning_rate": 7.45111777075839e-06,
      "loss": 0.0329,
      "step": 6653
    },
    {
      "epoch": 0.34360960495739734,
      "grad_norm": 1.7213293313980103,
      "learning_rate": 7.4504035857724945e-06,
      "loss": 0.0201,
      "step": 6654
    },
    {
      "epoch": 0.34366124451329716,
      "grad_norm": 4.185669422149658,
      "learning_rate": 7.449689334983278e-06,
      "loss": 0.0813,
      "step": 6655
    },
    {
      "epoch": 0.343712884069197,
      "grad_norm": 3.063854455947876,
      "learning_rate": 7.4489750184099165e-06,
      "loss": 0.0686,
      "step": 6656
    },
    {
      "epoch": 0.34376452362509685,
      "grad_norm": 2.311460494995117,
      "learning_rate": 7.448260636071595e-06,
      "loss": 0.0249,
      "step": 6657
    },
    {
      "epoch": 0.34381616318099667,
      "grad_norm": 9.937765121459961,
      "learning_rate": 7.4475461879874975e-06,
      "loss": 0.0971,
      "step": 6658
    },
    {
      "epoch": 0.3438678027368965,
      "grad_norm": 3.502530336380005,
      "learning_rate": 7.446831674176808e-06,
      "loss": 0.0444,
      "step": 6659
    },
    {
      "epoch": 0.3439194422927963,
      "grad_norm": 13.420979499816895,
      "learning_rate": 7.446117094658716e-06,
      "loss": 0.1468,
      "step": 6660
    },
    {
      "epoch": 0.3439710818486961,
      "grad_norm": 5.259005069732666,
      "learning_rate": 7.445402449452409e-06,
      "loss": 0.0539,
      "step": 6661
    },
    {
      "epoch": 0.34402272140459594,
      "grad_norm": 5.492711067199707,
      "learning_rate": 7.444687738577081e-06,
      "loss": 0.0499,
      "step": 6662
    },
    {
      "epoch": 0.34407436096049576,
      "grad_norm": 2.9750473499298096,
      "learning_rate": 7.443972962051923e-06,
      "loss": 0.0366,
      "step": 6663
    },
    {
      "epoch": 0.3441260005163956,
      "grad_norm": 4.194118976593018,
      "learning_rate": 7.443258119896129e-06,
      "loss": 0.0585,
      "step": 6664
    },
    {
      "epoch": 0.3441776400722954,
      "grad_norm": 9.031234741210938,
      "learning_rate": 7.442543212128896e-06,
      "loss": 0.0417,
      "step": 6665
    },
    {
      "epoch": 0.3442292796281952,
      "grad_norm": 4.511330604553223,
      "learning_rate": 7.441828238769424e-06,
      "loss": 0.0499,
      "step": 6666
    },
    {
      "epoch": 0.34428091918409504,
      "grad_norm": 7.747670650482178,
      "learning_rate": 7.441113199836912e-06,
      "loss": 0.0602,
      "step": 6667
    },
    {
      "epoch": 0.34433255873999485,
      "grad_norm": 5.248988628387451,
      "learning_rate": 7.4403980953505585e-06,
      "loss": 0.0757,
      "step": 6668
    },
    {
      "epoch": 0.3443841982958947,
      "grad_norm": 11.845100402832031,
      "learning_rate": 7.439682925329572e-06,
      "loss": 0.0815,
      "step": 6669
    },
    {
      "epoch": 0.3444358378517945,
      "grad_norm": 5.508826732635498,
      "learning_rate": 7.438967689793155e-06,
      "loss": 0.06,
      "step": 6670
    },
    {
      "epoch": 0.3444874774076943,
      "grad_norm": 2.0955145359039307,
      "learning_rate": 7.438252388760514e-06,
      "loss": 0.0421,
      "step": 6671
    },
    {
      "epoch": 0.3445391169635941,
      "grad_norm": 12.65635871887207,
      "learning_rate": 7.437537022250858e-06,
      "loss": 0.0709,
      "step": 6672
    },
    {
      "epoch": 0.34459075651949395,
      "grad_norm": 4.952422142028809,
      "learning_rate": 7.4368215902834e-06,
      "loss": 0.0522,
      "step": 6673
    },
    {
      "epoch": 0.34464239607539376,
      "grad_norm": 21.864469528198242,
      "learning_rate": 7.436106092877349e-06,
      "loss": 0.1424,
      "step": 6674
    },
    {
      "epoch": 0.3446940356312936,
      "grad_norm": 5.085047721862793,
      "learning_rate": 7.43539053005192e-06,
      "loss": 0.069,
      "step": 6675
    },
    {
      "epoch": 0.3447456751871934,
      "grad_norm": 12.230273246765137,
      "learning_rate": 7.434674901826329e-06,
      "loss": 0.1523,
      "step": 6676
    },
    {
      "epoch": 0.3447973147430932,
      "grad_norm": 15.921876907348633,
      "learning_rate": 7.433959208219796e-06,
      "loss": 0.0923,
      "step": 6677
    },
    {
      "epoch": 0.34484895429899304,
      "grad_norm": 10.702237129211426,
      "learning_rate": 7.4332434492515355e-06,
      "loss": 0.0641,
      "step": 6678
    },
    {
      "epoch": 0.34490059385489286,
      "grad_norm": 9.57306957244873,
      "learning_rate": 7.432527624940773e-06,
      "loss": 0.0909,
      "step": 6679
    },
    {
      "epoch": 0.3449522334107927,
      "grad_norm": 5.425619125366211,
      "learning_rate": 7.431811735306729e-06,
      "loss": 0.0501,
      "step": 6680
    },
    {
      "epoch": 0.3450038729666925,
      "grad_norm": 4.631422996520996,
      "learning_rate": 7.4310957803686264e-06,
      "loss": 0.0333,
      "step": 6681
    },
    {
      "epoch": 0.3450555125225923,
      "grad_norm": 4.030004978179932,
      "learning_rate": 7.430379760145696e-06,
      "loss": 0.0682,
      "step": 6682
    },
    {
      "epoch": 0.34510715207849213,
      "grad_norm": 9.209311485290527,
      "learning_rate": 7.4296636746571616e-06,
      "loss": 0.0899,
      "step": 6683
    },
    {
      "epoch": 0.34515879163439195,
      "grad_norm": 2.8257813453674316,
      "learning_rate": 7.428947523922256e-06,
      "loss": 0.0406,
      "step": 6684
    },
    {
      "epoch": 0.34521043119029177,
      "grad_norm": 15.869490623474121,
      "learning_rate": 7.428231307960209e-06,
      "loss": 0.1322,
      "step": 6685
    },
    {
      "epoch": 0.3452620707461916,
      "grad_norm": 4.972014427185059,
      "learning_rate": 7.427515026790255e-06,
      "loss": 0.0415,
      "step": 6686
    },
    {
      "epoch": 0.3453137103020914,
      "grad_norm": 4.540088176727295,
      "learning_rate": 7.426798680431626e-06,
      "loss": 0.0647,
      "step": 6687
    },
    {
      "epoch": 0.3453653498579912,
      "grad_norm": 3.342067003250122,
      "learning_rate": 7.426082268903565e-06,
      "loss": 0.0306,
      "step": 6688
    },
    {
      "epoch": 0.34541698941389104,
      "grad_norm": 4.60732889175415,
      "learning_rate": 7.425365792225305e-06,
      "loss": 0.0492,
      "step": 6689
    },
    {
      "epoch": 0.34546862896979086,
      "grad_norm": 4.660549640655518,
      "learning_rate": 7.424649250416089e-06,
      "loss": 0.039,
      "step": 6690
    },
    {
      "epoch": 0.3455202685256907,
      "grad_norm": 5.18461799621582,
      "learning_rate": 7.423932643495158e-06,
      "loss": 0.0709,
      "step": 6691
    },
    {
      "epoch": 0.3455719080815905,
      "grad_norm": 3.1712472438812256,
      "learning_rate": 7.423215971481754e-06,
      "loss": 0.0457,
      "step": 6692
    },
    {
      "epoch": 0.3456235476374903,
      "grad_norm": 2.6491243839263916,
      "learning_rate": 7.422499234395128e-06,
      "loss": 0.064,
      "step": 6693
    },
    {
      "epoch": 0.34567518719339013,
      "grad_norm": 6.193986892700195,
      "learning_rate": 7.421782432254522e-06,
      "loss": 0.0663,
      "step": 6694
    },
    {
      "epoch": 0.34572682674928995,
      "grad_norm": 1.7690969705581665,
      "learning_rate": 7.421065565079189e-06,
      "loss": 0.0109,
      "step": 6695
    },
    {
      "epoch": 0.34577846630518977,
      "grad_norm": 10.56119441986084,
      "learning_rate": 7.420348632888376e-06,
      "loss": 0.1575,
      "step": 6696
    },
    {
      "epoch": 0.3458301058610896,
      "grad_norm": 3.7573537826538086,
      "learning_rate": 7.419631635701338e-06,
      "loss": 0.0376,
      "step": 6697
    },
    {
      "epoch": 0.3458817454169894,
      "grad_norm": 4.099913120269775,
      "learning_rate": 7.418914573537328e-06,
      "loss": 0.0263,
      "step": 6698
    },
    {
      "epoch": 0.3459333849728892,
      "grad_norm": 18.043794631958008,
      "learning_rate": 7.4181974464156025e-06,
      "loss": 0.0829,
      "step": 6699
    },
    {
      "epoch": 0.34598502452878904,
      "grad_norm": 6.273175239562988,
      "learning_rate": 7.41748025435542e-06,
      "loss": 0.0284,
      "step": 6700
    },
    {
      "epoch": 0.34603666408468886,
      "grad_norm": 5.0496649742126465,
      "learning_rate": 7.4167629973760405e-06,
      "loss": 0.0734,
      "step": 6701
    },
    {
      "epoch": 0.3460883036405887,
      "grad_norm": 25.730974197387695,
      "learning_rate": 7.416045675496725e-06,
      "loss": 0.0909,
      "step": 6702
    },
    {
      "epoch": 0.3461399431964885,
      "grad_norm": 4.7668046951293945,
      "learning_rate": 7.415328288736734e-06,
      "loss": 0.0848,
      "step": 6703
    },
    {
      "epoch": 0.3461915827523883,
      "grad_norm": 26.417810440063477,
      "learning_rate": 7.414610837115335e-06,
      "loss": 0.0848,
      "step": 6704
    },
    {
      "epoch": 0.34624322230828813,
      "grad_norm": 5.827630519866943,
      "learning_rate": 7.4138933206517925e-06,
      "loss": 0.1135,
      "step": 6705
    },
    {
      "epoch": 0.34629486186418795,
      "grad_norm": 5.732945919036865,
      "learning_rate": 7.413175739365377e-06,
      "loss": 0.1163,
      "step": 6706
    },
    {
      "epoch": 0.34634650142008777,
      "grad_norm": 10.791389465332031,
      "learning_rate": 7.412458093275358e-06,
      "loss": 0.0627,
      "step": 6707
    },
    {
      "epoch": 0.3463981409759876,
      "grad_norm": 3.213329553604126,
      "learning_rate": 7.411740382401005e-06,
      "loss": 0.0467,
      "step": 6708
    },
    {
      "epoch": 0.3464497805318874,
      "grad_norm": 3.3654325008392334,
      "learning_rate": 7.4110226067615935e-06,
      "loss": 0.0512,
      "step": 6709
    },
    {
      "epoch": 0.3465014200877872,
      "grad_norm": 1.8778741359710693,
      "learning_rate": 7.410304766376399e-06,
      "loss": 0.0359,
      "step": 6710
    },
    {
      "epoch": 0.34655305964368704,
      "grad_norm": 3.366579532623291,
      "learning_rate": 7.409586861264696e-06,
      "loss": 0.0564,
      "step": 6711
    },
    {
      "epoch": 0.34660469919958686,
      "grad_norm": 3.7533860206604004,
      "learning_rate": 7.408868891445767e-06,
      "loss": 0.0483,
      "step": 6712
    },
    {
      "epoch": 0.3466563387554867,
      "grad_norm": 7.065947532653809,
      "learning_rate": 7.4081508569388895e-06,
      "loss": 0.0829,
      "step": 6713
    },
    {
      "epoch": 0.3467079783113865,
      "grad_norm": 3.9266233444213867,
      "learning_rate": 7.407432757763345e-06,
      "loss": 0.0739,
      "step": 6714
    },
    {
      "epoch": 0.3467596178672863,
      "grad_norm": 7.2995405197143555,
      "learning_rate": 7.40671459393842e-06,
      "loss": 0.0579,
      "step": 6715
    },
    {
      "epoch": 0.34681125742318614,
      "grad_norm": 2.936072826385498,
      "learning_rate": 7.4059963654833985e-06,
      "loss": 0.037,
      "step": 6716
    },
    {
      "epoch": 0.34686289697908596,
      "grad_norm": 9.241636276245117,
      "learning_rate": 7.40527807241757e-06,
      "loss": 0.102,
      "step": 6717
    },
    {
      "epoch": 0.3469145365349858,
      "grad_norm": 15.644750595092773,
      "learning_rate": 7.404559714760219e-06,
      "loss": 0.0716,
      "step": 6718
    },
    {
      "epoch": 0.3469661760908856,
      "grad_norm": 13.988494873046875,
      "learning_rate": 7.403841292530641e-06,
      "loss": 0.1618,
      "step": 6719
    },
    {
      "epoch": 0.3470178156467854,
      "grad_norm": 5.276730060577393,
      "learning_rate": 7.4031228057481265e-06,
      "loss": 0.091,
      "step": 6720
    },
    {
      "epoch": 0.3470694552026853,
      "grad_norm": 9.190788269042969,
      "learning_rate": 7.402404254431971e-06,
      "loss": 0.0818,
      "step": 6721
    },
    {
      "epoch": 0.3471210947585851,
      "grad_norm": 4.008764743804932,
      "learning_rate": 7.401685638601468e-06,
      "loss": 0.0605,
      "step": 6722
    },
    {
      "epoch": 0.3471727343144849,
      "grad_norm": 13.506451606750488,
      "learning_rate": 7.400966958275917e-06,
      "loss": 0.096,
      "step": 6723
    },
    {
      "epoch": 0.34722437387038474,
      "grad_norm": 4.528697967529297,
      "learning_rate": 7.400248213474619e-06,
      "loss": 0.0603,
      "step": 6724
    },
    {
      "epoch": 0.34727601342628456,
      "grad_norm": 6.188925743103027,
      "learning_rate": 7.399529404216872e-06,
      "loss": 0.0686,
      "step": 6725
    },
    {
      "epoch": 0.3473276529821844,
      "grad_norm": 4.407393932342529,
      "learning_rate": 7.398810530521982e-06,
      "loss": 0.0528,
      "step": 6726
    },
    {
      "epoch": 0.3473792925380842,
      "grad_norm": 13.31883716583252,
      "learning_rate": 7.39809159240925e-06,
      "loss": 0.178,
      "step": 6727
    },
    {
      "epoch": 0.347430932093984,
      "grad_norm": 3.281867265701294,
      "learning_rate": 7.397372589897986e-06,
      "loss": 0.0837,
      "step": 6728
    },
    {
      "epoch": 0.34748257164988383,
      "grad_norm": 17.733257293701172,
      "learning_rate": 7.396653523007498e-06,
      "loss": 0.1123,
      "step": 6729
    },
    {
      "epoch": 0.34753421120578365,
      "grad_norm": 3.9434990882873535,
      "learning_rate": 7.395934391757092e-06,
      "loss": 0.0617,
      "step": 6730
    },
    {
      "epoch": 0.34758585076168347,
      "grad_norm": 8.359476089477539,
      "learning_rate": 7.395215196166084e-06,
      "loss": 0.0711,
      "step": 6731
    },
    {
      "epoch": 0.3476374903175833,
      "grad_norm": 10.328963279724121,
      "learning_rate": 7.394495936253785e-06,
      "loss": 0.0915,
      "step": 6732
    },
    {
      "epoch": 0.3476891298734831,
      "grad_norm": 8.785053253173828,
      "learning_rate": 7.393776612039513e-06,
      "loss": 0.1056,
      "step": 6733
    },
    {
      "epoch": 0.3477407694293829,
      "grad_norm": 6.979983329772949,
      "learning_rate": 7.393057223542579e-06,
      "loss": 0.0956,
      "step": 6734
    },
    {
      "epoch": 0.34779240898528274,
      "grad_norm": 3.1725189685821533,
      "learning_rate": 7.392337770782308e-06,
      "loss": 0.0561,
      "step": 6735
    },
    {
      "epoch": 0.34784404854118256,
      "grad_norm": 13.707794189453125,
      "learning_rate": 7.391618253778015e-06,
      "loss": 0.1573,
      "step": 6736
    },
    {
      "epoch": 0.3478956880970824,
      "grad_norm": 10.902050971984863,
      "learning_rate": 7.390898672549025e-06,
      "loss": 0.124,
      "step": 6737
    },
    {
      "epoch": 0.3479473276529822,
      "grad_norm": 6.894827842712402,
      "learning_rate": 7.39017902711466e-06,
      "loss": 0.0794,
      "step": 6738
    },
    {
      "epoch": 0.347998967208882,
      "grad_norm": 8.778366088867188,
      "learning_rate": 7.389459317494248e-06,
      "loss": 0.1025,
      "step": 6739
    },
    {
      "epoch": 0.34805060676478183,
      "grad_norm": 2.4043707847595215,
      "learning_rate": 7.388739543707113e-06,
      "loss": 0.0434,
      "step": 6740
    },
    {
      "epoch": 0.34810224632068165,
      "grad_norm": 7.089170455932617,
      "learning_rate": 7.3880197057725856e-06,
      "loss": 0.0704,
      "step": 6741
    },
    {
      "epoch": 0.34815388587658147,
      "grad_norm": 6.926032066345215,
      "learning_rate": 7.387299803709996e-06,
      "loss": 0.0878,
      "step": 6742
    },
    {
      "epoch": 0.3482055254324813,
      "grad_norm": 6.2354207038879395,
      "learning_rate": 7.3865798375386775e-06,
      "loss": 0.0406,
      "step": 6743
    },
    {
      "epoch": 0.3482571649883811,
      "grad_norm": 2.0114541053771973,
      "learning_rate": 7.385859807277963e-06,
      "loss": 0.0537,
      "step": 6744
    },
    {
      "epoch": 0.3483088045442809,
      "grad_norm": 3.6255717277526855,
      "learning_rate": 7.385139712947187e-06,
      "loss": 0.0667,
      "step": 6745
    },
    {
      "epoch": 0.34836044410018074,
      "grad_norm": 4.881816864013672,
      "learning_rate": 7.38441955456569e-06,
      "loss": 0.0369,
      "step": 6746
    },
    {
      "epoch": 0.34841208365608056,
      "grad_norm": 4.124273777008057,
      "learning_rate": 7.38369933215281e-06,
      "loss": 0.0608,
      "step": 6747
    },
    {
      "epoch": 0.3484637232119804,
      "grad_norm": 3.5479376316070557,
      "learning_rate": 7.382979045727885e-06,
      "loss": 0.0738,
      "step": 6748
    },
    {
      "epoch": 0.3485153627678802,
      "grad_norm": 4.962547779083252,
      "learning_rate": 7.382258695310262e-06,
      "loss": 0.0505,
      "step": 6749
    },
    {
      "epoch": 0.34856700232378,
      "grad_norm": 8.703600883483887,
      "learning_rate": 7.3815382809192835e-06,
      "loss": 0.1308,
      "step": 6750
    },
    {
      "epoch": 0.34861864187967984,
      "grad_norm": 5.781859874725342,
      "learning_rate": 7.380817802574294e-06,
      "loss": 0.06,
      "step": 6751
    },
    {
      "epoch": 0.34867028143557965,
      "grad_norm": 9.915474891662598,
      "learning_rate": 7.380097260294645e-06,
      "loss": 0.073,
      "step": 6752
    },
    {
      "epoch": 0.3487219209914795,
      "grad_norm": 11.783187866210938,
      "learning_rate": 7.3793766540996825e-06,
      "loss": 0.1649,
      "step": 6753
    },
    {
      "epoch": 0.3487735605473793,
      "grad_norm": 8.353425979614258,
      "learning_rate": 7.37865598400876e-06,
      "loss": 0.103,
      "step": 6754
    },
    {
      "epoch": 0.3488252001032791,
      "grad_norm": 3.309572219848633,
      "learning_rate": 7.377935250041228e-06,
      "loss": 0.0255,
      "step": 6755
    },
    {
      "epoch": 0.34887683965917893,
      "grad_norm": 4.136860370635986,
      "learning_rate": 7.377214452216444e-06,
      "loss": 0.0876,
      "step": 6756
    },
    {
      "epoch": 0.34892847921507875,
      "grad_norm": 6.591407299041748,
      "learning_rate": 7.376493590553764e-06,
      "loss": 0.0908,
      "step": 6757
    },
    {
      "epoch": 0.34898011877097856,
      "grad_norm": 8.95669937133789,
      "learning_rate": 7.375772665072542e-06,
      "loss": 0.1142,
      "step": 6758
    },
    {
      "epoch": 0.3490317583268784,
      "grad_norm": 2.1998062133789062,
      "learning_rate": 7.375051675792143e-06,
      "loss": 0.0303,
      "step": 6759
    },
    {
      "epoch": 0.3490833978827782,
      "grad_norm": 19.867755889892578,
      "learning_rate": 7.374330622731925e-06,
      "loss": 0.1957,
      "step": 6760
    },
    {
      "epoch": 0.349135037438678,
      "grad_norm": 5.752110481262207,
      "learning_rate": 7.373609505911253e-06,
      "loss": 0.077,
      "step": 6761
    },
    {
      "epoch": 0.34918667699457784,
      "grad_norm": 2.4693427085876465,
      "learning_rate": 7.372888325349492e-06,
      "loss": 0.0389,
      "step": 6762
    },
    {
      "epoch": 0.34923831655047766,
      "grad_norm": 4.832016468048096,
      "learning_rate": 7.372167081066007e-06,
      "loss": 0.0511,
      "step": 6763
    },
    {
      "epoch": 0.3492899561063775,
      "grad_norm": 1.8561232089996338,
      "learning_rate": 7.3714457730801685e-06,
      "loss": 0.0374,
      "step": 6764
    },
    {
      "epoch": 0.3493415956622773,
      "grad_norm": 2.837902545928955,
      "learning_rate": 7.370724401411346e-06,
      "loss": 0.0249,
      "step": 6765
    },
    {
      "epoch": 0.3493932352181771,
      "grad_norm": 4.966628551483154,
      "learning_rate": 7.370002966078911e-06,
      "loss": 0.0279,
      "step": 6766
    },
    {
      "epoch": 0.34944487477407693,
      "grad_norm": 3.7871222496032715,
      "learning_rate": 7.369281467102234e-06,
      "loss": 0.0792,
      "step": 6767
    },
    {
      "epoch": 0.34949651432997675,
      "grad_norm": 3.9148528575897217,
      "learning_rate": 7.368559904500695e-06,
      "loss": 0.0494,
      "step": 6768
    },
    {
      "epoch": 0.34954815388587657,
      "grad_norm": 1.976881742477417,
      "learning_rate": 7.367838278293669e-06,
      "loss": 0.0103,
      "step": 6769
    },
    {
      "epoch": 0.3495997934417764,
      "grad_norm": 20.88225746154785,
      "learning_rate": 7.3671165885005336e-06,
      "loss": 0.1505,
      "step": 6770
    },
    {
      "epoch": 0.3496514329976762,
      "grad_norm": 2.0622782707214355,
      "learning_rate": 7.36639483514067e-06,
      "loss": 0.0382,
      "step": 6771
    },
    {
      "epoch": 0.349703072553576,
      "grad_norm": 2.05117130279541,
      "learning_rate": 7.3656730182334616e-06,
      "loss": 0.0295,
      "step": 6772
    },
    {
      "epoch": 0.34975471210947584,
      "grad_norm": 2.985480308532715,
      "learning_rate": 7.364951137798289e-06,
      "loss": 0.0636,
      "step": 6773
    },
    {
      "epoch": 0.34980635166537566,
      "grad_norm": 7.705911636352539,
      "learning_rate": 7.364229193854539e-06,
      "loss": 0.0417,
      "step": 6774
    },
    {
      "epoch": 0.3498579912212755,
      "grad_norm": 4.319073677062988,
      "learning_rate": 7.363507186421601e-06,
      "loss": 0.0501,
      "step": 6775
    },
    {
      "epoch": 0.3499096307771753,
      "grad_norm": 5.632095813751221,
      "learning_rate": 7.362785115518861e-06,
      "loss": 0.0905,
      "step": 6776
    },
    {
      "epoch": 0.3499612703330751,
      "grad_norm": 27.33164405822754,
      "learning_rate": 7.362062981165711e-06,
      "loss": 0.1048,
      "step": 6777
    },
    {
      "epoch": 0.35001290988897493,
      "grad_norm": 3.9077248573303223,
      "learning_rate": 7.361340783381542e-06,
      "loss": 0.0726,
      "step": 6778
    },
    {
      "epoch": 0.35006454944487475,
      "grad_norm": 17.83705711364746,
      "learning_rate": 7.360618522185749e-06,
      "loss": 0.1211,
      "step": 6779
    },
    {
      "epoch": 0.35011618900077457,
      "grad_norm": 7.744838714599609,
      "learning_rate": 7.359896197597727e-06,
      "loss": 0.0732,
      "step": 6780
    },
    {
      "epoch": 0.3501678285566744,
      "grad_norm": 2.9228875637054443,
      "learning_rate": 7.359173809636874e-06,
      "loss": 0.0246,
      "step": 6781
    },
    {
      "epoch": 0.3502194681125742,
      "grad_norm": 14.903768539428711,
      "learning_rate": 7.358451358322588e-06,
      "loss": 0.2087,
      "step": 6782
    },
    {
      "epoch": 0.350271107668474,
      "grad_norm": 2.4952852725982666,
      "learning_rate": 7.3577288436742725e-06,
      "loss": 0.0406,
      "step": 6783
    },
    {
      "epoch": 0.35032274722437384,
      "grad_norm": 3.1638779640197754,
      "learning_rate": 7.357006265711325e-06,
      "loss": 0.0329,
      "step": 6784
    },
    {
      "epoch": 0.3503743867802737,
      "grad_norm": 10.137405395507812,
      "learning_rate": 7.356283624453155e-06,
      "loss": 0.1783,
      "step": 6785
    },
    {
      "epoch": 0.35042602633617354,
      "grad_norm": 6.197693347930908,
      "learning_rate": 7.355560919919164e-06,
      "loss": 0.0711,
      "step": 6786
    },
    {
      "epoch": 0.35047766589207335,
      "grad_norm": 6.924429893493652,
      "learning_rate": 7.354838152128765e-06,
      "loss": 0.0877,
      "step": 6787
    },
    {
      "epoch": 0.35052930544797317,
      "grad_norm": 1.7508876323699951,
      "learning_rate": 7.354115321101361e-06,
      "loss": 0.0335,
      "step": 6788
    },
    {
      "epoch": 0.350580945003873,
      "grad_norm": 10.12918472290039,
      "learning_rate": 7.3533924268563664e-06,
      "loss": 0.1352,
      "step": 6789
    },
    {
      "epoch": 0.3506325845597728,
      "grad_norm": 1.7682335376739502,
      "learning_rate": 7.352669469413194e-06,
      "loss": 0.0326,
      "step": 6790
    },
    {
      "epoch": 0.3506842241156726,
      "grad_norm": 5.921465873718262,
      "learning_rate": 7.3519464487912565e-06,
      "loss": 0.0494,
      "step": 6791
    },
    {
      "epoch": 0.35073586367157245,
      "grad_norm": 6.327697277069092,
      "learning_rate": 7.351223365009972e-06,
      "loss": 0.0523,
      "step": 6792
    },
    {
      "epoch": 0.35078750322747226,
      "grad_norm": 12.43214225769043,
      "learning_rate": 7.350500218088754e-06,
      "loss": 0.1136,
      "step": 6793
    },
    {
      "epoch": 0.3508391427833721,
      "grad_norm": 8.64200210571289,
      "learning_rate": 7.349777008047028e-06,
      "loss": 0.098,
      "step": 6794
    },
    {
      "epoch": 0.3508907823392719,
      "grad_norm": 2.3676540851593018,
      "learning_rate": 7.349053734904212e-06,
      "loss": 0.0276,
      "step": 6795
    },
    {
      "epoch": 0.3509424218951717,
      "grad_norm": 5.261247634887695,
      "learning_rate": 7.348330398679729e-06,
      "loss": 0.0491,
      "step": 6796
    },
    {
      "epoch": 0.35099406145107154,
      "grad_norm": 2.0885350704193115,
      "learning_rate": 7.347606999393002e-06,
      "loss": 0.0273,
      "step": 6797
    },
    {
      "epoch": 0.35104570100697136,
      "grad_norm": 4.946341514587402,
      "learning_rate": 7.346883537063459e-06,
      "loss": 0.0329,
      "step": 6798
    },
    {
      "epoch": 0.3510973405628712,
      "grad_norm": 1.525205373764038,
      "learning_rate": 7.346160011710529e-06,
      "loss": 0.0246,
      "step": 6799
    },
    {
      "epoch": 0.351148980118771,
      "grad_norm": 9.697260856628418,
      "learning_rate": 7.345436423353638e-06,
      "loss": 0.0881,
      "step": 6800
    },
    {
      "epoch": 0.3512006196746708,
      "grad_norm": 7.053563594818115,
      "learning_rate": 7.344712772012222e-06,
      "loss": 0.0973,
      "step": 6801
    },
    {
      "epoch": 0.35125225923057063,
      "grad_norm": 3.1456565856933594,
      "learning_rate": 7.343989057705711e-06,
      "loss": 0.0633,
      "step": 6802
    },
    {
      "epoch": 0.35130389878647045,
      "grad_norm": 8.191737174987793,
      "learning_rate": 7.3432652804535395e-06,
      "loss": 0.0572,
      "step": 6803
    },
    {
      "epoch": 0.35135553834237027,
      "grad_norm": 2.8459227085113525,
      "learning_rate": 7.342541440275146e-06,
      "loss": 0.0522,
      "step": 6804
    },
    {
      "epoch": 0.3514071778982701,
      "grad_norm": 1.7971117496490479,
      "learning_rate": 7.341817537189966e-06,
      "loss": 0.0305,
      "step": 6805
    },
    {
      "epoch": 0.3514588174541699,
      "grad_norm": 5.590219020843506,
      "learning_rate": 7.3410935712174405e-06,
      "loss": 0.0638,
      "step": 6806
    },
    {
      "epoch": 0.3515104570100697,
      "grad_norm": 4.5846147537231445,
      "learning_rate": 7.340369542377011e-06,
      "loss": 0.113,
      "step": 6807
    },
    {
      "epoch": 0.35156209656596954,
      "grad_norm": 3.0425846576690674,
      "learning_rate": 7.33964545068812e-06,
      "loss": 0.0572,
      "step": 6808
    },
    {
      "epoch": 0.35161373612186936,
      "grad_norm": 11.672551155090332,
      "learning_rate": 7.338921296170214e-06,
      "loss": 0.1134,
      "step": 6809
    },
    {
      "epoch": 0.3516653756777692,
      "grad_norm": 17.967233657836914,
      "learning_rate": 7.338197078842737e-06,
      "loss": 0.161,
      "step": 6810
    },
    {
      "epoch": 0.351717015233669,
      "grad_norm": 2.505220651626587,
      "learning_rate": 7.337472798725138e-06,
      "loss": 0.035,
      "step": 6811
    },
    {
      "epoch": 0.3517686547895688,
      "grad_norm": 16.419282913208008,
      "learning_rate": 7.336748455836868e-06,
      "loss": 0.1681,
      "step": 6812
    },
    {
      "epoch": 0.35182029434546863,
      "grad_norm": 8.207189559936523,
      "learning_rate": 7.336024050197379e-06,
      "loss": 0.1381,
      "step": 6813
    },
    {
      "epoch": 0.35187193390136845,
      "grad_norm": 13.353719711303711,
      "learning_rate": 7.335299581826122e-06,
      "loss": 0.1699,
      "step": 6814
    },
    {
      "epoch": 0.35192357345726827,
      "grad_norm": 6.3255462646484375,
      "learning_rate": 7.334575050742554e-06,
      "loss": 0.1071,
      "step": 6815
    },
    {
      "epoch": 0.3519752130131681,
      "grad_norm": 3.993110179901123,
      "learning_rate": 7.333850456966129e-06,
      "loss": 0.0953,
      "step": 6816
    },
    {
      "epoch": 0.3520268525690679,
      "grad_norm": 6.049075126647949,
      "learning_rate": 7.333125800516308e-06,
      "loss": 0.0613,
      "step": 6817
    },
    {
      "epoch": 0.3520784921249677,
      "grad_norm": 14.330299377441406,
      "learning_rate": 7.332401081412551e-06,
      "loss": 0.1044,
      "step": 6818
    },
    {
      "epoch": 0.35213013168086754,
      "grad_norm": 3.4473307132720947,
      "learning_rate": 7.331676299674316e-06,
      "loss": 0.0589,
      "step": 6819
    },
    {
      "epoch": 0.35218177123676736,
      "grad_norm": 7.97144079208374,
      "learning_rate": 7.330951455321072e-06,
      "loss": 0.0547,
      "step": 6820
    },
    {
      "epoch": 0.3522334107926672,
      "grad_norm": 15.58449649810791,
      "learning_rate": 7.330226548372279e-06,
      "loss": 0.2069,
      "step": 6821
    },
    {
      "epoch": 0.352285050348567,
      "grad_norm": 3.918320417404175,
      "learning_rate": 7.329501578847407e-06,
      "loss": 0.0623,
      "step": 6822
    },
    {
      "epoch": 0.3523366899044668,
      "grad_norm": 15.538963317871094,
      "learning_rate": 7.328776546765922e-06,
      "loss": 0.1469,
      "step": 6823
    },
    {
      "epoch": 0.35238832946036663,
      "grad_norm": 3.055278778076172,
      "learning_rate": 7.3280514521472955e-06,
      "loss": 0.0656,
      "step": 6824
    },
    {
      "epoch": 0.35243996901626645,
      "grad_norm": 4.289205074310303,
      "learning_rate": 7.327326295010999e-06,
      "loss": 0.0706,
      "step": 6825
    },
    {
      "epoch": 0.35249160857216627,
      "grad_norm": 10.337058067321777,
      "learning_rate": 7.3266010753765045e-06,
      "loss": 0.0813,
      "step": 6826
    },
    {
      "epoch": 0.3525432481280661,
      "grad_norm": 3.205242395401001,
      "learning_rate": 7.325875793263291e-06,
      "loss": 0.0457,
      "step": 6827
    },
    {
      "epoch": 0.3525948876839659,
      "grad_norm": 3.079228639602661,
      "learning_rate": 7.325150448690832e-06,
      "loss": 0.0726,
      "step": 6828
    },
    {
      "epoch": 0.3526465272398657,
      "grad_norm": 3.6728131771087646,
      "learning_rate": 7.324425041678606e-06,
      "loss": 0.0685,
      "step": 6829
    },
    {
      "epoch": 0.35269816679576554,
      "grad_norm": 4.690028667449951,
      "learning_rate": 7.323699572246093e-06,
      "loss": 0.0871,
      "step": 6830
    },
    {
      "epoch": 0.35274980635166536,
      "grad_norm": 3.916332244873047,
      "learning_rate": 7.322974040412777e-06,
      "loss": 0.0488,
      "step": 6831
    },
    {
      "epoch": 0.3528014459075652,
      "grad_norm": 15.90157413482666,
      "learning_rate": 7.32224844619814e-06,
      "loss": 0.1004,
      "step": 6832
    },
    {
      "epoch": 0.352853085463465,
      "grad_norm": 9.344064712524414,
      "learning_rate": 7.321522789621666e-06,
      "loss": 0.0741,
      "step": 6833
    },
    {
      "epoch": 0.3529047250193648,
      "grad_norm": 6.585085868835449,
      "learning_rate": 7.320797070702844e-06,
      "loss": 0.0638,
      "step": 6834
    },
    {
      "epoch": 0.35295636457526464,
      "grad_norm": 7.241869926452637,
      "learning_rate": 7.320071289461161e-06,
      "loss": 0.0986,
      "step": 6835
    },
    {
      "epoch": 0.35300800413116445,
      "grad_norm": 4.483908176422119,
      "learning_rate": 7.3193454459161084e-06,
      "loss": 0.087,
      "step": 6836
    },
    {
      "epoch": 0.3530596436870643,
      "grad_norm": 5.114391803741455,
      "learning_rate": 7.318619540087177e-06,
      "loss": 0.0796,
      "step": 6837
    },
    {
      "epoch": 0.3531112832429641,
      "grad_norm": 3.926215171813965,
      "learning_rate": 7.317893571993861e-06,
      "loss": 0.0271,
      "step": 6838
    },
    {
      "epoch": 0.3531629227988639,
      "grad_norm": 5.356268405914307,
      "learning_rate": 7.317167541655657e-06,
      "loss": 0.0809,
      "step": 6839
    },
    {
      "epoch": 0.35321456235476373,
      "grad_norm": 3.945803165435791,
      "learning_rate": 7.316441449092058e-06,
      "loss": 0.0669,
      "step": 6840
    },
    {
      "epoch": 0.35326620191066355,
      "grad_norm": 3.575112819671631,
      "learning_rate": 7.315715294322565e-06,
      "loss": 0.0362,
      "step": 6841
    },
    {
      "epoch": 0.35331784146656336,
      "grad_norm": 6.075947284698486,
      "learning_rate": 7.314989077366678e-06,
      "loss": 0.0689,
      "step": 6842
    },
    {
      "epoch": 0.3533694810224632,
      "grad_norm": 7.839669704437256,
      "learning_rate": 7.314262798243901e-06,
      "loss": 0.0928,
      "step": 6843
    },
    {
      "epoch": 0.353421120578363,
      "grad_norm": 6.383562088012695,
      "learning_rate": 7.3135364569737335e-06,
      "loss": 0.0654,
      "step": 6844
    },
    {
      "epoch": 0.3534727601342628,
      "grad_norm": 6.227776050567627,
      "learning_rate": 7.312810053575683e-06,
      "loss": 0.0881,
      "step": 6845
    },
    {
      "epoch": 0.35352439969016264,
      "grad_norm": 2.6174118518829346,
      "learning_rate": 7.312083588069257e-06,
      "loss": 0.0435,
      "step": 6846
    },
    {
      "epoch": 0.35357603924606246,
      "grad_norm": 5.347812652587891,
      "learning_rate": 7.311357060473961e-06,
      "loss": 0.0816,
      "step": 6847
    },
    {
      "epoch": 0.35362767880196233,
      "grad_norm": 7.295329570770264,
      "learning_rate": 7.310630470809309e-06,
      "loss": 0.0993,
      "step": 6848
    },
    {
      "epoch": 0.35367931835786215,
      "grad_norm": 4.593784809112549,
      "learning_rate": 7.3099038190948114e-06,
      "loss": 0.0746,
      "step": 6849
    },
    {
      "epoch": 0.35373095791376197,
      "grad_norm": 17.40774154663086,
      "learning_rate": 7.30917710534998e-06,
      "loss": 0.1676,
      "step": 6850
    },
    {
      "epoch": 0.3537825974696618,
      "grad_norm": 3.8673489093780518,
      "learning_rate": 7.308450329594334e-06,
      "loss": 0.0625,
      "step": 6851
    },
    {
      "epoch": 0.3538342370255616,
      "grad_norm": 3.2089436054229736,
      "learning_rate": 7.307723491847386e-06,
      "loss": 0.0413,
      "step": 6852
    },
    {
      "epoch": 0.3538858765814614,
      "grad_norm": 5.939150333404541,
      "learning_rate": 7.3069965921286575e-06,
      "loss": 0.0392,
      "step": 6853
    },
    {
      "epoch": 0.35393751613736124,
      "grad_norm": 5.223714828491211,
      "learning_rate": 7.306269630457667e-06,
      "loss": 0.0699,
      "step": 6854
    },
    {
      "epoch": 0.35398915569326106,
      "grad_norm": 3.5562431812286377,
      "learning_rate": 7.305542606853938e-06,
      "loss": 0.0626,
      "step": 6855
    },
    {
      "epoch": 0.3540407952491609,
      "grad_norm": 7.040595054626465,
      "learning_rate": 7.304815521336992e-06,
      "loss": 0.0786,
      "step": 6856
    },
    {
      "epoch": 0.3540924348050607,
      "grad_norm": 10.47609806060791,
      "learning_rate": 7.304088373926357e-06,
      "loss": 0.1167,
      "step": 6857
    },
    {
      "epoch": 0.3541440743609605,
      "grad_norm": 4.295420169830322,
      "learning_rate": 7.303361164641557e-06,
      "loss": 0.0381,
      "step": 6858
    },
    {
      "epoch": 0.35419571391686033,
      "grad_norm": 17.015274047851562,
      "learning_rate": 7.302633893502121e-06,
      "loss": 0.1351,
      "step": 6859
    },
    {
      "epoch": 0.35424735347276015,
      "grad_norm": 3.8764445781707764,
      "learning_rate": 7.301906560527582e-06,
      "loss": 0.0563,
      "step": 6860
    },
    {
      "epoch": 0.35429899302865997,
      "grad_norm": 1.9395806789398193,
      "learning_rate": 7.301179165737468e-06,
      "loss": 0.0297,
      "step": 6861
    },
    {
      "epoch": 0.3543506325845598,
      "grad_norm": 4.000226020812988,
      "learning_rate": 7.300451709151315e-06,
      "loss": 0.0337,
      "step": 6862
    },
    {
      "epoch": 0.3544022721404596,
      "grad_norm": 7.554118633270264,
      "learning_rate": 7.299724190788658e-06,
      "loss": 0.0934,
      "step": 6863
    },
    {
      "epoch": 0.3544539116963594,
      "grad_norm": 3.2836270332336426,
      "learning_rate": 7.298996610669032e-06,
      "loss": 0.0602,
      "step": 6864
    },
    {
      "epoch": 0.35450555125225924,
      "grad_norm": 2.611286163330078,
      "learning_rate": 7.298268968811979e-06,
      "loss": 0.057,
      "step": 6865
    },
    {
      "epoch": 0.35455719080815906,
      "grad_norm": 8.574597358703613,
      "learning_rate": 7.297541265237034e-06,
      "loss": 0.1016,
      "step": 6866
    },
    {
      "epoch": 0.3546088303640589,
      "grad_norm": 5.490713119506836,
      "learning_rate": 7.296813499963744e-06,
      "loss": 0.0523,
      "step": 6867
    },
    {
      "epoch": 0.3546604699199587,
      "grad_norm": 5.909343242645264,
      "learning_rate": 7.296085673011648e-06,
      "loss": 0.0896,
      "step": 6868
    },
    {
      "epoch": 0.3547121094758585,
      "grad_norm": 4.972075462341309,
      "learning_rate": 7.2953577844002966e-06,
      "loss": 0.0962,
      "step": 6869
    },
    {
      "epoch": 0.35476374903175834,
      "grad_norm": 12.682941436767578,
      "learning_rate": 7.294629834149232e-06,
      "loss": 0.0855,
      "step": 6870
    },
    {
      "epoch": 0.35481538858765815,
      "grad_norm": 8.521936416625977,
      "learning_rate": 7.293901822278002e-06,
      "loss": 0.1431,
      "step": 6871
    },
    {
      "epoch": 0.35486702814355797,
      "grad_norm": 11.753921508789062,
      "learning_rate": 7.29317374880616e-06,
      "loss": 0.149,
      "step": 6872
    },
    {
      "epoch": 0.3549186676994578,
      "grad_norm": 2.7037599086761475,
      "learning_rate": 7.292445613753257e-06,
      "loss": 0.0733,
      "step": 6873
    },
    {
      "epoch": 0.3549703072553576,
      "grad_norm": 11.875494003295898,
      "learning_rate": 7.2917174171388465e-06,
      "loss": 0.1753,
      "step": 6874
    },
    {
      "epoch": 0.3550219468112574,
      "grad_norm": 3.5825233459472656,
      "learning_rate": 7.29098915898248e-06,
      "loss": 0.0397,
      "step": 6875
    },
    {
      "epoch": 0.35507358636715725,
      "grad_norm": 7.260391712188721,
      "learning_rate": 7.29026083930372e-06,
      "loss": 0.0654,
      "step": 6876
    },
    {
      "epoch": 0.35512522592305706,
      "grad_norm": 3.988100051879883,
      "learning_rate": 7.28953245812212e-06,
      "loss": 0.0816,
      "step": 6877
    },
    {
      "epoch": 0.3551768654789569,
      "grad_norm": 2.546776533126831,
      "learning_rate": 7.288804015457242e-06,
      "loss": 0.0404,
      "step": 6878
    },
    {
      "epoch": 0.3552285050348567,
      "grad_norm": 2.0070743560791016,
      "learning_rate": 7.288075511328649e-06,
      "loss": 0.0326,
      "step": 6879
    },
    {
      "epoch": 0.3552801445907565,
      "grad_norm": 4.831051349639893,
      "learning_rate": 7.287346945755902e-06,
      "loss": 0.0681,
      "step": 6880
    },
    {
      "epoch": 0.35533178414665634,
      "grad_norm": 4.149515628814697,
      "learning_rate": 7.286618318758566e-06,
      "loss": 0.0437,
      "step": 6881
    },
    {
      "epoch": 0.35538342370255616,
      "grad_norm": 5.615499496459961,
      "learning_rate": 7.2858896303562085e-06,
      "loss": 0.0675,
      "step": 6882
    },
    {
      "epoch": 0.355435063258456,
      "grad_norm": 6.492940425872803,
      "learning_rate": 7.285160880568399e-06,
      "loss": 0.0735,
      "step": 6883
    },
    {
      "epoch": 0.3554867028143558,
      "grad_norm": 6.195639610290527,
      "learning_rate": 7.284432069414707e-06,
      "loss": 0.0685,
      "step": 6884
    },
    {
      "epoch": 0.3555383423702556,
      "grad_norm": 3.0194737911224365,
      "learning_rate": 7.2837031969146994e-06,
      "loss": 0.0628,
      "step": 6885
    },
    {
      "epoch": 0.35558998192615543,
      "grad_norm": 4.8697686195373535,
      "learning_rate": 7.282974263087957e-06,
      "loss": 0.0498,
      "step": 6886
    },
    {
      "epoch": 0.35564162148205525,
      "grad_norm": 6.097408771514893,
      "learning_rate": 7.2822452679540466e-06,
      "loss": 0.08,
      "step": 6887
    },
    {
      "epoch": 0.35569326103795507,
      "grad_norm": 2.9481780529022217,
      "learning_rate": 7.281516211532551e-06,
      "loss": 0.0927,
      "step": 6888
    },
    {
      "epoch": 0.3557449005938549,
      "grad_norm": 4.49916410446167,
      "learning_rate": 7.280787093843047e-06,
      "loss": 0.0538,
      "step": 6889
    },
    {
      "epoch": 0.3557965401497547,
      "grad_norm": 1.9384945631027222,
      "learning_rate": 7.280057914905112e-06,
      "loss": 0.0337,
      "step": 6890
    },
    {
      "epoch": 0.3558481797056545,
      "grad_norm": 5.20751953125,
      "learning_rate": 7.279328674738331e-06,
      "loss": 0.08,
      "step": 6891
    },
    {
      "epoch": 0.35589981926155434,
      "grad_norm": 9.726935386657715,
      "learning_rate": 7.278599373362283e-06,
      "loss": 0.1395,
      "step": 6892
    },
    {
      "epoch": 0.35595145881745416,
      "grad_norm": 3.151015520095825,
      "learning_rate": 7.277870010796556e-06,
      "loss": 0.0364,
      "step": 6893
    },
    {
      "epoch": 0.356003098373354,
      "grad_norm": 8.829377174377441,
      "learning_rate": 7.277140587060734e-06,
      "loss": 0.1092,
      "step": 6894
    },
    {
      "epoch": 0.3560547379292538,
      "grad_norm": 13.518594741821289,
      "learning_rate": 7.276411102174408e-06,
      "loss": 0.0821,
      "step": 6895
    },
    {
      "epoch": 0.3561063774851536,
      "grad_norm": 8.893733024597168,
      "learning_rate": 7.275681556157164e-06,
      "loss": 0.0963,
      "step": 6896
    },
    {
      "epoch": 0.35615801704105343,
      "grad_norm": 5.145441055297852,
      "learning_rate": 7.274951949028596e-06,
      "loss": 0.0598,
      "step": 6897
    },
    {
      "epoch": 0.35620965659695325,
      "grad_norm": 4.37180757522583,
      "learning_rate": 7.2742222808082965e-06,
      "loss": 0.0886,
      "step": 6898
    },
    {
      "epoch": 0.35626129615285307,
      "grad_norm": 3.4707226753234863,
      "learning_rate": 7.273492551515857e-06,
      "loss": 0.0527,
      "step": 6899
    },
    {
      "epoch": 0.3563129357087529,
      "grad_norm": 5.177388668060303,
      "learning_rate": 7.2727627611708796e-06,
      "loss": 0.0972,
      "step": 6900
    },
    {
      "epoch": 0.3563645752646527,
      "grad_norm": 4.183944225311279,
      "learning_rate": 7.272032909792956e-06,
      "loss": 0.0677,
      "step": 6901
    },
    {
      "epoch": 0.3564162148205525,
      "grad_norm": 5.745948791503906,
      "learning_rate": 7.27130299740169e-06,
      "loss": 0.0599,
      "step": 6902
    },
    {
      "epoch": 0.35646785437645234,
      "grad_norm": 22.46046257019043,
      "learning_rate": 7.2705730240166805e-06,
      "loss": 0.1355,
      "step": 6903
    },
    {
      "epoch": 0.35651949393235216,
      "grad_norm": 3.446178674697876,
      "learning_rate": 7.269842989657532e-06,
      "loss": 0.0942,
      "step": 6904
    },
    {
      "epoch": 0.356571133488252,
      "grad_norm": 3.731905937194824,
      "learning_rate": 7.269112894343846e-06,
      "loss": 0.0342,
      "step": 6905
    },
    {
      "epoch": 0.3566227730441518,
      "grad_norm": 4.5023193359375,
      "learning_rate": 7.268382738095232e-06,
      "loss": 0.0517,
      "step": 6906
    },
    {
      "epoch": 0.3566744126000516,
      "grad_norm": 6.053398132324219,
      "learning_rate": 7.267652520931296e-06,
      "loss": 0.114,
      "step": 6907
    },
    {
      "epoch": 0.35672605215595143,
      "grad_norm": 1.723327875137329,
      "learning_rate": 7.266922242871646e-06,
      "loss": 0.0385,
      "step": 6908
    },
    {
      "epoch": 0.35677769171185125,
      "grad_norm": 8.392763137817383,
      "learning_rate": 7.266191903935896e-06,
      "loss": 0.1117,
      "step": 6909
    },
    {
      "epoch": 0.35682933126775107,
      "grad_norm": 3.6447012424468994,
      "learning_rate": 7.265461504143656e-06,
      "loss": 0.0673,
      "step": 6910
    },
    {
      "epoch": 0.3568809708236509,
      "grad_norm": 2.369144916534424,
      "learning_rate": 7.264731043514541e-06,
      "loss": 0.0444,
      "step": 6911
    },
    {
      "epoch": 0.35693261037955076,
      "grad_norm": 15.129598617553711,
      "learning_rate": 7.264000522068168e-06,
      "loss": 0.23,
      "step": 6912
    },
    {
      "epoch": 0.3569842499354506,
      "grad_norm": 4.351248264312744,
      "learning_rate": 7.263269939824153e-06,
      "loss": 0.0391,
      "step": 6913
    },
    {
      "epoch": 0.3570358894913504,
      "grad_norm": 7.8616228103637695,
      "learning_rate": 7.2625392968021145e-06,
      "loss": 0.0919,
      "step": 6914
    },
    {
      "epoch": 0.3570875290472502,
      "grad_norm": 6.543180465698242,
      "learning_rate": 7.261808593021675e-06,
      "loss": 0.0358,
      "step": 6915
    },
    {
      "epoch": 0.35713916860315004,
      "grad_norm": 12.619087219238281,
      "learning_rate": 7.261077828502456e-06,
      "loss": 0.0876,
      "step": 6916
    },
    {
      "epoch": 0.35719080815904986,
      "grad_norm": 2.446176528930664,
      "learning_rate": 7.260347003264083e-06,
      "loss": 0.0614,
      "step": 6917
    },
    {
      "epoch": 0.3572424477149497,
      "grad_norm": 6.402580738067627,
      "learning_rate": 7.259616117326178e-06,
      "loss": 0.1003,
      "step": 6918
    },
    {
      "epoch": 0.3572940872708495,
      "grad_norm": 7.537329196929932,
      "learning_rate": 7.258885170708372e-06,
      "loss": 0.0782,
      "step": 6919
    },
    {
      "epoch": 0.3573457268267493,
      "grad_norm": 3.6857194900512695,
      "learning_rate": 7.258154163430292e-06,
      "loss": 0.0633,
      "step": 6920
    },
    {
      "epoch": 0.35739736638264913,
      "grad_norm": 2.818056344985962,
      "learning_rate": 7.257423095511569e-06,
      "loss": 0.067,
      "step": 6921
    },
    {
      "epoch": 0.35744900593854895,
      "grad_norm": 2.187253713607788,
      "learning_rate": 7.256691966971835e-06,
      "loss": 0.027,
      "step": 6922
    },
    {
      "epoch": 0.35750064549444877,
      "grad_norm": 7.978191375732422,
      "learning_rate": 7.255960777830724e-06,
      "loss": 0.0921,
      "step": 6923
    },
    {
      "epoch": 0.3575522850503486,
      "grad_norm": 22.40583610534668,
      "learning_rate": 7.255229528107872e-06,
      "loss": 0.1141,
      "step": 6924
    },
    {
      "epoch": 0.3576039246062484,
      "grad_norm": 3.6737983226776123,
      "learning_rate": 7.254498217822914e-06,
      "loss": 0.0672,
      "step": 6925
    },
    {
      "epoch": 0.3576555641621482,
      "grad_norm": 7.893997669219971,
      "learning_rate": 7.2537668469954925e-06,
      "loss": 0.104,
      "step": 6926
    },
    {
      "epoch": 0.35770720371804804,
      "grad_norm": 4.609719753265381,
      "learning_rate": 7.253035415645242e-06,
      "loss": 0.0731,
      "step": 6927
    },
    {
      "epoch": 0.35775884327394786,
      "grad_norm": 1.311130404472351,
      "learning_rate": 7.2523039237918106e-06,
      "loss": 0.0131,
      "step": 6928
    },
    {
      "epoch": 0.3578104828298477,
      "grad_norm": 2.254099130630493,
      "learning_rate": 7.251572371454838e-06,
      "loss": 0.022,
      "step": 6929
    },
    {
      "epoch": 0.3578621223857475,
      "grad_norm": 8.77037525177002,
      "learning_rate": 7.2508407586539695e-06,
      "loss": 0.1007,
      "step": 6930
    },
    {
      "epoch": 0.3579137619416473,
      "grad_norm": 10.01802921295166,
      "learning_rate": 7.250109085408854e-06,
      "loss": 0.1124,
      "step": 6931
    },
    {
      "epoch": 0.35796540149754713,
      "grad_norm": 12.21018123626709,
      "learning_rate": 7.249377351739137e-06,
      "loss": 0.1409,
      "step": 6932
    },
    {
      "epoch": 0.35801704105344695,
      "grad_norm": 6.828376770019531,
      "learning_rate": 7.248645557664472e-06,
      "loss": 0.0737,
      "step": 6933
    },
    {
      "epoch": 0.35806868060934677,
      "grad_norm": 7.233665943145752,
      "learning_rate": 7.247913703204507e-06,
      "loss": 0.069,
      "step": 6934
    },
    {
      "epoch": 0.3581203201652466,
      "grad_norm": 2.6895031929016113,
      "learning_rate": 7.2471817883788984e-06,
      "loss": 0.0749,
      "step": 6935
    },
    {
      "epoch": 0.3581719597211464,
      "grad_norm": 14.69414234161377,
      "learning_rate": 7.246449813207299e-06,
      "loss": 0.1189,
      "step": 6936
    },
    {
      "epoch": 0.3582235992770462,
      "grad_norm": 4.732700347900391,
      "learning_rate": 7.245717777709367e-06,
      "loss": 0.0513,
      "step": 6937
    },
    {
      "epoch": 0.35827523883294604,
      "grad_norm": 2.0002903938293457,
      "learning_rate": 7.244985681904759e-06,
      "loss": 0.0153,
      "step": 6938
    },
    {
      "epoch": 0.35832687838884586,
      "grad_norm": 5.043787479400635,
      "learning_rate": 7.244253525813136e-06,
      "loss": 0.0759,
      "step": 6939
    },
    {
      "epoch": 0.3583785179447457,
      "grad_norm": 6.784912586212158,
      "learning_rate": 7.243521309454158e-06,
      "loss": 0.0905,
      "step": 6940
    },
    {
      "epoch": 0.3584301575006455,
      "grad_norm": 8.266741752624512,
      "learning_rate": 7.242789032847488e-06,
      "loss": 0.0649,
      "step": 6941
    },
    {
      "epoch": 0.3584817970565453,
      "grad_norm": 3.2939565181732178,
      "learning_rate": 7.242056696012794e-06,
      "loss": 0.0506,
      "step": 6942
    },
    {
      "epoch": 0.35853343661244513,
      "grad_norm": 10.10115909576416,
      "learning_rate": 7.241324298969738e-06,
      "loss": 0.1587,
      "step": 6943
    },
    {
      "epoch": 0.35858507616834495,
      "grad_norm": 2.107221841812134,
      "learning_rate": 7.240591841737989e-06,
      "loss": 0.0289,
      "step": 6944
    },
    {
      "epoch": 0.35863671572424477,
      "grad_norm": 2.321995973587036,
      "learning_rate": 7.239859324337216e-06,
      "loss": 0.0341,
      "step": 6945
    },
    {
      "epoch": 0.3586883552801446,
      "grad_norm": 7.368337631225586,
      "learning_rate": 7.239126746787093e-06,
      "loss": 0.0871,
      "step": 6946
    },
    {
      "epoch": 0.3587399948360444,
      "grad_norm": 3.82796311378479,
      "learning_rate": 7.238394109107289e-06,
      "loss": 0.059,
      "step": 6947
    },
    {
      "epoch": 0.3587916343919442,
      "grad_norm": 6.868191719055176,
      "learning_rate": 7.2376614113174816e-06,
      "loss": 0.0871,
      "step": 6948
    },
    {
      "epoch": 0.35884327394784404,
      "grad_norm": 4.051858425140381,
      "learning_rate": 7.236928653437344e-06,
      "loss": 0.0738,
      "step": 6949
    },
    {
      "epoch": 0.35889491350374386,
      "grad_norm": 34.786617279052734,
      "learning_rate": 7.236195835486556e-06,
      "loss": 0.1669,
      "step": 6950
    },
    {
      "epoch": 0.3589465530596437,
      "grad_norm": 3.629981756210327,
      "learning_rate": 7.235462957484795e-06,
      "loss": 0.0543,
      "step": 6951
    },
    {
      "epoch": 0.3589981926155435,
      "grad_norm": 7.876351833343506,
      "learning_rate": 7.234730019451741e-06,
      "loss": 0.0586,
      "step": 6952
    },
    {
      "epoch": 0.3590498321714433,
      "grad_norm": 3.0522005558013916,
      "learning_rate": 7.23399702140708e-06,
      "loss": 0.0347,
      "step": 6953
    },
    {
      "epoch": 0.35910147172734314,
      "grad_norm": 6.088164806365967,
      "learning_rate": 7.233263963370492e-06,
      "loss": 0.1193,
      "step": 6954
    },
    {
      "epoch": 0.35915311128324295,
      "grad_norm": 7.900387763977051,
      "learning_rate": 7.232530845361665e-06,
      "loss": 0.0711,
      "step": 6955
    },
    {
      "epoch": 0.3592047508391428,
      "grad_norm": 4.482982635498047,
      "learning_rate": 7.231797667400286e-06,
      "loss": 0.0721,
      "step": 6956
    },
    {
      "epoch": 0.3592563903950426,
      "grad_norm": 3.2499380111694336,
      "learning_rate": 7.231064429506043e-06,
      "loss": 0.0658,
      "step": 6957
    },
    {
      "epoch": 0.3593080299509424,
      "grad_norm": 10.96927261352539,
      "learning_rate": 7.230331131698625e-06,
      "loss": 0.1044,
      "step": 6958
    },
    {
      "epoch": 0.3593596695068422,
      "grad_norm": 14.946646690368652,
      "learning_rate": 7.229597773997728e-06,
      "loss": 0.2177,
      "step": 6959
    },
    {
      "epoch": 0.35941130906274205,
      "grad_norm": 2.9130239486694336,
      "learning_rate": 7.228864356423042e-06,
      "loss": 0.0407,
      "step": 6960
    },
    {
      "epoch": 0.35946294861864186,
      "grad_norm": 4.218596935272217,
      "learning_rate": 7.228130878994265e-06,
      "loss": 0.0559,
      "step": 6961
    },
    {
      "epoch": 0.3595145881745417,
      "grad_norm": 5.9694719314575195,
      "learning_rate": 7.227397341731093e-06,
      "loss": 0.0853,
      "step": 6962
    },
    {
      "epoch": 0.3595662277304415,
      "grad_norm": 3.682325601577759,
      "learning_rate": 7.226663744653222e-06,
      "loss": 0.0489,
      "step": 6963
    },
    {
      "epoch": 0.3596178672863413,
      "grad_norm": 8.549039840698242,
      "learning_rate": 7.225930087780356e-06,
      "loss": 0.1333,
      "step": 6964
    },
    {
      "epoch": 0.35966950684224114,
      "grad_norm": 15.94155216217041,
      "learning_rate": 7.2251963711321926e-06,
      "loss": 0.1097,
      "step": 6965
    },
    {
      "epoch": 0.35972114639814096,
      "grad_norm": 1.6936835050582886,
      "learning_rate": 7.224462594728439e-06,
      "loss": 0.027,
      "step": 6966
    },
    {
      "epoch": 0.3597727859540408,
      "grad_norm": 7.342230796813965,
      "learning_rate": 7.2237287585887975e-06,
      "loss": 0.0828,
      "step": 6967
    },
    {
      "epoch": 0.3598244255099406,
      "grad_norm": 5.4363861083984375,
      "learning_rate": 7.222994862732976e-06,
      "loss": 0.067,
      "step": 6968
    },
    {
      "epoch": 0.3598760650658404,
      "grad_norm": 2.9452476501464844,
      "learning_rate": 7.2222609071806815e-06,
      "loss": 0.0574,
      "step": 6969
    },
    {
      "epoch": 0.35992770462174023,
      "grad_norm": 19.12710189819336,
      "learning_rate": 7.221526891951624e-06,
      "loss": 0.1781,
      "step": 6970
    },
    {
      "epoch": 0.35997934417764005,
      "grad_norm": 3.346015691757202,
      "learning_rate": 7.220792817065515e-06,
      "loss": 0.0862,
      "step": 6971
    },
    {
      "epoch": 0.36003098373353987,
      "grad_norm": 2.590665340423584,
      "learning_rate": 7.2200586825420684e-06,
      "loss": 0.048,
      "step": 6972
    },
    {
      "epoch": 0.3600826232894397,
      "grad_norm": 2.7507643699645996,
      "learning_rate": 7.2193244884009985e-06,
      "loss": 0.0268,
      "step": 6973
    },
    {
      "epoch": 0.3601342628453395,
      "grad_norm": 3.510823965072632,
      "learning_rate": 7.218590234662017e-06,
      "loss": 0.0554,
      "step": 6974
    },
    {
      "epoch": 0.3601859024012393,
      "grad_norm": 2.4217844009399414,
      "learning_rate": 7.217855921344849e-06,
      "loss": 0.0539,
      "step": 6975
    },
    {
      "epoch": 0.3602375419571392,
      "grad_norm": 7.8929009437561035,
      "learning_rate": 7.217121548469208e-06,
      "loss": 0.0942,
      "step": 6976
    },
    {
      "epoch": 0.360289181513039,
      "grad_norm": 3.1024186611175537,
      "learning_rate": 7.216387116054817e-06,
      "loss": 0.0564,
      "step": 6977
    },
    {
      "epoch": 0.36034082106893883,
      "grad_norm": 3.35343861579895,
      "learning_rate": 7.215652624121399e-06,
      "loss": 0.0574,
      "step": 6978
    },
    {
      "epoch": 0.36039246062483865,
      "grad_norm": 3.8143763542175293,
      "learning_rate": 7.214918072688677e-06,
      "loss": 0.0328,
      "step": 6979
    },
    {
      "epoch": 0.36044410018073847,
      "grad_norm": 2.106609582901001,
      "learning_rate": 7.214183461776378e-06,
      "loss": 0.0286,
      "step": 6980
    },
    {
      "epoch": 0.3604957397366383,
      "grad_norm": 2.7549211978912354,
      "learning_rate": 7.213448791404227e-06,
      "loss": 0.036,
      "step": 6981
    },
    {
      "epoch": 0.3605473792925381,
      "grad_norm": 2.67411470413208,
      "learning_rate": 7.212714061591956e-06,
      "loss": 0.0503,
      "step": 6982
    },
    {
      "epoch": 0.3605990188484379,
      "grad_norm": 2.949570894241333,
      "learning_rate": 7.211979272359295e-06,
      "loss": 0.0726,
      "step": 6983
    },
    {
      "epoch": 0.36065065840433774,
      "grad_norm": 1.4177545309066772,
      "learning_rate": 7.211244423725974e-06,
      "loss": 0.0342,
      "step": 6984
    },
    {
      "epoch": 0.36070229796023756,
      "grad_norm": 2.558425188064575,
      "learning_rate": 7.210509515711727e-06,
      "loss": 0.0637,
      "step": 6985
    },
    {
      "epoch": 0.3607539375161374,
      "grad_norm": 4.413471698760986,
      "learning_rate": 7.2097745483362894e-06,
      "loss": 0.0919,
      "step": 6986
    },
    {
      "epoch": 0.3608055770720372,
      "grad_norm": 5.108150005340576,
      "learning_rate": 7.209039521619398e-06,
      "loss": 0.0609,
      "step": 6987
    },
    {
      "epoch": 0.360857216627937,
      "grad_norm": 3.218327760696411,
      "learning_rate": 7.2083044355807954e-06,
      "loss": 0.0574,
      "step": 6988
    },
    {
      "epoch": 0.36090885618383683,
      "grad_norm": 7.470418930053711,
      "learning_rate": 7.207569290240215e-06,
      "loss": 0.1159,
      "step": 6989
    },
    {
      "epoch": 0.36096049573973665,
      "grad_norm": 34.16117858886719,
      "learning_rate": 7.206834085617403e-06,
      "loss": 0.1245,
      "step": 6990
    },
    {
      "epoch": 0.36101213529563647,
      "grad_norm": 4.426228046417236,
      "learning_rate": 7.2060988217321e-06,
      "loss": 0.0679,
      "step": 6991
    },
    {
      "epoch": 0.3610637748515363,
      "grad_norm": 16.5479793548584,
      "learning_rate": 7.205363498604055e-06,
      "loss": 0.0432,
      "step": 6992
    },
    {
      "epoch": 0.3611154144074361,
      "grad_norm": 5.1734700202941895,
      "learning_rate": 7.204628116253007e-06,
      "loss": 0.0554,
      "step": 6993
    },
    {
      "epoch": 0.3611670539633359,
      "grad_norm": 19.16635513305664,
      "learning_rate": 7.2038926746987125e-06,
      "loss": 0.2023,
      "step": 6994
    },
    {
      "epoch": 0.36121869351923575,
      "grad_norm": 11.872953414916992,
      "learning_rate": 7.203157173960915e-06,
      "loss": 0.1358,
      "step": 6995
    },
    {
      "epoch": 0.36127033307513556,
      "grad_norm": 4.508788108825684,
      "learning_rate": 7.202421614059369e-06,
      "loss": 0.0556,
      "step": 6996
    },
    {
      "epoch": 0.3613219726310354,
      "grad_norm": 2.4329118728637695,
      "learning_rate": 7.201685995013826e-06,
      "loss": 0.0375,
      "step": 6997
    },
    {
      "epoch": 0.3613736121869352,
      "grad_norm": 12.64612865447998,
      "learning_rate": 7.2009503168440395e-06,
      "loss": 0.1046,
      "step": 6998
    },
    {
      "epoch": 0.361425251742835,
      "grad_norm": 3.9830739498138428,
      "learning_rate": 7.2002145795697675e-06,
      "loss": 0.0454,
      "step": 6999
    },
    {
      "epoch": 0.36147689129873484,
      "grad_norm": 5.6327128410339355,
      "learning_rate": 7.199478783210765e-06,
      "loss": 0.0683,
      "step": 7000
    },
    {
      "epoch": 0.36152853085463466,
      "grad_norm": 3.380302906036377,
      "learning_rate": 7.198742927786795e-06,
      "loss": 0.0471,
      "step": 7001
    },
    {
      "epoch": 0.3615801704105345,
      "grad_norm": 1.9176589250564575,
      "learning_rate": 7.198007013317614e-06,
      "loss": 0.0281,
      "step": 7002
    },
    {
      "epoch": 0.3616318099664343,
      "grad_norm": 8.466470718383789,
      "learning_rate": 7.197271039822986e-06,
      "loss": 0.0736,
      "step": 7003
    },
    {
      "epoch": 0.3616834495223341,
      "grad_norm": 6.145705223083496,
      "learning_rate": 7.196535007322675e-06,
      "loss": 0.0515,
      "step": 7004
    },
    {
      "epoch": 0.36173508907823393,
      "grad_norm": 7.450737476348877,
      "learning_rate": 7.195798915836447e-06,
      "loss": 0.0703,
      "step": 7005
    },
    {
      "epoch": 0.36178672863413375,
      "grad_norm": 22.244230270385742,
      "learning_rate": 7.195062765384069e-06,
      "loss": 0.0848,
      "step": 7006
    },
    {
      "epoch": 0.36183836819003357,
      "grad_norm": 8.676069259643555,
      "learning_rate": 7.1943265559853074e-06,
      "loss": 0.045,
      "step": 7007
    },
    {
      "epoch": 0.3618900077459334,
      "grad_norm": 2.8599636554718018,
      "learning_rate": 7.193590287659936e-06,
      "loss": 0.0717,
      "step": 7008
    },
    {
      "epoch": 0.3619416473018332,
      "grad_norm": 2.4456112384796143,
      "learning_rate": 7.192853960427724e-06,
      "loss": 0.0274,
      "step": 7009
    },
    {
      "epoch": 0.361993286857733,
      "grad_norm": 6.433627605438232,
      "learning_rate": 7.192117574308446e-06,
      "loss": 0.0693,
      "step": 7010
    },
    {
      "epoch": 0.36204492641363284,
      "grad_norm": 2.0633199214935303,
      "learning_rate": 7.191381129321876e-06,
      "loss": 0.028,
      "step": 7011
    },
    {
      "epoch": 0.36209656596953266,
      "grad_norm": 9.21800422668457,
      "learning_rate": 7.190644625487791e-06,
      "loss": 0.1303,
      "step": 7012
    },
    {
      "epoch": 0.3621482055254325,
      "grad_norm": 4.117283344268799,
      "learning_rate": 7.18990806282597e-06,
      "loss": 0.0578,
      "step": 7013
    },
    {
      "epoch": 0.3621998450813323,
      "grad_norm": 2.082700729370117,
      "learning_rate": 7.189171441356191e-06,
      "loss": 0.0367,
      "step": 7014
    },
    {
      "epoch": 0.3622514846372321,
      "grad_norm": 3.6514358520507812,
      "learning_rate": 7.188434761098236e-06,
      "loss": 0.0403,
      "step": 7015
    },
    {
      "epoch": 0.36230312419313193,
      "grad_norm": 3.7829246520996094,
      "learning_rate": 7.187698022071891e-06,
      "loss": 0.0437,
      "step": 7016
    },
    {
      "epoch": 0.36235476374903175,
      "grad_norm": 2.016279935836792,
      "learning_rate": 7.186961224296934e-06,
      "loss": 0.042,
      "step": 7017
    },
    {
      "epoch": 0.36240640330493157,
      "grad_norm": 3.8379721641540527,
      "learning_rate": 7.1862243677931565e-06,
      "loss": 0.0355,
      "step": 7018
    },
    {
      "epoch": 0.3624580428608314,
      "grad_norm": 3.63779878616333,
      "learning_rate": 7.1854874525803445e-06,
      "loss": 0.078,
      "step": 7019
    },
    {
      "epoch": 0.3625096824167312,
      "grad_norm": 3.975633382797241,
      "learning_rate": 7.184750478678285e-06,
      "loss": 0.0627,
      "step": 7020
    },
    {
      "epoch": 0.362561321972631,
      "grad_norm": 3.607642412185669,
      "learning_rate": 7.184013446106771e-06,
      "loss": 0.036,
      "step": 7021
    },
    {
      "epoch": 0.36261296152853084,
      "grad_norm": 13.072554588317871,
      "learning_rate": 7.183276354885596e-06,
      "loss": 0.1705,
      "step": 7022
    },
    {
      "epoch": 0.36266460108443066,
      "grad_norm": 1.8046274185180664,
      "learning_rate": 7.182539205034551e-06,
      "loss": 0.0387,
      "step": 7023
    },
    {
      "epoch": 0.3627162406403305,
      "grad_norm": 1.9705841541290283,
      "learning_rate": 7.181801996573432e-06,
      "loss": 0.0189,
      "step": 7024
    },
    {
      "epoch": 0.3627678801962303,
      "grad_norm": 12.113779067993164,
      "learning_rate": 7.181064729522039e-06,
      "loss": 0.0864,
      "step": 7025
    },
    {
      "epoch": 0.3628195197521301,
      "grad_norm": 3.6537423133850098,
      "learning_rate": 7.1803274039001666e-06,
      "loss": 0.0506,
      "step": 7026
    },
    {
      "epoch": 0.36287115930802993,
      "grad_norm": 13.137726783752441,
      "learning_rate": 7.179590019727619e-06,
      "loss": 0.0671,
      "step": 7027
    },
    {
      "epoch": 0.36292279886392975,
      "grad_norm": 5.938558578491211,
      "learning_rate": 7.178852577024195e-06,
      "loss": 0.0773,
      "step": 7028
    },
    {
      "epoch": 0.36297443841982957,
      "grad_norm": 6.3447723388671875,
      "learning_rate": 7.178115075809697e-06,
      "loss": 0.1158,
      "step": 7029
    },
    {
      "epoch": 0.3630260779757294,
      "grad_norm": 10.678462982177734,
      "learning_rate": 7.177377516103935e-06,
      "loss": 0.1365,
      "step": 7030
    },
    {
      "epoch": 0.3630777175316292,
      "grad_norm": 4.193188190460205,
      "learning_rate": 7.176639897926709e-06,
      "loss": 0.036,
      "step": 7031
    },
    {
      "epoch": 0.363129357087529,
      "grad_norm": 2.318502902984619,
      "learning_rate": 7.1759022212978326e-06,
      "loss": 0.035,
      "step": 7032
    },
    {
      "epoch": 0.36318099664342884,
      "grad_norm": 4.574371814727783,
      "learning_rate": 7.17516448623711e-06,
      "loss": 0.0391,
      "step": 7033
    },
    {
      "epoch": 0.36323263619932866,
      "grad_norm": 5.509261608123779,
      "learning_rate": 7.174426692764359e-06,
      "loss": 0.0655,
      "step": 7034
    },
    {
      "epoch": 0.3632842757552285,
      "grad_norm": 2.6297504901885986,
      "learning_rate": 7.173688840899388e-06,
      "loss": 0.0441,
      "step": 7035
    },
    {
      "epoch": 0.3633359153111283,
      "grad_norm": 5.668145179748535,
      "learning_rate": 7.172950930662011e-06,
      "loss": 0.0568,
      "step": 7036
    },
    {
      "epoch": 0.3633875548670281,
      "grad_norm": 5.335456848144531,
      "learning_rate": 7.172212962072046e-06,
      "loss": 0.0388,
      "step": 7037
    },
    {
      "epoch": 0.36343919442292794,
      "grad_norm": 2.6280264854431152,
      "learning_rate": 7.171474935149308e-06,
      "loss": 0.0408,
      "step": 7038
    },
    {
      "epoch": 0.3634908339788278,
      "grad_norm": 3.93019962310791,
      "learning_rate": 7.17073684991362e-06,
      "loss": 0.0339,
      "step": 7039
    },
    {
      "epoch": 0.36354247353472763,
      "grad_norm": 22.38318634033203,
      "learning_rate": 7.169998706384798e-06,
      "loss": 0.1012,
      "step": 7040
    },
    {
      "epoch": 0.36359411309062745,
      "grad_norm": 2.6801536083221436,
      "learning_rate": 7.169260504582667e-06,
      "loss": 0.0263,
      "step": 7041
    },
    {
      "epoch": 0.36364575264652726,
      "grad_norm": 9.010812759399414,
      "learning_rate": 7.16852224452705e-06,
      "loss": 0.0878,
      "step": 7042
    },
    {
      "epoch": 0.3636973922024271,
      "grad_norm": 4.4644951820373535,
      "learning_rate": 7.167783926237772e-06,
      "loss": 0.0784,
      "step": 7043
    },
    {
      "epoch": 0.3637490317583269,
      "grad_norm": 28.5306396484375,
      "learning_rate": 7.1670455497346614e-06,
      "loss": 0.1274,
      "step": 7044
    },
    {
      "epoch": 0.3638006713142267,
      "grad_norm": 4.942347526550293,
      "learning_rate": 7.166307115037544e-06,
      "loss": 0.0732,
      "step": 7045
    },
    {
      "epoch": 0.36385231087012654,
      "grad_norm": 4.523856163024902,
      "learning_rate": 7.165568622166253e-06,
      "loss": 0.0803,
      "step": 7046
    },
    {
      "epoch": 0.36390395042602636,
      "grad_norm": 3.2416253089904785,
      "learning_rate": 7.164830071140616e-06,
      "loss": 0.0429,
      "step": 7047
    },
    {
      "epoch": 0.3639555899819262,
      "grad_norm": 3.323402166366577,
      "learning_rate": 7.1640914619804694e-06,
      "loss": 0.0728,
      "step": 7048
    },
    {
      "epoch": 0.364007229537826,
      "grad_norm": 2.9267396926879883,
      "learning_rate": 7.163352794705647e-06,
      "loss": 0.0606,
      "step": 7049
    },
    {
      "epoch": 0.3640588690937258,
      "grad_norm": 2.937221050262451,
      "learning_rate": 7.162614069335983e-06,
      "loss": 0.063,
      "step": 7050
    },
    {
      "epoch": 0.36411050864962563,
      "grad_norm": 5.264703273773193,
      "learning_rate": 7.161875285891318e-06,
      "loss": 0.0952,
      "step": 7051
    },
    {
      "epoch": 0.36416214820552545,
      "grad_norm": 2.8369224071502686,
      "learning_rate": 7.161136444391491e-06,
      "loss": 0.0562,
      "step": 7052
    },
    {
      "epoch": 0.36421378776142527,
      "grad_norm": 3.0542080402374268,
      "learning_rate": 7.1603975448563415e-06,
      "loss": 0.0398,
      "step": 7053
    },
    {
      "epoch": 0.3642654273173251,
      "grad_norm": 4.43837308883667,
      "learning_rate": 7.159658587305712e-06,
      "loss": 0.0723,
      "step": 7054
    },
    {
      "epoch": 0.3643170668732249,
      "grad_norm": 6.48321533203125,
      "learning_rate": 7.158919571759448e-06,
      "loss": 0.0782,
      "step": 7055
    },
    {
      "epoch": 0.3643687064291247,
      "grad_norm": 10.452583312988281,
      "learning_rate": 7.158180498237395e-06,
      "loss": 0.0861,
      "step": 7056
    },
    {
      "epoch": 0.36442034598502454,
      "grad_norm": 5.38425350189209,
      "learning_rate": 7.157441366759397e-06,
      "loss": 0.0506,
      "step": 7057
    },
    {
      "epoch": 0.36447198554092436,
      "grad_norm": 1.6853370666503906,
      "learning_rate": 7.156702177345307e-06,
      "loss": 0.0394,
      "step": 7058
    },
    {
      "epoch": 0.3645236250968242,
      "grad_norm": 12.507524490356445,
      "learning_rate": 7.155962930014973e-06,
      "loss": 0.1414,
      "step": 7059
    },
    {
      "epoch": 0.364575264652724,
      "grad_norm": 4.261792182922363,
      "learning_rate": 7.155223624788247e-06,
      "loss": 0.0482,
      "step": 7060
    },
    {
      "epoch": 0.3646269042086238,
      "grad_norm": 5.040379047393799,
      "learning_rate": 7.154484261684982e-06,
      "loss": 0.0596,
      "step": 7061
    },
    {
      "epoch": 0.36467854376452363,
      "grad_norm": 3.4594569206237793,
      "learning_rate": 7.153744840725033e-06,
      "loss": 0.0677,
      "step": 7062
    },
    {
      "epoch": 0.36473018332042345,
      "grad_norm": 2.4039080142974854,
      "learning_rate": 7.153005361928258e-06,
      "loss": 0.0447,
      "step": 7063
    },
    {
      "epoch": 0.36478182287632327,
      "grad_norm": 1.369223713874817,
      "learning_rate": 7.152265825314513e-06,
      "loss": 0.0155,
      "step": 7064
    },
    {
      "epoch": 0.3648334624322231,
      "grad_norm": 3.500153064727783,
      "learning_rate": 7.15152623090366e-06,
      "loss": 0.0691,
      "step": 7065
    },
    {
      "epoch": 0.3648851019881229,
      "grad_norm": 3.605531930923462,
      "learning_rate": 7.150786578715556e-06,
      "loss": 0.0605,
      "step": 7066
    },
    {
      "epoch": 0.3649367415440227,
      "grad_norm": 6.411986351013184,
      "learning_rate": 7.1500468687700685e-06,
      "loss": 0.0429,
      "step": 7067
    },
    {
      "epoch": 0.36498838109992254,
      "grad_norm": 2.148454427719116,
      "learning_rate": 7.149307101087059e-06,
      "loss": 0.0376,
      "step": 7068
    },
    {
      "epoch": 0.36504002065582236,
      "grad_norm": 4.428370475769043,
      "learning_rate": 7.148567275686394e-06,
      "loss": 0.0313,
      "step": 7069
    },
    {
      "epoch": 0.3650916602117222,
      "grad_norm": 5.878023624420166,
      "learning_rate": 7.147827392587939e-06,
      "loss": 0.062,
      "step": 7070
    },
    {
      "epoch": 0.365143299767622,
      "grad_norm": 32.72817611694336,
      "learning_rate": 7.147087451811565e-06,
      "loss": 0.1914,
      "step": 7071
    },
    {
      "epoch": 0.3651949393235218,
      "grad_norm": 5.191134929656982,
      "learning_rate": 7.146347453377144e-06,
      "loss": 0.1127,
      "step": 7072
    },
    {
      "epoch": 0.36524657887942164,
      "grad_norm": 3.499952793121338,
      "learning_rate": 7.1456073973045425e-06,
      "loss": 0.0406,
      "step": 7073
    },
    {
      "epoch": 0.36529821843532145,
      "grad_norm": 3.340949535369873,
      "learning_rate": 7.14486728361364e-06,
      "loss": 0.0585,
      "step": 7074
    },
    {
      "epoch": 0.36534985799122127,
      "grad_norm": 16.900615692138672,
      "learning_rate": 7.144127112324307e-06,
      "loss": 0.113,
      "step": 7075
    },
    {
      "epoch": 0.3654014975471211,
      "grad_norm": 2.6938962936401367,
      "learning_rate": 7.143386883456424e-06,
      "loss": 0.0404,
      "step": 7076
    },
    {
      "epoch": 0.3654531371030209,
      "grad_norm": 3.5141563415527344,
      "learning_rate": 7.142646597029865e-06,
      "loss": 0.0438,
      "step": 7077
    },
    {
      "epoch": 0.3655047766589207,
      "grad_norm": 3.42097544670105,
      "learning_rate": 7.141906253064513e-06,
      "loss": 0.0495,
      "step": 7078
    },
    {
      "epoch": 0.36555641621482055,
      "grad_norm": 3.477457284927368,
      "learning_rate": 7.141165851580248e-06,
      "loss": 0.0472,
      "step": 7079
    },
    {
      "epoch": 0.36560805577072036,
      "grad_norm": 4.865344524383545,
      "learning_rate": 7.140425392596953e-06,
      "loss": 0.039,
      "step": 7080
    },
    {
      "epoch": 0.3656596953266202,
      "grad_norm": 7.640966892242432,
      "learning_rate": 7.139684876134514e-06,
      "loss": 0.0449,
      "step": 7081
    },
    {
      "epoch": 0.36571133488252,
      "grad_norm": 2.509735345840454,
      "learning_rate": 7.138944302212813e-06,
      "loss": 0.0373,
      "step": 7082
    },
    {
      "epoch": 0.3657629744384198,
      "grad_norm": 2.975234270095825,
      "learning_rate": 7.138203670851741e-06,
      "loss": 0.0224,
      "step": 7083
    },
    {
      "epoch": 0.36581461399431964,
      "grad_norm": 5.819434642791748,
      "learning_rate": 7.137462982071185e-06,
      "loss": 0.0962,
      "step": 7084
    },
    {
      "epoch": 0.36586625355021946,
      "grad_norm": 6.84705114364624,
      "learning_rate": 7.1367222358910355e-06,
      "loss": 0.0757,
      "step": 7085
    },
    {
      "epoch": 0.3659178931061193,
      "grad_norm": 7.3330864906311035,
      "learning_rate": 7.135981432331187e-06,
      "loss": 0.1012,
      "step": 7086
    },
    {
      "epoch": 0.3659695326620191,
      "grad_norm": 8.175727844238281,
      "learning_rate": 7.135240571411529e-06,
      "loss": 0.1224,
      "step": 7087
    },
    {
      "epoch": 0.3660211722179189,
      "grad_norm": 4.705692768096924,
      "learning_rate": 7.134499653151961e-06,
      "loss": 0.0688,
      "step": 7088
    },
    {
      "epoch": 0.36607281177381873,
      "grad_norm": 4.97346305847168,
      "learning_rate": 7.133758677572376e-06,
      "loss": 0.0543,
      "step": 7089
    },
    {
      "epoch": 0.36612445132971855,
      "grad_norm": 6.978935241699219,
      "learning_rate": 7.133017644692675e-06,
      "loss": 0.0648,
      "step": 7090
    },
    {
      "epoch": 0.36617609088561837,
      "grad_norm": 3.7614028453826904,
      "learning_rate": 7.1322765545327555e-06,
      "loss": 0.061,
      "step": 7091
    },
    {
      "epoch": 0.3662277304415182,
      "grad_norm": 2.5944788455963135,
      "learning_rate": 7.1315354071125195e-06,
      "loss": 0.0317,
      "step": 7092
    },
    {
      "epoch": 0.366279369997418,
      "grad_norm": 10.402219772338867,
      "learning_rate": 7.130794202451872e-06,
      "loss": 0.0816,
      "step": 7093
    },
    {
      "epoch": 0.3663310095533178,
      "grad_norm": 3.5475285053253174,
      "learning_rate": 7.130052940570714e-06,
      "loss": 0.0465,
      "step": 7094
    },
    {
      "epoch": 0.36638264910921764,
      "grad_norm": 10.00616455078125,
      "learning_rate": 7.129311621488953e-06,
      "loss": 0.0475,
      "step": 7095
    },
    {
      "epoch": 0.36643428866511746,
      "grad_norm": 15.044109344482422,
      "learning_rate": 7.128570245226497e-06,
      "loss": 0.1414,
      "step": 7096
    },
    {
      "epoch": 0.3664859282210173,
      "grad_norm": 25.31337547302246,
      "learning_rate": 7.127828811803253e-06,
      "loss": 0.2054,
      "step": 7097
    },
    {
      "epoch": 0.3665375677769171,
      "grad_norm": 6.481548309326172,
      "learning_rate": 7.127087321239136e-06,
      "loss": 0.0723,
      "step": 7098
    },
    {
      "epoch": 0.3665892073328169,
      "grad_norm": 4.977145195007324,
      "learning_rate": 7.126345773554051e-06,
      "loss": 0.067,
      "step": 7099
    },
    {
      "epoch": 0.36664084688871673,
      "grad_norm": 2.9655497074127197,
      "learning_rate": 7.125604168767919e-06,
      "loss": 0.0577,
      "step": 7100
    },
    {
      "epoch": 0.36669248644461655,
      "grad_norm": 1.910290241241455,
      "learning_rate": 7.12486250690065e-06,
      "loss": 0.0367,
      "step": 7101
    },
    {
      "epoch": 0.36674412600051637,
      "grad_norm": 12.610261917114258,
      "learning_rate": 7.124120787972161e-06,
      "loss": 0.0712,
      "step": 7102
    },
    {
      "epoch": 0.36679576555641624,
      "grad_norm": 11.89153003692627,
      "learning_rate": 7.1233790120023725e-06,
      "loss": 0.0597,
      "step": 7103
    },
    {
      "epoch": 0.36684740511231606,
      "grad_norm": 2.2541139125823975,
      "learning_rate": 7.1226371790112025e-06,
      "loss": 0.0266,
      "step": 7104
    },
    {
      "epoch": 0.3668990446682159,
      "grad_norm": 1.6811872720718384,
      "learning_rate": 7.121895289018574e-06,
      "loss": 0.0323,
      "step": 7105
    },
    {
      "epoch": 0.3669506842241157,
      "grad_norm": 2.4379358291625977,
      "learning_rate": 7.121153342044408e-06,
      "loss": 0.0487,
      "step": 7106
    },
    {
      "epoch": 0.3670023237800155,
      "grad_norm": 5.176153659820557,
      "learning_rate": 7.12041133810863e-06,
      "loss": 0.1049,
      "step": 7107
    },
    {
      "epoch": 0.36705396333591533,
      "grad_norm": 7.838663578033447,
      "learning_rate": 7.119669277231164e-06,
      "loss": 0.0819,
      "step": 7108
    },
    {
      "epoch": 0.36710560289181515,
      "grad_norm": 12.021346092224121,
      "learning_rate": 7.11892715943194e-06,
      "loss": 0.1265,
      "step": 7109
    },
    {
      "epoch": 0.36715724244771497,
      "grad_norm": 2.4932117462158203,
      "learning_rate": 7.118184984730885e-06,
      "loss": 0.0494,
      "step": 7110
    },
    {
      "epoch": 0.3672088820036148,
      "grad_norm": 5.976612091064453,
      "learning_rate": 7.1174427531479294e-06,
      "loss": 0.0739,
      "step": 7111
    },
    {
      "epoch": 0.3672605215595146,
      "grad_norm": 9.13451862335205,
      "learning_rate": 7.116700464703006e-06,
      "loss": 0.0784,
      "step": 7112
    },
    {
      "epoch": 0.3673121611154144,
      "grad_norm": 6.670460224151611,
      "learning_rate": 7.115958119416048e-06,
      "loss": 0.1109,
      "step": 7113
    },
    {
      "epoch": 0.36736380067131424,
      "grad_norm": 3.0940895080566406,
      "learning_rate": 7.115215717306991e-06,
      "loss": 0.0697,
      "step": 7114
    },
    {
      "epoch": 0.36741544022721406,
      "grad_norm": 2.3919968605041504,
      "learning_rate": 7.114473258395769e-06,
      "loss": 0.0386,
      "step": 7115
    },
    {
      "epoch": 0.3674670797831139,
      "grad_norm": 3.181422710418701,
      "learning_rate": 7.113730742702324e-06,
      "loss": 0.0354,
      "step": 7116
    },
    {
      "epoch": 0.3675187193390137,
      "grad_norm": 7.657106399536133,
      "learning_rate": 7.112988170246592e-06,
      "loss": 0.0822,
      "step": 7117
    },
    {
      "epoch": 0.3675703588949135,
      "grad_norm": 5.225619792938232,
      "learning_rate": 7.112245541048516e-06,
      "loss": 0.0568,
      "step": 7118
    },
    {
      "epoch": 0.36762199845081334,
      "grad_norm": 3.436042070388794,
      "learning_rate": 7.111502855128039e-06,
      "loss": 0.039,
      "step": 7119
    },
    {
      "epoch": 0.36767363800671315,
      "grad_norm": 5.93548059463501,
      "learning_rate": 7.110760112505103e-06,
      "loss": 0.0683,
      "step": 7120
    },
    {
      "epoch": 0.367725277562613,
      "grad_norm": 8.479704856872559,
      "learning_rate": 7.110017313199655e-06,
      "loss": 0.0879,
      "step": 7121
    },
    {
      "epoch": 0.3677769171185128,
      "grad_norm": 4.42561149597168,
      "learning_rate": 7.109274457231643e-06,
      "loss": 0.0742,
      "step": 7122
    },
    {
      "epoch": 0.3678285566744126,
      "grad_norm": 4.7595062255859375,
      "learning_rate": 7.108531544621015e-06,
      "loss": 0.0658,
      "step": 7123
    },
    {
      "epoch": 0.36788019623031243,
      "grad_norm": 2.613365888595581,
      "learning_rate": 7.107788575387722e-06,
      "loss": 0.0451,
      "step": 7124
    },
    {
      "epoch": 0.36793183578621225,
      "grad_norm": 3.720869302749634,
      "learning_rate": 7.107045549551712e-06,
      "loss": 0.0512,
      "step": 7125
    },
    {
      "epoch": 0.36798347534211207,
      "grad_norm": 5.108208656311035,
      "learning_rate": 7.106302467132944e-06,
      "loss": 0.0752,
      "step": 7126
    },
    {
      "epoch": 0.3680351148980119,
      "grad_norm": 1.7598090171813965,
      "learning_rate": 7.105559328151368e-06,
      "loss": 0.0293,
      "step": 7127
    },
    {
      "epoch": 0.3680867544539117,
      "grad_norm": 2.4610795974731445,
      "learning_rate": 7.104816132626943e-06,
      "loss": 0.0368,
      "step": 7128
    },
    {
      "epoch": 0.3681383940098115,
      "grad_norm": 4.012953758239746,
      "learning_rate": 7.1040728805796265e-06,
      "loss": 0.0821,
      "step": 7129
    },
    {
      "epoch": 0.36819003356571134,
      "grad_norm": 8.062206268310547,
      "learning_rate": 7.103329572029376e-06,
      "loss": 0.0835,
      "step": 7130
    },
    {
      "epoch": 0.36824167312161116,
      "grad_norm": 2.968303680419922,
      "learning_rate": 7.102586206996156e-06,
      "loss": 0.0531,
      "step": 7131
    },
    {
      "epoch": 0.368293312677511,
      "grad_norm": 7.8908162117004395,
      "learning_rate": 7.101842785499924e-06,
      "loss": 0.0853,
      "step": 7132
    },
    {
      "epoch": 0.3683449522334108,
      "grad_norm": 4.53231143951416,
      "learning_rate": 7.101099307560649e-06,
      "loss": 0.0436,
      "step": 7133
    },
    {
      "epoch": 0.3683965917893106,
      "grad_norm": 6.852601051330566,
      "learning_rate": 7.100355773198293e-06,
      "loss": 0.1033,
      "step": 7134
    },
    {
      "epoch": 0.36844823134521043,
      "grad_norm": 7.629488468170166,
      "learning_rate": 7.099612182432825e-06,
      "loss": 0.1252,
      "step": 7135
    },
    {
      "epoch": 0.36849987090111025,
      "grad_norm": 6.855085849761963,
      "learning_rate": 7.0988685352842115e-06,
      "loss": 0.12,
      "step": 7136
    },
    {
      "epoch": 0.36855151045701007,
      "grad_norm": 11.729083061218262,
      "learning_rate": 7.098124831772422e-06,
      "loss": 0.111,
      "step": 7137
    },
    {
      "epoch": 0.3686031500129099,
      "grad_norm": 17.202104568481445,
      "learning_rate": 7.097381071917432e-06,
      "loss": 0.138,
      "step": 7138
    },
    {
      "epoch": 0.3686547895688097,
      "grad_norm": 6.958043098449707,
      "learning_rate": 7.096637255739208e-06,
      "loss": 0.0613,
      "step": 7139
    },
    {
      "epoch": 0.3687064291247095,
      "grad_norm": 2.829737663269043,
      "learning_rate": 7.095893383257731e-06,
      "loss": 0.0363,
      "step": 7140
    },
    {
      "epoch": 0.36875806868060934,
      "grad_norm": 7.471146583557129,
      "learning_rate": 7.095149454492973e-06,
      "loss": 0.0785,
      "step": 7141
    },
    {
      "epoch": 0.36880970823650916,
      "grad_norm": 7.5821332931518555,
      "learning_rate": 7.094405469464916e-06,
      "loss": 0.0945,
      "step": 7142
    },
    {
      "epoch": 0.368861347792409,
      "grad_norm": 7.2295074462890625,
      "learning_rate": 7.093661428193532e-06,
      "loss": 0.0686,
      "step": 7143
    },
    {
      "epoch": 0.3689129873483088,
      "grad_norm": 4.471750736236572,
      "learning_rate": 7.092917330698807e-06,
      "loss": 0.0284,
      "step": 7144
    },
    {
      "epoch": 0.3689646269042086,
      "grad_norm": 14.192285537719727,
      "learning_rate": 7.092173177000721e-06,
      "loss": 0.0498,
      "step": 7145
    },
    {
      "epoch": 0.36901626646010843,
      "grad_norm": 2.357722520828247,
      "learning_rate": 7.091428967119258e-06,
      "loss": 0.0573,
      "step": 7146
    },
    {
      "epoch": 0.36906790601600825,
      "grad_norm": 7.228517055511475,
      "learning_rate": 7.090684701074404e-06,
      "loss": 0.0604,
      "step": 7147
    },
    {
      "epoch": 0.36911954557190807,
      "grad_norm": 4.69283390045166,
      "learning_rate": 7.089940378886142e-06,
      "loss": 0.1,
      "step": 7148
    },
    {
      "epoch": 0.3691711851278079,
      "grad_norm": 2.7421908378601074,
      "learning_rate": 7.089196000574465e-06,
      "loss": 0.0511,
      "step": 7149
    },
    {
      "epoch": 0.3692228246837077,
      "grad_norm": 7.130580425262451,
      "learning_rate": 7.08845156615936e-06,
      "loss": 0.0666,
      "step": 7150
    },
    {
      "epoch": 0.3692744642396075,
      "grad_norm": 4.530081748962402,
      "learning_rate": 7.087707075660818e-06,
      "loss": 0.0788,
      "step": 7151
    },
    {
      "epoch": 0.36932610379550734,
      "grad_norm": 2.256479501724243,
      "learning_rate": 7.086962529098832e-06,
      "loss": 0.0501,
      "step": 7152
    },
    {
      "epoch": 0.36937774335140716,
      "grad_norm": 2.8444128036499023,
      "learning_rate": 7.086217926493396e-06,
      "loss": 0.0564,
      "step": 7153
    },
    {
      "epoch": 0.369429382907307,
      "grad_norm": 2.5646777153015137,
      "learning_rate": 7.085473267864506e-06,
      "loss": 0.0655,
      "step": 7154
    },
    {
      "epoch": 0.3694810224632068,
      "grad_norm": 4.371525764465332,
      "learning_rate": 7.084728553232159e-06,
      "loss": 0.0723,
      "step": 7155
    },
    {
      "epoch": 0.3695326620191066,
      "grad_norm": 4.14456033706665,
      "learning_rate": 7.083983782616353e-06,
      "loss": 0.0513,
      "step": 7156
    },
    {
      "epoch": 0.36958430157500644,
      "grad_norm": 11.479211807250977,
      "learning_rate": 7.0832389560370895e-06,
      "loss": 0.1186,
      "step": 7157
    },
    {
      "epoch": 0.36963594113090625,
      "grad_norm": 4.097333908081055,
      "learning_rate": 7.082494073514369e-06,
      "loss": 0.0711,
      "step": 7158
    },
    {
      "epoch": 0.36968758068680607,
      "grad_norm": 2.518860340118408,
      "learning_rate": 7.081749135068194e-06,
      "loss": 0.0395,
      "step": 7159
    },
    {
      "epoch": 0.3697392202427059,
      "grad_norm": 6.6206488609313965,
      "learning_rate": 7.08100414071857e-06,
      "loss": 0.0852,
      "step": 7160
    },
    {
      "epoch": 0.3697908597986057,
      "grad_norm": 4.665274620056152,
      "learning_rate": 7.080259090485504e-06,
      "loss": 0.0695,
      "step": 7161
    },
    {
      "epoch": 0.3698424993545055,
      "grad_norm": 2.2018651962280273,
      "learning_rate": 7.079513984389001e-06,
      "loss": 0.0441,
      "step": 7162
    },
    {
      "epoch": 0.36989413891040535,
      "grad_norm": 5.920347690582275,
      "learning_rate": 7.078768822449074e-06,
      "loss": 0.0483,
      "step": 7163
    },
    {
      "epoch": 0.36994577846630516,
      "grad_norm": 3.2187657356262207,
      "learning_rate": 7.078023604685731e-06,
      "loss": 0.0479,
      "step": 7164
    },
    {
      "epoch": 0.369997418022205,
      "grad_norm": 2.447523593902588,
      "learning_rate": 7.077278331118984e-06,
      "loss": 0.0479,
      "step": 7165
    },
    {
      "epoch": 0.3700490575781048,
      "grad_norm": 2.279669761657715,
      "learning_rate": 7.076533001768848e-06,
      "loss": 0.0429,
      "step": 7166
    },
    {
      "epoch": 0.3701006971340047,
      "grad_norm": 21.26581573486328,
      "learning_rate": 7.075787616655336e-06,
      "loss": 0.0957,
      "step": 7167
    },
    {
      "epoch": 0.3701523366899045,
      "grad_norm": 8.164435386657715,
      "learning_rate": 7.075042175798468e-06,
      "loss": 0.0468,
      "step": 7168
    },
    {
      "epoch": 0.3702039762458043,
      "grad_norm": 5.365111827850342,
      "learning_rate": 7.07429667921826e-06,
      "loss": 0.0517,
      "step": 7169
    },
    {
      "epoch": 0.37025561580170413,
      "grad_norm": 2.6171090602874756,
      "learning_rate": 7.07355112693473e-06,
      "loss": 0.0571,
      "step": 7170
    },
    {
      "epoch": 0.37030725535760395,
      "grad_norm": 10.679443359375,
      "learning_rate": 7.072805518967904e-06,
      "loss": 0.1595,
      "step": 7171
    },
    {
      "epoch": 0.37035889491350377,
      "grad_norm": 6.720848560333252,
      "learning_rate": 7.072059855337797e-06,
      "loss": 0.0728,
      "step": 7172
    },
    {
      "epoch": 0.3704105344694036,
      "grad_norm": 4.332729339599609,
      "learning_rate": 7.071314136064442e-06,
      "loss": 0.0747,
      "step": 7173
    },
    {
      "epoch": 0.3704621740253034,
      "grad_norm": 4.311502933502197,
      "learning_rate": 7.070568361167857e-06,
      "loss": 0.0733,
      "step": 7174
    },
    {
      "epoch": 0.3705138135812032,
      "grad_norm": 1.2941595315933228,
      "learning_rate": 7.069822530668074e-06,
      "loss": 0.0296,
      "step": 7175
    },
    {
      "epoch": 0.37056545313710304,
      "grad_norm": 3.7859959602355957,
      "learning_rate": 7.06907664458512e-06,
      "loss": 0.0553,
      "step": 7176
    },
    {
      "epoch": 0.37061709269300286,
      "grad_norm": 5.066312313079834,
      "learning_rate": 7.0683307029390245e-06,
      "loss": 0.0819,
      "step": 7177
    },
    {
      "epoch": 0.3706687322489027,
      "grad_norm": 9.742552757263184,
      "learning_rate": 7.06758470574982e-06,
      "loss": 0.0963,
      "step": 7178
    },
    {
      "epoch": 0.3707203718048025,
      "grad_norm": 7.688744068145752,
      "learning_rate": 7.066838653037537e-06,
      "loss": 0.0994,
      "step": 7179
    },
    {
      "epoch": 0.3707720113607023,
      "grad_norm": 5.517281532287598,
      "learning_rate": 7.066092544822216e-06,
      "loss": 0.0709,
      "step": 7180
    },
    {
      "epoch": 0.37082365091660213,
      "grad_norm": 4.874201774597168,
      "learning_rate": 7.0653463811238866e-06,
      "loss": 0.0569,
      "step": 7181
    },
    {
      "epoch": 0.37087529047250195,
      "grad_norm": 4.0976691246032715,
      "learning_rate": 7.06460016196259e-06,
      "loss": 0.0713,
      "step": 7182
    },
    {
      "epoch": 0.37092693002840177,
      "grad_norm": 6.0706305503845215,
      "learning_rate": 7.063853887358364e-06,
      "loss": 0.0887,
      "step": 7183
    },
    {
      "epoch": 0.3709785695843016,
      "grad_norm": 13.586441040039062,
      "learning_rate": 7.063107557331249e-06,
      "loss": 0.1305,
      "step": 7184
    },
    {
      "epoch": 0.3710302091402014,
      "grad_norm": 15.25317096710205,
      "learning_rate": 7.062361171901287e-06,
      "loss": 0.1113,
      "step": 7185
    },
    {
      "epoch": 0.3710818486961012,
      "grad_norm": 9.431985855102539,
      "learning_rate": 7.061614731088523e-06,
      "loss": 0.1037,
      "step": 7186
    },
    {
      "epoch": 0.37113348825200104,
      "grad_norm": 11.283533096313477,
      "learning_rate": 7.060868234913e-06,
      "loss": 0.1911,
      "step": 7187
    },
    {
      "epoch": 0.37118512780790086,
      "grad_norm": 4.955937385559082,
      "learning_rate": 7.060121683394765e-06,
      "loss": 0.1432,
      "step": 7188
    },
    {
      "epoch": 0.3712367673638007,
      "grad_norm": 5.746227741241455,
      "learning_rate": 7.059375076553868e-06,
      "loss": 0.0336,
      "step": 7189
    },
    {
      "epoch": 0.3712884069197005,
      "grad_norm": 10.520020484924316,
      "learning_rate": 7.058628414410356e-06,
      "loss": 0.0689,
      "step": 7190
    },
    {
      "epoch": 0.3713400464756003,
      "grad_norm": 3.686001777648926,
      "learning_rate": 7.057881696984279e-06,
      "loss": 0.0524,
      "step": 7191
    },
    {
      "epoch": 0.37139168603150013,
      "grad_norm": 8.6550931930542,
      "learning_rate": 7.057134924295691e-06,
      "loss": 0.0696,
      "step": 7192
    },
    {
      "epoch": 0.37144332558739995,
      "grad_norm": 2.1477882862091064,
      "learning_rate": 7.056388096364647e-06,
      "loss": 0.0484,
      "step": 7193
    },
    {
      "epoch": 0.37149496514329977,
      "grad_norm": 2.9490671157836914,
      "learning_rate": 7.055641213211201e-06,
      "loss": 0.0641,
      "step": 7194
    },
    {
      "epoch": 0.3715466046991996,
      "grad_norm": 2.744724988937378,
      "learning_rate": 7.0548942748554094e-06,
      "loss": 0.0393,
      "step": 7195
    },
    {
      "epoch": 0.3715982442550994,
      "grad_norm": 3.1156749725341797,
      "learning_rate": 7.054147281317331e-06,
      "loss": 0.0543,
      "step": 7196
    },
    {
      "epoch": 0.3716498838109992,
      "grad_norm": 4.696217060089111,
      "learning_rate": 7.053400232617028e-06,
      "loss": 0.058,
      "step": 7197
    },
    {
      "epoch": 0.37170152336689904,
      "grad_norm": 2.5298237800598145,
      "learning_rate": 7.052653128774558e-06,
      "loss": 0.0425,
      "step": 7198
    },
    {
      "epoch": 0.37175316292279886,
      "grad_norm": 4.240698337554932,
      "learning_rate": 7.051905969809986e-06,
      "loss": 0.0767,
      "step": 7199
    },
    {
      "epoch": 0.3718048024786987,
      "grad_norm": 2.6463775634765625,
      "learning_rate": 7.051158755743375e-06,
      "loss": 0.02,
      "step": 7200
    },
    {
      "epoch": 0.3718564420345985,
      "grad_norm": 7.94753360748291,
      "learning_rate": 7.0504114865947914e-06,
      "loss": 0.0858,
      "step": 7201
    },
    {
      "epoch": 0.3719080815904983,
      "grad_norm": 3.794172525405884,
      "learning_rate": 7.049664162384302e-06,
      "loss": 0.0698,
      "step": 7202
    },
    {
      "epoch": 0.37195972114639814,
      "grad_norm": 2.2168657779693604,
      "learning_rate": 7.048916783131977e-06,
      "loss": 0.0399,
      "step": 7203
    },
    {
      "epoch": 0.37201136070229796,
      "grad_norm": 2.351621389389038,
      "learning_rate": 7.048169348857885e-06,
      "loss": 0.0471,
      "step": 7204
    },
    {
      "epoch": 0.3720630002581978,
      "grad_norm": 6.35506010055542,
      "learning_rate": 7.047421859582097e-06,
      "loss": 0.0677,
      "step": 7205
    },
    {
      "epoch": 0.3721146398140976,
      "grad_norm": 5.207657814025879,
      "learning_rate": 7.04667431532469e-06,
      "loss": 0.0583,
      "step": 7206
    },
    {
      "epoch": 0.3721662793699974,
      "grad_norm": 23.133115768432617,
      "learning_rate": 7.045926716105732e-06,
      "loss": 0.1657,
      "step": 7207
    },
    {
      "epoch": 0.37221791892589723,
      "grad_norm": 4.935194492340088,
      "learning_rate": 7.045179061945307e-06,
      "loss": 0.0528,
      "step": 7208
    },
    {
      "epoch": 0.37226955848179705,
      "grad_norm": 4.419788837432861,
      "learning_rate": 7.044431352863486e-06,
      "loss": 0.0657,
      "step": 7209
    },
    {
      "epoch": 0.37232119803769687,
      "grad_norm": 4.38740873336792,
      "learning_rate": 7.04368358888035e-06,
      "loss": 0.0743,
      "step": 7210
    },
    {
      "epoch": 0.3723728375935967,
      "grad_norm": 8.208558082580566,
      "learning_rate": 7.0429357700159826e-06,
      "loss": 0.1285,
      "step": 7211
    },
    {
      "epoch": 0.3724244771494965,
      "grad_norm": 15.915034294128418,
      "learning_rate": 7.04218789629046e-06,
      "loss": 0.1976,
      "step": 7212
    },
    {
      "epoch": 0.3724761167053963,
      "grad_norm": 4.142772197723389,
      "learning_rate": 7.041439967723872e-06,
      "loss": 0.0791,
      "step": 7213
    },
    {
      "epoch": 0.37252775626129614,
      "grad_norm": 18.048418045043945,
      "learning_rate": 7.040691984336297e-06,
      "loss": 0.0703,
      "step": 7214
    },
    {
      "epoch": 0.37257939581719596,
      "grad_norm": 4.638028144836426,
      "learning_rate": 7.039943946147828e-06,
      "loss": 0.0542,
      "step": 7215
    },
    {
      "epoch": 0.3726310353730958,
      "grad_norm": 4.731310844421387,
      "learning_rate": 7.039195853178549e-06,
      "loss": 0.0712,
      "step": 7216
    },
    {
      "epoch": 0.3726826749289956,
      "grad_norm": 3.0738277435302734,
      "learning_rate": 7.038447705448549e-06,
      "loss": 0.0545,
      "step": 7217
    },
    {
      "epoch": 0.3727343144848954,
      "grad_norm": 1.9848277568817139,
      "learning_rate": 7.03769950297792e-06,
      "loss": 0.0313,
      "step": 7218
    },
    {
      "epoch": 0.37278595404079523,
      "grad_norm": 11.871448516845703,
      "learning_rate": 7.036951245786755e-06,
      "loss": 0.0809,
      "step": 7219
    },
    {
      "epoch": 0.37283759359669505,
      "grad_norm": 3.249361276626587,
      "learning_rate": 7.036202933895146e-06,
      "loss": 0.0606,
      "step": 7220
    },
    {
      "epoch": 0.37288923315259487,
      "grad_norm": 10.91308307647705,
      "learning_rate": 7.03545456732319e-06,
      "loss": 0.1053,
      "step": 7221
    },
    {
      "epoch": 0.3729408727084947,
      "grad_norm": 3.896932601928711,
      "learning_rate": 7.034706146090983e-06,
      "loss": 0.0586,
      "step": 7222
    },
    {
      "epoch": 0.3729925122643945,
      "grad_norm": 9.243929862976074,
      "learning_rate": 7.033957670218622e-06,
      "loss": 0.1568,
      "step": 7223
    },
    {
      "epoch": 0.3730441518202943,
      "grad_norm": 4.453684329986572,
      "learning_rate": 7.033209139726208e-06,
      "loss": 0.072,
      "step": 7224
    },
    {
      "epoch": 0.37309579137619414,
      "grad_norm": 8.606400489807129,
      "learning_rate": 7.032460554633842e-06,
      "loss": 0.1301,
      "step": 7225
    },
    {
      "epoch": 0.37314743093209396,
      "grad_norm": 4.076364994049072,
      "learning_rate": 7.031711914961625e-06,
      "loss": 0.0429,
      "step": 7226
    },
    {
      "epoch": 0.3731990704879938,
      "grad_norm": 16.179004669189453,
      "learning_rate": 7.030963220729663e-06,
      "loss": 0.1443,
      "step": 7227
    },
    {
      "epoch": 0.3732507100438936,
      "grad_norm": 7.307550430297852,
      "learning_rate": 7.030214471958061e-06,
      "loss": 0.0852,
      "step": 7228
    },
    {
      "epoch": 0.3733023495997934,
      "grad_norm": 3.7389426231384277,
      "learning_rate": 7.029465668666926e-06,
      "loss": 0.0544,
      "step": 7229
    },
    {
      "epoch": 0.37335398915569323,
      "grad_norm": 4.541865825653076,
      "learning_rate": 7.028716810876366e-06,
      "loss": 0.075,
      "step": 7230
    },
    {
      "epoch": 0.3734056287115931,
      "grad_norm": 3.213852882385254,
      "learning_rate": 7.0279678986064905e-06,
      "loss": 0.0689,
      "step": 7231
    },
    {
      "epoch": 0.3734572682674929,
      "grad_norm": 7.40645694732666,
      "learning_rate": 7.027218931877412e-06,
      "loss": 0.0594,
      "step": 7232
    },
    {
      "epoch": 0.37350890782339274,
      "grad_norm": 1.3704578876495361,
      "learning_rate": 7.026469910709243e-06,
      "loss": 0.0411,
      "step": 7233
    },
    {
      "epoch": 0.37356054737929256,
      "grad_norm": 5.719512462615967,
      "learning_rate": 7.025720835122096e-06,
      "loss": 0.0847,
      "step": 7234
    },
    {
      "epoch": 0.3736121869351924,
      "grad_norm": 3.370656728744507,
      "learning_rate": 7.02497170513609e-06,
      "loss": 0.0587,
      "step": 7235
    },
    {
      "epoch": 0.3736638264910922,
      "grad_norm": 8.052159309387207,
      "learning_rate": 7.024222520771339e-06,
      "loss": 0.0945,
      "step": 7236
    },
    {
      "epoch": 0.373715466046992,
      "grad_norm": 2.7423624992370605,
      "learning_rate": 7.023473282047964e-06,
      "loss": 0.0347,
      "step": 7237
    },
    {
      "epoch": 0.37376710560289184,
      "grad_norm": 2.539919137954712,
      "learning_rate": 7.0227239889860835e-06,
      "loss": 0.0467,
      "step": 7238
    },
    {
      "epoch": 0.37381874515879165,
      "grad_norm": 2.025331497192383,
      "learning_rate": 7.02197464160582e-06,
      "loss": 0.0255,
      "step": 7239
    },
    {
      "epoch": 0.3738703847146915,
      "grad_norm": 2.484661340713501,
      "learning_rate": 7.021225239927295e-06,
      "loss": 0.0649,
      "step": 7240
    },
    {
      "epoch": 0.3739220242705913,
      "grad_norm": 12.631424903869629,
      "learning_rate": 7.020475783970637e-06,
      "loss": 0.1617,
      "step": 7241
    },
    {
      "epoch": 0.3739736638264911,
      "grad_norm": 3.297011137008667,
      "learning_rate": 7.0197262737559665e-06,
      "loss": 0.0735,
      "step": 7242
    },
    {
      "epoch": 0.37402530338239093,
      "grad_norm": 2.8255043029785156,
      "learning_rate": 7.018976709303416e-06,
      "loss": 0.0671,
      "step": 7243
    },
    {
      "epoch": 0.37407694293829075,
      "grad_norm": 2.564958095550537,
      "learning_rate": 7.01822709063311e-06,
      "loss": 0.0368,
      "step": 7244
    },
    {
      "epoch": 0.37412858249419056,
      "grad_norm": 3.320751667022705,
      "learning_rate": 7.0174774177651815e-06,
      "loss": 0.0533,
      "step": 7245
    },
    {
      "epoch": 0.3741802220500904,
      "grad_norm": 6.450133800506592,
      "learning_rate": 7.016727690719762e-06,
      "loss": 0.044,
      "step": 7246
    },
    {
      "epoch": 0.3742318616059902,
      "grad_norm": 6.111512184143066,
      "learning_rate": 7.015977909516982e-06,
      "loss": 0.0895,
      "step": 7247
    },
    {
      "epoch": 0.37428350116189,
      "grad_norm": 8.55594253540039,
      "learning_rate": 7.015228074176982e-06,
      "loss": 0.0663,
      "step": 7248
    },
    {
      "epoch": 0.37433514071778984,
      "grad_norm": 5.26528263092041,
      "learning_rate": 7.0144781847198924e-06,
      "loss": 0.0236,
      "step": 7249
    },
    {
      "epoch": 0.37438678027368966,
      "grad_norm": 2.439366579055786,
      "learning_rate": 7.013728241165854e-06,
      "loss": 0.0477,
      "step": 7250
    },
    {
      "epoch": 0.3744384198295895,
      "grad_norm": 2.361640691757202,
      "learning_rate": 7.012978243535003e-06,
      "loss": 0.0537,
      "step": 7251
    },
    {
      "epoch": 0.3744900593854893,
      "grad_norm": 1.4567054510116577,
      "learning_rate": 7.012228191847483e-06,
      "loss": 0.0424,
      "step": 7252
    },
    {
      "epoch": 0.3745416989413891,
      "grad_norm": 4.010318279266357,
      "learning_rate": 7.011478086123436e-06,
      "loss": 0.0442,
      "step": 7253
    },
    {
      "epoch": 0.37459333849728893,
      "grad_norm": 3.4350011348724365,
      "learning_rate": 7.0107279263830016e-06,
      "loss": 0.0776,
      "step": 7254
    },
    {
      "epoch": 0.37464497805318875,
      "grad_norm": 4.5132670402526855,
      "learning_rate": 7.009977712646329e-06,
      "loss": 0.0421,
      "step": 7255
    },
    {
      "epoch": 0.37469661760908857,
      "grad_norm": 3.845883369445801,
      "learning_rate": 7.009227444933561e-06,
      "loss": 0.0654,
      "step": 7256
    },
    {
      "epoch": 0.3747482571649884,
      "grad_norm": 4.75294828414917,
      "learning_rate": 7.008477123264849e-06,
      "loss": 0.0803,
      "step": 7257
    },
    {
      "epoch": 0.3747998967208882,
      "grad_norm": 28.68767738342285,
      "learning_rate": 7.007726747660338e-06,
      "loss": 0.1119,
      "step": 7258
    },
    {
      "epoch": 0.374851536276788,
      "grad_norm": 7.35662317276001,
      "learning_rate": 7.006976318140182e-06,
      "loss": 0.0597,
      "step": 7259
    },
    {
      "epoch": 0.37490317583268784,
      "grad_norm": 5.548770904541016,
      "learning_rate": 7.006225834724532e-06,
      "loss": 0.0373,
      "step": 7260
    },
    {
      "epoch": 0.37495481538858766,
      "grad_norm": 3.686448812484741,
      "learning_rate": 7.005475297433539e-06,
      "loss": 0.0638,
      "step": 7261
    },
    {
      "epoch": 0.3750064549444875,
      "grad_norm": 2.305507183074951,
      "learning_rate": 7.004724706287362e-06,
      "loss": 0.0206,
      "step": 7262
    },
    {
      "epoch": 0.3750580945003873,
      "grad_norm": 10.425378799438477,
      "learning_rate": 7.003974061306158e-06,
      "loss": 0.0876,
      "step": 7263
    },
    {
      "epoch": 0.3751097340562871,
      "grad_norm": 2.9822652339935303,
      "learning_rate": 7.00322336251008e-06,
      "loss": 0.0411,
      "step": 7264
    },
    {
      "epoch": 0.37516137361218693,
      "grad_norm": 7.4334940910339355,
      "learning_rate": 7.002472609919292e-06,
      "loss": 0.0551,
      "step": 7265
    },
    {
      "epoch": 0.37521301316808675,
      "grad_norm": 9.409704208374023,
      "learning_rate": 7.001721803553952e-06,
      "loss": 0.136,
      "step": 7266
    },
    {
      "epoch": 0.37526465272398657,
      "grad_norm": 3.3019206523895264,
      "learning_rate": 7.000970943434222e-06,
      "loss": 0.0386,
      "step": 7267
    },
    {
      "epoch": 0.3753162922798864,
      "grad_norm": 5.082183837890625,
      "learning_rate": 7.000220029580267e-06,
      "loss": 0.0685,
      "step": 7268
    },
    {
      "epoch": 0.3753679318357862,
      "grad_norm": 4.692168712615967,
      "learning_rate": 6.999469062012252e-06,
      "loss": 0.0532,
      "step": 7269
    },
    {
      "epoch": 0.375419571391686,
      "grad_norm": 2.305610179901123,
      "learning_rate": 6.998718040750346e-06,
      "loss": 0.0396,
      "step": 7270
    },
    {
      "epoch": 0.37547121094758584,
      "grad_norm": 7.579598426818848,
      "learning_rate": 6.9979669658147086e-06,
      "loss": 0.1224,
      "step": 7271
    },
    {
      "epoch": 0.37552285050348566,
      "grad_norm": 4.87555456161499,
      "learning_rate": 6.997215837225519e-06,
      "loss": 0.0828,
      "step": 7272
    },
    {
      "epoch": 0.3755744900593855,
      "grad_norm": 3.2797179222106934,
      "learning_rate": 6.996464655002943e-06,
      "loss": 0.0694,
      "step": 7273
    },
    {
      "epoch": 0.3756261296152853,
      "grad_norm": 4.260691165924072,
      "learning_rate": 6.995713419167154e-06,
      "loss": 0.0513,
      "step": 7274
    },
    {
      "epoch": 0.3756777691711851,
      "grad_norm": 4.997096061706543,
      "learning_rate": 6.994962129738325e-06,
      "loss": 0.0498,
      "step": 7275
    },
    {
      "epoch": 0.37572940872708493,
      "grad_norm": 14.891002655029297,
      "learning_rate": 6.994210786736633e-06,
      "loss": 0.112,
      "step": 7276
    },
    {
      "epoch": 0.37578104828298475,
      "grad_norm": 3.467146873474121,
      "learning_rate": 6.993459390182253e-06,
      "loss": 0.0481,
      "step": 7277
    },
    {
      "epoch": 0.37583268783888457,
      "grad_norm": 7.492284774780273,
      "learning_rate": 6.992707940095363e-06,
      "loss": 0.0436,
      "step": 7278
    },
    {
      "epoch": 0.3758843273947844,
      "grad_norm": 6.4944047927856445,
      "learning_rate": 6.991956436496144e-06,
      "loss": 0.077,
      "step": 7279
    },
    {
      "epoch": 0.3759359669506842,
      "grad_norm": 2.5213091373443604,
      "learning_rate": 6.991204879404774e-06,
      "loss": 0.0561,
      "step": 7280
    },
    {
      "epoch": 0.375987606506584,
      "grad_norm": 4.753364562988281,
      "learning_rate": 6.990453268841438e-06,
      "loss": 0.0471,
      "step": 7281
    },
    {
      "epoch": 0.37603924606248385,
      "grad_norm": 4.001603603363037,
      "learning_rate": 6.98970160482632e-06,
      "loss": 0.0558,
      "step": 7282
    },
    {
      "epoch": 0.37609088561838366,
      "grad_norm": 3.934965133666992,
      "learning_rate": 6.988949887379603e-06,
      "loss": 0.0392,
      "step": 7283
    },
    {
      "epoch": 0.3761425251742835,
      "grad_norm": 16.429832458496094,
      "learning_rate": 6.9881981165214765e-06,
      "loss": 0.1942,
      "step": 7284
    },
    {
      "epoch": 0.3761941647301833,
      "grad_norm": 2.4045310020446777,
      "learning_rate": 6.987446292272126e-06,
      "loss": 0.0555,
      "step": 7285
    },
    {
      "epoch": 0.3762458042860831,
      "grad_norm": 2.0055201053619385,
      "learning_rate": 6.9866944146517425e-06,
      "loss": 0.0393,
      "step": 7286
    },
    {
      "epoch": 0.37629744384198294,
      "grad_norm": 4.727054119110107,
      "learning_rate": 6.985942483680517e-06,
      "loss": 0.0508,
      "step": 7287
    },
    {
      "epoch": 0.37634908339788276,
      "grad_norm": 30.850276947021484,
      "learning_rate": 6.985190499378642e-06,
      "loss": 0.215,
      "step": 7288
    },
    {
      "epoch": 0.3764007229537826,
      "grad_norm": 4.966448783874512,
      "learning_rate": 6.98443846176631e-06,
      "loss": 0.0459,
      "step": 7289
    },
    {
      "epoch": 0.3764523625096824,
      "grad_norm": 2.7846357822418213,
      "learning_rate": 6.983686370863717e-06,
      "loss": 0.035,
      "step": 7290
    },
    {
      "epoch": 0.3765040020655822,
      "grad_norm": 2.296496868133545,
      "learning_rate": 6.98293422669106e-06,
      "loss": 0.0289,
      "step": 7291
    },
    {
      "epoch": 0.37655564162148203,
      "grad_norm": 1.3780903816223145,
      "learning_rate": 6.982182029268538e-06,
      "loss": 0.0103,
      "step": 7292
    },
    {
      "epoch": 0.37660728117738185,
      "grad_norm": 18.794099807739258,
      "learning_rate": 6.981429778616349e-06,
      "loss": 0.1507,
      "step": 7293
    },
    {
      "epoch": 0.3766589207332817,
      "grad_norm": 1.8931341171264648,
      "learning_rate": 6.980677474754694e-06,
      "loss": 0.0173,
      "step": 7294
    },
    {
      "epoch": 0.37671056028918154,
      "grad_norm": 1.964771032333374,
      "learning_rate": 6.979925117703776e-06,
      "loss": 0.0366,
      "step": 7295
    },
    {
      "epoch": 0.37676219984508136,
      "grad_norm": 5.4084930419921875,
      "learning_rate": 6.979172707483801e-06,
      "loss": 0.0705,
      "step": 7296
    },
    {
      "epoch": 0.3768138394009812,
      "grad_norm": 1.8741137981414795,
      "learning_rate": 6.978420244114969e-06,
      "loss": 0.0323,
      "step": 7297
    },
    {
      "epoch": 0.376865478956881,
      "grad_norm": 3.192857503890991,
      "learning_rate": 6.977667727617492e-06,
      "loss": 0.0436,
      "step": 7298
    },
    {
      "epoch": 0.3769171185127808,
      "grad_norm": 5.182126522064209,
      "learning_rate": 6.976915158011575e-06,
      "loss": 0.0781,
      "step": 7299
    },
    {
      "epoch": 0.37696875806868063,
      "grad_norm": 2.764791488647461,
      "learning_rate": 6.9761625353174285e-06,
      "loss": 0.0386,
      "step": 7300
    },
    {
      "epoch": 0.37702039762458045,
      "grad_norm": 7.069623947143555,
      "learning_rate": 6.9754098595552644e-06,
      "loss": 0.1303,
      "step": 7301
    },
    {
      "epoch": 0.37707203718048027,
      "grad_norm": 11.648918151855469,
      "learning_rate": 6.974657130745294e-06,
      "loss": 0.1037,
      "step": 7302
    },
    {
      "epoch": 0.3771236767363801,
      "grad_norm": 5.875183582305908,
      "learning_rate": 6.973904348907732e-06,
      "loss": 0.0723,
      "step": 7303
    },
    {
      "epoch": 0.3771753162922799,
      "grad_norm": 7.120603084564209,
      "learning_rate": 6.97315151406279e-06,
      "loss": 0.0488,
      "step": 7304
    },
    {
      "epoch": 0.3772269558481797,
      "grad_norm": 6.435311794281006,
      "learning_rate": 6.972398626230692e-06,
      "loss": 0.095,
      "step": 7305
    },
    {
      "epoch": 0.37727859540407954,
      "grad_norm": 8.149201393127441,
      "learning_rate": 6.971645685431649e-06,
      "loss": 0.0861,
      "step": 7306
    },
    {
      "epoch": 0.37733023495997936,
      "grad_norm": 4.893059253692627,
      "learning_rate": 6.970892691685884e-06,
      "loss": 0.0727,
      "step": 7307
    },
    {
      "epoch": 0.3773818745158792,
      "grad_norm": 7.6797099113464355,
      "learning_rate": 6.970139645013617e-06,
      "loss": 0.0595,
      "step": 7308
    },
    {
      "epoch": 0.377433514071779,
      "grad_norm": 4.20020055770874,
      "learning_rate": 6.96938654543507e-06,
      "loss": 0.0549,
      "step": 7309
    },
    {
      "epoch": 0.3774851536276788,
      "grad_norm": 17.542083740234375,
      "learning_rate": 6.96863339297047e-06,
      "loss": 0.1851,
      "step": 7310
    },
    {
      "epoch": 0.37753679318357863,
      "grad_norm": 5.867044925689697,
      "learning_rate": 6.9678801876400385e-06,
      "loss": 0.0482,
      "step": 7311
    },
    {
      "epoch": 0.37758843273947845,
      "grad_norm": 3.832447052001953,
      "learning_rate": 6.967126929464003e-06,
      "loss": 0.0567,
      "step": 7312
    },
    {
      "epoch": 0.37764007229537827,
      "grad_norm": 14.776345252990723,
      "learning_rate": 6.9663736184625915e-06,
      "loss": 0.1908,
      "step": 7313
    },
    {
      "epoch": 0.3776917118512781,
      "grad_norm": 3.890591859817505,
      "learning_rate": 6.9656202546560345e-06,
      "loss": 0.0783,
      "step": 7314
    },
    {
      "epoch": 0.3777433514071779,
      "grad_norm": 12.406164169311523,
      "learning_rate": 6.964866838064562e-06,
      "loss": 0.1061,
      "step": 7315
    },
    {
      "epoch": 0.3777949909630777,
      "grad_norm": 2.490525722503662,
      "learning_rate": 6.964113368708407e-06,
      "loss": 0.0617,
      "step": 7316
    },
    {
      "epoch": 0.37784663051897754,
      "grad_norm": 15.750693321228027,
      "learning_rate": 6.963359846607802e-06,
      "loss": 0.113,
      "step": 7317
    },
    {
      "epoch": 0.37789827007487736,
      "grad_norm": 3.2634615898132324,
      "learning_rate": 6.962606271782983e-06,
      "loss": 0.0702,
      "step": 7318
    },
    {
      "epoch": 0.3779499096307772,
      "grad_norm": 6.748265743255615,
      "learning_rate": 6.961852644254187e-06,
      "loss": 0.0709,
      "step": 7319
    },
    {
      "epoch": 0.378001549186677,
      "grad_norm": 4.597617149353027,
      "learning_rate": 6.9610989640416505e-06,
      "loss": 0.0817,
      "step": 7320
    },
    {
      "epoch": 0.3780531887425768,
      "grad_norm": 5.693789958953857,
      "learning_rate": 6.9603452311656165e-06,
      "loss": 0.081,
      "step": 7321
    },
    {
      "epoch": 0.37810482829847664,
      "grad_norm": 4.922207832336426,
      "learning_rate": 6.959591445646321e-06,
      "loss": 0.0615,
      "step": 7322
    },
    {
      "epoch": 0.37815646785437645,
      "grad_norm": 2.9608304500579834,
      "learning_rate": 6.958837607504007e-06,
      "loss": 0.0547,
      "step": 7323
    },
    {
      "epoch": 0.3782081074102763,
      "grad_norm": 15.901973724365234,
      "learning_rate": 6.958083716758921e-06,
      "loss": 0.0798,
      "step": 7324
    },
    {
      "epoch": 0.3782597469661761,
      "grad_norm": 7.255044937133789,
      "learning_rate": 6.957329773431307e-06,
      "loss": 0.0935,
      "step": 7325
    },
    {
      "epoch": 0.3783113865220759,
      "grad_norm": 3.334702968597412,
      "learning_rate": 6.95657577754141e-06,
      "loss": 0.0709,
      "step": 7326
    },
    {
      "epoch": 0.37836302607797573,
      "grad_norm": 4.255613327026367,
      "learning_rate": 6.955821729109479e-06,
      "loss": 0.055,
      "step": 7327
    },
    {
      "epoch": 0.37841466563387555,
      "grad_norm": 3.4713003635406494,
      "learning_rate": 6.955067628155763e-06,
      "loss": 0.04,
      "step": 7328
    },
    {
      "epoch": 0.37846630518977536,
      "grad_norm": 9.32766056060791,
      "learning_rate": 6.954313474700514e-06,
      "loss": 0.1872,
      "step": 7329
    },
    {
      "epoch": 0.3785179447456752,
      "grad_norm": 4.401544094085693,
      "learning_rate": 6.953559268763981e-06,
      "loss": 0.0661,
      "step": 7330
    },
    {
      "epoch": 0.378569584301575,
      "grad_norm": 3.3740906715393066,
      "learning_rate": 6.952805010366421e-06,
      "loss": 0.0664,
      "step": 7331
    },
    {
      "epoch": 0.3786212238574748,
      "grad_norm": 10.519262313842773,
      "learning_rate": 6.952050699528086e-06,
      "loss": 0.0711,
      "step": 7332
    },
    {
      "epoch": 0.37867286341337464,
      "grad_norm": 3.2878835201263428,
      "learning_rate": 6.951296336269235e-06,
      "loss": 0.0597,
      "step": 7333
    },
    {
      "epoch": 0.37872450296927446,
      "grad_norm": 3.8209776878356934,
      "learning_rate": 6.950541920610123e-06,
      "loss": 0.0753,
      "step": 7334
    },
    {
      "epoch": 0.3787761425251743,
      "grad_norm": 4.789769649505615,
      "learning_rate": 6.949787452571011e-06,
      "loss": 0.0576,
      "step": 7335
    },
    {
      "epoch": 0.3788277820810741,
      "grad_norm": 12.336158752441406,
      "learning_rate": 6.949032932172158e-06,
      "loss": 0.0833,
      "step": 7336
    },
    {
      "epoch": 0.3788794216369739,
      "grad_norm": 2.41304087638855,
      "learning_rate": 6.948278359433827e-06,
      "loss": 0.0397,
      "step": 7337
    },
    {
      "epoch": 0.37893106119287373,
      "grad_norm": 2.9384942054748535,
      "learning_rate": 6.947523734376285e-06,
      "loss": 0.0482,
      "step": 7338
    },
    {
      "epoch": 0.37898270074877355,
      "grad_norm": 14.60787296295166,
      "learning_rate": 6.946769057019788e-06,
      "loss": 0.1234,
      "step": 7339
    },
    {
      "epoch": 0.37903434030467337,
      "grad_norm": 13.065210342407227,
      "learning_rate": 6.946014327384611e-06,
      "loss": 0.0917,
      "step": 7340
    },
    {
      "epoch": 0.3790859798605732,
      "grad_norm": 6.74562406539917,
      "learning_rate": 6.945259545491015e-06,
      "loss": 0.0326,
      "step": 7341
    },
    {
      "epoch": 0.379137619416473,
      "grad_norm": 2.4285569190979004,
      "learning_rate": 6.944504711359273e-06,
      "loss": 0.0322,
      "step": 7342
    },
    {
      "epoch": 0.3791892589723728,
      "grad_norm": 3.352548599243164,
      "learning_rate": 6.943749825009654e-06,
      "loss": 0.0639,
      "step": 7343
    },
    {
      "epoch": 0.37924089852827264,
      "grad_norm": 2.6687238216400146,
      "learning_rate": 6.942994886462431e-06,
      "loss": 0.0199,
      "step": 7344
    },
    {
      "epoch": 0.37929253808417246,
      "grad_norm": 3.8325133323669434,
      "learning_rate": 6.942239895737876e-06,
      "loss": 0.0655,
      "step": 7345
    },
    {
      "epoch": 0.3793441776400723,
      "grad_norm": 11.10177993774414,
      "learning_rate": 6.9414848528562614e-06,
      "loss": 0.1387,
      "step": 7346
    },
    {
      "epoch": 0.3793958171959721,
      "grad_norm": 3.0341718196868896,
      "learning_rate": 6.940729757837868e-06,
      "loss": 0.0093,
      "step": 7347
    },
    {
      "epoch": 0.3794474567518719,
      "grad_norm": 3.6667377948760986,
      "learning_rate": 6.939974610702968e-06,
      "loss": 0.0744,
      "step": 7348
    },
    {
      "epoch": 0.37949909630777173,
      "grad_norm": 8.539514541625977,
      "learning_rate": 6.9392194114718446e-06,
      "loss": 0.0811,
      "step": 7349
    },
    {
      "epoch": 0.37955073586367155,
      "grad_norm": 4.137991905212402,
      "learning_rate": 6.938464160164776e-06,
      "loss": 0.0628,
      "step": 7350
    },
    {
      "epoch": 0.37960237541957137,
      "grad_norm": 5.790907382965088,
      "learning_rate": 6.937708856802045e-06,
      "loss": 0.076,
      "step": 7351
    },
    {
      "epoch": 0.3796540149754712,
      "grad_norm": 17.730632781982422,
      "learning_rate": 6.936953501403933e-06,
      "loss": 0.2015,
      "step": 7352
    },
    {
      "epoch": 0.379705654531371,
      "grad_norm": 6.150522232055664,
      "learning_rate": 6.936198093990722e-06,
      "loss": 0.0754,
      "step": 7353
    },
    {
      "epoch": 0.3797572940872708,
      "grad_norm": 9.310583114624023,
      "learning_rate": 6.9354426345827044e-06,
      "loss": 0.0595,
      "step": 7354
    },
    {
      "epoch": 0.37980893364317064,
      "grad_norm": 5.856411457061768,
      "learning_rate": 6.934687123200161e-06,
      "loss": 0.0602,
      "step": 7355
    },
    {
      "epoch": 0.37986057319907046,
      "grad_norm": 2.975215196609497,
      "learning_rate": 6.9339315598633845e-06,
      "loss": 0.0495,
      "step": 7356
    },
    {
      "epoch": 0.3799122127549703,
      "grad_norm": 2.3655295372009277,
      "learning_rate": 6.933175944592664e-06,
      "loss": 0.0526,
      "step": 7357
    },
    {
      "epoch": 0.37996385231087015,
      "grad_norm": 5.187709808349609,
      "learning_rate": 6.932420277408288e-06,
      "loss": 0.0662,
      "step": 7358
    },
    {
      "epoch": 0.38001549186676997,
      "grad_norm": 3.183253765106201,
      "learning_rate": 6.931664558330553e-06,
      "loss": 0.0613,
      "step": 7359
    },
    {
      "epoch": 0.3800671314226698,
      "grad_norm": 4.560736179351807,
      "learning_rate": 6.930908787379751e-06,
      "loss": 0.0819,
      "step": 7360
    },
    {
      "epoch": 0.3801187709785696,
      "grad_norm": 6.506054401397705,
      "learning_rate": 6.930152964576179e-06,
      "loss": 0.0792,
      "step": 7361
    },
    {
      "epoch": 0.3801704105344694,
      "grad_norm": 5.085020065307617,
      "learning_rate": 6.929397089940132e-06,
      "loss": 0.0416,
      "step": 7362
    },
    {
      "epoch": 0.38022205009036925,
      "grad_norm": 5.496973991394043,
      "learning_rate": 6.9286411634919105e-06,
      "loss": 0.0462,
      "step": 7363
    },
    {
      "epoch": 0.38027368964626906,
      "grad_norm": 5.953623294830322,
      "learning_rate": 6.927885185251813e-06,
      "loss": 0.0914,
      "step": 7364
    },
    {
      "epoch": 0.3803253292021689,
      "grad_norm": 3.290870428085327,
      "learning_rate": 6.927129155240139e-06,
      "loss": 0.0462,
      "step": 7365
    },
    {
      "epoch": 0.3803769687580687,
      "grad_norm": 3.3137686252593994,
      "learning_rate": 6.9263730734771924e-06,
      "loss": 0.0321,
      "step": 7366
    },
    {
      "epoch": 0.3804286083139685,
      "grad_norm": 17.176271438598633,
      "learning_rate": 6.92561693998328e-06,
      "loss": 0.2384,
      "step": 7367
    },
    {
      "epoch": 0.38048024786986834,
      "grad_norm": 5.859920501708984,
      "learning_rate": 6.924860754778702e-06,
      "loss": 0.0749,
      "step": 7368
    },
    {
      "epoch": 0.38053188742576816,
      "grad_norm": 3.9910073280334473,
      "learning_rate": 6.9241045178837684e-06,
      "loss": 0.0772,
      "step": 7369
    },
    {
      "epoch": 0.380583526981668,
      "grad_norm": 6.269284725189209,
      "learning_rate": 6.923348229318786e-06,
      "loss": 0.092,
      "step": 7370
    },
    {
      "epoch": 0.3806351665375678,
      "grad_norm": 10.30794906616211,
      "learning_rate": 6.922591889104066e-06,
      "loss": 0.126,
      "step": 7371
    },
    {
      "epoch": 0.3806868060934676,
      "grad_norm": 5.219051361083984,
      "learning_rate": 6.9218354972599145e-06,
      "loss": 0.0721,
      "step": 7372
    },
    {
      "epoch": 0.38073844564936743,
      "grad_norm": 2.9049627780914307,
      "learning_rate": 6.92107905380665e-06,
      "loss": 0.0419,
      "step": 7373
    },
    {
      "epoch": 0.38079008520526725,
      "grad_norm": 13.478208541870117,
      "learning_rate": 6.920322558764581e-06,
      "loss": 0.1429,
      "step": 7374
    },
    {
      "epoch": 0.38084172476116707,
      "grad_norm": 2.410071611404419,
      "learning_rate": 6.919566012154027e-06,
      "loss": 0.02,
      "step": 7375
    },
    {
      "epoch": 0.3808933643170669,
      "grad_norm": 4.21217155456543,
      "learning_rate": 6.918809413995299e-06,
      "loss": 0.082,
      "step": 7376
    },
    {
      "epoch": 0.3809450038729667,
      "grad_norm": 9.10663890838623,
      "learning_rate": 6.918052764308719e-06,
      "loss": 0.104,
      "step": 7377
    },
    {
      "epoch": 0.3809966434288665,
      "grad_norm": 12.671621322631836,
      "learning_rate": 6.917296063114605e-06,
      "loss": 0.1181,
      "step": 7378
    },
    {
      "epoch": 0.38104828298476634,
      "grad_norm": 17.61737823486328,
      "learning_rate": 6.916539310433275e-06,
      "loss": 0.1083,
      "step": 7379
    },
    {
      "epoch": 0.38109992254066616,
      "grad_norm": 5.318961143493652,
      "learning_rate": 6.915782506285057e-06,
      "loss": 0.0475,
      "step": 7380
    },
    {
      "epoch": 0.381151562096566,
      "grad_norm": 2.9791977405548096,
      "learning_rate": 6.915025650690267e-06,
      "loss": 0.0401,
      "step": 7381
    },
    {
      "epoch": 0.3812032016524658,
      "grad_norm": 4.2118353843688965,
      "learning_rate": 6.914268743669235e-06,
      "loss": 0.0582,
      "step": 7382
    },
    {
      "epoch": 0.3812548412083656,
      "grad_norm": 11.982508659362793,
      "learning_rate": 6.913511785242285e-06,
      "loss": 0.1106,
      "step": 7383
    },
    {
      "epoch": 0.38130648076426543,
      "grad_norm": 2.943591594696045,
      "learning_rate": 6.912754775429744e-06,
      "loss": 0.0631,
      "step": 7384
    },
    {
      "epoch": 0.38135812032016525,
      "grad_norm": 3.9022209644317627,
      "learning_rate": 6.9119977142519415e-06,
      "loss": 0.0392,
      "step": 7385
    },
    {
      "epoch": 0.38140975987606507,
      "grad_norm": 5.756685256958008,
      "learning_rate": 6.911240601729207e-06,
      "loss": 0.056,
      "step": 7386
    },
    {
      "epoch": 0.3814613994319649,
      "grad_norm": 4.395662784576416,
      "learning_rate": 6.910483437881873e-06,
      "loss": 0.0414,
      "step": 7387
    },
    {
      "epoch": 0.3815130389878647,
      "grad_norm": 7.231332302093506,
      "learning_rate": 6.909726222730271e-06,
      "loss": 0.0758,
      "step": 7388
    },
    {
      "epoch": 0.3815646785437645,
      "grad_norm": 5.86823844909668,
      "learning_rate": 6.9089689562947385e-06,
      "loss": 0.0783,
      "step": 7389
    },
    {
      "epoch": 0.38161631809966434,
      "grad_norm": 13.268861770629883,
      "learning_rate": 6.908211638595608e-06,
      "loss": 0.1196,
      "step": 7390
    },
    {
      "epoch": 0.38166795765556416,
      "grad_norm": 11.491470336914062,
      "learning_rate": 6.907454269653217e-06,
      "loss": 0.0665,
      "step": 7391
    },
    {
      "epoch": 0.381719597211464,
      "grad_norm": 11.504761695861816,
      "learning_rate": 6.906696849487904e-06,
      "loss": 0.0701,
      "step": 7392
    },
    {
      "epoch": 0.3817712367673638,
      "grad_norm": 8.060310363769531,
      "learning_rate": 6.90593937812001e-06,
      "loss": 0.0846,
      "step": 7393
    },
    {
      "epoch": 0.3818228763232636,
      "grad_norm": 12.673036575317383,
      "learning_rate": 6.905181855569876e-06,
      "loss": 0.1392,
      "step": 7394
    },
    {
      "epoch": 0.38187451587916343,
      "grad_norm": 5.469425678253174,
      "learning_rate": 6.904424281857843e-06,
      "loss": 0.0714,
      "step": 7395
    },
    {
      "epoch": 0.38192615543506325,
      "grad_norm": 6.4876508712768555,
      "learning_rate": 6.903666657004257e-06,
      "loss": 0.0621,
      "step": 7396
    },
    {
      "epoch": 0.38197779499096307,
      "grad_norm": 6.740936756134033,
      "learning_rate": 6.902908981029462e-06,
      "loss": 0.0704,
      "step": 7397
    },
    {
      "epoch": 0.3820294345468629,
      "grad_norm": 3.2207350730895996,
      "learning_rate": 6.902151253953806e-06,
      "loss": 0.0358,
      "step": 7398
    },
    {
      "epoch": 0.3820810741027627,
      "grad_norm": 7.447697162628174,
      "learning_rate": 6.9013934757976355e-06,
      "loss": 0.09,
      "step": 7399
    },
    {
      "epoch": 0.3821327136586625,
      "grad_norm": 7.40045166015625,
      "learning_rate": 6.9006356465813e-06,
      "loss": 0.0558,
      "step": 7400
    },
    {
      "epoch": 0.38218435321456234,
      "grad_norm": 4.5542755126953125,
      "learning_rate": 6.89987776632515e-06,
      "loss": 0.0569,
      "step": 7401
    },
    {
      "epoch": 0.38223599277046216,
      "grad_norm": 2.966949462890625,
      "learning_rate": 6.899119835049539e-06,
      "loss": 0.0806,
      "step": 7402
    },
    {
      "epoch": 0.382287632326362,
      "grad_norm": 4.130910396575928,
      "learning_rate": 6.898361852774821e-06,
      "loss": 0.0586,
      "step": 7403
    },
    {
      "epoch": 0.3823392718822618,
      "grad_norm": 9.003682136535645,
      "learning_rate": 6.89760381952135e-06,
      "loss": 0.1011,
      "step": 7404
    },
    {
      "epoch": 0.3823909114381616,
      "grad_norm": 2.7038731575012207,
      "learning_rate": 6.896845735309481e-06,
      "loss": 0.0632,
      "step": 7405
    },
    {
      "epoch": 0.38244255099406144,
      "grad_norm": 3.6572823524475098,
      "learning_rate": 6.896087600159576e-06,
      "loss": 0.0944,
      "step": 7406
    },
    {
      "epoch": 0.38249419054996125,
      "grad_norm": 10.140886306762695,
      "learning_rate": 6.895329414091987e-06,
      "loss": 0.0866,
      "step": 7407
    },
    {
      "epoch": 0.3825458301058611,
      "grad_norm": 4.8298540115356445,
      "learning_rate": 6.89457117712708e-06,
      "loss": 0.0921,
      "step": 7408
    },
    {
      "epoch": 0.3825974696617609,
      "grad_norm": 3.153381586074829,
      "learning_rate": 6.893812889285215e-06,
      "loss": 0.0387,
      "step": 7409
    },
    {
      "epoch": 0.3826491092176607,
      "grad_norm": 5.577614784240723,
      "learning_rate": 6.893054550586756e-06,
      "loss": 0.0754,
      "step": 7410
    },
    {
      "epoch": 0.38270074877356053,
      "grad_norm": 5.129656791687012,
      "learning_rate": 6.8922961610520654e-06,
      "loss": 0.0555,
      "step": 7411
    },
    {
      "epoch": 0.38275238832946035,
      "grad_norm": 8.034024238586426,
      "learning_rate": 6.89153772070151e-06,
      "loss": 0.1221,
      "step": 7412
    },
    {
      "epoch": 0.38280402788536017,
      "grad_norm": 2.4777464866638184,
      "learning_rate": 6.890779229555459e-06,
      "loss": 0.0674,
      "step": 7413
    },
    {
      "epoch": 0.38285566744126,
      "grad_norm": 2.361435890197754,
      "learning_rate": 6.890020687634277e-06,
      "loss": 0.0246,
      "step": 7414
    },
    {
      "epoch": 0.3829073069971598,
      "grad_norm": 17.22631072998047,
      "learning_rate": 6.8892620949583375e-06,
      "loss": 0.1822,
      "step": 7415
    },
    {
      "epoch": 0.3829589465530596,
      "grad_norm": 6.0657477378845215,
      "learning_rate": 6.888503451548011e-06,
      "loss": 0.1169,
      "step": 7416
    },
    {
      "epoch": 0.38301058610895944,
      "grad_norm": 9.060077667236328,
      "learning_rate": 6.887744757423669e-06,
      "loss": 0.1167,
      "step": 7417
    },
    {
      "epoch": 0.38306222566485926,
      "grad_norm": 2.365447998046875,
      "learning_rate": 6.886986012605688e-06,
      "loss": 0.0355,
      "step": 7418
    },
    {
      "epoch": 0.3831138652207591,
      "grad_norm": 7.603076457977295,
      "learning_rate": 6.8862272171144386e-06,
      "loss": 0.0715,
      "step": 7419
    },
    {
      "epoch": 0.3831655047766589,
      "grad_norm": 1.9459919929504395,
      "learning_rate": 6.885468370970303e-06,
      "loss": 0.0296,
      "step": 7420
    },
    {
      "epoch": 0.3832171443325587,
      "grad_norm": 2.1549272537231445,
      "learning_rate": 6.884709474193655e-06,
      "loss": 0.0191,
      "step": 7421
    },
    {
      "epoch": 0.3832687838884586,
      "grad_norm": 3.9201500415802,
      "learning_rate": 6.883950526804878e-06,
      "loss": 0.074,
      "step": 7422
    },
    {
      "epoch": 0.3833204234443584,
      "grad_norm": 3.5924232006073,
      "learning_rate": 6.8831915288243505e-06,
      "loss": 0.048,
      "step": 7423
    },
    {
      "epoch": 0.3833720630002582,
      "grad_norm": 4.235139846801758,
      "learning_rate": 6.882432480272453e-06,
      "loss": 0.0656,
      "step": 7424
    },
    {
      "epoch": 0.38342370255615804,
      "grad_norm": 1.2260642051696777,
      "learning_rate": 6.881673381169572e-06,
      "loss": 0.0147,
      "step": 7425
    },
    {
      "epoch": 0.38347534211205786,
      "grad_norm": 2.23142671585083,
      "learning_rate": 6.880914231536092e-06,
      "loss": 0.0415,
      "step": 7426
    },
    {
      "epoch": 0.3835269816679577,
      "grad_norm": 2.179253101348877,
      "learning_rate": 6.880155031392398e-06,
      "loss": 0.0391,
      "step": 7427
    },
    {
      "epoch": 0.3835786212238575,
      "grad_norm": 11.238953590393066,
      "learning_rate": 6.87939578075888e-06,
      "loss": 0.1177,
      "step": 7428
    },
    {
      "epoch": 0.3836302607797573,
      "grad_norm": 4.20920467376709,
      "learning_rate": 6.878636479655925e-06,
      "loss": 0.1081,
      "step": 7429
    },
    {
      "epoch": 0.38368190033565713,
      "grad_norm": 8.218926429748535,
      "learning_rate": 6.8778771281039215e-06,
      "loss": 0.0882,
      "step": 7430
    },
    {
      "epoch": 0.38373353989155695,
      "grad_norm": 5.961413383483887,
      "learning_rate": 6.877117726123264e-06,
      "loss": 0.0582,
      "step": 7431
    },
    {
      "epoch": 0.38378517944745677,
      "grad_norm": 3.7436697483062744,
      "learning_rate": 6.876358273734346e-06,
      "loss": 0.0435,
      "step": 7432
    },
    {
      "epoch": 0.3838368190033566,
      "grad_norm": 3.176680326461792,
      "learning_rate": 6.87559877095756e-06,
      "loss": 0.0403,
      "step": 7433
    },
    {
      "epoch": 0.3838884585592564,
      "grad_norm": 2.8936424255371094,
      "learning_rate": 6.8748392178133025e-06,
      "loss": 0.0747,
      "step": 7434
    },
    {
      "epoch": 0.3839400981151562,
      "grad_norm": 3.8305752277374268,
      "learning_rate": 6.87407961432197e-06,
      "loss": 0.1004,
      "step": 7435
    },
    {
      "epoch": 0.38399173767105604,
      "grad_norm": 6.074758529663086,
      "learning_rate": 6.873319960503962e-06,
      "loss": 0.0678,
      "step": 7436
    },
    {
      "epoch": 0.38404337722695586,
      "grad_norm": 15.441750526428223,
      "learning_rate": 6.872560256379679e-06,
      "loss": 0.1643,
      "step": 7437
    },
    {
      "epoch": 0.3840950167828557,
      "grad_norm": 1.434195876121521,
      "learning_rate": 6.871800501969519e-06,
      "loss": 0.0326,
      "step": 7438
    },
    {
      "epoch": 0.3841466563387555,
      "grad_norm": 2.0978803634643555,
      "learning_rate": 6.871040697293888e-06,
      "loss": 0.0215,
      "step": 7439
    },
    {
      "epoch": 0.3841982958946553,
      "grad_norm": 3.517343521118164,
      "learning_rate": 6.870280842373188e-06,
      "loss": 0.0571,
      "step": 7440
    },
    {
      "epoch": 0.38424993545055514,
      "grad_norm": 12.984092712402344,
      "learning_rate": 6.869520937227825e-06,
      "loss": 0.1467,
      "step": 7441
    },
    {
      "epoch": 0.38430157500645495,
      "grad_norm": 2.615190267562866,
      "learning_rate": 6.868760981878204e-06,
      "loss": 0.0473,
      "step": 7442
    },
    {
      "epoch": 0.3843532145623548,
      "grad_norm": 4.829897403717041,
      "learning_rate": 6.868000976344734e-06,
      "loss": 0.0609,
      "step": 7443
    },
    {
      "epoch": 0.3844048541182546,
      "grad_norm": 6.425693511962891,
      "learning_rate": 6.8672409206478255e-06,
      "loss": 0.0768,
      "step": 7444
    },
    {
      "epoch": 0.3844564936741544,
      "grad_norm": 8.27495288848877,
      "learning_rate": 6.866480814807885e-06,
      "loss": 0.0349,
      "step": 7445
    },
    {
      "epoch": 0.3845081332300542,
      "grad_norm": 7.404259204864502,
      "learning_rate": 6.865720658845331e-06,
      "loss": 0.1013,
      "step": 7446
    },
    {
      "epoch": 0.38455977278595405,
      "grad_norm": 4.503859043121338,
      "learning_rate": 6.86496045278057e-06,
      "loss": 0.0454,
      "step": 7447
    },
    {
      "epoch": 0.38461141234185386,
      "grad_norm": 2.8034892082214355,
      "learning_rate": 6.8642001966340225e-06,
      "loss": 0.0378,
      "step": 7448
    },
    {
      "epoch": 0.3846630518977537,
      "grad_norm": 10.23360824584961,
      "learning_rate": 6.8634398904261e-06,
      "loss": 0.1516,
      "step": 7449
    },
    {
      "epoch": 0.3847146914536535,
      "grad_norm": 2.8482673168182373,
      "learning_rate": 6.862679534177221e-06,
      "loss": 0.0589,
      "step": 7450
    },
    {
      "epoch": 0.3847663310095533,
      "grad_norm": 1.9434109926223755,
      "learning_rate": 6.861919127907806e-06,
      "loss": 0.0192,
      "step": 7451
    },
    {
      "epoch": 0.38481797056545314,
      "grad_norm": 4.52139139175415,
      "learning_rate": 6.861158671638272e-06,
      "loss": 0.0631,
      "step": 7452
    },
    {
      "epoch": 0.38486961012135296,
      "grad_norm": 12.201322555541992,
      "learning_rate": 6.8603981653890435e-06,
      "loss": 0.1026,
      "step": 7453
    },
    {
      "epoch": 0.3849212496772528,
      "grad_norm": 19.916778564453125,
      "learning_rate": 6.859637609180541e-06,
      "loss": 0.1484,
      "step": 7454
    },
    {
      "epoch": 0.3849728892331526,
      "grad_norm": 7.437620162963867,
      "learning_rate": 6.8588770030331895e-06,
      "loss": 0.0464,
      "step": 7455
    },
    {
      "epoch": 0.3850245287890524,
      "grad_norm": 4.3377814292907715,
      "learning_rate": 6.8581163469674135e-06,
      "loss": 0.0632,
      "step": 7456
    },
    {
      "epoch": 0.38507616834495223,
      "grad_norm": 2.397981882095337,
      "learning_rate": 6.8573556410036415e-06,
      "loss": 0.0164,
      "step": 7457
    },
    {
      "epoch": 0.38512780790085205,
      "grad_norm": 4.055746078491211,
      "learning_rate": 6.856594885162299e-06,
      "loss": 0.0476,
      "step": 7458
    },
    {
      "epoch": 0.38517944745675187,
      "grad_norm": 14.374281883239746,
      "learning_rate": 6.855834079463818e-06,
      "loss": 0.0842,
      "step": 7459
    },
    {
      "epoch": 0.3852310870126517,
      "grad_norm": 2.011359691619873,
      "learning_rate": 6.8550732239286266e-06,
      "loss": 0.0442,
      "step": 7460
    },
    {
      "epoch": 0.3852827265685515,
      "grad_norm": 4.497236251831055,
      "learning_rate": 6.854312318577159e-06,
      "loss": 0.0305,
      "step": 7461
    },
    {
      "epoch": 0.3853343661244513,
      "grad_norm": 3.939819812774658,
      "learning_rate": 6.853551363429849e-06,
      "loss": 0.0185,
      "step": 7462
    },
    {
      "epoch": 0.38538600568035114,
      "grad_norm": 2.984931468963623,
      "learning_rate": 6.852790358507129e-06,
      "loss": 0.0416,
      "step": 7463
    },
    {
      "epoch": 0.38543764523625096,
      "grad_norm": 2.8312158584594727,
      "learning_rate": 6.852029303829437e-06,
      "loss": 0.0433,
      "step": 7464
    },
    {
      "epoch": 0.3854892847921508,
      "grad_norm": 4.1056718826293945,
      "learning_rate": 6.851268199417209e-06,
      "loss": 0.0806,
      "step": 7465
    },
    {
      "epoch": 0.3855409243480506,
      "grad_norm": 12.857012748718262,
      "learning_rate": 6.850507045290884e-06,
      "loss": 0.273,
      "step": 7466
    },
    {
      "epoch": 0.3855925639039504,
      "grad_norm": 10.02283763885498,
      "learning_rate": 6.849745841470904e-06,
      "loss": 0.0653,
      "step": 7467
    },
    {
      "epoch": 0.38564420345985023,
      "grad_norm": 21.170412063598633,
      "learning_rate": 6.848984587977708e-06,
      "loss": 0.2826,
      "step": 7468
    },
    {
      "epoch": 0.38569584301575005,
      "grad_norm": 13.14614486694336,
      "learning_rate": 6.84822328483174e-06,
      "loss": 0.1174,
      "step": 7469
    },
    {
      "epoch": 0.38574748257164987,
      "grad_norm": 4.871288776397705,
      "learning_rate": 6.847461932053445e-06,
      "loss": 0.1093,
      "step": 7470
    },
    {
      "epoch": 0.3857991221275497,
      "grad_norm": 10.121049880981445,
      "learning_rate": 6.846700529663265e-06,
      "loss": 0.09,
      "step": 7471
    },
    {
      "epoch": 0.3858507616834495,
      "grad_norm": 2.296705484390259,
      "learning_rate": 6.845939077681652e-06,
      "loss": 0.0386,
      "step": 7472
    },
    {
      "epoch": 0.3859024012393493,
      "grad_norm": 11.499853134155273,
      "learning_rate": 6.845177576129048e-06,
      "loss": 0.1,
      "step": 7473
    },
    {
      "epoch": 0.38595404079524914,
      "grad_norm": 9.185375213623047,
      "learning_rate": 6.844416025025907e-06,
      "loss": 0.1286,
      "step": 7474
    },
    {
      "epoch": 0.38600568035114896,
      "grad_norm": 15.321711540222168,
      "learning_rate": 6.843654424392679e-06,
      "loss": 0.3167,
      "step": 7475
    },
    {
      "epoch": 0.3860573199070488,
      "grad_norm": 1.8249253034591675,
      "learning_rate": 6.842892774249814e-06,
      "loss": 0.0266,
      "step": 7476
    },
    {
      "epoch": 0.3861089594629486,
      "grad_norm": 7.4253458976745605,
      "learning_rate": 6.842131074617768e-06,
      "loss": 0.0671,
      "step": 7477
    },
    {
      "epoch": 0.3861605990188484,
      "grad_norm": 8.424676895141602,
      "learning_rate": 6.841369325516993e-06,
      "loss": 0.1242,
      "step": 7478
    },
    {
      "epoch": 0.38621223857474823,
      "grad_norm": 8.784598350524902,
      "learning_rate": 6.840607526967949e-06,
      "loss": 0.0588,
      "step": 7479
    },
    {
      "epoch": 0.38626387813064805,
      "grad_norm": 3.2396984100341797,
      "learning_rate": 6.839845678991089e-06,
      "loss": 0.0543,
      "step": 7480
    },
    {
      "epoch": 0.38631551768654787,
      "grad_norm": 2.2629268169403076,
      "learning_rate": 6.839083781606874e-06,
      "loss": 0.0421,
      "step": 7481
    },
    {
      "epoch": 0.3863671572424477,
      "grad_norm": 2.4812912940979004,
      "learning_rate": 6.838321834835765e-06,
      "loss": 0.0225,
      "step": 7482
    },
    {
      "epoch": 0.3864187967983475,
      "grad_norm": 42.713932037353516,
      "learning_rate": 6.837559838698221e-06,
      "loss": 0.0801,
      "step": 7483
    },
    {
      "epoch": 0.3864704363542473,
      "grad_norm": 3.148092031478882,
      "learning_rate": 6.836797793214707e-06,
      "loss": 0.032,
      "step": 7484
    },
    {
      "epoch": 0.3865220759101472,
      "grad_norm": 9.080384254455566,
      "learning_rate": 6.836035698405684e-06,
      "loss": 0.1056,
      "step": 7485
    },
    {
      "epoch": 0.386573715466047,
      "grad_norm": 3.9702603816986084,
      "learning_rate": 6.835273554291622e-06,
      "loss": 0.0586,
      "step": 7486
    },
    {
      "epoch": 0.38662535502194684,
      "grad_norm": 1.65451979637146,
      "learning_rate": 6.834511360892981e-06,
      "loss": 0.0311,
      "step": 7487
    },
    {
      "epoch": 0.38667699457784666,
      "grad_norm": 6.584075927734375,
      "learning_rate": 6.833749118230236e-06,
      "loss": 0.0911,
      "step": 7488
    },
    {
      "epoch": 0.3867286341337465,
      "grad_norm": 2.7955663204193115,
      "learning_rate": 6.832986826323853e-06,
      "loss": 0.0777,
      "step": 7489
    },
    {
      "epoch": 0.3867802736896463,
      "grad_norm": 2.879802942276001,
      "learning_rate": 6.832224485194302e-06,
      "loss": 0.0435,
      "step": 7490
    },
    {
      "epoch": 0.3868319132455461,
      "grad_norm": 4.458540439605713,
      "learning_rate": 6.831462094862056e-06,
      "loss": 0.0437,
      "step": 7491
    },
    {
      "epoch": 0.38688355280144593,
      "grad_norm": 11.028694152832031,
      "learning_rate": 6.830699655347588e-06,
      "loss": 0.0647,
      "step": 7492
    },
    {
      "epoch": 0.38693519235734575,
      "grad_norm": 6.911690711975098,
      "learning_rate": 6.829937166671372e-06,
      "loss": 0.065,
      "step": 7493
    },
    {
      "epoch": 0.38698683191324557,
      "grad_norm": 8.483264923095703,
      "learning_rate": 6.829174628853885e-06,
      "loss": 0.0741,
      "step": 7494
    },
    {
      "epoch": 0.3870384714691454,
      "grad_norm": 3.8362228870391846,
      "learning_rate": 6.828412041915607e-06,
      "loss": 0.0922,
      "step": 7495
    },
    {
      "epoch": 0.3870901110250452,
      "grad_norm": 2.970140218734741,
      "learning_rate": 6.82764940587701e-06,
      "loss": 0.0538,
      "step": 7496
    },
    {
      "epoch": 0.387141750580945,
      "grad_norm": 14.03698444366455,
      "learning_rate": 6.826886720758579e-06,
      "loss": 0.3357,
      "step": 7497
    },
    {
      "epoch": 0.38719339013684484,
      "grad_norm": 5.619626045227051,
      "learning_rate": 6.826123986580791e-06,
      "loss": 0.086,
      "step": 7498
    },
    {
      "epoch": 0.38724502969274466,
      "grad_norm": 9.336917877197266,
      "learning_rate": 6.825361203364133e-06,
      "loss": 0.1704,
      "step": 7499
    },
    {
      "epoch": 0.3872966692486445,
      "grad_norm": 19.063430786132812,
      "learning_rate": 6.824598371129088e-06,
      "loss": 0.1322,
      "step": 7500
    },
    {
      "epoch": 0.3873483088045443,
      "grad_norm": 4.2032551765441895,
      "learning_rate": 6.823835489896138e-06,
      "loss": 0.065,
      "step": 7501
    },
    {
      "epoch": 0.3873999483604441,
      "grad_norm": 4.974436283111572,
      "learning_rate": 6.823072559685773e-06,
      "loss": 0.0787,
      "step": 7502
    },
    {
      "epoch": 0.38745158791634393,
      "grad_norm": 4.905792713165283,
      "learning_rate": 6.822309580518479e-06,
      "loss": 0.033,
      "step": 7503
    },
    {
      "epoch": 0.38750322747224375,
      "grad_norm": 7.796116828918457,
      "learning_rate": 6.8215465524147464e-06,
      "loss": 0.0558,
      "step": 7504
    },
    {
      "epoch": 0.38755486702814357,
      "grad_norm": 4.1023125648498535,
      "learning_rate": 6.820783475395063e-06,
      "loss": 0.0705,
      "step": 7505
    },
    {
      "epoch": 0.3876065065840434,
      "grad_norm": 1.6420618295669556,
      "learning_rate": 6.820020349479922e-06,
      "loss": 0.0255,
      "step": 7506
    },
    {
      "epoch": 0.3876581461399432,
      "grad_norm": 1.56080162525177,
      "learning_rate": 6.819257174689818e-06,
      "loss": 0.0251,
      "step": 7507
    },
    {
      "epoch": 0.387709785695843,
      "grad_norm": 1.6670117378234863,
      "learning_rate": 6.818493951045243e-06,
      "loss": 0.0289,
      "step": 7508
    },
    {
      "epoch": 0.38776142525174284,
      "grad_norm": 4.495280742645264,
      "learning_rate": 6.817730678566694e-06,
      "loss": 0.0612,
      "step": 7509
    },
    {
      "epoch": 0.38781306480764266,
      "grad_norm": 8.11846923828125,
      "learning_rate": 6.816967357274668e-06,
      "loss": 0.0792,
      "step": 7510
    },
    {
      "epoch": 0.3878647043635425,
      "grad_norm": 4.050751209259033,
      "learning_rate": 6.816203987189661e-06,
      "loss": 0.0466,
      "step": 7511
    },
    {
      "epoch": 0.3879163439194423,
      "grad_norm": 8.996736526489258,
      "learning_rate": 6.8154405683321765e-06,
      "loss": 0.1437,
      "step": 7512
    },
    {
      "epoch": 0.3879679834753421,
      "grad_norm": 2.695706605911255,
      "learning_rate": 6.814677100722713e-06,
      "loss": 0.042,
      "step": 7513
    },
    {
      "epoch": 0.38801962303124193,
      "grad_norm": 3.075333595275879,
      "learning_rate": 6.813913584381774e-06,
      "loss": 0.0521,
      "step": 7514
    },
    {
      "epoch": 0.38807126258714175,
      "grad_norm": 12.801240921020508,
      "learning_rate": 6.81315001932986e-06,
      "loss": 0.0719,
      "step": 7515
    },
    {
      "epoch": 0.38812290214304157,
      "grad_norm": 4.329682350158691,
      "learning_rate": 6.81238640558748e-06,
      "loss": 0.0598,
      "step": 7516
    },
    {
      "epoch": 0.3881745416989414,
      "grad_norm": 9.640810012817383,
      "learning_rate": 6.811622743175138e-06,
      "loss": 0.1464,
      "step": 7517
    },
    {
      "epoch": 0.3882261812548412,
      "grad_norm": 4.086205959320068,
      "learning_rate": 6.81085903211334e-06,
      "loss": 0.0666,
      "step": 7518
    },
    {
      "epoch": 0.388277820810741,
      "grad_norm": 5.964380264282227,
      "learning_rate": 6.8100952724225975e-06,
      "loss": 0.029,
      "step": 7519
    },
    {
      "epoch": 0.38832946036664084,
      "grad_norm": 4.471297264099121,
      "learning_rate": 6.809331464123419e-06,
      "loss": 0.0654,
      "step": 7520
    },
    {
      "epoch": 0.38838109992254066,
      "grad_norm": 2.278707265853882,
      "learning_rate": 6.808567607236317e-06,
      "loss": 0.0456,
      "step": 7521
    },
    {
      "epoch": 0.3884327394784405,
      "grad_norm": 5.770811080932617,
      "learning_rate": 6.807803701781802e-06,
      "loss": 0.0831,
      "step": 7522
    },
    {
      "epoch": 0.3884843790343403,
      "grad_norm": 7.204294681549072,
      "learning_rate": 6.807039747780391e-06,
      "loss": 0.0616,
      "step": 7523
    },
    {
      "epoch": 0.3885360185902401,
      "grad_norm": 3.126631259918213,
      "learning_rate": 6.806275745252596e-06,
      "loss": 0.0289,
      "step": 7524
    },
    {
      "epoch": 0.38858765814613994,
      "grad_norm": 4.784875392913818,
      "learning_rate": 6.805511694218936e-06,
      "loss": 0.0719,
      "step": 7525
    },
    {
      "epoch": 0.38863929770203975,
      "grad_norm": 4.644952774047852,
      "learning_rate": 6.8047475946999295e-06,
      "loss": 0.033,
      "step": 7526
    },
    {
      "epoch": 0.3886909372579396,
      "grad_norm": 4.773207664489746,
      "learning_rate": 6.8039834467160925e-06,
      "loss": 0.0535,
      "step": 7527
    },
    {
      "epoch": 0.3887425768138394,
      "grad_norm": 8.034405708312988,
      "learning_rate": 6.803219250287949e-06,
      "loss": 0.163,
      "step": 7528
    },
    {
      "epoch": 0.3887942163697392,
      "grad_norm": 3.886758327484131,
      "learning_rate": 6.802455005436017e-06,
      "loss": 0.058,
      "step": 7529
    },
    {
      "epoch": 0.38884585592563903,
      "grad_norm": 5.106607437133789,
      "learning_rate": 6.801690712180824e-06,
      "loss": 0.0757,
      "step": 7530
    },
    {
      "epoch": 0.38889749548153885,
      "grad_norm": 5.400914192199707,
      "learning_rate": 6.800926370542891e-06,
      "loss": 0.0442,
      "step": 7531
    },
    {
      "epoch": 0.38894913503743866,
      "grad_norm": 3.34883975982666,
      "learning_rate": 6.800161980542743e-06,
      "loss": 0.0417,
      "step": 7532
    },
    {
      "epoch": 0.3890007745933385,
      "grad_norm": 6.852285861968994,
      "learning_rate": 6.799397542200911e-06,
      "loss": 0.0765,
      "step": 7533
    },
    {
      "epoch": 0.3890524141492383,
      "grad_norm": 2.5015952587127686,
      "learning_rate": 6.798633055537921e-06,
      "loss": 0.0458,
      "step": 7534
    },
    {
      "epoch": 0.3891040537051381,
      "grad_norm": 3.986078977584839,
      "learning_rate": 6.797868520574303e-06,
      "loss": 0.041,
      "step": 7535
    },
    {
      "epoch": 0.38915569326103794,
      "grad_norm": 9.321161270141602,
      "learning_rate": 6.797103937330587e-06,
      "loss": 0.0973,
      "step": 7536
    },
    {
      "epoch": 0.38920733281693776,
      "grad_norm": 9.792168617248535,
      "learning_rate": 6.796339305827304e-06,
      "loss": 0.1996,
      "step": 7537
    },
    {
      "epoch": 0.3892589723728376,
      "grad_norm": 2.6174216270446777,
      "learning_rate": 6.795574626084991e-06,
      "loss": 0.028,
      "step": 7538
    },
    {
      "epoch": 0.3893106119287374,
      "grad_norm": 11.561380386352539,
      "learning_rate": 6.794809898124181e-06,
      "loss": 0.044,
      "step": 7539
    },
    {
      "epoch": 0.3893622514846372,
      "grad_norm": 2.884209156036377,
      "learning_rate": 6.794045121965409e-06,
      "loss": 0.0695,
      "step": 7540
    },
    {
      "epoch": 0.38941389104053703,
      "grad_norm": 7.734256267547607,
      "learning_rate": 6.793280297629213e-06,
      "loss": 0.0814,
      "step": 7541
    },
    {
      "epoch": 0.38946553059643685,
      "grad_norm": 5.339278697967529,
      "learning_rate": 6.792515425136133e-06,
      "loss": 0.0944,
      "step": 7542
    },
    {
      "epoch": 0.38951717015233667,
      "grad_norm": 3.6020169258117676,
      "learning_rate": 6.791750504506709e-06,
      "loss": 0.0707,
      "step": 7543
    },
    {
      "epoch": 0.3895688097082365,
      "grad_norm": 3.1472058296203613,
      "learning_rate": 6.790985535761479e-06,
      "loss": 0.0537,
      "step": 7544
    },
    {
      "epoch": 0.3896204492641363,
      "grad_norm": 3.2553176879882812,
      "learning_rate": 6.790220518920989e-06,
      "loss": 0.0631,
      "step": 7545
    },
    {
      "epoch": 0.3896720888200361,
      "grad_norm": 4.180641174316406,
      "learning_rate": 6.78945545400578e-06,
      "loss": 0.0624,
      "step": 7546
    },
    {
      "epoch": 0.38972372837593594,
      "grad_norm": 3.3718552589416504,
      "learning_rate": 6.7886903410364016e-06,
      "loss": 0.0489,
      "step": 7547
    },
    {
      "epoch": 0.38977536793183576,
      "grad_norm": 2.3267176151275635,
      "learning_rate": 6.787925180033394e-06,
      "loss": 0.0525,
      "step": 7548
    },
    {
      "epoch": 0.38982700748773563,
      "grad_norm": 26.04310417175293,
      "learning_rate": 6.78715997101731e-06,
      "loss": 0.0923,
      "step": 7549
    },
    {
      "epoch": 0.38987864704363545,
      "grad_norm": 3.466068983078003,
      "learning_rate": 6.786394714008696e-06,
      "loss": 0.0501,
      "step": 7550
    },
    {
      "epoch": 0.38993028659953527,
      "grad_norm": 3.2217743396759033,
      "learning_rate": 6.785629409028104e-06,
      "loss": 0.0659,
      "step": 7551
    },
    {
      "epoch": 0.3899819261554351,
      "grad_norm": 7.611376762390137,
      "learning_rate": 6.784864056096084e-06,
      "loss": 0.0863,
      "step": 7552
    },
    {
      "epoch": 0.3900335657113349,
      "grad_norm": 5.5653910636901855,
      "learning_rate": 6.784098655233189e-06,
      "loss": 0.0509,
      "step": 7553
    },
    {
      "epoch": 0.3900852052672347,
      "grad_norm": 3.564077377319336,
      "learning_rate": 6.783333206459975e-06,
      "loss": 0.0641,
      "step": 7554
    },
    {
      "epoch": 0.39013684482313454,
      "grad_norm": 7.70734167098999,
      "learning_rate": 6.782567709796994e-06,
      "loss": 0.1194,
      "step": 7555
    },
    {
      "epoch": 0.39018848437903436,
      "grad_norm": 2.79093599319458,
      "learning_rate": 6.781802165264805e-06,
      "loss": 0.0641,
      "step": 7556
    },
    {
      "epoch": 0.3902401239349342,
      "grad_norm": 13.362868309020996,
      "learning_rate": 6.781036572883965e-06,
      "loss": 0.2186,
      "step": 7557
    },
    {
      "epoch": 0.390291763490834,
      "grad_norm": 12.180268287658691,
      "learning_rate": 6.780270932675036e-06,
      "loss": 0.135,
      "step": 7558
    },
    {
      "epoch": 0.3903434030467338,
      "grad_norm": 5.260475158691406,
      "learning_rate": 6.779505244658577e-06,
      "loss": 0.0409,
      "step": 7559
    },
    {
      "epoch": 0.39039504260263364,
      "grad_norm": 1.7392182350158691,
      "learning_rate": 6.7787395088551456e-06,
      "loss": 0.0386,
      "step": 7560
    },
    {
      "epoch": 0.39044668215853345,
      "grad_norm": 8.907304763793945,
      "learning_rate": 6.777973725285312e-06,
      "loss": 0.0793,
      "step": 7561
    },
    {
      "epoch": 0.39049832171443327,
      "grad_norm": 17.35108184814453,
      "learning_rate": 6.777207893969635e-06,
      "loss": 0.2016,
      "step": 7562
    },
    {
      "epoch": 0.3905499612703331,
      "grad_norm": 3.087421178817749,
      "learning_rate": 6.776442014928684e-06,
      "loss": 0.0507,
      "step": 7563
    },
    {
      "epoch": 0.3906016008262329,
      "grad_norm": 5.846672534942627,
      "learning_rate": 6.775676088183024e-06,
      "loss": 0.0473,
      "step": 7564
    },
    {
      "epoch": 0.3906532403821327,
      "grad_norm": 6.755026817321777,
      "learning_rate": 6.774910113753224e-06,
      "loss": 0.1091,
      "step": 7565
    },
    {
      "epoch": 0.39070487993803255,
      "grad_norm": 3.236663341522217,
      "learning_rate": 6.774144091659853e-06,
      "loss": 0.0526,
      "step": 7566
    },
    {
      "epoch": 0.39075651949393236,
      "grad_norm": 5.627609729766846,
      "learning_rate": 6.773378021923483e-06,
      "loss": 0.1564,
      "step": 7567
    },
    {
      "epoch": 0.3908081590498322,
      "grad_norm": 2.2500555515289307,
      "learning_rate": 6.772611904564683e-06,
      "loss": 0.0419,
      "step": 7568
    },
    {
      "epoch": 0.390859798605732,
      "grad_norm": 3.205740213394165,
      "learning_rate": 6.77184573960403e-06,
      "loss": 0.0526,
      "step": 7569
    },
    {
      "epoch": 0.3909114381616318,
      "grad_norm": 2.7412052154541016,
      "learning_rate": 6.771079527062099e-06,
      "loss": 0.0775,
      "step": 7570
    },
    {
      "epoch": 0.39096307771753164,
      "grad_norm": 6.44580078125,
      "learning_rate": 6.770313266959462e-06,
      "loss": 0.1065,
      "step": 7571
    },
    {
      "epoch": 0.39101471727343146,
      "grad_norm": 3.8327443599700928,
      "learning_rate": 6.7695469593167e-06,
      "loss": 0.07,
      "step": 7572
    },
    {
      "epoch": 0.3910663568293313,
      "grad_norm": 3.4688937664031982,
      "learning_rate": 6.768780604154388e-06,
      "loss": 0.0776,
      "step": 7573
    },
    {
      "epoch": 0.3911179963852311,
      "grad_norm": 7.97922420501709,
      "learning_rate": 6.768014201493108e-06,
      "loss": 0.0481,
      "step": 7574
    },
    {
      "epoch": 0.3911696359411309,
      "grad_norm": 3.310401201248169,
      "learning_rate": 6.767247751353441e-06,
      "loss": 0.066,
      "step": 7575
    },
    {
      "epoch": 0.39122127549703073,
      "grad_norm": 5.541158676147461,
      "learning_rate": 6.7664812537559684e-06,
      "loss": 0.0799,
      "step": 7576
    },
    {
      "epoch": 0.39127291505293055,
      "grad_norm": 5.739167213439941,
      "learning_rate": 6.765714708721276e-06,
      "loss": 0.1026,
      "step": 7577
    },
    {
      "epoch": 0.39132455460883037,
      "grad_norm": 2.3938121795654297,
      "learning_rate": 6.764948116269947e-06,
      "loss": 0.0507,
      "step": 7578
    },
    {
      "epoch": 0.3913761941647302,
      "grad_norm": 1.5269285440444946,
      "learning_rate": 6.764181476422564e-06,
      "loss": 0.0266,
      "step": 7579
    },
    {
      "epoch": 0.39142783372063,
      "grad_norm": 1.8005130290985107,
      "learning_rate": 6.763414789199722e-06,
      "loss": 0.0351,
      "step": 7580
    },
    {
      "epoch": 0.3914794732765298,
      "grad_norm": 6.607802867889404,
      "learning_rate": 6.7626480546220046e-06,
      "loss": 0.1215,
      "step": 7581
    },
    {
      "epoch": 0.39153111283242964,
      "grad_norm": 4.856544017791748,
      "learning_rate": 6.761881272710003e-06,
      "loss": 0.0473,
      "step": 7582
    },
    {
      "epoch": 0.39158275238832946,
      "grad_norm": 3.899553060531616,
      "learning_rate": 6.761114443484307e-06,
      "loss": 0.0583,
      "step": 7583
    },
    {
      "epoch": 0.3916343919442293,
      "grad_norm": 10.911115646362305,
      "learning_rate": 6.76034756696551e-06,
      "loss": 0.1307,
      "step": 7584
    },
    {
      "epoch": 0.3916860315001291,
      "grad_norm": 2.352059841156006,
      "learning_rate": 6.759580643174208e-06,
      "loss": 0.0492,
      "step": 7585
    },
    {
      "epoch": 0.3917376710560289,
      "grad_norm": 8.825244903564453,
      "learning_rate": 6.758813672130992e-06,
      "loss": 0.1475,
      "step": 7586
    },
    {
      "epoch": 0.39178931061192873,
      "grad_norm": 13.667717933654785,
      "learning_rate": 6.758046653856461e-06,
      "loss": 0.2019,
      "step": 7587
    },
    {
      "epoch": 0.39184095016782855,
      "grad_norm": 1.8218239545822144,
      "learning_rate": 6.757279588371213e-06,
      "loss": 0.0281,
      "step": 7588
    },
    {
      "epoch": 0.39189258972372837,
      "grad_norm": 4.36283016204834,
      "learning_rate": 6.756512475695844e-06,
      "loss": 0.0849,
      "step": 7589
    },
    {
      "epoch": 0.3919442292796282,
      "grad_norm": 6.988341808319092,
      "learning_rate": 6.755745315850958e-06,
      "loss": 0.1328,
      "step": 7590
    },
    {
      "epoch": 0.391995868835528,
      "grad_norm": 3.445974826812744,
      "learning_rate": 6.754978108857153e-06,
      "loss": 0.0628,
      "step": 7591
    },
    {
      "epoch": 0.3920475083914278,
      "grad_norm": 4.884437561035156,
      "learning_rate": 6.7542108547350334e-06,
      "loss": 0.0679,
      "step": 7592
    },
    {
      "epoch": 0.39209914794732764,
      "grad_norm": 6.52869176864624,
      "learning_rate": 6.753443553505201e-06,
      "loss": 0.0786,
      "step": 7593
    },
    {
      "epoch": 0.39215078750322746,
      "grad_norm": 3.5760014057159424,
      "learning_rate": 6.7526762051882655e-06,
      "loss": 0.0667,
      "step": 7594
    },
    {
      "epoch": 0.3922024270591273,
      "grad_norm": 3.4678854942321777,
      "learning_rate": 6.751908809804827e-06,
      "loss": 0.0478,
      "step": 7595
    },
    {
      "epoch": 0.3922540666150271,
      "grad_norm": 5.91292142868042,
      "learning_rate": 6.7511413673755e-06,
      "loss": 0.0602,
      "step": 7596
    },
    {
      "epoch": 0.3923057061709269,
      "grad_norm": 4.398767471313477,
      "learning_rate": 6.750373877920887e-06,
      "loss": 0.1078,
      "step": 7597
    },
    {
      "epoch": 0.39235734572682673,
      "grad_norm": 3.4032249450683594,
      "learning_rate": 6.749606341461603e-06,
      "loss": 0.0657,
      "step": 7598
    },
    {
      "epoch": 0.39240898528272655,
      "grad_norm": 1.8751609325408936,
      "learning_rate": 6.748838758018257e-06,
      "loss": 0.0364,
      "step": 7599
    },
    {
      "epoch": 0.39246062483862637,
      "grad_norm": 6.8157854080200195,
      "learning_rate": 6.748071127611463e-06,
      "loss": 0.0587,
      "step": 7600
    },
    {
      "epoch": 0.3925122643945262,
      "grad_norm": 2.0355634689331055,
      "learning_rate": 6.747303450261834e-06,
      "loss": 0.0359,
      "step": 7601
    },
    {
      "epoch": 0.392563903950426,
      "grad_norm": 3.586832046508789,
      "learning_rate": 6.7465357259899854e-06,
      "loss": 0.0556,
      "step": 7602
    },
    {
      "epoch": 0.3926155435063258,
      "grad_norm": 4.350131988525391,
      "learning_rate": 6.745767954816536e-06,
      "loss": 0.0384,
      "step": 7603
    },
    {
      "epoch": 0.39266718306222564,
      "grad_norm": 5.993349075317383,
      "learning_rate": 6.7450001367620995e-06,
      "loss": 0.0403,
      "step": 7604
    },
    {
      "epoch": 0.39271882261812546,
      "grad_norm": 3.3149077892303467,
      "learning_rate": 6.744232271847298e-06,
      "loss": 0.0833,
      "step": 7605
    },
    {
      "epoch": 0.3927704621740253,
      "grad_norm": 13.995813369750977,
      "learning_rate": 6.743464360092751e-06,
      "loss": 0.0939,
      "step": 7606
    },
    {
      "epoch": 0.3928221017299251,
      "grad_norm": 1.4435417652130127,
      "learning_rate": 6.74269640151908e-06,
      "loss": 0.0101,
      "step": 7607
    },
    {
      "epoch": 0.3928737412858249,
      "grad_norm": 2.726951837539673,
      "learning_rate": 6.741928396146906e-06,
      "loss": 0.0399,
      "step": 7608
    },
    {
      "epoch": 0.39292538084172474,
      "grad_norm": 3.5410678386688232,
      "learning_rate": 6.741160343996858e-06,
      "loss": 0.0405,
      "step": 7609
    },
    {
      "epoch": 0.39297702039762455,
      "grad_norm": 3.897367238998413,
      "learning_rate": 6.7403922450895555e-06,
      "loss": 0.0567,
      "step": 7610
    },
    {
      "epoch": 0.3930286599535244,
      "grad_norm": 3.857347011566162,
      "learning_rate": 6.73962409944563e-06,
      "loss": 0.051,
      "step": 7611
    },
    {
      "epoch": 0.3930802995094242,
      "grad_norm": 10.182173728942871,
      "learning_rate": 6.738855907085704e-06,
      "loss": 0.0935,
      "step": 7612
    },
    {
      "epoch": 0.39313193906532407,
      "grad_norm": 2.4893062114715576,
      "learning_rate": 6.738087668030413e-06,
      "loss": 0.0552,
      "step": 7613
    },
    {
      "epoch": 0.3931835786212239,
      "grad_norm": 6.399996757507324,
      "learning_rate": 6.7373193823003825e-06,
      "loss": 0.0703,
      "step": 7614
    },
    {
      "epoch": 0.3932352181771237,
      "grad_norm": 2.4875712394714355,
      "learning_rate": 6.7365510499162455e-06,
      "loss": 0.0209,
      "step": 7615
    },
    {
      "epoch": 0.3932868577330235,
      "grad_norm": 2.2249741554260254,
      "learning_rate": 6.735782670898636e-06,
      "loss": 0.0319,
      "step": 7616
    },
    {
      "epoch": 0.39333849728892334,
      "grad_norm": 2.3305065631866455,
      "learning_rate": 6.7350142452681865e-06,
      "loss": 0.0343,
      "step": 7617
    },
    {
      "epoch": 0.39339013684482316,
      "grad_norm": 4.791745662689209,
      "learning_rate": 6.734245773045534e-06,
      "loss": 0.0808,
      "step": 7618
    },
    {
      "epoch": 0.393441776400723,
      "grad_norm": 4.645274639129639,
      "learning_rate": 6.733477254251312e-06,
      "loss": 0.0592,
      "step": 7619
    },
    {
      "epoch": 0.3934934159566228,
      "grad_norm": 4.5992865562438965,
      "learning_rate": 6.732708688906162e-06,
      "loss": 0.0563,
      "step": 7620
    },
    {
      "epoch": 0.3935450555125226,
      "grad_norm": 2.9848265647888184,
      "learning_rate": 6.7319400770307205e-06,
      "loss": 0.0449,
      "step": 7621
    },
    {
      "epoch": 0.39359669506842243,
      "grad_norm": 5.887706756591797,
      "learning_rate": 6.73117141864563e-06,
      "loss": 0.0456,
      "step": 7622
    },
    {
      "epoch": 0.39364833462432225,
      "grad_norm": 14.004905700683594,
      "learning_rate": 6.730402713771531e-06,
      "loss": 0.0716,
      "step": 7623
    },
    {
      "epoch": 0.39369997418022207,
      "grad_norm": 8.375221252441406,
      "learning_rate": 6.7296339624290655e-06,
      "loss": 0.0622,
      "step": 7624
    },
    {
      "epoch": 0.3937516137361219,
      "grad_norm": 3.6953725814819336,
      "learning_rate": 6.72886516463888e-06,
      "loss": 0.0764,
      "step": 7625
    },
    {
      "epoch": 0.3938032532920217,
      "grad_norm": 1.8351396322250366,
      "learning_rate": 6.728096320421617e-06,
      "loss": 0.0373,
      "step": 7626
    },
    {
      "epoch": 0.3938548928479215,
      "grad_norm": 14.306344032287598,
      "learning_rate": 6.727327429797926e-06,
      "loss": 0.0937,
      "step": 7627
    },
    {
      "epoch": 0.39390653240382134,
      "grad_norm": 3.640151262283325,
      "learning_rate": 6.726558492788451e-06,
      "loss": 0.031,
      "step": 7628
    },
    {
      "epoch": 0.39395817195972116,
      "grad_norm": 9.376660346984863,
      "learning_rate": 6.725789509413845e-06,
      "loss": 0.0735,
      "step": 7629
    },
    {
      "epoch": 0.394009811515621,
      "grad_norm": 2.2434580326080322,
      "learning_rate": 6.725020479694756e-06,
      "loss": 0.0214,
      "step": 7630
    },
    {
      "epoch": 0.3940614510715208,
      "grad_norm": 7.118676662445068,
      "learning_rate": 6.724251403651836e-06,
      "loss": 0.0694,
      "step": 7631
    },
    {
      "epoch": 0.3941130906274206,
      "grad_norm": 4.808758735656738,
      "learning_rate": 6.723482281305739e-06,
      "loss": 0.0475,
      "step": 7632
    },
    {
      "epoch": 0.39416473018332043,
      "grad_norm": 1.3702157735824585,
      "learning_rate": 6.722713112677118e-06,
      "loss": 0.0247,
      "step": 7633
    },
    {
      "epoch": 0.39421636973922025,
      "grad_norm": 1.4795305728912354,
      "learning_rate": 6.721943897786629e-06,
      "loss": 0.0157,
      "step": 7634
    },
    {
      "epoch": 0.39426800929512007,
      "grad_norm": 5.391768932342529,
      "learning_rate": 6.721174636654927e-06,
      "loss": 0.0776,
      "step": 7635
    },
    {
      "epoch": 0.3943196488510199,
      "grad_norm": 2.314520835876465,
      "learning_rate": 6.720405329302673e-06,
      "loss": 0.0367,
      "step": 7636
    },
    {
      "epoch": 0.3943712884069197,
      "grad_norm": 10.83371353149414,
      "learning_rate": 6.719635975750523e-06,
      "loss": 0.0951,
      "step": 7637
    },
    {
      "epoch": 0.3944229279628195,
      "grad_norm": 7.180777072906494,
      "learning_rate": 6.718866576019136e-06,
      "loss": 0.1123,
      "step": 7638
    },
    {
      "epoch": 0.39447456751871934,
      "grad_norm": 8.95555591583252,
      "learning_rate": 6.718097130129177e-06,
      "loss": 0.1057,
      "step": 7639
    },
    {
      "epoch": 0.39452620707461916,
      "grad_norm": 2.1929659843444824,
      "learning_rate": 6.717327638101308e-06,
      "loss": 0.0374,
      "step": 7640
    },
    {
      "epoch": 0.394577846630519,
      "grad_norm": 3.7764151096343994,
      "learning_rate": 6.716558099956193e-06,
      "loss": 0.0789,
      "step": 7641
    },
    {
      "epoch": 0.3946294861864188,
      "grad_norm": 3.435203790664673,
      "learning_rate": 6.715788515714495e-06,
      "loss": 0.048,
      "step": 7642
    },
    {
      "epoch": 0.3946811257423186,
      "grad_norm": 6.87372350692749,
      "learning_rate": 6.715018885396885e-06,
      "loss": 0.0685,
      "step": 7643
    },
    {
      "epoch": 0.39473276529821844,
      "grad_norm": 2.8594424724578857,
      "learning_rate": 6.714249209024026e-06,
      "loss": 0.0649,
      "step": 7644
    },
    {
      "epoch": 0.39478440485411825,
      "grad_norm": 1.537437915802002,
      "learning_rate": 6.713479486616587e-06,
      "loss": 0.0273,
      "step": 7645
    },
    {
      "epoch": 0.39483604441001807,
      "grad_norm": 3.7478740215301514,
      "learning_rate": 6.712709718195244e-06,
      "loss": 0.0506,
      "step": 7646
    },
    {
      "epoch": 0.3948876839659179,
      "grad_norm": 3.777233362197876,
      "learning_rate": 6.7119399037806625e-06,
      "loss": 0.0343,
      "step": 7647
    },
    {
      "epoch": 0.3949393235218177,
      "grad_norm": 3.6993045806884766,
      "learning_rate": 6.711170043393516e-06,
      "loss": 0.0553,
      "step": 7648
    },
    {
      "epoch": 0.3949909630777175,
      "grad_norm": 2.4507033824920654,
      "learning_rate": 6.710400137054481e-06,
      "loss": 0.0228,
      "step": 7649
    },
    {
      "epoch": 0.39504260263361735,
      "grad_norm": 2.887458086013794,
      "learning_rate": 6.709630184784231e-06,
      "loss": 0.0359,
      "step": 7650
    },
    {
      "epoch": 0.39509424218951716,
      "grad_norm": 7.99492073059082,
      "learning_rate": 6.708860186603443e-06,
      "loss": 0.0704,
      "step": 7651
    },
    {
      "epoch": 0.395145881745417,
      "grad_norm": 4.34084415435791,
      "learning_rate": 6.7080901425327916e-06,
      "loss": 0.0802,
      "step": 7652
    },
    {
      "epoch": 0.3951975213013168,
      "grad_norm": 8.996927261352539,
      "learning_rate": 6.707320052592961e-06,
      "loss": 0.0634,
      "step": 7653
    },
    {
      "epoch": 0.3952491608572166,
      "grad_norm": 4.40005350112915,
      "learning_rate": 6.706549916804626e-06,
      "loss": 0.085,
      "step": 7654
    },
    {
      "epoch": 0.39530080041311644,
      "grad_norm": 3.5220260620117188,
      "learning_rate": 6.705779735188471e-06,
      "loss": 0.0575,
      "step": 7655
    },
    {
      "epoch": 0.39535243996901626,
      "grad_norm": 8.683252334594727,
      "learning_rate": 6.705009507765179e-06,
      "loss": 0.0987,
      "step": 7656
    },
    {
      "epoch": 0.3954040795249161,
      "grad_norm": 7.943546295166016,
      "learning_rate": 6.704239234555431e-06,
      "loss": 0.1427,
      "step": 7657
    },
    {
      "epoch": 0.3954557190808159,
      "grad_norm": 8.885808944702148,
      "learning_rate": 6.703468915579915e-06,
      "loss": 0.0941,
      "step": 7658
    },
    {
      "epoch": 0.3955073586367157,
      "grad_norm": 7.276912212371826,
      "learning_rate": 6.7026985508593135e-06,
      "loss": 0.0829,
      "step": 7659
    },
    {
      "epoch": 0.39555899819261553,
      "grad_norm": 11.73534870147705,
      "learning_rate": 6.701928140414318e-06,
      "loss": 0.1179,
      "step": 7660
    },
    {
      "epoch": 0.39561063774851535,
      "grad_norm": 6.655685901641846,
      "learning_rate": 6.701157684265613e-06,
      "loss": 0.064,
      "step": 7661
    },
    {
      "epoch": 0.39566227730441517,
      "grad_norm": 5.29547119140625,
      "learning_rate": 6.700387182433893e-06,
      "loss": 0.0657,
      "step": 7662
    },
    {
      "epoch": 0.395713916860315,
      "grad_norm": 4.461860179901123,
      "learning_rate": 6.699616634939846e-06,
      "loss": 0.0362,
      "step": 7663
    },
    {
      "epoch": 0.3957655564162148,
      "grad_norm": 15.38684368133545,
      "learning_rate": 6.698846041804164e-06,
      "loss": 0.062,
      "step": 7664
    },
    {
      "epoch": 0.3958171959721146,
      "grad_norm": 5.325438022613525,
      "learning_rate": 6.698075403047544e-06,
      "loss": 0.1413,
      "step": 7665
    },
    {
      "epoch": 0.39586883552801444,
      "grad_norm": 15.84714412689209,
      "learning_rate": 6.6973047186906755e-06,
      "loss": 0.1229,
      "step": 7666
    },
    {
      "epoch": 0.39592047508391426,
      "grad_norm": 6.240447521209717,
      "learning_rate": 6.69653398875426e-06,
      "loss": 0.049,
      "step": 7667
    },
    {
      "epoch": 0.3959721146398141,
      "grad_norm": 3.0764310359954834,
      "learning_rate": 6.69576321325899e-06,
      "loss": 0.0524,
      "step": 7668
    },
    {
      "epoch": 0.3960237541957139,
      "grad_norm": 2.538921594619751,
      "learning_rate": 6.694992392225569e-06,
      "loss": 0.0554,
      "step": 7669
    },
    {
      "epoch": 0.3960753937516137,
      "grad_norm": 3.611863851547241,
      "learning_rate": 6.6942215256746925e-06,
      "loss": 0.0667,
      "step": 7670
    },
    {
      "epoch": 0.39612703330751353,
      "grad_norm": 9.600946426391602,
      "learning_rate": 6.693450613627062e-06,
      "loss": 0.0828,
      "step": 7671
    },
    {
      "epoch": 0.39617867286341335,
      "grad_norm": 2.2761306762695312,
      "learning_rate": 6.692679656103381e-06,
      "loss": 0.041,
      "step": 7672
    },
    {
      "epoch": 0.39623031241931317,
      "grad_norm": 8.98764705657959,
      "learning_rate": 6.691908653124351e-06,
      "loss": 0.1092,
      "step": 7673
    },
    {
      "epoch": 0.396281951975213,
      "grad_norm": 2.056619644165039,
      "learning_rate": 6.69113760471068e-06,
      "loss": 0.0276,
      "step": 7674
    },
    {
      "epoch": 0.3963335915311128,
      "grad_norm": 3.265662431716919,
      "learning_rate": 6.690366510883071e-06,
      "loss": 0.0625,
      "step": 7675
    },
    {
      "epoch": 0.3963852310870127,
      "grad_norm": 2.857832908630371,
      "learning_rate": 6.689595371662231e-06,
      "loss": 0.0199,
      "step": 7676
    },
    {
      "epoch": 0.3964368706429125,
      "grad_norm": 3.721489429473877,
      "learning_rate": 6.68882418706887e-06,
      "loss": 0.0589,
      "step": 7677
    },
    {
      "epoch": 0.3964885101988123,
      "grad_norm": 3.6401219367980957,
      "learning_rate": 6.6880529571236955e-06,
      "loss": 0.0772,
      "step": 7678
    },
    {
      "epoch": 0.39654014975471213,
      "grad_norm": 2.7277398109436035,
      "learning_rate": 6.687281681847419e-06,
      "loss": 0.064,
      "step": 7679
    },
    {
      "epoch": 0.39659178931061195,
      "grad_norm": 5.669684410095215,
      "learning_rate": 6.6865103612607515e-06,
      "loss": 0.0655,
      "step": 7680
    },
    {
      "epoch": 0.39664342886651177,
      "grad_norm": 3.482053279876709,
      "learning_rate": 6.685738995384408e-06,
      "loss": 0.0747,
      "step": 7681
    },
    {
      "epoch": 0.3966950684224116,
      "grad_norm": 17.041086196899414,
      "learning_rate": 6.684967584239102e-06,
      "loss": 0.0968,
      "step": 7682
    },
    {
      "epoch": 0.3967467079783114,
      "grad_norm": 4.569214344024658,
      "learning_rate": 6.684196127845547e-06,
      "loss": 0.0453,
      "step": 7683
    },
    {
      "epoch": 0.3967983475342112,
      "grad_norm": 4.127387523651123,
      "learning_rate": 6.683424626224464e-06,
      "loss": 0.0486,
      "step": 7684
    },
    {
      "epoch": 0.39684998709011104,
      "grad_norm": 8.00877857208252,
      "learning_rate": 6.682653079396566e-06,
      "loss": 0.0692,
      "step": 7685
    },
    {
      "epoch": 0.39690162664601086,
      "grad_norm": 2.98148512840271,
      "learning_rate": 6.6818814873825775e-06,
      "loss": 0.0613,
      "step": 7686
    },
    {
      "epoch": 0.3969532662019107,
      "grad_norm": 6.975001811981201,
      "learning_rate": 6.681109850203214e-06,
      "loss": 0.0893,
      "step": 7687
    },
    {
      "epoch": 0.3970049057578105,
      "grad_norm": 18.61650276184082,
      "learning_rate": 6.6803381678792e-06,
      "loss": 0.0969,
      "step": 7688
    },
    {
      "epoch": 0.3970565453137103,
      "grad_norm": 4.276468276977539,
      "learning_rate": 6.679566440431257e-06,
      "loss": 0.0711,
      "step": 7689
    },
    {
      "epoch": 0.39710818486961014,
      "grad_norm": 3.9062113761901855,
      "learning_rate": 6.678794667880111e-06,
      "loss": 0.0592,
      "step": 7690
    },
    {
      "epoch": 0.39715982442550996,
      "grad_norm": 2.9512429237365723,
      "learning_rate": 6.678022850246485e-06,
      "loss": 0.0539,
      "step": 7691
    },
    {
      "epoch": 0.3972114639814098,
      "grad_norm": 5.29226016998291,
      "learning_rate": 6.677250987551105e-06,
      "loss": 0.0649,
      "step": 7692
    },
    {
      "epoch": 0.3972631035373096,
      "grad_norm": 9.893980979919434,
      "learning_rate": 6.676479079814702e-06,
      "loss": 0.1145,
      "step": 7693
    },
    {
      "epoch": 0.3973147430932094,
      "grad_norm": 3.3288075923919678,
      "learning_rate": 6.675707127058e-06,
      "loss": 0.04,
      "step": 7694
    },
    {
      "epoch": 0.39736638264910923,
      "grad_norm": 2.906282663345337,
      "learning_rate": 6.674935129301735e-06,
      "loss": 0.0701,
      "step": 7695
    },
    {
      "epoch": 0.39741802220500905,
      "grad_norm": 7.216241359710693,
      "learning_rate": 6.6741630865666326e-06,
      "loss": 0.0824,
      "step": 7696
    },
    {
      "epoch": 0.39746966176090887,
      "grad_norm": 7.035358905792236,
      "learning_rate": 6.673390998873429e-06,
      "loss": 0.0777,
      "step": 7697
    },
    {
      "epoch": 0.3975213013168087,
      "grad_norm": 19.25173568725586,
      "learning_rate": 6.672618866242857e-06,
      "loss": 0.1003,
      "step": 7698
    },
    {
      "epoch": 0.3975729408727085,
      "grad_norm": 9.010647773742676,
      "learning_rate": 6.67184668869565e-06,
      "loss": 0.0811,
      "step": 7699
    },
    {
      "epoch": 0.3976245804286083,
      "grad_norm": 3.3853049278259277,
      "learning_rate": 6.671074466252547e-06,
      "loss": 0.0555,
      "step": 7700
    },
    {
      "epoch": 0.39767621998450814,
      "grad_norm": 2.6749887466430664,
      "learning_rate": 6.670302198934282e-06,
      "loss": 0.0403,
      "step": 7701
    },
    {
      "epoch": 0.39772785954040796,
      "grad_norm": 4.95787239074707,
      "learning_rate": 6.669529886761597e-06,
      "loss": 0.0534,
      "step": 7702
    },
    {
      "epoch": 0.3977794990963078,
      "grad_norm": 3.8321714401245117,
      "learning_rate": 6.668757529755229e-06,
      "loss": 0.0484,
      "step": 7703
    },
    {
      "epoch": 0.3978311386522076,
      "grad_norm": 2.6489651203155518,
      "learning_rate": 6.667985127935918e-06,
      "loss": 0.0301,
      "step": 7704
    },
    {
      "epoch": 0.3978827782081074,
      "grad_norm": 2.0747158527374268,
      "learning_rate": 6.667212681324411e-06,
      "loss": 0.0342,
      "step": 7705
    },
    {
      "epoch": 0.39793441776400723,
      "grad_norm": 7.551504135131836,
      "learning_rate": 6.666440189941447e-06,
      "loss": 0.0749,
      "step": 7706
    },
    {
      "epoch": 0.39798605731990705,
      "grad_norm": 6.194829940795898,
      "learning_rate": 6.665667653807771e-06,
      "loss": 0.0581,
      "step": 7707
    },
    {
      "epoch": 0.39803769687580687,
      "grad_norm": 20.362390518188477,
      "learning_rate": 6.664895072944131e-06,
      "loss": 0.2724,
      "step": 7708
    },
    {
      "epoch": 0.3980893364317067,
      "grad_norm": 3.524958610534668,
      "learning_rate": 6.6641224473712716e-06,
      "loss": 0.0288,
      "step": 7709
    },
    {
      "epoch": 0.3981409759876065,
      "grad_norm": 7.07900333404541,
      "learning_rate": 6.663349777109943e-06,
      "loss": 0.1462,
      "step": 7710
    },
    {
      "epoch": 0.3981926155435063,
      "grad_norm": 6.425405979156494,
      "learning_rate": 6.662577062180893e-06,
      "loss": 0.0726,
      "step": 7711
    },
    {
      "epoch": 0.39824425509940614,
      "grad_norm": 10.662607192993164,
      "learning_rate": 6.661804302604871e-06,
      "loss": 0.0914,
      "step": 7712
    },
    {
      "epoch": 0.39829589465530596,
      "grad_norm": 12.464543342590332,
      "learning_rate": 6.661031498402633e-06,
      "loss": 0.0579,
      "step": 7713
    },
    {
      "epoch": 0.3983475342112058,
      "grad_norm": 2.6868820190429688,
      "learning_rate": 6.660258649594927e-06,
      "loss": 0.0461,
      "step": 7714
    },
    {
      "epoch": 0.3983991737671056,
      "grad_norm": 8.163639068603516,
      "learning_rate": 6.65948575620251e-06,
      "loss": 0.0973,
      "step": 7715
    },
    {
      "epoch": 0.3984508133230054,
      "grad_norm": 7.549313068389893,
      "learning_rate": 6.658712818246138e-06,
      "loss": 0.0909,
      "step": 7716
    },
    {
      "epoch": 0.39850245287890523,
      "grad_norm": 7.133005142211914,
      "learning_rate": 6.657939835746566e-06,
      "loss": 0.0604,
      "step": 7717
    },
    {
      "epoch": 0.39855409243480505,
      "grad_norm": 1.9500387907028198,
      "learning_rate": 6.657166808724551e-06,
      "loss": 0.0307,
      "step": 7718
    },
    {
      "epoch": 0.39860573199070487,
      "grad_norm": 5.887158393859863,
      "learning_rate": 6.6563937372008535e-06,
      "loss": 0.0623,
      "step": 7719
    },
    {
      "epoch": 0.3986573715466047,
      "grad_norm": 3.6039516925811768,
      "learning_rate": 6.6556206211962326e-06,
      "loss": 0.0458,
      "step": 7720
    },
    {
      "epoch": 0.3987090111025045,
      "grad_norm": 3.1899213790893555,
      "learning_rate": 6.65484746073145e-06,
      "loss": 0.0603,
      "step": 7721
    },
    {
      "epoch": 0.3987606506584043,
      "grad_norm": 6.641463756561279,
      "learning_rate": 6.65407425582727e-06,
      "loss": 0.0817,
      "step": 7722
    },
    {
      "epoch": 0.39881229021430414,
      "grad_norm": 8.289922714233398,
      "learning_rate": 6.653301006504454e-06,
      "loss": 0.0959,
      "step": 7723
    },
    {
      "epoch": 0.39886392977020396,
      "grad_norm": 6.470686912536621,
      "learning_rate": 6.652527712783767e-06,
      "loss": 0.125,
      "step": 7724
    },
    {
      "epoch": 0.3989155693261038,
      "grad_norm": 6.824835777282715,
      "learning_rate": 6.651754374685973e-06,
      "loss": 0.0528,
      "step": 7725
    },
    {
      "epoch": 0.3989672088820036,
      "grad_norm": 2.848121166229248,
      "learning_rate": 6.650980992231846e-06,
      "loss": 0.0473,
      "step": 7726
    },
    {
      "epoch": 0.3990188484379034,
      "grad_norm": 11.562200546264648,
      "learning_rate": 6.650207565442148e-06,
      "loss": 0.0502,
      "step": 7727
    },
    {
      "epoch": 0.39907048799380324,
      "grad_norm": 2.8495850563049316,
      "learning_rate": 6.649434094337652e-06,
      "loss": 0.0607,
      "step": 7728
    },
    {
      "epoch": 0.39912212754970305,
      "grad_norm": 10.315905570983887,
      "learning_rate": 6.648660578939128e-06,
      "loss": 0.0843,
      "step": 7729
    },
    {
      "epoch": 0.3991737671056029,
      "grad_norm": 3.3519511222839355,
      "learning_rate": 6.647887019267347e-06,
      "loss": 0.0589,
      "step": 7730
    },
    {
      "epoch": 0.3992254066615027,
      "grad_norm": 3.893627643585205,
      "learning_rate": 6.6471134153430826e-06,
      "loss": 0.0527,
      "step": 7731
    },
    {
      "epoch": 0.3992770462174025,
      "grad_norm": 3.1956377029418945,
      "learning_rate": 6.6463397671871114e-06,
      "loss": 0.0568,
      "step": 7732
    },
    {
      "epoch": 0.3993286857733023,
      "grad_norm": 8.80396842956543,
      "learning_rate": 6.6455660748202065e-06,
      "loss": 0.0417,
      "step": 7733
    },
    {
      "epoch": 0.39938032532920215,
      "grad_norm": 11.88211727142334,
      "learning_rate": 6.6447923382631454e-06,
      "loss": 0.0886,
      "step": 7734
    },
    {
      "epoch": 0.39943196488510196,
      "grad_norm": 4.0357232093811035,
      "learning_rate": 6.644018557536707e-06,
      "loss": 0.0524,
      "step": 7735
    },
    {
      "epoch": 0.3994836044410018,
      "grad_norm": 10.757272720336914,
      "learning_rate": 6.6432447326616685e-06,
      "loss": 0.075,
      "step": 7736
    },
    {
      "epoch": 0.3995352439969016,
      "grad_norm": 2.8609206676483154,
      "learning_rate": 6.642470863658812e-06,
      "loss": 0.042,
      "step": 7737
    },
    {
      "epoch": 0.3995868835528014,
      "grad_norm": 10.878232955932617,
      "learning_rate": 6.641696950548919e-06,
      "loss": 0.1398,
      "step": 7738
    },
    {
      "epoch": 0.39963852310870124,
      "grad_norm": 7.3918962478637695,
      "learning_rate": 6.640922993352771e-06,
      "loss": 0.105,
      "step": 7739
    },
    {
      "epoch": 0.3996901626646011,
      "grad_norm": 4.63099479675293,
      "learning_rate": 6.640148992091152e-06,
      "loss": 0.0818,
      "step": 7740
    },
    {
      "epoch": 0.39974180222050093,
      "grad_norm": 8.574441909790039,
      "learning_rate": 6.639374946784849e-06,
      "loss": 0.1128,
      "step": 7741
    },
    {
      "epoch": 0.39979344177640075,
      "grad_norm": 9.16493034362793,
      "learning_rate": 6.638600857454645e-06,
      "loss": 0.1089,
      "step": 7742
    },
    {
      "epoch": 0.39984508133230057,
      "grad_norm": 14.537910461425781,
      "learning_rate": 6.6378267241213326e-06,
      "loss": 0.1511,
      "step": 7743
    },
    {
      "epoch": 0.3998967208882004,
      "grad_norm": 5.897097587585449,
      "learning_rate": 6.637052546805696e-06,
      "loss": 0.0542,
      "step": 7744
    },
    {
      "epoch": 0.3999483604441002,
      "grad_norm": 4.297042369842529,
      "learning_rate": 6.636278325528524e-06,
      "loss": 0.0617,
      "step": 7745
    },
    {
      "epoch": 0.4,
      "grad_norm": 6.875668525695801,
      "learning_rate": 6.635504060310612e-06,
      "loss": 0.1128,
      "step": 7746
    },
    {
      "epoch": 0.40005163955589984,
      "grad_norm": 1.6854848861694336,
      "learning_rate": 6.63472975117275e-06,
      "loss": 0.0147,
      "step": 7747
    },
    {
      "epoch": 0.40010327911179966,
      "grad_norm": 2.3037002086639404,
      "learning_rate": 6.633955398135731e-06,
      "loss": 0.0544,
      "step": 7748
    },
    {
      "epoch": 0.40010327911179966,
      "eval_loss": 0.066283218562603,
      "eval_runtime": 34.1817,
      "eval_samples_per_second": 45.785,
      "eval_steps_per_second": 5.734,
      "step": 7748
    },
    {
      "epoch": 0.4001549186676995,
      "grad_norm": 1.7071576118469238,
      "learning_rate": 6.63318100122035e-06,
      "loss": 0.0403,
      "step": 7749
    },
    {
      "epoch": 0.4002065582235993,
      "grad_norm": 2.7193140983581543,
      "learning_rate": 6.632406560447403e-06,
      "loss": 0.0353,
      "step": 7750
    },
    {
      "epoch": 0.4002581977794991,
      "grad_norm": 4.1848063468933105,
      "learning_rate": 6.631632075837685e-06,
      "loss": 0.0661,
      "step": 7751
    },
    {
      "epoch": 0.40030983733539893,
      "grad_norm": 4.677114486694336,
      "learning_rate": 6.630857547411997e-06,
      "loss": 0.0747,
      "step": 7752
    },
    {
      "epoch": 0.40036147689129875,
      "grad_norm": 15.805781364440918,
      "learning_rate": 6.630082975191136e-06,
      "loss": 0.1667,
      "step": 7753
    },
    {
      "epoch": 0.40041311644719857,
      "grad_norm": 4.487678527832031,
      "learning_rate": 6.629308359195905e-06,
      "loss": 0.0892,
      "step": 7754
    },
    {
      "epoch": 0.4004647560030984,
      "grad_norm": 2.056459665298462,
      "learning_rate": 6.628533699447103e-06,
      "loss": 0.0373,
      "step": 7755
    },
    {
      "epoch": 0.4005163955589982,
      "grad_norm": 2.6692190170288086,
      "learning_rate": 6.627758995965533e-06,
      "loss": 0.0656,
      "step": 7756
    },
    {
      "epoch": 0.400568035114898,
      "grad_norm": 2.1292688846588135,
      "learning_rate": 6.626984248772e-06,
      "loss": 0.0527,
      "step": 7757
    },
    {
      "epoch": 0.40061967467079784,
      "grad_norm": 3.3814103603363037,
      "learning_rate": 6.626209457887309e-06,
      "loss": 0.0831,
      "step": 7758
    },
    {
      "epoch": 0.40067131422669766,
      "grad_norm": 8.919849395751953,
      "learning_rate": 6.625434623332266e-06,
      "loss": 0.2285,
      "step": 7759
    },
    {
      "epoch": 0.4007229537825975,
      "grad_norm": 3.168031930923462,
      "learning_rate": 6.624659745127676e-06,
      "loss": 0.073,
      "step": 7760
    },
    {
      "epoch": 0.4007745933384973,
      "grad_norm": 2.2920849323272705,
      "learning_rate": 6.623884823294353e-06,
      "loss": 0.0466,
      "step": 7761
    },
    {
      "epoch": 0.4008262328943971,
      "grad_norm": 4.948526859283447,
      "learning_rate": 6.623109857853103e-06,
      "loss": 0.0297,
      "step": 7762
    },
    {
      "epoch": 0.40087787245029693,
      "grad_norm": 3.16910719871521,
      "learning_rate": 6.6223348488247385e-06,
      "loss": 0.0404,
      "step": 7763
    },
    {
      "epoch": 0.40092951200619675,
      "grad_norm": 2.7950539588928223,
      "learning_rate": 6.621559796230071e-06,
      "loss": 0.0335,
      "step": 7764
    },
    {
      "epoch": 0.40098115156209657,
      "grad_norm": 2.887667179107666,
      "learning_rate": 6.620784700089913e-06,
      "loss": 0.0768,
      "step": 7765
    },
    {
      "epoch": 0.4010327911179964,
      "grad_norm": 4.969449520111084,
      "learning_rate": 6.62000956042508e-06,
      "loss": 0.0784,
      "step": 7766
    },
    {
      "epoch": 0.4010844306738962,
      "grad_norm": 1.828099012374878,
      "learning_rate": 6.619234377256387e-06,
      "loss": 0.0513,
      "step": 7767
    },
    {
      "epoch": 0.401136070229796,
      "grad_norm": 1.8938325643539429,
      "learning_rate": 6.6184591506046524e-06,
      "loss": 0.0372,
      "step": 7768
    },
    {
      "epoch": 0.40118770978569585,
      "grad_norm": 7.022898197174072,
      "learning_rate": 6.617683880490693e-06,
      "loss": 0.0631,
      "step": 7769
    },
    {
      "epoch": 0.40123934934159566,
      "grad_norm": 6.435731410980225,
      "learning_rate": 6.616908566935327e-06,
      "loss": 0.0663,
      "step": 7770
    },
    {
      "epoch": 0.4012909888974955,
      "grad_norm": 2.2650020122528076,
      "learning_rate": 6.616133209959376e-06,
      "loss": 0.0669,
      "step": 7771
    },
    {
      "epoch": 0.4013426284533953,
      "grad_norm": 2.4583523273468018,
      "learning_rate": 6.615357809583663e-06,
      "loss": 0.0524,
      "step": 7772
    },
    {
      "epoch": 0.4013942680092951,
      "grad_norm": 2.5939295291900635,
      "learning_rate": 6.614582365829008e-06,
      "loss": 0.0767,
      "step": 7773
    },
    {
      "epoch": 0.40144590756519494,
      "grad_norm": 3.4961719512939453,
      "learning_rate": 6.613806878716236e-06,
      "loss": 0.0776,
      "step": 7774
    },
    {
      "epoch": 0.40149754712109476,
      "grad_norm": 6.732507228851318,
      "learning_rate": 6.6130313482661715e-06,
      "loss": 0.1082,
      "step": 7775
    },
    {
      "epoch": 0.4015491866769946,
      "grad_norm": 4.338997840881348,
      "learning_rate": 6.612255774499642e-06,
      "loss": 0.0666,
      "step": 7776
    },
    {
      "epoch": 0.4016008262328944,
      "grad_norm": 4.136804103851318,
      "learning_rate": 6.611480157437472e-06,
      "loss": 0.0728,
      "step": 7777
    },
    {
      "epoch": 0.4016524657887942,
      "grad_norm": 4.745784759521484,
      "learning_rate": 6.610704497100493e-06,
      "loss": 0.0465,
      "step": 7778
    },
    {
      "epoch": 0.40170410534469403,
      "grad_norm": 8.998038291931152,
      "learning_rate": 6.609928793509532e-06,
      "loss": 0.1209,
      "step": 7779
    },
    {
      "epoch": 0.40175574490059385,
      "grad_norm": 3.264648199081421,
      "learning_rate": 6.609153046685423e-06,
      "loss": 0.0726,
      "step": 7780
    },
    {
      "epoch": 0.40180738445649367,
      "grad_norm": 2.872791290283203,
      "learning_rate": 6.608377256648995e-06,
      "loss": 0.0481,
      "step": 7781
    },
    {
      "epoch": 0.4018590240123935,
      "grad_norm": 4.603013038635254,
      "learning_rate": 6.607601423421082e-06,
      "loss": 0.0476,
      "step": 7782
    },
    {
      "epoch": 0.4019106635682933,
      "grad_norm": 2.7459568977355957,
      "learning_rate": 6.60682554702252e-06,
      "loss": 0.0283,
      "step": 7783
    },
    {
      "epoch": 0.4019623031241931,
      "grad_norm": 3.673192024230957,
      "learning_rate": 6.606049627474141e-06,
      "loss": 0.0737,
      "step": 7784
    },
    {
      "epoch": 0.40201394268009294,
      "grad_norm": 3.374666213989258,
      "learning_rate": 6.6052736647967855e-06,
      "loss": 0.0917,
      "step": 7785
    },
    {
      "epoch": 0.40206558223599276,
      "grad_norm": 14.565732955932617,
      "learning_rate": 6.604497659011288e-06,
      "loss": 0.0848,
      "step": 7786
    },
    {
      "epoch": 0.4021172217918926,
      "grad_norm": 2.1558592319488525,
      "learning_rate": 6.603721610138489e-06,
      "loss": 0.0235,
      "step": 7787
    },
    {
      "epoch": 0.4021688613477924,
      "grad_norm": 5.4347100257873535,
      "learning_rate": 6.602945518199228e-06,
      "loss": 0.0567,
      "step": 7788
    },
    {
      "epoch": 0.4022205009036922,
      "grad_norm": 2.8944690227508545,
      "learning_rate": 6.602169383214348e-06,
      "loss": 0.0706,
      "step": 7789
    },
    {
      "epoch": 0.40227214045959203,
      "grad_norm": 6.3298115730285645,
      "learning_rate": 6.6013932052046896e-06,
      "loss": 0.0863,
      "step": 7790
    },
    {
      "epoch": 0.40232378001549185,
      "grad_norm": 6.127263069152832,
      "learning_rate": 6.600616984191096e-06,
      "loss": 0.1242,
      "step": 7791
    },
    {
      "epoch": 0.40237541957139167,
      "grad_norm": 3.970487356185913,
      "learning_rate": 6.599840720194415e-06,
      "loss": 0.0554,
      "step": 7792
    },
    {
      "epoch": 0.4024270591272915,
      "grad_norm": 3.01509428024292,
      "learning_rate": 6.599064413235487e-06,
      "loss": 0.0324,
      "step": 7793
    },
    {
      "epoch": 0.4024786986831913,
      "grad_norm": 11.09738540649414,
      "learning_rate": 6.598288063335163e-06,
      "loss": 0.1212,
      "step": 7794
    },
    {
      "epoch": 0.4025303382390911,
      "grad_norm": 8.977396011352539,
      "learning_rate": 6.597511670514291e-06,
      "loss": 0.179,
      "step": 7795
    },
    {
      "epoch": 0.40258197779499094,
      "grad_norm": 17.58868408203125,
      "learning_rate": 6.5967352347937186e-06,
      "loss": 0.14,
      "step": 7796
    },
    {
      "epoch": 0.40263361735089076,
      "grad_norm": 14.218603134155273,
      "learning_rate": 6.5959587561942974e-06,
      "loss": 0.1167,
      "step": 7797
    },
    {
      "epoch": 0.4026852569067906,
      "grad_norm": 5.134237289428711,
      "learning_rate": 6.5951822347368785e-06,
      "loss": 0.0527,
      "step": 7798
    },
    {
      "epoch": 0.4027368964626904,
      "grad_norm": 2.478010892868042,
      "learning_rate": 6.594405670442317e-06,
      "loss": 0.0307,
      "step": 7799
    },
    {
      "epoch": 0.4027885360185902,
      "grad_norm": 4.76471471786499,
      "learning_rate": 6.593629063331462e-06,
      "loss": 0.0474,
      "step": 7800
    },
    {
      "epoch": 0.40284017557449003,
      "grad_norm": 8.494527816772461,
      "learning_rate": 6.592852413425173e-06,
      "loss": 0.1028,
      "step": 7801
    },
    {
      "epoch": 0.40289181513038985,
      "grad_norm": 1.5315653085708618,
      "learning_rate": 6.592075720744304e-06,
      "loss": 0.012,
      "step": 7802
    },
    {
      "epoch": 0.40294345468628967,
      "grad_norm": 2.427556037902832,
      "learning_rate": 6.591298985309713e-06,
      "loss": 0.0473,
      "step": 7803
    },
    {
      "epoch": 0.40299509424218954,
      "grad_norm": 3.1392345428466797,
      "learning_rate": 6.590522207142259e-06,
      "loss": 0.0399,
      "step": 7804
    },
    {
      "epoch": 0.40304673379808936,
      "grad_norm": 4.465517997741699,
      "learning_rate": 6.5897453862628e-06,
      "loss": 0.0656,
      "step": 7805
    },
    {
      "epoch": 0.4030983733539892,
      "grad_norm": 8.355802536010742,
      "learning_rate": 6.5889685226921994e-06,
      "loss": 0.0901,
      "step": 7806
    },
    {
      "epoch": 0.403150012909889,
      "grad_norm": 2.621584415435791,
      "learning_rate": 6.588191616451317e-06,
      "loss": 0.0296,
      "step": 7807
    },
    {
      "epoch": 0.4032016524657888,
      "grad_norm": 3.979525327682495,
      "learning_rate": 6.587414667561016e-06,
      "loss": 0.0472,
      "step": 7808
    },
    {
      "epoch": 0.40325329202168864,
      "grad_norm": 14.361175537109375,
      "learning_rate": 6.586637676042163e-06,
      "loss": 0.0841,
      "step": 7809
    },
    {
      "epoch": 0.40330493157758845,
      "grad_norm": 25.30734634399414,
      "learning_rate": 6.58586064191562e-06,
      "loss": 0.0647,
      "step": 7810
    },
    {
      "epoch": 0.4033565711334883,
      "grad_norm": 4.857710838317871,
      "learning_rate": 6.585083565202256e-06,
      "loss": 0.0455,
      "step": 7811
    },
    {
      "epoch": 0.4034082106893881,
      "grad_norm": 2.8449926376342773,
      "learning_rate": 6.5843064459229375e-06,
      "loss": 0.025,
      "step": 7812
    },
    {
      "epoch": 0.4034598502452879,
      "grad_norm": 11.401199340820312,
      "learning_rate": 6.583529284098533e-06,
      "loss": 0.0535,
      "step": 7813
    },
    {
      "epoch": 0.40351148980118773,
      "grad_norm": 8.616742134094238,
      "learning_rate": 6.582752079749914e-06,
      "loss": 0.088,
      "step": 7814
    },
    {
      "epoch": 0.40356312935708755,
      "grad_norm": 3.211503028869629,
      "learning_rate": 6.581974832897951e-06,
      "loss": 0.0334,
      "step": 7815
    },
    {
      "epoch": 0.40361476891298737,
      "grad_norm": 7.694411754608154,
      "learning_rate": 6.581197543563515e-06,
      "loss": 0.1048,
      "step": 7816
    },
    {
      "epoch": 0.4036664084688872,
      "grad_norm": 4.7496161460876465,
      "learning_rate": 6.58042021176748e-06,
      "loss": 0.0383,
      "step": 7817
    },
    {
      "epoch": 0.403718048024787,
      "grad_norm": 6.878429889678955,
      "learning_rate": 6.579642837530723e-06,
      "loss": 0.0845,
      "step": 7818
    },
    {
      "epoch": 0.4037696875806868,
      "grad_norm": 5.720396518707275,
      "learning_rate": 6.578865420874116e-06,
      "loss": 0.0532,
      "step": 7819
    },
    {
      "epoch": 0.40382132713658664,
      "grad_norm": 2.1392979621887207,
      "learning_rate": 6.5780879618185375e-06,
      "loss": 0.0342,
      "step": 7820
    },
    {
      "epoch": 0.40387296669248646,
      "grad_norm": 8.481103897094727,
      "learning_rate": 6.577310460384865e-06,
      "loss": 0.0685,
      "step": 7821
    },
    {
      "epoch": 0.4039246062483863,
      "grad_norm": 6.5702714920043945,
      "learning_rate": 6.5765329165939785e-06,
      "loss": 0.1411,
      "step": 7822
    },
    {
      "epoch": 0.4039762458042861,
      "grad_norm": 12.207686424255371,
      "learning_rate": 6.575755330466758e-06,
      "loss": 0.163,
      "step": 7823
    },
    {
      "epoch": 0.4040278853601859,
      "grad_norm": 5.178786277770996,
      "learning_rate": 6.574977702024082e-06,
      "loss": 0.0766,
      "step": 7824
    },
    {
      "epoch": 0.40407952491608573,
      "grad_norm": 3.484361171722412,
      "learning_rate": 6.5742000312868395e-06,
      "loss": 0.036,
      "step": 7825
    },
    {
      "epoch": 0.40413116447198555,
      "grad_norm": 18.52550506591797,
      "learning_rate": 6.573422318275906e-06,
      "loss": 0.163,
      "step": 7826
    },
    {
      "epoch": 0.40418280402788537,
      "grad_norm": 4.899090766906738,
      "learning_rate": 6.572644563012173e-06,
      "loss": 0.0459,
      "step": 7827
    },
    {
      "epoch": 0.4042344435837852,
      "grad_norm": 4.191385269165039,
      "learning_rate": 6.5718667655165226e-06,
      "loss": 0.0624,
      "step": 7828
    },
    {
      "epoch": 0.404286083139685,
      "grad_norm": 7.618969917297363,
      "learning_rate": 6.571088925809843e-06,
      "loss": 0.1117,
      "step": 7829
    },
    {
      "epoch": 0.4043377226955848,
      "grad_norm": 18.20553207397461,
      "learning_rate": 6.570311043913023e-06,
      "loss": 0.0997,
      "step": 7830
    },
    {
      "epoch": 0.40438936225148464,
      "grad_norm": 3.472386121749878,
      "learning_rate": 6.569533119846951e-06,
      "loss": 0.0496,
      "step": 7831
    },
    {
      "epoch": 0.40444100180738446,
      "grad_norm": 12.908809661865234,
      "learning_rate": 6.5687551536325176e-06,
      "loss": 0.0719,
      "step": 7832
    },
    {
      "epoch": 0.4044926413632843,
      "grad_norm": 2.311947822570801,
      "learning_rate": 6.567977145290614e-06,
      "loss": 0.0325,
      "step": 7833
    },
    {
      "epoch": 0.4045442809191841,
      "grad_norm": 3.3581106662750244,
      "learning_rate": 6.567199094842134e-06,
      "loss": 0.0451,
      "step": 7834
    },
    {
      "epoch": 0.4045959204750839,
      "grad_norm": 4.0370306968688965,
      "learning_rate": 6.5664210023079686e-06,
      "loss": 0.0725,
      "step": 7835
    },
    {
      "epoch": 0.40464756003098373,
      "grad_norm": 5.214371204376221,
      "learning_rate": 6.5656428677090174e-06,
      "loss": 0.0721,
      "step": 7836
    },
    {
      "epoch": 0.40469919958688355,
      "grad_norm": 10.528922080993652,
      "learning_rate": 6.564864691066174e-06,
      "loss": 0.1674,
      "step": 7837
    },
    {
      "epoch": 0.40475083914278337,
      "grad_norm": 3.8624229431152344,
      "learning_rate": 6.564086472400335e-06,
      "loss": 0.067,
      "step": 7838
    },
    {
      "epoch": 0.4048024786986832,
      "grad_norm": 3.1554157733917236,
      "learning_rate": 6.563308211732399e-06,
      "loss": 0.0516,
      "step": 7839
    },
    {
      "epoch": 0.404854118254583,
      "grad_norm": 3.0566093921661377,
      "learning_rate": 6.562529909083266e-06,
      "loss": 0.0506,
      "step": 7840
    },
    {
      "epoch": 0.4049057578104828,
      "grad_norm": 9.324934005737305,
      "learning_rate": 6.5617515644738375e-06,
      "loss": 0.0453,
      "step": 7841
    },
    {
      "epoch": 0.40495739736638264,
      "grad_norm": 2.7426278591156006,
      "learning_rate": 6.560973177925012e-06,
      "loss": 0.0349,
      "step": 7842
    },
    {
      "epoch": 0.40500903692228246,
      "grad_norm": 6.2509541511535645,
      "learning_rate": 6.560194749457697e-06,
      "loss": 0.0932,
      "step": 7843
    },
    {
      "epoch": 0.4050606764781823,
      "grad_norm": 7.171485424041748,
      "learning_rate": 6.559416279092792e-06,
      "loss": 0.046,
      "step": 7844
    },
    {
      "epoch": 0.4051123160340821,
      "grad_norm": 13.03868293762207,
      "learning_rate": 6.558637766851205e-06,
      "loss": 0.1605,
      "step": 7845
    },
    {
      "epoch": 0.4051639555899819,
      "grad_norm": 21.320796966552734,
      "learning_rate": 6.557859212753841e-06,
      "loss": 0.1358,
      "step": 7846
    },
    {
      "epoch": 0.40521559514588174,
      "grad_norm": 4.1820573806762695,
      "learning_rate": 6.557080616821607e-06,
      "loss": 0.0734,
      "step": 7847
    },
    {
      "epoch": 0.40526723470178155,
      "grad_norm": 11.788965225219727,
      "learning_rate": 6.556301979075412e-06,
      "loss": 0.0533,
      "step": 7848
    },
    {
      "epoch": 0.40531887425768137,
      "grad_norm": 4.369396209716797,
      "learning_rate": 6.555523299536167e-06,
      "loss": 0.0564,
      "step": 7849
    },
    {
      "epoch": 0.4053705138135812,
      "grad_norm": 2.2302792072296143,
      "learning_rate": 6.554744578224781e-06,
      "loss": 0.0394,
      "step": 7850
    },
    {
      "epoch": 0.405422153369481,
      "grad_norm": 3.711782693862915,
      "learning_rate": 6.553965815162167e-06,
      "loss": 0.078,
      "step": 7851
    },
    {
      "epoch": 0.4054737929253808,
      "grad_norm": 2.565800666809082,
      "learning_rate": 6.553187010369236e-06,
      "loss": 0.0453,
      "step": 7852
    },
    {
      "epoch": 0.40552543248128065,
      "grad_norm": 10.49715518951416,
      "learning_rate": 6.552408163866904e-06,
      "loss": 0.0652,
      "step": 7853
    },
    {
      "epoch": 0.40557707203718046,
      "grad_norm": 8.325732231140137,
      "learning_rate": 6.551629275676086e-06,
      "loss": 0.0639,
      "step": 7854
    },
    {
      "epoch": 0.4056287115930803,
      "grad_norm": 3.4982526302337646,
      "learning_rate": 6.550850345817697e-06,
      "loss": 0.0492,
      "step": 7855
    },
    {
      "epoch": 0.4056803511489801,
      "grad_norm": 3.155531167984009,
      "learning_rate": 6.550071374312654e-06,
      "loss": 0.0625,
      "step": 7856
    },
    {
      "epoch": 0.4057319907048799,
      "grad_norm": 3.070504665374756,
      "learning_rate": 6.549292361181879e-06,
      "loss": 0.0636,
      "step": 7857
    },
    {
      "epoch": 0.40578363026077974,
      "grad_norm": 4.298681259155273,
      "learning_rate": 6.5485133064462905e-06,
      "loss": 0.0773,
      "step": 7858
    },
    {
      "epoch": 0.40583526981667956,
      "grad_norm": 3.1811912059783936,
      "learning_rate": 6.547734210126805e-06,
      "loss": 0.0451,
      "step": 7859
    },
    {
      "epoch": 0.4058869093725794,
      "grad_norm": 2.5212955474853516,
      "learning_rate": 6.54695507224435e-06,
      "loss": 0.051,
      "step": 7860
    },
    {
      "epoch": 0.4059385489284792,
      "grad_norm": 7.525758266448975,
      "learning_rate": 6.546175892819845e-06,
      "loss": 0.1488,
      "step": 7861
    },
    {
      "epoch": 0.405990188484379,
      "grad_norm": 5.935540199279785,
      "learning_rate": 6.545396671874217e-06,
      "loss": 0.1025,
      "step": 7862
    },
    {
      "epoch": 0.40604182804027883,
      "grad_norm": 1.94930899143219,
      "learning_rate": 6.5446174094283885e-06,
      "loss": 0.0557,
      "step": 7863
    },
    {
      "epoch": 0.40609346759617865,
      "grad_norm": 4.793895721435547,
      "learning_rate": 6.543838105503287e-06,
      "loss": 0.0875,
      "step": 7864
    },
    {
      "epoch": 0.40614510715207847,
      "grad_norm": 6.221998691558838,
      "learning_rate": 6.543058760119843e-06,
      "loss": 0.0777,
      "step": 7865
    },
    {
      "epoch": 0.4061967467079783,
      "grad_norm": 9.72350788116455,
      "learning_rate": 6.542279373298978e-06,
      "loss": 0.1407,
      "step": 7866
    },
    {
      "epoch": 0.4062483862638781,
      "grad_norm": 4.306913375854492,
      "learning_rate": 6.541499945061628e-06,
      "loss": 0.0873,
      "step": 7867
    },
    {
      "epoch": 0.406300025819778,
      "grad_norm": 4.035315036773682,
      "learning_rate": 6.540720475428719e-06,
      "loss": 0.0732,
      "step": 7868
    },
    {
      "epoch": 0.4063516653756778,
      "grad_norm": 7.949227333068848,
      "learning_rate": 6.5399409644211895e-06,
      "loss": 0.0836,
      "step": 7869
    },
    {
      "epoch": 0.4064033049315776,
      "grad_norm": 2.1604416370391846,
      "learning_rate": 6.539161412059967e-06,
      "loss": 0.0334,
      "step": 7870
    },
    {
      "epoch": 0.40645494448747743,
      "grad_norm": 2.396528959274292,
      "learning_rate": 6.538381818365987e-06,
      "loss": 0.0474,
      "step": 7871
    },
    {
      "epoch": 0.40650658404337725,
      "grad_norm": 5.333260536193848,
      "learning_rate": 6.5376021833601845e-06,
      "loss": 0.0617,
      "step": 7872
    },
    {
      "epoch": 0.40655822359927707,
      "grad_norm": 5.672069549560547,
      "learning_rate": 6.536822507063497e-06,
      "loss": 0.0828,
      "step": 7873
    },
    {
      "epoch": 0.4066098631551769,
      "grad_norm": 8.641366004943848,
      "learning_rate": 6.536042789496863e-06,
      "loss": 0.119,
      "step": 7874
    },
    {
      "epoch": 0.4066615027110767,
      "grad_norm": 2.2784786224365234,
      "learning_rate": 6.535263030681216e-06,
      "loss": 0.0492,
      "step": 7875
    },
    {
      "epoch": 0.4067131422669765,
      "grad_norm": 3.269979476928711,
      "learning_rate": 6.534483230637503e-06,
      "loss": 0.0745,
      "step": 7876
    },
    {
      "epoch": 0.40676478182287634,
      "grad_norm": 5.526833534240723,
      "learning_rate": 6.533703389386659e-06,
      "loss": 0.0988,
      "step": 7877
    },
    {
      "epoch": 0.40681642137877616,
      "grad_norm": 2.3434245586395264,
      "learning_rate": 6.532923506949629e-06,
      "loss": 0.0216,
      "step": 7878
    },
    {
      "epoch": 0.406868060934676,
      "grad_norm": 13.740533828735352,
      "learning_rate": 6.532143583347354e-06,
      "loss": 0.0902,
      "step": 7879
    },
    {
      "epoch": 0.4069197004905758,
      "grad_norm": 2.078032970428467,
      "learning_rate": 6.531363618600779e-06,
      "loss": 0.0403,
      "step": 7880
    },
    {
      "epoch": 0.4069713400464756,
      "grad_norm": 4.540918827056885,
      "learning_rate": 6.53058361273085e-06,
      "loss": 0.0551,
      "step": 7881
    },
    {
      "epoch": 0.40702297960237543,
      "grad_norm": 7.764376640319824,
      "learning_rate": 6.529803565758513e-06,
      "loss": 0.1107,
      "step": 7882
    },
    {
      "epoch": 0.40707461915827525,
      "grad_norm": 5.016412734985352,
      "learning_rate": 6.529023477704714e-06,
      "loss": 0.0773,
      "step": 7883
    },
    {
      "epoch": 0.40712625871417507,
      "grad_norm": 6.790655612945557,
      "learning_rate": 6.528243348590404e-06,
      "loss": 0.0498,
      "step": 7884
    },
    {
      "epoch": 0.4071778982700749,
      "grad_norm": 2.1763112545013428,
      "learning_rate": 6.527463178436531e-06,
      "loss": 0.0323,
      "step": 7885
    },
    {
      "epoch": 0.4072295378259747,
      "grad_norm": 3.0235559940338135,
      "learning_rate": 6.526682967264045e-06,
      "loss": 0.0243,
      "step": 7886
    },
    {
      "epoch": 0.4072811773818745,
      "grad_norm": 2.985929250717163,
      "learning_rate": 6.525902715093899e-06,
      "loss": 0.0376,
      "step": 7887
    },
    {
      "epoch": 0.40733281693777434,
      "grad_norm": 3.5253007411956787,
      "learning_rate": 6.5251224219470475e-06,
      "loss": 0.0555,
      "step": 7888
    },
    {
      "epoch": 0.40738445649367416,
      "grad_norm": 3.1651434898376465,
      "learning_rate": 6.524342087844442e-06,
      "loss": 0.0455,
      "step": 7889
    },
    {
      "epoch": 0.407436096049574,
      "grad_norm": 3.0517497062683105,
      "learning_rate": 6.523561712807038e-06,
      "loss": 0.0238,
      "step": 7890
    },
    {
      "epoch": 0.4074877356054738,
      "grad_norm": 3.3294525146484375,
      "learning_rate": 6.522781296855793e-06,
      "loss": 0.0422,
      "step": 7891
    },
    {
      "epoch": 0.4075393751613736,
      "grad_norm": 5.0309739112854,
      "learning_rate": 6.522000840011663e-06,
      "loss": 0.088,
      "step": 7892
    },
    {
      "epoch": 0.40759101471727344,
      "grad_norm": 7.296426773071289,
      "learning_rate": 6.521220342295609e-06,
      "loss": 0.0559,
      "step": 7893
    },
    {
      "epoch": 0.40764265427317325,
      "grad_norm": 14.243233680725098,
      "learning_rate": 6.5204398037285855e-06,
      "loss": 0.1148,
      "step": 7894
    },
    {
      "epoch": 0.4076942938290731,
      "grad_norm": 8.938983917236328,
      "learning_rate": 6.519659224331558e-06,
      "loss": 0.1341,
      "step": 7895
    },
    {
      "epoch": 0.4077459333849729,
      "grad_norm": 4.157503604888916,
      "learning_rate": 6.518878604125487e-06,
      "loss": 0.0979,
      "step": 7896
    },
    {
      "epoch": 0.4077975729408727,
      "grad_norm": 2.50849986076355,
      "learning_rate": 6.518097943131334e-06,
      "loss": 0.0293,
      "step": 7897
    },
    {
      "epoch": 0.40784921249677253,
      "grad_norm": 6.749706745147705,
      "learning_rate": 6.517317241370066e-06,
      "loss": 0.0449,
      "step": 7898
    },
    {
      "epoch": 0.40790085205267235,
      "grad_norm": 24.025888442993164,
      "learning_rate": 6.516536498862642e-06,
      "loss": 0.1608,
      "step": 7899
    },
    {
      "epoch": 0.40795249160857217,
      "grad_norm": 4.227044105529785,
      "learning_rate": 6.515755715630036e-06,
      "loss": 0.0622,
      "step": 7900
    },
    {
      "epoch": 0.408004131164472,
      "grad_norm": 4.097754001617432,
      "learning_rate": 6.5149748916932075e-06,
      "loss": 0.0475,
      "step": 7901
    },
    {
      "epoch": 0.4080557707203718,
      "grad_norm": 13.319456100463867,
      "learning_rate": 6.514194027073131e-06,
      "loss": 0.1435,
      "step": 7902
    },
    {
      "epoch": 0.4081074102762716,
      "grad_norm": 44.416934967041016,
      "learning_rate": 6.513413121790773e-06,
      "loss": 0.1473,
      "step": 7903
    },
    {
      "epoch": 0.40815904983217144,
      "grad_norm": 8.114594459533691,
      "learning_rate": 6.512632175867103e-06,
      "loss": 0.1162,
      "step": 7904
    },
    {
      "epoch": 0.40821068938807126,
      "grad_norm": 11.514503479003906,
      "learning_rate": 6.511851189323095e-06,
      "loss": 0.1197,
      "step": 7905
    },
    {
      "epoch": 0.4082623289439711,
      "grad_norm": 3.9405574798583984,
      "learning_rate": 6.51107016217972e-06,
      "loss": 0.0675,
      "step": 7906
    },
    {
      "epoch": 0.4083139684998709,
      "grad_norm": 4.8573760986328125,
      "learning_rate": 6.510289094457953e-06,
      "loss": 0.045,
      "step": 7907
    },
    {
      "epoch": 0.4083656080557707,
      "grad_norm": 5.4721598625183105,
      "learning_rate": 6.509507986178768e-06,
      "loss": 0.08,
      "step": 7908
    },
    {
      "epoch": 0.40841724761167053,
      "grad_norm": 3.6225197315216064,
      "learning_rate": 6.5087268373631415e-06,
      "loss": 0.0523,
      "step": 7909
    },
    {
      "epoch": 0.40846888716757035,
      "grad_norm": 10.121859550476074,
      "learning_rate": 6.5079456480320504e-06,
      "loss": 0.1141,
      "step": 7910
    },
    {
      "epoch": 0.40852052672347017,
      "grad_norm": 6.688459873199463,
      "learning_rate": 6.507164418206471e-06,
      "loss": 0.0796,
      "step": 7911
    },
    {
      "epoch": 0.40857216627937,
      "grad_norm": 2.25030779838562,
      "learning_rate": 6.506383147907385e-06,
      "loss": 0.0679,
      "step": 7912
    },
    {
      "epoch": 0.4086238058352698,
      "grad_norm": 2.8572752475738525,
      "learning_rate": 6.5056018371557725e-06,
      "loss": 0.0485,
      "step": 7913
    },
    {
      "epoch": 0.4086754453911696,
      "grad_norm": 7.79918909072876,
      "learning_rate": 6.504820485972613e-06,
      "loss": 0.0951,
      "step": 7914
    },
    {
      "epoch": 0.40872708494706944,
      "grad_norm": 7.349212169647217,
      "learning_rate": 6.504039094378891e-06,
      "loss": 0.0623,
      "step": 7915
    },
    {
      "epoch": 0.40877872450296926,
      "grad_norm": 4.0391082763671875,
      "learning_rate": 6.503257662395588e-06,
      "loss": 0.0446,
      "step": 7916
    },
    {
      "epoch": 0.4088303640588691,
      "grad_norm": 10.699840545654297,
      "learning_rate": 6.502476190043692e-06,
      "loss": 0.0653,
      "step": 7917
    },
    {
      "epoch": 0.4088820036147689,
      "grad_norm": 3.2542295455932617,
      "learning_rate": 6.501694677344186e-06,
      "loss": 0.0555,
      "step": 7918
    },
    {
      "epoch": 0.4089336431706687,
      "grad_norm": 1.2006971836090088,
      "learning_rate": 6.500913124318055e-06,
      "loss": 0.0281,
      "step": 7919
    },
    {
      "epoch": 0.40898528272656853,
      "grad_norm": 2.629173755645752,
      "learning_rate": 6.500131530986291e-06,
      "loss": 0.0359,
      "step": 7920
    },
    {
      "epoch": 0.40903692228246835,
      "grad_norm": 3.295419216156006,
      "learning_rate": 6.499349897369882e-06,
      "loss": 0.0403,
      "step": 7921
    },
    {
      "epoch": 0.40908856183836817,
      "grad_norm": 4.548181056976318,
      "learning_rate": 6.498568223489816e-06,
      "loss": 0.0727,
      "step": 7922
    },
    {
      "epoch": 0.409140201394268,
      "grad_norm": 4.07267427444458,
      "learning_rate": 6.497786509367085e-06,
      "loss": 0.0723,
      "step": 7923
    },
    {
      "epoch": 0.4091918409501678,
      "grad_norm": 3.968486785888672,
      "learning_rate": 6.497004755022684e-06,
      "loss": 0.0568,
      "step": 7924
    },
    {
      "epoch": 0.4092434805060676,
      "grad_norm": 9.254122734069824,
      "learning_rate": 6.4962229604776e-06,
      "loss": 0.0857,
      "step": 7925
    },
    {
      "epoch": 0.40929512006196744,
      "grad_norm": 9.459792137145996,
      "learning_rate": 6.495441125752835e-06,
      "loss": 0.1075,
      "step": 7926
    },
    {
      "epoch": 0.40934675961786726,
      "grad_norm": 8.27361011505127,
      "learning_rate": 6.494659250869378e-06,
      "loss": 0.0861,
      "step": 7927
    },
    {
      "epoch": 0.4093983991737671,
      "grad_norm": 5.346077919006348,
      "learning_rate": 6.49387733584823e-06,
      "loss": 0.0905,
      "step": 7928
    },
    {
      "epoch": 0.4094500387296669,
      "grad_norm": 8.438179016113281,
      "learning_rate": 6.493095380710386e-06,
      "loss": 0.1197,
      "step": 7929
    },
    {
      "epoch": 0.4095016782855667,
      "grad_norm": 19.83772087097168,
      "learning_rate": 6.492313385476845e-06,
      "loss": 0.1246,
      "step": 7930
    },
    {
      "epoch": 0.4095533178414666,
      "grad_norm": 22.423585891723633,
      "learning_rate": 6.4915313501686096e-06,
      "loss": 0.0786,
      "step": 7931
    },
    {
      "epoch": 0.4096049573973664,
      "grad_norm": 11.54234790802002,
      "learning_rate": 6.4907492748066746e-06,
      "loss": 0.1566,
      "step": 7932
    },
    {
      "epoch": 0.4096565969532662,
      "grad_norm": 3.6474809646606445,
      "learning_rate": 6.48996715941205e-06,
      "loss": 0.07,
      "step": 7933
    },
    {
      "epoch": 0.40970823650916605,
      "grad_norm": 13.612740516662598,
      "learning_rate": 6.489185004005729e-06,
      "loss": 0.1531,
      "step": 7934
    },
    {
      "epoch": 0.40975987606506586,
      "grad_norm": 5.789085388183594,
      "learning_rate": 6.488402808608726e-06,
      "loss": 0.04,
      "step": 7935
    },
    {
      "epoch": 0.4098115156209657,
      "grad_norm": 4.774659156799316,
      "learning_rate": 6.487620573242039e-06,
      "loss": 0.0559,
      "step": 7936
    },
    {
      "epoch": 0.4098631551768655,
      "grad_norm": 6.437483787536621,
      "learning_rate": 6.486838297926677e-06,
      "loss": 0.1064,
      "step": 7937
    },
    {
      "epoch": 0.4099147947327653,
      "grad_norm": 6.113980770111084,
      "learning_rate": 6.486055982683646e-06,
      "loss": 0.1254,
      "step": 7938
    },
    {
      "epoch": 0.40996643428866514,
      "grad_norm": 5.307106018066406,
      "learning_rate": 6.485273627533955e-06,
      "loss": 0.0396,
      "step": 7939
    },
    {
      "epoch": 0.41001807384456496,
      "grad_norm": 3.645639657974243,
      "learning_rate": 6.484491232498615e-06,
      "loss": 0.0551,
      "step": 7940
    },
    {
      "epoch": 0.4100697134004648,
      "grad_norm": 5.918747901916504,
      "learning_rate": 6.483708797598632e-06,
      "loss": 0.1096,
      "step": 7941
    },
    {
      "epoch": 0.4101213529563646,
      "grad_norm": 11.229751586914062,
      "learning_rate": 6.482926322855024e-06,
      "loss": 0.0785,
      "step": 7942
    },
    {
      "epoch": 0.4101729925122644,
      "grad_norm": 1.8256217241287231,
      "learning_rate": 6.482143808288797e-06,
      "loss": 0.0452,
      "step": 7943
    },
    {
      "epoch": 0.41022463206816423,
      "grad_norm": 6.294797420501709,
      "learning_rate": 6.48136125392097e-06,
      "loss": 0.1329,
      "step": 7944
    },
    {
      "epoch": 0.41027627162406405,
      "grad_norm": 3.8629963397979736,
      "learning_rate": 6.480578659772555e-06,
      "loss": 0.0845,
      "step": 7945
    },
    {
      "epoch": 0.41032791117996387,
      "grad_norm": 4.615016937255859,
      "learning_rate": 6.479796025864569e-06,
      "loss": 0.0707,
      "step": 7946
    },
    {
      "epoch": 0.4103795507358637,
      "grad_norm": 3.6457889080047607,
      "learning_rate": 6.4790133522180266e-06,
      "loss": 0.06,
      "step": 7947
    },
    {
      "epoch": 0.4104311902917635,
      "grad_norm": 6.700169086456299,
      "learning_rate": 6.4782306388539494e-06,
      "loss": 0.0697,
      "step": 7948
    },
    {
      "epoch": 0.4104828298476633,
      "grad_norm": 6.675643444061279,
      "learning_rate": 6.477447885793353e-06,
      "loss": 0.0444,
      "step": 7949
    },
    {
      "epoch": 0.41053446940356314,
      "grad_norm": 4.02327823638916,
      "learning_rate": 6.476665093057262e-06,
      "loss": 0.0522,
      "step": 7950
    },
    {
      "epoch": 0.41058610895946296,
      "grad_norm": 9.175921440124512,
      "learning_rate": 6.4758822606666926e-06,
      "loss": 0.1226,
      "step": 7951
    },
    {
      "epoch": 0.4106377485153628,
      "grad_norm": 10.688812255859375,
      "learning_rate": 6.475099388642669e-06,
      "loss": 0.1181,
      "step": 7952
    },
    {
      "epoch": 0.4106893880712626,
      "grad_norm": 16.56269073486328,
      "learning_rate": 6.474316477006215e-06,
      "loss": 0.0751,
      "step": 7953
    },
    {
      "epoch": 0.4107410276271624,
      "grad_norm": 5.090817928314209,
      "learning_rate": 6.473533525778355e-06,
      "loss": 0.0553,
      "step": 7954
    },
    {
      "epoch": 0.41079266718306223,
      "grad_norm": 2.3224987983703613,
      "learning_rate": 6.472750534980113e-06,
      "loss": 0.037,
      "step": 7955
    },
    {
      "epoch": 0.41084430673896205,
      "grad_norm": 2.6849324703216553,
      "learning_rate": 6.471967504632517e-06,
      "loss": 0.0504,
      "step": 7956
    },
    {
      "epoch": 0.41089594629486187,
      "grad_norm": 2.9724280834198,
      "learning_rate": 6.471184434756595e-06,
      "loss": 0.0275,
      "step": 7957
    },
    {
      "epoch": 0.4109475858507617,
      "grad_norm": 1.6853492259979248,
      "learning_rate": 6.470401325373373e-06,
      "loss": 0.0393,
      "step": 7958
    },
    {
      "epoch": 0.4109992254066615,
      "grad_norm": 3.9500622749328613,
      "learning_rate": 6.469618176503884e-06,
      "loss": 0.0437,
      "step": 7959
    },
    {
      "epoch": 0.4110508649625613,
      "grad_norm": 3.007007598876953,
      "learning_rate": 6.468834988169156e-06,
      "loss": 0.0756,
      "step": 7960
    },
    {
      "epoch": 0.41110250451846114,
      "grad_norm": 6.815203666687012,
      "learning_rate": 6.468051760390224e-06,
      "loss": 0.1224,
      "step": 7961
    },
    {
      "epoch": 0.41115414407436096,
      "grad_norm": 2.5584182739257812,
      "learning_rate": 6.4672684931881165e-06,
      "loss": 0.0555,
      "step": 7962
    },
    {
      "epoch": 0.4112057836302608,
      "grad_norm": 7.363744735717773,
      "learning_rate": 6.466485186583871e-06,
      "loss": 0.1274,
      "step": 7963
    },
    {
      "epoch": 0.4112574231861606,
      "grad_norm": 5.965040683746338,
      "learning_rate": 6.465701840598522e-06,
      "loss": 0.0753,
      "step": 7964
    },
    {
      "epoch": 0.4113090627420604,
      "grad_norm": 1.618241786956787,
      "learning_rate": 6.464918455253102e-06,
      "loss": 0.0615,
      "step": 7965
    },
    {
      "epoch": 0.41136070229796023,
      "grad_norm": 3.8040876388549805,
      "learning_rate": 6.464135030568652e-06,
      "loss": 0.0322,
      "step": 7966
    },
    {
      "epoch": 0.41141234185386005,
      "grad_norm": 9.125747680664062,
      "learning_rate": 6.463351566566208e-06,
      "loss": 0.058,
      "step": 7967
    },
    {
      "epoch": 0.41146398140975987,
      "grad_norm": 7.072242259979248,
      "learning_rate": 6.462568063266812e-06,
      "loss": 0.0816,
      "step": 7968
    },
    {
      "epoch": 0.4115156209656597,
      "grad_norm": 1.383079171180725,
      "learning_rate": 6.461784520691502e-06,
      "loss": 0.0236,
      "step": 7969
    },
    {
      "epoch": 0.4115672605215595,
      "grad_norm": 2.0213937759399414,
      "learning_rate": 6.461000938861319e-06,
      "loss": 0.0496,
      "step": 7970
    },
    {
      "epoch": 0.4116189000774593,
      "grad_norm": 11.939763069152832,
      "learning_rate": 6.460217317797306e-06,
      "loss": 0.0992,
      "step": 7971
    },
    {
      "epoch": 0.41167053963335914,
      "grad_norm": 2.9150190353393555,
      "learning_rate": 6.459433657520508e-06,
      "loss": 0.0666,
      "step": 7972
    },
    {
      "epoch": 0.41172217918925896,
      "grad_norm": 3.20157790184021,
      "learning_rate": 6.458649958051968e-06,
      "loss": 0.1014,
      "step": 7973
    },
    {
      "epoch": 0.4117738187451588,
      "grad_norm": 2.2765181064605713,
      "learning_rate": 6.457866219412729e-06,
      "loss": 0.0489,
      "step": 7974
    },
    {
      "epoch": 0.4118254583010586,
      "grad_norm": 5.681329727172852,
      "learning_rate": 6.457082441623842e-06,
      "loss": 0.092,
      "step": 7975
    },
    {
      "epoch": 0.4118770978569584,
      "grad_norm": 5.358357906341553,
      "learning_rate": 6.456298624706352e-06,
      "loss": 0.0813,
      "step": 7976
    },
    {
      "epoch": 0.41192873741285824,
      "grad_norm": 3.2330703735351562,
      "learning_rate": 6.455514768681308e-06,
      "loss": 0.0324,
      "step": 7977
    },
    {
      "epoch": 0.41198037696875806,
      "grad_norm": 3.59649920463562,
      "learning_rate": 6.454730873569761e-06,
      "loss": 0.0845,
      "step": 7978
    },
    {
      "epoch": 0.4120320165246579,
      "grad_norm": 4.111301898956299,
      "learning_rate": 6.453946939392761e-06,
      "loss": 0.046,
      "step": 7979
    },
    {
      "epoch": 0.4120836560805577,
      "grad_norm": 2.2174060344696045,
      "learning_rate": 6.453162966171359e-06,
      "loss": 0.0425,
      "step": 7980
    },
    {
      "epoch": 0.4121352956364575,
      "grad_norm": 4.672916412353516,
      "learning_rate": 6.452378953926608e-06,
      "loss": 0.0358,
      "step": 7981
    },
    {
      "epoch": 0.41218693519235733,
      "grad_norm": 6.385516166687012,
      "learning_rate": 6.451594902679566e-06,
      "loss": 0.0703,
      "step": 7982
    },
    {
      "epoch": 0.41223857474825715,
      "grad_norm": 7.288697242736816,
      "learning_rate": 6.4508108124512805e-06,
      "loss": 0.0674,
      "step": 7983
    },
    {
      "epoch": 0.41229021430415697,
      "grad_norm": 6.7194294929504395,
      "learning_rate": 6.4500266832628136e-06,
      "loss": 0.0654,
      "step": 7984
    },
    {
      "epoch": 0.4123418538600568,
      "grad_norm": 13.322674751281738,
      "learning_rate": 6.449242515135219e-06,
      "loss": 0.1997,
      "step": 7985
    },
    {
      "epoch": 0.4123934934159566,
      "grad_norm": 2.1796951293945312,
      "learning_rate": 6.448458308089556e-06,
      "loss": 0.0306,
      "step": 7986
    },
    {
      "epoch": 0.4124451329718564,
      "grad_norm": 5.003080368041992,
      "learning_rate": 6.447674062146885e-06,
      "loss": 0.1219,
      "step": 7987
    },
    {
      "epoch": 0.41249677252775624,
      "grad_norm": 8.158489227294922,
      "learning_rate": 6.446889777328265e-06,
      "loss": 0.078,
      "step": 7988
    },
    {
      "epoch": 0.41254841208365606,
      "grad_norm": 2.272671699523926,
      "learning_rate": 6.446105453654757e-06,
      "loss": 0.0346,
      "step": 7989
    },
    {
      "epoch": 0.4126000516395559,
      "grad_norm": 10.692462921142578,
      "learning_rate": 6.445321091147423e-06,
      "loss": 0.0879,
      "step": 7990
    },
    {
      "epoch": 0.4126516911954557,
      "grad_norm": 4.617941379547119,
      "learning_rate": 6.444536689827326e-06,
      "loss": 0.0939,
      "step": 7991
    },
    {
      "epoch": 0.4127033307513555,
      "grad_norm": 4.515783786773682,
      "learning_rate": 6.4437522497155336e-06,
      "loss": 0.0651,
      "step": 7992
    },
    {
      "epoch": 0.41275497030725533,
      "grad_norm": 5.267086029052734,
      "learning_rate": 6.4429677708331075e-06,
      "loss": 0.0604,
      "step": 7993
    },
    {
      "epoch": 0.41280660986315515,
      "grad_norm": 2.651379346847534,
      "learning_rate": 6.442183253201116e-06,
      "loss": 0.0518,
      "step": 7994
    },
    {
      "epoch": 0.412858249419055,
      "grad_norm": 6.492796897888184,
      "learning_rate": 6.441398696840626e-06,
      "loss": 0.0875,
      "step": 7995
    },
    {
      "epoch": 0.41290988897495484,
      "grad_norm": 8.250231742858887,
      "learning_rate": 6.440614101772707e-06,
      "loss": 0.0967,
      "step": 7996
    },
    {
      "epoch": 0.41296152853085466,
      "grad_norm": 2.5745534896850586,
      "learning_rate": 6.439829468018427e-06,
      "loss": 0.043,
      "step": 7997
    },
    {
      "epoch": 0.4130131680867545,
      "grad_norm": 4.255914211273193,
      "learning_rate": 6.439044795598857e-06,
      "loss": 0.0915,
      "step": 7998
    },
    {
      "epoch": 0.4130648076426543,
      "grad_norm": 2.0490665435791016,
      "learning_rate": 6.438260084535069e-06,
      "loss": 0.0435,
      "step": 7999
    },
    {
      "epoch": 0.4131164471985541,
      "grad_norm": 4.324810981750488,
      "learning_rate": 6.437475334848135e-06,
      "loss": 0.0302,
      "step": 8000
    },
    {
      "epoch": 0.41316808675445393,
      "grad_norm": 7.022913932800293,
      "learning_rate": 6.4366905465591315e-06,
      "loss": 0.1285,
      "step": 8001
    },
    {
      "epoch": 0.41321972631035375,
      "grad_norm": 12.561385154724121,
      "learning_rate": 6.4359057196891305e-06,
      "loss": 0.0651,
      "step": 8002
    },
    {
      "epoch": 0.41327136586625357,
      "grad_norm": 1.5027997493743896,
      "learning_rate": 6.4351208542592065e-06,
      "loss": 0.0316,
      "step": 8003
    },
    {
      "epoch": 0.4133230054221534,
      "grad_norm": 2.8615379333496094,
      "learning_rate": 6.434335950290442e-06,
      "loss": 0.0614,
      "step": 8004
    },
    {
      "epoch": 0.4133746449780532,
      "grad_norm": 7.847778797149658,
      "learning_rate": 6.4335510078039075e-06,
      "loss": 0.0883,
      "step": 8005
    },
    {
      "epoch": 0.413426284533953,
      "grad_norm": 7.893518924713135,
      "learning_rate": 6.432766026820689e-06,
      "loss": 0.0959,
      "step": 8006
    },
    {
      "epoch": 0.41347792408985284,
      "grad_norm": 5.910889148712158,
      "learning_rate": 6.431981007361861e-06,
      "loss": 0.1105,
      "step": 8007
    },
    {
      "epoch": 0.41352956364575266,
      "grad_norm": 5.150066375732422,
      "learning_rate": 6.4311959494485064e-06,
      "loss": 0.0486,
      "step": 8008
    },
    {
      "epoch": 0.4135812032016525,
      "grad_norm": 7.3907084465026855,
      "learning_rate": 6.430410853101708e-06,
      "loss": 0.1118,
      "step": 8009
    },
    {
      "epoch": 0.4136328427575523,
      "grad_norm": 22.878223419189453,
      "learning_rate": 6.429625718342548e-06,
      "loss": 0.1518,
      "step": 8010
    },
    {
      "epoch": 0.4136844823134521,
      "grad_norm": 3.218618392944336,
      "learning_rate": 6.42884054519211e-06,
      "loss": 0.0377,
      "step": 8011
    },
    {
      "epoch": 0.41373612186935194,
      "grad_norm": 6.268611907958984,
      "learning_rate": 6.42805533367148e-06,
      "loss": 0.0894,
      "step": 8012
    },
    {
      "epoch": 0.41378776142525175,
      "grad_norm": 4.75072717666626,
      "learning_rate": 6.427270083801744e-06,
      "loss": 0.0394,
      "step": 8013
    },
    {
      "epoch": 0.4138394009811516,
      "grad_norm": 1.6810778379440308,
      "learning_rate": 6.42648479560399e-06,
      "loss": 0.031,
      "step": 8014
    },
    {
      "epoch": 0.4138910405370514,
      "grad_norm": 9.795029640197754,
      "learning_rate": 6.425699469099304e-06,
      "loss": 0.0636,
      "step": 8015
    },
    {
      "epoch": 0.4139426800929512,
      "grad_norm": 17.874832153320312,
      "learning_rate": 6.424914104308778e-06,
      "loss": 0.2152,
      "step": 8016
    },
    {
      "epoch": 0.41399431964885103,
      "grad_norm": 3.087587833404541,
      "learning_rate": 6.424128701253499e-06,
      "loss": 0.078,
      "step": 8017
    },
    {
      "epoch": 0.41404595920475085,
      "grad_norm": 8.360746383666992,
      "learning_rate": 6.423343259954561e-06,
      "loss": 0.0795,
      "step": 8018
    },
    {
      "epoch": 0.41409759876065066,
      "grad_norm": 2.3858156204223633,
      "learning_rate": 6.422557780433054e-06,
      "loss": 0.0249,
      "step": 8019
    },
    {
      "epoch": 0.4141492383165505,
      "grad_norm": 3.3217906951904297,
      "learning_rate": 6.421772262710073e-06,
      "loss": 0.0361,
      "step": 8020
    },
    {
      "epoch": 0.4142008778724503,
      "grad_norm": 8.369490623474121,
      "learning_rate": 6.420986706806712e-06,
      "loss": 0.0734,
      "step": 8021
    },
    {
      "epoch": 0.4142525174283501,
      "grad_norm": 6.6931939125061035,
      "learning_rate": 6.4202011127440665e-06,
      "loss": 0.1072,
      "step": 8022
    },
    {
      "epoch": 0.41430415698424994,
      "grad_norm": 15.93560791015625,
      "learning_rate": 6.419415480543234e-06,
      "loss": 0.0859,
      "step": 8023
    },
    {
      "epoch": 0.41435579654014976,
      "grad_norm": 9.771568298339844,
      "learning_rate": 6.4186298102253076e-06,
      "loss": 0.1014,
      "step": 8024
    },
    {
      "epoch": 0.4144074360960496,
      "grad_norm": 6.715365409851074,
      "learning_rate": 6.417844101811393e-06,
      "loss": 0.1086,
      "step": 8025
    },
    {
      "epoch": 0.4144590756519494,
      "grad_norm": 6.009362697601318,
      "learning_rate": 6.417058355322582e-06,
      "loss": 0.0872,
      "step": 8026
    },
    {
      "epoch": 0.4145107152078492,
      "grad_norm": 4.499631404876709,
      "learning_rate": 6.4162725707799785e-06,
      "loss": 0.0702,
      "step": 8027
    },
    {
      "epoch": 0.41456235476374903,
      "grad_norm": 1.122538685798645,
      "learning_rate": 6.415486748204686e-06,
      "loss": 0.0194,
      "step": 8028
    },
    {
      "epoch": 0.41461399431964885,
      "grad_norm": 4.330747127532959,
      "learning_rate": 6.414700887617804e-06,
      "loss": 0.0481,
      "step": 8029
    },
    {
      "epoch": 0.41466563387554867,
      "grad_norm": 11.501932144165039,
      "learning_rate": 6.413914989040437e-06,
      "loss": 0.0941,
      "step": 8030
    },
    {
      "epoch": 0.4147172734314485,
      "grad_norm": 14.654263496398926,
      "learning_rate": 6.413129052493689e-06,
      "loss": 0.0948,
      "step": 8031
    },
    {
      "epoch": 0.4147689129873483,
      "grad_norm": 5.218995094299316,
      "learning_rate": 6.4123430779986695e-06,
      "loss": 0.0675,
      "step": 8032
    },
    {
      "epoch": 0.4148205525432481,
      "grad_norm": 4.326454162597656,
      "learning_rate": 6.411557065576478e-06,
      "loss": 0.063,
      "step": 8033
    },
    {
      "epoch": 0.41487219209914794,
      "grad_norm": 5.176865100860596,
      "learning_rate": 6.410771015248229e-06,
      "loss": 0.063,
      "step": 8034
    },
    {
      "epoch": 0.41492383165504776,
      "grad_norm": 4.743167400360107,
      "learning_rate": 6.409984927035027e-06,
      "loss": 0.103,
      "step": 8035
    },
    {
      "epoch": 0.4149754712109476,
      "grad_norm": 3.9371144771575928,
      "learning_rate": 6.409198800957983e-06,
      "loss": 0.0667,
      "step": 8036
    },
    {
      "epoch": 0.4150271107668474,
      "grad_norm": 3.478121280670166,
      "learning_rate": 6.408412637038208e-06,
      "loss": 0.069,
      "step": 8037
    },
    {
      "epoch": 0.4150787503227472,
      "grad_norm": 3.8371822834014893,
      "learning_rate": 6.407626435296813e-06,
      "loss": 0.0492,
      "step": 8038
    },
    {
      "epoch": 0.41513038987864703,
      "grad_norm": 4.164485931396484,
      "learning_rate": 6.406840195754912e-06,
      "loss": 0.0526,
      "step": 8039
    },
    {
      "epoch": 0.41518202943454685,
      "grad_norm": 3.4718923568725586,
      "learning_rate": 6.406053918433616e-06,
      "loss": 0.0291,
      "step": 8040
    },
    {
      "epoch": 0.41523366899044667,
      "grad_norm": 3.5733563899993896,
      "learning_rate": 6.405267603354044e-06,
      "loss": 0.0318,
      "step": 8041
    },
    {
      "epoch": 0.4152853085463465,
      "grad_norm": 5.604750633239746,
      "learning_rate": 6.404481250537307e-06,
      "loss": 0.0285,
      "step": 8042
    },
    {
      "epoch": 0.4153369481022463,
      "grad_norm": 7.924130916595459,
      "learning_rate": 6.403694860004525e-06,
      "loss": 0.0418,
      "step": 8043
    },
    {
      "epoch": 0.4153885876581461,
      "grad_norm": 8.884010314941406,
      "learning_rate": 6.402908431776814e-06,
      "loss": 0.0974,
      "step": 8044
    },
    {
      "epoch": 0.41544022721404594,
      "grad_norm": 2.8104467391967773,
      "learning_rate": 6.402121965875294e-06,
      "loss": 0.0593,
      "step": 8045
    },
    {
      "epoch": 0.41549186676994576,
      "grad_norm": 3.7043232917785645,
      "learning_rate": 6.401335462321086e-06,
      "loss": 0.0566,
      "step": 8046
    },
    {
      "epoch": 0.4155435063258456,
      "grad_norm": 2.0283007621765137,
      "learning_rate": 6.4005489211353074e-06,
      "loss": 0.0359,
      "step": 8047
    },
    {
      "epoch": 0.4155951458817454,
      "grad_norm": 20.032196044921875,
      "learning_rate": 6.399762342339084e-06,
      "loss": 0.1346,
      "step": 8048
    },
    {
      "epoch": 0.4156467854376452,
      "grad_norm": 5.057328701019287,
      "learning_rate": 6.3989757259535355e-06,
      "loss": 0.1075,
      "step": 8049
    },
    {
      "epoch": 0.41569842499354503,
      "grad_norm": 15.395879745483398,
      "learning_rate": 6.398189071999787e-06,
      "loss": 0.1187,
      "step": 8050
    },
    {
      "epoch": 0.41575006454944485,
      "grad_norm": 4.044466972351074,
      "learning_rate": 6.397402380498961e-06,
      "loss": 0.0611,
      "step": 8051
    },
    {
      "epoch": 0.41580170410534467,
      "grad_norm": 12.873085021972656,
      "learning_rate": 6.396615651472188e-06,
      "loss": 0.2157,
      "step": 8052
    },
    {
      "epoch": 0.4158533436612445,
      "grad_norm": 2.477963924407959,
      "learning_rate": 6.395828884940591e-06,
      "loss": 0.037,
      "step": 8053
    },
    {
      "epoch": 0.4159049832171443,
      "grad_norm": 3.2780895233154297,
      "learning_rate": 6.395042080925301e-06,
      "loss": 0.0347,
      "step": 8054
    },
    {
      "epoch": 0.4159566227730441,
      "grad_norm": 3.414680242538452,
      "learning_rate": 6.394255239447444e-06,
      "loss": 0.0465,
      "step": 8055
    },
    {
      "epoch": 0.41600826232894395,
      "grad_norm": 3.8049728870391846,
      "learning_rate": 6.3934683605281524e-06,
      "loss": 0.0742,
      "step": 8056
    },
    {
      "epoch": 0.41605990188484376,
      "grad_norm": 2.8867783546447754,
      "learning_rate": 6.392681444188555e-06,
      "loss": 0.0371,
      "step": 8057
    },
    {
      "epoch": 0.4161115414407436,
      "grad_norm": 8.698801040649414,
      "learning_rate": 6.391894490449786e-06,
      "loss": 0.1095,
      "step": 8058
    },
    {
      "epoch": 0.41616318099664346,
      "grad_norm": 3.155466318130493,
      "learning_rate": 6.3911074993329745e-06,
      "loss": 0.0525,
      "step": 8059
    },
    {
      "epoch": 0.4162148205525433,
      "grad_norm": 7.457042217254639,
      "learning_rate": 6.390320470859258e-06,
      "loss": 0.0883,
      "step": 8060
    },
    {
      "epoch": 0.4162664601084431,
      "grad_norm": 1.5093024969100952,
      "learning_rate": 6.38953340504977e-06,
      "loss": 0.0176,
      "step": 8061
    },
    {
      "epoch": 0.4163180996643429,
      "grad_norm": 3.7167420387268066,
      "learning_rate": 6.3887463019256466e-06,
      "loss": 0.0566,
      "step": 8062
    },
    {
      "epoch": 0.41636973922024273,
      "grad_norm": 6.892951011657715,
      "learning_rate": 6.387959161508025e-06,
      "loss": 0.1428,
      "step": 8063
    },
    {
      "epoch": 0.41642137877614255,
      "grad_norm": 4.46142578125,
      "learning_rate": 6.387171983818043e-06,
      "loss": 0.0638,
      "step": 8064
    },
    {
      "epoch": 0.41647301833204237,
      "grad_norm": 11.002480506896973,
      "learning_rate": 6.38638476887684e-06,
      "loss": 0.0787,
      "step": 8065
    },
    {
      "epoch": 0.4165246578879422,
      "grad_norm": 4.773715972900391,
      "learning_rate": 6.385597516705553e-06,
      "loss": 0.098,
      "step": 8066
    },
    {
      "epoch": 0.416576297443842,
      "grad_norm": 19.83466911315918,
      "learning_rate": 6.384810227325329e-06,
      "loss": 0.2961,
      "step": 8067
    },
    {
      "epoch": 0.4166279369997418,
      "grad_norm": 10.370271682739258,
      "learning_rate": 6.384022900757303e-06,
      "loss": 0.0885,
      "step": 8068
    },
    {
      "epoch": 0.41667957655564164,
      "grad_norm": 3.519925117492676,
      "learning_rate": 6.383235537022623e-06,
      "loss": 0.0472,
      "step": 8069
    },
    {
      "epoch": 0.41673121611154146,
      "grad_norm": 3.6645939350128174,
      "learning_rate": 6.38244813614243e-06,
      "loss": 0.0467,
      "step": 8070
    },
    {
      "epoch": 0.4167828556674413,
      "grad_norm": 5.261351108551025,
      "learning_rate": 6.3816606981378705e-06,
      "loss": 0.0644,
      "step": 8071
    },
    {
      "epoch": 0.4168344952233411,
      "grad_norm": 11.848467826843262,
      "learning_rate": 6.380873223030091e-06,
      "loss": 0.106,
      "step": 8072
    },
    {
      "epoch": 0.4168861347792409,
      "grad_norm": 4.99653434753418,
      "learning_rate": 6.380085710840235e-06,
      "loss": 0.0911,
      "step": 8073
    },
    {
      "epoch": 0.41693777433514073,
      "grad_norm": 6.862636089324951,
      "learning_rate": 6.379298161589454e-06,
      "loss": 0.056,
      "step": 8074
    },
    {
      "epoch": 0.41698941389104055,
      "grad_norm": 5.5509209632873535,
      "learning_rate": 6.378510575298895e-06,
      "loss": 0.0724,
      "step": 8075
    },
    {
      "epoch": 0.41704105344694037,
      "grad_norm": 5.761579513549805,
      "learning_rate": 6.37772295198971e-06,
      "loss": 0.0828,
      "step": 8076
    },
    {
      "epoch": 0.4170926930028402,
      "grad_norm": 11.77230453491211,
      "learning_rate": 6.3769352916830475e-06,
      "loss": 0.128,
      "step": 8077
    },
    {
      "epoch": 0.41714433255874,
      "grad_norm": 2.3417646884918213,
      "learning_rate": 6.37614759440006e-06,
      "loss": 0.0576,
      "step": 8078
    },
    {
      "epoch": 0.4171959721146398,
      "grad_norm": 10.218048095703125,
      "learning_rate": 6.375359860161901e-06,
      "loss": 0.1247,
      "step": 8079
    },
    {
      "epoch": 0.41724761167053964,
      "grad_norm": 4.654983043670654,
      "learning_rate": 6.374572088989724e-06,
      "loss": 0.0847,
      "step": 8080
    },
    {
      "epoch": 0.41729925122643946,
      "grad_norm": 2.1852197647094727,
      "learning_rate": 6.373784280904685e-06,
      "loss": 0.0451,
      "step": 8081
    },
    {
      "epoch": 0.4173508907823393,
      "grad_norm": 3.074894428253174,
      "learning_rate": 6.372996435927937e-06,
      "loss": 0.0379,
      "step": 8082
    },
    {
      "epoch": 0.4174025303382391,
      "grad_norm": 2.8676917552948,
      "learning_rate": 6.372208554080641e-06,
      "loss": 0.0415,
      "step": 8083
    },
    {
      "epoch": 0.4174541698941389,
      "grad_norm": 2.343186378479004,
      "learning_rate": 6.37142063538395e-06,
      "loss": 0.0497,
      "step": 8084
    },
    {
      "epoch": 0.41750580945003873,
      "grad_norm": 4.710056781768799,
      "learning_rate": 6.370632679859027e-06,
      "loss": 0.0541,
      "step": 8085
    },
    {
      "epoch": 0.41755744900593855,
      "grad_norm": 5.006488800048828,
      "learning_rate": 6.369844687527029e-06,
      "loss": 0.0788,
      "step": 8086
    },
    {
      "epoch": 0.41760908856183837,
      "grad_norm": 5.237309455871582,
      "learning_rate": 6.3690566584091184e-06,
      "loss": 0.0494,
      "step": 8087
    },
    {
      "epoch": 0.4176607281177382,
      "grad_norm": 12.501692771911621,
      "learning_rate": 6.368268592526457e-06,
      "loss": 0.1582,
      "step": 8088
    },
    {
      "epoch": 0.417712367673638,
      "grad_norm": 4.655656814575195,
      "learning_rate": 6.3674804899002074e-06,
      "loss": 0.0458,
      "step": 8089
    },
    {
      "epoch": 0.4177640072295378,
      "grad_norm": 7.2081732749938965,
      "learning_rate": 6.366692350551532e-06,
      "loss": 0.0982,
      "step": 8090
    },
    {
      "epoch": 0.41781564678543764,
      "grad_norm": 3.1486215591430664,
      "learning_rate": 6.365904174501598e-06,
      "loss": 0.0423,
      "step": 8091
    },
    {
      "epoch": 0.41786728634133746,
      "grad_norm": 10.307414054870605,
      "learning_rate": 6.36511596177157e-06,
      "loss": 0.1273,
      "step": 8092
    },
    {
      "epoch": 0.4179189258972373,
      "grad_norm": 1.9823532104492188,
      "learning_rate": 6.364327712382613e-06,
      "loss": 0.0248,
      "step": 8093
    },
    {
      "epoch": 0.4179705654531371,
      "grad_norm": 4.930261135101318,
      "learning_rate": 6.363539426355897e-06,
      "loss": 0.076,
      "step": 8094
    },
    {
      "epoch": 0.4180222050090369,
      "grad_norm": 6.9197916984558105,
      "learning_rate": 6.362751103712589e-06,
      "loss": 0.085,
      "step": 8095
    },
    {
      "epoch": 0.41807384456493674,
      "grad_norm": 4.488452911376953,
      "learning_rate": 6.3619627444738605e-06,
      "loss": 0.0543,
      "step": 8096
    },
    {
      "epoch": 0.41812548412083655,
      "grad_norm": 13.885909080505371,
      "learning_rate": 6.361174348660881e-06,
      "loss": 0.1683,
      "step": 8097
    },
    {
      "epoch": 0.4181771236767364,
      "grad_norm": 12.352155685424805,
      "learning_rate": 6.360385916294824e-06,
      "loss": 0.1325,
      "step": 8098
    },
    {
      "epoch": 0.4182287632326362,
      "grad_norm": 4.9010772705078125,
      "learning_rate": 6.359597447396857e-06,
      "loss": 0.0526,
      "step": 8099
    },
    {
      "epoch": 0.418280402788536,
      "grad_norm": 7.020668983459473,
      "learning_rate": 6.358808941988161e-06,
      "loss": 0.1123,
      "step": 8100
    },
    {
      "epoch": 0.41833204234443583,
      "grad_norm": 2.4129064083099365,
      "learning_rate": 6.3580204000899035e-06,
      "loss": 0.0175,
      "step": 8101
    },
    {
      "epoch": 0.41838368190033565,
      "grad_norm": 2.1322357654571533,
      "learning_rate": 6.357231821723265e-06,
      "loss": 0.069,
      "step": 8102
    },
    {
      "epoch": 0.41843532145623547,
      "grad_norm": 7.601905822753906,
      "learning_rate": 6.3564432069094195e-06,
      "loss": 0.1303,
      "step": 8103
    },
    {
      "epoch": 0.4184869610121353,
      "grad_norm": 1.4018996953964233,
      "learning_rate": 6.355654555669546e-06,
      "loss": 0.0149,
      "step": 8104
    },
    {
      "epoch": 0.4185386005680351,
      "grad_norm": 11.40701675415039,
      "learning_rate": 6.354865868024823e-06,
      "loss": 0.0961,
      "step": 8105
    },
    {
      "epoch": 0.4185902401239349,
      "grad_norm": 6.537300109863281,
      "learning_rate": 6.354077143996428e-06,
      "loss": 0.0831,
      "step": 8106
    },
    {
      "epoch": 0.41864187967983474,
      "grad_norm": 5.276995658874512,
      "learning_rate": 6.3532883836055435e-06,
      "loss": 0.0904,
      "step": 8107
    },
    {
      "epoch": 0.41869351923573456,
      "grad_norm": 3.0175061225891113,
      "learning_rate": 6.352499586873348e-06,
      "loss": 0.0647,
      "step": 8108
    },
    {
      "epoch": 0.4187451587916344,
      "grad_norm": 2.837634801864624,
      "learning_rate": 6.35171075382103e-06,
      "loss": 0.0359,
      "step": 8109
    },
    {
      "epoch": 0.4187967983475342,
      "grad_norm": 4.618889331817627,
      "learning_rate": 6.350921884469768e-06,
      "loss": 0.0679,
      "step": 8110
    },
    {
      "epoch": 0.418848437903434,
      "grad_norm": 3.437870979309082,
      "learning_rate": 6.350132978840745e-06,
      "loss": 0.0508,
      "step": 8111
    },
    {
      "epoch": 0.41890007745933383,
      "grad_norm": 2.5084164142608643,
      "learning_rate": 6.3493440369551495e-06,
      "loss": 0.0287,
      "step": 8112
    },
    {
      "epoch": 0.41895171701523365,
      "grad_norm": 8.939214706420898,
      "learning_rate": 6.348555058834168e-06,
      "loss": 0.2057,
      "step": 8113
    },
    {
      "epoch": 0.41900335657113347,
      "grad_norm": 3.8196799755096436,
      "learning_rate": 6.347766044498988e-06,
      "loss": 0.0679,
      "step": 8114
    },
    {
      "epoch": 0.4190549961270333,
      "grad_norm": 4.0092549324035645,
      "learning_rate": 6.346976993970792e-06,
      "loss": 0.0689,
      "step": 8115
    },
    {
      "epoch": 0.4191066356829331,
      "grad_norm": 6.563676834106445,
      "learning_rate": 6.3461879072707776e-06,
      "loss": 0.08,
      "step": 8116
    },
    {
      "epoch": 0.4191582752388329,
      "grad_norm": 2.722853660583496,
      "learning_rate": 6.345398784420132e-06,
      "loss": 0.0478,
      "step": 8117
    },
    {
      "epoch": 0.41920991479473274,
      "grad_norm": 16.006887435913086,
      "learning_rate": 6.344609625440043e-06,
      "loss": 0.0601,
      "step": 8118
    },
    {
      "epoch": 0.41926155435063256,
      "grad_norm": 7.959610462188721,
      "learning_rate": 6.343820430351705e-06,
      "loss": 0.0819,
      "step": 8119
    },
    {
      "epoch": 0.4193131939065324,
      "grad_norm": 4.133063316345215,
      "learning_rate": 6.343031199176312e-06,
      "loss": 0.0731,
      "step": 8120
    },
    {
      "epoch": 0.4193648334624322,
      "grad_norm": 7.036531448364258,
      "learning_rate": 6.342241931935058e-06,
      "loss": 0.1099,
      "step": 8121
    },
    {
      "epoch": 0.41941647301833207,
      "grad_norm": 1.6577407121658325,
      "learning_rate": 6.341452628649136e-06,
      "loss": 0.0201,
      "step": 8122
    },
    {
      "epoch": 0.4194681125742319,
      "grad_norm": 2.714820623397827,
      "learning_rate": 6.340663289339744e-06,
      "loss": 0.0513,
      "step": 8123
    },
    {
      "epoch": 0.4195197521301317,
      "grad_norm": 3.5424880981445312,
      "learning_rate": 6.339873914028081e-06,
      "loss": 0.0599,
      "step": 8124
    },
    {
      "epoch": 0.4195713916860315,
      "grad_norm": 2.5459988117218018,
      "learning_rate": 6.33908450273534e-06,
      "loss": 0.0368,
      "step": 8125
    },
    {
      "epoch": 0.41962303124193134,
      "grad_norm": 12.240992546081543,
      "learning_rate": 6.338295055482722e-06,
      "loss": 0.0922,
      "step": 8126
    },
    {
      "epoch": 0.41967467079783116,
      "grad_norm": 3.791640281677246,
      "learning_rate": 6.337505572291428e-06,
      "loss": 0.0892,
      "step": 8127
    },
    {
      "epoch": 0.419726310353731,
      "grad_norm": 2.3310351371765137,
      "learning_rate": 6.336716053182658e-06,
      "loss": 0.03,
      "step": 8128
    },
    {
      "epoch": 0.4197779499096308,
      "grad_norm": 4.327939033508301,
      "learning_rate": 6.3359264981776135e-06,
      "loss": 0.0465,
      "step": 8129
    },
    {
      "epoch": 0.4198295894655306,
      "grad_norm": 5.7221221923828125,
      "learning_rate": 6.3351369072974965e-06,
      "loss": 0.0874,
      "step": 8130
    },
    {
      "epoch": 0.41988122902143044,
      "grad_norm": 2.135674476623535,
      "learning_rate": 6.334347280563514e-06,
      "loss": 0.029,
      "step": 8131
    },
    {
      "epoch": 0.41993286857733025,
      "grad_norm": 4.140853404998779,
      "learning_rate": 6.333557617996867e-06,
      "loss": 0.0291,
      "step": 8132
    },
    {
      "epoch": 0.4199845081332301,
      "grad_norm": 5.666293144226074,
      "learning_rate": 6.332767919618764e-06,
      "loss": 0.0539,
      "step": 8133
    },
    {
      "epoch": 0.4200361476891299,
      "grad_norm": 7.732719421386719,
      "learning_rate": 6.331978185450409e-06,
      "loss": 0.1034,
      "step": 8134
    },
    {
      "epoch": 0.4200877872450297,
      "grad_norm": 1.7184669971466064,
      "learning_rate": 6.331188415513012e-06,
      "loss": 0.0237,
      "step": 8135
    },
    {
      "epoch": 0.4201394268009295,
      "grad_norm": 2.1328203678131104,
      "learning_rate": 6.330398609827779e-06,
      "loss": 0.029,
      "step": 8136
    },
    {
      "epoch": 0.42019106635682935,
      "grad_norm": 14.240873336791992,
      "learning_rate": 6.329608768415923e-06,
      "loss": 0.0735,
      "step": 8137
    },
    {
      "epoch": 0.42024270591272916,
      "grad_norm": 12.084718704223633,
      "learning_rate": 6.3288188912986514e-06,
      "loss": 0.0465,
      "step": 8138
    },
    {
      "epoch": 0.420294345468629,
      "grad_norm": 3.4451355934143066,
      "learning_rate": 6.3280289784971746e-06,
      "loss": 0.0224,
      "step": 8139
    },
    {
      "epoch": 0.4203459850245288,
      "grad_norm": 2.5692999362945557,
      "learning_rate": 6.32723903003271e-06,
      "loss": 0.0679,
      "step": 8140
    },
    {
      "epoch": 0.4203976245804286,
      "grad_norm": 4.364395618438721,
      "learning_rate": 6.326449045926465e-06,
      "loss": 0.0348,
      "step": 8141
    },
    {
      "epoch": 0.42044926413632844,
      "grad_norm": 3.1778807640075684,
      "learning_rate": 6.32565902619966e-06,
      "loss": 0.0543,
      "step": 8142
    },
    {
      "epoch": 0.42050090369222826,
      "grad_norm": 3.85063099861145,
      "learning_rate": 6.3248689708735045e-06,
      "loss": 0.0618,
      "step": 8143
    },
    {
      "epoch": 0.4205525432481281,
      "grad_norm": 7.13122034072876,
      "learning_rate": 6.324078879969219e-06,
      "loss": 0.0913,
      "step": 8144
    },
    {
      "epoch": 0.4206041828040279,
      "grad_norm": 6.178515434265137,
      "learning_rate": 6.323288753508019e-06,
      "loss": 0.0744,
      "step": 8145
    },
    {
      "epoch": 0.4206558223599277,
      "grad_norm": 15.11433219909668,
      "learning_rate": 6.3224985915111205e-06,
      "loss": 0.0737,
      "step": 8146
    },
    {
      "epoch": 0.42070746191582753,
      "grad_norm": 9.573338508605957,
      "learning_rate": 6.321708393999746e-06,
      "loss": 0.1214,
      "step": 8147
    },
    {
      "epoch": 0.42075910147172735,
      "grad_norm": 5.093291759490967,
      "learning_rate": 6.3209181609951135e-06,
      "loss": 0.0766,
      "step": 8148
    },
    {
      "epoch": 0.42081074102762717,
      "grad_norm": 6.06862735748291,
      "learning_rate": 6.320127892518446e-06,
      "loss": 0.1176,
      "step": 8149
    },
    {
      "epoch": 0.420862380583527,
      "grad_norm": 8.889730453491211,
      "learning_rate": 6.319337588590962e-06,
      "loss": 0.1177,
      "step": 8150
    },
    {
      "epoch": 0.4209140201394268,
      "grad_norm": 4.388431549072266,
      "learning_rate": 6.318547249233888e-06,
      "loss": 0.0469,
      "step": 8151
    },
    {
      "epoch": 0.4209656596953266,
      "grad_norm": 6.822268486022949,
      "learning_rate": 6.317756874468446e-06,
      "loss": 0.1043,
      "step": 8152
    },
    {
      "epoch": 0.42101729925122644,
      "grad_norm": 3.199284076690674,
      "learning_rate": 6.3169664643158604e-06,
      "loss": 0.0378,
      "step": 8153
    },
    {
      "epoch": 0.42106893880712626,
      "grad_norm": 4.755743980407715,
      "learning_rate": 6.316176018797358e-06,
      "loss": 0.0455,
      "step": 8154
    },
    {
      "epoch": 0.4211205783630261,
      "grad_norm": 2.3729519844055176,
      "learning_rate": 6.315385537934165e-06,
      "loss": 0.0424,
      "step": 8155
    },
    {
      "epoch": 0.4211722179189259,
      "grad_norm": 5.3819661140441895,
      "learning_rate": 6.314595021747509e-06,
      "loss": 0.0345,
      "step": 8156
    },
    {
      "epoch": 0.4212238574748257,
      "grad_norm": 5.642824172973633,
      "learning_rate": 6.31380447025862e-06,
      "loss": 0.0988,
      "step": 8157
    },
    {
      "epoch": 0.42127549703072553,
      "grad_norm": 12.809242248535156,
      "learning_rate": 6.3130138834887245e-06,
      "loss": 0.1254,
      "step": 8158
    },
    {
      "epoch": 0.42132713658662535,
      "grad_norm": 2.6194589138031006,
      "learning_rate": 6.312223261459055e-06,
      "loss": 0.0365,
      "step": 8159
    },
    {
      "epoch": 0.42137877614252517,
      "grad_norm": 4.279966354370117,
      "learning_rate": 6.311432604190844e-06,
      "loss": 0.0477,
      "step": 8160
    },
    {
      "epoch": 0.421430415698425,
      "grad_norm": 3.988537311553955,
      "learning_rate": 6.3106419117053195e-06,
      "loss": 0.0488,
      "step": 8161
    },
    {
      "epoch": 0.4214820552543248,
      "grad_norm": 2.3396663665771484,
      "learning_rate": 6.30985118402372e-06,
      "loss": 0.0293,
      "step": 8162
    },
    {
      "epoch": 0.4215336948102246,
      "grad_norm": 6.375918865203857,
      "learning_rate": 6.309060421167277e-06,
      "loss": 0.0847,
      "step": 8163
    },
    {
      "epoch": 0.42158533436612444,
      "grad_norm": 2.279365301132202,
      "learning_rate": 6.308269623157227e-06,
      "loss": 0.0535,
      "step": 8164
    },
    {
      "epoch": 0.42163697392202426,
      "grad_norm": 7.494387149810791,
      "learning_rate": 6.307478790014804e-06,
      "loss": 0.1023,
      "step": 8165
    },
    {
      "epoch": 0.4216886134779241,
      "grad_norm": 6.892366886138916,
      "learning_rate": 6.306687921761248e-06,
      "loss": 0.0654,
      "step": 8166
    },
    {
      "epoch": 0.4217402530338239,
      "grad_norm": 10.3541259765625,
      "learning_rate": 6.305897018417795e-06,
      "loss": 0.179,
      "step": 8167
    },
    {
      "epoch": 0.4217918925897237,
      "grad_norm": 15.201872825622559,
      "learning_rate": 6.305106080005684e-06,
      "loss": 0.1618,
      "step": 8168
    },
    {
      "epoch": 0.42184353214562353,
      "grad_norm": 9.08816146850586,
      "learning_rate": 6.304315106546156e-06,
      "loss": 0.1079,
      "step": 8169
    },
    {
      "epoch": 0.42189517170152335,
      "grad_norm": 5.819034099578857,
      "learning_rate": 6.303524098060451e-06,
      "loss": 0.0819,
      "step": 8170
    },
    {
      "epoch": 0.42194681125742317,
      "grad_norm": 5.654734134674072,
      "learning_rate": 6.302733054569811e-06,
      "loss": 0.0618,
      "step": 8171
    },
    {
      "epoch": 0.421998450813323,
      "grad_norm": 5.586622714996338,
      "learning_rate": 6.301941976095478e-06,
      "loss": 0.0448,
      "step": 8172
    },
    {
      "epoch": 0.4220500903692228,
      "grad_norm": 2.31811261177063,
      "learning_rate": 6.301150862658699e-06,
      "loss": 0.0373,
      "step": 8173
    },
    {
      "epoch": 0.4221017299251226,
      "grad_norm": 1.4357517957687378,
      "learning_rate": 6.3003597142807135e-06,
      "loss": 0.0178,
      "step": 8174
    },
    {
      "epoch": 0.42215336948102244,
      "grad_norm": 17.282297134399414,
      "learning_rate": 6.2995685309827714e-06,
      "loss": 0.146,
      "step": 8175
    },
    {
      "epoch": 0.42220500903692226,
      "grad_norm": 8.21143913269043,
      "learning_rate": 6.298777312786117e-06,
      "loss": 0.0937,
      "step": 8176
    },
    {
      "epoch": 0.4222566485928221,
      "grad_norm": 5.304648399353027,
      "learning_rate": 6.297986059711998e-06,
      "loss": 0.1021,
      "step": 8177
    },
    {
      "epoch": 0.4223082881487219,
      "grad_norm": 3.0560028553009033,
      "learning_rate": 6.297194771781665e-06,
      "loss": 0.0386,
      "step": 8178
    },
    {
      "epoch": 0.4223599277046217,
      "grad_norm": 6.5678486824035645,
      "learning_rate": 6.296403449016362e-06,
      "loss": 0.0508,
      "step": 8179
    },
    {
      "epoch": 0.42241156726052154,
      "grad_norm": 5.785799980163574,
      "learning_rate": 6.2956120914373445e-06,
      "loss": 0.0402,
      "step": 8180
    },
    {
      "epoch": 0.42246320681642135,
      "grad_norm": 5.1951494216918945,
      "learning_rate": 6.29482069906586e-06,
      "loss": 0.0581,
      "step": 8181
    },
    {
      "epoch": 0.4225148463723212,
      "grad_norm": 4.116623878479004,
      "learning_rate": 6.294029271923165e-06,
      "loss": 0.0724,
      "step": 8182
    },
    {
      "epoch": 0.422566485928221,
      "grad_norm": 2.400690793991089,
      "learning_rate": 6.293237810030508e-06,
      "loss": 0.0197,
      "step": 8183
    },
    {
      "epoch": 0.4226181254841208,
      "grad_norm": 4.865714073181152,
      "learning_rate": 6.292446313409145e-06,
      "loss": 0.0404,
      "step": 8184
    },
    {
      "epoch": 0.42266976504002063,
      "grad_norm": 6.9544501304626465,
      "learning_rate": 6.291654782080332e-06,
      "loss": 0.0745,
      "step": 8185
    },
    {
      "epoch": 0.4227214045959205,
      "grad_norm": 2.522458553314209,
      "learning_rate": 6.290863216065321e-06,
      "loss": 0.0456,
      "step": 8186
    },
    {
      "epoch": 0.4227730441518203,
      "grad_norm": 9.976761817932129,
      "learning_rate": 6.290071615385374e-06,
      "loss": 0.1055,
      "step": 8187
    },
    {
      "epoch": 0.42282468370772014,
      "grad_norm": 4.437987327575684,
      "learning_rate": 6.289279980061745e-06,
      "loss": 0.0281,
      "step": 8188
    },
    {
      "epoch": 0.42287632326361996,
      "grad_norm": 10.354000091552734,
      "learning_rate": 6.288488310115695e-06,
      "loss": 0.141,
      "step": 8189
    },
    {
      "epoch": 0.4229279628195198,
      "grad_norm": 2.464442253112793,
      "learning_rate": 6.2876966055684805e-06,
      "loss": 0.0164,
      "step": 8190
    },
    {
      "epoch": 0.4229796023754196,
      "grad_norm": 3.3421545028686523,
      "learning_rate": 6.286904866441365e-06,
      "loss": 0.0566,
      "step": 8191
    },
    {
      "epoch": 0.4230312419313194,
      "grad_norm": 3.298427104949951,
      "learning_rate": 6.286113092755608e-06,
      "loss": 0.0484,
      "step": 8192
    },
    {
      "epoch": 0.42308288148721923,
      "grad_norm": 3.881664276123047,
      "learning_rate": 6.285321284532473e-06,
      "loss": 0.0627,
      "step": 8193
    },
    {
      "epoch": 0.42313452104311905,
      "grad_norm": 3.6485025882720947,
      "learning_rate": 6.284529441793221e-06,
      "loss": 0.0409,
      "step": 8194
    },
    {
      "epoch": 0.42318616059901887,
      "grad_norm": 3.1683876514434814,
      "learning_rate": 6.2837375645591205e-06,
      "loss": 0.0578,
      "step": 8195
    },
    {
      "epoch": 0.4232378001549187,
      "grad_norm": 3.657475471496582,
      "learning_rate": 6.282945652851433e-06,
      "loss": 0.0417,
      "step": 8196
    },
    {
      "epoch": 0.4232894397108185,
      "grad_norm": 10.750496864318848,
      "learning_rate": 6.282153706691427e-06,
      "loss": 0.1168,
      "step": 8197
    },
    {
      "epoch": 0.4233410792667183,
      "grad_norm": 8.57402515411377,
      "learning_rate": 6.281361726100366e-06,
      "loss": 0.0831,
      "step": 8198
    },
    {
      "epoch": 0.42339271882261814,
      "grad_norm": 5.895503044128418,
      "learning_rate": 6.2805697110995225e-06,
      "loss": 0.0857,
      "step": 8199
    },
    {
      "epoch": 0.42344435837851796,
      "grad_norm": 6.863577842712402,
      "learning_rate": 6.2797776617101615e-06,
      "loss": 0.0909,
      "step": 8200
    },
    {
      "epoch": 0.4234959979344178,
      "grad_norm": 4.5212082862854,
      "learning_rate": 6.278985577953553e-06,
      "loss": 0.093,
      "step": 8201
    },
    {
      "epoch": 0.4235476374903176,
      "grad_norm": 5.303221702575684,
      "learning_rate": 6.278193459850971e-06,
      "loss": 0.038,
      "step": 8202
    },
    {
      "epoch": 0.4235992770462174,
      "grad_norm": 6.326028823852539,
      "learning_rate": 6.277401307423684e-06,
      "loss": 0.0544,
      "step": 8203
    },
    {
      "epoch": 0.42365091660211723,
      "grad_norm": 5.238138675689697,
      "learning_rate": 6.276609120692967e-06,
      "loss": 0.0857,
      "step": 8204
    },
    {
      "epoch": 0.42370255615801705,
      "grad_norm": 2.0042717456817627,
      "learning_rate": 6.275816899680089e-06,
      "loss": 0.0368,
      "step": 8205
    },
    {
      "epoch": 0.42375419571391687,
      "grad_norm": 8.594839096069336,
      "learning_rate": 6.27502464440633e-06,
      "loss": 0.0709,
      "step": 8206
    },
    {
      "epoch": 0.4238058352698167,
      "grad_norm": 3.3103818893432617,
      "learning_rate": 6.274232354892961e-06,
      "loss": 0.0528,
      "step": 8207
    },
    {
      "epoch": 0.4238574748257165,
      "grad_norm": 3.3459553718566895,
      "learning_rate": 6.273440031161261e-06,
      "loss": 0.0472,
      "step": 8208
    },
    {
      "epoch": 0.4239091143816163,
      "grad_norm": 11.198323249816895,
      "learning_rate": 6.272647673232506e-06,
      "loss": 0.1036,
      "step": 8209
    },
    {
      "epoch": 0.42396075393751614,
      "grad_norm": 2.7701079845428467,
      "learning_rate": 6.271855281127972e-06,
      "loss": 0.0519,
      "step": 8210
    },
    {
      "epoch": 0.42401239349341596,
      "grad_norm": 2.762864351272583,
      "learning_rate": 6.271062854868942e-06,
      "loss": 0.0291,
      "step": 8211
    },
    {
      "epoch": 0.4240640330493158,
      "grad_norm": 1.8916165828704834,
      "learning_rate": 6.270270394476692e-06,
      "loss": 0.0294,
      "step": 8212
    },
    {
      "epoch": 0.4241156726052156,
      "grad_norm": 3.602412700653076,
      "learning_rate": 6.269477899972507e-06,
      "loss": 0.036,
      "step": 8213
    },
    {
      "epoch": 0.4241673121611154,
      "grad_norm": 2.768305778503418,
      "learning_rate": 6.268685371377663e-06,
      "loss": 0.0474,
      "step": 8214
    },
    {
      "epoch": 0.42421895171701524,
      "grad_norm": 7.234560012817383,
      "learning_rate": 6.267892808713449e-06,
      "loss": 0.0612,
      "step": 8215
    },
    {
      "epoch": 0.42427059127291505,
      "grad_norm": 4.157402038574219,
      "learning_rate": 6.2671002120011435e-06,
      "loss": 0.0759,
      "step": 8216
    },
    {
      "epoch": 0.4243222308288149,
      "grad_norm": 3.93926739692688,
      "learning_rate": 6.266307581262035e-06,
      "loss": 0.0813,
      "step": 8217
    },
    {
      "epoch": 0.4243738703847147,
      "grad_norm": 4.396510601043701,
      "learning_rate": 6.265514916517403e-06,
      "loss": 0.0646,
      "step": 8218
    },
    {
      "epoch": 0.4244255099406145,
      "grad_norm": 2.967012405395508,
      "learning_rate": 6.26472221778854e-06,
      "loss": 0.0325,
      "step": 8219
    },
    {
      "epoch": 0.4244771494965143,
      "grad_norm": 8.646254539489746,
      "learning_rate": 6.2639294850967316e-06,
      "loss": 0.0333,
      "step": 8220
    },
    {
      "epoch": 0.42452878905241415,
      "grad_norm": 5.578742980957031,
      "learning_rate": 6.263136718463263e-06,
      "loss": 0.1355,
      "step": 8221
    },
    {
      "epoch": 0.42458042860831396,
      "grad_norm": 2.6547157764434814,
      "learning_rate": 6.262343917909428e-06,
      "loss": 0.0397,
      "step": 8222
    },
    {
      "epoch": 0.4246320681642138,
      "grad_norm": 2.632378339767456,
      "learning_rate": 6.261551083456512e-06,
      "loss": 0.0493,
      "step": 8223
    },
    {
      "epoch": 0.4246837077201136,
      "grad_norm": 14.20128345489502,
      "learning_rate": 6.260758215125807e-06,
      "loss": 0.088,
      "step": 8224
    },
    {
      "epoch": 0.4247353472760134,
      "grad_norm": 3.030291795730591,
      "learning_rate": 6.259965312938605e-06,
      "loss": 0.066,
      "step": 8225
    },
    {
      "epoch": 0.42478698683191324,
      "grad_norm": 3.5011956691741943,
      "learning_rate": 6.259172376916201e-06,
      "loss": 0.0411,
      "step": 8226
    },
    {
      "epoch": 0.42483862638781306,
      "grad_norm": 6.01527214050293,
      "learning_rate": 6.258379407079885e-06,
      "loss": 0.067,
      "step": 8227
    },
    {
      "epoch": 0.4248902659437129,
      "grad_norm": 4.363064289093018,
      "learning_rate": 6.2575864034509535e-06,
      "loss": 0.047,
      "step": 8228
    },
    {
      "epoch": 0.4249419054996127,
      "grad_norm": 3.0688083171844482,
      "learning_rate": 6.2567933660507015e-06,
      "loss": 0.0383,
      "step": 8229
    },
    {
      "epoch": 0.4249935450555125,
      "grad_norm": 2.871236801147461,
      "learning_rate": 6.256000294900426e-06,
      "loss": 0.0413,
      "step": 8230
    },
    {
      "epoch": 0.42504518461141233,
      "grad_norm": 6.1482062339782715,
      "learning_rate": 6.255207190021421e-06,
      "loss": 0.0627,
      "step": 8231
    },
    {
      "epoch": 0.42509682416731215,
      "grad_norm": 6.285787582397461,
      "learning_rate": 6.25441405143499e-06,
      "loss": 0.0992,
      "step": 8232
    },
    {
      "epoch": 0.42514846372321197,
      "grad_norm": 4.432332992553711,
      "learning_rate": 6.253620879162428e-06,
      "loss": 0.0742,
      "step": 8233
    },
    {
      "epoch": 0.4252001032791118,
      "grad_norm": 12.020630836486816,
      "learning_rate": 6.252827673225035e-06,
      "loss": 0.1186,
      "step": 8234
    },
    {
      "epoch": 0.4252517428350116,
      "grad_norm": 9.08611011505127,
      "learning_rate": 6.252034433644114e-06,
      "loss": 0.0807,
      "step": 8235
    },
    {
      "epoch": 0.4253033823909114,
      "grad_norm": 7.352167129516602,
      "learning_rate": 6.251241160440964e-06,
      "loss": 0.0652,
      "step": 8236
    },
    {
      "epoch": 0.42535502194681124,
      "grad_norm": 2.9219017028808594,
      "learning_rate": 6.250447853636891e-06,
      "loss": 0.0584,
      "step": 8237
    },
    {
      "epoch": 0.42540666150271106,
      "grad_norm": 3.946847438812256,
      "learning_rate": 6.2496545132531935e-06,
      "loss": 0.0573,
      "step": 8238
    },
    {
      "epoch": 0.4254583010586109,
      "grad_norm": 4.0584564208984375,
      "learning_rate": 6.248861139311182e-06,
      "loss": 0.0459,
      "step": 8239
    },
    {
      "epoch": 0.4255099406145107,
      "grad_norm": 2.287903070449829,
      "learning_rate": 6.248067731832157e-06,
      "loss": 0.0366,
      "step": 8240
    },
    {
      "epoch": 0.4255615801704105,
      "grad_norm": 5.638911247253418,
      "learning_rate": 6.2472742908374284e-06,
      "loss": 0.0386,
      "step": 8241
    },
    {
      "epoch": 0.42561321972631033,
      "grad_norm": 3.607661008834839,
      "learning_rate": 6.2464808163483e-06,
      "loss": 0.0512,
      "step": 8242
    },
    {
      "epoch": 0.42566485928221015,
      "grad_norm": 6.270362854003906,
      "learning_rate": 6.245687308386081e-06,
      "loss": 0.0507,
      "step": 8243
    },
    {
      "epoch": 0.42571649883810997,
      "grad_norm": 11.60391616821289,
      "learning_rate": 6.2448937669720836e-06,
      "loss": 0.2758,
      "step": 8244
    },
    {
      "epoch": 0.4257681383940098,
      "grad_norm": 3.377446413040161,
      "learning_rate": 6.244100192127611e-06,
      "loss": 0.0655,
      "step": 8245
    },
    {
      "epoch": 0.4258197779499096,
      "grad_norm": 2.997814893722534,
      "learning_rate": 6.2433065838739795e-06,
      "loss": 0.0232,
      "step": 8246
    },
    {
      "epoch": 0.4258714175058094,
      "grad_norm": 3.917966842651367,
      "learning_rate": 6.242512942232497e-06,
      "loss": 0.0504,
      "step": 8247
    },
    {
      "epoch": 0.42592305706170924,
      "grad_norm": 12.66628360748291,
      "learning_rate": 6.241719267224479e-06,
      "loss": 0.1498,
      "step": 8248
    },
    {
      "epoch": 0.42597469661760906,
      "grad_norm": 3.8108763694763184,
      "learning_rate": 6.240925558871239e-06,
      "loss": 0.0528,
      "step": 8249
    },
    {
      "epoch": 0.42602633617350893,
      "grad_norm": 14.615692138671875,
      "learning_rate": 6.240131817194089e-06,
      "loss": 0.1135,
      "step": 8250
    },
    {
      "epoch": 0.42607797572940875,
      "grad_norm": 3.6343588829040527,
      "learning_rate": 6.2393380422143445e-06,
      "loss": 0.0718,
      "step": 8251
    },
    {
      "epoch": 0.42612961528530857,
      "grad_norm": 2.515071392059326,
      "learning_rate": 6.238544233953323e-06,
      "loss": 0.0757,
      "step": 8252
    },
    {
      "epoch": 0.4261812548412084,
      "grad_norm": 4.152977466583252,
      "learning_rate": 6.23775039243234e-06,
      "loss": 0.057,
      "step": 8253
    },
    {
      "epoch": 0.4262328943971082,
      "grad_norm": 4.292190074920654,
      "learning_rate": 6.236956517672714e-06,
      "loss": 0.0571,
      "step": 8254
    },
    {
      "epoch": 0.426284533953008,
      "grad_norm": 2.9105796813964844,
      "learning_rate": 6.236162609695766e-06,
      "loss": 0.0757,
      "step": 8255
    },
    {
      "epoch": 0.42633617350890785,
      "grad_norm": 2.7347590923309326,
      "learning_rate": 6.235368668522812e-06,
      "loss": 0.0499,
      "step": 8256
    },
    {
      "epoch": 0.42638781306480766,
      "grad_norm": 1.8130820989608765,
      "learning_rate": 6.234574694175175e-06,
      "loss": 0.0342,
      "step": 8257
    },
    {
      "epoch": 0.4264394526207075,
      "grad_norm": 4.522795677185059,
      "learning_rate": 6.233780686674175e-06,
      "loss": 0.0565,
      "step": 8258
    },
    {
      "epoch": 0.4264910921766073,
      "grad_norm": 9.386220932006836,
      "learning_rate": 6.232986646041135e-06,
      "loss": 0.0752,
      "step": 8259
    },
    {
      "epoch": 0.4265427317325071,
      "grad_norm": 2.3272016048431396,
      "learning_rate": 6.232192572297377e-06,
      "loss": 0.0284,
      "step": 8260
    },
    {
      "epoch": 0.42659437128840694,
      "grad_norm": 1.2481039762496948,
      "learning_rate": 6.231398465464228e-06,
      "loss": 0.0168,
      "step": 8261
    },
    {
      "epoch": 0.42664601084430676,
      "grad_norm": 2.871997117996216,
      "learning_rate": 6.230604325563011e-06,
      "loss": 0.0534,
      "step": 8262
    },
    {
      "epoch": 0.4266976504002066,
      "grad_norm": 2.600229024887085,
      "learning_rate": 6.229810152615052e-06,
      "loss": 0.0619,
      "step": 8263
    },
    {
      "epoch": 0.4267492899561064,
      "grad_norm": 5.362118244171143,
      "learning_rate": 6.2290159466416775e-06,
      "loss": 0.0517,
      "step": 8264
    },
    {
      "epoch": 0.4268009295120062,
      "grad_norm": 12.063247680664062,
      "learning_rate": 6.228221707664217e-06,
      "loss": 0.1003,
      "step": 8265
    },
    {
      "epoch": 0.42685256906790603,
      "grad_norm": 5.9434027671813965,
      "learning_rate": 6.227427435703997e-06,
      "loss": 0.1172,
      "step": 8266
    },
    {
      "epoch": 0.42690420862380585,
      "grad_norm": 10.063441276550293,
      "learning_rate": 6.226633130782347e-06,
      "loss": 0.1346,
      "step": 8267
    },
    {
      "epoch": 0.42695584817970567,
      "grad_norm": 4.091145992279053,
      "learning_rate": 6.225838792920598e-06,
      "loss": 0.0667,
      "step": 8268
    },
    {
      "epoch": 0.4270074877356055,
      "grad_norm": 3.444845199584961,
      "learning_rate": 6.2250444221400816e-06,
      "loss": 0.0482,
      "step": 8269
    },
    {
      "epoch": 0.4270591272915053,
      "grad_norm": 5.143360614776611,
      "learning_rate": 6.22425001846213e-06,
      "loss": 0.1041,
      "step": 8270
    },
    {
      "epoch": 0.4271107668474051,
      "grad_norm": 2.5959064960479736,
      "learning_rate": 6.223455581908075e-06,
      "loss": 0.0389,
      "step": 8271
    },
    {
      "epoch": 0.42716240640330494,
      "grad_norm": 4.557258605957031,
      "learning_rate": 6.222661112499251e-06,
      "loss": 0.0583,
      "step": 8272
    },
    {
      "epoch": 0.42721404595920476,
      "grad_norm": 5.878100395202637,
      "learning_rate": 6.221866610256993e-06,
      "loss": 0.1007,
      "step": 8273
    },
    {
      "epoch": 0.4272656855151046,
      "grad_norm": 6.547023773193359,
      "learning_rate": 6.221072075202637e-06,
      "loss": 0.0753,
      "step": 8274
    },
    {
      "epoch": 0.4273173250710044,
      "grad_norm": 7.450534820556641,
      "learning_rate": 6.2202775073575184e-06,
      "loss": 0.1444,
      "step": 8275
    },
    {
      "epoch": 0.4273689646269042,
      "grad_norm": 4.846990585327148,
      "learning_rate": 6.219482906742975e-06,
      "loss": 0.0761,
      "step": 8276
    },
    {
      "epoch": 0.42742060418280403,
      "grad_norm": 16.785934448242188,
      "learning_rate": 6.2186882733803445e-06,
      "loss": 0.1569,
      "step": 8277
    },
    {
      "epoch": 0.42747224373870385,
      "grad_norm": 2.0387656688690186,
      "learning_rate": 6.217893607290968e-06,
      "loss": 0.0208,
      "step": 8278
    },
    {
      "epoch": 0.42752388329460367,
      "grad_norm": 12.837311744689941,
      "learning_rate": 6.217098908496185e-06,
      "loss": 0.2558,
      "step": 8279
    },
    {
      "epoch": 0.4275755228505035,
      "grad_norm": 2.2097041606903076,
      "learning_rate": 6.216304177017334e-06,
      "loss": 0.0383,
      "step": 8280
    },
    {
      "epoch": 0.4276271624064033,
      "grad_norm": 2.7124834060668945,
      "learning_rate": 6.21550941287576e-06,
      "loss": 0.0457,
      "step": 8281
    },
    {
      "epoch": 0.4276788019623031,
      "grad_norm": 9.1881742477417,
      "learning_rate": 6.214714616092803e-06,
      "loss": 0.095,
      "step": 8282
    },
    {
      "epoch": 0.42773044151820294,
      "grad_norm": 3.345924139022827,
      "learning_rate": 6.213919786689807e-06,
      "loss": 0.0579,
      "step": 8283
    },
    {
      "epoch": 0.42778208107410276,
      "grad_norm": 3.795280694961548,
      "learning_rate": 6.213124924688118e-06,
      "loss": 0.0811,
      "step": 8284
    },
    {
      "epoch": 0.4278337206300026,
      "grad_norm": 5.19576358795166,
      "learning_rate": 6.212330030109081e-06,
      "loss": 0.0696,
      "step": 8285
    },
    {
      "epoch": 0.4278853601859024,
      "grad_norm": 5.207030296325684,
      "learning_rate": 6.21153510297404e-06,
      "loss": 0.0359,
      "step": 8286
    },
    {
      "epoch": 0.4279369997418022,
      "grad_norm": 4.568475723266602,
      "learning_rate": 6.210740143304343e-06,
      "loss": 0.0869,
      "step": 8287
    },
    {
      "epoch": 0.42798863929770203,
      "grad_norm": 5.028355121612549,
      "learning_rate": 6.209945151121341e-06,
      "loss": 0.0838,
      "step": 8288
    },
    {
      "epoch": 0.42804027885360185,
      "grad_norm": 4.6619672775268555,
      "learning_rate": 6.209150126446379e-06,
      "loss": 0.0515,
      "step": 8289
    },
    {
      "epoch": 0.42809191840950167,
      "grad_norm": 4.820230007171631,
      "learning_rate": 6.208355069300809e-06,
      "loss": 0.1215,
      "step": 8290
    },
    {
      "epoch": 0.4281435579654015,
      "grad_norm": 5.211330413818359,
      "learning_rate": 6.2075599797059785e-06,
      "loss": 0.0929,
      "step": 8291
    },
    {
      "epoch": 0.4281951975213013,
      "grad_norm": 6.298826694488525,
      "learning_rate": 6.206764857683242e-06,
      "loss": 0.1452,
      "step": 8292
    },
    {
      "epoch": 0.4282468370772011,
      "grad_norm": 2.6269397735595703,
      "learning_rate": 6.205969703253949e-06,
      "loss": 0.0501,
      "step": 8293
    },
    {
      "epoch": 0.42829847663310094,
      "grad_norm": 7.399572849273682,
      "learning_rate": 6.205174516439456e-06,
      "loss": 0.0967,
      "step": 8294
    },
    {
      "epoch": 0.42835011618900076,
      "grad_norm": 3.474771022796631,
      "learning_rate": 6.204379297261114e-06,
      "loss": 0.0512,
      "step": 8295
    },
    {
      "epoch": 0.4284017557449006,
      "grad_norm": 2.035367012023926,
      "learning_rate": 6.20358404574028e-06,
      "loss": 0.0291,
      "step": 8296
    },
    {
      "epoch": 0.4284533953008004,
      "grad_norm": 4.567334175109863,
      "learning_rate": 6.202788761898309e-06,
      "loss": 0.0717,
      "step": 8297
    },
    {
      "epoch": 0.4285050348567002,
      "grad_norm": 2.860589027404785,
      "learning_rate": 6.201993445756557e-06,
      "loss": 0.0566,
      "step": 8298
    },
    {
      "epoch": 0.42855667441260004,
      "grad_norm": 2.4486324787139893,
      "learning_rate": 6.2011980973363816e-06,
      "loss": 0.0417,
      "step": 8299
    },
    {
      "epoch": 0.42860831396849985,
      "grad_norm": 2.8856637477874756,
      "learning_rate": 6.200402716659142e-06,
      "loss": 0.0479,
      "step": 8300
    },
    {
      "epoch": 0.4286599535243997,
      "grad_norm": 3.8907008171081543,
      "learning_rate": 6.199607303746197e-06,
      "loss": 0.0697,
      "step": 8301
    },
    {
      "epoch": 0.4287115930802995,
      "grad_norm": 2.781877279281616,
      "learning_rate": 6.198811858618905e-06,
      "loss": 0.0451,
      "step": 8302
    },
    {
      "epoch": 0.4287632326361993,
      "grad_norm": 2.9656503200531006,
      "learning_rate": 6.198016381298631e-06,
      "loss": 0.0745,
      "step": 8303
    },
    {
      "epoch": 0.42881487219209913,
      "grad_norm": 5.66721248626709,
      "learning_rate": 6.197220871806732e-06,
      "loss": 0.0708,
      "step": 8304
    },
    {
      "epoch": 0.42886651174799895,
      "grad_norm": 2.980456590652466,
      "learning_rate": 6.196425330164575e-06,
      "loss": 0.0706,
      "step": 8305
    },
    {
      "epoch": 0.42891815130389876,
      "grad_norm": 5.8127899169921875,
      "learning_rate": 6.19562975639352e-06,
      "loss": 0.0637,
      "step": 8306
    },
    {
      "epoch": 0.4289697908597986,
      "grad_norm": 3.8741095066070557,
      "learning_rate": 6.194834150514933e-06,
      "loss": 0.0524,
      "step": 8307
    },
    {
      "epoch": 0.4290214304156984,
      "grad_norm": 4.5211076736450195,
      "learning_rate": 6.19403851255018e-06,
      "loss": 0.0478,
      "step": 8308
    },
    {
      "epoch": 0.4290730699715982,
      "grad_norm": 9.759635925292969,
      "learning_rate": 6.193242842520626e-06,
      "loss": 0.1134,
      "step": 8309
    },
    {
      "epoch": 0.42912470952749804,
      "grad_norm": 6.734475135803223,
      "learning_rate": 6.192447140447638e-06,
      "loss": 0.0426,
      "step": 8310
    },
    {
      "epoch": 0.42917634908339786,
      "grad_norm": 5.121449947357178,
      "learning_rate": 6.191651406352585e-06,
      "loss": 0.065,
      "step": 8311
    },
    {
      "epoch": 0.4292279886392977,
      "grad_norm": 8.289196014404297,
      "learning_rate": 6.190855640256834e-06,
      "loss": 0.1189,
      "step": 8312
    },
    {
      "epoch": 0.42927962819519755,
      "grad_norm": 11.484190940856934,
      "learning_rate": 6.190059842181755e-06,
      "loss": 0.0793,
      "step": 8313
    },
    {
      "epoch": 0.42933126775109737,
      "grad_norm": 5.8566508293151855,
      "learning_rate": 6.18926401214872e-06,
      "loss": 0.0787,
      "step": 8314
    },
    {
      "epoch": 0.4293829073069972,
      "grad_norm": 3.5822644233703613,
      "learning_rate": 6.188468150179099e-06,
      "loss": 0.0739,
      "step": 8315
    },
    {
      "epoch": 0.429434546862897,
      "grad_norm": 2.4035725593566895,
      "learning_rate": 6.187672256294264e-06,
      "loss": 0.035,
      "step": 8316
    },
    {
      "epoch": 0.4294861864187968,
      "grad_norm": 2.7102701663970947,
      "learning_rate": 6.186876330515588e-06,
      "loss": 0.0497,
      "step": 8317
    },
    {
      "epoch": 0.42953782597469664,
      "grad_norm": 3.2517282962799072,
      "learning_rate": 6.186080372864445e-06,
      "loss": 0.0654,
      "step": 8318
    },
    {
      "epoch": 0.42958946553059646,
      "grad_norm": 2.9567954540252686,
      "learning_rate": 6.1852843833622104e-06,
      "loss": 0.0375,
      "step": 8319
    },
    {
      "epoch": 0.4296411050864963,
      "grad_norm": 2.591292142868042,
      "learning_rate": 6.1844883620302575e-06,
      "loss": 0.0338,
      "step": 8320
    },
    {
      "epoch": 0.4296927446423961,
      "grad_norm": 2.8529417514801025,
      "learning_rate": 6.183692308889967e-06,
      "loss": 0.0246,
      "step": 8321
    },
    {
      "epoch": 0.4297443841982959,
      "grad_norm": 5.247343063354492,
      "learning_rate": 6.18289622396271e-06,
      "loss": 0.0424,
      "step": 8322
    },
    {
      "epoch": 0.42979602375419573,
      "grad_norm": 1.5193383693695068,
      "learning_rate": 6.182100107269872e-06,
      "loss": 0.0218,
      "step": 8323
    },
    {
      "epoch": 0.42984766331009555,
      "grad_norm": 9.716737747192383,
      "learning_rate": 6.181303958832826e-06,
      "loss": 0.0349,
      "step": 8324
    },
    {
      "epoch": 0.42989930286599537,
      "grad_norm": 2.7309207916259766,
      "learning_rate": 6.180507778672954e-06,
      "loss": 0.0558,
      "step": 8325
    },
    {
      "epoch": 0.4299509424218952,
      "grad_norm": 11.05296516418457,
      "learning_rate": 6.179711566811637e-06,
      "loss": 0.1562,
      "step": 8326
    },
    {
      "epoch": 0.430002581977795,
      "grad_norm": 8.274296760559082,
      "learning_rate": 6.178915323270256e-06,
      "loss": 0.1204,
      "step": 8327
    },
    {
      "epoch": 0.4300542215336948,
      "grad_norm": 6.888729572296143,
      "learning_rate": 6.1781190480701935e-06,
      "loss": 0.0925,
      "step": 8328
    },
    {
      "epoch": 0.43010586108959464,
      "grad_norm": 17.53856086730957,
      "learning_rate": 6.177322741232833e-06,
      "loss": 0.1402,
      "step": 8329
    },
    {
      "epoch": 0.43015750064549446,
      "grad_norm": 1.584218978881836,
      "learning_rate": 6.17652640277956e-06,
      "loss": 0.0251,
      "step": 8330
    },
    {
      "epoch": 0.4302091402013943,
      "grad_norm": 4.9615912437438965,
      "learning_rate": 6.175730032731756e-06,
      "loss": 0.0609,
      "step": 8331
    },
    {
      "epoch": 0.4302607797572941,
      "grad_norm": 1.9321115016937256,
      "learning_rate": 6.1749336311108085e-06,
      "loss": 0.04,
      "step": 8332
    },
    {
      "epoch": 0.4303124193131939,
      "grad_norm": 6.897035598754883,
      "learning_rate": 6.174137197938105e-06,
      "loss": 0.0421,
      "step": 8333
    },
    {
      "epoch": 0.43036405886909374,
      "grad_norm": 12.065384864807129,
      "learning_rate": 6.173340733235031e-06,
      "loss": 0.095,
      "step": 8334
    },
    {
      "epoch": 0.43041569842499355,
      "grad_norm": 4.757604598999023,
      "learning_rate": 6.1725442370229775e-06,
      "loss": 0.0588,
      "step": 8335
    },
    {
      "epoch": 0.43046733798089337,
      "grad_norm": 3.9672913551330566,
      "learning_rate": 6.171747709323331e-06,
      "loss": 0.1032,
      "step": 8336
    },
    {
      "epoch": 0.4305189775367932,
      "grad_norm": 10.63889217376709,
      "learning_rate": 6.170951150157484e-06,
      "loss": 0.097,
      "step": 8337
    },
    {
      "epoch": 0.430570617092693,
      "grad_norm": 15.565240859985352,
      "learning_rate": 6.170154559546826e-06,
      "loss": 0.1063,
      "step": 8338
    },
    {
      "epoch": 0.4306222566485928,
      "grad_norm": 3.0556223392486572,
      "learning_rate": 6.169357937512748e-06,
      "loss": 0.0596,
      "step": 8339
    },
    {
      "epoch": 0.43067389620449265,
      "grad_norm": 18.965110778808594,
      "learning_rate": 6.168561284076644e-06,
      "loss": 0.0594,
      "step": 8340
    },
    {
      "epoch": 0.43072553576039246,
      "grad_norm": 7.424616813659668,
      "learning_rate": 6.167764599259906e-06,
      "loss": 0.1061,
      "step": 8341
    },
    {
      "epoch": 0.4307771753162923,
      "grad_norm": 3.2654974460601807,
      "learning_rate": 6.16696788308393e-06,
      "loss": 0.0617,
      "step": 8342
    },
    {
      "epoch": 0.4308288148721921,
      "grad_norm": 5.275254249572754,
      "learning_rate": 6.166171135570108e-06,
      "loss": 0.0736,
      "step": 8343
    },
    {
      "epoch": 0.4308804544280919,
      "grad_norm": 2.9954781532287598,
      "learning_rate": 6.165374356739839e-06,
      "loss": 0.0279,
      "step": 8344
    },
    {
      "epoch": 0.43093209398399174,
      "grad_norm": 10.683895111083984,
      "learning_rate": 6.164577546614521e-06,
      "loss": 0.072,
      "step": 8345
    },
    {
      "epoch": 0.43098373353989156,
      "grad_norm": 10.63056468963623,
      "learning_rate": 6.163780705215545e-06,
      "loss": 0.0827,
      "step": 8346
    },
    {
      "epoch": 0.4310353730957914,
      "grad_norm": 15.51634407043457,
      "learning_rate": 6.162983832564318e-06,
      "loss": 0.1593,
      "step": 8347
    },
    {
      "epoch": 0.4310870126516912,
      "grad_norm": 9.167920112609863,
      "learning_rate": 6.162186928682231e-06,
      "loss": 0.0661,
      "step": 8348
    },
    {
      "epoch": 0.431138652207591,
      "grad_norm": 11.271193504333496,
      "learning_rate": 6.1613899935906916e-06,
      "loss": 0.2169,
      "step": 8349
    },
    {
      "epoch": 0.43119029176349083,
      "grad_norm": 3.947460651397705,
      "learning_rate": 6.160593027311096e-06,
      "loss": 0.0575,
      "step": 8350
    },
    {
      "epoch": 0.43124193131939065,
      "grad_norm": 7.3339433670043945,
      "learning_rate": 6.1597960298648475e-06,
      "loss": 0.0804,
      "step": 8351
    },
    {
      "epoch": 0.43129357087529047,
      "grad_norm": 5.233100414276123,
      "learning_rate": 6.158999001273349e-06,
      "loss": 0.0573,
      "step": 8352
    },
    {
      "epoch": 0.4313452104311903,
      "grad_norm": 3.105621576309204,
      "learning_rate": 6.1582019415580015e-06,
      "loss": 0.0458,
      "step": 8353
    },
    {
      "epoch": 0.4313968499870901,
      "grad_norm": 10.000883102416992,
      "learning_rate": 6.157404850740214e-06,
      "loss": 0.0857,
      "step": 8354
    },
    {
      "epoch": 0.4314484895429899,
      "grad_norm": 8.040955543518066,
      "learning_rate": 6.156607728841386e-06,
      "loss": 0.0706,
      "step": 8355
    },
    {
      "epoch": 0.43150012909888974,
      "grad_norm": 3.8671584129333496,
      "learning_rate": 6.155810575882929e-06,
      "loss": 0.0819,
      "step": 8356
    },
    {
      "epoch": 0.43155176865478956,
      "grad_norm": 6.337374210357666,
      "learning_rate": 6.155013391886245e-06,
      "loss": 0.0797,
      "step": 8357
    },
    {
      "epoch": 0.4316034082106894,
      "grad_norm": 24.89757537841797,
      "learning_rate": 6.154216176872746e-06,
      "loss": 0.2253,
      "step": 8358
    },
    {
      "epoch": 0.4316550477665892,
      "grad_norm": 5.077173233032227,
      "learning_rate": 6.153418930863837e-06,
      "loss": 0.0788,
      "step": 8359
    },
    {
      "epoch": 0.431706687322489,
      "grad_norm": 4.319528102874756,
      "learning_rate": 6.152621653880929e-06,
      "loss": 0.0572,
      "step": 8360
    },
    {
      "epoch": 0.43175832687838883,
      "grad_norm": 3.6017956733703613,
      "learning_rate": 6.151824345945432e-06,
      "loss": 0.0375,
      "step": 8361
    },
    {
      "epoch": 0.43180996643428865,
      "grad_norm": 4.781859874725342,
      "learning_rate": 6.151027007078756e-06,
      "loss": 0.0577,
      "step": 8362
    },
    {
      "epoch": 0.43186160599018847,
      "grad_norm": 5.863338947296143,
      "learning_rate": 6.150229637302315e-06,
      "loss": 0.0769,
      "step": 8363
    },
    {
      "epoch": 0.4319132455460883,
      "grad_norm": 7.9391655921936035,
      "learning_rate": 6.149432236637519e-06,
      "loss": 0.0573,
      "step": 8364
    },
    {
      "epoch": 0.4319648851019881,
      "grad_norm": 2.016286611557007,
      "learning_rate": 6.148634805105782e-06,
      "loss": 0.0574,
      "step": 8365
    },
    {
      "epoch": 0.4320165246578879,
      "grad_norm": 2.847080945968628,
      "learning_rate": 6.1478373427285195e-06,
      "loss": 0.0516,
      "step": 8366
    },
    {
      "epoch": 0.43206816421378774,
      "grad_norm": 6.180270195007324,
      "learning_rate": 6.1470398495271465e-06,
      "loss": 0.0726,
      "step": 8367
    },
    {
      "epoch": 0.43211980376968756,
      "grad_norm": 1.9910178184509277,
      "learning_rate": 6.146242325523078e-06,
      "loss": 0.0449,
      "step": 8368
    },
    {
      "epoch": 0.4321714433255874,
      "grad_norm": 3.610579013824463,
      "learning_rate": 6.145444770737732e-06,
      "loss": 0.0468,
      "step": 8369
    },
    {
      "epoch": 0.4322230828814872,
      "grad_norm": 5.920179843902588,
      "learning_rate": 6.144647185192524e-06,
      "loss": 0.0673,
      "step": 8370
    },
    {
      "epoch": 0.432274722437387,
      "grad_norm": 4.449520111083984,
      "learning_rate": 6.1438495689088754e-06,
      "loss": 0.0242,
      "step": 8371
    },
    {
      "epoch": 0.43232636199328683,
      "grad_norm": 2.7309536933898926,
      "learning_rate": 6.1430519219082015e-06,
      "loss": 0.0477,
      "step": 8372
    },
    {
      "epoch": 0.43237800154918665,
      "grad_norm": 2.4179046154022217,
      "learning_rate": 6.142254244211927e-06,
      "loss": 0.0224,
      "step": 8373
    },
    {
      "epoch": 0.43242964110508647,
      "grad_norm": 4.065674304962158,
      "learning_rate": 6.1414565358414704e-06,
      "loss": 0.0551,
      "step": 8374
    },
    {
      "epoch": 0.4324812806609863,
      "grad_norm": 4.2344279289245605,
      "learning_rate": 6.140658796818252e-06,
      "loss": 0.0476,
      "step": 8375
    },
    {
      "epoch": 0.4325329202168861,
      "grad_norm": 5.823686122894287,
      "learning_rate": 6.139861027163696e-06,
      "loss": 0.0538,
      "step": 8376
    },
    {
      "epoch": 0.432584559772786,
      "grad_norm": 4.612803936004639,
      "learning_rate": 6.139063226899227e-06,
      "loss": 0.065,
      "step": 8377
    },
    {
      "epoch": 0.4326361993286858,
      "grad_norm": 3.397488832473755,
      "learning_rate": 6.138265396046267e-06,
      "loss": 0.0926,
      "step": 8378
    },
    {
      "epoch": 0.4326878388845856,
      "grad_norm": 3.57853364944458,
      "learning_rate": 6.1374675346262405e-06,
      "loss": 0.037,
      "step": 8379
    },
    {
      "epoch": 0.43273947844048544,
      "grad_norm": 2.501534938812256,
      "learning_rate": 6.136669642660576e-06,
      "loss": 0.0503,
      "step": 8380
    },
    {
      "epoch": 0.43279111799638526,
      "grad_norm": 3.9881105422973633,
      "learning_rate": 6.135871720170698e-06,
      "loss": 0.0746,
      "step": 8381
    },
    {
      "epoch": 0.4328427575522851,
      "grad_norm": 3.9940311908721924,
      "learning_rate": 6.135073767178037e-06,
      "loss": 0.0249,
      "step": 8382
    },
    {
      "epoch": 0.4328943971081849,
      "grad_norm": 4.640454292297363,
      "learning_rate": 6.134275783704017e-06,
      "loss": 0.0669,
      "step": 8383
    },
    {
      "epoch": 0.4329460366640847,
      "grad_norm": 3.061835527420044,
      "learning_rate": 6.13347776977007e-06,
      "loss": 0.0516,
      "step": 8384
    },
    {
      "epoch": 0.43299767621998453,
      "grad_norm": 2.0551884174346924,
      "learning_rate": 6.132679725397624e-06,
      "loss": 0.0282,
      "step": 8385
    },
    {
      "epoch": 0.43304931577588435,
      "grad_norm": 11.4074068069458,
      "learning_rate": 6.131881650608111e-06,
      "loss": 0.1463,
      "step": 8386
    },
    {
      "epoch": 0.43310095533178417,
      "grad_norm": 6.594322204589844,
      "learning_rate": 6.131083545422965e-06,
      "loss": 0.0662,
      "step": 8387
    },
    {
      "epoch": 0.433152594887684,
      "grad_norm": 2.509852647781372,
      "learning_rate": 6.130285409863611e-06,
      "loss": 0.0442,
      "step": 8388
    },
    {
      "epoch": 0.4332042344435838,
      "grad_norm": 6.976001262664795,
      "learning_rate": 6.129487243951491e-06,
      "loss": 0.025,
      "step": 8389
    },
    {
      "epoch": 0.4332558739994836,
      "grad_norm": 6.965019226074219,
      "learning_rate": 6.128689047708033e-06,
      "loss": 0.1383,
      "step": 8390
    },
    {
      "epoch": 0.43330751355538344,
      "grad_norm": 3.067819833755493,
      "learning_rate": 6.127890821154675e-06,
      "loss": 0.0466,
      "step": 8391
    },
    {
      "epoch": 0.43335915311128326,
      "grad_norm": 2.8604228496551514,
      "learning_rate": 6.12709256431285e-06,
      "loss": 0.0391,
      "step": 8392
    },
    {
      "epoch": 0.4334107926671831,
      "grad_norm": 4.2960968017578125,
      "learning_rate": 6.1262942772039964e-06,
      "loss": 0.0421,
      "step": 8393
    },
    {
      "epoch": 0.4334624322230829,
      "grad_norm": 3.838400363922119,
      "learning_rate": 6.1254959598495515e-06,
      "loss": 0.0289,
      "step": 8394
    },
    {
      "epoch": 0.4335140717789827,
      "grad_norm": 19.08318519592285,
      "learning_rate": 6.1246976122709525e-06,
      "loss": 0.1362,
      "step": 8395
    },
    {
      "epoch": 0.43356571133488253,
      "grad_norm": 1.9503341913223267,
      "learning_rate": 6.123899234489639e-06,
      "loss": 0.0346,
      "step": 8396
    },
    {
      "epoch": 0.43361735089078235,
      "grad_norm": 4.801864147186279,
      "learning_rate": 6.1231008265270494e-06,
      "loss": 0.0703,
      "step": 8397
    },
    {
      "epoch": 0.43366899044668217,
      "grad_norm": 3.082279920578003,
      "learning_rate": 6.122302388404626e-06,
      "loss": 0.0341,
      "step": 8398
    },
    {
      "epoch": 0.433720630002582,
      "grad_norm": 3.4709439277648926,
      "learning_rate": 6.121503920143809e-06,
      "loss": 0.0493,
      "step": 8399
    },
    {
      "epoch": 0.4337722695584818,
      "grad_norm": 9.474787712097168,
      "learning_rate": 6.12070542176604e-06,
      "loss": 0.057,
      "step": 8400
    },
    {
      "epoch": 0.4338239091143816,
      "grad_norm": 2.8189165592193604,
      "learning_rate": 6.119906893292763e-06,
      "loss": 0.0536,
      "step": 8401
    },
    {
      "epoch": 0.43387554867028144,
      "grad_norm": 5.553689956665039,
      "learning_rate": 6.1191083347454215e-06,
      "loss": 0.0844,
      "step": 8402
    },
    {
      "epoch": 0.43392718822618126,
      "grad_norm": 2.650730848312378,
      "learning_rate": 6.11830974614546e-06,
      "loss": 0.0552,
      "step": 8403
    },
    {
      "epoch": 0.4339788277820811,
      "grad_norm": 5.548437118530273,
      "learning_rate": 6.117511127514325e-06,
      "loss": 0.0669,
      "step": 8404
    },
    {
      "epoch": 0.4340304673379809,
      "grad_norm": 3.8248136043548584,
      "learning_rate": 6.116712478873459e-06,
      "loss": 0.0599,
      "step": 8405
    },
    {
      "epoch": 0.4340821068938807,
      "grad_norm": 1.8187698125839233,
      "learning_rate": 6.115913800244314e-06,
      "loss": 0.0274,
      "step": 8406
    },
    {
      "epoch": 0.43413374644978053,
      "grad_norm": 4.707709312438965,
      "learning_rate": 6.115115091648333e-06,
      "loss": 0.0523,
      "step": 8407
    },
    {
      "epoch": 0.43418538600568035,
      "grad_norm": 3.1184916496276855,
      "learning_rate": 6.11431635310697e-06,
      "loss": 0.0481,
      "step": 8408
    },
    {
      "epoch": 0.43423702556158017,
      "grad_norm": 17.479839324951172,
      "learning_rate": 6.113517584641668e-06,
      "loss": 0.1681,
      "step": 8409
    },
    {
      "epoch": 0.43428866511748,
      "grad_norm": 5.410426616668701,
      "learning_rate": 6.1127187862738814e-06,
      "loss": 0.1059,
      "step": 8410
    },
    {
      "epoch": 0.4343403046733798,
      "grad_norm": 5.226466178894043,
      "learning_rate": 6.1119199580250625e-06,
      "loss": 0.0892,
      "step": 8411
    },
    {
      "epoch": 0.4343919442292796,
      "grad_norm": 2.7325146198272705,
      "learning_rate": 6.111121099916657e-06,
      "loss": 0.0511,
      "step": 8412
    },
    {
      "epoch": 0.43444358378517944,
      "grad_norm": 2.8466951847076416,
      "learning_rate": 6.110322211970124e-06,
      "loss": 0.0267,
      "step": 8413
    },
    {
      "epoch": 0.43449522334107926,
      "grad_norm": 5.831629753112793,
      "learning_rate": 6.109523294206911e-06,
      "loss": 0.0944,
      "step": 8414
    },
    {
      "epoch": 0.4345468628969791,
      "grad_norm": 5.718925952911377,
      "learning_rate": 6.108724346648479e-06,
      "loss": 0.0522,
      "step": 8415
    },
    {
      "epoch": 0.4345985024528789,
      "grad_norm": 4.006691932678223,
      "learning_rate": 6.107925369316279e-06,
      "loss": 0.0235,
      "step": 8416
    },
    {
      "epoch": 0.4346501420087787,
      "grad_norm": 3.7568461894989014,
      "learning_rate": 6.107126362231765e-06,
      "loss": 0.0886,
      "step": 8417
    },
    {
      "epoch": 0.43470178156467854,
      "grad_norm": 2.3940677642822266,
      "learning_rate": 6.106327325416398e-06,
      "loss": 0.0142,
      "step": 8418
    },
    {
      "epoch": 0.43475342112057835,
      "grad_norm": 3.4901680946350098,
      "learning_rate": 6.105528258891631e-06,
      "loss": 0.0799,
      "step": 8419
    },
    {
      "epoch": 0.4348050606764782,
      "grad_norm": 4.119497299194336,
      "learning_rate": 6.1047291626789264e-06,
      "loss": 0.0407,
      "step": 8420
    },
    {
      "epoch": 0.434856700232378,
      "grad_norm": 3.8434553146362305,
      "learning_rate": 6.103930036799739e-06,
      "loss": 0.0386,
      "step": 8421
    },
    {
      "epoch": 0.4349083397882778,
      "grad_norm": 6.8983306884765625,
      "learning_rate": 6.103130881275533e-06,
      "loss": 0.0927,
      "step": 8422
    },
    {
      "epoch": 0.4349599793441776,
      "grad_norm": 8.58923625946045,
      "learning_rate": 6.102331696127765e-06,
      "loss": 0.1479,
      "step": 8423
    },
    {
      "epoch": 0.43501161890007745,
      "grad_norm": 3.323134183883667,
      "learning_rate": 6.101532481377899e-06,
      "loss": 0.0687,
      "step": 8424
    },
    {
      "epoch": 0.43506325845597726,
      "grad_norm": 3.209399700164795,
      "learning_rate": 6.100733237047396e-06,
      "loss": 0.0522,
      "step": 8425
    },
    {
      "epoch": 0.4351148980118771,
      "grad_norm": 7.420602321624756,
      "learning_rate": 6.0999339631577195e-06,
      "loss": 0.0405,
      "step": 8426
    },
    {
      "epoch": 0.4351665375677769,
      "grad_norm": 7.586044788360596,
      "learning_rate": 6.099134659730332e-06,
      "loss": 0.0533,
      "step": 8427
    },
    {
      "epoch": 0.4352181771236767,
      "grad_norm": 6.103119850158691,
      "learning_rate": 6.098335326786701e-06,
      "loss": 0.1299,
      "step": 8428
    },
    {
      "epoch": 0.43526981667957654,
      "grad_norm": 2.431201934814453,
      "learning_rate": 6.097535964348289e-06,
      "loss": 0.0362,
      "step": 8429
    },
    {
      "epoch": 0.43532145623547636,
      "grad_norm": 4.723580837249756,
      "learning_rate": 6.096736572436562e-06,
      "loss": 0.046,
      "step": 8430
    },
    {
      "epoch": 0.4353730957913762,
      "grad_norm": 2.489363193511963,
      "learning_rate": 6.095937151072989e-06,
      "loss": 0.0594,
      "step": 8431
    },
    {
      "epoch": 0.435424735347276,
      "grad_norm": 3.3745803833007812,
      "learning_rate": 6.095137700279037e-06,
      "loss": 0.0557,
      "step": 8432
    },
    {
      "epoch": 0.4354763749031758,
      "grad_norm": 3.6566407680511475,
      "learning_rate": 6.094338220076173e-06,
      "loss": 0.0753,
      "step": 8433
    },
    {
      "epoch": 0.43552801445907563,
      "grad_norm": 12.57571029663086,
      "learning_rate": 6.093538710485869e-06,
      "loss": 0.0621,
      "step": 8434
    },
    {
      "epoch": 0.43557965401497545,
      "grad_norm": 5.214595794677734,
      "learning_rate": 6.092739171529592e-06,
      "loss": 0.093,
      "step": 8435
    },
    {
      "epoch": 0.43563129357087527,
      "grad_norm": 9.58856201171875,
      "learning_rate": 6.091939603228816e-06,
      "loss": 0.0941,
      "step": 8436
    },
    {
      "epoch": 0.4356829331267751,
      "grad_norm": 13.584360122680664,
      "learning_rate": 6.09114000560501e-06,
      "loss": 0.1169,
      "step": 8437
    },
    {
      "epoch": 0.4357345726826749,
      "grad_norm": 2.838750123977661,
      "learning_rate": 6.090340378679648e-06,
      "loss": 0.0455,
      "step": 8438
    },
    {
      "epoch": 0.4357862122385747,
      "grad_norm": 2.8351337909698486,
      "learning_rate": 6.089540722474203e-06,
      "loss": 0.0783,
      "step": 8439
    },
    {
      "epoch": 0.43583785179447454,
      "grad_norm": 3.309387445449829,
      "learning_rate": 6.0887410370101495e-06,
      "loss": 0.0477,
      "step": 8440
    },
    {
      "epoch": 0.4358894913503744,
      "grad_norm": 12.848662376403809,
      "learning_rate": 6.0879413223089616e-06,
      "loss": 0.1174,
      "step": 8441
    },
    {
      "epoch": 0.43594113090627423,
      "grad_norm": 2.705279588699341,
      "learning_rate": 6.087141578392114e-06,
      "loss": 0.0147,
      "step": 8442
    },
    {
      "epoch": 0.43599277046217405,
      "grad_norm": 15.550169944763184,
      "learning_rate": 6.086341805281084e-06,
      "loss": 0.1723,
      "step": 8443
    },
    {
      "epoch": 0.43604441001807387,
      "grad_norm": 5.594913959503174,
      "learning_rate": 6.08554200299735e-06,
      "loss": 0.0829,
      "step": 8444
    },
    {
      "epoch": 0.4360960495739737,
      "grad_norm": 5.790196895599365,
      "learning_rate": 6.084742171562387e-06,
      "loss": 0.0659,
      "step": 8445
    },
    {
      "epoch": 0.4361476891298735,
      "grad_norm": 2.265727996826172,
      "learning_rate": 6.0839423109976784e-06,
      "loss": 0.0357,
      "step": 8446
    },
    {
      "epoch": 0.4361993286857733,
      "grad_norm": 5.979286193847656,
      "learning_rate": 6.083142421324697e-06,
      "loss": 0.0995,
      "step": 8447
    },
    {
      "epoch": 0.43625096824167314,
      "grad_norm": 8.3359375,
      "learning_rate": 6.082342502564931e-06,
      "loss": 0.0634,
      "step": 8448
    },
    {
      "epoch": 0.43630260779757296,
      "grad_norm": 1.8484644889831543,
      "learning_rate": 6.081542554739855e-06,
      "loss": 0.0368,
      "step": 8449
    },
    {
      "epoch": 0.4363542473534728,
      "grad_norm": 10.9989595413208,
      "learning_rate": 6.080742577870953e-06,
      "loss": 0.0723,
      "step": 8450
    },
    {
      "epoch": 0.4364058869093726,
      "grad_norm": 4.9601569175720215,
      "learning_rate": 6.07994257197971e-06,
      "loss": 0.0903,
      "step": 8451
    },
    {
      "epoch": 0.4364575264652724,
      "grad_norm": 2.882539987564087,
      "learning_rate": 6.079142537087605e-06,
      "loss": 0.0584,
      "step": 8452
    },
    {
      "epoch": 0.43650916602117223,
      "grad_norm": 3.4347281455993652,
      "learning_rate": 6.078342473216126e-06,
      "loss": 0.0423,
      "step": 8453
    },
    {
      "epoch": 0.43656080557707205,
      "grad_norm": 4.813358783721924,
      "learning_rate": 6.077542380386755e-06,
      "loss": 0.0723,
      "step": 8454
    },
    {
      "epoch": 0.43661244513297187,
      "grad_norm": 1.9310829639434814,
      "learning_rate": 6.076742258620981e-06,
      "loss": 0.0254,
      "step": 8455
    },
    {
      "epoch": 0.4366640846888717,
      "grad_norm": 3.950336456298828,
      "learning_rate": 6.075942107940288e-06,
      "loss": 0.0656,
      "step": 8456
    },
    {
      "epoch": 0.4367157242447715,
      "grad_norm": 3.585022211074829,
      "learning_rate": 6.075141928366164e-06,
      "loss": 0.0512,
      "step": 8457
    },
    {
      "epoch": 0.4367673638006713,
      "grad_norm": 3.8574583530426025,
      "learning_rate": 6.074341719920098e-06,
      "loss": 0.0445,
      "step": 8458
    },
    {
      "epoch": 0.43681900335657114,
      "grad_norm": 3.5882129669189453,
      "learning_rate": 6.073541482623577e-06,
      "loss": 0.0552,
      "step": 8459
    },
    {
      "epoch": 0.43687064291247096,
      "grad_norm": 36.90419006347656,
      "learning_rate": 6.0727412164980925e-06,
      "loss": 0.1777,
      "step": 8460
    },
    {
      "epoch": 0.4369222824683708,
      "grad_norm": 2.123990058898926,
      "learning_rate": 6.0719409215651335e-06,
      "loss": 0.0282,
      "step": 8461
    },
    {
      "epoch": 0.4369739220242706,
      "grad_norm": 5.18138313293457,
      "learning_rate": 6.071140597846193e-06,
      "loss": 0.0781,
      "step": 8462
    },
    {
      "epoch": 0.4370255615801704,
      "grad_norm": 4.413154125213623,
      "learning_rate": 6.07034024536276e-06,
      "loss": 0.0579,
      "step": 8463
    },
    {
      "epoch": 0.43707720113607024,
      "grad_norm": 7.788779258728027,
      "learning_rate": 6.06953986413633e-06,
      "loss": 0.1045,
      "step": 8464
    },
    {
      "epoch": 0.43712884069197006,
      "grad_norm": 3.1073782444000244,
      "learning_rate": 6.068739454188395e-06,
      "loss": 0.0418,
      "step": 8465
    },
    {
      "epoch": 0.4371804802478699,
      "grad_norm": 24.6136531829834,
      "learning_rate": 6.067939015540449e-06,
      "loss": 0.1041,
      "step": 8466
    },
    {
      "epoch": 0.4372321198037697,
      "grad_norm": 3.474628210067749,
      "learning_rate": 6.067138548213989e-06,
      "loss": 0.0431,
      "step": 8467
    },
    {
      "epoch": 0.4372837593596695,
      "grad_norm": 10.699019432067871,
      "learning_rate": 6.066338052230509e-06,
      "loss": 0.1879,
      "step": 8468
    },
    {
      "epoch": 0.43733539891556933,
      "grad_norm": 10.86746883392334,
      "learning_rate": 6.065537527611506e-06,
      "loss": 0.0871,
      "step": 8469
    },
    {
      "epoch": 0.43738703847146915,
      "grad_norm": 2.083181142807007,
      "learning_rate": 6.064736974378479e-06,
      "loss": 0.0287,
      "step": 8470
    },
    {
      "epoch": 0.43743867802736897,
      "grad_norm": 3.4574661254882812,
      "learning_rate": 6.063936392552922e-06,
      "loss": 0.0353,
      "step": 8471
    },
    {
      "epoch": 0.4374903175832688,
      "grad_norm": 14.83729076385498,
      "learning_rate": 6.063135782156339e-06,
      "loss": 0.0965,
      "step": 8472
    },
    {
      "epoch": 0.4375419571391686,
      "grad_norm": 2.6635055541992188,
      "learning_rate": 6.0623351432102264e-06,
      "loss": 0.0605,
      "step": 8473
    },
    {
      "epoch": 0.4375935966950684,
      "grad_norm": 4.116119861602783,
      "learning_rate": 6.061534475736084e-06,
      "loss": 0.0713,
      "step": 8474
    },
    {
      "epoch": 0.43764523625096824,
      "grad_norm": 6.596070289611816,
      "learning_rate": 6.060733779755415e-06,
      "loss": 0.0764,
      "step": 8475
    },
    {
      "epoch": 0.43769687580686806,
      "grad_norm": 7.186230659484863,
      "learning_rate": 6.059933055289722e-06,
      "loss": 0.0986,
      "step": 8476
    },
    {
      "epoch": 0.4377485153627679,
      "grad_norm": 19.695213317871094,
      "learning_rate": 6.059132302360506e-06,
      "loss": 0.0355,
      "step": 8477
    },
    {
      "epoch": 0.4378001549186677,
      "grad_norm": 8.031477928161621,
      "learning_rate": 6.058331520989269e-06,
      "loss": 0.112,
      "step": 8478
    },
    {
      "epoch": 0.4378517944745675,
      "grad_norm": 4.406612873077393,
      "learning_rate": 6.057530711197519e-06,
      "loss": 0.0725,
      "step": 8479
    },
    {
      "epoch": 0.43790343403046733,
      "grad_norm": 3.9144670963287354,
      "learning_rate": 6.056729873006758e-06,
      "loss": 0.078,
      "step": 8480
    },
    {
      "epoch": 0.43795507358636715,
      "grad_norm": 4.398853778839111,
      "learning_rate": 6.055929006438495e-06,
      "loss": 0.0335,
      "step": 8481
    },
    {
      "epoch": 0.43800671314226697,
      "grad_norm": 5.3488078117370605,
      "learning_rate": 6.055128111514233e-06,
      "loss": 0.0734,
      "step": 8482
    },
    {
      "epoch": 0.4380583526981668,
      "grad_norm": 4.405507564544678,
      "learning_rate": 6.05432718825548e-06,
      "loss": 0.0696,
      "step": 8483
    },
    {
      "epoch": 0.4381099922540666,
      "grad_norm": 5.472805500030518,
      "learning_rate": 6.053526236683747e-06,
      "loss": 0.0477,
      "step": 8484
    },
    {
      "epoch": 0.4381616318099664,
      "grad_norm": 7.324010372161865,
      "learning_rate": 6.0527252568205376e-06,
      "loss": 0.05,
      "step": 8485
    },
    {
      "epoch": 0.43821327136586624,
      "grad_norm": 2.7975547313690186,
      "learning_rate": 6.051924248687367e-06,
      "loss": 0.0428,
      "step": 8486
    },
    {
      "epoch": 0.43826491092176606,
      "grad_norm": 10.688128471374512,
      "learning_rate": 6.05112321230574e-06,
      "loss": 0.1417,
      "step": 8487
    },
    {
      "epoch": 0.4383165504776659,
      "grad_norm": 4.259797096252441,
      "learning_rate": 6.0503221476971745e-06,
      "loss": 0.0415,
      "step": 8488
    },
    {
      "epoch": 0.4383681900335657,
      "grad_norm": 8.72636604309082,
      "learning_rate": 6.0495210548831765e-06,
      "loss": 0.0862,
      "step": 8489
    },
    {
      "epoch": 0.4384198295894655,
      "grad_norm": 2.5453193187713623,
      "learning_rate": 6.04871993388526e-06,
      "loss": 0.0675,
      "step": 8490
    },
    {
      "epoch": 0.43847146914536533,
      "grad_norm": 3.3778462409973145,
      "learning_rate": 6.047918784724939e-06,
      "loss": 0.0323,
      "step": 8491
    },
    {
      "epoch": 0.43852310870126515,
      "grad_norm": 23.521814346313477,
      "learning_rate": 6.047117607423728e-06,
      "loss": 0.0691,
      "step": 8492
    },
    {
      "epoch": 0.43857474825716497,
      "grad_norm": 6.883060932159424,
      "learning_rate": 6.046316402003142e-06,
      "loss": 0.116,
      "step": 8493
    },
    {
      "epoch": 0.4386263878130648,
      "grad_norm": 5.39824104309082,
      "learning_rate": 6.045515168484695e-06,
      "loss": 0.1282,
      "step": 8494
    },
    {
      "epoch": 0.4386780273689646,
      "grad_norm": 2.851182222366333,
      "learning_rate": 6.044713906889905e-06,
      "loss": 0.0292,
      "step": 8495
    },
    {
      "epoch": 0.4387296669248644,
      "grad_norm": 12.54285717010498,
      "learning_rate": 6.043912617240288e-06,
      "loss": 0.1889,
      "step": 8496
    },
    {
      "epoch": 0.43878130648076424,
      "grad_norm": 1.9274318218231201,
      "learning_rate": 6.043111299557364e-06,
      "loss": 0.0433,
      "step": 8497
    },
    {
      "epoch": 0.43883294603666406,
      "grad_norm": 2.718459367752075,
      "learning_rate": 6.0423099538626486e-06,
      "loss": 0.0386,
      "step": 8498
    },
    {
      "epoch": 0.4388845855925639,
      "grad_norm": 2.7833495140075684,
      "learning_rate": 6.041508580177663e-06,
      "loss": 0.0499,
      "step": 8499
    },
    {
      "epoch": 0.4389362251484637,
      "grad_norm": 11.809680938720703,
      "learning_rate": 6.040707178523927e-06,
      "loss": 0.1069,
      "step": 8500
    },
    {
      "epoch": 0.4389878647043635,
      "grad_norm": 1.4748389720916748,
      "learning_rate": 6.0399057489229615e-06,
      "loss": 0.0291,
      "step": 8501
    },
    {
      "epoch": 0.43903950426026334,
      "grad_norm": 6.260873794555664,
      "learning_rate": 6.0391042913962885e-06,
      "loss": 0.0672,
      "step": 8502
    },
    {
      "epoch": 0.43909114381616315,
      "grad_norm": 3.16662859916687,
      "learning_rate": 6.038302805965431e-06,
      "loss": 0.0697,
      "step": 8503
    },
    {
      "epoch": 0.439142783372063,
      "grad_norm": 1.5668582916259766,
      "learning_rate": 6.037501292651911e-06,
      "loss": 0.0302,
      "step": 8504
    },
    {
      "epoch": 0.43919442292796285,
      "grad_norm": 1.80147385597229,
      "learning_rate": 6.036699751477252e-06,
      "loss": 0.0475,
      "step": 8505
    },
    {
      "epoch": 0.43924606248386266,
      "grad_norm": 8.502724647521973,
      "learning_rate": 6.035898182462979e-06,
      "loss": 0.084,
      "step": 8506
    },
    {
      "epoch": 0.4392977020397625,
      "grad_norm": 1.5189002752304077,
      "learning_rate": 6.0350965856306175e-06,
      "loss": 0.0202,
      "step": 8507
    },
    {
      "epoch": 0.4393493415956623,
      "grad_norm": 4.21268892288208,
      "learning_rate": 6.034294961001695e-06,
      "loss": 0.044,
      "step": 8508
    },
    {
      "epoch": 0.4394009811515621,
      "grad_norm": 5.797810077667236,
      "learning_rate": 6.0334933085977365e-06,
      "loss": 0.0441,
      "step": 8509
    },
    {
      "epoch": 0.43945262070746194,
      "grad_norm": 15.536725997924805,
      "learning_rate": 6.032691628440271e-06,
      "loss": 0.1543,
      "step": 8510
    },
    {
      "epoch": 0.43950426026336176,
      "grad_norm": 8.624882698059082,
      "learning_rate": 6.031889920550825e-06,
      "loss": 0.0928,
      "step": 8511
    },
    {
      "epoch": 0.4395558998192616,
      "grad_norm": 5.268145561218262,
      "learning_rate": 6.031088184950931e-06,
      "loss": 0.0649,
      "step": 8512
    },
    {
      "epoch": 0.4396075393751614,
      "grad_norm": 2.8639132976531982,
      "learning_rate": 6.030286421662114e-06,
      "loss": 0.0579,
      "step": 8513
    },
    {
      "epoch": 0.4396591789310612,
      "grad_norm": 10.36865234375,
      "learning_rate": 6.02948463070591e-06,
      "loss": 0.0923,
      "step": 8514
    },
    {
      "epoch": 0.43971081848696103,
      "grad_norm": 21.811697006225586,
      "learning_rate": 6.0286828121038465e-06,
      "loss": 0.2421,
      "step": 8515
    },
    {
      "epoch": 0.43976245804286085,
      "grad_norm": 1.8171391487121582,
      "learning_rate": 6.027880965877457e-06,
      "loss": 0.0306,
      "step": 8516
    },
    {
      "epoch": 0.43981409759876067,
      "grad_norm": 3.705148458480835,
      "learning_rate": 6.027079092048274e-06,
      "loss": 0.0469,
      "step": 8517
    },
    {
      "epoch": 0.4398657371546605,
      "grad_norm": 2.826024293899536,
      "learning_rate": 6.02627719063783e-06,
      "loss": 0.0377,
      "step": 8518
    },
    {
      "epoch": 0.4399173767105603,
      "grad_norm": 5.822463035583496,
      "learning_rate": 6.025475261667664e-06,
      "loss": 0.0569,
      "step": 8519
    },
    {
      "epoch": 0.4399690162664601,
      "grad_norm": 10.85580825805664,
      "learning_rate": 6.024673305159304e-06,
      "loss": 0.11,
      "step": 8520
    },
    {
      "epoch": 0.44002065582235994,
      "grad_norm": 2.7623159885406494,
      "learning_rate": 6.023871321134292e-06,
      "loss": 0.0262,
      "step": 8521
    },
    {
      "epoch": 0.44007229537825976,
      "grad_norm": 2.6435484886169434,
      "learning_rate": 6.02306930961416e-06,
      "loss": 0.0179,
      "step": 8522
    },
    {
      "epoch": 0.4401239349341596,
      "grad_norm": 6.174746036529541,
      "learning_rate": 6.022267270620449e-06,
      "loss": 0.0604,
      "step": 8523
    },
    {
      "epoch": 0.4401755744900594,
      "grad_norm": 1.5082367658615112,
      "learning_rate": 6.021465204174693e-06,
      "loss": 0.0146,
      "step": 8524
    },
    {
      "epoch": 0.4402272140459592,
      "grad_norm": 2.5911872386932373,
      "learning_rate": 6.020663110298435e-06,
      "loss": 0.0312,
      "step": 8525
    },
    {
      "epoch": 0.44027885360185903,
      "grad_norm": 3.6183183193206787,
      "learning_rate": 6.019860989013213e-06,
      "loss": 0.0366,
      "step": 8526
    },
    {
      "epoch": 0.44033049315775885,
      "grad_norm": 7.326127052307129,
      "learning_rate": 6.0190588403405634e-06,
      "loss": 0.0259,
      "step": 8527
    },
    {
      "epoch": 0.44038213271365867,
      "grad_norm": 3.5305545330047607,
      "learning_rate": 6.018256664302035e-06,
      "loss": 0.0482,
      "step": 8528
    },
    {
      "epoch": 0.4404337722695585,
      "grad_norm": 3.437055826187134,
      "learning_rate": 6.017454460919162e-06,
      "loss": 0.0446,
      "step": 8529
    },
    {
      "epoch": 0.4404854118254583,
      "grad_norm": 4.447775840759277,
      "learning_rate": 6.01665223021349e-06,
      "loss": 0.0602,
      "step": 8530
    },
    {
      "epoch": 0.4405370513813581,
      "grad_norm": 4.852179527282715,
      "learning_rate": 6.015849972206562e-06,
      "loss": 0.0532,
      "step": 8531
    },
    {
      "epoch": 0.44058869093725794,
      "grad_norm": 5.729658126831055,
      "learning_rate": 6.015047686919922e-06,
      "loss": 0.0815,
      "step": 8532
    },
    {
      "epoch": 0.44064033049315776,
      "grad_norm": 6.978706359863281,
      "learning_rate": 6.014245374375115e-06,
      "loss": 0.1082,
      "step": 8533
    },
    {
      "epoch": 0.4406919700490576,
      "grad_norm": 2.905447244644165,
      "learning_rate": 6.013443034593685e-06,
      "loss": 0.0606,
      "step": 8534
    },
    {
      "epoch": 0.4407436096049574,
      "grad_norm": 1.7954435348510742,
      "learning_rate": 6.01264066759718e-06,
      "loss": 0.0217,
      "step": 8535
    },
    {
      "epoch": 0.4407952491608572,
      "grad_norm": 2.640037775039673,
      "learning_rate": 6.011838273407147e-06,
      "loss": 0.0389,
      "step": 8536
    },
    {
      "epoch": 0.44084688871675703,
      "grad_norm": 20.22869873046875,
      "learning_rate": 6.0110358520451305e-06,
      "loss": 0.112,
      "step": 8537
    },
    {
      "epoch": 0.44089852827265685,
      "grad_norm": 23.77724838256836,
      "learning_rate": 6.01023340353268e-06,
      "loss": 0.0962,
      "step": 8538
    },
    {
      "epoch": 0.44095016782855667,
      "grad_norm": 5.143096923828125,
      "learning_rate": 6.009430927891347e-06,
      "loss": 0.0759,
      "step": 8539
    },
    {
      "epoch": 0.4410018073844565,
      "grad_norm": 2.5498206615448,
      "learning_rate": 6.008628425142677e-06,
      "loss": 0.029,
      "step": 8540
    },
    {
      "epoch": 0.4410534469403563,
      "grad_norm": 1.719252586364746,
      "learning_rate": 6.007825895308225e-06,
      "loss": 0.016,
      "step": 8541
    },
    {
      "epoch": 0.4411050864962561,
      "grad_norm": 3.7245047092437744,
      "learning_rate": 6.007023338409539e-06,
      "loss": 0.0531,
      "step": 8542
    },
    {
      "epoch": 0.44115672605215595,
      "grad_norm": 4.04728889465332,
      "learning_rate": 6.006220754468172e-06,
      "loss": 0.068,
      "step": 8543
    },
    {
      "epoch": 0.44120836560805576,
      "grad_norm": 6.3530049324035645,
      "learning_rate": 6.0054181435056765e-06,
      "loss": 0.0724,
      "step": 8544
    },
    {
      "epoch": 0.4412600051639556,
      "grad_norm": 2.8365371227264404,
      "learning_rate": 6.004615505543607e-06,
      "loss": 0.0543,
      "step": 8545
    },
    {
      "epoch": 0.4413116447198554,
      "grad_norm": 19.383014678955078,
      "learning_rate": 6.003812840603514e-06,
      "loss": 0.1135,
      "step": 8546
    },
    {
      "epoch": 0.4413632842757552,
      "grad_norm": 1.5834596157073975,
      "learning_rate": 6.003010148706958e-06,
      "loss": 0.0309,
      "step": 8547
    },
    {
      "epoch": 0.44141492383165504,
      "grad_norm": 13.806816101074219,
      "learning_rate": 6.00220742987549e-06,
      "loss": 0.0835,
      "step": 8548
    },
    {
      "epoch": 0.44146656338755486,
      "grad_norm": 5.095125675201416,
      "learning_rate": 6.001404684130669e-06,
      "loss": 0.0581,
      "step": 8549
    },
    {
      "epoch": 0.4415182029434547,
      "grad_norm": 4.438575744628906,
      "learning_rate": 6.000601911494049e-06,
      "loss": 0.0416,
      "step": 8550
    },
    {
      "epoch": 0.4415698424993545,
      "grad_norm": 4.156148433685303,
      "learning_rate": 5.99979911198719e-06,
      "loss": 0.045,
      "step": 8551
    },
    {
      "epoch": 0.4416214820552543,
      "grad_norm": 19.983692169189453,
      "learning_rate": 5.998996285631651e-06,
      "loss": 0.1412,
      "step": 8552
    },
    {
      "epoch": 0.44167312161115413,
      "grad_norm": 8.446587562561035,
      "learning_rate": 5.998193432448989e-06,
      "loss": 0.0624,
      "step": 8553
    },
    {
      "epoch": 0.44172476116705395,
      "grad_norm": 13.769830703735352,
      "learning_rate": 5.997390552460766e-06,
      "loss": 0.1264,
      "step": 8554
    },
    {
      "epoch": 0.44177640072295377,
      "grad_norm": 7.29681921005249,
      "learning_rate": 5.99658764568854e-06,
      "loss": 0.0797,
      "step": 8555
    },
    {
      "epoch": 0.4418280402788536,
      "grad_norm": 1.9103591442108154,
      "learning_rate": 5.9957847121538736e-06,
      "loss": 0.0302,
      "step": 8556
    },
    {
      "epoch": 0.4418796798347534,
      "grad_norm": 2.22746205329895,
      "learning_rate": 5.9949817518783306e-06,
      "loss": 0.0322,
      "step": 8557
    },
    {
      "epoch": 0.4419313193906532,
      "grad_norm": 6.111743450164795,
      "learning_rate": 5.994178764883471e-06,
      "loss": 0.0828,
      "step": 8558
    },
    {
      "epoch": 0.44198295894655304,
      "grad_norm": 2.8940227031707764,
      "learning_rate": 5.993375751190861e-06,
      "loss": 0.0759,
      "step": 8559
    },
    {
      "epoch": 0.44203459850245286,
      "grad_norm": 10.913725852966309,
      "learning_rate": 5.992572710822061e-06,
      "loss": 0.1254,
      "step": 8560
    },
    {
      "epoch": 0.4420862380583527,
      "grad_norm": 3.8969242572784424,
      "learning_rate": 5.99176964379864e-06,
      "loss": 0.0755,
      "step": 8561
    },
    {
      "epoch": 0.4421378776142525,
      "grad_norm": 6.234708786010742,
      "learning_rate": 5.9909665501421625e-06,
      "loss": 0.0761,
      "step": 8562
    },
    {
      "epoch": 0.4421895171701523,
      "grad_norm": 9.300835609436035,
      "learning_rate": 5.990163429874193e-06,
      "loss": 0.083,
      "step": 8563
    },
    {
      "epoch": 0.44224115672605213,
      "grad_norm": 3.0794949531555176,
      "learning_rate": 5.989360283016299e-06,
      "loss": 0.0326,
      "step": 8564
    },
    {
      "epoch": 0.44229279628195195,
      "grad_norm": 5.5024919509887695,
      "learning_rate": 5.98855710959005e-06,
      "loss": 0.0418,
      "step": 8565
    },
    {
      "epoch": 0.44234443583785177,
      "grad_norm": 8.458760261535645,
      "learning_rate": 5.987753909617014e-06,
      "loss": 0.0654,
      "step": 8566
    },
    {
      "epoch": 0.4423960753937516,
      "grad_norm": 5.494351387023926,
      "learning_rate": 5.986950683118758e-06,
      "loss": 0.0193,
      "step": 8567
    },
    {
      "epoch": 0.44244771494965146,
      "grad_norm": 3.1955440044403076,
      "learning_rate": 5.986147430116855e-06,
      "loss": 0.0366,
      "step": 8568
    },
    {
      "epoch": 0.4424993545055513,
      "grad_norm": 5.115545272827148,
      "learning_rate": 5.985344150632872e-06,
      "loss": 0.0614,
      "step": 8569
    },
    {
      "epoch": 0.4425509940614511,
      "grad_norm": 8.100386619567871,
      "learning_rate": 5.984540844688386e-06,
      "loss": 0.076,
      "step": 8570
    },
    {
      "epoch": 0.4426026336173509,
      "grad_norm": 2.222357749938965,
      "learning_rate": 5.983737512304964e-06,
      "loss": 0.0142,
      "step": 8571
    },
    {
      "epoch": 0.44265427317325073,
      "grad_norm": 3.5464887619018555,
      "learning_rate": 5.98293415350418e-06,
      "loss": 0.0572,
      "step": 8572
    },
    {
      "epoch": 0.44270591272915055,
      "grad_norm": 4.896358489990234,
      "learning_rate": 5.9821307683076065e-06,
      "loss": 0.092,
      "step": 8573
    },
    {
      "epoch": 0.44275755228505037,
      "grad_norm": 4.551328182220459,
      "learning_rate": 5.981327356736822e-06,
      "loss": 0.0427,
      "step": 8574
    },
    {
      "epoch": 0.4428091918409502,
      "grad_norm": 7.455621242523193,
      "learning_rate": 5.980523918813396e-06,
      "loss": 0.0356,
      "step": 8575
    },
    {
      "epoch": 0.44286083139685,
      "grad_norm": 6.7953691482543945,
      "learning_rate": 5.979720454558906e-06,
      "loss": 0.0788,
      "step": 8576
    },
    {
      "epoch": 0.4429124709527498,
      "grad_norm": 2.952148914337158,
      "learning_rate": 5.978916963994929e-06,
      "loss": 0.0693,
      "step": 8577
    },
    {
      "epoch": 0.44296411050864964,
      "grad_norm": 5.414305686950684,
      "learning_rate": 5.978113447143043e-06,
      "loss": 0.041,
      "step": 8578
    },
    {
      "epoch": 0.44301575006454946,
      "grad_norm": 10.008529663085938,
      "learning_rate": 5.977309904024823e-06,
      "loss": 0.0726,
      "step": 8579
    },
    {
      "epoch": 0.4430673896204493,
      "grad_norm": 3.25370192527771,
      "learning_rate": 5.97650633466185e-06,
      "loss": 0.0639,
      "step": 8580
    },
    {
      "epoch": 0.4431190291763491,
      "grad_norm": 4.4624104499816895,
      "learning_rate": 5.975702739075702e-06,
      "loss": 0.0802,
      "step": 8581
    },
    {
      "epoch": 0.4431706687322489,
      "grad_norm": 6.142836570739746,
      "learning_rate": 5.9748991172879576e-06,
      "loss": 0.0511,
      "step": 8582
    },
    {
      "epoch": 0.44322230828814874,
      "grad_norm": 5.708972454071045,
      "learning_rate": 5.974095469320199e-06,
      "loss": 0.0373,
      "step": 8583
    },
    {
      "epoch": 0.44327394784404855,
      "grad_norm": 4.838665962219238,
      "learning_rate": 5.973291795194006e-06,
      "loss": 0.0598,
      "step": 8584
    },
    {
      "epoch": 0.4433255873999484,
      "grad_norm": 8.498270034790039,
      "learning_rate": 5.972488094930964e-06,
      "loss": 0.0659,
      "step": 8585
    },
    {
      "epoch": 0.4433772269558482,
      "grad_norm": 2.1785664558410645,
      "learning_rate": 5.971684368552649e-06,
      "loss": 0.0415,
      "step": 8586
    },
    {
      "epoch": 0.443428866511748,
      "grad_norm": 2.469003200531006,
      "learning_rate": 5.970880616080652e-06,
      "loss": 0.0605,
      "step": 8587
    },
    {
      "epoch": 0.44348050606764783,
      "grad_norm": 16.38486099243164,
      "learning_rate": 5.97007683753655e-06,
      "loss": 0.1535,
      "step": 8588
    },
    {
      "epoch": 0.44353214562354765,
      "grad_norm": 4.255702018737793,
      "learning_rate": 5.969273032941935e-06,
      "loss": 0.1065,
      "step": 8589
    },
    {
      "epoch": 0.44358378517944747,
      "grad_norm": 3.7771716117858887,
      "learning_rate": 5.968469202318386e-06,
      "loss": 0.0481,
      "step": 8590
    },
    {
      "epoch": 0.4436354247353473,
      "grad_norm": 2.7604739665985107,
      "learning_rate": 5.967665345687492e-06,
      "loss": 0.0544,
      "step": 8591
    },
    {
      "epoch": 0.4436870642912471,
      "grad_norm": 4.860602378845215,
      "learning_rate": 5.96686146307084e-06,
      "loss": 0.0694,
      "step": 8592
    },
    {
      "epoch": 0.4437387038471469,
      "grad_norm": 10.818163871765137,
      "learning_rate": 5.966057554490016e-06,
      "loss": 0.0934,
      "step": 8593
    },
    {
      "epoch": 0.44379034340304674,
      "grad_norm": 3.2635324001312256,
      "learning_rate": 5.965253619966611e-06,
      "loss": 0.0813,
      "step": 8594
    },
    {
      "epoch": 0.44384198295894656,
      "grad_norm": 14.71575927734375,
      "learning_rate": 5.964449659522209e-06,
      "loss": 0.1065,
      "step": 8595
    },
    {
      "epoch": 0.4438936225148464,
      "grad_norm": 7.842198371887207,
      "learning_rate": 5.963645673178406e-06,
      "loss": 0.1118,
      "step": 8596
    },
    {
      "epoch": 0.4439452620707462,
      "grad_norm": 10.230777740478516,
      "learning_rate": 5.962841660956787e-06,
      "loss": 0.0642,
      "step": 8597
    },
    {
      "epoch": 0.443996901626646,
      "grad_norm": 3.679582357406616,
      "learning_rate": 5.962037622878947e-06,
      "loss": 0.0801,
      "step": 8598
    },
    {
      "epoch": 0.44404854118254583,
      "grad_norm": 22.18338394165039,
      "learning_rate": 5.961233558966474e-06,
      "loss": 0.0723,
      "step": 8599
    },
    {
      "epoch": 0.44410018073844565,
      "grad_norm": 8.479523658752441,
      "learning_rate": 5.960429469240963e-06,
      "loss": 0.061,
      "step": 8600
    },
    {
      "epoch": 0.44415182029434547,
      "grad_norm": 1.9885298013687134,
      "learning_rate": 5.959625353724007e-06,
      "loss": 0.052,
      "step": 8601
    },
    {
      "epoch": 0.4442034598502453,
      "grad_norm": 2.398676633834839,
      "learning_rate": 5.958821212437197e-06,
      "loss": 0.0319,
      "step": 8602
    },
    {
      "epoch": 0.4442550994061451,
      "grad_norm": 3.6168906688690186,
      "learning_rate": 5.958017045402132e-06,
      "loss": 0.0746,
      "step": 8603
    },
    {
      "epoch": 0.4443067389620449,
      "grad_norm": 3.44826078414917,
      "learning_rate": 5.957212852640403e-06,
      "loss": 0.0319,
      "step": 8604
    },
    {
      "epoch": 0.44435837851794474,
      "grad_norm": 4.265072822570801,
      "learning_rate": 5.956408634173607e-06,
      "loss": 0.0351,
      "step": 8605
    },
    {
      "epoch": 0.44441001807384456,
      "grad_norm": 31.084741592407227,
      "learning_rate": 5.955604390023342e-06,
      "loss": 0.1288,
      "step": 8606
    },
    {
      "epoch": 0.4444616576297444,
      "grad_norm": 2.38008451461792,
      "learning_rate": 5.954800120211204e-06,
      "loss": 0.0343,
      "step": 8607
    },
    {
      "epoch": 0.4445132971856442,
      "grad_norm": 7.930346965789795,
      "learning_rate": 5.953995824758792e-06,
      "loss": 0.1114,
      "step": 8608
    },
    {
      "epoch": 0.444564936741544,
      "grad_norm": 27.940807342529297,
      "learning_rate": 5.953191503687703e-06,
      "loss": 0.0967,
      "step": 8609
    },
    {
      "epoch": 0.44461657629744383,
      "grad_norm": 4.781363487243652,
      "learning_rate": 5.952387157019537e-06,
      "loss": 0.0564,
      "step": 8610
    },
    {
      "epoch": 0.44466821585334365,
      "grad_norm": 2.385366916656494,
      "learning_rate": 5.951582784775896e-06,
      "loss": 0.0316,
      "step": 8611
    },
    {
      "epoch": 0.44471985540924347,
      "grad_norm": 5.935311794281006,
      "learning_rate": 5.950778386978376e-06,
      "loss": 0.0579,
      "step": 8612
    },
    {
      "epoch": 0.4447714949651433,
      "grad_norm": 3.6608357429504395,
      "learning_rate": 5.949973963648583e-06,
      "loss": 0.0314,
      "step": 8613
    },
    {
      "epoch": 0.4448231345210431,
      "grad_norm": 8.100924491882324,
      "learning_rate": 5.949169514808117e-06,
      "loss": 0.0449,
      "step": 8614
    },
    {
      "epoch": 0.4448747740769429,
      "grad_norm": 4.276215553283691,
      "learning_rate": 5.9483650404785794e-06,
      "loss": 0.0342,
      "step": 8615
    },
    {
      "epoch": 0.44492641363284274,
      "grad_norm": 9.008519172668457,
      "learning_rate": 5.947560540681576e-06,
      "loss": 0.0915,
      "step": 8616
    },
    {
      "epoch": 0.44497805318874256,
      "grad_norm": 4.8331170082092285,
      "learning_rate": 5.94675601543871e-06,
      "loss": 0.0543,
      "step": 8617
    },
    {
      "epoch": 0.4450296927446424,
      "grad_norm": 3.1471915245056152,
      "learning_rate": 5.945951464771589e-06,
      "loss": 0.0329,
      "step": 8618
    },
    {
      "epoch": 0.4450813323005422,
      "grad_norm": 10.734402656555176,
      "learning_rate": 5.9451468887018116e-06,
      "loss": 0.0879,
      "step": 8619
    },
    {
      "epoch": 0.445132971856442,
      "grad_norm": 15.663076400756836,
      "learning_rate": 5.94434228725099e-06,
      "loss": 0.1367,
      "step": 8620
    },
    {
      "epoch": 0.44518461141234184,
      "grad_norm": 5.313734531402588,
      "learning_rate": 5.943537660440728e-06,
      "loss": 0.1082,
      "step": 8621
    },
    {
      "epoch": 0.44523625096824165,
      "grad_norm": 6.452117919921875,
      "learning_rate": 5.942733008292636e-06,
      "loss": 0.0755,
      "step": 8622
    },
    {
      "epoch": 0.44528789052414147,
      "grad_norm": 2.874630928039551,
      "learning_rate": 5.941928330828319e-06,
      "loss": 0.0443,
      "step": 8623
    },
    {
      "epoch": 0.4453395300800413,
      "grad_norm": 4.049482345581055,
      "learning_rate": 5.941123628069388e-06,
      "loss": 0.0706,
      "step": 8624
    },
    {
      "epoch": 0.4453911696359411,
      "grad_norm": 11.190672874450684,
      "learning_rate": 5.940318900037452e-06,
      "loss": 0.1088,
      "step": 8625
    },
    {
      "epoch": 0.4454428091918409,
      "grad_norm": 4.636573791503906,
      "learning_rate": 5.9395141467541184e-06,
      "loss": 0.0542,
      "step": 8626
    },
    {
      "epoch": 0.44549444874774075,
      "grad_norm": 5.328283309936523,
      "learning_rate": 5.938709368241003e-06,
      "loss": 0.0435,
      "step": 8627
    },
    {
      "epoch": 0.44554608830364056,
      "grad_norm": 2.6855506896972656,
      "learning_rate": 5.937904564519714e-06,
      "loss": 0.0463,
      "step": 8628
    },
    {
      "epoch": 0.4455977278595404,
      "grad_norm": 4.2283935546875,
      "learning_rate": 5.937099735611867e-06,
      "loss": 0.038,
      "step": 8629
    },
    {
      "epoch": 0.4456493674154402,
      "grad_norm": 4.069630146026611,
      "learning_rate": 5.936294881539071e-06,
      "loss": 0.0689,
      "step": 8630
    },
    {
      "epoch": 0.44570100697134,
      "grad_norm": 2.13753080368042,
      "learning_rate": 5.935490002322941e-06,
      "loss": 0.0113,
      "step": 8631
    },
    {
      "epoch": 0.4457526465272399,
      "grad_norm": 9.27065372467041,
      "learning_rate": 5.934685097985092e-06,
      "loss": 0.0866,
      "step": 8632
    },
    {
      "epoch": 0.4458042860831397,
      "grad_norm": 41.451210021972656,
      "learning_rate": 5.9338801685471385e-06,
      "loss": 0.1354,
      "step": 8633
    },
    {
      "epoch": 0.44585592563903953,
      "grad_norm": 3.7993924617767334,
      "learning_rate": 5.9330752140306966e-06,
      "loss": 0.0378,
      "step": 8634
    },
    {
      "epoch": 0.44590756519493935,
      "grad_norm": 23.530214309692383,
      "learning_rate": 5.932270234457381e-06,
      "loss": 0.1781,
      "step": 8635
    },
    {
      "epoch": 0.44595920475083917,
      "grad_norm": 13.128122329711914,
      "learning_rate": 5.9314652298488116e-06,
      "loss": 0.1377,
      "step": 8636
    },
    {
      "epoch": 0.446010844306739,
      "grad_norm": 4.919586658477783,
      "learning_rate": 5.930660200226604e-06,
      "loss": 0.0403,
      "step": 8637
    },
    {
      "epoch": 0.4460624838626388,
      "grad_norm": 3.0594606399536133,
      "learning_rate": 5.929855145612377e-06,
      "loss": 0.0524,
      "step": 8638
    },
    {
      "epoch": 0.4461141234185386,
      "grad_norm": 4.349763870239258,
      "learning_rate": 5.929050066027748e-06,
      "loss": 0.0621,
      "step": 8639
    },
    {
      "epoch": 0.44616576297443844,
      "grad_norm": 1.9476205110549927,
      "learning_rate": 5.928244961494338e-06,
      "loss": 0.0319,
      "step": 8640
    },
    {
      "epoch": 0.44621740253033826,
      "grad_norm": 3.068929433822632,
      "learning_rate": 5.9274398320337685e-06,
      "loss": 0.0276,
      "step": 8641
    },
    {
      "epoch": 0.4462690420862381,
      "grad_norm": 2.6987340450286865,
      "learning_rate": 5.92663467766766e-06,
      "loss": 0.0544,
      "step": 8642
    },
    {
      "epoch": 0.4463206816421379,
      "grad_norm": 3.3427982330322266,
      "learning_rate": 5.925829498417632e-06,
      "loss": 0.0422,
      "step": 8643
    },
    {
      "epoch": 0.4463723211980377,
      "grad_norm": 4.18455696105957,
      "learning_rate": 5.925024294305312e-06,
      "loss": 0.0842,
      "step": 8644
    },
    {
      "epoch": 0.44642396075393753,
      "grad_norm": 11.468692779541016,
      "learning_rate": 5.924219065352316e-06,
      "loss": 0.08,
      "step": 8645
    },
    {
      "epoch": 0.44647560030983735,
      "grad_norm": 13.02080249786377,
      "learning_rate": 5.923413811580272e-06,
      "loss": 0.1939,
      "step": 8646
    },
    {
      "epoch": 0.44652723986573717,
      "grad_norm": 2.304753541946411,
      "learning_rate": 5.922608533010804e-06,
      "loss": 0.0406,
      "step": 8647
    },
    {
      "epoch": 0.446578879421637,
      "grad_norm": 3.5967228412628174,
      "learning_rate": 5.921803229665536e-06,
      "loss": 0.0252,
      "step": 8648
    },
    {
      "epoch": 0.4466305189775368,
      "grad_norm": 4.530757904052734,
      "learning_rate": 5.920997901566095e-06,
      "loss": 0.0334,
      "step": 8649
    },
    {
      "epoch": 0.4466821585334366,
      "grad_norm": 3.174354314804077,
      "learning_rate": 5.920192548734105e-06,
      "loss": 0.0523,
      "step": 8650
    },
    {
      "epoch": 0.44673379808933644,
      "grad_norm": 25.429523468017578,
      "learning_rate": 5.919387171191196e-06,
      "loss": 0.064,
      "step": 8651
    },
    {
      "epoch": 0.44678543764523626,
      "grad_norm": 9.425468444824219,
      "learning_rate": 5.918581768958994e-06,
      "loss": 0.0612,
      "step": 8652
    },
    {
      "epoch": 0.4468370772011361,
      "grad_norm": 7.913191795349121,
      "learning_rate": 5.917776342059128e-06,
      "loss": 0.0811,
      "step": 8653
    },
    {
      "epoch": 0.4468887167570359,
      "grad_norm": 3.330310344696045,
      "learning_rate": 5.916970890513225e-06,
      "loss": 0.0328,
      "step": 8654
    },
    {
      "epoch": 0.4469403563129357,
      "grad_norm": 2.8467190265655518,
      "learning_rate": 5.916165414342918e-06,
      "loss": 0.0544,
      "step": 8655
    },
    {
      "epoch": 0.44699199586883553,
      "grad_norm": 7.196476459503174,
      "learning_rate": 5.9153599135698345e-06,
      "loss": 0.0635,
      "step": 8656
    },
    {
      "epoch": 0.44704363542473535,
      "grad_norm": 4.471527099609375,
      "learning_rate": 5.914554388215607e-06,
      "loss": 0.0729,
      "step": 8657
    },
    {
      "epoch": 0.44709527498063517,
      "grad_norm": 13.857057571411133,
      "learning_rate": 5.9137488383018674e-06,
      "loss": 0.0749,
      "step": 8658
    },
    {
      "epoch": 0.447146914536535,
      "grad_norm": 4.010155200958252,
      "learning_rate": 5.9129432638502446e-06,
      "loss": 0.0877,
      "step": 8659
    },
    {
      "epoch": 0.4471985540924348,
      "grad_norm": 3.5138654708862305,
      "learning_rate": 5.912137664882378e-06,
      "loss": 0.0608,
      "step": 8660
    },
    {
      "epoch": 0.4472501936483346,
      "grad_norm": 3.202913284301758,
      "learning_rate": 5.911332041419894e-06,
      "loss": 0.0842,
      "step": 8661
    },
    {
      "epoch": 0.44730183320423444,
      "grad_norm": 13.106644630432129,
      "learning_rate": 5.9105263934844325e-06,
      "loss": 0.1885,
      "step": 8662
    },
    {
      "epoch": 0.44735347276013426,
      "grad_norm": 2.8670549392700195,
      "learning_rate": 5.909720721097625e-06,
      "loss": 0.0628,
      "step": 8663
    },
    {
      "epoch": 0.4474051123160341,
      "grad_norm": 5.557792663574219,
      "learning_rate": 5.908915024281109e-06,
      "loss": 0.0426,
      "step": 8664
    },
    {
      "epoch": 0.4474567518719339,
      "grad_norm": 2.380358934402466,
      "learning_rate": 5.90810930305652e-06,
      "loss": 0.0441,
      "step": 8665
    },
    {
      "epoch": 0.4475083914278337,
      "grad_norm": 6.7757062911987305,
      "learning_rate": 5.907303557445494e-06,
      "loss": 0.0796,
      "step": 8666
    },
    {
      "epoch": 0.44756003098373354,
      "grad_norm": 1.1898688077926636,
      "learning_rate": 5.906497787469672e-06,
      "loss": 0.0163,
      "step": 8667
    },
    {
      "epoch": 0.44761167053963336,
      "grad_norm": 6.099295616149902,
      "learning_rate": 5.9056919931506864e-06,
      "loss": 0.0708,
      "step": 8668
    },
    {
      "epoch": 0.4476633100955332,
      "grad_norm": 2.8142642974853516,
      "learning_rate": 5.9048861745101815e-06,
      "loss": 0.0498,
      "step": 8669
    },
    {
      "epoch": 0.447714949651433,
      "grad_norm": 12.969675064086914,
      "learning_rate": 5.904080331569795e-06,
      "loss": 0.2191,
      "step": 8670
    },
    {
      "epoch": 0.4477665892073328,
      "grad_norm": 6.5752272605896,
      "learning_rate": 5.903274464351165e-06,
      "loss": 0.0633,
      "step": 8671
    },
    {
      "epoch": 0.44781822876323263,
      "grad_norm": 16.594684600830078,
      "learning_rate": 5.902468572875936e-06,
      "loss": 0.2179,
      "step": 8672
    },
    {
      "epoch": 0.44786986831913245,
      "grad_norm": 1.991579294204712,
      "learning_rate": 5.901662657165745e-06,
      "loss": 0.0274,
      "step": 8673
    },
    {
      "epoch": 0.44792150787503227,
      "grad_norm": 5.410712718963623,
      "learning_rate": 5.90085671724224e-06,
      "loss": 0.0538,
      "step": 8674
    },
    {
      "epoch": 0.4479731474309321,
      "grad_norm": 3.1725685596466064,
      "learning_rate": 5.9000507531270575e-06,
      "loss": 0.0862,
      "step": 8675
    },
    {
      "epoch": 0.4480247869868319,
      "grad_norm": 3.925203800201416,
      "learning_rate": 5.899244764841845e-06,
      "loss": 0.0339,
      "step": 8676
    },
    {
      "epoch": 0.4480764265427317,
      "grad_norm": 12.871936798095703,
      "learning_rate": 5.898438752408246e-06,
      "loss": 0.1369,
      "step": 8677
    },
    {
      "epoch": 0.44812806609863154,
      "grad_norm": 2.700068235397339,
      "learning_rate": 5.897632715847905e-06,
      "loss": 0.0429,
      "step": 8678
    },
    {
      "epoch": 0.44817970565453136,
      "grad_norm": 2.6036665439605713,
      "learning_rate": 5.896826655182466e-06,
      "loss": 0.0312,
      "step": 8679
    },
    {
      "epoch": 0.4482313452104312,
      "grad_norm": 3.764233112335205,
      "learning_rate": 5.896020570433576e-06,
      "loss": 0.0734,
      "step": 8680
    },
    {
      "epoch": 0.448282984766331,
      "grad_norm": 12.358491897583008,
      "learning_rate": 5.895214461622882e-06,
      "loss": 0.1032,
      "step": 8681
    },
    {
      "epoch": 0.4483346243222308,
      "grad_norm": 13.117043495178223,
      "learning_rate": 5.894408328772031e-06,
      "loss": 0.1211,
      "step": 8682
    },
    {
      "epoch": 0.44838626387813063,
      "grad_norm": 1.4124096632003784,
      "learning_rate": 5.893602171902671e-06,
      "loss": 0.0271,
      "step": 8683
    },
    {
      "epoch": 0.44843790343403045,
      "grad_norm": 5.034396648406982,
      "learning_rate": 5.892795991036452e-06,
      "loss": 0.0795,
      "step": 8684
    },
    {
      "epoch": 0.44848954298993027,
      "grad_norm": 5.299707889556885,
      "learning_rate": 5.891989786195018e-06,
      "loss": 0.0486,
      "step": 8685
    },
    {
      "epoch": 0.4485411825458301,
      "grad_norm": 4.802944183349609,
      "learning_rate": 5.8911835574000265e-06,
      "loss": 0.0393,
      "step": 8686
    },
    {
      "epoch": 0.4485928221017299,
      "grad_norm": 5.802323341369629,
      "learning_rate": 5.890377304673121e-06,
      "loss": 0.0955,
      "step": 8687
    },
    {
      "epoch": 0.4486444616576297,
      "grad_norm": 5.1606831550598145,
      "learning_rate": 5.88957102803596e-06,
      "loss": 0.0608,
      "step": 8688
    },
    {
      "epoch": 0.44869610121352954,
      "grad_norm": 1.9837409257888794,
      "learning_rate": 5.888764727510188e-06,
      "loss": 0.0315,
      "step": 8689
    },
    {
      "epoch": 0.44874774076942936,
      "grad_norm": 14.819045066833496,
      "learning_rate": 5.887958403117462e-06,
      "loss": 0.0616,
      "step": 8690
    },
    {
      "epoch": 0.4487993803253292,
      "grad_norm": 5.921828269958496,
      "learning_rate": 5.887152054879434e-06,
      "loss": 0.0725,
      "step": 8691
    },
    {
      "epoch": 0.448851019881229,
      "grad_norm": 4.770100116729736,
      "learning_rate": 5.8863456828177554e-06,
      "loss": 0.0947,
      "step": 8692
    },
    {
      "epoch": 0.4489026594371288,
      "grad_norm": 3.049551010131836,
      "learning_rate": 5.8855392869540854e-06,
      "loss": 0.0801,
      "step": 8693
    },
    {
      "epoch": 0.44895429899302863,
      "grad_norm": 5.358504295349121,
      "learning_rate": 5.884732867310074e-06,
      "loss": 0.07,
      "step": 8694
    },
    {
      "epoch": 0.44900593854892845,
      "grad_norm": 6.4302897453308105,
      "learning_rate": 5.88392642390738e-06,
      "loss": 0.072,
      "step": 8695
    },
    {
      "epoch": 0.4490575781048283,
      "grad_norm": 6.945916175842285,
      "learning_rate": 5.88311995676766e-06,
      "loss": 0.0583,
      "step": 8696
    },
    {
      "epoch": 0.44910921766072814,
      "grad_norm": 10.69011402130127,
      "learning_rate": 5.882313465912568e-06,
      "loss": 0.1434,
      "step": 8697
    },
    {
      "epoch": 0.44916085721662796,
      "grad_norm": 9.788777351379395,
      "learning_rate": 5.881506951363763e-06,
      "loss": 0.1471,
      "step": 8698
    },
    {
      "epoch": 0.4492124967725278,
      "grad_norm": 5.780505180358887,
      "learning_rate": 5.8807004131429055e-06,
      "loss": 0.0739,
      "step": 8699
    },
    {
      "epoch": 0.4492641363284276,
      "grad_norm": 2.2290730476379395,
      "learning_rate": 5.879893851271652e-06,
      "loss": 0.0152,
      "step": 8700
    },
    {
      "epoch": 0.4493157758843274,
      "grad_norm": 7.8631391525268555,
      "learning_rate": 5.8790872657716615e-06,
      "loss": 0.0473,
      "step": 8701
    },
    {
      "epoch": 0.44936741544022724,
      "grad_norm": 9.486306190490723,
      "learning_rate": 5.878280656664596e-06,
      "loss": 0.0615,
      "step": 8702
    },
    {
      "epoch": 0.44941905499612705,
      "grad_norm": 4.0891337394714355,
      "learning_rate": 5.877474023972114e-06,
      "loss": 0.0851,
      "step": 8703
    },
    {
      "epoch": 0.4494706945520269,
      "grad_norm": 3.553443431854248,
      "learning_rate": 5.876667367715878e-06,
      "loss": 0.0742,
      "step": 8704
    },
    {
      "epoch": 0.4495223341079267,
      "grad_norm": 6.5733642578125,
      "learning_rate": 5.87586068791755e-06,
      "loss": 0.0832,
      "step": 8705
    },
    {
      "epoch": 0.4495739736638265,
      "grad_norm": 5.065418243408203,
      "learning_rate": 5.8750539845987945e-06,
      "loss": 0.0447,
      "step": 8706
    },
    {
      "epoch": 0.44962561321972633,
      "grad_norm": 2.896895408630371,
      "learning_rate": 5.874247257781271e-06,
      "loss": 0.0436,
      "step": 8707
    },
    {
      "epoch": 0.44967725277562615,
      "grad_norm": 2.9298176765441895,
      "learning_rate": 5.873440507486646e-06,
      "loss": 0.0478,
      "step": 8708
    },
    {
      "epoch": 0.44972889233152596,
      "grad_norm": 3.817828416824341,
      "learning_rate": 5.872633733736584e-06,
      "loss": 0.0419,
      "step": 8709
    },
    {
      "epoch": 0.4497805318874258,
      "grad_norm": 3.2003531455993652,
      "learning_rate": 5.871826936552749e-06,
      "loss": 0.0846,
      "step": 8710
    },
    {
      "epoch": 0.4498321714433256,
      "grad_norm": 5.270711898803711,
      "learning_rate": 5.871020115956808e-06,
      "loss": 0.0517,
      "step": 8711
    },
    {
      "epoch": 0.4498838109992254,
      "grad_norm": 4.56484317779541,
      "learning_rate": 5.870213271970427e-06,
      "loss": 0.0898,
      "step": 8712
    },
    {
      "epoch": 0.44993545055512524,
      "grad_norm": 4.444637298583984,
      "learning_rate": 5.869406404615272e-06,
      "loss": 0.0254,
      "step": 8713
    },
    {
      "epoch": 0.44998709011102506,
      "grad_norm": 6.241086006164551,
      "learning_rate": 5.868599513913011e-06,
      "loss": 0.062,
      "step": 8714
    },
    {
      "epoch": 0.4500387296669249,
      "grad_norm": 5.944522380828857,
      "learning_rate": 5.867792599885314e-06,
      "loss": 0.0785,
      "step": 8715
    },
    {
      "epoch": 0.4500903692228247,
      "grad_norm": 2.941539764404297,
      "learning_rate": 5.866985662553848e-06,
      "loss": 0.0523,
      "step": 8716
    },
    {
      "epoch": 0.4501420087787245,
      "grad_norm": 5.919492244720459,
      "learning_rate": 5.866178701940285e-06,
      "loss": 0.0822,
      "step": 8717
    },
    {
      "epoch": 0.45019364833462433,
      "grad_norm": 7.082779884338379,
      "learning_rate": 5.86537171806629e-06,
      "loss": 0.0585,
      "step": 8718
    },
    {
      "epoch": 0.45024528789052415,
      "grad_norm": 3.272571563720703,
      "learning_rate": 5.864564710953541e-06,
      "loss": 0.0583,
      "step": 8719
    },
    {
      "epoch": 0.45029692744642397,
      "grad_norm": 2.568535566329956,
      "learning_rate": 5.863757680623703e-06,
      "loss": 0.0164,
      "step": 8720
    },
    {
      "epoch": 0.4503485670023238,
      "grad_norm": 4.753647804260254,
      "learning_rate": 5.862950627098453e-06,
      "loss": 0.0497,
      "step": 8721
    },
    {
      "epoch": 0.4504002065582236,
      "grad_norm": 3.5452418327331543,
      "learning_rate": 5.862143550399461e-06,
      "loss": 0.0331,
      "step": 8722
    },
    {
      "epoch": 0.4504518461141234,
      "grad_norm": 1.196209192276001,
      "learning_rate": 5.8613364505484e-06,
      "loss": 0.0126,
      "step": 8723
    },
    {
      "epoch": 0.45050348567002324,
      "grad_norm": 5.439576148986816,
      "learning_rate": 5.8605293275669465e-06,
      "loss": 0.0749,
      "step": 8724
    },
    {
      "epoch": 0.45055512522592306,
      "grad_norm": 13.584896087646484,
      "learning_rate": 5.85972218147677e-06,
      "loss": 0.0682,
      "step": 8725
    },
    {
      "epoch": 0.4506067647818229,
      "grad_norm": 26.927806854248047,
      "learning_rate": 5.858915012299551e-06,
      "loss": 0.0702,
      "step": 8726
    },
    {
      "epoch": 0.4506584043377227,
      "grad_norm": 4.2512640953063965,
      "learning_rate": 5.858107820056962e-06,
      "loss": 0.0316,
      "step": 8727
    },
    {
      "epoch": 0.4507100438936225,
      "grad_norm": 8.177205085754395,
      "learning_rate": 5.857300604770681e-06,
      "loss": 0.0691,
      "step": 8728
    },
    {
      "epoch": 0.45076168344952233,
      "grad_norm": 7.786367893218994,
      "learning_rate": 5.856493366462386e-06,
      "loss": 0.0386,
      "step": 8729
    },
    {
      "epoch": 0.45081332300542215,
      "grad_norm": 4.513723373413086,
      "learning_rate": 5.855686105153751e-06,
      "loss": 0.052,
      "step": 8730
    },
    {
      "epoch": 0.45086496256132197,
      "grad_norm": 5.058846473693848,
      "learning_rate": 5.854878820866458e-06,
      "loss": 0.0367,
      "step": 8731
    },
    {
      "epoch": 0.4509166021172218,
      "grad_norm": 4.220285892486572,
      "learning_rate": 5.8540715136221845e-06,
      "loss": 0.0497,
      "step": 8732
    },
    {
      "epoch": 0.4509682416731216,
      "grad_norm": 12.456282615661621,
      "learning_rate": 5.853264183442611e-06,
      "loss": 0.0571,
      "step": 8733
    },
    {
      "epoch": 0.4510198812290214,
      "grad_norm": 8.420051574707031,
      "learning_rate": 5.8524568303494136e-06,
      "loss": 0.0463,
      "step": 8734
    },
    {
      "epoch": 0.45107152078492124,
      "grad_norm": 3.6984169483184814,
      "learning_rate": 5.851649454364278e-06,
      "loss": 0.0376,
      "step": 8735
    },
    {
      "epoch": 0.45112316034082106,
      "grad_norm": 2.6928772926330566,
      "learning_rate": 5.850842055508882e-06,
      "loss": 0.036,
      "step": 8736
    },
    {
      "epoch": 0.4511747998967209,
      "grad_norm": 3.3368189334869385,
      "learning_rate": 5.85003463380491e-06,
      "loss": 0.0321,
      "step": 8737
    },
    {
      "epoch": 0.4512264394526207,
      "grad_norm": 11.938965797424316,
      "learning_rate": 5.849227189274042e-06,
      "loss": 0.1474,
      "step": 8738
    },
    {
      "epoch": 0.4512780790085205,
      "grad_norm": 14.081238746643066,
      "learning_rate": 5.848419721937963e-06,
      "loss": 0.0718,
      "step": 8739
    },
    {
      "epoch": 0.45132971856442033,
      "grad_norm": 13.846327781677246,
      "learning_rate": 5.847612231818358e-06,
      "loss": 0.1167,
      "step": 8740
    },
    {
      "epoch": 0.45138135812032015,
      "grad_norm": 3.752999782562256,
      "learning_rate": 5.846804718936909e-06,
      "loss": 0.0554,
      "step": 8741
    },
    {
      "epoch": 0.45143299767621997,
      "grad_norm": 4.875956058502197,
      "learning_rate": 5.845997183315302e-06,
      "loss": 0.058,
      "step": 8742
    },
    {
      "epoch": 0.4514846372321198,
      "grad_norm": 5.769262313842773,
      "learning_rate": 5.845189624975224e-06,
      "loss": 0.0674,
      "step": 8743
    },
    {
      "epoch": 0.4515362767880196,
      "grad_norm": 3.081350803375244,
      "learning_rate": 5.844382043938358e-06,
      "loss": 0.0372,
      "step": 8744
    },
    {
      "epoch": 0.4515879163439194,
      "grad_norm": 5.533491611480713,
      "learning_rate": 5.843574440226393e-06,
      "loss": 0.0458,
      "step": 8745
    },
    {
      "epoch": 0.45163955589981924,
      "grad_norm": 5.16412878036499,
      "learning_rate": 5.842766813861016e-06,
      "loss": 0.0864,
      "step": 8746
    },
    {
      "epoch": 0.45169119545571906,
      "grad_norm": 6.454485893249512,
      "learning_rate": 5.841959164863915e-06,
      "loss": 0.0787,
      "step": 8747
    },
    {
      "epoch": 0.4517428350116189,
      "grad_norm": 2.2551989555358887,
      "learning_rate": 5.841151493256779e-06,
      "loss": 0.0206,
      "step": 8748
    },
    {
      "epoch": 0.4517944745675187,
      "grad_norm": 9.978150367736816,
      "learning_rate": 5.840343799061296e-06,
      "loss": 0.1643,
      "step": 8749
    },
    {
      "epoch": 0.4518461141234185,
      "grad_norm": 15.918255805969238,
      "learning_rate": 5.83953608229916e-06,
      "loss": 0.1832,
      "step": 8750
    },
    {
      "epoch": 0.45189775367931834,
      "grad_norm": 2.462430953979492,
      "learning_rate": 5.8387283429920545e-06,
      "loss": 0.0474,
      "step": 8751
    },
    {
      "epoch": 0.45194939323521816,
      "grad_norm": 7.872053146362305,
      "learning_rate": 5.837920581161678e-06,
      "loss": 0.0299,
      "step": 8752
    },
    {
      "epoch": 0.452001032791118,
      "grad_norm": 5.344137668609619,
      "learning_rate": 5.837112796829716e-06,
      "loss": 0.0541,
      "step": 8753
    },
    {
      "epoch": 0.4520526723470178,
      "grad_norm": 4.770347595214844,
      "learning_rate": 5.836304990017865e-06,
      "loss": 0.0812,
      "step": 8754
    },
    {
      "epoch": 0.4521043119029176,
      "grad_norm": 2.0987541675567627,
      "learning_rate": 5.835497160747816e-06,
      "loss": 0.0354,
      "step": 8755
    },
    {
      "epoch": 0.45215595145881743,
      "grad_norm": 7.922534942626953,
      "learning_rate": 5.8346893090412635e-06,
      "loss": 0.0907,
      "step": 8756
    },
    {
      "epoch": 0.45220759101471725,
      "grad_norm": 1.9618879556655884,
      "learning_rate": 5.833881434919901e-06,
      "loss": 0.0444,
      "step": 8757
    },
    {
      "epoch": 0.45225923057061707,
      "grad_norm": 6.157760143280029,
      "learning_rate": 5.833073538405422e-06,
      "loss": 0.0573,
      "step": 8758
    },
    {
      "epoch": 0.45231087012651694,
      "grad_norm": 2.702219247817993,
      "learning_rate": 5.832265619519525e-06,
      "loss": 0.0346,
      "step": 8759
    },
    {
      "epoch": 0.45236250968241676,
      "grad_norm": 13.591460227966309,
      "learning_rate": 5.831457678283902e-06,
      "loss": 0.0768,
      "step": 8760
    },
    {
      "epoch": 0.4524141492383166,
      "grad_norm": 13.133918762207031,
      "learning_rate": 5.830649714720253e-06,
      "loss": 0.1313,
      "step": 8761
    },
    {
      "epoch": 0.4524657887942164,
      "grad_norm": 2.856807231903076,
      "learning_rate": 5.829841728850273e-06,
      "loss": 0.0556,
      "step": 8762
    },
    {
      "epoch": 0.4525174283501162,
      "grad_norm": 9.397391319274902,
      "learning_rate": 5.8290337206956605e-06,
      "loss": 0.1473,
      "step": 8763
    },
    {
      "epoch": 0.45256906790601603,
      "grad_norm": 5.3887553215026855,
      "learning_rate": 5.828225690278113e-06,
      "loss": 0.0637,
      "step": 8764
    },
    {
      "epoch": 0.45262070746191585,
      "grad_norm": 2.415964365005493,
      "learning_rate": 5.82741763761933e-06,
      "loss": 0.0225,
      "step": 8765
    },
    {
      "epoch": 0.45267234701781567,
      "grad_norm": 1.9863998889923096,
      "learning_rate": 5.826609562741012e-06,
      "loss": 0.0326,
      "step": 8766
    },
    {
      "epoch": 0.4527239865737155,
      "grad_norm": 5.60715913772583,
      "learning_rate": 5.825801465664857e-06,
      "loss": 0.1018,
      "step": 8767
    },
    {
      "epoch": 0.4527756261296153,
      "grad_norm": 1.819960594177246,
      "learning_rate": 5.824993346412569e-06,
      "loss": 0.0241,
      "step": 8768
    },
    {
      "epoch": 0.4528272656855151,
      "grad_norm": 7.661189556121826,
      "learning_rate": 5.824185205005846e-06,
      "loss": 0.07,
      "step": 8769
    },
    {
      "epoch": 0.45287890524141494,
      "grad_norm": 6.9946770668029785,
      "learning_rate": 5.823377041466391e-06,
      "loss": 0.0884,
      "step": 8770
    },
    {
      "epoch": 0.45293054479731476,
      "grad_norm": 4.2497382164001465,
      "learning_rate": 5.8225688558159055e-06,
      "loss": 0.1,
      "step": 8771
    },
    {
      "epoch": 0.4529821843532146,
      "grad_norm": 6.106378078460693,
      "learning_rate": 5.821760648076094e-06,
      "loss": 0.1411,
      "step": 8772
    },
    {
      "epoch": 0.4530338239091144,
      "grad_norm": 6.882951736450195,
      "learning_rate": 5.820952418268661e-06,
      "loss": 0.0506,
      "step": 8773
    },
    {
      "epoch": 0.4530854634650142,
      "grad_norm": 7.713958263397217,
      "learning_rate": 5.8201441664153095e-06,
      "loss": 0.0681,
      "step": 8774
    },
    {
      "epoch": 0.45313710302091403,
      "grad_norm": 3.6531822681427,
      "learning_rate": 5.8193358925377445e-06,
      "loss": 0.0678,
      "step": 8775
    },
    {
      "epoch": 0.45318874257681385,
      "grad_norm": 3.707155704498291,
      "learning_rate": 5.8185275966576694e-06,
      "loss": 0.0316,
      "step": 8776
    },
    {
      "epoch": 0.45324038213271367,
      "grad_norm": 4.66381311416626,
      "learning_rate": 5.817719278796794e-06,
      "loss": 0.0638,
      "step": 8777
    },
    {
      "epoch": 0.4532920216886135,
      "grad_norm": 2.585273027420044,
      "learning_rate": 5.816910938976822e-06,
      "loss": 0.0364,
      "step": 8778
    },
    {
      "epoch": 0.4533436612445133,
      "grad_norm": 55.71826171875,
      "learning_rate": 5.8161025772194624e-06,
      "loss": 0.1044,
      "step": 8779
    },
    {
      "epoch": 0.4533953008004131,
      "grad_norm": 4.863447666168213,
      "learning_rate": 5.815294193546422e-06,
      "loss": 0.0953,
      "step": 8780
    },
    {
      "epoch": 0.45344694035631294,
      "grad_norm": 6.653972148895264,
      "learning_rate": 5.814485787979409e-06,
      "loss": 0.0575,
      "step": 8781
    },
    {
      "epoch": 0.45349857991221276,
      "grad_norm": 3.7088871002197266,
      "learning_rate": 5.813677360540134e-06,
      "loss": 0.0516,
      "step": 8782
    },
    {
      "epoch": 0.4535502194681126,
      "grad_norm": 22.089054107666016,
      "learning_rate": 5.812868911250307e-06,
      "loss": 0.1287,
      "step": 8783
    },
    {
      "epoch": 0.4536018590240124,
      "grad_norm": 3.8002946376800537,
      "learning_rate": 5.812060440131633e-06,
      "loss": 0.0966,
      "step": 8784
    },
    {
      "epoch": 0.4536534985799122,
      "grad_norm": 2.438011884689331,
      "learning_rate": 5.811251947205829e-06,
      "loss": 0.0679,
      "step": 8785
    },
    {
      "epoch": 0.45370513813581204,
      "grad_norm": 3.3592371940612793,
      "learning_rate": 5.810443432494603e-06,
      "loss": 0.0636,
      "step": 8786
    },
    {
      "epoch": 0.45375677769171185,
      "grad_norm": 3.0747909545898438,
      "learning_rate": 5.809634896019668e-06,
      "loss": 0.0595,
      "step": 8787
    },
    {
      "epoch": 0.4538084172476117,
      "grad_norm": 2.5666050910949707,
      "learning_rate": 5.808826337802735e-06,
      "loss": 0.0345,
      "step": 8788
    },
    {
      "epoch": 0.4538600568035115,
      "grad_norm": 3.9302139282226562,
      "learning_rate": 5.8080177578655195e-06,
      "loss": 0.0338,
      "step": 8789
    },
    {
      "epoch": 0.4539116963594113,
      "grad_norm": 11.74552059173584,
      "learning_rate": 5.807209156229734e-06,
      "loss": 0.1864,
      "step": 8790
    },
    {
      "epoch": 0.45396333591531113,
      "grad_norm": 8.349235534667969,
      "learning_rate": 5.806400532917092e-06,
      "loss": 0.0733,
      "step": 8791
    },
    {
      "epoch": 0.45401497547121095,
      "grad_norm": 9.207596778869629,
      "learning_rate": 5.805591887949309e-06,
      "loss": 0.0989,
      "step": 8792
    },
    {
      "epoch": 0.45406661502711076,
      "grad_norm": 3.001682996749878,
      "learning_rate": 5.8047832213481e-06,
      "loss": 0.0467,
      "step": 8793
    },
    {
      "epoch": 0.4541182545830106,
      "grad_norm": 2.491731882095337,
      "learning_rate": 5.803974533135183e-06,
      "loss": 0.0555,
      "step": 8794
    },
    {
      "epoch": 0.4541698941389104,
      "grad_norm": 9.066567420959473,
      "learning_rate": 5.8031658233322715e-06,
      "loss": 0.082,
      "step": 8795
    },
    {
      "epoch": 0.4542215336948102,
      "grad_norm": 3.144120931625366,
      "learning_rate": 5.802357091961084e-06,
      "loss": 0.0632,
      "step": 8796
    },
    {
      "epoch": 0.45427317325071004,
      "grad_norm": 5.6394195556640625,
      "learning_rate": 5.801548339043339e-06,
      "loss": 0.0528,
      "step": 8797
    },
    {
      "epoch": 0.45432481280660986,
      "grad_norm": 3.6026809215545654,
      "learning_rate": 5.800739564600753e-06,
      "loss": 0.035,
      "step": 8798
    },
    {
      "epoch": 0.4543764523625097,
      "grad_norm": 4.65293550491333,
      "learning_rate": 5.799930768655049e-06,
      "loss": 0.0487,
      "step": 8799
    },
    {
      "epoch": 0.4544280919184095,
      "grad_norm": 2.3843934535980225,
      "learning_rate": 5.799121951227941e-06,
      "loss": 0.0652,
      "step": 8800
    },
    {
      "epoch": 0.4544797314743093,
      "grad_norm": 1.8209894895553589,
      "learning_rate": 5.798313112341153e-06,
      "loss": 0.0329,
      "step": 8801
    },
    {
      "epoch": 0.45453137103020913,
      "grad_norm": 6.447046279907227,
      "learning_rate": 5.7975042520164025e-06,
      "loss": 0.0812,
      "step": 8802
    },
    {
      "epoch": 0.45458301058610895,
      "grad_norm": 3.3561415672302246,
      "learning_rate": 5.7966953702754135e-06,
      "loss": 0.0747,
      "step": 8803
    },
    {
      "epoch": 0.45463465014200877,
      "grad_norm": 5.216127395629883,
      "learning_rate": 5.795886467139906e-06,
      "loss": 0.0969,
      "step": 8804
    },
    {
      "epoch": 0.4546862896979086,
      "grad_norm": 17.02448081970215,
      "learning_rate": 5.795077542631603e-06,
      "loss": 0.0889,
      "step": 8805
    },
    {
      "epoch": 0.4547379292538084,
      "grad_norm": 1.536291241645813,
      "learning_rate": 5.794268596772228e-06,
      "loss": 0.0354,
      "step": 8806
    },
    {
      "epoch": 0.4547895688097082,
      "grad_norm": 7.656494140625,
      "learning_rate": 5.793459629583503e-06,
      "loss": 0.0854,
      "step": 8807
    },
    {
      "epoch": 0.45484120836560804,
      "grad_norm": 2.8923110961914062,
      "learning_rate": 5.792650641087155e-06,
      "loss": 0.0569,
      "step": 8808
    },
    {
      "epoch": 0.45489284792150786,
      "grad_norm": 2.8298192024230957,
      "learning_rate": 5.791841631304904e-06,
      "loss": 0.0423,
      "step": 8809
    },
    {
      "epoch": 0.4549444874774077,
      "grad_norm": 2.707249164581299,
      "learning_rate": 5.79103260025848e-06,
      "loss": 0.0303,
      "step": 8810
    },
    {
      "epoch": 0.4549961270333075,
      "grad_norm": 1.7410287857055664,
      "learning_rate": 5.790223547969607e-06,
      "loss": 0.0334,
      "step": 8811
    },
    {
      "epoch": 0.4550477665892073,
      "grad_norm": 3.030691385269165,
      "learning_rate": 5.789414474460009e-06,
      "loss": 0.0711,
      "step": 8812
    },
    {
      "epoch": 0.45509940614510713,
      "grad_norm": 3.75598406791687,
      "learning_rate": 5.788605379751415e-06,
      "loss": 0.0538,
      "step": 8813
    },
    {
      "epoch": 0.45515104570100695,
      "grad_norm": 3.9380061626434326,
      "learning_rate": 5.787796263865554e-06,
      "loss": 0.0437,
      "step": 8814
    },
    {
      "epoch": 0.45520268525690677,
      "grad_norm": 7.155029773712158,
      "learning_rate": 5.78698712682415e-06,
      "loss": 0.1228,
      "step": 8815
    },
    {
      "epoch": 0.4552543248128066,
      "grad_norm": 2.6400721073150635,
      "learning_rate": 5.786177968648936e-06,
      "loss": 0.056,
      "step": 8816
    },
    {
      "epoch": 0.4553059643687064,
      "grad_norm": 8.821884155273438,
      "learning_rate": 5.785368789361638e-06,
      "loss": 0.0664,
      "step": 8817
    },
    {
      "epoch": 0.4553576039246062,
      "grad_norm": 4.678064823150635,
      "learning_rate": 5.784559588983989e-06,
      "loss": 0.0816,
      "step": 8818
    },
    {
      "epoch": 0.45540924348050604,
      "grad_norm": 4.396125793457031,
      "learning_rate": 5.783750367537717e-06,
      "loss": 0.0546,
      "step": 8819
    },
    {
      "epoch": 0.45546088303640586,
      "grad_norm": 4.988765239715576,
      "learning_rate": 5.782941125044552e-06,
      "loss": 0.0652,
      "step": 8820
    },
    {
      "epoch": 0.4555125225923057,
      "grad_norm": 9.461504936218262,
      "learning_rate": 5.782131861526226e-06,
      "loss": 0.06,
      "step": 8821
    },
    {
      "epoch": 0.4555641621482055,
      "grad_norm": 3.2986977100372314,
      "learning_rate": 5.781322577004473e-06,
      "loss": 0.0372,
      "step": 8822
    },
    {
      "epoch": 0.45561580170410537,
      "grad_norm": 7.293397903442383,
      "learning_rate": 5.780513271501025e-06,
      "loss": 0.0666,
      "step": 8823
    },
    {
      "epoch": 0.4556674412600052,
      "grad_norm": 2.6255319118499756,
      "learning_rate": 5.779703945037613e-06,
      "loss": 0.0477,
      "step": 8824
    },
    {
      "epoch": 0.455719080815905,
      "grad_norm": 8.26521110534668,
      "learning_rate": 5.778894597635973e-06,
      "loss": 0.0583,
      "step": 8825
    },
    {
      "epoch": 0.4557707203718048,
      "grad_norm": 11.142658233642578,
      "learning_rate": 5.778085229317838e-06,
      "loss": 0.0625,
      "step": 8826
    },
    {
      "epoch": 0.45582235992770465,
      "grad_norm": 10.340659141540527,
      "learning_rate": 5.777275840104944e-06,
      "loss": 0.0736,
      "step": 8827
    },
    {
      "epoch": 0.45587399948360446,
      "grad_norm": 6.55143404006958,
      "learning_rate": 5.776466430019025e-06,
      "loss": 0.0779,
      "step": 8828
    },
    {
      "epoch": 0.4559256390395043,
      "grad_norm": 3.471808910369873,
      "learning_rate": 5.7756569990818175e-06,
      "loss": 0.0505,
      "step": 8829
    },
    {
      "epoch": 0.4559772785954041,
      "grad_norm": 4.47812032699585,
      "learning_rate": 5.774847547315059e-06,
      "loss": 0.1023,
      "step": 8830
    },
    {
      "epoch": 0.4560289181513039,
      "grad_norm": 4.2005839347839355,
      "learning_rate": 5.774038074740485e-06,
      "loss": 0.0717,
      "step": 8831
    },
    {
      "epoch": 0.45608055770720374,
      "grad_norm": 2.4027175903320312,
      "learning_rate": 5.773228581379835e-06,
      "loss": 0.029,
      "step": 8832
    },
    {
      "epoch": 0.45613219726310356,
      "grad_norm": 10.05518913269043,
      "learning_rate": 5.772419067254844e-06,
      "loss": 0.1014,
      "step": 8833
    },
    {
      "epoch": 0.4561838368190034,
      "grad_norm": 1.722062587738037,
      "learning_rate": 5.771609532387256e-06,
      "loss": 0.0317,
      "step": 8834
    },
    {
      "epoch": 0.4562354763749032,
      "grad_norm": 4.916800022125244,
      "learning_rate": 5.7707999767988035e-06,
      "loss": 0.0951,
      "step": 8835
    },
    {
      "epoch": 0.456287115930803,
      "grad_norm": 5.791032314300537,
      "learning_rate": 5.769990400511233e-06,
      "loss": 0.0639,
      "step": 8836
    },
    {
      "epoch": 0.45633875548670283,
      "grad_norm": 4.3041205406188965,
      "learning_rate": 5.76918080354628e-06,
      "loss": 0.0317,
      "step": 8837
    },
    {
      "epoch": 0.45639039504260265,
      "grad_norm": 9.894245147705078,
      "learning_rate": 5.768371185925687e-06,
      "loss": 0.1179,
      "step": 8838
    },
    {
      "epoch": 0.45644203459850247,
      "grad_norm": 6.34933614730835,
      "learning_rate": 5.767561547671197e-06,
      "loss": 0.0429,
      "step": 8839
    },
    {
      "epoch": 0.4564936741544023,
      "grad_norm": 3.2877676486968994,
      "learning_rate": 5.766751888804551e-06,
      "loss": 0.0453,
      "step": 8840
    },
    {
      "epoch": 0.4565453137103021,
      "grad_norm": 8.33021354675293,
      "learning_rate": 5.765942209347494e-06,
      "loss": 0.0702,
      "step": 8841
    },
    {
      "epoch": 0.4565969532662019,
      "grad_norm": 3.213456630706787,
      "learning_rate": 5.7651325093217626e-06,
      "loss": 0.085,
      "step": 8842
    },
    {
      "epoch": 0.45664859282210174,
      "grad_norm": 10.040897369384766,
      "learning_rate": 5.764322788749108e-06,
      "loss": 0.0974,
      "step": 8843
    },
    {
      "epoch": 0.45670023237800156,
      "grad_norm": 14.19941520690918,
      "learning_rate": 5.7635130476512705e-06,
      "loss": 0.0781,
      "step": 8844
    },
    {
      "epoch": 0.4567518719339014,
      "grad_norm": 9.143765449523926,
      "learning_rate": 5.7627032860499954e-06,
      "loss": 0.1214,
      "step": 8845
    },
    {
      "epoch": 0.4568035114898012,
      "grad_norm": 3.9479126930236816,
      "learning_rate": 5.761893503967028e-06,
      "loss": 0.0402,
      "step": 8846
    },
    {
      "epoch": 0.456855151045701,
      "grad_norm": 5.636626243591309,
      "learning_rate": 5.761083701424115e-06,
      "loss": 0.0868,
      "step": 8847
    },
    {
      "epoch": 0.45690679060160083,
      "grad_norm": 8.813619613647461,
      "learning_rate": 5.760273878443003e-06,
      "loss": 0.0526,
      "step": 8848
    },
    {
      "epoch": 0.45695843015750065,
      "grad_norm": 2.187086582183838,
      "learning_rate": 5.759464035045439e-06,
      "loss": 0.0534,
      "step": 8849
    },
    {
      "epoch": 0.45701006971340047,
      "grad_norm": 2.573946475982666,
      "learning_rate": 5.758654171253168e-06,
      "loss": 0.0536,
      "step": 8850
    },
    {
      "epoch": 0.4570617092693003,
      "grad_norm": 8.450675964355469,
      "learning_rate": 5.757844287087944e-06,
      "loss": 0.12,
      "step": 8851
    },
    {
      "epoch": 0.4571133488252001,
      "grad_norm": 2.65299391746521,
      "learning_rate": 5.757034382571511e-06,
      "loss": 0.0713,
      "step": 8852
    },
    {
      "epoch": 0.4571649883810999,
      "grad_norm": 5.18906307220459,
      "learning_rate": 5.756224457725618e-06,
      "loss": 0.067,
      "step": 8853
    },
    {
      "epoch": 0.45721662793699974,
      "grad_norm": 6.769356727600098,
      "learning_rate": 5.755414512572016e-06,
      "loss": 0.0888,
      "step": 8854
    },
    {
      "epoch": 0.45726826749289956,
      "grad_norm": 9.521285057067871,
      "learning_rate": 5.754604547132456e-06,
      "loss": 0.2491,
      "step": 8855
    },
    {
      "epoch": 0.4573199070487994,
      "grad_norm": 4.790517330169678,
      "learning_rate": 5.753794561428688e-06,
      "loss": 0.0949,
      "step": 8856
    },
    {
      "epoch": 0.4573715466046992,
      "grad_norm": 3.5649924278259277,
      "learning_rate": 5.752984555482465e-06,
      "loss": 0.0743,
      "step": 8857
    },
    {
      "epoch": 0.457423186160599,
      "grad_norm": 1.2120543718338013,
      "learning_rate": 5.7521745293155376e-06,
      "loss": 0.0333,
      "step": 8858
    },
    {
      "epoch": 0.45747482571649883,
      "grad_norm": 8.30726432800293,
      "learning_rate": 5.751364482949656e-06,
      "loss": 0.0707,
      "step": 8859
    },
    {
      "epoch": 0.45752646527239865,
      "grad_norm": 1.878822684288025,
      "learning_rate": 5.750554416406579e-06,
      "loss": 0.0488,
      "step": 8860
    },
    {
      "epoch": 0.45757810482829847,
      "grad_norm": 4.942776679992676,
      "learning_rate": 5.749744329708054e-06,
      "loss": 0.0532,
      "step": 8861
    },
    {
      "epoch": 0.4576297443841983,
      "grad_norm": 3.3606607913970947,
      "learning_rate": 5.74893422287584e-06,
      "loss": 0.0461,
      "step": 8862
    },
    {
      "epoch": 0.4576813839400981,
      "grad_norm": 5.89153528213501,
      "learning_rate": 5.748124095931688e-06,
      "loss": 0.0683,
      "step": 8863
    },
    {
      "epoch": 0.4577330234959979,
      "grad_norm": 4.456990718841553,
      "learning_rate": 5.747313948897355e-06,
      "loss": 0.0465,
      "step": 8864
    },
    {
      "epoch": 0.45778466305189774,
      "grad_norm": 3.091308355331421,
      "learning_rate": 5.746503781794598e-06,
      "loss": 0.0678,
      "step": 8865
    },
    {
      "epoch": 0.45783630260779756,
      "grad_norm": 2.975998640060425,
      "learning_rate": 5.74569359464517e-06,
      "loss": 0.0568,
      "step": 8866
    },
    {
      "epoch": 0.4578879421636974,
      "grad_norm": 1.905035376548767,
      "learning_rate": 5.744883387470832e-06,
      "loss": 0.0431,
      "step": 8867
    },
    {
      "epoch": 0.4579395817195972,
      "grad_norm": 4.608496189117432,
      "learning_rate": 5.744073160293337e-06,
      "loss": 0.0613,
      "step": 8868
    },
    {
      "epoch": 0.457991221275497,
      "grad_norm": 6.443349361419678,
      "learning_rate": 5.743262913134446e-06,
      "loss": 0.0653,
      "step": 8869
    },
    {
      "epoch": 0.45804286083139684,
      "grad_norm": 10.409623146057129,
      "learning_rate": 5.742452646015916e-06,
      "loss": 0.0467,
      "step": 8870
    },
    {
      "epoch": 0.45809450038729665,
      "grad_norm": 7.220547199249268,
      "learning_rate": 5.741642358959506e-06,
      "loss": 0.0788,
      "step": 8871
    },
    {
      "epoch": 0.4581461399431965,
      "grad_norm": 5.097150802612305,
      "learning_rate": 5.740832051986977e-06,
      "loss": 0.1014,
      "step": 8872
    },
    {
      "epoch": 0.4581977794990963,
      "grad_norm": 1.6013474464416504,
      "learning_rate": 5.740021725120087e-06,
      "loss": 0.0177,
      "step": 8873
    },
    {
      "epoch": 0.4582494190549961,
      "grad_norm": 2.1007509231567383,
      "learning_rate": 5.7392113783805995e-06,
      "loss": 0.0499,
      "step": 8874
    },
    {
      "epoch": 0.45830105861089593,
      "grad_norm": 4.36593770980835,
      "learning_rate": 5.73840101179027e-06,
      "loss": 0.0642,
      "step": 8875
    },
    {
      "epoch": 0.45835269816679575,
      "grad_norm": 6.5990681648254395,
      "learning_rate": 5.737590625370867e-06,
      "loss": 0.0664,
      "step": 8876
    },
    {
      "epoch": 0.45840433772269557,
      "grad_norm": 2.7650299072265625,
      "learning_rate": 5.736780219144148e-06,
      "loss": 0.0478,
      "step": 8877
    },
    {
      "epoch": 0.4584559772785954,
      "grad_norm": 4.523682117462158,
      "learning_rate": 5.735969793131877e-06,
      "loss": 0.049,
      "step": 8878
    },
    {
      "epoch": 0.4585076168344952,
      "grad_norm": 3.000079870223999,
      "learning_rate": 5.735159347355818e-06,
      "loss": 0.045,
      "step": 8879
    },
    {
      "epoch": 0.458559256390395,
      "grad_norm": 3.736158847808838,
      "learning_rate": 5.7343488818377334e-06,
      "loss": 0.038,
      "step": 8880
    },
    {
      "epoch": 0.45861089594629484,
      "grad_norm": 6.5319414138793945,
      "learning_rate": 5.7335383965993895e-06,
      "loss": 0.0591,
      "step": 8881
    },
    {
      "epoch": 0.45866253550219466,
      "grad_norm": 2.2762503623962402,
      "learning_rate": 5.732727891662548e-06,
      "loss": 0.0353,
      "step": 8882
    },
    {
      "epoch": 0.4587141750580945,
      "grad_norm": 6.131175994873047,
      "learning_rate": 5.731917367048978e-06,
      "loss": 0.0488,
      "step": 8883
    },
    {
      "epoch": 0.4587658146139943,
      "grad_norm": 10.496789932250977,
      "learning_rate": 5.731106822780445e-06,
      "loss": 0.0633,
      "step": 8884
    },
    {
      "epoch": 0.4588174541698941,
      "grad_norm": 2.2988500595092773,
      "learning_rate": 5.730296258878712e-06,
      "loss": 0.0257,
      "step": 8885
    },
    {
      "epoch": 0.45886909372579393,
      "grad_norm": 5.849758148193359,
      "learning_rate": 5.729485675365547e-06,
      "loss": 0.0792,
      "step": 8886
    },
    {
      "epoch": 0.4589207332816938,
      "grad_norm": 4.867733478546143,
      "learning_rate": 5.728675072262719e-06,
      "loss": 0.0545,
      "step": 8887
    },
    {
      "epoch": 0.4589723728375936,
      "grad_norm": 4.250921726226807,
      "learning_rate": 5.7278644495919955e-06,
      "loss": 0.0377,
      "step": 8888
    },
    {
      "epoch": 0.45902401239349344,
      "grad_norm": 14.90375804901123,
      "learning_rate": 5.727053807375145e-06,
      "loss": 0.0522,
      "step": 8889
    },
    {
      "epoch": 0.45907565194939326,
      "grad_norm": 9.984821319580078,
      "learning_rate": 5.726243145633938e-06,
      "loss": 0.1067,
      "step": 8890
    },
    {
      "epoch": 0.4591272915052931,
      "grad_norm": 4.271368980407715,
      "learning_rate": 5.725432464390141e-06,
      "loss": 0.0603,
      "step": 8891
    },
    {
      "epoch": 0.4591789310611929,
      "grad_norm": 5.874868392944336,
      "learning_rate": 5.7246217636655244e-06,
      "loss": 0.037,
      "step": 8892
    },
    {
      "epoch": 0.4592305706170927,
      "grad_norm": 8.6014404296875,
      "learning_rate": 5.723811043481863e-06,
      "loss": 0.1426,
      "step": 8893
    },
    {
      "epoch": 0.45928221017299253,
      "grad_norm": 3.2768263816833496,
      "learning_rate": 5.723000303860922e-06,
      "loss": 0.0529,
      "step": 8894
    },
    {
      "epoch": 0.45933384972889235,
      "grad_norm": 4.693364143371582,
      "learning_rate": 5.722189544824478e-06,
      "loss": 0.0656,
      "step": 8895
    },
    {
      "epoch": 0.45938548928479217,
      "grad_norm": 3.4063496589660645,
      "learning_rate": 5.721378766394301e-06,
      "loss": 0.0574,
      "step": 8896
    },
    {
      "epoch": 0.459437128840692,
      "grad_norm": 12.549174308776855,
      "learning_rate": 5.720567968592163e-06,
      "loss": 0.0716,
      "step": 8897
    },
    {
      "epoch": 0.4594887683965918,
      "grad_norm": 4.5529327392578125,
      "learning_rate": 5.71975715143984e-06,
      "loss": 0.0499,
      "step": 8898
    },
    {
      "epoch": 0.4595404079524916,
      "grad_norm": 16.08479881286621,
      "learning_rate": 5.718946314959102e-06,
      "loss": 0.0749,
      "step": 8899
    },
    {
      "epoch": 0.45959204750839144,
      "grad_norm": 9.410578727722168,
      "learning_rate": 5.718135459171726e-06,
      "loss": 0.0696,
      "step": 8900
    },
    {
      "epoch": 0.45964368706429126,
      "grad_norm": 8.182133674621582,
      "learning_rate": 5.717324584099485e-06,
      "loss": 0.0613,
      "step": 8901
    },
    {
      "epoch": 0.4596953266201911,
      "grad_norm": 4.764195442199707,
      "learning_rate": 5.716513689764157e-06,
      "loss": 0.0516,
      "step": 8902
    },
    {
      "epoch": 0.4597469661760909,
      "grad_norm": 3.594081401824951,
      "learning_rate": 5.715702776187515e-06,
      "loss": 0.04,
      "step": 8903
    },
    {
      "epoch": 0.4597986057319907,
      "grad_norm": 3.906562566757202,
      "learning_rate": 5.714891843391337e-06,
      "loss": 0.0336,
      "step": 8904
    },
    {
      "epoch": 0.45985024528789054,
      "grad_norm": 4.644001007080078,
      "learning_rate": 5.7140808913973985e-06,
      "loss": 0.095,
      "step": 8905
    },
    {
      "epoch": 0.45990188484379035,
      "grad_norm": 4.401449203491211,
      "learning_rate": 5.7132699202274785e-06,
      "loss": 0.0411,
      "step": 8906
    },
    {
      "epoch": 0.4599535243996902,
      "grad_norm": 10.803647994995117,
      "learning_rate": 5.712458929903355e-06,
      "loss": 0.1327,
      "step": 8907
    },
    {
      "epoch": 0.46000516395559,
      "grad_norm": 4.068695545196533,
      "learning_rate": 5.711647920446802e-06,
      "loss": 0.0627,
      "step": 8908
    },
    {
      "epoch": 0.4600568035114898,
      "grad_norm": 15.109938621520996,
      "learning_rate": 5.710836891879605e-06,
      "loss": 0.0752,
      "step": 8909
    },
    {
      "epoch": 0.4601084430673896,
      "grad_norm": 3.6280579566955566,
      "learning_rate": 5.710025844223539e-06,
      "loss": 0.0561,
      "step": 8910
    },
    {
      "epoch": 0.46016008262328945,
      "grad_norm": 5.51384162902832,
      "learning_rate": 5.709214777500385e-06,
      "loss": 0.0324,
      "step": 8911
    },
    {
      "epoch": 0.46021172217918926,
      "grad_norm": 4.178517818450928,
      "learning_rate": 5.708403691731924e-06,
      "loss": 0.0513,
      "step": 8912
    },
    {
      "epoch": 0.4602633617350891,
      "grad_norm": 3.5215885639190674,
      "learning_rate": 5.7075925869399366e-06,
      "loss": 0.0513,
      "step": 8913
    },
    {
      "epoch": 0.4603150012909889,
      "grad_norm": 4.737057685852051,
      "learning_rate": 5.706781463146203e-06,
      "loss": 0.0621,
      "step": 8914
    },
    {
      "epoch": 0.4603666408468887,
      "grad_norm": 6.916408538818359,
      "learning_rate": 5.705970320372508e-06,
      "loss": 0.081,
      "step": 8915
    },
    {
      "epoch": 0.46041828040278854,
      "grad_norm": 6.301009654998779,
      "learning_rate": 5.705159158640633e-06,
      "loss": 0.0499,
      "step": 8916
    },
    {
      "epoch": 0.46046991995868836,
      "grad_norm": 2.299752950668335,
      "learning_rate": 5.7043479779723596e-06,
      "loss": 0.0345,
      "step": 8917
    },
    {
      "epoch": 0.4605215595145882,
      "grad_norm": 7.030038833618164,
      "learning_rate": 5.703536778389472e-06,
      "loss": 0.1156,
      "step": 8918
    },
    {
      "epoch": 0.460573199070488,
      "grad_norm": 5.81937313079834,
      "learning_rate": 5.702725559913754e-06,
      "loss": 0.057,
      "step": 8919
    },
    {
      "epoch": 0.4606248386263878,
      "grad_norm": 5.915347576141357,
      "learning_rate": 5.701914322566992e-06,
      "loss": 0.0664,
      "step": 8920
    },
    {
      "epoch": 0.46067647818228763,
      "grad_norm": 7.552699565887451,
      "learning_rate": 5.701103066370968e-06,
      "loss": 0.0688,
      "step": 8921
    },
    {
      "epoch": 0.46072811773818745,
      "grad_norm": 6.436141014099121,
      "learning_rate": 5.70029179134747e-06,
      "loss": 0.1047,
      "step": 8922
    },
    {
      "epoch": 0.46077975729408727,
      "grad_norm": 8.214173316955566,
      "learning_rate": 5.699480497518283e-06,
      "loss": 0.1451,
      "step": 8923
    },
    {
      "epoch": 0.4608313968499871,
      "grad_norm": 5.75526762008667,
      "learning_rate": 5.698669184905195e-06,
      "loss": 0.043,
      "step": 8924
    },
    {
      "epoch": 0.4608830364058869,
      "grad_norm": 2.766780138015747,
      "learning_rate": 5.697857853529989e-06,
      "loss": 0.0641,
      "step": 8925
    },
    {
      "epoch": 0.4609346759617867,
      "grad_norm": 3.861675500869751,
      "learning_rate": 5.697046503414459e-06,
      "loss": 0.0801,
      "step": 8926
    },
    {
      "epoch": 0.46098631551768654,
      "grad_norm": 3.357700824737549,
      "learning_rate": 5.696235134580386e-06,
      "loss": 0.0269,
      "step": 8927
    },
    {
      "epoch": 0.46103795507358636,
      "grad_norm": 6.057356834411621,
      "learning_rate": 5.695423747049563e-06,
      "loss": 0.0547,
      "step": 8928
    },
    {
      "epoch": 0.4610895946294862,
      "grad_norm": 3.8700389862060547,
      "learning_rate": 5.694612340843778e-06,
      "loss": 0.0528,
      "step": 8929
    },
    {
      "epoch": 0.461141234185386,
      "grad_norm": 5.444501876831055,
      "learning_rate": 5.69380091598482e-06,
      "loss": 0.0937,
      "step": 8930
    },
    {
      "epoch": 0.4611928737412858,
      "grad_norm": 4.042153358459473,
      "learning_rate": 5.692989472494482e-06,
      "loss": 0.0605,
      "step": 8931
    },
    {
      "epoch": 0.46124451329718563,
      "grad_norm": 2.2840864658355713,
      "learning_rate": 5.6921780103945504e-06,
      "loss": 0.0338,
      "step": 8932
    },
    {
      "epoch": 0.46129615285308545,
      "grad_norm": 5.371262550354004,
      "learning_rate": 5.691366529706819e-06,
      "loss": 0.0711,
      "step": 8933
    },
    {
      "epoch": 0.46134779240898527,
      "grad_norm": 2.825218677520752,
      "learning_rate": 5.690555030453075e-06,
      "loss": 0.0536,
      "step": 8934
    },
    {
      "epoch": 0.4613994319648851,
      "grad_norm": 19.501277923583984,
      "learning_rate": 5.689743512655118e-06,
      "loss": 0.1223,
      "step": 8935
    },
    {
      "epoch": 0.4614510715207849,
      "grad_norm": 3.9786295890808105,
      "learning_rate": 5.688931976334735e-06,
      "loss": 0.0611,
      "step": 8936
    },
    {
      "epoch": 0.4615027110766847,
      "grad_norm": 2.475705623626709,
      "learning_rate": 5.6881204215137205e-06,
      "loss": 0.0546,
      "step": 8937
    },
    {
      "epoch": 0.46155435063258454,
      "grad_norm": 8.039628982543945,
      "learning_rate": 5.6873088482138695e-06,
      "loss": 0.0604,
      "step": 8938
    },
    {
      "epoch": 0.46160599018848436,
      "grad_norm": 2.6693062782287598,
      "learning_rate": 5.686497256456972e-06,
      "loss": 0.0312,
      "step": 8939
    },
    {
      "epoch": 0.4616576297443842,
      "grad_norm": 2.674344778060913,
      "learning_rate": 5.685685646264827e-06,
      "loss": 0.045,
      "step": 8940
    },
    {
      "epoch": 0.461709269300284,
      "grad_norm": 3.416799545288086,
      "learning_rate": 5.684874017659225e-06,
      "loss": 0.0378,
      "step": 8941
    },
    {
      "epoch": 0.4617609088561838,
      "grad_norm": 14.624905586242676,
      "learning_rate": 5.684062370661968e-06,
      "loss": 0.1168,
      "step": 8942
    },
    {
      "epoch": 0.46181254841208363,
      "grad_norm": 6.731444835662842,
      "learning_rate": 5.683250705294847e-06,
      "loss": 0.0406,
      "step": 8943
    },
    {
      "epoch": 0.46186418796798345,
      "grad_norm": 4.293112277984619,
      "learning_rate": 5.6824390215796585e-06,
      "loss": 0.0747,
      "step": 8944
    },
    {
      "epoch": 0.46191582752388327,
      "grad_norm": 6.744246006011963,
      "learning_rate": 5.681627319538202e-06,
      "loss": 0.0498,
      "step": 8945
    },
    {
      "epoch": 0.4619674670797831,
      "grad_norm": 11.304388046264648,
      "learning_rate": 5.680815599192273e-06,
      "loss": 0.0643,
      "step": 8946
    },
    {
      "epoch": 0.4620191066356829,
      "grad_norm": 13.462407112121582,
      "learning_rate": 5.680003860563671e-06,
      "loss": 0.0973,
      "step": 8947
    },
    {
      "epoch": 0.4620707461915827,
      "grad_norm": 9.368328094482422,
      "learning_rate": 5.679192103674194e-06,
      "loss": 0.084,
      "step": 8948
    },
    {
      "epoch": 0.46212238574748254,
      "grad_norm": 8.430912971496582,
      "learning_rate": 5.67838032854564e-06,
      "loss": 0.0439,
      "step": 8949
    },
    {
      "epoch": 0.4621740253033824,
      "grad_norm": 6.462446212768555,
      "learning_rate": 5.67756853519981e-06,
      "loss": 0.0652,
      "step": 8950
    },
    {
      "epoch": 0.46222566485928224,
      "grad_norm": 2.6501076221466064,
      "learning_rate": 5.676756723658503e-06,
      "loss": 0.0494,
      "step": 8951
    },
    {
      "epoch": 0.46227730441518206,
      "grad_norm": 6.106532573699951,
      "learning_rate": 5.675944893943519e-06,
      "loss": 0.1032,
      "step": 8952
    },
    {
      "epoch": 0.4623289439710819,
      "grad_norm": 10.431337356567383,
      "learning_rate": 5.67513304607666e-06,
      "loss": 0.1797,
      "step": 8953
    },
    {
      "epoch": 0.4623805835269817,
      "grad_norm": 5.543924808502197,
      "learning_rate": 5.674321180079727e-06,
      "loss": 0.0404,
      "step": 8954
    },
    {
      "epoch": 0.4624322230828815,
      "grad_norm": 3.9599905014038086,
      "learning_rate": 5.673509295974522e-06,
      "loss": 0.0437,
      "step": 8955
    },
    {
      "epoch": 0.46248386263878133,
      "grad_norm": 1.4560872316360474,
      "learning_rate": 5.672697393782847e-06,
      "loss": 0.0248,
      "step": 8956
    },
    {
      "epoch": 0.46253550219468115,
      "grad_norm": 6.150458812713623,
      "learning_rate": 5.671885473526507e-06,
      "loss": 0.0715,
      "step": 8957
    },
    {
      "epoch": 0.46258714175058097,
      "grad_norm": 6.578962802886963,
      "learning_rate": 5.6710735352273005e-06,
      "loss": 0.0819,
      "step": 8958
    },
    {
      "epoch": 0.4626387813064808,
      "grad_norm": 14.538914680480957,
      "learning_rate": 5.670261578907037e-06,
      "loss": 0.1979,
      "step": 8959
    },
    {
      "epoch": 0.4626904208623806,
      "grad_norm": 14.834625244140625,
      "learning_rate": 5.669449604587518e-06,
      "loss": 0.1403,
      "step": 8960
    },
    {
      "epoch": 0.4627420604182804,
      "grad_norm": 2.7043168544769287,
      "learning_rate": 5.6686376122905474e-06,
      "loss": 0.0253,
      "step": 8961
    },
    {
      "epoch": 0.46279369997418024,
      "grad_norm": 4.1778483390808105,
      "learning_rate": 5.6678256020379315e-06,
      "loss": 0.0687,
      "step": 8962
    },
    {
      "epoch": 0.46284533953008006,
      "grad_norm": 3.495955228805542,
      "learning_rate": 5.6670135738514775e-06,
      "loss": 0.0664,
      "step": 8963
    },
    {
      "epoch": 0.4628969790859799,
      "grad_norm": 2.494381904602051,
      "learning_rate": 5.666201527752991e-06,
      "loss": 0.0458,
      "step": 8964
    },
    {
      "epoch": 0.4629486186418797,
      "grad_norm": 2.6931490898132324,
      "learning_rate": 5.665389463764276e-06,
      "loss": 0.0365,
      "step": 8965
    },
    {
      "epoch": 0.4630002581977795,
      "grad_norm": 3.9731292724609375,
      "learning_rate": 5.6645773819071434e-06,
      "loss": 0.0614,
      "step": 8966
    },
    {
      "epoch": 0.46305189775367933,
      "grad_norm": 6.284189224243164,
      "learning_rate": 5.663765282203399e-06,
      "loss": 0.0679,
      "step": 8967
    },
    {
      "epoch": 0.46310353730957915,
      "grad_norm": 3.0977859497070312,
      "learning_rate": 5.662953164674851e-06,
      "loss": 0.059,
      "step": 8968
    },
    {
      "epoch": 0.46315517686547897,
      "grad_norm": 9.498186111450195,
      "learning_rate": 5.66214102934331e-06,
      "loss": 0.0736,
      "step": 8969
    },
    {
      "epoch": 0.4632068164213788,
      "grad_norm": 6.372063636779785,
      "learning_rate": 5.661328876230583e-06,
      "loss": 0.0918,
      "step": 8970
    },
    {
      "epoch": 0.4632584559772786,
      "grad_norm": 6.798797607421875,
      "learning_rate": 5.66051670535848e-06,
      "loss": 0.0981,
      "step": 8971
    },
    {
      "epoch": 0.4633100955331784,
      "grad_norm": 2.9186394214630127,
      "learning_rate": 5.659704516748809e-06,
      "loss": 0.0331,
      "step": 8972
    },
    {
      "epoch": 0.46336173508907824,
      "grad_norm": 3.1252307891845703,
      "learning_rate": 5.658892310423387e-06,
      "loss": 0.0363,
      "step": 8973
    },
    {
      "epoch": 0.46341337464497806,
      "grad_norm": 3.4175565242767334,
      "learning_rate": 5.658080086404019e-06,
      "loss": 0.0562,
      "step": 8974
    },
    {
      "epoch": 0.4634650142008779,
      "grad_norm": 3.778993606567383,
      "learning_rate": 5.657267844712518e-06,
      "loss": 0.0365,
      "step": 8975
    },
    {
      "epoch": 0.4635166537567777,
      "grad_norm": 4.001579761505127,
      "learning_rate": 5.656455585370698e-06,
      "loss": 0.0807,
      "step": 8976
    },
    {
      "epoch": 0.4635682933126775,
      "grad_norm": 6.045547008514404,
      "learning_rate": 5.655643308400369e-06,
      "loss": 0.0688,
      "step": 8977
    },
    {
      "epoch": 0.46361993286857733,
      "grad_norm": 2.9079058170318604,
      "learning_rate": 5.654831013823344e-06,
      "loss": 0.0623,
      "step": 8978
    },
    {
      "epoch": 0.46367157242447715,
      "grad_norm": 9.148455619812012,
      "learning_rate": 5.654018701661439e-06,
      "loss": 0.0568,
      "step": 8979
    },
    {
      "epoch": 0.46372321198037697,
      "grad_norm": 11.561121940612793,
      "learning_rate": 5.653206371936465e-06,
      "loss": 0.1717,
      "step": 8980
    },
    {
      "epoch": 0.4637748515362768,
      "grad_norm": 3.5879013538360596,
      "learning_rate": 5.652394024670238e-06,
      "loss": 0.0409,
      "step": 8981
    },
    {
      "epoch": 0.4638264910921766,
      "grad_norm": 3.0039947032928467,
      "learning_rate": 5.651581659884574e-06,
      "loss": 0.0433,
      "step": 8982
    },
    {
      "epoch": 0.4638781306480764,
      "grad_norm": 3.8158180713653564,
      "learning_rate": 5.6507692776012846e-06,
      "loss": 0.0696,
      "step": 8983
    },
    {
      "epoch": 0.46392977020397624,
      "grad_norm": 11.380865097045898,
      "learning_rate": 5.649956877842189e-06,
      "loss": 0.052,
      "step": 8984
    },
    {
      "epoch": 0.46398140975987606,
      "grad_norm": 3.547295331954956,
      "learning_rate": 5.649144460629101e-06,
      "loss": 0.0324,
      "step": 8985
    },
    {
      "epoch": 0.4640330493157759,
      "grad_norm": 2.9027936458587646,
      "learning_rate": 5.648332025983839e-06,
      "loss": 0.0349,
      "step": 8986
    },
    {
      "epoch": 0.4640846888716757,
      "grad_norm": 2.4910635948181152,
      "learning_rate": 5.64751957392822e-06,
      "loss": 0.0385,
      "step": 8987
    },
    {
      "epoch": 0.4641363284275755,
      "grad_norm": 3.8851702213287354,
      "learning_rate": 5.646707104484061e-06,
      "loss": 0.0642,
      "step": 8988
    },
    {
      "epoch": 0.46418796798347534,
      "grad_norm": 2.474289655685425,
      "learning_rate": 5.64589461767318e-06,
      "loss": 0.0336,
      "step": 8989
    },
    {
      "epoch": 0.46423960753937515,
      "grad_norm": 8.808304786682129,
      "learning_rate": 5.6450821135173976e-06,
      "loss": 0.0563,
      "step": 8990
    },
    {
      "epoch": 0.464291247095275,
      "grad_norm": 3.065432548522949,
      "learning_rate": 5.644269592038528e-06,
      "loss": 0.0468,
      "step": 8991
    },
    {
      "epoch": 0.4643428866511748,
      "grad_norm": 4.3079986572265625,
      "learning_rate": 5.643457053258397e-06,
      "loss": 0.0576,
      "step": 8992
    },
    {
      "epoch": 0.4643945262070746,
      "grad_norm": 3.213611125946045,
      "learning_rate": 5.642644497198821e-06,
      "loss": 0.0437,
      "step": 8993
    },
    {
      "epoch": 0.46444616576297443,
      "grad_norm": 3.105473041534424,
      "learning_rate": 5.64183192388162e-06,
      "loss": 0.0589,
      "step": 8994
    },
    {
      "epoch": 0.46449780531887425,
      "grad_norm": 3.303598165512085,
      "learning_rate": 5.641019333328615e-06,
      "loss": 0.0363,
      "step": 8995
    },
    {
      "epoch": 0.46454944487477406,
      "grad_norm": 6.209897994995117,
      "learning_rate": 5.64020672556163e-06,
      "loss": 0.1055,
      "step": 8996
    },
    {
      "epoch": 0.4646010844306739,
      "grad_norm": 6.046745777130127,
      "learning_rate": 5.639394100602487e-06,
      "loss": 0.0479,
      "step": 8997
    },
    {
      "epoch": 0.4646527239865737,
      "grad_norm": 2.862152338027954,
      "learning_rate": 5.638581458473003e-06,
      "loss": 0.0487,
      "step": 8998
    },
    {
      "epoch": 0.4647043635424735,
      "grad_norm": 2.331977605819702,
      "learning_rate": 5.637768799195004e-06,
      "loss": 0.051,
      "step": 8999
    },
    {
      "epoch": 0.46475600309837334,
      "grad_norm": 5.8444390296936035,
      "learning_rate": 5.636956122790315e-06,
      "loss": 0.0683,
      "step": 9000
    },
    {
      "epoch": 0.46480764265427316,
      "grad_norm": 4.130648136138916,
      "learning_rate": 5.636143429280758e-06,
      "loss": 0.0773,
      "step": 9001
    },
    {
      "epoch": 0.464859282210173,
      "grad_norm": 4.034487724304199,
      "learning_rate": 5.6353307186881566e-06,
      "loss": 0.0693,
      "step": 9002
    },
    {
      "epoch": 0.4649109217660728,
      "grad_norm": 3.4974637031555176,
      "learning_rate": 5.634517991034336e-06,
      "loss": 0.1018,
      "step": 9003
    },
    {
      "epoch": 0.4649625613219726,
      "grad_norm": 10.576922416687012,
      "learning_rate": 5.633705246341123e-06,
      "loss": 0.3047,
      "step": 9004
    },
    {
      "epoch": 0.46501420087787243,
      "grad_norm": 25.507335662841797,
      "learning_rate": 5.632892484630337e-06,
      "loss": 0.1255,
      "step": 9005
    },
    {
      "epoch": 0.46506584043377225,
      "grad_norm": 11.33364200592041,
      "learning_rate": 5.632079705923812e-06,
      "loss": 0.0747,
      "step": 9006
    },
    {
      "epoch": 0.46511747998967207,
      "grad_norm": 38.3812141418457,
      "learning_rate": 5.631266910243368e-06,
      "loss": 0.1662,
      "step": 9007
    },
    {
      "epoch": 0.4651691195455719,
      "grad_norm": 11.478097915649414,
      "learning_rate": 5.630454097610838e-06,
      "loss": 0.0845,
      "step": 9008
    },
    {
      "epoch": 0.4652207591014717,
      "grad_norm": 5.534789562225342,
      "learning_rate": 5.6296412680480425e-06,
      "loss": 0.0459,
      "step": 9009
    },
    {
      "epoch": 0.4652723986573715,
      "grad_norm": 4.459678649902344,
      "learning_rate": 5.628828421576815e-06,
      "loss": 0.0559,
      "step": 9010
    },
    {
      "epoch": 0.46532403821327134,
      "grad_norm": 11.39450454711914,
      "learning_rate": 5.62801555821898e-06,
      "loss": 0.1247,
      "step": 9011
    },
    {
      "epoch": 0.46537567776917116,
      "grad_norm": 6.559330463409424,
      "learning_rate": 5.627202677996369e-06,
      "loss": 0.1064,
      "step": 9012
    },
    {
      "epoch": 0.465427317325071,
      "grad_norm": 12.298895835876465,
      "learning_rate": 5.626389780930808e-06,
      "loss": 0.1051,
      "step": 9013
    },
    {
      "epoch": 0.46547895688097085,
      "grad_norm": 1.5604439973831177,
      "learning_rate": 5.625576867044129e-06,
      "loss": 0.0367,
      "step": 9014
    },
    {
      "epoch": 0.46553059643687067,
      "grad_norm": 3.9956350326538086,
      "learning_rate": 5.6247639363581615e-06,
      "loss": 0.0994,
      "step": 9015
    },
    {
      "epoch": 0.4655822359927705,
      "grad_norm": 4.20225191116333,
      "learning_rate": 5.623950988894737e-06,
      "loss": 0.0584,
      "step": 9016
    },
    {
      "epoch": 0.4656338755486703,
      "grad_norm": 12.591989517211914,
      "learning_rate": 5.6231380246756836e-06,
      "loss": 0.1092,
      "step": 9017
    },
    {
      "epoch": 0.4656855151045701,
      "grad_norm": 2.229903221130371,
      "learning_rate": 5.622325043722836e-06,
      "loss": 0.0259,
      "step": 9018
    },
    {
      "epoch": 0.46573715466046994,
      "grad_norm": 4.687196254730225,
      "learning_rate": 5.621512046058023e-06,
      "loss": 0.0496,
      "step": 9019
    },
    {
      "epoch": 0.46578879421636976,
      "grad_norm": 3.33077335357666,
      "learning_rate": 5.6206990317030795e-06,
      "loss": 0.0573,
      "step": 9020
    },
    {
      "epoch": 0.4658404337722696,
      "grad_norm": 11.219880104064941,
      "learning_rate": 5.619886000679837e-06,
      "loss": 0.1003,
      "step": 9021
    },
    {
      "epoch": 0.4658920733281694,
      "grad_norm": 9.180246353149414,
      "learning_rate": 5.6190729530101286e-06,
      "loss": 0.0788,
      "step": 9022
    },
    {
      "epoch": 0.4659437128840692,
      "grad_norm": 3.5697617530822754,
      "learning_rate": 5.618259888715789e-06,
      "loss": 0.0489,
      "step": 9023
    },
    {
      "epoch": 0.46599535243996903,
      "grad_norm": 2.2813668251037598,
      "learning_rate": 5.6174468078186496e-06,
      "loss": 0.0497,
      "step": 9024
    },
    {
      "epoch": 0.46604699199586885,
      "grad_norm": 8.281752586364746,
      "learning_rate": 5.616633710340549e-06,
      "loss": 0.0897,
      "step": 9025
    },
    {
      "epoch": 0.46609863155176867,
      "grad_norm": 2.346449136734009,
      "learning_rate": 5.615820596303319e-06,
      "loss": 0.0368,
      "step": 9026
    },
    {
      "epoch": 0.4661502711076685,
      "grad_norm": 5.538516044616699,
      "learning_rate": 5.615007465728796e-06,
      "loss": 0.0713,
      "step": 9027
    },
    {
      "epoch": 0.4662019106635683,
      "grad_norm": 2.0529940128326416,
      "learning_rate": 5.614194318638816e-06,
      "loss": 0.0125,
      "step": 9028
    },
    {
      "epoch": 0.4662535502194681,
      "grad_norm": 10.373835563659668,
      "learning_rate": 5.613381155055215e-06,
      "loss": 0.0624,
      "step": 9029
    },
    {
      "epoch": 0.46630518977536795,
      "grad_norm": 2.0108492374420166,
      "learning_rate": 5.61256797499983e-06,
      "loss": 0.0472,
      "step": 9030
    },
    {
      "epoch": 0.46635682933126776,
      "grad_norm": 1.6161161661148071,
      "learning_rate": 5.611754778494499e-06,
      "loss": 0.0219,
      "step": 9031
    },
    {
      "epoch": 0.4664084688871676,
      "grad_norm": 3.5262317657470703,
      "learning_rate": 5.610941565561059e-06,
      "loss": 0.0646,
      "step": 9032
    },
    {
      "epoch": 0.4664601084430674,
      "grad_norm": 3.9658243656158447,
      "learning_rate": 5.610128336221347e-06,
      "loss": 0.0475,
      "step": 9033
    },
    {
      "epoch": 0.4665117479989672,
      "grad_norm": 2.3021697998046875,
      "learning_rate": 5.6093150904972025e-06,
      "loss": 0.041,
      "step": 9034
    },
    {
      "epoch": 0.46656338755486704,
      "grad_norm": 14.637970924377441,
      "learning_rate": 5.6085018284104646e-06,
      "loss": 0.0714,
      "step": 9035
    },
    {
      "epoch": 0.46661502711076686,
      "grad_norm": 3.130023717880249,
      "learning_rate": 5.607688549982973e-06,
      "loss": 0.0391,
      "step": 9036
    },
    {
      "epoch": 0.4666666666666667,
      "grad_norm": 11.21743106842041,
      "learning_rate": 5.6068752552365655e-06,
      "loss": 0.2183,
      "step": 9037
    },
    {
      "epoch": 0.4667183062225665,
      "grad_norm": 8.2517728805542,
      "learning_rate": 5.606061944193085e-06,
      "loss": 0.0818,
      "step": 9038
    },
    {
      "epoch": 0.4667699457784663,
      "grad_norm": 4.725163459777832,
      "learning_rate": 5.605248616874372e-06,
      "loss": 0.0516,
      "step": 9039
    },
    {
      "epoch": 0.46682158533436613,
      "grad_norm": 4.036818981170654,
      "learning_rate": 5.604435273302264e-06,
      "loss": 0.043,
      "step": 9040
    },
    {
      "epoch": 0.46687322489026595,
      "grad_norm": 2.948056697845459,
      "learning_rate": 5.603621913498608e-06,
      "loss": 0.0461,
      "step": 9041
    },
    {
      "epoch": 0.46692486444616577,
      "grad_norm": 3.3416905403137207,
      "learning_rate": 5.602808537485242e-06,
      "loss": 0.061,
      "step": 9042
    },
    {
      "epoch": 0.4669765040020656,
      "grad_norm": 9.796069145202637,
      "learning_rate": 5.601995145284011e-06,
      "loss": 0.0654,
      "step": 9043
    },
    {
      "epoch": 0.4670281435579654,
      "grad_norm": 9.217329025268555,
      "learning_rate": 5.601181736916756e-06,
      "loss": 0.0818,
      "step": 9044
    },
    {
      "epoch": 0.4670797831138652,
      "grad_norm": 2.855821132659912,
      "learning_rate": 5.600368312405321e-06,
      "loss": 0.0459,
      "step": 9045
    },
    {
      "epoch": 0.46713142266976504,
      "grad_norm": 4.710914611816406,
      "learning_rate": 5.59955487177155e-06,
      "loss": 0.0371,
      "step": 9046
    },
    {
      "epoch": 0.46718306222566486,
      "grad_norm": 11.94002628326416,
      "learning_rate": 5.5987414150372875e-06,
      "loss": 0.1116,
      "step": 9047
    },
    {
      "epoch": 0.4672347017815647,
      "grad_norm": 6.4204583168029785,
      "learning_rate": 5.597927942224378e-06,
      "loss": 0.0508,
      "step": 9048
    },
    {
      "epoch": 0.4672863413374645,
      "grad_norm": 4.33486795425415,
      "learning_rate": 5.597114453354665e-06,
      "loss": 0.0812,
      "step": 9049
    },
    {
      "epoch": 0.4673379808933643,
      "grad_norm": 4.2193684577941895,
      "learning_rate": 5.596300948449996e-06,
      "loss": 0.0498,
      "step": 9050
    },
    {
      "epoch": 0.46738962044926413,
      "grad_norm": 12.52634334564209,
      "learning_rate": 5.595487427532216e-06,
      "loss": 0.076,
      "step": 9051
    },
    {
      "epoch": 0.46744126000516395,
      "grad_norm": 27.378040313720703,
      "learning_rate": 5.594673890623171e-06,
      "loss": 0.1538,
      "step": 9052
    },
    {
      "epoch": 0.46749289956106377,
      "grad_norm": 8.627789497375488,
      "learning_rate": 5.593860337744708e-06,
      "loss": 0.211,
      "step": 9053
    },
    {
      "epoch": 0.4675445391169636,
      "grad_norm": 7.2560014724731445,
      "learning_rate": 5.593046768918675e-06,
      "loss": 0.0568,
      "step": 9054
    },
    {
      "epoch": 0.4675961786728634,
      "grad_norm": 4.019079208374023,
      "learning_rate": 5.592233184166919e-06,
      "loss": 0.0607,
      "step": 9055
    },
    {
      "epoch": 0.4676478182287632,
      "grad_norm": 5.437445163726807,
      "learning_rate": 5.591419583511289e-06,
      "loss": 0.0561,
      "step": 9056
    },
    {
      "epoch": 0.46769945778466304,
      "grad_norm": 2.859036445617676,
      "learning_rate": 5.590605966973632e-06,
      "loss": 0.0652,
      "step": 9057
    },
    {
      "epoch": 0.46775109734056286,
      "grad_norm": 2.6270618438720703,
      "learning_rate": 5.589792334575798e-06,
      "loss": 0.0344,
      "step": 9058
    },
    {
      "epoch": 0.4678027368964627,
      "grad_norm": 2.833503246307373,
      "learning_rate": 5.588978686339637e-06,
      "loss": 0.0486,
      "step": 9059
    },
    {
      "epoch": 0.4678543764523625,
      "grad_norm": 3.1970391273498535,
      "learning_rate": 5.588165022286996e-06,
      "loss": 0.0629,
      "step": 9060
    },
    {
      "epoch": 0.4679060160082623,
      "grad_norm": 2.149174690246582,
      "learning_rate": 5.587351342439728e-06,
      "loss": 0.0528,
      "step": 9061
    },
    {
      "epoch": 0.46795765556416213,
      "grad_norm": 3.4382827281951904,
      "learning_rate": 5.586537646819681e-06,
      "loss": 0.0525,
      "step": 9062
    },
    {
      "epoch": 0.46800929512006195,
      "grad_norm": 1.56455397605896,
      "learning_rate": 5.585723935448708e-06,
      "loss": 0.0208,
      "step": 9063
    },
    {
      "epoch": 0.46806093467596177,
      "grad_norm": 1.6975469589233398,
      "learning_rate": 5.58491020834866e-06,
      "loss": 0.0213,
      "step": 9064
    },
    {
      "epoch": 0.4681125742318616,
      "grad_norm": 7.208748817443848,
      "learning_rate": 5.584096465541391e-06,
      "loss": 0.0955,
      "step": 9065
    },
    {
      "epoch": 0.4681642137877614,
      "grad_norm": 2.0135817527770996,
      "learning_rate": 5.583282707048748e-06,
      "loss": 0.0366,
      "step": 9066
    },
    {
      "epoch": 0.4682158533436612,
      "grad_norm": 7.473797798156738,
      "learning_rate": 5.5824689328925905e-06,
      "loss": 0.096,
      "step": 9067
    },
    {
      "epoch": 0.46826749289956104,
      "grad_norm": 3.6165554523468018,
      "learning_rate": 5.581655143094766e-06,
      "loss": 0.0412,
      "step": 9068
    },
    {
      "epoch": 0.46831913245546086,
      "grad_norm": 3.672520637512207,
      "learning_rate": 5.580841337677131e-06,
      "loss": 0.0591,
      "step": 9069
    },
    {
      "epoch": 0.4683707720113607,
      "grad_norm": 3.606165885925293,
      "learning_rate": 5.580027516661538e-06,
      "loss": 0.1044,
      "step": 9070
    },
    {
      "epoch": 0.4684224115672605,
      "grad_norm": 5.489877223968506,
      "learning_rate": 5.579213680069842e-06,
      "loss": 0.0906,
      "step": 9071
    },
    {
      "epoch": 0.4684740511231603,
      "grad_norm": 7.216740608215332,
      "learning_rate": 5.5783998279239e-06,
      "loss": 0.1285,
      "step": 9072
    },
    {
      "epoch": 0.46852569067906014,
      "grad_norm": 3.855234384536743,
      "learning_rate": 5.577585960245562e-06,
      "loss": 0.038,
      "step": 9073
    },
    {
      "epoch": 0.46857733023495995,
      "grad_norm": 2.546739101409912,
      "learning_rate": 5.576772077056689e-06,
      "loss": 0.0401,
      "step": 9074
    },
    {
      "epoch": 0.4686289697908598,
      "grad_norm": 4.931612014770508,
      "learning_rate": 5.575958178379133e-06,
      "loss": 0.0624,
      "step": 9075
    },
    {
      "epoch": 0.4686806093467596,
      "grad_norm": 4.59319543838501,
      "learning_rate": 5.575144264234755e-06,
      "loss": 0.0743,
      "step": 9076
    },
    {
      "epoch": 0.4687322489026594,
      "grad_norm": 2.7431445121765137,
      "learning_rate": 5.574330334645408e-06,
      "loss": 0.0649,
      "step": 9077
    },
    {
      "epoch": 0.4687838884585593,
      "grad_norm": 6.727207660675049,
      "learning_rate": 5.573516389632951e-06,
      "loss": 0.0514,
      "step": 9078
    },
    {
      "epoch": 0.4688355280144591,
      "grad_norm": 8.008198738098145,
      "learning_rate": 5.572702429219242e-06,
      "loss": 0.1035,
      "step": 9079
    },
    {
      "epoch": 0.4688871675703589,
      "grad_norm": 19.87447166442871,
      "learning_rate": 5.571888453426137e-06,
      "loss": 0.1125,
      "step": 9080
    },
    {
      "epoch": 0.46893880712625874,
      "grad_norm": 6.630444526672363,
      "learning_rate": 5.571074462275497e-06,
      "loss": 0.0829,
      "step": 9081
    },
    {
      "epoch": 0.46899044668215856,
      "grad_norm": 5.020312309265137,
      "learning_rate": 5.570260455789179e-06,
      "loss": 0.0558,
      "step": 9082
    },
    {
      "epoch": 0.4690420862380584,
      "grad_norm": 4.362886905670166,
      "learning_rate": 5.569446433989045e-06,
      "loss": 0.0603,
      "step": 9083
    },
    {
      "epoch": 0.4690937257939582,
      "grad_norm": 3.9282681941986084,
      "learning_rate": 5.568632396896953e-06,
      "loss": 0.0392,
      "step": 9084
    },
    {
      "epoch": 0.469145365349858,
      "grad_norm": 4.603585243225098,
      "learning_rate": 5.567818344534763e-06,
      "loss": 0.0708,
      "step": 9085
    },
    {
      "epoch": 0.46919700490575783,
      "grad_norm": 4.518601417541504,
      "learning_rate": 5.5670042769243375e-06,
      "loss": 0.0944,
      "step": 9086
    },
    {
      "epoch": 0.46924864446165765,
      "grad_norm": 1.7757569551467896,
      "learning_rate": 5.566190194087535e-06,
      "loss": 0.0426,
      "step": 9087
    },
    {
      "epoch": 0.46930028401755747,
      "grad_norm": 7.402650356292725,
      "learning_rate": 5.565376096046218e-06,
      "loss": 0.0684,
      "step": 9088
    },
    {
      "epoch": 0.4693519235734573,
      "grad_norm": 4.760990619659424,
      "learning_rate": 5.564561982822248e-06,
      "loss": 0.0676,
      "step": 9089
    },
    {
      "epoch": 0.4694035631293571,
      "grad_norm": 6.904789924621582,
      "learning_rate": 5.56374785443749e-06,
      "loss": 0.0899,
      "step": 9090
    },
    {
      "epoch": 0.4694552026852569,
      "grad_norm": 8.703197479248047,
      "learning_rate": 5.562933710913802e-06,
      "loss": 0.0952,
      "step": 9091
    },
    {
      "epoch": 0.46950684224115674,
      "grad_norm": 3.1669461727142334,
      "learning_rate": 5.562119552273051e-06,
      "loss": 0.0344,
      "step": 9092
    },
    {
      "epoch": 0.46955848179705656,
      "grad_norm": 10.597832679748535,
      "learning_rate": 5.561305378537099e-06,
      "loss": 0.2018,
      "step": 9093
    },
    {
      "epoch": 0.4696101213529564,
      "grad_norm": 6.2573113441467285,
      "learning_rate": 5.560491189727809e-06,
      "loss": 0.0553,
      "step": 9094
    },
    {
      "epoch": 0.4696617609088562,
      "grad_norm": 6.526391506195068,
      "learning_rate": 5.559676985867046e-06,
      "loss": 0.0768,
      "step": 9095
    },
    {
      "epoch": 0.469713400464756,
      "grad_norm": 8.156620025634766,
      "learning_rate": 5.558862766976675e-06,
      "loss": 0.1274,
      "step": 9096
    },
    {
      "epoch": 0.46976504002065583,
      "grad_norm": 2.5615391731262207,
      "learning_rate": 5.55804853307856e-06,
      "loss": 0.0607,
      "step": 9097
    },
    {
      "epoch": 0.46981667957655565,
      "grad_norm": 10.521595001220703,
      "learning_rate": 5.5572342841945704e-06,
      "loss": 0.0554,
      "step": 9098
    },
    {
      "epoch": 0.46986831913245547,
      "grad_norm": 5.867640018463135,
      "learning_rate": 5.556420020346565e-06,
      "loss": 0.0679,
      "step": 9099
    },
    {
      "epoch": 0.4699199586883553,
      "grad_norm": 5.425601482391357,
      "learning_rate": 5.555605741556418e-06,
      "loss": 0.0249,
      "step": 9100
    },
    {
      "epoch": 0.4699715982442551,
      "grad_norm": 3.647430896759033,
      "learning_rate": 5.554791447845991e-06,
      "loss": 0.056,
      "step": 9101
    },
    {
      "epoch": 0.4700232378001549,
      "grad_norm": 1.7930032014846802,
      "learning_rate": 5.553977139237152e-06,
      "loss": 0.0197,
      "step": 9102
    },
    {
      "epoch": 0.47007487735605474,
      "grad_norm": 1.647397756576538,
      "learning_rate": 5.553162815751769e-06,
      "loss": 0.0386,
      "step": 9103
    },
    {
      "epoch": 0.47012651691195456,
      "grad_norm": 8.270790100097656,
      "learning_rate": 5.5523484774117094e-06,
      "loss": 0.0807,
      "step": 9104
    },
    {
      "epoch": 0.4701781564678544,
      "grad_norm": 2.9077327251434326,
      "learning_rate": 5.551534124238843e-06,
      "loss": 0.0776,
      "step": 9105
    },
    {
      "epoch": 0.4702297960237542,
      "grad_norm": 4.342676639556885,
      "learning_rate": 5.550719756255037e-06,
      "loss": 0.0721,
      "step": 9106
    },
    {
      "epoch": 0.470281435579654,
      "grad_norm": 4.08773136138916,
      "learning_rate": 5.549905373482161e-06,
      "loss": 0.0384,
      "step": 9107
    },
    {
      "epoch": 0.47033307513555384,
      "grad_norm": 4.9416022300720215,
      "learning_rate": 5.549090975942083e-06,
      "loss": 0.082,
      "step": 9108
    },
    {
      "epoch": 0.47038471469145365,
      "grad_norm": 2.574310064315796,
      "learning_rate": 5.548276563656676e-06,
      "loss": 0.041,
      "step": 9109
    },
    {
      "epoch": 0.47043635424735347,
      "grad_norm": 2.55466890335083,
      "learning_rate": 5.547462136647809e-06,
      "loss": 0.0606,
      "step": 9110
    },
    {
      "epoch": 0.4704879938032533,
      "grad_norm": 1.1902101039886475,
      "learning_rate": 5.54664769493735e-06,
      "loss": 0.0161,
      "step": 9111
    },
    {
      "epoch": 0.4705396333591531,
      "grad_norm": 4.678977012634277,
      "learning_rate": 5.545833238547176e-06,
      "loss": 0.069,
      "step": 9112
    },
    {
      "epoch": 0.4705912729150529,
      "grad_norm": 4.73297119140625,
      "learning_rate": 5.545018767499151e-06,
      "loss": 0.0641,
      "step": 9113
    },
    {
      "epoch": 0.47064291247095275,
      "grad_norm": 17.20794677734375,
      "learning_rate": 5.544204281815153e-06,
      "loss": 0.0559,
      "step": 9114
    },
    {
      "epoch": 0.47069455202685256,
      "grad_norm": 6.669273376464844,
      "learning_rate": 5.54338978151705e-06,
      "loss": 0.0498,
      "step": 9115
    },
    {
      "epoch": 0.4707461915827524,
      "grad_norm": 2.1619036197662354,
      "learning_rate": 5.54257526662672e-06,
      "loss": 0.0206,
      "step": 9116
    },
    {
      "epoch": 0.4707978311386522,
      "grad_norm": 4.691880702972412,
      "learning_rate": 5.54176073716603e-06,
      "loss": 0.0531,
      "step": 9117
    },
    {
      "epoch": 0.470849470694552,
      "grad_norm": 28.431684494018555,
      "learning_rate": 5.540946193156856e-06,
      "loss": 0.0777,
      "step": 9118
    },
    {
      "epoch": 0.47090111025045184,
      "grad_norm": 15.482056617736816,
      "learning_rate": 5.540131634621073e-06,
      "loss": 0.133,
      "step": 9119
    },
    {
      "epoch": 0.47095274980635166,
      "grad_norm": 9.11209487915039,
      "learning_rate": 5.5393170615805534e-06,
      "loss": 0.0595,
      "step": 9120
    },
    {
      "epoch": 0.4710043893622515,
      "grad_norm": 4.069493293762207,
      "learning_rate": 5.538502474057174e-06,
      "loss": 0.0894,
      "step": 9121
    },
    {
      "epoch": 0.4710560289181513,
      "grad_norm": 4.952864646911621,
      "learning_rate": 5.537687872072806e-06,
      "loss": 0.0302,
      "step": 9122
    },
    {
      "epoch": 0.4711076684740511,
      "grad_norm": 9.862452507019043,
      "learning_rate": 5.536873255649331e-06,
      "loss": 0.0814,
      "step": 9123
    },
    {
      "epoch": 0.47115930802995093,
      "grad_norm": 3.7460713386535645,
      "learning_rate": 5.5360586248086175e-06,
      "loss": 0.057,
      "step": 9124
    },
    {
      "epoch": 0.47121094758585075,
      "grad_norm": 6.301356315612793,
      "learning_rate": 5.535243979572546e-06,
      "loss": 0.0609,
      "step": 9125
    },
    {
      "epoch": 0.47126258714175057,
      "grad_norm": 3.4769785404205322,
      "learning_rate": 5.534429319962993e-06,
      "loss": 0.0522,
      "step": 9126
    },
    {
      "epoch": 0.4713142266976504,
      "grad_norm": 10.822174072265625,
      "learning_rate": 5.533614646001833e-06,
      "loss": 0.0746,
      "step": 9127
    },
    {
      "epoch": 0.4713658662535502,
      "grad_norm": 3.929283618927002,
      "learning_rate": 5.532799957710945e-06,
      "loss": 0.0643,
      "step": 9128
    },
    {
      "epoch": 0.47141750580945,
      "grad_norm": 7.8704729080200195,
      "learning_rate": 5.5319852551122074e-06,
      "loss": 0.0802,
      "step": 9129
    },
    {
      "epoch": 0.47146914536534984,
      "grad_norm": 3.870943546295166,
      "learning_rate": 5.531170538227497e-06,
      "loss": 0.0258,
      "step": 9130
    },
    {
      "epoch": 0.47152078492124966,
      "grad_norm": 2.584268808364868,
      "learning_rate": 5.5303558070786935e-06,
      "loss": 0.0413,
      "step": 9131
    },
    {
      "epoch": 0.4715724244771495,
      "grad_norm": 13.563692092895508,
      "learning_rate": 5.529541061687673e-06,
      "loss": 0.114,
      "step": 9132
    },
    {
      "epoch": 0.4716240640330493,
      "grad_norm": 3.7607338428497314,
      "learning_rate": 5.528726302076319e-06,
      "loss": 0.0309,
      "step": 9133
    },
    {
      "epoch": 0.4716757035889491,
      "grad_norm": 8.118681907653809,
      "learning_rate": 5.527911528266508e-06,
      "loss": 0.1577,
      "step": 9134
    },
    {
      "epoch": 0.47172734314484893,
      "grad_norm": 4.451262474060059,
      "learning_rate": 5.527096740280121e-06,
      "loss": 0.0609,
      "step": 9135
    },
    {
      "epoch": 0.47177898270074875,
      "grad_norm": 3.511754035949707,
      "learning_rate": 5.526281938139037e-06,
      "loss": 0.0799,
      "step": 9136
    },
    {
      "epoch": 0.47183062225664857,
      "grad_norm": 1.7137887477874756,
      "learning_rate": 5.525467121865139e-06,
      "loss": 0.0416,
      "step": 9137
    },
    {
      "epoch": 0.4718822618125484,
      "grad_norm": 14.587715148925781,
      "learning_rate": 5.524652291480308e-06,
      "loss": 0.0555,
      "step": 9138
    },
    {
      "epoch": 0.4719339013684482,
      "grad_norm": 3.7373387813568115,
      "learning_rate": 5.5238374470064226e-06,
      "loss": 0.0496,
      "step": 9139
    },
    {
      "epoch": 0.471985540924348,
      "grad_norm": 2.6476714611053467,
      "learning_rate": 5.523022588465369e-06,
      "loss": 0.0306,
      "step": 9140
    },
    {
      "epoch": 0.47203718048024784,
      "grad_norm": 14.26809310913086,
      "learning_rate": 5.522207715879024e-06,
      "loss": 0.0739,
      "step": 9141
    },
    {
      "epoch": 0.4720888200361477,
      "grad_norm": 6.085570335388184,
      "learning_rate": 5.521392829269276e-06,
      "loss": 0.0722,
      "step": 9142
    },
    {
      "epoch": 0.47214045959204753,
      "grad_norm": 1.5798245668411255,
      "learning_rate": 5.520577928658005e-06,
      "loss": 0.0411,
      "step": 9143
    },
    {
      "epoch": 0.47219209914794735,
      "grad_norm": 4.1414713859558105,
      "learning_rate": 5.519763014067094e-06,
      "loss": 0.1067,
      "step": 9144
    },
    {
      "epoch": 0.47224373870384717,
      "grad_norm": 4.824182033538818,
      "learning_rate": 5.5189480855184295e-06,
      "loss": 0.0695,
      "step": 9145
    },
    {
      "epoch": 0.472295378259747,
      "grad_norm": 5.8475117683410645,
      "learning_rate": 5.518133143033892e-06,
      "loss": 0.0854,
      "step": 9146
    },
    {
      "epoch": 0.4723470178156468,
      "grad_norm": 8.85750961303711,
      "learning_rate": 5.517318186635369e-06,
      "loss": 0.1158,
      "step": 9147
    },
    {
      "epoch": 0.4723986573715466,
      "grad_norm": 2.076413869857788,
      "learning_rate": 5.516503216344742e-06,
      "loss": 0.0388,
      "step": 9148
    },
    {
      "epoch": 0.47245029692744644,
      "grad_norm": 4.516427516937256,
      "learning_rate": 5.515688232183901e-06,
      "loss": 0.0269,
      "step": 9149
    },
    {
      "epoch": 0.47250193648334626,
      "grad_norm": 3.782144069671631,
      "learning_rate": 5.514873234174728e-06,
      "loss": 0.0424,
      "step": 9150
    },
    {
      "epoch": 0.4725535760392461,
      "grad_norm": 3.1192433834075928,
      "learning_rate": 5.51405822233911e-06,
      "loss": 0.0535,
      "step": 9151
    },
    {
      "epoch": 0.4726052155951459,
      "grad_norm": 2.60695481300354,
      "learning_rate": 5.513243196698934e-06,
      "loss": 0.0437,
      "step": 9152
    },
    {
      "epoch": 0.4726568551510457,
      "grad_norm": 2.82485032081604,
      "learning_rate": 5.5124281572760865e-06,
      "loss": 0.0392,
      "step": 9153
    },
    {
      "epoch": 0.47270849470694554,
      "grad_norm": 13.072650909423828,
      "learning_rate": 5.511613104092454e-06,
      "loss": 0.0451,
      "step": 9154
    },
    {
      "epoch": 0.47276013426284536,
      "grad_norm": 2.1960551738739014,
      "learning_rate": 5.510798037169924e-06,
      "loss": 0.0387,
      "step": 9155
    },
    {
      "epoch": 0.4728117738187452,
      "grad_norm": 8.540434837341309,
      "learning_rate": 5.509982956530387e-06,
      "loss": 0.0531,
      "step": 9156
    },
    {
      "epoch": 0.472863413374645,
      "grad_norm": 3.287870407104492,
      "learning_rate": 5.509167862195727e-06,
      "loss": 0.0934,
      "step": 9157
    },
    {
      "epoch": 0.4729150529305448,
      "grad_norm": 18.091211318969727,
      "learning_rate": 5.508352754187835e-06,
      "loss": 0.0936,
      "step": 9158
    },
    {
      "epoch": 0.47296669248644463,
      "grad_norm": 1.589706301689148,
      "learning_rate": 5.507537632528599e-06,
      "loss": 0.0324,
      "step": 9159
    },
    {
      "epoch": 0.47301833204234445,
      "grad_norm": 2.5977513790130615,
      "learning_rate": 5.506722497239909e-06,
      "loss": 0.0476,
      "step": 9160
    },
    {
      "epoch": 0.47306997159824427,
      "grad_norm": 3.6780831813812256,
      "learning_rate": 5.505907348343656e-06,
      "loss": 0.0329,
      "step": 9161
    },
    {
      "epoch": 0.4731216111541441,
      "grad_norm": 6.640153884887695,
      "learning_rate": 5.505092185861726e-06,
      "loss": 0.0651,
      "step": 9162
    },
    {
      "epoch": 0.4731732507100439,
      "grad_norm": 3.9421021938323975,
      "learning_rate": 5.5042770098160145e-06,
      "loss": 0.0486,
      "step": 9163
    },
    {
      "epoch": 0.4732248902659437,
      "grad_norm": 5.642761707305908,
      "learning_rate": 5.503461820228412e-06,
      "loss": 0.08,
      "step": 9164
    },
    {
      "epoch": 0.47327652982184354,
      "grad_norm": 7.388014793395996,
      "learning_rate": 5.5026466171208035e-06,
      "loss": 0.0424,
      "step": 9165
    },
    {
      "epoch": 0.47332816937774336,
      "grad_norm": 12.934983253479004,
      "learning_rate": 5.5018314005150875e-06,
      "loss": 0.0963,
      "step": 9166
    },
    {
      "epoch": 0.4733798089336432,
      "grad_norm": 1.8723957538604736,
      "learning_rate": 5.501016170433152e-06,
      "loss": 0.0115,
      "step": 9167
    },
    {
      "epoch": 0.473431448489543,
      "grad_norm": 4.699037075042725,
      "learning_rate": 5.500200926896891e-06,
      "loss": 0.0813,
      "step": 9168
    },
    {
      "epoch": 0.4734830880454428,
      "grad_norm": 4.800393104553223,
      "learning_rate": 5.499385669928195e-06,
      "loss": 0.0546,
      "step": 9169
    },
    {
      "epoch": 0.47353472760134263,
      "grad_norm": 6.9215407371521,
      "learning_rate": 5.498570399548958e-06,
      "loss": 0.082,
      "step": 9170
    },
    {
      "epoch": 0.47358636715724245,
      "grad_norm": 13.118595123291016,
      "learning_rate": 5.497755115781076e-06,
      "loss": 0.1066,
      "step": 9171
    },
    {
      "epoch": 0.47363800671314227,
      "grad_norm": 5.691271781921387,
      "learning_rate": 5.496939818646439e-06,
      "loss": 0.0678,
      "step": 9172
    },
    {
      "epoch": 0.4736896462690421,
      "grad_norm": 4.070315837860107,
      "learning_rate": 5.496124508166943e-06,
      "loss": 0.0472,
      "step": 9173
    },
    {
      "epoch": 0.4737412858249419,
      "grad_norm": 2.7872982025146484,
      "learning_rate": 5.49530918436448e-06,
      "loss": 0.0539,
      "step": 9174
    },
    {
      "epoch": 0.4737929253808417,
      "grad_norm": 4.010873317718506,
      "learning_rate": 5.494493847260949e-06,
      "loss": 0.0452,
      "step": 9175
    },
    {
      "epoch": 0.47384456493674154,
      "grad_norm": 4.12465763092041,
      "learning_rate": 5.493678496878242e-06,
      "loss": 0.1101,
      "step": 9176
    },
    {
      "epoch": 0.47389620449264136,
      "grad_norm": 2.3569161891937256,
      "learning_rate": 5.492863133238255e-06,
      "loss": 0.0212,
      "step": 9177
    },
    {
      "epoch": 0.4739478440485412,
      "grad_norm": 9.703729629516602,
      "learning_rate": 5.492047756362885e-06,
      "loss": 0.0969,
      "step": 9178
    },
    {
      "epoch": 0.473999483604441,
      "grad_norm": 4.004023551940918,
      "learning_rate": 5.491232366274025e-06,
      "loss": 0.0584,
      "step": 9179
    },
    {
      "epoch": 0.4740511231603408,
      "grad_norm": 6.352883815765381,
      "learning_rate": 5.490416962993577e-06,
      "loss": 0.0736,
      "step": 9180
    },
    {
      "epoch": 0.47410276271624063,
      "grad_norm": 4.437172889709473,
      "learning_rate": 5.48960154654343e-06,
      "loss": 0.0502,
      "step": 9181
    },
    {
      "epoch": 0.47415440227214045,
      "grad_norm": 3.2460789680480957,
      "learning_rate": 5.48878611694549e-06,
      "loss": 0.057,
      "step": 9182
    },
    {
      "epoch": 0.47420604182804027,
      "grad_norm": 9.757979393005371,
      "learning_rate": 5.48797067422165e-06,
      "loss": 0.08,
      "step": 9183
    },
    {
      "epoch": 0.4742576813839401,
      "grad_norm": 5.134242057800293,
      "learning_rate": 5.487155218393808e-06,
      "loss": 0.081,
      "step": 9184
    },
    {
      "epoch": 0.4743093209398399,
      "grad_norm": 2.056753158569336,
      "learning_rate": 5.486339749483863e-06,
      "loss": 0.0247,
      "step": 9185
    },
    {
      "epoch": 0.4743609604957397,
      "grad_norm": 10.391523361206055,
      "learning_rate": 5.485524267513713e-06,
      "loss": 0.0917,
      "step": 9186
    },
    {
      "epoch": 0.47441260005163954,
      "grad_norm": 3.4432075023651123,
      "learning_rate": 5.484708772505258e-06,
      "loss": 0.0699,
      "step": 9187
    },
    {
      "epoch": 0.47446423960753936,
      "grad_norm": 12.265400886535645,
      "learning_rate": 5.483893264480397e-06,
      "loss": 0.195,
      "step": 9188
    },
    {
      "epoch": 0.4745158791634392,
      "grad_norm": 3.858429193496704,
      "learning_rate": 5.48307774346103e-06,
      "loss": 0.0326,
      "step": 9189
    },
    {
      "epoch": 0.474567518719339,
      "grad_norm": 6.119214057922363,
      "learning_rate": 5.482262209469056e-06,
      "loss": 0.0776,
      "step": 9190
    },
    {
      "epoch": 0.4746191582752388,
      "grad_norm": 4.969279766082764,
      "learning_rate": 5.4814466625263766e-06,
      "loss": 0.0888,
      "step": 9191
    },
    {
      "epoch": 0.47467079783113864,
      "grad_norm": 14.49027156829834,
      "learning_rate": 5.480631102654891e-06,
      "loss": 0.2942,
      "step": 9192
    },
    {
      "epoch": 0.47472243738703845,
      "grad_norm": 3.803656578063965,
      "learning_rate": 5.4798155298765024e-06,
      "loss": 0.0951,
      "step": 9193
    },
    {
      "epoch": 0.4747740769429383,
      "grad_norm": 7.429201126098633,
      "learning_rate": 5.47899994421311e-06,
      "loss": 0.107,
      "step": 9194
    },
    {
      "epoch": 0.4748257164988381,
      "grad_norm": 3.1788382530212402,
      "learning_rate": 5.478184345686618e-06,
      "loss": 0.0538,
      "step": 9195
    },
    {
      "epoch": 0.4748773560547379,
      "grad_norm": 2.0682849884033203,
      "learning_rate": 5.477368734318926e-06,
      "loss": 0.0306,
      "step": 9196
    },
    {
      "epoch": 0.4749289956106377,
      "grad_norm": 5.532351493835449,
      "learning_rate": 5.47655311013194e-06,
      "loss": 0.0956,
      "step": 9197
    },
    {
      "epoch": 0.47498063516653755,
      "grad_norm": 3.7506096363067627,
      "learning_rate": 5.475737473147559e-06,
      "loss": 0.0673,
      "step": 9198
    },
    {
      "epoch": 0.47503227472243736,
      "grad_norm": 3.257295846939087,
      "learning_rate": 5.474921823387689e-06,
      "loss": 0.0484,
      "step": 9199
    },
    {
      "epoch": 0.4750839142783372,
      "grad_norm": 3.8947317600250244,
      "learning_rate": 5.474106160874232e-06,
      "loss": 0.0324,
      "step": 9200
    },
    {
      "epoch": 0.475135553834237,
      "grad_norm": 4.7660441398620605,
      "learning_rate": 5.473290485629091e-06,
      "loss": 0.0608,
      "step": 9201
    },
    {
      "epoch": 0.4751871933901368,
      "grad_norm": 8.861259460449219,
      "learning_rate": 5.472474797674172e-06,
      "loss": 0.1282,
      "step": 9202
    },
    {
      "epoch": 0.47523883294603664,
      "grad_norm": 8.673484802246094,
      "learning_rate": 5.47165909703138e-06,
      "loss": 0.0611,
      "step": 9203
    },
    {
      "epoch": 0.47529047250193646,
      "grad_norm": 6.425966262817383,
      "learning_rate": 5.47084338372262e-06,
      "loss": 0.0574,
      "step": 9204
    },
    {
      "epoch": 0.47534211205783633,
      "grad_norm": 1.3791255950927734,
      "learning_rate": 5.470027657769792e-06,
      "loss": 0.014,
      "step": 9205
    },
    {
      "epoch": 0.47539375161373615,
      "grad_norm": 3.0876777172088623,
      "learning_rate": 5.469211919194809e-06,
      "loss": 0.0574,
      "step": 9206
    },
    {
      "epoch": 0.47544539116963597,
      "grad_norm": 1.8131515979766846,
      "learning_rate": 5.46839616801957e-06,
      "loss": 0.0508,
      "step": 9207
    },
    {
      "epoch": 0.4754970307255358,
      "grad_norm": 9.567988395690918,
      "learning_rate": 5.46758040426599e-06,
      "loss": 0.1647,
      "step": 9208
    },
    {
      "epoch": 0.4755486702814356,
      "grad_norm": 2.4731626510620117,
      "learning_rate": 5.466764627955967e-06,
      "loss": 0.0385,
      "step": 9209
    },
    {
      "epoch": 0.4756003098373354,
      "grad_norm": 1.8233246803283691,
      "learning_rate": 5.465948839111412e-06,
      "loss": 0.0352,
      "step": 9210
    },
    {
      "epoch": 0.47565194939323524,
      "grad_norm": 4.838618278503418,
      "learning_rate": 5.465133037754232e-06,
      "loss": 0.0606,
      "step": 9211
    },
    {
      "epoch": 0.47570358894913506,
      "grad_norm": 2.3528521060943604,
      "learning_rate": 5.464317223906333e-06,
      "loss": 0.0284,
      "step": 9212
    },
    {
      "epoch": 0.4757552285050349,
      "grad_norm": 2.7208635807037354,
      "learning_rate": 5.463501397589626e-06,
      "loss": 0.0639,
      "step": 9213
    },
    {
      "epoch": 0.4758068680609347,
      "grad_norm": 2.269965887069702,
      "learning_rate": 5.462685558826015e-06,
      "loss": 0.0266,
      "step": 9214
    },
    {
      "epoch": 0.4758585076168345,
      "grad_norm": 6.36165714263916,
      "learning_rate": 5.461869707637413e-06,
      "loss": 0.0828,
      "step": 9215
    },
    {
      "epoch": 0.47591014717273433,
      "grad_norm": 6.215334892272949,
      "learning_rate": 5.461053844045726e-06,
      "loss": 0.0525,
      "step": 9216
    },
    {
      "epoch": 0.47596178672863415,
      "grad_norm": 2.96010684967041,
      "learning_rate": 5.460237968072864e-06,
      "loss": 0.0404,
      "step": 9217
    },
    {
      "epoch": 0.47601342628453397,
      "grad_norm": 7.802079200744629,
      "learning_rate": 5.459422079740737e-06,
      "loss": 0.0582,
      "step": 9218
    },
    {
      "epoch": 0.4760650658404338,
      "grad_norm": 2.26242733001709,
      "learning_rate": 5.458606179071254e-06,
      "loss": 0.0502,
      "step": 9219
    },
    {
      "epoch": 0.4761167053963336,
      "grad_norm": 3.9120070934295654,
      "learning_rate": 5.457790266086326e-06,
      "loss": 0.0615,
      "step": 9220
    },
    {
      "epoch": 0.4761683449522334,
      "grad_norm": 5.898311614990234,
      "learning_rate": 5.4569743408078646e-06,
      "loss": 0.0401,
      "step": 9221
    },
    {
      "epoch": 0.47621998450813324,
      "grad_norm": 17.139249801635742,
      "learning_rate": 5.45615840325778e-06,
      "loss": 0.1328,
      "step": 9222
    },
    {
      "epoch": 0.47627162406403306,
      "grad_norm": 4.986030101776123,
      "learning_rate": 5.455342453457981e-06,
      "loss": 0.0424,
      "step": 9223
    },
    {
      "epoch": 0.4763232636199329,
      "grad_norm": 3.044917106628418,
      "learning_rate": 5.454526491430383e-06,
      "loss": 0.0373,
      "step": 9224
    },
    {
      "epoch": 0.4763749031758327,
      "grad_norm": 5.508610248565674,
      "learning_rate": 5.453710517196895e-06,
      "loss": 0.0766,
      "step": 9225
    },
    {
      "epoch": 0.4764265427317325,
      "grad_norm": 7.5340423583984375,
      "learning_rate": 5.4528945307794314e-06,
      "loss": 0.0922,
      "step": 9226
    },
    {
      "epoch": 0.47647818228763233,
      "grad_norm": 4.0864152908325195,
      "learning_rate": 5.452078532199903e-06,
      "loss": 0.0492,
      "step": 9227
    },
    {
      "epoch": 0.47652982184353215,
      "grad_norm": 2.667128801345825,
      "learning_rate": 5.4512625214802225e-06,
      "loss": 0.0284,
      "step": 9228
    },
    {
      "epoch": 0.47658146139943197,
      "grad_norm": 16.58591079711914,
      "learning_rate": 5.450446498642305e-06,
      "loss": 0.0787,
      "step": 9229
    },
    {
      "epoch": 0.4766331009553318,
      "grad_norm": 15.315214157104492,
      "learning_rate": 5.449630463708064e-06,
      "loss": 0.1827,
      "step": 9230
    },
    {
      "epoch": 0.4766847405112316,
      "grad_norm": 4.443333625793457,
      "learning_rate": 5.448814416699413e-06,
      "loss": 0.0726,
      "step": 9231
    },
    {
      "epoch": 0.4767363800671314,
      "grad_norm": 3.449812650680542,
      "learning_rate": 5.447998357638263e-06,
      "loss": 0.0234,
      "step": 9232
    },
    {
      "epoch": 0.47678801962303125,
      "grad_norm": 3.100701332092285,
      "learning_rate": 5.447182286546533e-06,
      "loss": 0.0431,
      "step": 9233
    },
    {
      "epoch": 0.47683965917893106,
      "grad_norm": 4.0139899253845215,
      "learning_rate": 5.446366203446135e-06,
      "loss": 0.0493,
      "step": 9234
    },
    {
      "epoch": 0.4768912987348309,
      "grad_norm": 20.971105575561523,
      "learning_rate": 5.445550108358986e-06,
      "loss": 0.0895,
      "step": 9235
    },
    {
      "epoch": 0.4769429382907307,
      "grad_norm": 2.772704839706421,
      "learning_rate": 5.444734001307001e-06,
      "loss": 0.0195,
      "step": 9236
    },
    {
      "epoch": 0.4769945778466305,
      "grad_norm": 7.461729526519775,
      "learning_rate": 5.443917882312095e-06,
      "loss": 0.0864,
      "step": 9237
    },
    {
      "epoch": 0.47704621740253034,
      "grad_norm": 3.241259813308716,
      "learning_rate": 5.4431017513961824e-06,
      "loss": 0.0241,
      "step": 9238
    },
    {
      "epoch": 0.47709785695843016,
      "grad_norm": 16.452577590942383,
      "learning_rate": 5.442285608581185e-06,
      "loss": 0.056,
      "step": 9239
    },
    {
      "epoch": 0.47714949651433,
      "grad_norm": 5.332261085510254,
      "learning_rate": 5.441469453889015e-06,
      "loss": 0.0515,
      "step": 9240
    },
    {
      "epoch": 0.4772011360702298,
      "grad_norm": 4.832604885101318,
      "learning_rate": 5.4406532873415905e-06,
      "loss": 0.0426,
      "step": 9241
    },
    {
      "epoch": 0.4772527756261296,
      "grad_norm": 4.85270357131958,
      "learning_rate": 5.4398371089608305e-06,
      "loss": 0.0609,
      "step": 9242
    },
    {
      "epoch": 0.47730441518202943,
      "grad_norm": 2.6658523082733154,
      "learning_rate": 5.439020918768651e-06,
      "loss": 0.0451,
      "step": 9243
    },
    {
      "epoch": 0.47735605473792925,
      "grad_norm": 3.4897119998931885,
      "learning_rate": 5.43820471678697e-06,
      "loss": 0.056,
      "step": 9244
    },
    {
      "epoch": 0.47740769429382907,
      "grad_norm": 4.087083339691162,
      "learning_rate": 5.437388503037705e-06,
      "loss": 0.0915,
      "step": 9245
    },
    {
      "epoch": 0.4774593338497289,
      "grad_norm": 9.449213027954102,
      "learning_rate": 5.436572277542779e-06,
      "loss": 0.1325,
      "step": 9246
    },
    {
      "epoch": 0.4775109734056287,
      "grad_norm": 4.141473770141602,
      "learning_rate": 5.435756040324105e-06,
      "loss": 0.0767,
      "step": 9247
    },
    {
      "epoch": 0.4775626129615285,
      "grad_norm": 4.559571266174316,
      "learning_rate": 5.434939791403607e-06,
      "loss": 0.0555,
      "step": 9248
    },
    {
      "epoch": 0.47761425251742834,
      "grad_norm": 2.423129081726074,
      "learning_rate": 5.434123530803202e-06,
      "loss": 0.0243,
      "step": 9249
    },
    {
      "epoch": 0.47766589207332816,
      "grad_norm": 1.8109157085418701,
      "learning_rate": 5.4333072585448124e-06,
      "loss": 0.035,
      "step": 9250
    },
    {
      "epoch": 0.477717531629228,
      "grad_norm": 3.999758005142212,
      "learning_rate": 5.432490974650355e-06,
      "loss": 0.0125,
      "step": 9251
    },
    {
      "epoch": 0.4777691711851278,
      "grad_norm": 4.961874008178711,
      "learning_rate": 5.431674679141753e-06,
      "loss": 0.0627,
      "step": 9252
    },
    {
      "epoch": 0.4778208107410276,
      "grad_norm": 10.016940116882324,
      "learning_rate": 5.430858372040927e-06,
      "loss": 0.1979,
      "step": 9253
    },
    {
      "epoch": 0.47787245029692743,
      "grad_norm": 2.4291622638702393,
      "learning_rate": 5.4300420533697965e-06,
      "loss": 0.0378,
      "step": 9254
    },
    {
      "epoch": 0.47792408985282725,
      "grad_norm": 8.5559663772583,
      "learning_rate": 5.4292257231502855e-06,
      "loss": 0.0961,
      "step": 9255
    },
    {
      "epoch": 0.47797572940872707,
      "grad_norm": 5.220100402832031,
      "learning_rate": 5.428409381404313e-06,
      "loss": 0.0574,
      "step": 9256
    },
    {
      "epoch": 0.4780273689646269,
      "grad_norm": 4.914578914642334,
      "learning_rate": 5.4275930281538034e-06,
      "loss": 0.0846,
      "step": 9257
    },
    {
      "epoch": 0.4780790085205267,
      "grad_norm": 5.175954818725586,
      "learning_rate": 5.426776663420677e-06,
      "loss": 0.0571,
      "step": 9258
    },
    {
      "epoch": 0.4781306480764265,
      "grad_norm": 9.447514533996582,
      "learning_rate": 5.425960287226859e-06,
      "loss": 0.0626,
      "step": 9259
    },
    {
      "epoch": 0.47818228763232634,
      "grad_norm": 5.113521575927734,
      "learning_rate": 5.425143899594271e-06,
      "loss": 0.0297,
      "step": 9260
    },
    {
      "epoch": 0.47823392718822616,
      "grad_norm": 10.972641944885254,
      "learning_rate": 5.424327500544836e-06,
      "loss": 0.18,
      "step": 9261
    },
    {
      "epoch": 0.478285566744126,
      "grad_norm": 13.015661239624023,
      "learning_rate": 5.4235110901004785e-06,
      "loss": 0.1505,
      "step": 9262
    },
    {
      "epoch": 0.4783372063000258,
      "grad_norm": 2.662174701690674,
      "learning_rate": 5.422694668283123e-06,
      "loss": 0.0285,
      "step": 9263
    },
    {
      "epoch": 0.4783888458559256,
      "grad_norm": 7.893030643463135,
      "learning_rate": 5.421878235114691e-06,
      "loss": 0.0548,
      "step": 9264
    },
    {
      "epoch": 0.47844048541182543,
      "grad_norm": 2.069404125213623,
      "learning_rate": 5.42106179061711e-06,
      "loss": 0.0563,
      "step": 9265
    },
    {
      "epoch": 0.47849212496772525,
      "grad_norm": 1.6758536100387573,
      "learning_rate": 5.420245334812304e-06,
      "loss": 0.0194,
      "step": 9266
    },
    {
      "epoch": 0.47854376452362507,
      "grad_norm": 2.7121493816375732,
      "learning_rate": 5.419428867722197e-06,
      "loss": 0.0697,
      "step": 9267
    },
    {
      "epoch": 0.4785954040795249,
      "grad_norm": 5.265437602996826,
      "learning_rate": 5.418612389368716e-06,
      "loss": 0.0617,
      "step": 9268
    },
    {
      "epoch": 0.47864704363542476,
      "grad_norm": 5.507636070251465,
      "learning_rate": 5.4177958997737854e-06,
      "loss": 0.0958,
      "step": 9269
    },
    {
      "epoch": 0.4786986831913246,
      "grad_norm": 3.1163721084594727,
      "learning_rate": 5.416979398959334e-06,
      "loss": 0.0512,
      "step": 9270
    },
    {
      "epoch": 0.4787503227472244,
      "grad_norm": 4.533683776855469,
      "learning_rate": 5.416162886947284e-06,
      "loss": 0.0433,
      "step": 9271
    },
    {
      "epoch": 0.4788019623031242,
      "grad_norm": 7.417527198791504,
      "learning_rate": 5.415346363759566e-06,
      "loss": 0.0677,
      "step": 9272
    },
    {
      "epoch": 0.47885360185902404,
      "grad_norm": 5.996039867401123,
      "learning_rate": 5.414529829418104e-06,
      "loss": 0.0903,
      "step": 9273
    },
    {
      "epoch": 0.47890524141492385,
      "grad_norm": 3.6134073734283447,
      "learning_rate": 5.413713283944827e-06,
      "loss": 0.0546,
      "step": 9274
    },
    {
      "epoch": 0.4789568809708237,
      "grad_norm": 2.399406909942627,
      "learning_rate": 5.412896727361663e-06,
      "loss": 0.0478,
      "step": 9275
    },
    {
      "epoch": 0.4790085205267235,
      "grad_norm": 7.268205165863037,
      "learning_rate": 5.412080159690537e-06,
      "loss": 0.1309,
      "step": 9276
    },
    {
      "epoch": 0.4790601600826233,
      "grad_norm": 2.593878746032715,
      "learning_rate": 5.411263580953381e-06,
      "loss": 0.0335,
      "step": 9277
    },
    {
      "epoch": 0.47911179963852313,
      "grad_norm": 9.805397987365723,
      "learning_rate": 5.41044699117212e-06,
      "loss": 0.0898,
      "step": 9278
    },
    {
      "epoch": 0.47916343919442295,
      "grad_norm": 15.062356948852539,
      "learning_rate": 5.409630390368687e-06,
      "loss": 0.1424,
      "step": 9279
    },
    {
      "epoch": 0.47921507875032276,
      "grad_norm": 10.574895858764648,
      "learning_rate": 5.408813778565005e-06,
      "loss": 0.1557,
      "step": 9280
    },
    {
      "epoch": 0.4792667183062226,
      "grad_norm": 8.881555557250977,
      "learning_rate": 5.4079971557830105e-06,
      "loss": 0.0901,
      "step": 9281
    },
    {
      "epoch": 0.4793183578621224,
      "grad_norm": 1.4254140853881836,
      "learning_rate": 5.4071805220446276e-06,
      "loss": 0.0179,
      "step": 9282
    },
    {
      "epoch": 0.4793699974180222,
      "grad_norm": 4.5455217361450195,
      "learning_rate": 5.406363877371787e-06,
      "loss": 0.0341,
      "step": 9283
    },
    {
      "epoch": 0.47942163697392204,
      "grad_norm": 5.72359037399292,
      "learning_rate": 5.405547221786421e-06,
      "loss": 0.0482,
      "step": 9284
    },
    {
      "epoch": 0.47947327652982186,
      "grad_norm": 2.924238443374634,
      "learning_rate": 5.40473055531046e-06,
      "loss": 0.0645,
      "step": 9285
    },
    {
      "epoch": 0.4795249160857217,
      "grad_norm": 2.556774139404297,
      "learning_rate": 5.403913877965834e-06,
      "loss": 0.0286,
      "step": 9286
    },
    {
      "epoch": 0.4795765556416215,
      "grad_norm": 2.809436321258545,
      "learning_rate": 5.4030971897744726e-06,
      "loss": 0.0301,
      "step": 9287
    },
    {
      "epoch": 0.4796281951975213,
      "grad_norm": 5.123134613037109,
      "learning_rate": 5.4022804907583106e-06,
      "loss": 0.0815,
      "step": 9288
    },
    {
      "epoch": 0.47967983475342113,
      "grad_norm": 2.2778398990631104,
      "learning_rate": 5.401463780939277e-06,
      "loss": 0.0428,
      "step": 9289
    },
    {
      "epoch": 0.47973147430932095,
      "grad_norm": 5.372151851654053,
      "learning_rate": 5.400647060339303e-06,
      "loss": 0.0529,
      "step": 9290
    },
    {
      "epoch": 0.47978311386522077,
      "grad_norm": 3.991192579269409,
      "learning_rate": 5.399830328980324e-06,
      "loss": 0.0759,
      "step": 9291
    },
    {
      "epoch": 0.4798347534211206,
      "grad_norm": 4.294178485870361,
      "learning_rate": 5.399013586884272e-06,
      "loss": 0.0644,
      "step": 9292
    },
    {
      "epoch": 0.4798863929770204,
      "grad_norm": 1.976833462715149,
      "learning_rate": 5.398196834073077e-06,
      "loss": 0.0541,
      "step": 9293
    },
    {
      "epoch": 0.4799380325329202,
      "grad_norm": 4.836922645568848,
      "learning_rate": 5.397380070568675e-06,
      "loss": 0.0835,
      "step": 9294
    },
    {
      "epoch": 0.47998967208882004,
      "grad_norm": 9.581297874450684,
      "learning_rate": 5.396563296392998e-06,
      "loss": 0.0652,
      "step": 9295
    },
    {
      "epoch": 0.48004131164471986,
      "grad_norm": 6.9780683517456055,
      "learning_rate": 5.395746511567981e-06,
      "loss": 0.0523,
      "step": 9296
    },
    {
      "epoch": 0.4800929512006197,
      "grad_norm": 1.7959703207015991,
      "learning_rate": 5.394929716115558e-06,
      "loss": 0.0092,
      "step": 9297
    },
    {
      "epoch": 0.4801445907565195,
      "grad_norm": 2.962221145629883,
      "learning_rate": 5.39411291005766e-06,
      "loss": 0.0357,
      "step": 9298
    },
    {
      "epoch": 0.4801962303124193,
      "grad_norm": 21.157251358032227,
      "learning_rate": 5.393296093416226e-06,
      "loss": 0.0957,
      "step": 9299
    },
    {
      "epoch": 0.48024786986831913,
      "grad_norm": 3.260005474090576,
      "learning_rate": 5.392479266213188e-06,
      "loss": 0.0483,
      "step": 9300
    },
    {
      "epoch": 0.48029950942421895,
      "grad_norm": 8.126713752746582,
      "learning_rate": 5.391662428470482e-06,
      "loss": 0.089,
      "step": 9301
    },
    {
      "epoch": 0.48035114898011877,
      "grad_norm": 2.781550645828247,
      "learning_rate": 5.390845580210044e-06,
      "loss": 0.0559,
      "step": 9302
    },
    {
      "epoch": 0.4804027885360186,
      "grad_norm": 5.331477165222168,
      "learning_rate": 5.3900287214538084e-06,
      "loss": 0.0625,
      "step": 9303
    },
    {
      "epoch": 0.4804544280919184,
      "grad_norm": 3.177828311920166,
      "learning_rate": 5.389211852223713e-06,
      "loss": 0.0316,
      "step": 9304
    },
    {
      "epoch": 0.4805060676478182,
      "grad_norm": 2.323394536972046,
      "learning_rate": 5.388394972541693e-06,
      "loss": 0.0437,
      "step": 9305
    },
    {
      "epoch": 0.48055770720371804,
      "grad_norm": 7.077534198760986,
      "learning_rate": 5.387578082429683e-06,
      "loss": 0.0804,
      "step": 9306
    },
    {
      "epoch": 0.48060934675961786,
      "grad_norm": 3.0186588764190674,
      "learning_rate": 5.386761181909625e-06,
      "loss": 0.0396,
      "step": 9307
    },
    {
      "epoch": 0.4806609863155177,
      "grad_norm": 7.527096748352051,
      "learning_rate": 5.3859442710034505e-06,
      "loss": 0.037,
      "step": 9308
    },
    {
      "epoch": 0.4807126258714175,
      "grad_norm": 8.520776748657227,
      "learning_rate": 5.385127349733099e-06,
      "loss": 0.0694,
      "step": 9309
    },
    {
      "epoch": 0.4807642654273173,
      "grad_norm": 3.1343796253204346,
      "learning_rate": 5.38431041812051e-06,
      "loss": 0.066,
      "step": 9310
    },
    {
      "epoch": 0.48081590498321714,
      "grad_norm": 3.4458160400390625,
      "learning_rate": 5.383493476187619e-06,
      "loss": 0.0821,
      "step": 9311
    },
    {
      "epoch": 0.48086754453911695,
      "grad_norm": 5.911301612854004,
      "learning_rate": 5.382676523956366e-06,
      "loss": 0.064,
      "step": 9312
    },
    {
      "epoch": 0.48091918409501677,
      "grad_norm": 4.804929256439209,
      "learning_rate": 5.381859561448688e-06,
      "loss": 0.0419,
      "step": 9313
    },
    {
      "epoch": 0.4809708236509166,
      "grad_norm": 16.221149444580078,
      "learning_rate": 5.381042588686525e-06,
      "loss": 0.1509,
      "step": 9314
    },
    {
      "epoch": 0.4810224632068164,
      "grad_norm": 5.538272857666016,
      "learning_rate": 5.3802256056918135e-06,
      "loss": 0.0373,
      "step": 9315
    },
    {
      "epoch": 0.4810741027627162,
      "grad_norm": 5.723348140716553,
      "learning_rate": 5.379408612486498e-06,
      "loss": 0.0788,
      "step": 9316
    },
    {
      "epoch": 0.48112574231861605,
      "grad_norm": 2.036885976791382,
      "learning_rate": 5.378591609092513e-06,
      "loss": 0.0169,
      "step": 9317
    },
    {
      "epoch": 0.48117738187451586,
      "grad_norm": 2.3380796909332275,
      "learning_rate": 5.377774595531799e-06,
      "loss": 0.0364,
      "step": 9318
    },
    {
      "epoch": 0.4812290214304157,
      "grad_norm": 4.623100280761719,
      "learning_rate": 5.3769575718263e-06,
      "loss": 0.0486,
      "step": 9319
    },
    {
      "epoch": 0.4812806609863155,
      "grad_norm": 8.071537017822266,
      "learning_rate": 5.376140537997951e-06,
      "loss": 0.0537,
      "step": 9320
    },
    {
      "epoch": 0.4813323005422153,
      "grad_norm": 2.4138708114624023,
      "learning_rate": 5.3753234940686985e-06,
      "loss": 0.0361,
      "step": 9321
    },
    {
      "epoch": 0.48138394009811514,
      "grad_norm": 2.635572671890259,
      "learning_rate": 5.374506440060478e-06,
      "loss": 0.0509,
      "step": 9322
    },
    {
      "epoch": 0.48143557965401496,
      "grad_norm": 8.538670539855957,
      "learning_rate": 5.373689375995235e-06,
      "loss": 0.0483,
      "step": 9323
    },
    {
      "epoch": 0.4814872192099148,
      "grad_norm": 3.8419198989868164,
      "learning_rate": 5.372872301894908e-06,
      "loss": 0.0478,
      "step": 9324
    },
    {
      "epoch": 0.4815388587658146,
      "grad_norm": 6.227538585662842,
      "learning_rate": 5.37205521778144e-06,
      "loss": 0.0528,
      "step": 9325
    },
    {
      "epoch": 0.4815904983217144,
      "grad_norm": 11.089951515197754,
      "learning_rate": 5.371238123676773e-06,
      "loss": 0.0539,
      "step": 9326
    },
    {
      "epoch": 0.48164213787761423,
      "grad_norm": 10.239590644836426,
      "learning_rate": 5.37042101960285e-06,
      "loss": 0.0831,
      "step": 9327
    },
    {
      "epoch": 0.48169377743351405,
      "grad_norm": 5.002222537994385,
      "learning_rate": 5.369603905581612e-06,
      "loss": 0.0542,
      "step": 9328
    },
    {
      "epoch": 0.48174541698941387,
      "grad_norm": 2.0752038955688477,
      "learning_rate": 5.368786781635004e-06,
      "loss": 0.0153,
      "step": 9329
    },
    {
      "epoch": 0.4817970565453137,
      "grad_norm": 2.7959165573120117,
      "learning_rate": 5.367969647784967e-06,
      "loss": 0.0242,
      "step": 9330
    },
    {
      "epoch": 0.4818486961012135,
      "grad_norm": 7.475262641906738,
      "learning_rate": 5.367152504053446e-06,
      "loss": 0.1032,
      "step": 9331
    },
    {
      "epoch": 0.4819003356571133,
      "grad_norm": 4.007564544677734,
      "learning_rate": 5.366335350462382e-06,
      "loss": 0.0334,
      "step": 9332
    },
    {
      "epoch": 0.4819519752130132,
      "grad_norm": 6.009946346282959,
      "learning_rate": 5.365518187033722e-06,
      "loss": 0.0406,
      "step": 9333
    },
    {
      "epoch": 0.482003614768913,
      "grad_norm": 6.091977596282959,
      "learning_rate": 5.364701013789408e-06,
      "loss": 0.0594,
      "step": 9334
    },
    {
      "epoch": 0.48205525432481283,
      "grad_norm": 3.928856611251831,
      "learning_rate": 5.363883830751386e-06,
      "loss": 0.0387,
      "step": 9335
    },
    {
      "epoch": 0.48210689388071265,
      "grad_norm": 3.5436480045318604,
      "learning_rate": 5.3630666379415995e-06,
      "loss": 0.0438,
      "step": 9336
    },
    {
      "epoch": 0.48215853343661247,
      "grad_norm": 6.373096942901611,
      "learning_rate": 5.362249435381994e-06,
      "loss": 0.0592,
      "step": 9337
    },
    {
      "epoch": 0.4822101729925123,
      "grad_norm": 2.2774765491485596,
      "learning_rate": 5.361432223094517e-06,
      "loss": 0.0183,
      "step": 9338
    },
    {
      "epoch": 0.4822618125484121,
      "grad_norm": 6.564982891082764,
      "learning_rate": 5.360615001101108e-06,
      "loss": 0.0606,
      "step": 9339
    },
    {
      "epoch": 0.4823134521043119,
      "grad_norm": 4.08029317855835,
      "learning_rate": 5.35979776942372e-06,
      "loss": 0.0706,
      "step": 9340
    },
    {
      "epoch": 0.48236509166021174,
      "grad_norm": 1.6540473699569702,
      "learning_rate": 5.3589805280842945e-06,
      "loss": 0.0368,
      "step": 9341
    },
    {
      "epoch": 0.48241673121611156,
      "grad_norm": 15.841660499572754,
      "learning_rate": 5.3581632771047774e-06,
      "loss": 0.1014,
      "step": 9342
    },
    {
      "epoch": 0.4824683707720114,
      "grad_norm": 3.9594204425811768,
      "learning_rate": 5.3573460165071165e-06,
      "loss": 0.0332,
      "step": 9343
    },
    {
      "epoch": 0.4825200103279112,
      "grad_norm": 6.033344268798828,
      "learning_rate": 5.35652874631326e-06,
      "loss": 0.0372,
      "step": 9344
    },
    {
      "epoch": 0.482571649883811,
      "grad_norm": 7.736143112182617,
      "learning_rate": 5.355711466545152e-06,
      "loss": 0.066,
      "step": 9345
    },
    {
      "epoch": 0.48262328943971083,
      "grad_norm": 3.7348108291625977,
      "learning_rate": 5.354894177224742e-06,
      "loss": 0.0647,
      "step": 9346
    },
    {
      "epoch": 0.48267492899561065,
      "grad_norm": 13.48324203491211,
      "learning_rate": 5.354076878373978e-06,
      "loss": 0.2098,
      "step": 9347
    },
    {
      "epoch": 0.48272656855151047,
      "grad_norm": 14.14892578125,
      "learning_rate": 5.353259570014804e-06,
      "loss": 0.1143,
      "step": 9348
    },
    {
      "epoch": 0.4827782081074103,
      "grad_norm": 2.927208185195923,
      "learning_rate": 5.352442252169173e-06,
      "loss": 0.0584,
      "step": 9349
    },
    {
      "epoch": 0.4828298476633101,
      "grad_norm": 3.778743267059326,
      "learning_rate": 5.351624924859031e-06,
      "loss": 0.0486,
      "step": 9350
    },
    {
      "epoch": 0.4828814872192099,
      "grad_norm": 2.7919085025787354,
      "learning_rate": 5.350807588106326e-06,
      "loss": 0.0282,
      "step": 9351
    },
    {
      "epoch": 0.48293312677510974,
      "grad_norm": 9.22179889678955,
      "learning_rate": 5.349990241933009e-06,
      "loss": 0.0912,
      "step": 9352
    },
    {
      "epoch": 0.48298476633100956,
      "grad_norm": 3.5909500122070312,
      "learning_rate": 5.349172886361024e-06,
      "loss": 0.0605,
      "step": 9353
    },
    {
      "epoch": 0.4830364058869094,
      "grad_norm": 12.168095588684082,
      "learning_rate": 5.348355521412327e-06,
      "loss": 0.0963,
      "step": 9354
    },
    {
      "epoch": 0.4830880454428092,
      "grad_norm": 2.592155933380127,
      "learning_rate": 5.347538147108862e-06,
      "loss": 0.0445,
      "step": 9355
    },
    {
      "epoch": 0.483139684998709,
      "grad_norm": 8.377578735351562,
      "learning_rate": 5.346720763472583e-06,
      "loss": 0.1134,
      "step": 9356
    },
    {
      "epoch": 0.48319132455460884,
      "grad_norm": 6.634963035583496,
      "learning_rate": 5.345903370525438e-06,
      "loss": 0.1425,
      "step": 9357
    },
    {
      "epoch": 0.48324296411050865,
      "grad_norm": 2.6438634395599365,
      "learning_rate": 5.345085968289378e-06,
      "loss": 0.0446,
      "step": 9358
    },
    {
      "epoch": 0.4832946036664085,
      "grad_norm": 3.2651097774505615,
      "learning_rate": 5.344268556786352e-06,
      "loss": 0.0415,
      "step": 9359
    },
    {
      "epoch": 0.4833462432223083,
      "grad_norm": 2.225430488586426,
      "learning_rate": 5.343451136038313e-06,
      "loss": 0.0394,
      "step": 9360
    },
    {
      "epoch": 0.4833978827782081,
      "grad_norm": 2.6921944618225098,
      "learning_rate": 5.342633706067211e-06,
      "loss": 0.0286,
      "step": 9361
    },
    {
      "epoch": 0.48344952233410793,
      "grad_norm": 3.3332958221435547,
      "learning_rate": 5.341816266894998e-06,
      "loss": 0.0516,
      "step": 9362
    },
    {
      "epoch": 0.48350116189000775,
      "grad_norm": 3.718372344970703,
      "learning_rate": 5.340998818543626e-06,
      "loss": 0.0627,
      "step": 9363
    },
    {
      "epoch": 0.48355280144590757,
      "grad_norm": 3.397547483444214,
      "learning_rate": 5.340181361035044e-06,
      "loss": 0.0443,
      "step": 9364
    },
    {
      "epoch": 0.4836044410018074,
      "grad_norm": 11.442346572875977,
      "learning_rate": 5.339363894391206e-06,
      "loss": 0.0728,
      "step": 9365
    },
    {
      "epoch": 0.4836560805577072,
      "grad_norm": 2.0034470558166504,
      "learning_rate": 5.338546418634064e-06,
      "loss": 0.0374,
      "step": 9366
    },
    {
      "epoch": 0.483707720113607,
      "grad_norm": 3.0492169857025146,
      "learning_rate": 5.337728933785572e-06,
      "loss": 0.062,
      "step": 9367
    },
    {
      "epoch": 0.48375935966950684,
      "grad_norm": 2.22271466255188,
      "learning_rate": 5.336911439867681e-06,
      "loss": 0.0279,
      "step": 9368
    },
    {
      "epoch": 0.48381099922540666,
      "grad_norm": 2.1601173877716064,
      "learning_rate": 5.336093936902345e-06,
      "loss": 0.0247,
      "step": 9369
    },
    {
      "epoch": 0.4838626387813065,
      "grad_norm": 5.562097549438477,
      "learning_rate": 5.335276424911516e-06,
      "loss": 0.0631,
      "step": 9370
    },
    {
      "epoch": 0.4839142783372063,
      "grad_norm": 3.0823116302490234,
      "learning_rate": 5.33445890391715e-06,
      "loss": 0.0507,
      "step": 9371
    },
    {
      "epoch": 0.4839659178931061,
      "grad_norm": 3.4993464946746826,
      "learning_rate": 5.3336413739411986e-06,
      "loss": 0.0792,
      "step": 9372
    },
    {
      "epoch": 0.48401755744900593,
      "grad_norm": 1.7267358303070068,
      "learning_rate": 5.332823835005617e-06,
      "loss": 0.0303,
      "step": 9373
    },
    {
      "epoch": 0.48406919700490575,
      "grad_norm": 2.4399709701538086,
      "learning_rate": 5.332006287132359e-06,
      "loss": 0.0425,
      "step": 9374
    },
    {
      "epoch": 0.48412083656080557,
      "grad_norm": 4.301270961761475,
      "learning_rate": 5.331188730343377e-06,
      "loss": 0.0596,
      "step": 9375
    },
    {
      "epoch": 0.4841724761167054,
      "grad_norm": 5.527924537658691,
      "learning_rate": 5.33037116466063e-06,
      "loss": 0.0237,
      "step": 9376
    },
    {
      "epoch": 0.4842241156726052,
      "grad_norm": 1.5991865396499634,
      "learning_rate": 5.329553590106069e-06,
      "loss": 0.0215,
      "step": 9377
    },
    {
      "epoch": 0.484275755228505,
      "grad_norm": 14.65755558013916,
      "learning_rate": 5.328736006701652e-06,
      "loss": 0.0781,
      "step": 9378
    },
    {
      "epoch": 0.48432739478440484,
      "grad_norm": 2.9059438705444336,
      "learning_rate": 5.327918414469332e-06,
      "loss": 0.0336,
      "step": 9379
    },
    {
      "epoch": 0.48437903434030466,
      "grad_norm": 3.0810747146606445,
      "learning_rate": 5.327100813431068e-06,
      "loss": 0.0498,
      "step": 9380
    },
    {
      "epoch": 0.4844306738962045,
      "grad_norm": 5.883467674255371,
      "learning_rate": 5.326283203608812e-06,
      "loss": 0.0512,
      "step": 9381
    },
    {
      "epoch": 0.4844823134521043,
      "grad_norm": 8.646099090576172,
      "learning_rate": 5.325465585024524e-06,
      "loss": 0.0387,
      "step": 9382
    },
    {
      "epoch": 0.4845339530080041,
      "grad_norm": 2.915940999984741,
      "learning_rate": 5.3246479577001575e-06,
      "loss": 0.0277,
      "step": 9383
    },
    {
      "epoch": 0.48458559256390393,
      "grad_norm": 3.833702564239502,
      "learning_rate": 5.32383032165767e-06,
      "loss": 0.0326,
      "step": 9384
    },
    {
      "epoch": 0.48463723211980375,
      "grad_norm": 7.068485260009766,
      "learning_rate": 5.32301267691902e-06,
      "loss": 0.1469,
      "step": 9385
    },
    {
      "epoch": 0.48468887167570357,
      "grad_norm": 3.1662440299987793,
      "learning_rate": 5.322195023506161e-06,
      "loss": 0.0449,
      "step": 9386
    },
    {
      "epoch": 0.4847405112316034,
      "grad_norm": 2.0858335494995117,
      "learning_rate": 5.321377361441055e-06,
      "loss": 0.0306,
      "step": 9387
    },
    {
      "epoch": 0.4847921507875032,
      "grad_norm": 11.752142906188965,
      "learning_rate": 5.320559690745654e-06,
      "loss": 0.1062,
      "step": 9388
    },
    {
      "epoch": 0.484843790343403,
      "grad_norm": 4.012402534484863,
      "learning_rate": 5.319742011441921e-06,
      "loss": 0.0509,
      "step": 9389
    },
    {
      "epoch": 0.48489542989930284,
      "grad_norm": 2.4933724403381348,
      "learning_rate": 5.318924323551811e-06,
      "loss": 0.0273,
      "step": 9390
    },
    {
      "epoch": 0.48494706945520266,
      "grad_norm": 5.345174312591553,
      "learning_rate": 5.318106627097282e-06,
      "loss": 0.0454,
      "step": 9391
    },
    {
      "epoch": 0.4849987090111025,
      "grad_norm": 8.378063201904297,
      "learning_rate": 5.317288922100296e-06,
      "loss": 0.2024,
      "step": 9392
    },
    {
      "epoch": 0.4850503485670023,
      "grad_norm": 5.526030540466309,
      "learning_rate": 5.316471208582807e-06,
      "loss": 0.0608,
      "step": 9393
    },
    {
      "epoch": 0.4851019881229021,
      "grad_norm": 5.648934841156006,
      "learning_rate": 5.3156534865667775e-06,
      "loss": 0.101,
      "step": 9394
    },
    {
      "epoch": 0.48515362767880194,
      "grad_norm": 5.344666004180908,
      "learning_rate": 5.314835756074164e-06,
      "loss": 0.0489,
      "step": 9395
    },
    {
      "epoch": 0.4852052672347018,
      "grad_norm": 4.384760856628418,
      "learning_rate": 5.314018017126929e-06,
      "loss": 0.0425,
      "step": 9396
    },
    {
      "epoch": 0.4852569067906016,
      "grad_norm": 19.366249084472656,
      "learning_rate": 5.31320026974703e-06,
      "loss": 0.0806,
      "step": 9397
    },
    {
      "epoch": 0.48530854634650145,
      "grad_norm": 4.858604431152344,
      "learning_rate": 5.312382513956425e-06,
      "loss": 0.0408,
      "step": 9398
    },
    {
      "epoch": 0.48536018590240126,
      "grad_norm": 1.2122942209243774,
      "learning_rate": 5.311564749777078e-06,
      "loss": 0.0272,
      "step": 9399
    },
    {
      "epoch": 0.4854118254583011,
      "grad_norm": 7.9657883644104,
      "learning_rate": 5.310746977230948e-06,
      "loss": 0.0664,
      "step": 9400
    },
    {
      "epoch": 0.4854634650142009,
      "grad_norm": 38.25862503051758,
      "learning_rate": 5.309929196339995e-06,
      "loss": 0.1309,
      "step": 9401
    },
    {
      "epoch": 0.4855151045701007,
      "grad_norm": 4.902371406555176,
      "learning_rate": 5.309111407126179e-06,
      "loss": 0.0997,
      "step": 9402
    },
    {
      "epoch": 0.48556674412600054,
      "grad_norm": 2.6675772666931152,
      "learning_rate": 5.3082936096114626e-06,
      "loss": 0.0683,
      "step": 9403
    },
    {
      "epoch": 0.48561838368190036,
      "grad_norm": 3.6944587230682373,
      "learning_rate": 5.307475803817806e-06,
      "loss": 0.1042,
      "step": 9404
    },
    {
      "epoch": 0.4856700232378002,
      "grad_norm": 4.867122650146484,
      "learning_rate": 5.306657989767171e-06,
      "loss": 0.0491,
      "step": 9405
    },
    {
      "epoch": 0.4857216627937,
      "grad_norm": 7.458892822265625,
      "learning_rate": 5.305840167481519e-06,
      "loss": 0.0541,
      "step": 9406
    },
    {
      "epoch": 0.4857733023495998,
      "grad_norm": 5.497622013092041,
      "learning_rate": 5.305022336982812e-06,
      "loss": 0.1434,
      "step": 9407
    },
    {
      "epoch": 0.48582494190549963,
      "grad_norm": 9.998059272766113,
      "learning_rate": 5.3042044982930115e-06,
      "loss": 0.0692,
      "step": 9408
    },
    {
      "epoch": 0.48587658146139945,
      "grad_norm": 3.216501235961914,
      "learning_rate": 5.303386651434081e-06,
      "loss": 0.0664,
      "step": 9409
    },
    {
      "epoch": 0.48592822101729927,
      "grad_norm": 4.059179782867432,
      "learning_rate": 5.302568796427983e-06,
      "loss": 0.055,
      "step": 9410
    },
    {
      "epoch": 0.4859798605731991,
      "grad_norm": 7.2860426902771,
      "learning_rate": 5.30175093329668e-06,
      "loss": 0.0685,
      "step": 9411
    },
    {
      "epoch": 0.4860315001290989,
      "grad_norm": 4.344616413116455,
      "learning_rate": 5.300933062062131e-06,
      "loss": 0.0459,
      "step": 9412
    },
    {
      "epoch": 0.4860831396849987,
      "grad_norm": 3.5825507640838623,
      "learning_rate": 5.300115182746307e-06,
      "loss": 0.0421,
      "step": 9413
    },
    {
      "epoch": 0.48613477924089854,
      "grad_norm": 4.7846293449401855,
      "learning_rate": 5.299297295371164e-06,
      "loss": 0.0418,
      "step": 9414
    },
    {
      "epoch": 0.48618641879679836,
      "grad_norm": 3.163480758666992,
      "learning_rate": 5.2984793999586716e-06,
      "loss": 0.0562,
      "step": 9415
    },
    {
      "epoch": 0.4862380583526982,
      "grad_norm": 3.0771522521972656,
      "learning_rate": 5.297661496530789e-06,
      "loss": 0.0514,
      "step": 9416
    },
    {
      "epoch": 0.486289697908598,
      "grad_norm": 5.818752765655518,
      "learning_rate": 5.296843585109482e-06,
      "loss": 0.0656,
      "step": 9417
    },
    {
      "epoch": 0.4863413374644978,
      "grad_norm": 5.226281642913818,
      "learning_rate": 5.296025665716716e-06,
      "loss": 0.0613,
      "step": 9418
    },
    {
      "epoch": 0.48639297702039763,
      "grad_norm": 8.103858947753906,
      "learning_rate": 5.2952077383744526e-06,
      "loss": 0.0838,
      "step": 9419
    },
    {
      "epoch": 0.48644461657629745,
      "grad_norm": 6.331757068634033,
      "learning_rate": 5.29438980310466e-06,
      "loss": 0.0694,
      "step": 9420
    },
    {
      "epoch": 0.48649625613219727,
      "grad_norm": 2.9538559913635254,
      "learning_rate": 5.2935718599292994e-06,
      "loss": 0.0672,
      "step": 9421
    },
    {
      "epoch": 0.4865478956880971,
      "grad_norm": 4.211157321929932,
      "learning_rate": 5.2927539088703395e-06,
      "loss": 0.0469,
      "step": 9422
    },
    {
      "epoch": 0.4865995352439969,
      "grad_norm": 2.3708739280700684,
      "learning_rate": 5.291935949949743e-06,
      "loss": 0.0296,
      "step": 9423
    },
    {
      "epoch": 0.4866511747998967,
      "grad_norm": 2.4928762912750244,
      "learning_rate": 5.291117983189477e-06,
      "loss": 0.0542,
      "step": 9424
    },
    {
      "epoch": 0.48670281435579654,
      "grad_norm": 4.184813499450684,
      "learning_rate": 5.290300008611506e-06,
      "loss": 0.0603,
      "step": 9425
    },
    {
      "epoch": 0.48675445391169636,
      "grad_norm": 1.896555781364441,
      "learning_rate": 5.289482026237797e-06,
      "loss": 0.0216,
      "step": 9426
    },
    {
      "epoch": 0.4868060934675962,
      "grad_norm": 5.386655807495117,
      "learning_rate": 5.288664036090316e-06,
      "loss": 0.0931,
      "step": 9427
    },
    {
      "epoch": 0.486857733023496,
      "grad_norm": 3.7137832641601562,
      "learning_rate": 5.287846038191028e-06,
      "loss": 0.0455,
      "step": 9428
    },
    {
      "epoch": 0.4869093725793958,
      "grad_norm": 11.20224666595459,
      "learning_rate": 5.287028032561903e-06,
      "loss": 0.1503,
      "step": 9429
    },
    {
      "epoch": 0.48696101213529563,
      "grad_norm": 3.4227702617645264,
      "learning_rate": 5.286210019224904e-06,
      "loss": 0.0433,
      "step": 9430
    },
    {
      "epoch": 0.48701265169119545,
      "grad_norm": 11.430937767028809,
      "learning_rate": 5.285391998201999e-06,
      "loss": 0.1526,
      "step": 9431
    },
    {
      "epoch": 0.48706429124709527,
      "grad_norm": 1.9213981628417969,
      "learning_rate": 5.284573969515155e-06,
      "loss": 0.0255,
      "step": 9432
    },
    {
      "epoch": 0.4871159308029951,
      "grad_norm": 5.668910026550293,
      "learning_rate": 5.283755933186341e-06,
      "loss": 0.1066,
      "step": 9433
    },
    {
      "epoch": 0.4871675703588949,
      "grad_norm": 6.60839319229126,
      "learning_rate": 5.282937889237524e-06,
      "loss": 0.0695,
      "step": 9434
    },
    {
      "epoch": 0.4872192099147947,
      "grad_norm": 8.899527549743652,
      "learning_rate": 5.282119837690671e-06,
      "loss": 0.1355,
      "step": 9435
    },
    {
      "epoch": 0.48727084947069454,
      "grad_norm": 2.177746057510376,
      "learning_rate": 5.28130177856775e-06,
      "loss": 0.0261,
      "step": 9436
    },
    {
      "epoch": 0.48732248902659436,
      "grad_norm": 5.098538398742676,
      "learning_rate": 5.280483711890732e-06,
      "loss": 0.0482,
      "step": 9437
    },
    {
      "epoch": 0.4873741285824942,
      "grad_norm": 4.106692314147949,
      "learning_rate": 5.279665637681581e-06,
      "loss": 0.0375,
      "step": 9438
    },
    {
      "epoch": 0.487425768138394,
      "grad_norm": 12.905440330505371,
      "learning_rate": 5.278847555962268e-06,
      "loss": 0.1308,
      "step": 9439
    },
    {
      "epoch": 0.4874774076942938,
      "grad_norm": 3.088921070098877,
      "learning_rate": 5.278029466754762e-06,
      "loss": 0.0256,
      "step": 9440
    },
    {
      "epoch": 0.48752904725019364,
      "grad_norm": 8.403629302978516,
      "learning_rate": 5.277211370081031e-06,
      "loss": 0.1409,
      "step": 9441
    },
    {
      "epoch": 0.48758068680609346,
      "grad_norm": 2.669351100921631,
      "learning_rate": 5.2763932659630456e-06,
      "loss": 0.0196,
      "step": 9442
    },
    {
      "epoch": 0.4876323263619933,
      "grad_norm": 9.797435760498047,
      "learning_rate": 5.275575154422775e-06,
      "loss": 0.0979,
      "step": 9443
    },
    {
      "epoch": 0.4876839659178931,
      "grad_norm": 5.609067440032959,
      "learning_rate": 5.274757035482189e-06,
      "loss": 0.0387,
      "step": 9444
    },
    {
      "epoch": 0.4877356054737929,
      "grad_norm": 2.069563150405884,
      "learning_rate": 5.273938909163253e-06,
      "loss": 0.0477,
      "step": 9445
    },
    {
      "epoch": 0.48778724502969273,
      "grad_norm": 6.577366352081299,
      "learning_rate": 5.273120775487944e-06,
      "loss": 0.102,
      "step": 9446
    },
    {
      "epoch": 0.48783888458559255,
      "grad_norm": 6.524982452392578,
      "learning_rate": 5.272302634478228e-06,
      "loss": 0.0757,
      "step": 9447
    },
    {
      "epoch": 0.48789052414149237,
      "grad_norm": 3.29390025138855,
      "learning_rate": 5.271484486156078e-06,
      "loss": 0.0392,
      "step": 9448
    },
    {
      "epoch": 0.4879421636973922,
      "grad_norm": 4.92128324508667,
      "learning_rate": 5.27066633054346e-06,
      "loss": 0.0519,
      "step": 9449
    },
    {
      "epoch": 0.487993803253292,
      "grad_norm": 10.760160446166992,
      "learning_rate": 5.26984816766235e-06,
      "loss": 0.0783,
      "step": 9450
    },
    {
      "epoch": 0.4880454428091918,
      "grad_norm": 4.090414047241211,
      "learning_rate": 5.269029997534717e-06,
      "loss": 0.0872,
      "step": 9451
    },
    {
      "epoch": 0.48809708236509164,
      "grad_norm": 2.1465981006622314,
      "learning_rate": 5.2682118201825304e-06,
      "loss": 0.037,
      "step": 9452
    },
    {
      "epoch": 0.48814872192099146,
      "grad_norm": 3.562701940536499,
      "learning_rate": 5.267393635627765e-06,
      "loss": 0.0784,
      "step": 9453
    },
    {
      "epoch": 0.4882003614768913,
      "grad_norm": 6.85812520980835,
      "learning_rate": 5.266575443892389e-06,
      "loss": 0.129,
      "step": 9454
    },
    {
      "epoch": 0.4882520010327911,
      "grad_norm": 4.914527416229248,
      "learning_rate": 5.265757244998377e-06,
      "loss": 0.0444,
      "step": 9455
    },
    {
      "epoch": 0.4883036405886909,
      "grad_norm": 16.90673828125,
      "learning_rate": 5.2649390389677e-06,
      "loss": 0.0991,
      "step": 9456
    },
    {
      "epoch": 0.48835528014459073,
      "grad_norm": 17.275999069213867,
      "learning_rate": 5.26412082582233e-06,
      "loss": 0.1501,
      "step": 9457
    },
    {
      "epoch": 0.48840691970049055,
      "grad_norm": 9.460308074951172,
      "learning_rate": 5.2633026055842385e-06,
      "loss": 0.0923,
      "step": 9458
    },
    {
      "epoch": 0.48845855925639037,
      "grad_norm": 10.29520320892334,
      "learning_rate": 5.262484378275399e-06,
      "loss": 0.0853,
      "step": 9459
    },
    {
      "epoch": 0.48851019881229024,
      "grad_norm": 2.289821147918701,
      "learning_rate": 5.261666143917785e-06,
      "loss": 0.0541,
      "step": 9460
    },
    {
      "epoch": 0.48856183836819006,
      "grad_norm": 5.534786701202393,
      "learning_rate": 5.260847902533367e-06,
      "loss": 0.0816,
      "step": 9461
    },
    {
      "epoch": 0.4886134779240899,
      "grad_norm": 9.10323715209961,
      "learning_rate": 5.260029654144121e-06,
      "loss": 0.0799,
      "step": 9462
    },
    {
      "epoch": 0.4886651174799897,
      "grad_norm": 4.87069034576416,
      "learning_rate": 5.2592113987720185e-06,
      "loss": 0.0843,
      "step": 9463
    },
    {
      "epoch": 0.4887167570358895,
      "grad_norm": 7.112527370452881,
      "learning_rate": 5.258393136439034e-06,
      "loss": 0.0717,
      "step": 9464
    },
    {
      "epoch": 0.48876839659178933,
      "grad_norm": 4.846081256866455,
      "learning_rate": 5.25757486716714e-06,
      "loss": 0.0692,
      "step": 9465
    },
    {
      "epoch": 0.48882003614768915,
      "grad_norm": 6.559377193450928,
      "learning_rate": 5.25675659097831e-06,
      "loss": 0.0743,
      "step": 9466
    },
    {
      "epoch": 0.48887167570358897,
      "grad_norm": 3.3175740242004395,
      "learning_rate": 5.25593830789452e-06,
      "loss": 0.0703,
      "step": 9467
    },
    {
      "epoch": 0.4889233152594888,
      "grad_norm": 1.7948203086853027,
      "learning_rate": 5.255120017937743e-06,
      "loss": 0.0502,
      "step": 9468
    },
    {
      "epoch": 0.4889749548153886,
      "grad_norm": 1.4360841512680054,
      "learning_rate": 5.254301721129955e-06,
      "loss": 0.0296,
      "step": 9469
    },
    {
      "epoch": 0.4890265943712884,
      "grad_norm": 3.3137311935424805,
      "learning_rate": 5.253483417493129e-06,
      "loss": 0.0733,
      "step": 9470
    },
    {
      "epoch": 0.48907823392718824,
      "grad_norm": 4.123202800750732,
      "learning_rate": 5.25266510704924e-06,
      "loss": 0.0943,
      "step": 9471
    },
    {
      "epoch": 0.48912987348308806,
      "grad_norm": 2.389296531677246,
      "learning_rate": 5.251846789820262e-06,
      "loss": 0.054,
      "step": 9472
    },
    {
      "epoch": 0.4891815130389879,
      "grad_norm": 2.147125005722046,
      "learning_rate": 5.251028465828172e-06,
      "loss": 0.0649,
      "step": 9473
    },
    {
      "epoch": 0.4892331525948877,
      "grad_norm": 3.027660608291626,
      "learning_rate": 5.250210135094944e-06,
      "loss": 0.0458,
      "step": 9474
    },
    {
      "epoch": 0.4892847921507875,
      "grad_norm": 4.3661603927612305,
      "learning_rate": 5.249391797642555e-06,
      "loss": 0.0439,
      "step": 9475
    },
    {
      "epoch": 0.48933643170668734,
      "grad_norm": 9.809958457946777,
      "learning_rate": 5.2485734534929784e-06,
      "loss": 0.103,
      "step": 9476
    },
    {
      "epoch": 0.48938807126258715,
      "grad_norm": 2.7984228134155273,
      "learning_rate": 5.247755102668194e-06,
      "loss": 0.0393,
      "step": 9477
    },
    {
      "epoch": 0.489439710818487,
      "grad_norm": 3.947716236114502,
      "learning_rate": 5.246936745190172e-06,
      "loss": 0.0615,
      "step": 9478
    },
    {
      "epoch": 0.4894913503743868,
      "grad_norm": 6.103151798248291,
      "learning_rate": 5.246118381080894e-06,
      "loss": 0.0908,
      "step": 9479
    },
    {
      "epoch": 0.4895429899302866,
      "grad_norm": 5.452713489532471,
      "learning_rate": 5.245300010362333e-06,
      "loss": 0.0921,
      "step": 9480
    },
    {
      "epoch": 0.48959462948618643,
      "grad_norm": 7.447551727294922,
      "learning_rate": 5.244481633056468e-06,
      "loss": 0.0601,
      "step": 9481
    },
    {
      "epoch": 0.48964626904208625,
      "grad_norm": 3.849071502685547,
      "learning_rate": 5.243663249185276e-06,
      "loss": 0.0309,
      "step": 9482
    },
    {
      "epoch": 0.48969790859798606,
      "grad_norm": 9.08964729309082,
      "learning_rate": 5.24284485877073e-06,
      "loss": 0.1128,
      "step": 9483
    },
    {
      "epoch": 0.4897495481538859,
      "grad_norm": 2.918447971343994,
      "learning_rate": 5.242026461834813e-06,
      "loss": 0.0564,
      "step": 9484
    },
    {
      "epoch": 0.4898011877097857,
      "grad_norm": 2.902137517929077,
      "learning_rate": 5.241208058399495e-06,
      "loss": 0.0683,
      "step": 9485
    },
    {
      "epoch": 0.4898528272656855,
      "grad_norm": 4.247832775115967,
      "learning_rate": 5.24038964848676e-06,
      "loss": 0.0549,
      "step": 9486
    },
    {
      "epoch": 0.48990446682158534,
      "grad_norm": 4.107526779174805,
      "learning_rate": 5.239571232118582e-06,
      "loss": 0.0528,
      "step": 9487
    },
    {
      "epoch": 0.48995610637748516,
      "grad_norm": 4.465251922607422,
      "learning_rate": 5.238752809316942e-06,
      "loss": 0.0485,
      "step": 9488
    },
    {
      "epoch": 0.490007745933385,
      "grad_norm": 20.5277099609375,
      "learning_rate": 5.237934380103814e-06,
      "loss": 0.1646,
      "step": 9489
    },
    {
      "epoch": 0.4900593854892848,
      "grad_norm": 3.3098859786987305,
      "learning_rate": 5.237115944501179e-06,
      "loss": 0.0833,
      "step": 9490
    },
    {
      "epoch": 0.4901110250451846,
      "grad_norm": 6.1838555335998535,
      "learning_rate": 5.236297502531014e-06,
      "loss": 0.0929,
      "step": 9491
    },
    {
      "epoch": 0.49016266460108443,
      "grad_norm": 7.148126125335693,
      "learning_rate": 5.235479054215298e-06,
      "loss": 0.0491,
      "step": 9492
    },
    {
      "epoch": 0.49021430415698425,
      "grad_norm": 1.4512810707092285,
      "learning_rate": 5.23466059957601e-06,
      "loss": 0.0186,
      "step": 9493
    },
    {
      "epoch": 0.49026594371288407,
      "grad_norm": 2.907400131225586,
      "learning_rate": 5.233842138635128e-06,
      "loss": 0.0282,
      "step": 9494
    },
    {
      "epoch": 0.4903175832687839,
      "grad_norm": 4.968508720397949,
      "learning_rate": 5.2330236714146325e-06,
      "loss": 0.1007,
      "step": 9495
    },
    {
      "epoch": 0.4903692228246837,
      "grad_norm": 4.053833484649658,
      "learning_rate": 5.2322051979365006e-06,
      "loss": 0.0538,
      "step": 9496
    },
    {
      "epoch": 0.4904208623805835,
      "grad_norm": 3.191922426223755,
      "learning_rate": 5.231386718222713e-06,
      "loss": 0.0448,
      "step": 9497
    },
    {
      "epoch": 0.49047250193648334,
      "grad_norm": 9.002484321594238,
      "learning_rate": 5.230568232295249e-06,
      "loss": 0.0804,
      "step": 9498
    },
    {
      "epoch": 0.49052414149238316,
      "grad_norm": 3.35044002532959,
      "learning_rate": 5.229749740176088e-06,
      "loss": 0.0571,
      "step": 9499
    },
    {
      "epoch": 0.490575781048283,
      "grad_norm": 2.143328905105591,
      "learning_rate": 5.22893124188721e-06,
      "loss": 0.0263,
      "step": 9500
    },
    {
      "epoch": 0.4906274206041828,
      "grad_norm": 13.552701950073242,
      "learning_rate": 5.228112737450594e-06,
      "loss": 0.1458,
      "step": 9501
    },
    {
      "epoch": 0.4906790601600826,
      "grad_norm": 4.321488380432129,
      "learning_rate": 5.227294226888223e-06,
      "loss": 0.085,
      "step": 9502
    },
    {
      "epoch": 0.49073069971598243,
      "grad_norm": 2.2445621490478516,
      "learning_rate": 5.226475710222077e-06,
      "loss": 0.0458,
      "step": 9503
    },
    {
      "epoch": 0.49078233927188225,
      "grad_norm": 8.950820922851562,
      "learning_rate": 5.225657187474132e-06,
      "loss": 0.15,
      "step": 9504
    },
    {
      "epoch": 0.49083397882778207,
      "grad_norm": 2.5464835166931152,
      "learning_rate": 5.224838658666373e-06,
      "loss": 0.0448,
      "step": 9505
    },
    {
      "epoch": 0.4908856183836819,
      "grad_norm": 3.2320072650909424,
      "learning_rate": 5.22402012382078e-06,
      "loss": 0.048,
      "step": 9506
    },
    {
      "epoch": 0.4909372579395817,
      "grad_norm": 5.016849040985107,
      "learning_rate": 5.2232015829593345e-06,
      "loss": 0.0535,
      "step": 9507
    },
    {
      "epoch": 0.4909888974954815,
      "grad_norm": 8.943963050842285,
      "learning_rate": 5.222383036104015e-06,
      "loss": 0.0568,
      "step": 9508
    },
    {
      "epoch": 0.49104053705138134,
      "grad_norm": 11.602883338928223,
      "learning_rate": 5.221564483276805e-06,
      "loss": 0.1375,
      "step": 9509
    },
    {
      "epoch": 0.49109217660728116,
      "grad_norm": 3.832106113433838,
      "learning_rate": 5.220745924499686e-06,
      "loss": 0.0791,
      "step": 9510
    },
    {
      "epoch": 0.491143816163181,
      "grad_norm": 2.3807005882263184,
      "learning_rate": 5.2199273597946385e-06,
      "loss": 0.0506,
      "step": 9511
    },
    {
      "epoch": 0.4911954557190808,
      "grad_norm": 4.296934604644775,
      "learning_rate": 5.219108789183647e-06,
      "loss": 0.0382,
      "step": 9512
    },
    {
      "epoch": 0.4912470952749806,
      "grad_norm": 6.179748058319092,
      "learning_rate": 5.218290212688688e-06,
      "loss": 0.076,
      "step": 9513
    },
    {
      "epoch": 0.49129873483088043,
      "grad_norm": 7.107131481170654,
      "learning_rate": 5.217471630331751e-06,
      "loss": 0.0999,
      "step": 9514
    },
    {
      "epoch": 0.49135037438678025,
      "grad_norm": 1.9595441818237305,
      "learning_rate": 5.216653042134811e-06,
      "loss": 0.0387,
      "step": 9515
    },
    {
      "epoch": 0.49140201394268007,
      "grad_norm": 3.0928940773010254,
      "learning_rate": 5.215834448119855e-06,
      "loss": 0.0587,
      "step": 9516
    },
    {
      "epoch": 0.4914536534985799,
      "grad_norm": 2.9482107162475586,
      "learning_rate": 5.215015848308865e-06,
      "loss": 0.0391,
      "step": 9517
    },
    {
      "epoch": 0.4915052930544797,
      "grad_norm": 1.6973038911819458,
      "learning_rate": 5.214197242723823e-06,
      "loss": 0.0262,
      "step": 9518
    },
    {
      "epoch": 0.4915569326103795,
      "grad_norm": 5.377299785614014,
      "learning_rate": 5.213378631386713e-06,
      "loss": 0.0387,
      "step": 9519
    },
    {
      "epoch": 0.49160857216627935,
      "grad_norm": 13.144336700439453,
      "learning_rate": 5.212560014319514e-06,
      "loss": 0.1115,
      "step": 9520
    },
    {
      "epoch": 0.49166021172217916,
      "grad_norm": 2.1392436027526855,
      "learning_rate": 5.211741391544215e-06,
      "loss": 0.0424,
      "step": 9521
    },
    {
      "epoch": 0.491711851278079,
      "grad_norm": 2.1084144115448,
      "learning_rate": 5.210922763082797e-06,
      "loss": 0.0297,
      "step": 9522
    },
    {
      "epoch": 0.4917634908339788,
      "grad_norm": 0.3484754264354706,
      "learning_rate": 5.2101041289572404e-06,
      "loss": 0.0019,
      "step": 9523
    },
    {
      "epoch": 0.4918151303898787,
      "grad_norm": 4.506324768066406,
      "learning_rate": 5.2092854891895325e-06,
      "loss": 0.0469,
      "step": 9524
    },
    {
      "epoch": 0.4918667699457785,
      "grad_norm": 2.153334617614746,
      "learning_rate": 5.208466843801657e-06,
      "loss": 0.0323,
      "step": 9525
    },
    {
      "epoch": 0.4919184095016783,
      "grad_norm": 2.6429173946380615,
      "learning_rate": 5.207648192815599e-06,
      "loss": 0.0395,
      "step": 9526
    },
    {
      "epoch": 0.49197004905757813,
      "grad_norm": 1.7297308444976807,
      "learning_rate": 5.206829536253336e-06,
      "loss": 0.0226,
      "step": 9527
    },
    {
      "epoch": 0.49202168861347795,
      "grad_norm": 3.207791566848755,
      "learning_rate": 5.206010874136861e-06,
      "loss": 0.02,
      "step": 9528
    },
    {
      "epoch": 0.49207332816937777,
      "grad_norm": 25.135786056518555,
      "learning_rate": 5.205192206488152e-06,
      "loss": 0.0682,
      "step": 9529
    },
    {
      "epoch": 0.4921249677252776,
      "grad_norm": 3.6192786693573,
      "learning_rate": 5.204373533329198e-06,
      "loss": 0.0369,
      "step": 9530
    },
    {
      "epoch": 0.4921766072811774,
      "grad_norm": 4.010157585144043,
      "learning_rate": 5.203554854681979e-06,
      "loss": 0.0739,
      "step": 9531
    },
    {
      "epoch": 0.4922282468370772,
      "grad_norm": 11.120738983154297,
      "learning_rate": 5.202736170568484e-06,
      "loss": 0.0797,
      "step": 9532
    },
    {
      "epoch": 0.49227988639297704,
      "grad_norm": 2.550393581390381,
      "learning_rate": 5.201917481010697e-06,
      "loss": 0.0396,
      "step": 9533
    },
    {
      "epoch": 0.49233152594887686,
      "grad_norm": 5.880678176879883,
      "learning_rate": 5.201098786030601e-06,
      "loss": 0.1111,
      "step": 9534
    },
    {
      "epoch": 0.4923831655047767,
      "grad_norm": 20.37388801574707,
      "learning_rate": 5.200280085650185e-06,
      "loss": 0.0813,
      "step": 9535
    },
    {
      "epoch": 0.4924348050606765,
      "grad_norm": 5.844775676727295,
      "learning_rate": 5.1994613798914324e-06,
      "loss": 0.0729,
      "step": 9536
    },
    {
      "epoch": 0.4924864446165763,
      "grad_norm": 4.212052345275879,
      "learning_rate": 5.198642668776328e-06,
      "loss": 0.0522,
      "step": 9537
    },
    {
      "epoch": 0.49253808417247613,
      "grad_norm": 17.721647262573242,
      "learning_rate": 5.197823952326858e-06,
      "loss": 0.1391,
      "step": 9538
    },
    {
      "epoch": 0.49258972372837595,
      "grad_norm": 18.46426010131836,
      "learning_rate": 5.197005230565008e-06,
      "loss": 0.1687,
      "step": 9539
    },
    {
      "epoch": 0.49264136328427577,
      "grad_norm": 4.462006568908691,
      "learning_rate": 5.196186503512766e-06,
      "loss": 0.0497,
      "step": 9540
    },
    {
      "epoch": 0.4926930028401756,
      "grad_norm": 4.632849216461182,
      "learning_rate": 5.195367771192116e-06,
      "loss": 0.0568,
      "step": 9541
    },
    {
      "epoch": 0.4927446423960754,
      "grad_norm": 4.3341965675354,
      "learning_rate": 5.1945490336250445e-06,
      "loss": 0.0643,
      "step": 9542
    },
    {
      "epoch": 0.4927962819519752,
      "grad_norm": 2.145374059677124,
      "learning_rate": 5.19373029083354e-06,
      "loss": 0.0371,
      "step": 9543
    },
    {
      "epoch": 0.49284792150787504,
      "grad_norm": 3.2671244144439697,
      "learning_rate": 5.192911542839587e-06,
      "loss": 0.0794,
      "step": 9544
    },
    {
      "epoch": 0.49289956106377486,
      "grad_norm": 2.8963141441345215,
      "learning_rate": 5.192092789665173e-06,
      "loss": 0.0425,
      "step": 9545
    },
    {
      "epoch": 0.4929512006196747,
      "grad_norm": 15.137103080749512,
      "learning_rate": 5.191274031332284e-06,
      "loss": 0.2653,
      "step": 9546
    },
    {
      "epoch": 0.4930028401755745,
      "grad_norm": 6.5461931228637695,
      "learning_rate": 5.190455267862908e-06,
      "loss": 0.0975,
      "step": 9547
    },
    {
      "epoch": 0.4930544797314743,
      "grad_norm": 6.439833164215088,
      "learning_rate": 5.189636499279033e-06,
      "loss": 0.0787,
      "step": 9548
    },
    {
      "epoch": 0.49310611928737413,
      "grad_norm": 4.81509256362915,
      "learning_rate": 5.188817725602643e-06,
      "loss": 0.0471,
      "step": 9549
    },
    {
      "epoch": 0.49315775884327395,
      "grad_norm": 8.178382873535156,
      "learning_rate": 5.187998946855728e-06,
      "loss": 0.0376,
      "step": 9550
    },
    {
      "epoch": 0.49320939839917377,
      "grad_norm": 4.148234844207764,
      "learning_rate": 5.187180163060276e-06,
      "loss": 0.0513,
      "step": 9551
    },
    {
      "epoch": 0.4932610379550736,
      "grad_norm": 2.871347188949585,
      "learning_rate": 5.186361374238274e-06,
      "loss": 0.0748,
      "step": 9552
    },
    {
      "epoch": 0.4933126775109734,
      "grad_norm": 2.826986789703369,
      "learning_rate": 5.185542580411707e-06,
      "loss": 0.0659,
      "step": 9553
    },
    {
      "epoch": 0.4933643170668732,
      "grad_norm": 12.306708335876465,
      "learning_rate": 5.1847237816025685e-06,
      "loss": 0.0452,
      "step": 9554
    },
    {
      "epoch": 0.49341595662277304,
      "grad_norm": 3.0156097412109375,
      "learning_rate": 5.183904977832842e-06,
      "loss": 0.0395,
      "step": 9555
    },
    {
      "epoch": 0.49346759617867286,
      "grad_norm": 8.39310359954834,
      "learning_rate": 5.183086169124517e-06,
      "loss": 0.0808,
      "step": 9556
    },
    {
      "epoch": 0.4935192357345727,
      "grad_norm": 3.1227245330810547,
      "learning_rate": 5.182267355499583e-06,
      "loss": 0.0434,
      "step": 9557
    },
    {
      "epoch": 0.4935708752904725,
      "grad_norm": 3.5298123359680176,
      "learning_rate": 5.181448536980027e-06,
      "loss": 0.0424,
      "step": 9558
    },
    {
      "epoch": 0.4936225148463723,
      "grad_norm": 9.330513954162598,
      "learning_rate": 5.180629713587839e-06,
      "loss": 0.0726,
      "step": 9559
    },
    {
      "epoch": 0.49367415440227214,
      "grad_norm": 2.233762741088867,
      "learning_rate": 5.179810885345005e-06,
      "loss": 0.0251,
      "step": 9560
    },
    {
      "epoch": 0.49372579395817195,
      "grad_norm": 2.7727737426757812,
      "learning_rate": 5.178992052273519e-06,
      "loss": 0.0666,
      "step": 9561
    },
    {
      "epoch": 0.4937774335140718,
      "grad_norm": 4.656167984008789,
      "learning_rate": 5.178173214395365e-06,
      "loss": 0.0566,
      "step": 9562
    },
    {
      "epoch": 0.4938290730699716,
      "grad_norm": 4.89655065536499,
      "learning_rate": 5.177354371732535e-06,
      "loss": 0.0772,
      "step": 9563
    },
    {
      "epoch": 0.4938807126258714,
      "grad_norm": 2.4307894706726074,
      "learning_rate": 5.176535524307016e-06,
      "loss": 0.0496,
      "step": 9564
    },
    {
      "epoch": 0.49393235218177123,
      "grad_norm": 2.2604031562805176,
      "learning_rate": 5.1757166721407995e-06,
      "loss": 0.0376,
      "step": 9565
    },
    {
      "epoch": 0.49398399173767105,
      "grad_norm": 1.877524495124817,
      "learning_rate": 5.174897815255873e-06,
      "loss": 0.0222,
      "step": 9566
    },
    {
      "epoch": 0.49403563129357086,
      "grad_norm": 6.671752452850342,
      "learning_rate": 5.174078953674228e-06,
      "loss": 0.0321,
      "step": 9567
    },
    {
      "epoch": 0.4940872708494707,
      "grad_norm": 6.121087074279785,
      "learning_rate": 5.173260087417855e-06,
      "loss": 0.1038,
      "step": 9568
    },
    {
      "epoch": 0.4941389104053705,
      "grad_norm": 2.342700719833374,
      "learning_rate": 5.17244121650874e-06,
      "loss": 0.0448,
      "step": 9569
    },
    {
      "epoch": 0.4941905499612703,
      "grad_norm": 6.392960548400879,
      "learning_rate": 5.171622340968877e-06,
      "loss": 0.0245,
      "step": 9570
    },
    {
      "epoch": 0.49424218951717014,
      "grad_norm": 0.993524432182312,
      "learning_rate": 5.170803460820255e-06,
      "loss": 0.011,
      "step": 9571
    },
    {
      "epoch": 0.49429382907306996,
      "grad_norm": 2.7771830558776855,
      "learning_rate": 5.169984576084863e-06,
      "loss": 0.0244,
      "step": 9572
    },
    {
      "epoch": 0.4943454686289698,
      "grad_norm": 3.9338531494140625,
      "learning_rate": 5.169165686784693e-06,
      "loss": 0.0701,
      "step": 9573
    },
    {
      "epoch": 0.4943971081848696,
      "grad_norm": 2.8797624111175537,
      "learning_rate": 5.168346792941734e-06,
      "loss": 0.0414,
      "step": 9574
    },
    {
      "epoch": 0.4944487477407694,
      "grad_norm": 8.245657920837402,
      "learning_rate": 5.167527894577977e-06,
      "loss": 0.0629,
      "step": 9575
    },
    {
      "epoch": 0.49450038729666923,
      "grad_norm": 3.7760934829711914,
      "learning_rate": 5.166708991715414e-06,
      "loss": 0.0569,
      "step": 9576
    },
    {
      "epoch": 0.49455202685256905,
      "grad_norm": 3.4219861030578613,
      "learning_rate": 5.165890084376035e-06,
      "loss": 0.0362,
      "step": 9577
    },
    {
      "epoch": 0.49460366640846887,
      "grad_norm": 5.642984390258789,
      "learning_rate": 5.165071172581833e-06,
      "loss": 0.0882,
      "step": 9578
    },
    {
      "epoch": 0.4946553059643687,
      "grad_norm": 3.0953855514526367,
      "learning_rate": 5.164252256354796e-06,
      "loss": 0.0608,
      "step": 9579
    },
    {
      "epoch": 0.4947069455202685,
      "grad_norm": 10.540903091430664,
      "learning_rate": 5.163433335716915e-06,
      "loss": 0.0506,
      "step": 9580
    },
    {
      "epoch": 0.4947585850761683,
      "grad_norm": 12.870244026184082,
      "learning_rate": 5.162614410690184e-06,
      "loss": 0.2611,
      "step": 9581
    },
    {
      "epoch": 0.49481022463206814,
      "grad_norm": 22.295085906982422,
      "learning_rate": 5.161795481296593e-06,
      "loss": 0.0654,
      "step": 9582
    },
    {
      "epoch": 0.49486186418796796,
      "grad_norm": 14.604349136352539,
      "learning_rate": 5.1609765475581345e-06,
      "loss": 0.1086,
      "step": 9583
    },
    {
      "epoch": 0.4949135037438678,
      "grad_norm": 7.813544273376465,
      "learning_rate": 5.160157609496798e-06,
      "loss": 0.0547,
      "step": 9584
    },
    {
      "epoch": 0.4949651432997676,
      "grad_norm": 3.762922525405884,
      "learning_rate": 5.159338667134579e-06,
      "loss": 0.0448,
      "step": 9585
    },
    {
      "epoch": 0.4950167828556674,
      "grad_norm": 5.8878984451293945,
      "learning_rate": 5.1585197204934655e-06,
      "loss": 0.0796,
      "step": 9586
    },
    {
      "epoch": 0.4950684224115673,
      "grad_norm": 5.12908935546875,
      "learning_rate": 5.157700769595453e-06,
      "loss": 0.0579,
      "step": 9587
    },
    {
      "epoch": 0.4951200619674671,
      "grad_norm": 7.0842413902282715,
      "learning_rate": 5.156881814462531e-06,
      "loss": 0.1171,
      "step": 9588
    },
    {
      "epoch": 0.4951717015233669,
      "grad_norm": 3.0280370712280273,
      "learning_rate": 5.156062855116693e-06,
      "loss": 0.0453,
      "step": 9589
    },
    {
      "epoch": 0.49522334107926674,
      "grad_norm": 13.79735279083252,
      "learning_rate": 5.1552438915799316e-06,
      "loss": 0.0681,
      "step": 9590
    },
    {
      "epoch": 0.49527498063516656,
      "grad_norm": 7.409729480743408,
      "learning_rate": 5.15442492387424e-06,
      "loss": 0.0812,
      "step": 9591
    },
    {
      "epoch": 0.4953266201910664,
      "grad_norm": 4.214283466339111,
      "learning_rate": 5.153605952021608e-06,
      "loss": 0.058,
      "step": 9592
    },
    {
      "epoch": 0.4953782597469662,
      "grad_norm": 8.610305786132812,
      "learning_rate": 5.152786976044031e-06,
      "loss": 0.0881,
      "step": 9593
    },
    {
      "epoch": 0.495429899302866,
      "grad_norm": 3.62813401222229,
      "learning_rate": 5.151967995963502e-06,
      "loss": 0.0345,
      "step": 9594
    },
    {
      "epoch": 0.49548153885876584,
      "grad_norm": 1.6322511434555054,
      "learning_rate": 5.15114901180201e-06,
      "loss": 0.0307,
      "step": 9595
    },
    {
      "epoch": 0.49553317841466565,
      "grad_norm": 6.245715618133545,
      "learning_rate": 5.150330023581553e-06,
      "loss": 0.0753,
      "step": 9596
    },
    {
      "epoch": 0.49558481797056547,
      "grad_norm": 9.645544052124023,
      "learning_rate": 5.1495110313241225e-06,
      "loss": 0.0817,
      "step": 9597
    },
    {
      "epoch": 0.4956364575264653,
      "grad_norm": 3.5286848545074463,
      "learning_rate": 5.14869203505171e-06,
      "loss": 0.0447,
      "step": 9598
    },
    {
      "epoch": 0.4956880970823651,
      "grad_norm": 2.023192882537842,
      "learning_rate": 5.147873034786311e-06,
      "loss": 0.0296,
      "step": 9599
    },
    {
      "epoch": 0.4957397366382649,
      "grad_norm": 5.105980396270752,
      "learning_rate": 5.147054030549918e-06,
      "loss": 0.0438,
      "step": 9600
    },
    {
      "epoch": 0.49579137619416475,
      "grad_norm": 1.8172681331634521,
      "learning_rate": 5.1462350223645276e-06,
      "loss": 0.0165,
      "step": 9601
    },
    {
      "epoch": 0.49584301575006456,
      "grad_norm": 8.671263694763184,
      "learning_rate": 5.145416010252126e-06,
      "loss": 0.0696,
      "step": 9602
    },
    {
      "epoch": 0.4958946553059644,
      "grad_norm": 8.511857032775879,
      "learning_rate": 5.144596994234716e-06,
      "loss": 0.0537,
      "step": 9603
    },
    {
      "epoch": 0.4959462948618642,
      "grad_norm": 4.6194987297058105,
      "learning_rate": 5.1437779743342855e-06,
      "loss": 0.0635,
      "step": 9604
    },
    {
      "epoch": 0.495997934417764,
      "grad_norm": 4.125694751739502,
      "learning_rate": 5.142958950572831e-06,
      "loss": 0.0997,
      "step": 9605
    },
    {
      "epoch": 0.49604957397366384,
      "grad_norm": 5.13722562789917,
      "learning_rate": 5.142139922972346e-06,
      "loss": 0.0671,
      "step": 9606
    },
    {
      "epoch": 0.49610121352956366,
      "grad_norm": 2.503434181213379,
      "learning_rate": 5.141320891554824e-06,
      "loss": 0.0278,
      "step": 9607
    },
    {
      "epoch": 0.4961528530854635,
      "grad_norm": 4.816253662109375,
      "learning_rate": 5.140501856342261e-06,
      "loss": 0.0565,
      "step": 9608
    },
    {
      "epoch": 0.4962044926413633,
      "grad_norm": 6.484734535217285,
      "learning_rate": 5.139682817356648e-06,
      "loss": 0.0998,
      "step": 9609
    },
    {
      "epoch": 0.4962561321972631,
      "grad_norm": 3.339468002319336,
      "learning_rate": 5.138863774619984e-06,
      "loss": 0.0564,
      "step": 9610
    },
    {
      "epoch": 0.49630777175316293,
      "grad_norm": 4.730429172515869,
      "learning_rate": 5.1380447281542636e-06,
      "loss": 0.0946,
      "step": 9611
    },
    {
      "epoch": 0.49635941130906275,
      "grad_norm": 4.7337141036987305,
      "learning_rate": 5.137225677981478e-06,
      "loss": 0.0711,
      "step": 9612
    },
    {
      "epoch": 0.49641105086496257,
      "grad_norm": 1.4177056550979614,
      "learning_rate": 5.136406624123622e-06,
      "loss": 0.0193,
      "step": 9613
    },
    {
      "epoch": 0.4964626904208624,
      "grad_norm": 3.047802448272705,
      "learning_rate": 5.135587566602694e-06,
      "loss": 0.0376,
      "step": 9614
    },
    {
      "epoch": 0.4965143299767622,
      "grad_norm": 4.112879753112793,
      "learning_rate": 5.134768505440686e-06,
      "loss": 0.0447,
      "step": 9615
    },
    {
      "epoch": 0.496565969532662,
      "grad_norm": 4.437479496002197,
      "learning_rate": 5.1339494406595954e-06,
      "loss": 0.0539,
      "step": 9616
    },
    {
      "epoch": 0.49661760908856184,
      "grad_norm": 2.7817766666412354,
      "learning_rate": 5.133130372281416e-06,
      "loss": 0.0316,
      "step": 9617
    },
    {
      "epoch": 0.49666924864446166,
      "grad_norm": 6.460988521575928,
      "learning_rate": 5.132311300328145e-06,
      "loss": 0.0386,
      "step": 9618
    },
    {
      "epoch": 0.4967208882003615,
      "grad_norm": 5.628809452056885,
      "learning_rate": 5.131492224821773e-06,
      "loss": 0.088,
      "step": 9619
    },
    {
      "epoch": 0.4967725277562613,
      "grad_norm": 7.643799304962158,
      "learning_rate": 5.130673145784302e-06,
      "loss": 0.048,
      "step": 9620
    },
    {
      "epoch": 0.4968241673121611,
      "grad_norm": 5.93029260635376,
      "learning_rate": 5.129854063237721e-06,
      "loss": 0.0765,
      "step": 9621
    },
    {
      "epoch": 0.49687580686806093,
      "grad_norm": 4.100226879119873,
      "learning_rate": 5.129034977204033e-06,
      "loss": 0.1019,
      "step": 9622
    },
    {
      "epoch": 0.49692744642396075,
      "grad_norm": 5.385263919830322,
      "learning_rate": 5.128215887705228e-06,
      "loss": 0.0513,
      "step": 9623
    },
    {
      "epoch": 0.49697908597986057,
      "grad_norm": 4.258175849914551,
      "learning_rate": 5.127396794763304e-06,
      "loss": 0.0632,
      "step": 9624
    },
    {
      "epoch": 0.4970307255357604,
      "grad_norm": 3.2834391593933105,
      "learning_rate": 5.126577698400258e-06,
      "loss": 0.0475,
      "step": 9625
    },
    {
      "epoch": 0.4970823650916602,
      "grad_norm": 10.463172912597656,
      "learning_rate": 5.125758598638082e-06,
      "loss": 0.0551,
      "step": 9626
    },
    {
      "epoch": 0.49713400464756,
      "grad_norm": 5.806438446044922,
      "learning_rate": 5.124939495498778e-06,
      "loss": 0.1291,
      "step": 9627
    },
    {
      "epoch": 0.49718564420345984,
      "grad_norm": 6.860909938812256,
      "learning_rate": 5.124120389004337e-06,
      "loss": 0.0916,
      "step": 9628
    },
    {
      "epoch": 0.49723728375935966,
      "grad_norm": 2.4048619270324707,
      "learning_rate": 5.12330127917676e-06,
      "loss": 0.0464,
      "step": 9629
    },
    {
      "epoch": 0.4972889233152595,
      "grad_norm": 4.054447174072266,
      "learning_rate": 5.12248216603804e-06,
      "loss": 0.0555,
      "step": 9630
    },
    {
      "epoch": 0.4973405628711593,
      "grad_norm": 5.215097427368164,
      "learning_rate": 5.1216630496101734e-06,
      "loss": 0.0472,
      "step": 9631
    },
    {
      "epoch": 0.4973922024270591,
      "grad_norm": 5.482669353485107,
      "learning_rate": 5.12084392991516e-06,
      "loss": 0.1,
      "step": 9632
    },
    {
      "epoch": 0.49744384198295893,
      "grad_norm": 2.014514684677124,
      "learning_rate": 5.120024806974994e-06,
      "loss": 0.029,
      "step": 9633
    },
    {
      "epoch": 0.49749548153885875,
      "grad_norm": 3.3870604038238525,
      "learning_rate": 5.119205680811673e-06,
      "loss": 0.0528,
      "step": 9634
    },
    {
      "epoch": 0.49754712109475857,
      "grad_norm": 2.5847809314727783,
      "learning_rate": 5.118386551447192e-06,
      "loss": 0.0557,
      "step": 9635
    },
    {
      "epoch": 0.4975987606506584,
      "grad_norm": 2.007676124572754,
      "learning_rate": 5.117567418903552e-06,
      "loss": 0.0173,
      "step": 9636
    },
    {
      "epoch": 0.4976504002065582,
      "grad_norm": 8.377432823181152,
      "learning_rate": 5.116748283202746e-06,
      "loss": 0.1575,
      "step": 9637
    },
    {
      "epoch": 0.497702039762458,
      "grad_norm": 16.526155471801758,
      "learning_rate": 5.115929144366772e-06,
      "loss": 0.1142,
      "step": 9638
    },
    {
      "epoch": 0.49775367931835784,
      "grad_norm": 1.4338945150375366,
      "learning_rate": 5.11511000241763e-06,
      "loss": 0.026,
      "step": 9639
    },
    {
      "epoch": 0.49780531887425766,
      "grad_norm": 6.500817775726318,
      "learning_rate": 5.114290857377313e-06,
      "loss": 0.0473,
      "step": 9640
    },
    {
      "epoch": 0.4978569584301575,
      "grad_norm": 5.114504337310791,
      "learning_rate": 5.1134717092678225e-06,
      "loss": 0.0467,
      "step": 9641
    },
    {
      "epoch": 0.4979085979860573,
      "grad_norm": 1.2390002012252808,
      "learning_rate": 5.112652558111153e-06,
      "loss": 0.0128,
      "step": 9642
    },
    {
      "epoch": 0.4979602375419571,
      "grad_norm": 2.190232992172241,
      "learning_rate": 5.111833403929304e-06,
      "loss": 0.0183,
      "step": 9643
    },
    {
      "epoch": 0.49801187709785694,
      "grad_norm": 2.5450122356414795,
      "learning_rate": 5.1110142467442735e-06,
      "loss": 0.0372,
      "step": 9644
    },
    {
      "epoch": 0.49806351665375675,
      "grad_norm": 9.236104965209961,
      "learning_rate": 5.110195086578056e-06,
      "loss": 0.0425,
      "step": 9645
    },
    {
      "epoch": 0.4981151562096566,
      "grad_norm": 2.8659870624542236,
      "learning_rate": 5.109375923452652e-06,
      "loss": 0.0162,
      "step": 9646
    },
    {
      "epoch": 0.4981667957655564,
      "grad_norm": 9.758719444274902,
      "learning_rate": 5.108556757390059e-06,
      "loss": 0.1496,
      "step": 9647
    },
    {
      "epoch": 0.4982184353214562,
      "grad_norm": 2.9784727096557617,
      "learning_rate": 5.107737588412274e-06,
      "loss": 0.0575,
      "step": 9648
    },
    {
      "epoch": 0.49827007487735603,
      "grad_norm": 6.636659145355225,
      "learning_rate": 5.106918416541295e-06,
      "loss": 0.0856,
      "step": 9649
    },
    {
      "epoch": 0.49832171443325585,
      "grad_norm": 5.269540309906006,
      "learning_rate": 5.106099241799122e-06,
      "loss": 0.0558,
      "step": 9650
    },
    {
      "epoch": 0.4983733539891557,
      "grad_norm": 15.309910774230957,
      "learning_rate": 5.1052800642077515e-06,
      "loss": 0.1436,
      "step": 9651
    },
    {
      "epoch": 0.49842499354505554,
      "grad_norm": 5.3090386390686035,
      "learning_rate": 5.104460883789181e-06,
      "loss": 0.0791,
      "step": 9652
    },
    {
      "epoch": 0.49847663310095536,
      "grad_norm": 3.5850348472595215,
      "learning_rate": 5.103641700565413e-06,
      "loss": 0.0581,
      "step": 9653
    },
    {
      "epoch": 0.4985282726568552,
      "grad_norm": 13.441446304321289,
      "learning_rate": 5.1028225145584385e-06,
      "loss": 0.0759,
      "step": 9654
    },
    {
      "epoch": 0.498579912212755,
      "grad_norm": 5.218605995178223,
      "learning_rate": 5.1020033257902645e-06,
      "loss": 0.0551,
      "step": 9655
    },
    {
      "epoch": 0.4986315517686548,
      "grad_norm": 3.1154675483703613,
      "learning_rate": 5.101184134282884e-06,
      "loss": 0.0364,
      "step": 9656
    },
    {
      "epoch": 0.49868319132455463,
      "grad_norm": 8.662484169006348,
      "learning_rate": 5.100364940058297e-06,
      "loss": 0.0769,
      "step": 9657
    },
    {
      "epoch": 0.49873483088045445,
      "grad_norm": 11.99320125579834,
      "learning_rate": 5.099545743138503e-06,
      "loss": 0.1257,
      "step": 9658
    },
    {
      "epoch": 0.49878647043635427,
      "grad_norm": 3.6654624938964844,
      "learning_rate": 5.098726543545498e-06,
      "loss": 0.0733,
      "step": 9659
    },
    {
      "epoch": 0.4988381099922541,
      "grad_norm": 2.9110357761383057,
      "learning_rate": 5.097907341301285e-06,
      "loss": 0.0466,
      "step": 9660
    },
    {
      "epoch": 0.4988897495481539,
      "grad_norm": 3.5869526863098145,
      "learning_rate": 5.097088136427859e-06,
      "loss": 0.0281,
      "step": 9661
    },
    {
      "epoch": 0.4989413891040537,
      "grad_norm": 4.395628929138184,
      "learning_rate": 5.096268928947223e-06,
      "loss": 0.05,
      "step": 9662
    },
    {
      "epoch": 0.49899302865995354,
      "grad_norm": 3.0538618564605713,
      "learning_rate": 5.095449718881372e-06,
      "loss": 0.0577,
      "step": 9663
    },
    {
      "epoch": 0.49904466821585336,
      "grad_norm": 16.936105728149414,
      "learning_rate": 5.094630506252308e-06,
      "loss": 0.1799,
      "step": 9664
    },
    {
      "epoch": 0.4990963077717532,
      "grad_norm": 8.845166206359863,
      "learning_rate": 5.093811291082029e-06,
      "loss": 0.0899,
      "step": 9665
    },
    {
      "epoch": 0.499147947327653,
      "grad_norm": 2.275660276412964,
      "learning_rate": 5.092992073392534e-06,
      "loss": 0.0395,
      "step": 9666
    },
    {
      "epoch": 0.4991995868835528,
      "grad_norm": 2.4676709175109863,
      "learning_rate": 5.092172853205825e-06,
      "loss": 0.028,
      "step": 9667
    },
    {
      "epoch": 0.49925122643945263,
      "grad_norm": 3.3151471614837646,
      "learning_rate": 5.091353630543896e-06,
      "loss": 0.0584,
      "step": 9668
    },
    {
      "epoch": 0.49930286599535245,
      "grad_norm": 1.8424506187438965,
      "learning_rate": 5.090534405428753e-06,
      "loss": 0.0475,
      "step": 9669
    },
    {
      "epoch": 0.49935450555125227,
      "grad_norm": 7.178469181060791,
      "learning_rate": 5.08971517788239e-06,
      "loss": 0.0841,
      "step": 9670
    },
    {
      "epoch": 0.4994061451071521,
      "grad_norm": 3.242473840713501,
      "learning_rate": 5.088895947926809e-06,
      "loss": 0.0748,
      "step": 9671
    },
    {
      "epoch": 0.4994577846630519,
      "grad_norm": 3.1708083152770996,
      "learning_rate": 5.08807671558401e-06,
      "loss": 0.0476,
      "step": 9672
    },
    {
      "epoch": 0.4995094242189517,
      "grad_norm": 1.4286847114562988,
      "learning_rate": 5.087257480875991e-06,
      "loss": 0.0278,
      "step": 9673
    },
    {
      "epoch": 0.49956106377485154,
      "grad_norm": 3.0383787155151367,
      "learning_rate": 5.086438243824754e-06,
      "loss": 0.0342,
      "step": 9674
    },
    {
      "epoch": 0.49961270333075136,
      "grad_norm": 7.170670032501221,
      "learning_rate": 5.085619004452298e-06,
      "loss": 0.1441,
      "step": 9675
    },
    {
      "epoch": 0.4996643428866512,
      "grad_norm": 4.23670768737793,
      "learning_rate": 5.084799762780622e-06,
      "loss": 0.076,
      "step": 9676
    },
    {
      "epoch": 0.499715982442551,
      "grad_norm": 4.034765243530273,
      "learning_rate": 5.083980518831729e-06,
      "loss": 0.0505,
      "step": 9677
    },
    {
      "epoch": 0.4997676219984508,
      "grad_norm": 23.236581802368164,
      "learning_rate": 5.083161272627615e-06,
      "loss": 0.1513,
      "step": 9678
    },
    {
      "epoch": 0.49981926155435064,
      "grad_norm": 7.034396648406982,
      "learning_rate": 5.082342024190282e-06,
      "loss": 0.0602,
      "step": 9679
    },
    {
      "epoch": 0.49987090111025045,
      "grad_norm": 4.100638389587402,
      "learning_rate": 5.081522773541729e-06,
      "loss": 0.0398,
      "step": 9680
    },
    {
      "epoch": 0.4999225406661503,
      "grad_norm": 6.494370937347412,
      "learning_rate": 5.080703520703959e-06,
      "loss": 0.0721,
      "step": 9681
    },
    {
      "epoch": 0.4999741802220501,
      "grad_norm": 3.3346102237701416,
      "learning_rate": 5.07988426569897e-06,
      "loss": 0.0435,
      "step": 9682
    },
    {
      "epoch": 0.5000258197779499,
      "grad_norm": 15.689030647277832,
      "learning_rate": 5.0790650085487625e-06,
      "loss": 0.056,
      "step": 9683
    },
    {
      "epoch": 0.5000774593338497,
      "grad_norm": 3.220404863357544,
      "learning_rate": 5.078245749275338e-06,
      "loss": 0.0347,
      "step": 9684
    },
    {
      "epoch": 0.5001290988897495,
      "grad_norm": 7.440262794494629,
      "learning_rate": 5.077426487900694e-06,
      "loss": 0.1169,
      "step": 9685
    },
    {
      "epoch": 0.5001290988897495,
      "eval_loss": 0.06227559968829155,
      "eval_runtime": 34.1774,
      "eval_samples_per_second": 45.791,
      "eval_steps_per_second": 5.735,
      "step": 9685
    },
    {
      "epoch": 0.5001807384456494,
      "grad_norm": 9.1025390625,
      "learning_rate": 5.076607224446837e-06,
      "loss": 0.3386,
      "step": 9686
    },
    {
      "epoch": 0.5002323780015492,
      "grad_norm": 2.5032460689544678,
      "learning_rate": 5.075787958935759e-06,
      "loss": 0.0389,
      "step": 9687
    },
    {
      "epoch": 0.500284017557449,
      "grad_norm": 5.168651103973389,
      "learning_rate": 5.0749686913894706e-06,
      "loss": 0.079,
      "step": 9688
    },
    {
      "epoch": 0.5003356571133488,
      "grad_norm": 5.639957427978516,
      "learning_rate": 5.074149421829963e-06,
      "loss": 0.0662,
      "step": 9689
    },
    {
      "epoch": 0.5003872966692486,
      "grad_norm": 5.176469326019287,
      "learning_rate": 5.073330150279244e-06,
      "loss": 0.071,
      "step": 9690
    },
    {
      "epoch": 0.5004389362251485,
      "grad_norm": 5.539826393127441,
      "learning_rate": 5.072510876759311e-06,
      "loss": 0.0473,
      "step": 9691
    },
    {
      "epoch": 0.5004905757810483,
      "grad_norm": 2.3306221961975098,
      "learning_rate": 5.071691601292165e-06,
      "loss": 0.0339,
      "step": 9692
    },
    {
      "epoch": 0.5005422153369481,
      "grad_norm": 5.97498893737793,
      "learning_rate": 5.070872323899808e-06,
      "loss": 0.0542,
      "step": 9693
    },
    {
      "epoch": 0.5005938548928479,
      "grad_norm": 5.328975677490234,
      "learning_rate": 5.070053044604238e-06,
      "loss": 0.0772,
      "step": 9694
    },
    {
      "epoch": 0.5006454944487477,
      "grad_norm": 3.613624334335327,
      "learning_rate": 5.069233763427461e-06,
      "loss": 0.0804,
      "step": 9695
    },
    {
      "epoch": 0.5006971340046475,
      "grad_norm": 9.922307014465332,
      "learning_rate": 5.068414480391474e-06,
      "loss": 0.1177,
      "step": 9696
    },
    {
      "epoch": 0.5007487735605474,
      "grad_norm": 2.996788263320923,
      "learning_rate": 5.067595195518279e-06,
      "loss": 0.0592,
      "step": 9697
    },
    {
      "epoch": 0.5008004131164472,
      "grad_norm": 5.201539993286133,
      "learning_rate": 5.066775908829879e-06,
      "loss": 0.1039,
      "step": 9698
    },
    {
      "epoch": 0.500852052672347,
      "grad_norm": 8.813072204589844,
      "learning_rate": 5.065956620348271e-06,
      "loss": 0.0856,
      "step": 9699
    },
    {
      "epoch": 0.5009036922282468,
      "grad_norm": 8.986552238464355,
      "learning_rate": 5.065137330095462e-06,
      "loss": 0.1656,
      "step": 9700
    },
    {
      "epoch": 0.5009553317841466,
      "grad_norm": 3.4545514583587646,
      "learning_rate": 5.064318038093448e-06,
      "loss": 0.0479,
      "step": 9701
    },
    {
      "epoch": 0.5010069713400465,
      "grad_norm": 1.8494380712509155,
      "learning_rate": 5.063498744364234e-06,
      "loss": 0.0382,
      "step": 9702
    },
    {
      "epoch": 0.5010586108959463,
      "grad_norm": 3.0043253898620605,
      "learning_rate": 5.062679448929819e-06,
      "loss": 0.0468,
      "step": 9703
    },
    {
      "epoch": 0.5011102504518461,
      "grad_norm": 2.0506622791290283,
      "learning_rate": 5.061860151812206e-06,
      "loss": 0.0447,
      "step": 9704
    },
    {
      "epoch": 0.5011618900077459,
      "grad_norm": 3.9210801124572754,
      "learning_rate": 5.0610408530333945e-06,
      "loss": 0.069,
      "step": 9705
    },
    {
      "epoch": 0.5012135295636457,
      "grad_norm": 2.078378915786743,
      "learning_rate": 5.060221552615388e-06,
      "loss": 0.0521,
      "step": 9706
    },
    {
      "epoch": 0.5012651691195456,
      "grad_norm": 4.921864032745361,
      "learning_rate": 5.059402250580188e-06,
      "loss": 0.0595,
      "step": 9707
    },
    {
      "epoch": 0.5013168086754454,
      "grad_norm": 6.514870643615723,
      "learning_rate": 5.058582946949794e-06,
      "loss": 0.0502,
      "step": 9708
    },
    {
      "epoch": 0.5013684482313452,
      "grad_norm": 2.7147092819213867,
      "learning_rate": 5.0577636417462104e-06,
      "loss": 0.0728,
      "step": 9709
    },
    {
      "epoch": 0.501420087787245,
      "grad_norm": 8.312749862670898,
      "learning_rate": 5.056944334991438e-06,
      "loss": 0.0711,
      "step": 9710
    },
    {
      "epoch": 0.5014717273431448,
      "grad_norm": 9.351927757263184,
      "learning_rate": 5.056125026707477e-06,
      "loss": 0.0837,
      "step": 9711
    },
    {
      "epoch": 0.5015233668990446,
      "grad_norm": 7.780438423156738,
      "learning_rate": 5.05530571691633e-06,
      "loss": 0.0535,
      "step": 9712
    },
    {
      "epoch": 0.5015750064549445,
      "grad_norm": 4.898398399353027,
      "learning_rate": 5.05448640564e-06,
      "loss": 0.0337,
      "step": 9713
    },
    {
      "epoch": 0.5016266460108443,
      "grad_norm": 3.1481964588165283,
      "learning_rate": 5.0536670929004876e-06,
      "loss": 0.0488,
      "step": 9714
    },
    {
      "epoch": 0.5016782855667441,
      "grad_norm": 7.243237495422363,
      "learning_rate": 5.052847778719795e-06,
      "loss": 0.0756,
      "step": 9715
    },
    {
      "epoch": 0.5017299251226439,
      "grad_norm": 2.4997076988220215,
      "learning_rate": 5.0520284631199234e-06,
      "loss": 0.0461,
      "step": 9716
    },
    {
      "epoch": 0.5017815646785437,
      "grad_norm": 1.8444393873214722,
      "learning_rate": 5.051209146122877e-06,
      "loss": 0.0412,
      "step": 9717
    },
    {
      "epoch": 0.5018332042344436,
      "grad_norm": 5.825371265411377,
      "learning_rate": 5.050389827750655e-06,
      "loss": 0.0601,
      "step": 9718
    },
    {
      "epoch": 0.5018848437903434,
      "grad_norm": 3.1333096027374268,
      "learning_rate": 5.049570508025262e-06,
      "loss": 0.0555,
      "step": 9719
    },
    {
      "epoch": 0.5019364833462432,
      "grad_norm": 2.2741589546203613,
      "learning_rate": 5.048751186968699e-06,
      "loss": 0.0466,
      "step": 9720
    },
    {
      "epoch": 0.501988122902143,
      "grad_norm": 4.258922576904297,
      "learning_rate": 5.047931864602966e-06,
      "loss": 0.0461,
      "step": 9721
    },
    {
      "epoch": 0.5020397624580428,
      "grad_norm": 4.088797569274902,
      "learning_rate": 5.047112540950069e-06,
      "loss": 0.0525,
      "step": 9722
    },
    {
      "epoch": 0.5020914020139426,
      "grad_norm": 3.152836799621582,
      "learning_rate": 5.046293216032006e-06,
      "loss": 0.0599,
      "step": 9723
    },
    {
      "epoch": 0.5021430415698425,
      "grad_norm": 2.505476951599121,
      "learning_rate": 5.0454738898707844e-06,
      "loss": 0.0316,
      "step": 9724
    },
    {
      "epoch": 0.5021946811257423,
      "grad_norm": 2.6895980834960938,
      "learning_rate": 5.044654562488401e-06,
      "loss": 0.0236,
      "step": 9725
    },
    {
      "epoch": 0.5022463206816421,
      "grad_norm": 5.3506951332092285,
      "learning_rate": 5.043835233906863e-06,
      "loss": 0.0827,
      "step": 9726
    },
    {
      "epoch": 0.5022979602375419,
      "grad_norm": 15.700155258178711,
      "learning_rate": 5.043015904148167e-06,
      "loss": 0.125,
      "step": 9727
    },
    {
      "epoch": 0.5023495997934417,
      "grad_norm": 5.132328033447266,
      "learning_rate": 5.042196573234322e-06,
      "loss": 0.0501,
      "step": 9728
    },
    {
      "epoch": 0.5024012393493416,
      "grad_norm": 3.0214951038360596,
      "learning_rate": 5.041377241187326e-06,
      "loss": 0.0472,
      "step": 9729
    },
    {
      "epoch": 0.5024528789052414,
      "grad_norm": 5.933135986328125,
      "learning_rate": 5.040557908029182e-06,
      "loss": 0.0765,
      "step": 9730
    },
    {
      "epoch": 0.5025045184611412,
      "grad_norm": 1.8679231405258179,
      "learning_rate": 5.039738573781894e-06,
      "loss": 0.032,
      "step": 9731
    },
    {
      "epoch": 0.502556158017041,
      "grad_norm": 7.202311992645264,
      "learning_rate": 5.03891923846746e-06,
      "loss": 0.1005,
      "step": 9732
    },
    {
      "epoch": 0.5026077975729408,
      "grad_norm": 4.191286087036133,
      "learning_rate": 5.038099902107889e-06,
      "loss": 0.0735,
      "step": 9733
    },
    {
      "epoch": 0.5026594371288406,
      "grad_norm": 9.269256591796875,
      "learning_rate": 5.037280564725179e-06,
      "loss": 0.055,
      "step": 9734
    },
    {
      "epoch": 0.5027110766847405,
      "grad_norm": 3.414630651473999,
      "learning_rate": 5.0364612263413345e-06,
      "loss": 0.0557,
      "step": 9735
    },
    {
      "epoch": 0.5027627162406403,
      "grad_norm": 17.38247299194336,
      "learning_rate": 5.035641886978357e-06,
      "loss": 0.0309,
      "step": 9736
    },
    {
      "epoch": 0.5028143557965401,
      "grad_norm": 4.763491630554199,
      "learning_rate": 5.03482254665825e-06,
      "loss": 0.056,
      "step": 9737
    },
    {
      "epoch": 0.5028659953524399,
      "grad_norm": 2.6507070064544678,
      "learning_rate": 5.034003205403017e-06,
      "loss": 0.0214,
      "step": 9738
    },
    {
      "epoch": 0.5029176349083397,
      "grad_norm": 10.526296615600586,
      "learning_rate": 5.033183863234656e-06,
      "loss": 0.1291,
      "step": 9739
    },
    {
      "epoch": 0.5029692744642396,
      "grad_norm": 10.770461082458496,
      "learning_rate": 5.032364520175176e-06,
      "loss": 0.0908,
      "step": 9740
    },
    {
      "epoch": 0.5030209140201394,
      "grad_norm": 3.5355820655822754,
      "learning_rate": 5.031545176246575e-06,
      "loss": 0.0263,
      "step": 9741
    },
    {
      "epoch": 0.5030725535760392,
      "grad_norm": 4.30959939956665,
      "learning_rate": 5.030725831470858e-06,
      "loss": 0.0816,
      "step": 9742
    },
    {
      "epoch": 0.503124193131939,
      "grad_norm": 3.1751322746276855,
      "learning_rate": 5.029906485870028e-06,
      "loss": 0.0427,
      "step": 9743
    },
    {
      "epoch": 0.5031758326878388,
      "grad_norm": 4.863467693328857,
      "learning_rate": 5.029087139466086e-06,
      "loss": 0.0695,
      "step": 9744
    },
    {
      "epoch": 0.5032274722437386,
      "grad_norm": 3.4657156467437744,
      "learning_rate": 5.028267792281036e-06,
      "loss": 0.0457,
      "step": 9745
    },
    {
      "epoch": 0.5032791117996385,
      "grad_norm": 9.544281959533691,
      "learning_rate": 5.02744844433688e-06,
      "loss": 0.0554,
      "step": 9746
    },
    {
      "epoch": 0.5033307513555384,
      "grad_norm": 5.770705223083496,
      "learning_rate": 5.026629095655622e-06,
      "loss": 0.0662,
      "step": 9747
    },
    {
      "epoch": 0.5033823909114382,
      "grad_norm": 5.265528202056885,
      "learning_rate": 5.025809746259264e-06,
      "loss": 0.0517,
      "step": 9748
    },
    {
      "epoch": 0.503434030467338,
      "grad_norm": 8.819525718688965,
      "learning_rate": 5.0249903961698086e-06,
      "loss": 0.1821,
      "step": 9749
    },
    {
      "epoch": 0.5034856700232379,
      "grad_norm": 2.1431519985198975,
      "learning_rate": 5.024171045409262e-06,
      "loss": 0.0222,
      "step": 9750
    },
    {
      "epoch": 0.5035373095791377,
      "grad_norm": 1.9239392280578613,
      "learning_rate": 5.023351693999621e-06,
      "loss": 0.0282,
      "step": 9751
    },
    {
      "epoch": 0.5035889491350375,
      "grad_norm": 6.771427631378174,
      "learning_rate": 5.022532341962894e-06,
      "loss": 0.0445,
      "step": 9752
    },
    {
      "epoch": 0.5036405886909373,
      "grad_norm": 6.500406265258789,
      "learning_rate": 5.021712989321081e-06,
      "loss": 0.1041,
      "step": 9753
    },
    {
      "epoch": 0.5036922282468371,
      "grad_norm": 3.1799633502960205,
      "learning_rate": 5.020893636096185e-06,
      "loss": 0.0402,
      "step": 9754
    },
    {
      "epoch": 0.5037438678027369,
      "grad_norm": 14.36490249633789,
      "learning_rate": 5.02007428231021e-06,
      "loss": 0.1053,
      "step": 9755
    },
    {
      "epoch": 0.5037955073586368,
      "grad_norm": 4.56748628616333,
      "learning_rate": 5.019254927985159e-06,
      "loss": 0.0689,
      "step": 9756
    },
    {
      "epoch": 0.5038471469145366,
      "grad_norm": 2.0009968280792236,
      "learning_rate": 5.018435573143035e-06,
      "loss": 0.0333,
      "step": 9757
    },
    {
      "epoch": 0.5038987864704364,
      "grad_norm": 5.2596211433410645,
      "learning_rate": 5.01761621780584e-06,
      "loss": 0.0633,
      "step": 9758
    },
    {
      "epoch": 0.5039504260263362,
      "grad_norm": 5.6517653465271,
      "learning_rate": 5.016796861995578e-06,
      "loss": 0.0505,
      "step": 9759
    },
    {
      "epoch": 0.504002065582236,
      "grad_norm": 8.859572410583496,
      "learning_rate": 5.015977505734252e-06,
      "loss": 0.1218,
      "step": 9760
    },
    {
      "epoch": 0.5040537051381359,
      "grad_norm": 7.83172082901001,
      "learning_rate": 5.015158149043865e-06,
      "loss": 0.078,
      "step": 9761
    },
    {
      "epoch": 0.5041053446940357,
      "grad_norm": 3.9918503761291504,
      "learning_rate": 5.014338791946421e-06,
      "loss": 0.0353,
      "step": 9762
    },
    {
      "epoch": 0.5041569842499355,
      "grad_norm": 1.8970136642456055,
      "learning_rate": 5.0135194344639194e-06,
      "loss": 0.0391,
      "step": 9763
    },
    {
      "epoch": 0.5042086238058353,
      "grad_norm": 2.2565746307373047,
      "learning_rate": 5.012700076618367e-06,
      "loss": 0.0477,
      "step": 9764
    },
    {
      "epoch": 0.5042602633617351,
      "grad_norm": 2.906761884689331,
      "learning_rate": 5.0118807184317665e-06,
      "loss": 0.0585,
      "step": 9765
    },
    {
      "epoch": 0.504311902917635,
      "grad_norm": 2.2614200115203857,
      "learning_rate": 5.011061359926121e-06,
      "loss": 0.0215,
      "step": 9766
    },
    {
      "epoch": 0.5043635424735348,
      "grad_norm": 1.8808274269104004,
      "learning_rate": 5.01024200112343e-06,
      "loss": 0.0244,
      "step": 9767
    },
    {
      "epoch": 0.5044151820294346,
      "grad_norm": 1.8687059879302979,
      "learning_rate": 5.009422642045701e-06,
      "loss": 0.0415,
      "step": 9768
    },
    {
      "epoch": 0.5044668215853344,
      "grad_norm": 5.877493381500244,
      "learning_rate": 5.008603282714937e-06,
      "loss": 0.0381,
      "step": 9769
    },
    {
      "epoch": 0.5045184611412342,
      "grad_norm": 2.6282503604888916,
      "learning_rate": 5.007783923153138e-06,
      "loss": 0.0556,
      "step": 9770
    },
    {
      "epoch": 0.504570100697134,
      "grad_norm": 2.4848220348358154,
      "learning_rate": 5.006964563382309e-06,
      "loss": 0.0275,
      "step": 9771
    },
    {
      "epoch": 0.5046217402530339,
      "grad_norm": 12.390652656555176,
      "learning_rate": 5.006145203424454e-06,
      "loss": 0.1831,
      "step": 9772
    },
    {
      "epoch": 0.5046733798089337,
      "grad_norm": 5.842953681945801,
      "learning_rate": 5.005325843301575e-06,
      "loss": 0.071,
      "step": 9773
    },
    {
      "epoch": 0.5047250193648335,
      "grad_norm": 3.3471524715423584,
      "learning_rate": 5.0045064830356745e-06,
      "loss": 0.0725,
      "step": 9774
    },
    {
      "epoch": 0.5047766589207333,
      "grad_norm": 2.7182021141052246,
      "learning_rate": 5.003687122648759e-06,
      "loss": 0.0304,
      "step": 9775
    },
    {
      "epoch": 0.5048282984766331,
      "grad_norm": 2.373563289642334,
      "learning_rate": 5.002867762162827e-06,
      "loss": 0.0306,
      "step": 9776
    },
    {
      "epoch": 0.504879938032533,
      "grad_norm": 15.224905014038086,
      "learning_rate": 5.002048401599885e-06,
      "loss": 0.1247,
      "step": 9777
    },
    {
      "epoch": 0.5049315775884328,
      "grad_norm": 24.518672943115234,
      "learning_rate": 5.001229040981934e-06,
      "loss": 0.2997,
      "step": 9778
    },
    {
      "epoch": 0.5049832171443326,
      "grad_norm": 2.695246458053589,
      "learning_rate": 5.000409680330979e-06,
      "loss": 0.0451,
      "step": 9779
    },
    {
      "epoch": 0.5050348567002324,
      "grad_norm": 6.477114677429199,
      "learning_rate": 4.999590319669022e-06,
      "loss": 0.0697,
      "step": 9780
    },
    {
      "epoch": 0.5050864962561322,
      "grad_norm": 6.173931121826172,
      "learning_rate": 4.998770959018067e-06,
      "loss": 0.0482,
      "step": 9781
    },
    {
      "epoch": 0.505138135812032,
      "grad_norm": 1.309415340423584,
      "learning_rate": 4.997951598400117e-06,
      "loss": 0.0142,
      "step": 9782
    },
    {
      "epoch": 0.5051897753679319,
      "grad_norm": 7.2112579345703125,
      "learning_rate": 4.997132237837174e-06,
      "loss": 0.0472,
      "step": 9783
    },
    {
      "epoch": 0.5052414149238317,
      "grad_norm": 1.116686463356018,
      "learning_rate": 4.996312877351242e-06,
      "loss": 0.013,
      "step": 9784
    },
    {
      "epoch": 0.5052930544797315,
      "grad_norm": 3.4861037731170654,
      "learning_rate": 4.995493516964326e-06,
      "loss": 0.0556,
      "step": 9785
    },
    {
      "epoch": 0.5053446940356313,
      "grad_norm": 6.01624059677124,
      "learning_rate": 4.994674156698427e-06,
      "loss": 0.0478,
      "step": 9786
    },
    {
      "epoch": 0.5053963335915311,
      "grad_norm": 2.6059603691101074,
      "learning_rate": 4.9938547965755475e-06,
      "loss": 0.0386,
      "step": 9787
    },
    {
      "epoch": 0.505447973147431,
      "grad_norm": 2.530843734741211,
      "learning_rate": 4.993035436617692e-06,
      "loss": 0.0396,
      "step": 9788
    },
    {
      "epoch": 0.5054996127033308,
      "grad_norm": 2.609922409057617,
      "learning_rate": 4.992216076846864e-06,
      "loss": 0.0566,
      "step": 9789
    },
    {
      "epoch": 0.5055512522592306,
      "grad_norm": 5.238768100738525,
      "learning_rate": 4.991396717285065e-06,
      "loss": 0.0534,
      "step": 9790
    },
    {
      "epoch": 0.5056028918151304,
      "grad_norm": 16.37824058532715,
      "learning_rate": 4.9905773579542996e-06,
      "loss": 0.1041,
      "step": 9791
    },
    {
      "epoch": 0.5056545313710302,
      "grad_norm": 16.798137664794922,
      "learning_rate": 4.98975799887657e-06,
      "loss": 0.0409,
      "step": 9792
    },
    {
      "epoch": 0.50570617092693,
      "grad_norm": 4.200066566467285,
      "learning_rate": 4.988938640073882e-06,
      "loss": 0.0359,
      "step": 9793
    },
    {
      "epoch": 0.5057578104828299,
      "grad_norm": 2.625094175338745,
      "learning_rate": 4.988119281568236e-06,
      "loss": 0.0576,
      "step": 9794
    },
    {
      "epoch": 0.5058094500387297,
      "grad_norm": 3.4050183296203613,
      "learning_rate": 4.987299923381634e-06,
      "loss": 0.0824,
      "step": 9795
    },
    {
      "epoch": 0.5058610895946295,
      "grad_norm": 9.403806686401367,
      "learning_rate": 4.986480565536083e-06,
      "loss": 0.0878,
      "step": 9796
    },
    {
      "epoch": 0.5059127291505293,
      "grad_norm": 1.7156982421875,
      "learning_rate": 4.985661208053582e-06,
      "loss": 0.0265,
      "step": 9797
    },
    {
      "epoch": 0.5059643687064291,
      "grad_norm": 5.395516395568848,
      "learning_rate": 4.984841850956136e-06,
      "loss": 0.0805,
      "step": 9798
    },
    {
      "epoch": 0.506016008262329,
      "grad_norm": 6.542854309082031,
      "learning_rate": 4.984022494265749e-06,
      "loss": 0.0486,
      "step": 9799
    },
    {
      "epoch": 0.5060676478182288,
      "grad_norm": 3.115104913711548,
      "learning_rate": 4.983203138004424e-06,
      "loss": 0.0533,
      "step": 9800
    },
    {
      "epoch": 0.5061192873741286,
      "grad_norm": 19.955806732177734,
      "learning_rate": 4.982383782194162e-06,
      "loss": 0.1417,
      "step": 9801
    },
    {
      "epoch": 0.5061709269300284,
      "grad_norm": 22.944595336914062,
      "learning_rate": 4.9815644268569655e-06,
      "loss": 0.1628,
      "step": 9802
    },
    {
      "epoch": 0.5062225664859282,
      "grad_norm": 8.613909721374512,
      "learning_rate": 4.980745072014841e-06,
      "loss": 0.1761,
      "step": 9803
    },
    {
      "epoch": 0.506274206041828,
      "grad_norm": 9.184671401977539,
      "learning_rate": 4.979925717689791e-06,
      "loss": 0.0936,
      "step": 9804
    },
    {
      "epoch": 0.5063258455977279,
      "grad_norm": 3.86041522026062,
      "learning_rate": 4.979106363903817e-06,
      "loss": 0.0533,
      "step": 9805
    },
    {
      "epoch": 0.5063774851536277,
      "grad_norm": 1.7098274230957031,
      "learning_rate": 4.978287010678921e-06,
      "loss": 0.0298,
      "step": 9806
    },
    {
      "epoch": 0.5064291247095275,
      "grad_norm": 7.201534271240234,
      "learning_rate": 4.977467658037109e-06,
      "loss": 0.0744,
      "step": 9807
    },
    {
      "epoch": 0.5064807642654273,
      "grad_norm": 6.248540878295898,
      "learning_rate": 4.976648306000381e-06,
      "loss": 0.0836,
      "step": 9808
    },
    {
      "epoch": 0.5065324038213271,
      "grad_norm": 9.987733840942383,
      "learning_rate": 4.97582895459074e-06,
      "loss": 0.1167,
      "step": 9809
    },
    {
      "epoch": 0.506584043377227,
      "grad_norm": 9.468426704406738,
      "learning_rate": 4.9750096038301906e-06,
      "loss": 0.084,
      "step": 9810
    },
    {
      "epoch": 0.5066356829331268,
      "grad_norm": 6.376951694488525,
      "learning_rate": 4.974190253740737e-06,
      "loss": 0.1279,
      "step": 9811
    },
    {
      "epoch": 0.5066873224890266,
      "grad_norm": 5.724995136260986,
      "learning_rate": 4.97337090434438e-06,
      "loss": 0.0632,
      "step": 9812
    },
    {
      "epoch": 0.5067389620449264,
      "grad_norm": 3.148501396179199,
      "learning_rate": 4.972551555663121e-06,
      "loss": 0.0564,
      "step": 9813
    },
    {
      "epoch": 0.5067906016008262,
      "grad_norm": 12.623642921447754,
      "learning_rate": 4.971732207718966e-06,
      "loss": 0.1374,
      "step": 9814
    },
    {
      "epoch": 0.506842241156726,
      "grad_norm": 7.729351997375488,
      "learning_rate": 4.970912860533916e-06,
      "loss": 0.1064,
      "step": 9815
    },
    {
      "epoch": 0.5068938807126259,
      "grad_norm": 3.77490234375,
      "learning_rate": 4.970093514129973e-06,
      "loss": 0.0741,
      "step": 9816
    },
    {
      "epoch": 0.5069455202685257,
      "grad_norm": 21.04941177368164,
      "learning_rate": 4.969274168529143e-06,
      "loss": 0.2741,
      "step": 9817
    },
    {
      "epoch": 0.5069971598244255,
      "grad_norm": 2.1400575637817383,
      "learning_rate": 4.9684548237534266e-06,
      "loss": 0.0242,
      "step": 9818
    },
    {
      "epoch": 0.5070487993803253,
      "grad_norm": 4.159761428833008,
      "learning_rate": 4.9676354798248264e-06,
      "loss": 0.0593,
      "step": 9819
    },
    {
      "epoch": 0.5071004389362251,
      "grad_norm": 1.573097586631775,
      "learning_rate": 4.966816136765345e-06,
      "loss": 0.0245,
      "step": 9820
    },
    {
      "epoch": 0.507152078492125,
      "grad_norm": 3.9231679439544678,
      "learning_rate": 4.965996794596986e-06,
      "loss": 0.0452,
      "step": 9821
    },
    {
      "epoch": 0.5072037180480248,
      "grad_norm": 16.02167510986328,
      "learning_rate": 4.965177453341752e-06,
      "loss": 0.0515,
      "step": 9822
    },
    {
      "epoch": 0.5072553576039246,
      "grad_norm": 14.154837608337402,
      "learning_rate": 4.964358113021644e-06,
      "loss": 0.1557,
      "step": 9823
    },
    {
      "epoch": 0.5073069971598244,
      "grad_norm": 11.169137001037598,
      "learning_rate": 4.963538773658666e-06,
      "loss": 0.1229,
      "step": 9824
    },
    {
      "epoch": 0.5073586367157242,
      "grad_norm": 2.586637020111084,
      "learning_rate": 4.962719435274822e-06,
      "loss": 0.0668,
      "step": 9825
    },
    {
      "epoch": 0.507410276271624,
      "grad_norm": 1.7997125387191772,
      "learning_rate": 4.961900097892113e-06,
      "loss": 0.036,
      "step": 9826
    },
    {
      "epoch": 0.5074619158275239,
      "grad_norm": 4.977719783782959,
      "learning_rate": 4.9610807615325404e-06,
      "loss": 0.0545,
      "step": 9827
    },
    {
      "epoch": 0.5075135553834237,
      "grad_norm": 2.0155584812164307,
      "learning_rate": 4.960261426218108e-06,
      "loss": 0.0488,
      "step": 9828
    },
    {
      "epoch": 0.5075651949393235,
      "grad_norm": 5.051807403564453,
      "learning_rate": 4.959442091970821e-06,
      "loss": 0.0659,
      "step": 9829
    },
    {
      "epoch": 0.5076168344952233,
      "grad_norm": 4.754261016845703,
      "learning_rate": 4.958622758812676e-06,
      "loss": 0.0746,
      "step": 9830
    },
    {
      "epoch": 0.5076684740511231,
      "grad_norm": 1.6276891231536865,
      "learning_rate": 4.957803426765679e-06,
      "loss": 0.0241,
      "step": 9831
    },
    {
      "epoch": 0.507720113607023,
      "grad_norm": 11.083399772644043,
      "learning_rate": 4.956984095851833e-06,
      "loss": 0.1515,
      "step": 9832
    },
    {
      "epoch": 0.5077717531629228,
      "grad_norm": 3.5339550971984863,
      "learning_rate": 4.956164766093139e-06,
      "loss": 0.0851,
      "step": 9833
    },
    {
      "epoch": 0.5078233927188226,
      "grad_norm": 3.439706325531006,
      "learning_rate": 4.9553454375116e-06,
      "loss": 0.0556,
      "step": 9834
    },
    {
      "epoch": 0.5078750322747224,
      "grad_norm": 3.5690531730651855,
      "learning_rate": 4.954526110129216e-06,
      "loss": 0.0345,
      "step": 9835
    },
    {
      "epoch": 0.5079266718306222,
      "grad_norm": 7.627254009246826,
      "learning_rate": 4.953706783967993e-06,
      "loss": 0.1205,
      "step": 9836
    },
    {
      "epoch": 0.507978311386522,
      "grad_norm": 2.2539632320404053,
      "learning_rate": 4.9528874590499326e-06,
      "loss": 0.0411,
      "step": 9837
    },
    {
      "epoch": 0.5080299509424219,
      "grad_norm": 3.0227181911468506,
      "learning_rate": 4.9520681353970345e-06,
      "loss": 0.061,
      "step": 9838
    },
    {
      "epoch": 0.5080815904983217,
      "grad_norm": 3.1811506748199463,
      "learning_rate": 4.951248813031303e-06,
      "loss": 0.0522,
      "step": 9839
    },
    {
      "epoch": 0.5081332300542215,
      "grad_norm": 3.8505496978759766,
      "learning_rate": 4.95042949197474e-06,
      "loss": 0.0778,
      "step": 9840
    },
    {
      "epoch": 0.5081848696101213,
      "grad_norm": 2.862596035003662,
      "learning_rate": 4.949610172249346e-06,
      "loss": 0.0338,
      "step": 9841
    },
    {
      "epoch": 0.5082365091660211,
      "grad_norm": 3.856435775756836,
      "learning_rate": 4.9487908538771235e-06,
      "loss": 0.0815,
      "step": 9842
    },
    {
      "epoch": 0.508288148721921,
      "grad_norm": 3.5693111419677734,
      "learning_rate": 4.9479715368800765e-06,
      "loss": 0.037,
      "step": 9843
    },
    {
      "epoch": 0.5083397882778208,
      "grad_norm": 4.1970953941345215,
      "learning_rate": 4.947152221280206e-06,
      "loss": 0.0599,
      "step": 9844
    },
    {
      "epoch": 0.5083914278337206,
      "grad_norm": 3.2441623210906982,
      "learning_rate": 4.946332907099514e-06,
      "loss": 0.0466,
      "step": 9845
    },
    {
      "epoch": 0.5084430673896204,
      "grad_norm": 6.119664669036865,
      "learning_rate": 4.945513594360001e-06,
      "loss": 0.0654,
      "step": 9846
    },
    {
      "epoch": 0.5084947069455202,
      "grad_norm": 12.013288497924805,
      "learning_rate": 4.94469428308367e-06,
      "loss": 0.0709,
      "step": 9847
    },
    {
      "epoch": 0.50854634650142,
      "grad_norm": 1.7731878757476807,
      "learning_rate": 4.943874973292525e-06,
      "loss": 0.0325,
      "step": 9848
    },
    {
      "epoch": 0.5085979860573199,
      "grad_norm": 4.564419269561768,
      "learning_rate": 4.943055665008564e-06,
      "loss": 0.062,
      "step": 9849
    },
    {
      "epoch": 0.5086496256132197,
      "grad_norm": 11.815194129943848,
      "learning_rate": 4.94223635825379e-06,
      "loss": 0.1023,
      "step": 9850
    },
    {
      "epoch": 0.5087012651691195,
      "grad_norm": 3.652096748352051,
      "learning_rate": 4.941417053050208e-06,
      "loss": 0.0548,
      "step": 9851
    },
    {
      "epoch": 0.5087529047250193,
      "grad_norm": 3.86242413520813,
      "learning_rate": 4.940597749419814e-06,
      "loss": 0.0828,
      "step": 9852
    },
    {
      "epoch": 0.5088045442809191,
      "grad_norm": 3.878910541534424,
      "learning_rate": 4.939778447384614e-06,
      "loss": 0.0585,
      "step": 9853
    },
    {
      "epoch": 0.508856183836819,
      "grad_norm": 10.390105247497559,
      "learning_rate": 4.938959146966606e-06,
      "loss": 0.0965,
      "step": 9854
    },
    {
      "epoch": 0.5089078233927188,
      "grad_norm": 3.045750856399536,
      "learning_rate": 4.938139848187797e-06,
      "loss": 0.0312,
      "step": 9855
    },
    {
      "epoch": 0.5089594629486186,
      "grad_norm": 9.95605182647705,
      "learning_rate": 4.9373205510701824e-06,
      "loss": 0.0567,
      "step": 9856
    },
    {
      "epoch": 0.5090111025045184,
      "grad_norm": 3.4588463306427,
      "learning_rate": 4.936501255635768e-06,
      "loss": 0.067,
      "step": 9857
    },
    {
      "epoch": 0.5090627420604182,
      "grad_norm": 4.49181604385376,
      "learning_rate": 4.935681961906552e-06,
      "loss": 0.0571,
      "step": 9858
    },
    {
      "epoch": 0.5091143816163181,
      "grad_norm": 8.405349731445312,
      "learning_rate": 4.9348626699045405e-06,
      "loss": 0.174,
      "step": 9859
    },
    {
      "epoch": 0.5091660211722179,
      "grad_norm": 4.906909465789795,
      "learning_rate": 4.9340433796517305e-06,
      "loss": 0.0591,
      "step": 9860
    },
    {
      "epoch": 0.5092176607281177,
      "grad_norm": 3.806025505065918,
      "learning_rate": 4.933224091170122e-06,
      "loss": 0.0755,
      "step": 9861
    },
    {
      "epoch": 0.5092693002840175,
      "grad_norm": 16.359649658203125,
      "learning_rate": 4.932404804481723e-06,
      "loss": 0.0861,
      "step": 9862
    },
    {
      "epoch": 0.5093209398399173,
      "grad_norm": 2.905748128890991,
      "learning_rate": 4.931585519608528e-06,
      "loss": 0.0377,
      "step": 9863
    },
    {
      "epoch": 0.5093725793958171,
      "grad_norm": 2.598121404647827,
      "learning_rate": 4.93076623657254e-06,
      "loss": 0.0299,
      "step": 9864
    },
    {
      "epoch": 0.509424218951717,
      "grad_norm": 5.401383399963379,
      "learning_rate": 4.929946955395763e-06,
      "loss": 0.058,
      "step": 9865
    },
    {
      "epoch": 0.5094758585076168,
      "grad_norm": 7.516396522521973,
      "learning_rate": 4.929127676100195e-06,
      "loss": 0.0626,
      "step": 9866
    },
    {
      "epoch": 0.5095274980635166,
      "grad_norm": 5.917552471160889,
      "learning_rate": 4.9283083987078376e-06,
      "loss": 0.0832,
      "step": 9867
    },
    {
      "epoch": 0.5095791376194164,
      "grad_norm": 18.046058654785156,
      "learning_rate": 4.92748912324069e-06,
      "loss": 0.1221,
      "step": 9868
    },
    {
      "epoch": 0.5096307771753162,
      "grad_norm": 3.272660493850708,
      "learning_rate": 4.926669849720756e-06,
      "loss": 0.0545,
      "step": 9869
    },
    {
      "epoch": 0.5096824167312161,
      "grad_norm": 5.089259624481201,
      "learning_rate": 4.925850578170037e-06,
      "loss": 0.1086,
      "step": 9870
    },
    {
      "epoch": 0.5097340562871159,
      "grad_norm": 7.90696382522583,
      "learning_rate": 4.925031308610531e-06,
      "loss": 0.1335,
      "step": 9871
    },
    {
      "epoch": 0.5097856958430157,
      "grad_norm": 2.036220073699951,
      "learning_rate": 4.92421204106424e-06,
      "loss": 0.0225,
      "step": 9872
    },
    {
      "epoch": 0.5098373353989155,
      "grad_norm": 3.700446367263794,
      "learning_rate": 4.923392775553167e-06,
      "loss": 0.0369,
      "step": 9873
    },
    {
      "epoch": 0.5098889749548154,
      "grad_norm": 5.815946578979492,
      "learning_rate": 4.922573512099307e-06,
      "loss": 0.0411,
      "step": 9874
    },
    {
      "epoch": 0.5099406145107153,
      "grad_norm": 2.3110010623931885,
      "learning_rate": 4.921754250724664e-06,
      "loss": 0.019,
      "step": 9875
    },
    {
      "epoch": 0.5099922540666151,
      "grad_norm": 7.857186794281006,
      "learning_rate": 4.920934991451238e-06,
      "loss": 0.0613,
      "step": 9876
    },
    {
      "epoch": 0.5100438936225149,
      "grad_norm": 3.2512831687927246,
      "learning_rate": 4.920115734301032e-06,
      "loss": 0.0628,
      "step": 9877
    },
    {
      "epoch": 0.5100955331784147,
      "grad_norm": 7.205137729644775,
      "learning_rate": 4.919296479296042e-06,
      "loss": 0.0902,
      "step": 9878
    },
    {
      "epoch": 0.5101471727343145,
      "grad_norm": 12.527265548706055,
      "learning_rate": 4.918477226458272e-06,
      "loss": 0.1454,
      "step": 9879
    },
    {
      "epoch": 0.5101988122902144,
      "grad_norm": 6.9284749031066895,
      "learning_rate": 4.917657975809719e-06,
      "loss": 0.0532,
      "step": 9880
    },
    {
      "epoch": 0.5102504518461142,
      "grad_norm": 6.285793304443359,
      "learning_rate": 4.9168387273723875e-06,
      "loss": 0.081,
      "step": 9881
    },
    {
      "epoch": 0.510302091402014,
      "grad_norm": 2.7131786346435547,
      "learning_rate": 4.916019481168272e-06,
      "loss": 0.0478,
      "step": 9882
    },
    {
      "epoch": 0.5103537309579138,
      "grad_norm": 0.8289482593536377,
      "learning_rate": 4.915200237219378e-06,
      "loss": 0.008,
      "step": 9883
    },
    {
      "epoch": 0.5104053705138136,
      "grad_norm": 3.2596817016601562,
      "learning_rate": 4.914380995547703e-06,
      "loss": 0.05,
      "step": 9884
    },
    {
      "epoch": 0.5104570100697134,
      "grad_norm": 6.39797830581665,
      "learning_rate": 4.9135617561752465e-06,
      "loss": 0.0565,
      "step": 9885
    },
    {
      "epoch": 0.5105086496256133,
      "grad_norm": 4.230409145355225,
      "learning_rate": 4.91274251912401e-06,
      "loss": 0.0428,
      "step": 9886
    },
    {
      "epoch": 0.5105602891815131,
      "grad_norm": 4.785290241241455,
      "learning_rate": 4.91192328441599e-06,
      "loss": 0.0736,
      "step": 9887
    },
    {
      "epoch": 0.5106119287374129,
      "grad_norm": 2.476675033569336,
      "learning_rate": 4.911104052073193e-06,
      "loss": 0.0483,
      "step": 9888
    },
    {
      "epoch": 0.5106635682933127,
      "grad_norm": 1.741945505142212,
      "learning_rate": 4.910284822117611e-06,
      "loss": 0.0249,
      "step": 9889
    },
    {
      "epoch": 0.5107152078492125,
      "grad_norm": 3.2593274116516113,
      "learning_rate": 4.909465594571249e-06,
      "loss": 0.0701,
      "step": 9890
    },
    {
      "epoch": 0.5107668474051124,
      "grad_norm": 3.6728146076202393,
      "learning_rate": 4.908646369456103e-06,
      "loss": 0.0459,
      "step": 9891
    },
    {
      "epoch": 0.5108184869610122,
      "grad_norm": 5.48045015335083,
      "learning_rate": 4.907827146794177e-06,
      "loss": 0.077,
      "step": 9892
    },
    {
      "epoch": 0.510870126516912,
      "grad_norm": 2.740675687789917,
      "learning_rate": 4.9070079266074666e-06,
      "loss": 0.0592,
      "step": 9893
    },
    {
      "epoch": 0.5109217660728118,
      "grad_norm": 5.394093990325928,
      "learning_rate": 4.906188708917971e-06,
      "loss": 0.0394,
      "step": 9894
    },
    {
      "epoch": 0.5109734056287116,
      "grad_norm": 2.9713306427001953,
      "learning_rate": 4.9053694937476945e-06,
      "loss": 0.062,
      "step": 9895
    },
    {
      "epoch": 0.5110250451846114,
      "grad_norm": 9.859883308410645,
      "learning_rate": 4.9045502811186295e-06,
      "loss": 0.129,
      "step": 9896
    },
    {
      "epoch": 0.5110766847405113,
      "grad_norm": 10.864999771118164,
      "learning_rate": 4.9037310710527785e-06,
      "loss": 0.0715,
      "step": 9897
    },
    {
      "epoch": 0.5111283242964111,
      "grad_norm": 3.8715267181396484,
      "learning_rate": 4.9029118635721416e-06,
      "loss": 0.0851,
      "step": 9898
    },
    {
      "epoch": 0.5111799638523109,
      "grad_norm": 4.096114635467529,
      "learning_rate": 4.902092658698717e-06,
      "loss": 0.0687,
      "step": 9899
    },
    {
      "epoch": 0.5112316034082107,
      "grad_norm": 8.219914436340332,
      "learning_rate": 4.901273456454504e-06,
      "loss": 0.0267,
      "step": 9900
    },
    {
      "epoch": 0.5112832429641105,
      "grad_norm": 12.490543365478516,
      "learning_rate": 4.900454256861499e-06,
      "loss": 0.1584,
      "step": 9901
    },
    {
      "epoch": 0.5113348825200104,
      "grad_norm": 5.749051094055176,
      "learning_rate": 4.899635059941704e-06,
      "loss": 0.068,
      "step": 9902
    },
    {
      "epoch": 0.5113865220759102,
      "grad_norm": 2.3580539226531982,
      "learning_rate": 4.898815865717118e-06,
      "loss": 0.0347,
      "step": 9903
    },
    {
      "epoch": 0.51143816163181,
      "grad_norm": 5.339820384979248,
      "learning_rate": 4.897996674209737e-06,
      "loss": 0.0792,
      "step": 9904
    },
    {
      "epoch": 0.5114898011877098,
      "grad_norm": 12.506937980651855,
      "learning_rate": 4.8971774854415615e-06,
      "loss": 0.0825,
      "step": 9905
    },
    {
      "epoch": 0.5115414407436096,
      "grad_norm": 2.5390098094940186,
      "learning_rate": 4.8963582994345904e-06,
      "loss": 0.0409,
      "step": 9906
    },
    {
      "epoch": 0.5115930802995095,
      "grad_norm": 6.176205635070801,
      "learning_rate": 4.8955391162108206e-06,
      "loss": 0.0831,
      "step": 9907
    },
    {
      "epoch": 0.5116447198554093,
      "grad_norm": 12.093402862548828,
      "learning_rate": 4.89471993579225e-06,
      "loss": 0.2056,
      "step": 9908
    },
    {
      "epoch": 0.5116963594113091,
      "grad_norm": 2.616330146789551,
      "learning_rate": 4.893900758200879e-06,
      "loss": 0.0431,
      "step": 9909
    },
    {
      "epoch": 0.5117479989672089,
      "grad_norm": 26.57526969909668,
      "learning_rate": 4.893081583458706e-06,
      "loss": 0.077,
      "step": 9910
    },
    {
      "epoch": 0.5117996385231087,
      "grad_norm": 2.367231845855713,
      "learning_rate": 4.892262411587728e-06,
      "loss": 0.0453,
      "step": 9911
    },
    {
      "epoch": 0.5118512780790085,
      "grad_norm": 2.5221617221832275,
      "learning_rate": 4.891443242609944e-06,
      "loss": 0.0236,
      "step": 9912
    },
    {
      "epoch": 0.5119029176349084,
      "grad_norm": 6.194677829742432,
      "learning_rate": 4.890624076547348e-06,
      "loss": 0.086,
      "step": 9913
    },
    {
      "epoch": 0.5119545571908082,
      "grad_norm": 2.444186210632324,
      "learning_rate": 4.889804913421947e-06,
      "loss": 0.0551,
      "step": 9914
    },
    {
      "epoch": 0.512006196746708,
      "grad_norm": 3.7019922733306885,
      "learning_rate": 4.888985753255728e-06,
      "loss": 0.0589,
      "step": 9915
    },
    {
      "epoch": 0.5120578363026078,
      "grad_norm": 2.237302303314209,
      "learning_rate": 4.888166596070696e-06,
      "loss": 0.0421,
      "step": 9916
    },
    {
      "epoch": 0.5121094758585076,
      "grad_norm": 2.0469284057617188,
      "learning_rate": 4.887347441888848e-06,
      "loss": 0.0346,
      "step": 9917
    },
    {
      "epoch": 0.5121611154144075,
      "grad_norm": 4.559978008270264,
      "learning_rate": 4.886528290732178e-06,
      "loss": 0.0591,
      "step": 9918
    },
    {
      "epoch": 0.5122127549703073,
      "grad_norm": 9.567520141601562,
      "learning_rate": 4.885709142622688e-06,
      "loss": 0.0878,
      "step": 9919
    },
    {
      "epoch": 0.5122643945262071,
      "grad_norm": 13.123839378356934,
      "learning_rate": 4.884889997582371e-06,
      "loss": 0.0719,
      "step": 9920
    },
    {
      "epoch": 0.5123160340821069,
      "grad_norm": 3.2849814891815186,
      "learning_rate": 4.88407085563323e-06,
      "loss": 0.0808,
      "step": 9921
    },
    {
      "epoch": 0.5123676736380067,
      "grad_norm": 2.854323625564575,
      "learning_rate": 4.883251716797256e-06,
      "loss": 0.0485,
      "step": 9922
    },
    {
      "epoch": 0.5124193131939065,
      "grad_norm": 8.291495323181152,
      "learning_rate": 4.88243258109645e-06,
      "loss": 0.0755,
      "step": 9923
    },
    {
      "epoch": 0.5124709527498064,
      "grad_norm": 2.859118938446045,
      "learning_rate": 4.881613448552809e-06,
      "loss": 0.0567,
      "step": 9924
    },
    {
      "epoch": 0.5125225923057062,
      "grad_norm": 3.7928740978240967,
      "learning_rate": 4.8807943191883295e-06,
      "loss": 0.0544,
      "step": 9925
    },
    {
      "epoch": 0.512574231861606,
      "grad_norm": 1.983853816986084,
      "learning_rate": 4.879975193025008e-06,
      "loss": 0.0403,
      "step": 9926
    },
    {
      "epoch": 0.5126258714175058,
      "grad_norm": 2.6937122344970703,
      "learning_rate": 4.87915607008484e-06,
      "loss": 0.0498,
      "step": 9927
    },
    {
      "epoch": 0.5126775109734056,
      "grad_norm": 3.6207048892974854,
      "learning_rate": 4.878336950389829e-06,
      "loss": 0.0531,
      "step": 9928
    },
    {
      "epoch": 0.5127291505293055,
      "grad_norm": 2.740993022918701,
      "learning_rate": 4.877517833961962e-06,
      "loss": 0.0495,
      "step": 9929
    },
    {
      "epoch": 0.5127807900852053,
      "grad_norm": 7.045472145080566,
      "learning_rate": 4.876698720823241e-06,
      "loss": 0.0667,
      "step": 9930
    },
    {
      "epoch": 0.5128324296411051,
      "grad_norm": 9.664972305297852,
      "learning_rate": 4.875879610995663e-06,
      "loss": 0.0729,
      "step": 9931
    },
    {
      "epoch": 0.5128840691970049,
      "grad_norm": 2.5543813705444336,
      "learning_rate": 4.875060504501224e-06,
      "loss": 0.0474,
      "step": 9932
    },
    {
      "epoch": 0.5129357087529047,
      "grad_norm": 5.7463860511779785,
      "learning_rate": 4.874241401361919e-06,
      "loss": 0.0739,
      "step": 9933
    },
    {
      "epoch": 0.5129873483088045,
      "grad_norm": 4.735226631164551,
      "learning_rate": 4.8734223015997435e-06,
      "loss": 0.0629,
      "step": 9934
    },
    {
      "epoch": 0.5130389878647044,
      "grad_norm": 4.848759174346924,
      "learning_rate": 4.872603205236696e-06,
      "loss": 0.063,
      "step": 9935
    },
    {
      "epoch": 0.5130906274206042,
      "grad_norm": 3.634347915649414,
      "learning_rate": 4.871784112294773e-06,
      "loss": 0.0533,
      "step": 9936
    },
    {
      "epoch": 0.513142266976504,
      "grad_norm": 3.265815496444702,
      "learning_rate": 4.870965022795969e-06,
      "loss": 0.0419,
      "step": 9937
    },
    {
      "epoch": 0.5131939065324038,
      "grad_norm": 8.714712142944336,
      "learning_rate": 4.870145936762279e-06,
      "loss": 0.079,
      "step": 9938
    },
    {
      "epoch": 0.5132455460883036,
      "grad_norm": 2.142246723175049,
      "learning_rate": 4.869326854215701e-06,
      "loss": 0.0276,
      "step": 9939
    },
    {
      "epoch": 0.5132971856442035,
      "grad_norm": 8.430251121520996,
      "learning_rate": 4.8685077751782284e-06,
      "loss": 0.0526,
      "step": 9940
    },
    {
      "epoch": 0.5133488252001033,
      "grad_norm": 4.101236343383789,
      "learning_rate": 4.867688699671857e-06,
      "loss": 0.0304,
      "step": 9941
    },
    {
      "epoch": 0.5134004647560031,
      "grad_norm": 12.556289672851562,
      "learning_rate": 4.866869627718585e-06,
      "loss": 0.0945,
      "step": 9942
    },
    {
      "epoch": 0.5134521043119029,
      "grad_norm": 6.183897018432617,
      "learning_rate": 4.866050559340406e-06,
      "loss": 0.0641,
      "step": 9943
    },
    {
      "epoch": 0.5135037438678027,
      "grad_norm": 1.5531456470489502,
      "learning_rate": 4.865231494559315e-06,
      "loss": 0.044,
      "step": 9944
    },
    {
      "epoch": 0.5135553834237025,
      "grad_norm": 3.328003168106079,
      "learning_rate": 4.864412433397308e-06,
      "loss": 0.0411,
      "step": 9945
    },
    {
      "epoch": 0.5136070229796024,
      "grad_norm": 5.587536334991455,
      "learning_rate": 4.863593375876377e-06,
      "loss": 0.0559,
      "step": 9946
    },
    {
      "epoch": 0.5136586625355022,
      "grad_norm": 1.9512587785720825,
      "learning_rate": 4.8627743220185255e-06,
      "loss": 0.0288,
      "step": 9947
    },
    {
      "epoch": 0.513710302091402,
      "grad_norm": 3.5863301753997803,
      "learning_rate": 4.861955271845738e-06,
      "loss": 0.0659,
      "step": 9948
    },
    {
      "epoch": 0.5137619416473018,
      "grad_norm": 7.679018020629883,
      "learning_rate": 4.861136225380016e-06,
      "loss": 0.0461,
      "step": 9949
    },
    {
      "epoch": 0.5138135812032016,
      "grad_norm": 8.862351417541504,
      "learning_rate": 4.860317182643353e-06,
      "loss": 0.0748,
      "step": 9950
    },
    {
      "epoch": 0.5138652207591015,
      "grad_norm": 7.932219505310059,
      "learning_rate": 4.859498143657742e-06,
      "loss": 0.0847,
      "step": 9951
    },
    {
      "epoch": 0.5139168603150013,
      "grad_norm": 3.00929594039917,
      "learning_rate": 4.858679108445178e-06,
      "loss": 0.0639,
      "step": 9952
    },
    {
      "epoch": 0.5139684998709011,
      "grad_norm": 23.67083168029785,
      "learning_rate": 4.857860077027655e-06,
      "loss": 0.3173,
      "step": 9953
    },
    {
      "epoch": 0.5140201394268009,
      "grad_norm": 8.901535034179688,
      "learning_rate": 4.857041049427172e-06,
      "loss": 0.0855,
      "step": 9954
    },
    {
      "epoch": 0.5140717789827007,
      "grad_norm": 2.7845776081085205,
      "learning_rate": 4.856222025665715e-06,
      "loss": 0.0628,
      "step": 9955
    },
    {
      "epoch": 0.5141234185386006,
      "grad_norm": 5.8236212730407715,
      "learning_rate": 4.855403005765285e-06,
      "loss": 0.0511,
      "step": 9956
    },
    {
      "epoch": 0.5141750580945004,
      "grad_norm": 4.833427429199219,
      "learning_rate": 4.854583989747874e-06,
      "loss": 0.0478,
      "step": 9957
    },
    {
      "epoch": 0.5142266976504002,
      "grad_norm": 2.825819253921509,
      "learning_rate": 4.853764977635476e-06,
      "loss": 0.05,
      "step": 9958
    },
    {
      "epoch": 0.5142783372063,
      "grad_norm": 1.5792558193206787,
      "learning_rate": 4.852945969450083e-06,
      "loss": 0.0372,
      "step": 9959
    },
    {
      "epoch": 0.5143299767621998,
      "grad_norm": 14.247705459594727,
      "learning_rate": 4.852126965213689e-06,
      "loss": 0.12,
      "step": 9960
    },
    {
      "epoch": 0.5143816163180996,
      "grad_norm": 3.9812824726104736,
      "learning_rate": 4.851307964948292e-06,
      "loss": 0.0687,
      "step": 9961
    },
    {
      "epoch": 0.5144332558739995,
      "grad_norm": 5.793576717376709,
      "learning_rate": 4.850488968675879e-06,
      "loss": 0.086,
      "step": 9962
    },
    {
      "epoch": 0.5144848954298993,
      "grad_norm": 7.275843143463135,
      "learning_rate": 4.849669976418448e-06,
      "loss": 0.0512,
      "step": 9963
    },
    {
      "epoch": 0.5145365349857991,
      "grad_norm": 2.4833343029022217,
      "learning_rate": 4.8488509881979905e-06,
      "loss": 0.0547,
      "step": 9964
    },
    {
      "epoch": 0.5145881745416989,
      "grad_norm": 2.215620517730713,
      "learning_rate": 4.848032004036502e-06,
      "loss": 0.0368,
      "step": 9965
    },
    {
      "epoch": 0.5146398140975987,
      "grad_norm": 4.09289026260376,
      "learning_rate": 4.847213023955971e-06,
      "loss": 0.05,
      "step": 9966
    },
    {
      "epoch": 0.5146914536534986,
      "grad_norm": 1.729902982711792,
      "learning_rate": 4.8463940479783925e-06,
      "loss": 0.0498,
      "step": 9967
    },
    {
      "epoch": 0.5147430932093984,
      "grad_norm": 3.1916863918304443,
      "learning_rate": 4.845575076125761e-06,
      "loss": 0.0376,
      "step": 9968
    },
    {
      "epoch": 0.5147947327652982,
      "grad_norm": 3.606356143951416,
      "learning_rate": 4.844756108420069e-06,
      "loss": 0.0298,
      "step": 9969
    },
    {
      "epoch": 0.514846372321198,
      "grad_norm": 5.335087299346924,
      "learning_rate": 4.8439371448833075e-06,
      "loss": 0.0758,
      "step": 9970
    },
    {
      "epoch": 0.5148980118770978,
      "grad_norm": 4.686010360717773,
      "learning_rate": 4.84311818553747e-06,
      "loss": 0.083,
      "step": 9971
    },
    {
      "epoch": 0.5149496514329976,
      "grad_norm": 8.21250057220459,
      "learning_rate": 4.842299230404549e-06,
      "loss": 0.0951,
      "step": 9972
    },
    {
      "epoch": 0.5150012909888975,
      "grad_norm": 3.005338430404663,
      "learning_rate": 4.841480279506536e-06,
      "loss": 0.0401,
      "step": 9973
    },
    {
      "epoch": 0.5150529305447973,
      "grad_norm": 22.050025939941406,
      "learning_rate": 4.840661332865422e-06,
      "loss": 0.1075,
      "step": 9974
    },
    {
      "epoch": 0.5151045701006971,
      "grad_norm": 2.2289867401123047,
      "learning_rate": 4.839842390503202e-06,
      "loss": 0.03,
      "step": 9975
    },
    {
      "epoch": 0.5151562096565969,
      "grad_norm": 15.073965072631836,
      "learning_rate": 4.839023452441868e-06,
      "loss": 0.0772,
      "step": 9976
    },
    {
      "epoch": 0.5152078492124967,
      "grad_norm": 3.7592275142669678,
      "learning_rate": 4.8382045187034085e-06,
      "loss": 0.041,
      "step": 9977
    },
    {
      "epoch": 0.5152594887683966,
      "grad_norm": 1.9769407510757446,
      "learning_rate": 4.837385589309817e-06,
      "loss": 0.019,
      "step": 9978
    },
    {
      "epoch": 0.5153111283242964,
      "grad_norm": 3.3125860691070557,
      "learning_rate": 4.836566664283085e-06,
      "loss": 0.0428,
      "step": 9979
    },
    {
      "epoch": 0.5153627678801962,
      "grad_norm": 2.476170778274536,
      "learning_rate": 4.835747743645207e-06,
      "loss": 0.0556,
      "step": 9980
    },
    {
      "epoch": 0.515414407436096,
      "grad_norm": 2.657937526702881,
      "learning_rate": 4.834928827418168e-06,
      "loss": 0.0175,
      "step": 9981
    },
    {
      "epoch": 0.5154660469919958,
      "grad_norm": 5.48227071762085,
      "learning_rate": 4.834109915623965e-06,
      "loss": 0.0544,
      "step": 9982
    },
    {
      "epoch": 0.5155176865478956,
      "grad_norm": 3.9999513626098633,
      "learning_rate": 4.8332910082845865e-06,
      "loss": 0.0889,
      "step": 9983
    },
    {
      "epoch": 0.5155693261037955,
      "grad_norm": 3.774740219116211,
      "learning_rate": 4.832472105422024e-06,
      "loss": 0.0635,
      "step": 9984
    },
    {
      "epoch": 0.5156209656596953,
      "grad_norm": 3.396251916885376,
      "learning_rate": 4.8316532070582674e-06,
      "loss": 0.019,
      "step": 9985
    },
    {
      "epoch": 0.5156726052155951,
      "grad_norm": 11.968196868896484,
      "learning_rate": 4.830834313215308e-06,
      "loss": 0.0819,
      "step": 9986
    },
    {
      "epoch": 0.5157242447714949,
      "grad_norm": 7.757645130157471,
      "learning_rate": 4.8300154239151395e-06,
      "loss": 0.0662,
      "step": 9987
    },
    {
      "epoch": 0.5157758843273947,
      "grad_norm": 2.229419469833374,
      "learning_rate": 4.829196539179747e-06,
      "loss": 0.0476,
      "step": 9988
    },
    {
      "epoch": 0.5158275238832946,
      "grad_norm": 4.747222423553467,
      "learning_rate": 4.828377659031124e-06,
      "loss": 0.0513,
      "step": 9989
    },
    {
      "epoch": 0.5158791634391944,
      "grad_norm": 3.6231689453125,
      "learning_rate": 4.827558783491261e-06,
      "loss": 0.0723,
      "step": 9990
    },
    {
      "epoch": 0.5159308029950942,
      "grad_norm": 5.309370040893555,
      "learning_rate": 4.826739912582148e-06,
      "loss": 0.065,
      "step": 9991
    },
    {
      "epoch": 0.515982442550994,
      "grad_norm": 3.0315465927124023,
      "learning_rate": 4.825921046325773e-06,
      "loss": 0.0668,
      "step": 9992
    },
    {
      "epoch": 0.5160340821068938,
      "grad_norm": 3.1238784790039062,
      "learning_rate": 4.825102184744127e-06,
      "loss": 0.0229,
      "step": 9993
    },
    {
      "epoch": 0.5160857216627937,
      "grad_norm": 4.227876663208008,
      "learning_rate": 4.824283327859204e-06,
      "loss": 0.1099,
      "step": 9994
    },
    {
      "epoch": 0.5161373612186935,
      "grad_norm": 8.241179466247559,
      "learning_rate": 4.823464475692985e-06,
      "loss": 0.0931,
      "step": 9995
    },
    {
      "epoch": 0.5161890007745933,
      "grad_norm": 3.6492807865142822,
      "learning_rate": 4.822645628267467e-06,
      "loss": 0.0322,
      "step": 9996
    },
    {
      "epoch": 0.5162406403304931,
      "grad_norm": 3.654799461364746,
      "learning_rate": 4.821826785604636e-06,
      "loss": 0.0544,
      "step": 9997
    },
    {
      "epoch": 0.5162922798863929,
      "grad_norm": 1.6309815645217896,
      "learning_rate": 4.8210079477264834e-06,
      "loss": 0.0311,
      "step": 9998
    },
    {
      "epoch": 0.5163439194422927,
      "grad_norm": 12.882993698120117,
      "learning_rate": 4.8201891146549955e-06,
      "loss": 0.0451,
      "step": 9999
    },
    {
      "epoch": 0.5163955589981926,
      "grad_norm": 8.956052780151367,
      "learning_rate": 4.819370286412162e-06,
      "loss": 0.0491,
      "step": 10000
    },
    {
      "epoch": 0.5164471985540924,
      "grad_norm": 22.69967269897461,
      "learning_rate": 4.818551463019973e-06,
      "loss": 0.3643,
      "step": 10001
    },
    {
      "epoch": 0.5164988381099923,
      "grad_norm": 1.3976207971572876,
      "learning_rate": 4.8177326445004185e-06,
      "loss": 0.0166,
      "step": 10002
    },
    {
      "epoch": 0.5165504776658921,
      "grad_norm": 3.550302028656006,
      "learning_rate": 4.816913830875484e-06,
      "loss": 0.0726,
      "step": 10003
    },
    {
      "epoch": 0.516602117221792,
      "grad_norm": 4.633478164672852,
      "learning_rate": 4.81609502216716e-06,
      "loss": 0.0837,
      "step": 10004
    },
    {
      "epoch": 0.5166537567776918,
      "grad_norm": 10.677408218383789,
      "learning_rate": 4.815276218397434e-06,
      "loss": 0.1601,
      "step": 10005
    },
    {
      "epoch": 0.5167053963335916,
      "grad_norm": 9.301130294799805,
      "learning_rate": 4.814457419588294e-06,
      "loss": 0.1343,
      "step": 10006
    },
    {
      "epoch": 0.5167570358894914,
      "grad_norm": 2.7378950119018555,
      "learning_rate": 4.813638625761728e-06,
      "loss": 0.0526,
      "step": 10007
    },
    {
      "epoch": 0.5168086754453912,
      "grad_norm": 3.3656983375549316,
      "learning_rate": 4.812819836939725e-06,
      "loss": 0.0797,
      "step": 10008
    },
    {
      "epoch": 0.516860315001291,
      "grad_norm": 2.7463219165802,
      "learning_rate": 4.812001053144274e-06,
      "loss": 0.0489,
      "step": 10009
    },
    {
      "epoch": 0.5169119545571909,
      "grad_norm": 8.497456550598145,
      "learning_rate": 4.811182274397359e-06,
      "loss": 0.1307,
      "step": 10010
    },
    {
      "epoch": 0.5169635941130907,
      "grad_norm": 11.07149887084961,
      "learning_rate": 4.810363500720969e-06,
      "loss": 0.0823,
      "step": 10011
    },
    {
      "epoch": 0.5170152336689905,
      "grad_norm": 5.366752624511719,
      "learning_rate": 4.809544732137092e-06,
      "loss": 0.0688,
      "step": 10012
    },
    {
      "epoch": 0.5170668732248903,
      "grad_norm": 3.0878491401672363,
      "learning_rate": 4.808725968667718e-06,
      "loss": 0.0356,
      "step": 10013
    },
    {
      "epoch": 0.5171185127807901,
      "grad_norm": 4.797057628631592,
      "learning_rate": 4.807907210334828e-06,
      "loss": 0.0769,
      "step": 10014
    },
    {
      "epoch": 0.51717015233669,
      "grad_norm": 7.66999626159668,
      "learning_rate": 4.807088457160414e-06,
      "loss": 0.1175,
      "step": 10015
    },
    {
      "epoch": 0.5172217918925898,
      "grad_norm": 8.289076805114746,
      "learning_rate": 4.806269709166462e-06,
      "loss": 0.0752,
      "step": 10016
    },
    {
      "epoch": 0.5172734314484896,
      "grad_norm": 9.042210578918457,
      "learning_rate": 4.805450966374956e-06,
      "loss": 0.0845,
      "step": 10017
    },
    {
      "epoch": 0.5173250710043894,
      "grad_norm": 3.0119051933288574,
      "learning_rate": 4.804632228807886e-06,
      "loss": 0.0734,
      "step": 10018
    },
    {
      "epoch": 0.5173767105602892,
      "grad_norm": 10.197592735290527,
      "learning_rate": 4.803813496487234e-06,
      "loss": 0.0759,
      "step": 10019
    },
    {
      "epoch": 0.517428350116189,
      "grad_norm": 3.7813258171081543,
      "learning_rate": 4.802994769434994e-06,
      "loss": 0.0564,
      "step": 10020
    },
    {
      "epoch": 0.5174799896720889,
      "grad_norm": 1.8862216472625732,
      "learning_rate": 4.802176047673143e-06,
      "loss": 0.0384,
      "step": 10021
    },
    {
      "epoch": 0.5175316292279887,
      "grad_norm": 6.029797077178955,
      "learning_rate": 4.801357331223673e-06,
      "loss": 0.0617,
      "step": 10022
    },
    {
      "epoch": 0.5175832687838885,
      "grad_norm": 5.388339996337891,
      "learning_rate": 4.800538620108569e-06,
      "loss": 0.0847,
      "step": 10023
    },
    {
      "epoch": 0.5176349083397883,
      "grad_norm": 9.161788940429688,
      "learning_rate": 4.799719914349816e-06,
      "loss": 0.0629,
      "step": 10024
    },
    {
      "epoch": 0.5176865478956881,
      "grad_norm": 4.5153069496154785,
      "learning_rate": 4.7989012139693995e-06,
      "loss": 0.0275,
      "step": 10025
    },
    {
      "epoch": 0.517738187451588,
      "grad_norm": 4.9635844230651855,
      "learning_rate": 4.798082518989303e-06,
      "loss": 0.0436,
      "step": 10026
    },
    {
      "epoch": 0.5177898270074878,
      "grad_norm": 4.532450199127197,
      "learning_rate": 4.797263829431517e-06,
      "loss": 0.0794,
      "step": 10027
    },
    {
      "epoch": 0.5178414665633876,
      "grad_norm": 2.041635036468506,
      "learning_rate": 4.796445145318022e-06,
      "loss": 0.0318,
      "step": 10028
    },
    {
      "epoch": 0.5178931061192874,
      "grad_norm": 3.5275049209594727,
      "learning_rate": 4.795626466670804e-06,
      "loss": 0.0729,
      "step": 10029
    },
    {
      "epoch": 0.5179447456751872,
      "grad_norm": 2.160052537918091,
      "learning_rate": 4.7948077935118485e-06,
      "loss": 0.0412,
      "step": 10030
    },
    {
      "epoch": 0.517996385231087,
      "grad_norm": 7.171811580657959,
      "learning_rate": 4.793989125863141e-06,
      "loss": 0.1051,
      "step": 10031
    },
    {
      "epoch": 0.5180480247869869,
      "grad_norm": 7.783716201782227,
      "learning_rate": 4.793170463746665e-06,
      "loss": 0.0751,
      "step": 10032
    },
    {
      "epoch": 0.5180996643428867,
      "grad_norm": 10.017923355102539,
      "learning_rate": 4.7923518071844036e-06,
      "loss": 0.1252,
      "step": 10033
    },
    {
      "epoch": 0.5181513038987865,
      "grad_norm": 4.781060695648193,
      "learning_rate": 4.791533156198343e-06,
      "loss": 0.0694,
      "step": 10034
    },
    {
      "epoch": 0.5182029434546863,
      "grad_norm": 2.727421522140503,
      "learning_rate": 4.790714510810468e-06,
      "loss": 0.0342,
      "step": 10035
    },
    {
      "epoch": 0.5182545830105861,
      "grad_norm": 2.7712111473083496,
      "learning_rate": 4.78989587104276e-06,
      "loss": 0.0342,
      "step": 10036
    },
    {
      "epoch": 0.518306222566486,
      "grad_norm": 2.532686710357666,
      "learning_rate": 4.789077236917206e-06,
      "loss": 0.0442,
      "step": 10037
    },
    {
      "epoch": 0.5183578621223858,
      "grad_norm": 6.035699367523193,
      "learning_rate": 4.788258608455788e-06,
      "loss": 0.073,
      "step": 10038
    },
    {
      "epoch": 0.5184095016782856,
      "grad_norm": 2.3527848720550537,
      "learning_rate": 4.787439985680487e-06,
      "loss": 0.0237,
      "step": 10039
    },
    {
      "epoch": 0.5184611412341854,
      "grad_norm": 3.3876187801361084,
      "learning_rate": 4.786621368613289e-06,
      "loss": 0.0329,
      "step": 10040
    },
    {
      "epoch": 0.5185127807900852,
      "grad_norm": 3.7658982276916504,
      "learning_rate": 4.785802757276177e-06,
      "loss": 0.0164,
      "step": 10041
    },
    {
      "epoch": 0.518564420345985,
      "grad_norm": 6.924946308135986,
      "learning_rate": 4.784984151691136e-06,
      "loss": 0.0473,
      "step": 10042
    },
    {
      "epoch": 0.5186160599018849,
      "grad_norm": 6.011969089508057,
      "learning_rate": 4.784165551880146e-06,
      "loss": 0.046,
      "step": 10043
    },
    {
      "epoch": 0.5186676994577847,
      "grad_norm": 5.990752220153809,
      "learning_rate": 4.7833469578651895e-06,
      "loss": 0.0415,
      "step": 10044
    },
    {
      "epoch": 0.5187193390136845,
      "grad_norm": 9.007098197937012,
      "learning_rate": 4.78252836966825e-06,
      "loss": 0.0479,
      "step": 10045
    },
    {
      "epoch": 0.5187709785695843,
      "grad_norm": 4.326048374176025,
      "learning_rate": 4.781709787311313e-06,
      "loss": 0.0668,
      "step": 10046
    },
    {
      "epoch": 0.5188226181254841,
      "grad_norm": 14.306863784790039,
      "learning_rate": 4.7808912108163555e-06,
      "loss": 0.1543,
      "step": 10047
    },
    {
      "epoch": 0.518874257681384,
      "grad_norm": 2.7771739959716797,
      "learning_rate": 4.780072640205362e-06,
      "loss": 0.0334,
      "step": 10048
    },
    {
      "epoch": 0.5189258972372838,
      "grad_norm": 3.1777431964874268,
      "learning_rate": 4.779254075500316e-06,
      "loss": 0.0567,
      "step": 10049
    },
    {
      "epoch": 0.5189775367931836,
      "grad_norm": 3.7604904174804688,
      "learning_rate": 4.778435516723196e-06,
      "loss": 0.064,
      "step": 10050
    },
    {
      "epoch": 0.5190291763490834,
      "grad_norm": 3.420616626739502,
      "learning_rate": 4.777616963895987e-06,
      "loss": 0.0733,
      "step": 10051
    },
    {
      "epoch": 0.5190808159049832,
      "grad_norm": 9.610817909240723,
      "learning_rate": 4.776798417040666e-06,
      "loss": 0.1052,
      "step": 10052
    },
    {
      "epoch": 0.519132455460883,
      "grad_norm": 4.892521858215332,
      "learning_rate": 4.775979876179222e-06,
      "loss": 0.0202,
      "step": 10053
    },
    {
      "epoch": 0.5191840950167829,
      "grad_norm": 3.8016834259033203,
      "learning_rate": 4.7751613413336275e-06,
      "loss": 0.0336,
      "step": 10054
    },
    {
      "epoch": 0.5192357345726827,
      "grad_norm": 5.123043060302734,
      "learning_rate": 4.774342812525869e-06,
      "loss": 0.0821,
      "step": 10055
    },
    {
      "epoch": 0.5192873741285825,
      "grad_norm": 7.248293876647949,
      "learning_rate": 4.773524289777925e-06,
      "loss": 0.0814,
      "step": 10056
    },
    {
      "epoch": 0.5193390136844823,
      "grad_norm": 2.451361656188965,
      "learning_rate": 4.7727057731117784e-06,
      "loss": 0.0623,
      "step": 10057
    },
    {
      "epoch": 0.5193906532403821,
      "grad_norm": 2.9833297729492188,
      "learning_rate": 4.7718872625494065e-06,
      "loss": 0.0216,
      "step": 10058
    },
    {
      "epoch": 0.519442292796282,
      "grad_norm": 4.0794901847839355,
      "learning_rate": 4.77106875811279e-06,
      "loss": 0.0447,
      "step": 10059
    },
    {
      "epoch": 0.5194939323521818,
      "grad_norm": 2.551821231842041,
      "learning_rate": 4.770250259823915e-06,
      "loss": 0.0129,
      "step": 10060
    },
    {
      "epoch": 0.5195455719080816,
      "grad_norm": 2.309805154800415,
      "learning_rate": 4.769431767704753e-06,
      "loss": 0.0245,
      "step": 10061
    },
    {
      "epoch": 0.5195972114639814,
      "grad_norm": 3.2251646518707275,
      "learning_rate": 4.768613281777289e-06,
      "loss": 0.0415,
      "step": 10062
    },
    {
      "epoch": 0.5196488510198812,
      "grad_norm": 3.0117695331573486,
      "learning_rate": 4.767794802063501e-06,
      "loss": 0.0568,
      "step": 10063
    },
    {
      "epoch": 0.519700490575781,
      "grad_norm": 4.39605188369751,
      "learning_rate": 4.76697632858537e-06,
      "loss": 0.0882,
      "step": 10064
    },
    {
      "epoch": 0.5197521301316809,
      "grad_norm": 8.358922958374023,
      "learning_rate": 4.766157861364874e-06,
      "loss": 0.1108,
      "step": 10065
    },
    {
      "epoch": 0.5198037696875807,
      "grad_norm": 12.422364234924316,
      "learning_rate": 4.765339400423991e-06,
      "loss": 0.143,
      "step": 10066
    },
    {
      "epoch": 0.5198554092434805,
      "grad_norm": 9.19175910949707,
      "learning_rate": 4.764520945784702e-06,
      "loss": 0.092,
      "step": 10067
    },
    {
      "epoch": 0.5199070487993803,
      "grad_norm": 9.108630180358887,
      "learning_rate": 4.763702497468988e-06,
      "loss": 0.082,
      "step": 10068
    },
    {
      "epoch": 0.5199586883552801,
      "grad_norm": 4.0271711349487305,
      "learning_rate": 4.762884055498823e-06,
      "loss": 0.0508,
      "step": 10069
    },
    {
      "epoch": 0.52001032791118,
      "grad_norm": 9.27385139465332,
      "learning_rate": 4.7620656198961876e-06,
      "loss": 0.0684,
      "step": 10070
    },
    {
      "epoch": 0.5200619674670798,
      "grad_norm": 4.570675373077393,
      "learning_rate": 4.761247190683061e-06,
      "loss": 0.0675,
      "step": 10071
    },
    {
      "epoch": 0.5201136070229796,
      "grad_norm": 4.1162190437316895,
      "learning_rate": 4.76042876788142e-06,
      "loss": 0.0753,
      "step": 10072
    },
    {
      "epoch": 0.5201652465788794,
      "grad_norm": 4.851659297943115,
      "learning_rate": 4.759610351513241e-06,
      "loss": 0.0642,
      "step": 10073
    },
    {
      "epoch": 0.5202168861347792,
      "grad_norm": 10.258072853088379,
      "learning_rate": 4.758791941600505e-06,
      "loss": 0.1339,
      "step": 10074
    },
    {
      "epoch": 0.520268525690679,
      "grad_norm": 5.086020469665527,
      "learning_rate": 4.7579735381651904e-06,
      "loss": 0.0603,
      "step": 10075
    },
    {
      "epoch": 0.5203201652465789,
      "grad_norm": 2.0318615436553955,
      "learning_rate": 4.757155141229271e-06,
      "loss": 0.0378,
      "step": 10076
    },
    {
      "epoch": 0.5203718048024787,
      "grad_norm": 2.325406789779663,
      "learning_rate": 4.7563367508147265e-06,
      "loss": 0.032,
      "step": 10077
    },
    {
      "epoch": 0.5204234443583785,
      "grad_norm": 3.6107592582702637,
      "learning_rate": 4.7555183669435316e-06,
      "loss": 0.0782,
      "step": 10078
    },
    {
      "epoch": 0.5204750839142783,
      "grad_norm": 5.8917341232299805,
      "learning_rate": 4.754699989637669e-06,
      "loss": 0.0317,
      "step": 10079
    },
    {
      "epoch": 0.5205267234701781,
      "grad_norm": 4.275318145751953,
      "learning_rate": 4.7538816189191065e-06,
      "loss": 0.0557,
      "step": 10080
    },
    {
      "epoch": 0.520578363026078,
      "grad_norm": 9.2618989944458,
      "learning_rate": 4.753063254809828e-06,
      "loss": 0.0512,
      "step": 10081
    },
    {
      "epoch": 0.5206300025819778,
      "grad_norm": 5.132842063903809,
      "learning_rate": 4.752244897331809e-06,
      "loss": 0.0626,
      "step": 10082
    },
    {
      "epoch": 0.5206816421378776,
      "grad_norm": 2.8242781162261963,
      "learning_rate": 4.751426546507022e-06,
      "loss": 0.0212,
      "step": 10083
    },
    {
      "epoch": 0.5207332816937774,
      "grad_norm": 6.339079856872559,
      "learning_rate": 4.750608202357447e-06,
      "loss": 0.0669,
      "step": 10084
    },
    {
      "epoch": 0.5207849212496772,
      "grad_norm": 7.304733753204346,
      "learning_rate": 4.749789864905056e-06,
      "loss": 0.0508,
      "step": 10085
    },
    {
      "epoch": 0.520836560805577,
      "grad_norm": 9.981334686279297,
      "learning_rate": 4.748971534171831e-06,
      "loss": 0.0937,
      "step": 10086
    },
    {
      "epoch": 0.5208882003614769,
      "grad_norm": 7.214250564575195,
      "learning_rate": 4.748153210179739e-06,
      "loss": 0.1052,
      "step": 10087
    },
    {
      "epoch": 0.5209398399173767,
      "grad_norm": 3.329162836074829,
      "learning_rate": 4.747334892950762e-06,
      "loss": 0.0277,
      "step": 10088
    },
    {
      "epoch": 0.5209914794732765,
      "grad_norm": 11.294472694396973,
      "learning_rate": 4.746516582506872e-06,
      "loss": 0.131,
      "step": 10089
    },
    {
      "epoch": 0.5210431190291763,
      "grad_norm": 3.778944492340088,
      "learning_rate": 4.745698278870046e-06,
      "loss": 0.0464,
      "step": 10090
    },
    {
      "epoch": 0.5210947585850761,
      "grad_norm": 27.395627975463867,
      "learning_rate": 4.7448799820622575e-06,
      "loss": 0.1703,
      "step": 10091
    },
    {
      "epoch": 0.521146398140976,
      "grad_norm": 0.9440184235572815,
      "learning_rate": 4.74406169210548e-06,
      "loss": 0.0114,
      "step": 10092
    },
    {
      "epoch": 0.5211980376968758,
      "grad_norm": 8.733306884765625,
      "learning_rate": 4.743243409021692e-06,
      "loss": 0.0658,
      "step": 10093
    },
    {
      "epoch": 0.5212496772527756,
      "grad_norm": 5.495108127593994,
      "learning_rate": 4.742425132832862e-06,
      "loss": 0.0665,
      "step": 10094
    },
    {
      "epoch": 0.5213013168086754,
      "grad_norm": 5.163908004760742,
      "learning_rate": 4.741606863560968e-06,
      "loss": 0.0636,
      "step": 10095
    },
    {
      "epoch": 0.5213529563645752,
      "grad_norm": 2.930980920791626,
      "learning_rate": 4.740788601227982e-06,
      "loss": 0.0484,
      "step": 10096
    },
    {
      "epoch": 0.5214045959204751,
      "grad_norm": 12.521047592163086,
      "learning_rate": 4.739970345855881e-06,
      "loss": 0.0813,
      "step": 10097
    },
    {
      "epoch": 0.5214562354763749,
      "grad_norm": 8.810359001159668,
      "learning_rate": 4.739152097466634e-06,
      "loss": 0.0776,
      "step": 10098
    },
    {
      "epoch": 0.5215078750322747,
      "grad_norm": 5.951861381530762,
      "learning_rate": 4.738333856082216e-06,
      "loss": 0.0848,
      "step": 10099
    },
    {
      "epoch": 0.5215595145881745,
      "grad_norm": 5.492519378662109,
      "learning_rate": 4.737515621724603e-06,
      "loss": 0.0581,
      "step": 10100
    },
    {
      "epoch": 0.5216111541440743,
      "grad_norm": 5.00434684753418,
      "learning_rate": 4.736697394415763e-06,
      "loss": 0.048,
      "step": 10101
    },
    {
      "epoch": 0.5216627936999741,
      "grad_norm": 4.176885604858398,
      "learning_rate": 4.735879174177672e-06,
      "loss": 0.0573,
      "step": 10102
    },
    {
      "epoch": 0.521714433255874,
      "grad_norm": 5.134427070617676,
      "learning_rate": 4.7350609610323015e-06,
      "loss": 0.0755,
      "step": 10103
    },
    {
      "epoch": 0.5217660728117738,
      "grad_norm": 1.1416176557540894,
      "learning_rate": 4.734242755001624e-06,
      "loss": 0.0107,
      "step": 10104
    },
    {
      "epoch": 0.5218177123676736,
      "grad_norm": 2.5321643352508545,
      "learning_rate": 4.733424556107612e-06,
      "loss": 0.0313,
      "step": 10105
    },
    {
      "epoch": 0.5218693519235734,
      "grad_norm": 11.991033554077148,
      "learning_rate": 4.732606364372236e-06,
      "loss": 0.1734,
      "step": 10106
    },
    {
      "epoch": 0.5219209914794732,
      "grad_norm": 5.376570701599121,
      "learning_rate": 4.7317881798174695e-06,
      "loss": 0.0996,
      "step": 10107
    },
    {
      "epoch": 0.5219726310353731,
      "grad_norm": 3.3230676651000977,
      "learning_rate": 4.730970002465285e-06,
      "loss": 0.0247,
      "step": 10108
    },
    {
      "epoch": 0.5220242705912729,
      "grad_norm": 4.968777656555176,
      "learning_rate": 4.730151832337652e-06,
      "loss": 0.025,
      "step": 10109
    },
    {
      "epoch": 0.5220759101471727,
      "grad_norm": 4.62604284286499,
      "learning_rate": 4.72933366945654e-06,
      "loss": 0.0593,
      "step": 10110
    },
    {
      "epoch": 0.5221275497030725,
      "grad_norm": 1.1669585704803467,
      "learning_rate": 4.7285155138439256e-06,
      "loss": 0.0154,
      "step": 10111
    },
    {
      "epoch": 0.5221791892589723,
      "grad_norm": 14.090775489807129,
      "learning_rate": 4.727697365521774e-06,
      "loss": 0.1517,
      "step": 10112
    },
    {
      "epoch": 0.5222308288148722,
      "grad_norm": 2.811998128890991,
      "learning_rate": 4.7268792245120564e-06,
      "loss": 0.048,
      "step": 10113
    },
    {
      "epoch": 0.522282468370772,
      "grad_norm": 3.785583257675171,
      "learning_rate": 4.726061090836746e-06,
      "loss": 0.0659,
      "step": 10114
    },
    {
      "epoch": 0.5223341079266718,
      "grad_norm": 8.776924133300781,
      "learning_rate": 4.725242964517815e-06,
      "loss": 0.0587,
      "step": 10115
    },
    {
      "epoch": 0.5223857474825716,
      "grad_norm": 3.209359645843506,
      "learning_rate": 4.724424845577227e-06,
      "loss": 0.0299,
      "step": 10116
    },
    {
      "epoch": 0.5224373870384714,
      "grad_norm": 5.492812156677246,
      "learning_rate": 4.723606734036955e-06,
      "loss": 0.0512,
      "step": 10117
    },
    {
      "epoch": 0.5224890265943712,
      "grad_norm": 4.91958475112915,
      "learning_rate": 4.722788629918968e-06,
      "loss": 0.0274,
      "step": 10118
    },
    {
      "epoch": 0.5225406661502711,
      "grad_norm": 3.0703463554382324,
      "learning_rate": 4.72197053324524e-06,
      "loss": 0.0661,
      "step": 10119
    },
    {
      "epoch": 0.5225923057061709,
      "grad_norm": 8.223358154296875,
      "learning_rate": 4.721152444037733e-06,
      "loss": 0.1536,
      "step": 10120
    },
    {
      "epoch": 0.5226439452620707,
      "grad_norm": 5.863459587097168,
      "learning_rate": 4.72033436231842e-06,
      "loss": 0.0876,
      "step": 10121
    },
    {
      "epoch": 0.5226955848179705,
      "grad_norm": 2.876469135284424,
      "learning_rate": 4.719516288109271e-06,
      "loss": 0.0515,
      "step": 10122
    },
    {
      "epoch": 0.5227472243738703,
      "grad_norm": 5.44970703125,
      "learning_rate": 4.718698221432251e-06,
      "loss": 0.0363,
      "step": 10123
    },
    {
      "epoch": 0.5227988639297702,
      "grad_norm": 3.8298661708831787,
      "learning_rate": 4.7178801623093305e-06,
      "loss": 0.0749,
      "step": 10124
    },
    {
      "epoch": 0.52285050348567,
      "grad_norm": 2.5218751430511475,
      "learning_rate": 4.717062110762476e-06,
      "loss": 0.0451,
      "step": 10125
    },
    {
      "epoch": 0.5229021430415698,
      "grad_norm": 2.911221981048584,
      "learning_rate": 4.716244066813662e-06,
      "loss": 0.0221,
      "step": 10126
    },
    {
      "epoch": 0.5229537825974696,
      "grad_norm": 1.4887923002243042,
      "learning_rate": 4.715426030484846e-06,
      "loss": 0.0153,
      "step": 10127
    },
    {
      "epoch": 0.5230054221533694,
      "grad_norm": 1.7347780466079712,
      "learning_rate": 4.714608001798003e-06,
      "loss": 0.0374,
      "step": 10128
    },
    {
      "epoch": 0.5230570617092694,
      "grad_norm": 2.4263596534729004,
      "learning_rate": 4.713789980775098e-06,
      "loss": 0.0403,
      "step": 10129
    },
    {
      "epoch": 0.5231087012651692,
      "grad_norm": 10.602900505065918,
      "learning_rate": 4.7129719674381e-06,
      "loss": 0.0903,
      "step": 10130
    },
    {
      "epoch": 0.523160340821069,
      "grad_norm": 7.414513111114502,
      "learning_rate": 4.712153961808974e-06,
      "loss": 0.0468,
      "step": 10131
    },
    {
      "epoch": 0.5232119803769688,
      "grad_norm": 9.440914154052734,
      "learning_rate": 4.711335963909685e-06,
      "loss": 0.1379,
      "step": 10132
    },
    {
      "epoch": 0.5232636199328686,
      "grad_norm": 7.164198875427246,
      "learning_rate": 4.710517973762206e-06,
      "loss": 0.0318,
      "step": 10133
    },
    {
      "epoch": 0.5233152594887684,
      "grad_norm": 4.602572441101074,
      "learning_rate": 4.7096999913884955e-06,
      "loss": 0.0572,
      "step": 10134
    },
    {
      "epoch": 0.5233668990446683,
      "grad_norm": 5.134298324584961,
      "learning_rate": 4.708882016810524e-06,
      "loss": 0.0432,
      "step": 10135
    },
    {
      "epoch": 0.5234185386005681,
      "grad_norm": 7.887735843658447,
      "learning_rate": 4.708064050050258e-06,
      "loss": 0.1082,
      "step": 10136
    },
    {
      "epoch": 0.5234701781564679,
      "grad_norm": 4.7226080894470215,
      "learning_rate": 4.707246091129663e-06,
      "loss": 0.0493,
      "step": 10137
    },
    {
      "epoch": 0.5235218177123677,
      "grad_norm": 2.4052906036376953,
      "learning_rate": 4.706428140070702e-06,
      "loss": 0.0114,
      "step": 10138
    },
    {
      "epoch": 0.5235734572682675,
      "grad_norm": 8.895233154296875,
      "learning_rate": 4.705610196895341e-06,
      "loss": 0.1061,
      "step": 10139
    },
    {
      "epoch": 0.5236250968241674,
      "grad_norm": 2.9074554443359375,
      "learning_rate": 4.704792261625548e-06,
      "loss": 0.0451,
      "step": 10140
    },
    {
      "epoch": 0.5236767363800672,
      "grad_norm": 8.921539306640625,
      "learning_rate": 4.703974334283286e-06,
      "loss": 0.0711,
      "step": 10141
    },
    {
      "epoch": 0.523728375935967,
      "grad_norm": 4.6724934577941895,
      "learning_rate": 4.703156414890519e-06,
      "loss": 0.032,
      "step": 10142
    },
    {
      "epoch": 0.5237800154918668,
      "grad_norm": 12.574089050292969,
      "learning_rate": 4.7023385034692125e-06,
      "loss": 0.2219,
      "step": 10143
    },
    {
      "epoch": 0.5238316550477666,
      "grad_norm": 5.475788116455078,
      "learning_rate": 4.701520600041332e-06,
      "loss": 0.0416,
      "step": 10144
    },
    {
      "epoch": 0.5238832946036664,
      "grad_norm": 4.13463830947876,
      "learning_rate": 4.7007027046288375e-06,
      "loss": 0.0907,
      "step": 10145
    },
    {
      "epoch": 0.5239349341595663,
      "grad_norm": 3.812722682952881,
      "learning_rate": 4.699884817253694e-06,
      "loss": 0.0739,
      "step": 10146
    },
    {
      "epoch": 0.5239865737154661,
      "grad_norm": 3.5714545249938965,
      "learning_rate": 4.699066937937868e-06,
      "loss": 0.0649,
      "step": 10147
    },
    {
      "epoch": 0.5240382132713659,
      "grad_norm": 9.736143112182617,
      "learning_rate": 4.698249066703323e-06,
      "loss": 0.1501,
      "step": 10148
    },
    {
      "epoch": 0.5240898528272657,
      "grad_norm": 1.742552638053894,
      "learning_rate": 4.697431203572019e-06,
      "loss": 0.0289,
      "step": 10149
    },
    {
      "epoch": 0.5241414923831655,
      "grad_norm": 3.6654977798461914,
      "learning_rate": 4.69661334856592e-06,
      "loss": 0.0664,
      "step": 10150
    },
    {
      "epoch": 0.5241931319390654,
      "grad_norm": 11.915450096130371,
      "learning_rate": 4.695795501706988e-06,
      "loss": 0.1275,
      "step": 10151
    },
    {
      "epoch": 0.5242447714949652,
      "grad_norm": 7.581373691558838,
      "learning_rate": 4.6949776630171904e-06,
      "loss": 0.0791,
      "step": 10152
    },
    {
      "epoch": 0.524296411050865,
      "grad_norm": 1.7286165952682495,
      "learning_rate": 4.6941598325184814e-06,
      "loss": 0.0366,
      "step": 10153
    },
    {
      "epoch": 0.5243480506067648,
      "grad_norm": 2.137376308441162,
      "learning_rate": 4.69334201023283e-06,
      "loss": 0.0215,
      "step": 10154
    },
    {
      "epoch": 0.5243996901626646,
      "grad_norm": 3.8991758823394775,
      "learning_rate": 4.692524196182196e-06,
      "loss": 0.065,
      "step": 10155
    },
    {
      "epoch": 0.5244513297185645,
      "grad_norm": 6.135923862457275,
      "learning_rate": 4.691706390388539e-06,
      "loss": 0.0752,
      "step": 10156
    },
    {
      "epoch": 0.5245029692744643,
      "grad_norm": 3.2335093021392822,
      "learning_rate": 4.6908885928738224e-06,
      "loss": 0.0552,
      "step": 10157
    },
    {
      "epoch": 0.5245546088303641,
      "grad_norm": 3.588066339492798,
      "learning_rate": 4.690070803660006e-06,
      "loss": 0.0952,
      "step": 10158
    },
    {
      "epoch": 0.5246062483862639,
      "grad_norm": 4.307742595672607,
      "learning_rate": 4.6892530227690544e-06,
      "loss": 0.0585,
      "step": 10159
    },
    {
      "epoch": 0.5246578879421637,
      "grad_norm": 3.36706805229187,
      "learning_rate": 4.688435250222923e-06,
      "loss": 0.0298,
      "step": 10160
    },
    {
      "epoch": 0.5247095274980635,
      "grad_norm": 3.699171304702759,
      "learning_rate": 4.6876174860435755e-06,
      "loss": 0.0688,
      "step": 10161
    },
    {
      "epoch": 0.5247611670539634,
      "grad_norm": 4.241973876953125,
      "learning_rate": 4.6867997302529726e-06,
      "loss": 0.0525,
      "step": 10162
    },
    {
      "epoch": 0.5248128066098632,
      "grad_norm": 9.51194953918457,
      "learning_rate": 4.685981982873074e-06,
      "loss": 0.1598,
      "step": 10163
    },
    {
      "epoch": 0.524864446165763,
      "grad_norm": 5.320539474487305,
      "learning_rate": 4.685164243925838e-06,
      "loss": 0.0761,
      "step": 10164
    },
    {
      "epoch": 0.5249160857216628,
      "grad_norm": 3.2337005138397217,
      "learning_rate": 4.684346513433223e-06,
      "loss": 0.0534,
      "step": 10165
    },
    {
      "epoch": 0.5249677252775626,
      "grad_norm": 5.43710994720459,
      "learning_rate": 4.683528791417196e-06,
      "loss": 0.0408,
      "step": 10166
    },
    {
      "epoch": 0.5250193648334625,
      "grad_norm": 7.018599987030029,
      "learning_rate": 4.682711077899707e-06,
      "loss": 0.1184,
      "step": 10167
    },
    {
      "epoch": 0.5250710043893623,
      "grad_norm": 3.778221845626831,
      "learning_rate": 4.681893372902719e-06,
      "loss": 0.0621,
      "step": 10168
    },
    {
      "epoch": 0.5251226439452621,
      "grad_norm": 7.295832633972168,
      "learning_rate": 4.681075676448191e-06,
      "loss": 0.0986,
      "step": 10169
    },
    {
      "epoch": 0.5251742835011619,
      "grad_norm": 7.013349533081055,
      "learning_rate": 4.680257988558082e-06,
      "loss": 0.0792,
      "step": 10170
    },
    {
      "epoch": 0.5252259230570617,
      "grad_norm": 17.597200393676758,
      "learning_rate": 4.679440309254347e-06,
      "loss": 0.1582,
      "step": 10171
    },
    {
      "epoch": 0.5252775626129615,
      "grad_norm": 2.486276626586914,
      "learning_rate": 4.678622638558947e-06,
      "loss": 0.0383,
      "step": 10172
    },
    {
      "epoch": 0.5253292021688614,
      "grad_norm": 14.121395111083984,
      "learning_rate": 4.677804976493839e-06,
      "loss": 0.0833,
      "step": 10173
    },
    {
      "epoch": 0.5253808417247612,
      "grad_norm": 2.1778361797332764,
      "learning_rate": 4.676987323080982e-06,
      "loss": 0.059,
      "step": 10174
    },
    {
      "epoch": 0.525432481280661,
      "grad_norm": 3.931981086730957,
      "learning_rate": 4.676169678342331e-06,
      "loss": 0.0556,
      "step": 10175
    },
    {
      "epoch": 0.5254841208365608,
      "grad_norm": 3.68290638923645,
      "learning_rate": 4.675352042299843e-06,
      "loss": 0.0577,
      "step": 10176
    },
    {
      "epoch": 0.5255357603924606,
      "grad_norm": 10.496454238891602,
      "learning_rate": 4.674534414975478e-06,
      "loss": 0.1137,
      "step": 10177
    },
    {
      "epoch": 0.5255873999483605,
      "grad_norm": 3.3784546852111816,
      "learning_rate": 4.673716796391189e-06,
      "loss": 0.0444,
      "step": 10178
    },
    {
      "epoch": 0.5256390395042603,
      "grad_norm": 3.864816665649414,
      "learning_rate": 4.672899186568933e-06,
      "loss": 0.0738,
      "step": 10179
    },
    {
      "epoch": 0.5256906790601601,
      "grad_norm": 2.7941524982452393,
      "learning_rate": 4.672081585530668e-06,
      "loss": 0.0352,
      "step": 10180
    },
    {
      "epoch": 0.5257423186160599,
      "grad_norm": 2.3248097896575928,
      "learning_rate": 4.67126399329835e-06,
      "loss": 0.0456,
      "step": 10181
    },
    {
      "epoch": 0.5257939581719597,
      "grad_norm": 8.138144493103027,
      "learning_rate": 4.670446409893933e-06,
      "loss": 0.088,
      "step": 10182
    },
    {
      "epoch": 0.5258455977278595,
      "grad_norm": 5.622934818267822,
      "learning_rate": 4.669628835339372e-06,
      "loss": 0.0436,
      "step": 10183
    },
    {
      "epoch": 0.5258972372837594,
      "grad_norm": 1.693249225616455,
      "learning_rate": 4.668811269656623e-06,
      "loss": 0.0543,
      "step": 10184
    },
    {
      "epoch": 0.5259488768396592,
      "grad_norm": 25.799842834472656,
      "learning_rate": 4.667993712867645e-06,
      "loss": 0.0731,
      "step": 10185
    },
    {
      "epoch": 0.526000516395559,
      "grad_norm": 5.083909034729004,
      "learning_rate": 4.667176164994385e-06,
      "loss": 0.0551,
      "step": 10186
    },
    {
      "epoch": 0.5260521559514588,
      "grad_norm": 7.012645244598389,
      "learning_rate": 4.666358626058802e-06,
      "loss": 0.0606,
      "step": 10187
    },
    {
      "epoch": 0.5261037955073586,
      "grad_norm": 8.213972091674805,
      "learning_rate": 4.665541096082852e-06,
      "loss": 0.0735,
      "step": 10188
    },
    {
      "epoch": 0.5261554350632585,
      "grad_norm": 3.0509517192840576,
      "learning_rate": 4.664723575088485e-06,
      "loss": 0.0285,
      "step": 10189
    },
    {
      "epoch": 0.5262070746191583,
      "grad_norm": 4.473606109619141,
      "learning_rate": 4.663906063097657e-06,
      "loss": 0.0612,
      "step": 10190
    },
    {
      "epoch": 0.5262587141750581,
      "grad_norm": 4.8949432373046875,
      "learning_rate": 4.6630885601323196e-06,
      "loss": 0.0787,
      "step": 10191
    },
    {
      "epoch": 0.5263103537309579,
      "grad_norm": 3.36238694190979,
      "learning_rate": 4.662271066214431e-06,
      "loss": 0.0293,
      "step": 10192
    },
    {
      "epoch": 0.5263619932868577,
      "grad_norm": 3.156980276107788,
      "learning_rate": 4.661453581365937e-06,
      "loss": 0.0605,
      "step": 10193
    },
    {
      "epoch": 0.5264136328427576,
      "grad_norm": 4.195350646972656,
      "learning_rate": 4.6606361056087955e-06,
      "loss": 0.0338,
      "step": 10194
    },
    {
      "epoch": 0.5264652723986574,
      "grad_norm": 3.8324689865112305,
      "learning_rate": 4.659818638964958e-06,
      "loss": 0.0344,
      "step": 10195
    },
    {
      "epoch": 0.5265169119545572,
      "grad_norm": 2.903521776199341,
      "learning_rate": 4.659001181456378e-06,
      "loss": 0.0347,
      "step": 10196
    },
    {
      "epoch": 0.526568551510457,
      "grad_norm": 4.720805644989014,
      "learning_rate": 4.658183733105004e-06,
      "loss": 0.0652,
      "step": 10197
    },
    {
      "epoch": 0.5266201910663568,
      "grad_norm": 6.618748188018799,
      "learning_rate": 4.657366293932789e-06,
      "loss": 0.063,
      "step": 10198
    },
    {
      "epoch": 0.5266718306222566,
      "grad_norm": 2.857363700866699,
      "learning_rate": 4.65654886396169e-06,
      "loss": 0.0267,
      "step": 10199
    },
    {
      "epoch": 0.5267234701781565,
      "grad_norm": 6.707945823669434,
      "learning_rate": 4.655731443213649e-06,
      "loss": 0.0324,
      "step": 10200
    },
    {
      "epoch": 0.5267751097340563,
      "grad_norm": 6.200930118560791,
      "learning_rate": 4.654914031710624e-06,
      "loss": 0.0627,
      "step": 10201
    },
    {
      "epoch": 0.5268267492899561,
      "grad_norm": 6.036828994750977,
      "learning_rate": 4.654096629474563e-06,
      "loss": 0.0616,
      "step": 10202
    },
    {
      "epoch": 0.5268783888458559,
      "grad_norm": 5.005843162536621,
      "learning_rate": 4.653279236527419e-06,
      "loss": 0.0664,
      "step": 10203
    },
    {
      "epoch": 0.5269300284017557,
      "grad_norm": 1.873557448387146,
      "learning_rate": 4.65246185289114e-06,
      "loss": 0.0143,
      "step": 10204
    },
    {
      "epoch": 0.5269816679576556,
      "grad_norm": 4.472450256347656,
      "learning_rate": 4.651644478587674e-06,
      "loss": 0.0739,
      "step": 10205
    },
    {
      "epoch": 0.5270333075135554,
      "grad_norm": 3.1661624908447266,
      "learning_rate": 4.650827113638976e-06,
      "loss": 0.0499,
      "step": 10206
    },
    {
      "epoch": 0.5270849470694552,
      "grad_norm": 7.00613260269165,
      "learning_rate": 4.650009758066995e-06,
      "loss": 0.0478,
      "step": 10207
    },
    {
      "epoch": 0.527136586625355,
      "grad_norm": 11.79852294921875,
      "learning_rate": 4.649192411893676e-06,
      "loss": 0.0995,
      "step": 10208
    },
    {
      "epoch": 0.5271882261812548,
      "grad_norm": 6.124041557312012,
      "learning_rate": 4.6483750751409706e-06,
      "loss": 0.0346,
      "step": 10209
    },
    {
      "epoch": 0.5272398657371546,
      "grad_norm": 24.075721740722656,
      "learning_rate": 4.6475577478308295e-06,
      "loss": 0.1859,
      "step": 10210
    },
    {
      "epoch": 0.5272915052930545,
      "grad_norm": 1.8985190391540527,
      "learning_rate": 4.646740429985197e-06,
      "loss": 0.0386,
      "step": 10211
    },
    {
      "epoch": 0.5273431448489543,
      "grad_norm": 1.8133071660995483,
      "learning_rate": 4.645923121626024e-06,
      "loss": 0.0303,
      "step": 10212
    },
    {
      "epoch": 0.5273947844048541,
      "grad_norm": 3.4530551433563232,
      "learning_rate": 4.645105822775258e-06,
      "loss": 0.065,
      "step": 10213
    },
    {
      "epoch": 0.5274464239607539,
      "grad_norm": 11.458757400512695,
      "learning_rate": 4.6442885334548495e-06,
      "loss": 0.0601,
      "step": 10214
    },
    {
      "epoch": 0.5274980635166537,
      "grad_norm": 4.2300801277160645,
      "learning_rate": 4.6434712536867424e-06,
      "loss": 0.0357,
      "step": 10215
    },
    {
      "epoch": 0.5275497030725536,
      "grad_norm": 6.756396293640137,
      "learning_rate": 4.642653983492884e-06,
      "loss": 0.0607,
      "step": 10216
    },
    {
      "epoch": 0.5276013426284534,
      "grad_norm": 9.630913734436035,
      "learning_rate": 4.641836722895223e-06,
      "loss": 0.1426,
      "step": 10217
    },
    {
      "epoch": 0.5276529821843532,
      "grad_norm": 3.5614826679229736,
      "learning_rate": 4.641019471915709e-06,
      "loss": 0.0545,
      "step": 10218
    },
    {
      "epoch": 0.527704621740253,
      "grad_norm": 2.7670631408691406,
      "learning_rate": 4.640202230576281e-06,
      "loss": 0.047,
      "step": 10219
    },
    {
      "epoch": 0.5277562612961528,
      "grad_norm": 5.2236647605896,
      "learning_rate": 4.639384998898892e-06,
      "loss": 0.0491,
      "step": 10220
    },
    {
      "epoch": 0.5278079008520526,
      "grad_norm": 5.007957458496094,
      "learning_rate": 4.638567776905486e-06,
      "loss": 0.0613,
      "step": 10221
    },
    {
      "epoch": 0.5278595404079525,
      "grad_norm": 3.2076950073242188,
      "learning_rate": 4.637750564618007e-06,
      "loss": 0.0962,
      "step": 10222
    },
    {
      "epoch": 0.5279111799638523,
      "grad_norm": 2.2981631755828857,
      "learning_rate": 4.636933362058401e-06,
      "loss": 0.056,
      "step": 10223
    },
    {
      "epoch": 0.5279628195197521,
      "grad_norm": 2.1092045307159424,
      "learning_rate": 4.636116169248614e-06,
      "loss": 0.0322,
      "step": 10224
    },
    {
      "epoch": 0.5280144590756519,
      "grad_norm": 6.293689250946045,
      "learning_rate": 4.635298986210593e-06,
      "loss": 0.0347,
      "step": 10225
    },
    {
      "epoch": 0.5280660986315517,
      "grad_norm": 3.6705379486083984,
      "learning_rate": 4.63448181296628e-06,
      "loss": 0.0761,
      "step": 10226
    },
    {
      "epoch": 0.5281177381874516,
      "grad_norm": 10.128661155700684,
      "learning_rate": 4.633664649537619e-06,
      "loss": 0.1376,
      "step": 10227
    },
    {
      "epoch": 0.5281693777433514,
      "grad_norm": 4.457966327667236,
      "learning_rate": 4.632847495946557e-06,
      "loss": 0.099,
      "step": 10228
    },
    {
      "epoch": 0.5282210172992512,
      "grad_norm": 2.672793388366699,
      "learning_rate": 4.632030352215035e-06,
      "loss": 0.0241,
      "step": 10229
    },
    {
      "epoch": 0.528272656855151,
      "grad_norm": 6.398762226104736,
      "learning_rate": 4.6312132183649985e-06,
      "loss": 0.0524,
      "step": 10230
    },
    {
      "epoch": 0.5283242964110508,
      "grad_norm": 4.2523980140686035,
      "learning_rate": 4.630396094418388e-06,
      "loss": 0.0488,
      "step": 10231
    },
    {
      "epoch": 0.5283759359669506,
      "grad_norm": 12.928823471069336,
      "learning_rate": 4.629578980397151e-06,
      "loss": 0.1602,
      "step": 10232
    },
    {
      "epoch": 0.5284275755228505,
      "grad_norm": 1.758288860321045,
      "learning_rate": 4.628761876323228e-06,
      "loss": 0.0258,
      "step": 10233
    },
    {
      "epoch": 0.5284792150787503,
      "grad_norm": 5.355262279510498,
      "learning_rate": 4.627944782218561e-06,
      "loss": 0.0432,
      "step": 10234
    },
    {
      "epoch": 0.5285308546346501,
      "grad_norm": 2.215352773666382,
      "learning_rate": 4.6271276981050935e-06,
      "loss": 0.0422,
      "step": 10235
    },
    {
      "epoch": 0.5285824941905499,
      "grad_norm": 5.764080047607422,
      "learning_rate": 4.626310624004768e-06,
      "loss": 0.0444,
      "step": 10236
    },
    {
      "epoch": 0.5286341337464497,
      "grad_norm": 5.678531646728516,
      "learning_rate": 4.625493559939523e-06,
      "loss": 0.0727,
      "step": 10237
    },
    {
      "epoch": 0.5286857733023496,
      "grad_norm": 1.9677014350891113,
      "learning_rate": 4.624676505931303e-06,
      "loss": 0.0258,
      "step": 10238
    },
    {
      "epoch": 0.5287374128582494,
      "grad_norm": 7.746038436889648,
      "learning_rate": 4.6238594620020486e-06,
      "loss": 0.0655,
      "step": 10239
    },
    {
      "epoch": 0.5287890524141492,
      "grad_norm": 4.070205211639404,
      "learning_rate": 4.6230424281737025e-06,
      "loss": 0.0378,
      "step": 10240
    },
    {
      "epoch": 0.528840691970049,
      "grad_norm": 3.6576220989227295,
      "learning_rate": 4.6222254044682015e-06,
      "loss": 0.0305,
      "step": 10241
    },
    {
      "epoch": 0.5288923315259488,
      "grad_norm": 2.627260684967041,
      "learning_rate": 4.6214083909074896e-06,
      "loss": 0.0325,
      "step": 10242
    },
    {
      "epoch": 0.5289439710818487,
      "grad_norm": 10.808601379394531,
      "learning_rate": 4.620591387513505e-06,
      "loss": 0.0729,
      "step": 10243
    },
    {
      "epoch": 0.5289956106377485,
      "grad_norm": 2.004659652709961,
      "learning_rate": 4.619774394308188e-06,
      "loss": 0.0226,
      "step": 10244
    },
    {
      "epoch": 0.5290472501936483,
      "grad_norm": 9.886353492736816,
      "learning_rate": 4.618957411313477e-06,
      "loss": 0.1245,
      "step": 10245
    },
    {
      "epoch": 0.5290988897495481,
      "grad_norm": 9.029800415039062,
      "learning_rate": 4.618140438551313e-06,
      "loss": 0.0725,
      "step": 10246
    },
    {
      "epoch": 0.5291505293054479,
      "grad_norm": 4.076279640197754,
      "learning_rate": 4.617323476043636e-06,
      "loss": 0.0372,
      "step": 10247
    },
    {
      "epoch": 0.5292021688613477,
      "grad_norm": 9.988113403320312,
      "learning_rate": 4.616506523812382e-06,
      "loss": 0.1345,
      "step": 10248
    },
    {
      "epoch": 0.5292538084172476,
      "grad_norm": 3.3171300888061523,
      "learning_rate": 4.615689581879492e-06,
      "loss": 0.0369,
      "step": 10249
    },
    {
      "epoch": 0.5293054479731474,
      "grad_norm": 2.6803557872772217,
      "learning_rate": 4.6148726502669e-06,
      "loss": 0.0335,
      "step": 10250
    },
    {
      "epoch": 0.5293570875290472,
      "grad_norm": 9.433267593383789,
      "learning_rate": 4.614055728996551e-06,
      "loss": 0.0584,
      "step": 10251
    },
    {
      "epoch": 0.529408727084947,
      "grad_norm": 1.7193621397018433,
      "learning_rate": 4.613238818090378e-06,
      "loss": 0.0305,
      "step": 10252
    },
    {
      "epoch": 0.5294603666408468,
      "grad_norm": 4.666865825653076,
      "learning_rate": 4.612421917570317e-06,
      "loss": 0.0659,
      "step": 10253
    },
    {
      "epoch": 0.5295120061967467,
      "grad_norm": 4.08522367477417,
      "learning_rate": 4.61160502745831e-06,
      "loss": 0.0534,
      "step": 10254
    },
    {
      "epoch": 0.5295636457526465,
      "grad_norm": 3.7155909538269043,
      "learning_rate": 4.61078814777629e-06,
      "loss": 0.0471,
      "step": 10255
    },
    {
      "epoch": 0.5296152853085464,
      "grad_norm": 3.0964572429656982,
      "learning_rate": 4.609971278546193e-06,
      "loss": 0.0376,
      "step": 10256
    },
    {
      "epoch": 0.5296669248644462,
      "grad_norm": 4.9848785400390625,
      "learning_rate": 4.609154419789957e-06,
      "loss": 0.06,
      "step": 10257
    },
    {
      "epoch": 0.529718564420346,
      "grad_norm": 2.0814871788024902,
      "learning_rate": 4.608337571529519e-06,
      "loss": 0.036,
      "step": 10258
    },
    {
      "epoch": 0.5297702039762459,
      "grad_norm": 2.1295247077941895,
      "learning_rate": 4.607520733786813e-06,
      "loss": 0.0297,
      "step": 10259
    },
    {
      "epoch": 0.5298218435321457,
      "grad_norm": 3.390932083129883,
      "learning_rate": 4.606703906583776e-06,
      "loss": 0.0324,
      "step": 10260
    },
    {
      "epoch": 0.5298734830880455,
      "grad_norm": 21.012672424316406,
      "learning_rate": 4.6058870899423406e-06,
      "loss": 0.218,
      "step": 10261
    },
    {
      "epoch": 0.5299251226439453,
      "grad_norm": 25.4072322845459,
      "learning_rate": 4.605070283884446e-06,
      "loss": 0.1967,
      "step": 10262
    },
    {
      "epoch": 0.5299767621998451,
      "grad_norm": 2.7890000343322754,
      "learning_rate": 4.604253488432021e-06,
      "loss": 0.0187,
      "step": 10263
    },
    {
      "epoch": 0.530028401755745,
      "grad_norm": 3.4625394344329834,
      "learning_rate": 4.603436703607002e-06,
      "loss": 0.0606,
      "step": 10264
    },
    {
      "epoch": 0.5300800413116448,
      "grad_norm": 8.174161911010742,
      "learning_rate": 4.6026199294313265e-06,
      "loss": 0.0903,
      "step": 10265
    },
    {
      "epoch": 0.5301316808675446,
      "grad_norm": 2.7551095485687256,
      "learning_rate": 4.601803165926924e-06,
      "loss": 0.0407,
      "step": 10266
    },
    {
      "epoch": 0.5301833204234444,
      "grad_norm": 11.571508407592773,
      "learning_rate": 4.60098641311573e-06,
      "loss": 0.0408,
      "step": 10267
    },
    {
      "epoch": 0.5302349599793442,
      "grad_norm": 2.6672708988189697,
      "learning_rate": 4.600169671019677e-06,
      "loss": 0.0258,
      "step": 10268
    },
    {
      "epoch": 0.530286599535244,
      "grad_norm": 3.7388038635253906,
      "learning_rate": 4.5993529396606984e-06,
      "loss": 0.0499,
      "step": 10269
    },
    {
      "epoch": 0.5303382390911439,
      "grad_norm": 14.772917747497559,
      "learning_rate": 4.598536219060725e-06,
      "loss": 0.1126,
      "step": 10270
    },
    {
      "epoch": 0.5303898786470437,
      "grad_norm": 7.6744256019592285,
      "learning_rate": 4.597719509241691e-06,
      "loss": 0.0564,
      "step": 10271
    },
    {
      "epoch": 0.5304415182029435,
      "grad_norm": 5.447461128234863,
      "learning_rate": 4.596902810225527e-06,
      "loss": 0.057,
      "step": 10272
    },
    {
      "epoch": 0.5304931577588433,
      "grad_norm": 7.75293493270874,
      "learning_rate": 4.596086122034168e-06,
      "loss": 0.0934,
      "step": 10273
    },
    {
      "epoch": 0.5305447973147431,
      "grad_norm": 13.412622451782227,
      "learning_rate": 4.595269444689541e-06,
      "loss": 0.1419,
      "step": 10274
    },
    {
      "epoch": 0.530596436870643,
      "grad_norm": 2.890249252319336,
      "learning_rate": 4.59445277821358e-06,
      "loss": 0.0324,
      "step": 10275
    },
    {
      "epoch": 0.5306480764265428,
      "grad_norm": 2.8772897720336914,
      "learning_rate": 4.593636122628215e-06,
      "loss": 0.0418,
      "step": 10276
    },
    {
      "epoch": 0.5306997159824426,
      "grad_norm": 5.290698051452637,
      "learning_rate": 4.592819477955374e-06,
      "loss": 0.0557,
      "step": 10277
    },
    {
      "epoch": 0.5307513555383424,
      "grad_norm": 8.293206214904785,
      "learning_rate": 4.592002844216991e-06,
      "loss": 0.0902,
      "step": 10278
    },
    {
      "epoch": 0.5308029950942422,
      "grad_norm": 14.486292839050293,
      "learning_rate": 4.591186221434995e-06,
      "loss": 0.0906,
      "step": 10279
    },
    {
      "epoch": 0.530854634650142,
      "grad_norm": 5.7178215980529785,
      "learning_rate": 4.590369609631315e-06,
      "loss": 0.0919,
      "step": 10280
    },
    {
      "epoch": 0.5309062742060419,
      "grad_norm": 3.87528133392334,
      "learning_rate": 4.5895530088278805e-06,
      "loss": 0.0372,
      "step": 10281
    },
    {
      "epoch": 0.5309579137619417,
      "grad_norm": 5.24851131439209,
      "learning_rate": 4.5887364190466206e-06,
      "loss": 0.0547,
      "step": 10282
    },
    {
      "epoch": 0.5310095533178415,
      "grad_norm": 3.414196252822876,
      "learning_rate": 4.587919840309462e-06,
      "loss": 0.0558,
      "step": 10283
    },
    {
      "epoch": 0.5310611928737413,
      "grad_norm": 2.0549304485321045,
      "learning_rate": 4.587103272638339e-06,
      "loss": 0.0328,
      "step": 10284
    },
    {
      "epoch": 0.5311128324296411,
      "grad_norm": 18.67033576965332,
      "learning_rate": 4.586286716055174e-06,
      "loss": 0.0541,
      "step": 10285
    },
    {
      "epoch": 0.531164471985541,
      "grad_norm": 5.476624011993408,
      "learning_rate": 4.585470170581897e-06,
      "loss": 0.0606,
      "step": 10286
    },
    {
      "epoch": 0.5312161115414408,
      "grad_norm": 2.3471264839172363,
      "learning_rate": 4.5846536362404365e-06,
      "loss": 0.0415,
      "step": 10287
    },
    {
      "epoch": 0.5312677510973406,
      "grad_norm": 3.1072299480438232,
      "learning_rate": 4.583837113052718e-06,
      "loss": 0.0556,
      "step": 10288
    },
    {
      "epoch": 0.5313193906532404,
      "grad_norm": 12.125080108642578,
      "learning_rate": 4.583020601040667e-06,
      "loss": 0.0605,
      "step": 10289
    },
    {
      "epoch": 0.5313710302091402,
      "grad_norm": 5.445802688598633,
      "learning_rate": 4.5822041002262145e-06,
      "loss": 0.0605,
      "step": 10290
    },
    {
      "epoch": 0.53142266976504,
      "grad_norm": 3.0585458278656006,
      "learning_rate": 4.581387610631286e-06,
      "loss": 0.0449,
      "step": 10291
    },
    {
      "epoch": 0.5314743093209399,
      "grad_norm": 3.306288003921509,
      "learning_rate": 4.5805711322778045e-06,
      "loss": 0.0623,
      "step": 10292
    },
    {
      "epoch": 0.5315259488768397,
      "grad_norm": 16.564411163330078,
      "learning_rate": 4.579754665187698e-06,
      "loss": 0.0958,
      "step": 10293
    },
    {
      "epoch": 0.5315775884327395,
      "grad_norm": 0.9306775331497192,
      "learning_rate": 4.578938209382891e-06,
      "loss": 0.0091,
      "step": 10294
    },
    {
      "epoch": 0.5316292279886393,
      "grad_norm": 4.377257347106934,
      "learning_rate": 4.578121764885311e-06,
      "loss": 0.0852,
      "step": 10295
    },
    {
      "epoch": 0.5316808675445391,
      "grad_norm": 2.043048143386841,
      "learning_rate": 4.577305331716879e-06,
      "loss": 0.0359,
      "step": 10296
    },
    {
      "epoch": 0.531732507100439,
      "grad_norm": 17.466793060302734,
      "learning_rate": 4.576488909899522e-06,
      "loss": 0.1054,
      "step": 10297
    },
    {
      "epoch": 0.5317841466563388,
      "grad_norm": 5.227063179016113,
      "learning_rate": 4.575672499455165e-06,
      "loss": 0.0507,
      "step": 10298
    },
    {
      "epoch": 0.5318357862122386,
      "grad_norm": 15.833327293395996,
      "learning_rate": 4.574856100405731e-06,
      "loss": 0.1272,
      "step": 10299
    },
    {
      "epoch": 0.5318874257681384,
      "grad_norm": 4.573526859283447,
      "learning_rate": 4.574039712773143e-06,
      "loss": 0.0479,
      "step": 10300
    },
    {
      "epoch": 0.5319390653240382,
      "grad_norm": 5.231513500213623,
      "learning_rate": 4.573223336579324e-06,
      "loss": 0.0666,
      "step": 10301
    },
    {
      "epoch": 0.531990704879938,
      "grad_norm": 5.3791704177856445,
      "learning_rate": 4.572406971846199e-06,
      "loss": 0.0715,
      "step": 10302
    },
    {
      "epoch": 0.5320423444358379,
      "grad_norm": 1.8696130514144897,
      "learning_rate": 4.571590618595688e-06,
      "loss": 0.0324,
      "step": 10303
    },
    {
      "epoch": 0.5320939839917377,
      "grad_norm": 6.399851322174072,
      "learning_rate": 4.570774276849716e-06,
      "loss": 0.0773,
      "step": 10304
    },
    {
      "epoch": 0.5321456235476375,
      "grad_norm": 8.041019439697266,
      "learning_rate": 4.569957946630204e-06,
      "loss": 0.0603,
      "step": 10305
    },
    {
      "epoch": 0.5321972631035373,
      "grad_norm": 3.9829261302948,
      "learning_rate": 4.5691416279590754e-06,
      "loss": 0.048,
      "step": 10306
    },
    {
      "epoch": 0.5322489026594371,
      "grad_norm": 9.378796577453613,
      "learning_rate": 4.568325320858249e-06,
      "loss": 0.0422,
      "step": 10307
    },
    {
      "epoch": 0.532300542215337,
      "grad_norm": 4.141866207122803,
      "learning_rate": 4.567509025349647e-06,
      "loss": 0.0731,
      "step": 10308
    },
    {
      "epoch": 0.5323521817712368,
      "grad_norm": 2.588203191757202,
      "learning_rate": 4.566692741455191e-06,
      "loss": 0.06,
      "step": 10309
    },
    {
      "epoch": 0.5324038213271366,
      "grad_norm": 2.774794101715088,
      "learning_rate": 4.565876469196799e-06,
      "loss": 0.0508,
      "step": 10310
    },
    {
      "epoch": 0.5324554608830364,
      "grad_norm": 13.22331714630127,
      "learning_rate": 4.565060208596394e-06,
      "loss": 0.0745,
      "step": 10311
    },
    {
      "epoch": 0.5325071004389362,
      "grad_norm": 3.3898885250091553,
      "learning_rate": 4.564243959675896e-06,
      "loss": 0.0571,
      "step": 10312
    },
    {
      "epoch": 0.532558739994836,
      "grad_norm": 3.3797807693481445,
      "learning_rate": 4.563427722457224e-06,
      "loss": 0.0366,
      "step": 10313
    },
    {
      "epoch": 0.5326103795507359,
      "grad_norm": 2.8339927196502686,
      "learning_rate": 4.562611496962297e-06,
      "loss": 0.0371,
      "step": 10314
    },
    {
      "epoch": 0.5326620191066357,
      "grad_norm": 4.749706745147705,
      "learning_rate": 4.561795283213031e-06,
      "loss": 0.0606,
      "step": 10315
    },
    {
      "epoch": 0.5327136586625355,
      "grad_norm": 8.189620018005371,
      "learning_rate": 4.56097908123135e-06,
      "loss": 0.0454,
      "step": 10316
    },
    {
      "epoch": 0.5327652982184353,
      "grad_norm": 10.104328155517578,
      "learning_rate": 4.560162891039171e-06,
      "loss": 0.075,
      "step": 10317
    },
    {
      "epoch": 0.5328169377743351,
      "grad_norm": 2.0418457984924316,
      "learning_rate": 4.55934671265841e-06,
      "loss": 0.0587,
      "step": 10318
    },
    {
      "epoch": 0.532868577330235,
      "grad_norm": 4.058380126953125,
      "learning_rate": 4.558530546110986e-06,
      "loss": 0.0516,
      "step": 10319
    },
    {
      "epoch": 0.5329202168861348,
      "grad_norm": 11.063628196716309,
      "learning_rate": 4.557714391418817e-06,
      "loss": 0.0899,
      "step": 10320
    },
    {
      "epoch": 0.5329718564420346,
      "grad_norm": 3.1540238857269287,
      "learning_rate": 4.556898248603818e-06,
      "loss": 0.0361,
      "step": 10321
    },
    {
      "epoch": 0.5330234959979344,
      "grad_norm": 3.0984835624694824,
      "learning_rate": 4.5560821176879065e-06,
      "loss": 0.06,
      "step": 10322
    },
    {
      "epoch": 0.5330751355538342,
      "grad_norm": 4.1723551750183105,
      "learning_rate": 4.555265998693e-06,
      "loss": 0.0462,
      "step": 10323
    },
    {
      "epoch": 0.533126775109734,
      "grad_norm": 10.422955513000488,
      "learning_rate": 4.554449891641015e-06,
      "loss": 0.0745,
      "step": 10324
    },
    {
      "epoch": 0.5331784146656339,
      "grad_norm": 3.7778244018554688,
      "learning_rate": 4.553633796553866e-06,
      "loss": 0.0585,
      "step": 10325
    },
    {
      "epoch": 0.5332300542215337,
      "grad_norm": 3.738999843597412,
      "learning_rate": 4.5528177134534684e-06,
      "loss": 0.0372,
      "step": 10326
    },
    {
      "epoch": 0.5332816937774335,
      "grad_norm": 12.752863883972168,
      "learning_rate": 4.552001642361738e-06,
      "loss": 0.0985,
      "step": 10327
    },
    {
      "epoch": 0.5333333333333333,
      "grad_norm": 3.727832555770874,
      "learning_rate": 4.5511855833005904e-06,
      "loss": 0.0471,
      "step": 10328
    },
    {
      "epoch": 0.5333849728892331,
      "grad_norm": 2.622981071472168,
      "learning_rate": 4.550369536291937e-06,
      "loss": 0.0577,
      "step": 10329
    },
    {
      "epoch": 0.533436612445133,
      "grad_norm": 2.4274744987487793,
      "learning_rate": 4.549553501357695e-06,
      "loss": 0.0279,
      "step": 10330
    },
    {
      "epoch": 0.5334882520010328,
      "grad_norm": 4.3314995765686035,
      "learning_rate": 4.548737478519778e-06,
      "loss": 0.0778,
      "step": 10331
    },
    {
      "epoch": 0.5335398915569326,
      "grad_norm": 7.117590427398682,
      "learning_rate": 4.547921467800099e-06,
      "loss": 0.0835,
      "step": 10332
    },
    {
      "epoch": 0.5335915311128324,
      "grad_norm": 2.1574008464813232,
      "learning_rate": 4.547105469220571e-06,
      "loss": 0.0499,
      "step": 10333
    },
    {
      "epoch": 0.5336431706687322,
      "grad_norm": 2.7667877674102783,
      "learning_rate": 4.5462894828031056e-06,
      "loss": 0.0471,
      "step": 10334
    },
    {
      "epoch": 0.533694810224632,
      "grad_norm": 1.3986648321151733,
      "learning_rate": 4.54547350856962e-06,
      "loss": 0.0208,
      "step": 10335
    },
    {
      "epoch": 0.5337464497805319,
      "grad_norm": 4.227139949798584,
      "learning_rate": 4.5446575465420204e-06,
      "loss": 0.0512,
      "step": 10336
    },
    {
      "epoch": 0.5337980893364317,
      "grad_norm": 5.001458168029785,
      "learning_rate": 4.543841596742222e-06,
      "loss": 0.0448,
      "step": 10337
    },
    {
      "epoch": 0.5338497288923315,
      "grad_norm": 3.415398120880127,
      "learning_rate": 4.543025659192136e-06,
      "loss": 0.0509,
      "step": 10338
    },
    {
      "epoch": 0.5339013684482313,
      "grad_norm": 10.042815208435059,
      "learning_rate": 4.542209733913675e-06,
      "loss": 0.1414,
      "step": 10339
    },
    {
      "epoch": 0.5339530080041311,
      "grad_norm": 2.3918628692626953,
      "learning_rate": 4.541393820928747e-06,
      "loss": 0.038,
      "step": 10340
    },
    {
      "epoch": 0.534004647560031,
      "grad_norm": 6.989917755126953,
      "learning_rate": 4.540577920259264e-06,
      "loss": 0.0645,
      "step": 10341
    },
    {
      "epoch": 0.5340562871159308,
      "grad_norm": 2.2078070640563965,
      "learning_rate": 4.539762031927139e-06,
      "loss": 0.0559,
      "step": 10342
    },
    {
      "epoch": 0.5341079266718306,
      "grad_norm": 11.000527381896973,
      "learning_rate": 4.5389461559542755e-06,
      "loss": 0.0526,
      "step": 10343
    },
    {
      "epoch": 0.5341595662277304,
      "grad_norm": 3.049975633621216,
      "learning_rate": 4.5381302923625885e-06,
      "loss": 0.0746,
      "step": 10344
    },
    {
      "epoch": 0.5342112057836302,
      "grad_norm": 2.584928035736084,
      "learning_rate": 4.5373144411739856e-06,
      "loss": 0.0415,
      "step": 10345
    },
    {
      "epoch": 0.5342628453395301,
      "grad_norm": 3.2177717685699463,
      "learning_rate": 4.536498602410377e-06,
      "loss": 0.0565,
      "step": 10346
    },
    {
      "epoch": 0.5343144848954299,
      "grad_norm": 0.9131182432174683,
      "learning_rate": 4.535682776093668e-06,
      "loss": 0.006,
      "step": 10347
    },
    {
      "epoch": 0.5343661244513297,
      "grad_norm": 6.299177646636963,
      "learning_rate": 4.534866962245769e-06,
      "loss": 0.0605,
      "step": 10348
    },
    {
      "epoch": 0.5344177640072295,
      "grad_norm": 3.630715847015381,
      "learning_rate": 4.534051160888589e-06,
      "loss": 0.0319,
      "step": 10349
    },
    {
      "epoch": 0.5344694035631293,
      "grad_norm": 3.7248470783233643,
      "learning_rate": 4.533235372044035e-06,
      "loss": 0.0596,
      "step": 10350
    },
    {
      "epoch": 0.5345210431190291,
      "grad_norm": 10.456079483032227,
      "learning_rate": 4.5324195957340125e-06,
      "loss": 0.08,
      "step": 10351
    },
    {
      "epoch": 0.534572682674929,
      "grad_norm": 2.4312243461608887,
      "learning_rate": 4.531603831980429e-06,
      "loss": 0.0371,
      "step": 10352
    },
    {
      "epoch": 0.5346243222308288,
      "grad_norm": 8.096512794494629,
      "learning_rate": 4.530788080805194e-06,
      "loss": 0.07,
      "step": 10353
    },
    {
      "epoch": 0.5346759617867286,
      "grad_norm": 8.412033081054688,
      "learning_rate": 4.52997234223021e-06,
      "loss": 0.0746,
      "step": 10354
    },
    {
      "epoch": 0.5347276013426284,
      "grad_norm": 4.315621376037598,
      "learning_rate": 4.529156616277383e-06,
      "loss": 0.0624,
      "step": 10355
    },
    {
      "epoch": 0.5347792408985282,
      "grad_norm": 5.777555465698242,
      "learning_rate": 4.528340902968621e-06,
      "loss": 0.0559,
      "step": 10356
    },
    {
      "epoch": 0.5348308804544281,
      "grad_norm": 8.40592098236084,
      "learning_rate": 4.527525202325829e-06,
      "loss": 0.1162,
      "step": 10357
    },
    {
      "epoch": 0.5348825200103279,
      "grad_norm": 2.606861114501953,
      "learning_rate": 4.52670951437091e-06,
      "loss": 0.0226,
      "step": 10358
    },
    {
      "epoch": 0.5349341595662277,
      "grad_norm": 19.201335906982422,
      "learning_rate": 4.52589383912577e-06,
      "loss": 0.1412,
      "step": 10359
    },
    {
      "epoch": 0.5349857991221275,
      "grad_norm": 5.052907466888428,
      "learning_rate": 4.525078176612311e-06,
      "loss": 0.0341,
      "step": 10360
    },
    {
      "epoch": 0.5350374386780273,
      "grad_norm": 4.722988605499268,
      "learning_rate": 4.524262526852443e-06,
      "loss": 0.0662,
      "step": 10361
    },
    {
      "epoch": 0.5350890782339272,
      "grad_norm": 2.3000786304473877,
      "learning_rate": 4.523446889868061e-06,
      "loss": 0.0457,
      "step": 10362
    },
    {
      "epoch": 0.535140717789827,
      "grad_norm": 3.196355104446411,
      "learning_rate": 4.522631265681074e-06,
      "loss": 0.067,
      "step": 10363
    },
    {
      "epoch": 0.5351923573457268,
      "grad_norm": 11.681177139282227,
      "learning_rate": 4.5218156543133834e-06,
      "loss": 0.0818,
      "step": 10364
    },
    {
      "epoch": 0.5352439969016266,
      "grad_norm": 5.621140003204346,
      "learning_rate": 4.521000055786891e-06,
      "loss": 0.0969,
      "step": 10365
    },
    {
      "epoch": 0.5352956364575264,
      "grad_norm": 3.6963491439819336,
      "learning_rate": 4.520184470123499e-06,
      "loss": 0.0492,
      "step": 10366
    },
    {
      "epoch": 0.5353472760134262,
      "grad_norm": 11.176844596862793,
      "learning_rate": 4.519368897345109e-06,
      "loss": 0.2851,
      "step": 10367
    },
    {
      "epoch": 0.5353989155693261,
      "grad_norm": 3.1525566577911377,
      "learning_rate": 4.518553337473626e-06,
      "loss": 0.0683,
      "step": 10368
    },
    {
      "epoch": 0.5354505551252259,
      "grad_norm": 3.307741165161133,
      "learning_rate": 4.517737790530946e-06,
      "loss": 0.0478,
      "step": 10369
    },
    {
      "epoch": 0.5355021946811257,
      "grad_norm": 8.900744438171387,
      "learning_rate": 4.5169222565389715e-06,
      "loss": 0.1,
      "step": 10370
    },
    {
      "epoch": 0.5355538342370255,
      "grad_norm": 7.575302600860596,
      "learning_rate": 4.516106735519603e-06,
      "loss": 0.0371,
      "step": 10371
    },
    {
      "epoch": 0.5356054737929253,
      "grad_norm": 3.2482011318206787,
      "learning_rate": 4.515291227494744e-06,
      "loss": 0.0469,
      "step": 10372
    },
    {
      "epoch": 0.5356571133488252,
      "grad_norm": 3.5368919372558594,
      "learning_rate": 4.514475732486288e-06,
      "loss": 0.0796,
      "step": 10373
    },
    {
      "epoch": 0.535708752904725,
      "grad_norm": 2.2565531730651855,
      "learning_rate": 4.513660250516138e-06,
      "loss": 0.0281,
      "step": 10374
    },
    {
      "epoch": 0.5357603924606248,
      "grad_norm": 3.6346709728240967,
      "learning_rate": 4.512844781606194e-06,
      "loss": 0.0275,
      "step": 10375
    },
    {
      "epoch": 0.5358120320165246,
      "grad_norm": 8.766057014465332,
      "learning_rate": 4.512029325778351e-06,
      "loss": 0.0512,
      "step": 10376
    },
    {
      "epoch": 0.5358636715724244,
      "grad_norm": 4.014192581176758,
      "learning_rate": 4.5112138830545106e-06,
      "loss": 0.0621,
      "step": 10377
    },
    {
      "epoch": 0.5359153111283242,
      "grad_norm": 12.59426498413086,
      "learning_rate": 4.51039845345657e-06,
      "loss": 0.1089,
      "step": 10378
    },
    {
      "epoch": 0.5359669506842241,
      "grad_norm": 3.9060189723968506,
      "learning_rate": 4.509583037006427e-06,
      "loss": 0.0338,
      "step": 10379
    },
    {
      "epoch": 0.5360185902401239,
      "grad_norm": 3.090074062347412,
      "learning_rate": 4.508767633725976e-06,
      "loss": 0.0327,
      "step": 10380
    },
    {
      "epoch": 0.5360702297960237,
      "grad_norm": 1.4125936031341553,
      "learning_rate": 4.507952243637117e-06,
      "loss": 0.0184,
      "step": 10381
    },
    {
      "epoch": 0.5361218693519235,
      "grad_norm": 3.709813117980957,
      "learning_rate": 4.507136866761746e-06,
      "loss": 0.0639,
      "step": 10382
    },
    {
      "epoch": 0.5361735089078233,
      "grad_norm": 4.444406986236572,
      "learning_rate": 4.50632150312176e-06,
      "loss": 0.0795,
      "step": 10383
    },
    {
      "epoch": 0.5362251484637233,
      "grad_norm": 5.007440567016602,
      "learning_rate": 4.505506152739053e-06,
      "loss": 0.0517,
      "step": 10384
    },
    {
      "epoch": 0.5362767880196231,
      "grad_norm": 2.3439888954162598,
      "learning_rate": 4.504690815635521e-06,
      "loss": 0.0498,
      "step": 10385
    },
    {
      "epoch": 0.5363284275755229,
      "grad_norm": 3.615445137023926,
      "learning_rate": 4.50387549183306e-06,
      "loss": 0.0837,
      "step": 10386
    },
    {
      "epoch": 0.5363800671314227,
      "grad_norm": 1.852551817893982,
      "learning_rate": 4.503060181353564e-06,
      "loss": 0.0258,
      "step": 10387
    },
    {
      "epoch": 0.5364317066873225,
      "grad_norm": 2.3211703300476074,
      "learning_rate": 4.5022448842189254e-06,
      "loss": 0.0407,
      "step": 10388
    },
    {
      "epoch": 0.5364833462432224,
      "grad_norm": 3.406791925430298,
      "learning_rate": 4.501429600451042e-06,
      "loss": 0.0485,
      "step": 10389
    },
    {
      "epoch": 0.5365349857991222,
      "grad_norm": 3.973168134689331,
      "learning_rate": 4.500614330071807e-06,
      "loss": 0.0492,
      "step": 10390
    },
    {
      "epoch": 0.536586625355022,
      "grad_norm": 4.146574974060059,
      "learning_rate": 4.499799073103112e-06,
      "loss": 0.07,
      "step": 10391
    },
    {
      "epoch": 0.5366382649109218,
      "grad_norm": 5.265380859375,
      "learning_rate": 4.49898382956685e-06,
      "loss": 0.0605,
      "step": 10392
    },
    {
      "epoch": 0.5366899044668216,
      "grad_norm": 5.732239246368408,
      "learning_rate": 4.4981685994849125e-06,
      "loss": 0.1176,
      "step": 10393
    },
    {
      "epoch": 0.5367415440227215,
      "grad_norm": 5.3816237449646,
      "learning_rate": 4.497353382879198e-06,
      "loss": 0.0707,
      "step": 10394
    },
    {
      "epoch": 0.5367931835786213,
      "grad_norm": 2.5111374855041504,
      "learning_rate": 4.49653817977159e-06,
      "loss": 0.028,
      "step": 10395
    },
    {
      "epoch": 0.5368448231345211,
      "grad_norm": 2.973123550415039,
      "learning_rate": 4.495722990183985e-06,
      "loss": 0.0451,
      "step": 10396
    },
    {
      "epoch": 0.5368964626904209,
      "grad_norm": 2.813530445098877,
      "learning_rate": 4.4949078141382744e-06,
      "loss": 0.0399,
      "step": 10397
    },
    {
      "epoch": 0.5369481022463207,
      "grad_norm": 3.2951087951660156,
      "learning_rate": 4.494092651656346e-06,
      "loss": 0.0415,
      "step": 10398
    },
    {
      "epoch": 0.5369997418022205,
      "grad_norm": 3.6401283740997314,
      "learning_rate": 4.493277502760092e-06,
      "loss": 0.0395,
      "step": 10399
    },
    {
      "epoch": 0.5370513813581204,
      "grad_norm": 25.158462524414062,
      "learning_rate": 4.492462367471401e-06,
      "loss": 0.1819,
      "step": 10400
    },
    {
      "epoch": 0.5371030209140202,
      "grad_norm": 8.166300773620605,
      "learning_rate": 4.491647245812169e-06,
      "loss": 0.0918,
      "step": 10401
    },
    {
      "epoch": 0.53715466046992,
      "grad_norm": 16.629865646362305,
      "learning_rate": 4.490832137804275e-06,
      "loss": 0.0456,
      "step": 10402
    },
    {
      "epoch": 0.5372063000258198,
      "grad_norm": 2.373434066772461,
      "learning_rate": 4.490017043469616e-06,
      "loss": 0.0269,
      "step": 10403
    },
    {
      "epoch": 0.5372579395817196,
      "grad_norm": 3.2012245655059814,
      "learning_rate": 4.4892019628300765e-06,
      "loss": 0.0565,
      "step": 10404
    },
    {
      "epoch": 0.5373095791376195,
      "grad_norm": 6.8422136306762695,
      "learning_rate": 4.488386895907548e-06,
      "loss": 0.0481,
      "step": 10405
    },
    {
      "epoch": 0.5373612186935193,
      "grad_norm": 5.059622764587402,
      "learning_rate": 4.487571842723916e-06,
      "loss": 0.0802,
      "step": 10406
    },
    {
      "epoch": 0.5374128582494191,
      "grad_norm": 3.705444812774658,
      "learning_rate": 4.486756803301067e-06,
      "loss": 0.0907,
      "step": 10407
    },
    {
      "epoch": 0.5374644978053189,
      "grad_norm": 3.072267770767212,
      "learning_rate": 4.485941777660892e-06,
      "loss": 0.0465,
      "step": 10408
    },
    {
      "epoch": 0.5375161373612187,
      "grad_norm": 7.631485939025879,
      "learning_rate": 4.4851267658252736e-06,
      "loss": 0.0534,
      "step": 10409
    },
    {
      "epoch": 0.5375677769171185,
      "grad_norm": 5.402566909790039,
      "learning_rate": 4.484311767816101e-06,
      "loss": 0.0303,
      "step": 10410
    },
    {
      "epoch": 0.5376194164730184,
      "grad_norm": 4.124183177947998,
      "learning_rate": 4.483496783655258e-06,
      "loss": 0.0449,
      "step": 10411
    },
    {
      "epoch": 0.5376710560289182,
      "grad_norm": 4.9545488357543945,
      "learning_rate": 4.4826818133646336e-06,
      "loss": 0.0928,
      "step": 10412
    },
    {
      "epoch": 0.537722695584818,
      "grad_norm": 8.522509574890137,
      "learning_rate": 4.48186685696611e-06,
      "loss": 0.0558,
      "step": 10413
    },
    {
      "epoch": 0.5377743351407178,
      "grad_norm": 3.575145959854126,
      "learning_rate": 4.481051914481572e-06,
      "loss": 0.0506,
      "step": 10414
    },
    {
      "epoch": 0.5378259746966176,
      "grad_norm": 12.325858116149902,
      "learning_rate": 4.480236985932908e-06,
      "loss": 0.1706,
      "step": 10415
    },
    {
      "epoch": 0.5378776142525175,
      "grad_norm": 2.5592846870422363,
      "learning_rate": 4.479422071341996e-06,
      "loss": 0.047,
      "step": 10416
    },
    {
      "epoch": 0.5379292538084173,
      "grad_norm": 18.483707427978516,
      "learning_rate": 4.478607170730725e-06,
      "loss": 0.1205,
      "step": 10417
    },
    {
      "epoch": 0.5379808933643171,
      "grad_norm": 2.8158621788024902,
      "learning_rate": 4.477792284120977e-06,
      "loss": 0.0365,
      "step": 10418
    },
    {
      "epoch": 0.5380325329202169,
      "grad_norm": 1.749864935874939,
      "learning_rate": 4.4769774115346345e-06,
      "loss": 0.0371,
      "step": 10419
    },
    {
      "epoch": 0.5380841724761167,
      "grad_norm": 16.47124671936035,
      "learning_rate": 4.476162552993579e-06,
      "loss": 0.1755,
      "step": 10420
    },
    {
      "epoch": 0.5381358120320165,
      "grad_norm": 3.8538761138916016,
      "learning_rate": 4.475347708519694e-06,
      "loss": 0.0694,
      "step": 10421
    },
    {
      "epoch": 0.5381874515879164,
      "grad_norm": 3.044057607650757,
      "learning_rate": 4.474532878134861e-06,
      "loss": 0.0347,
      "step": 10422
    },
    {
      "epoch": 0.5382390911438162,
      "grad_norm": 9.98094654083252,
      "learning_rate": 4.4737180618609635e-06,
      "loss": 0.0916,
      "step": 10423
    },
    {
      "epoch": 0.538290730699716,
      "grad_norm": 2.037360191345215,
      "learning_rate": 4.472903259719881e-06,
      "loss": 0.0508,
      "step": 10424
    },
    {
      "epoch": 0.5383423702556158,
      "grad_norm": 3.613231897354126,
      "learning_rate": 4.472088471733493e-06,
      "loss": 0.0384,
      "step": 10425
    },
    {
      "epoch": 0.5383940098115156,
      "grad_norm": 8.144108772277832,
      "learning_rate": 4.471273697923683e-06,
      "loss": 0.0823,
      "step": 10426
    },
    {
      "epoch": 0.5384456493674155,
      "grad_norm": 3.06264066696167,
      "learning_rate": 4.470458938312328e-06,
      "loss": 0.0545,
      "step": 10427
    },
    {
      "epoch": 0.5384972889233153,
      "grad_norm": 2.476052761077881,
      "learning_rate": 4.469644192921307e-06,
      "loss": 0.0535,
      "step": 10428
    },
    {
      "epoch": 0.5385489284792151,
      "grad_norm": 2.2898495197296143,
      "learning_rate": 4.468829461772504e-06,
      "loss": 0.0309,
      "step": 10429
    },
    {
      "epoch": 0.5386005680351149,
      "grad_norm": 1.7801182270050049,
      "learning_rate": 4.468014744887794e-06,
      "loss": 0.0389,
      "step": 10430
    },
    {
      "epoch": 0.5386522075910147,
      "grad_norm": 10.559471130371094,
      "learning_rate": 4.467200042289056e-06,
      "loss": 0.0799,
      "step": 10431
    },
    {
      "epoch": 0.5387038471469145,
      "grad_norm": 5.134768009185791,
      "learning_rate": 4.466385353998168e-06,
      "loss": 0.0586,
      "step": 10432
    },
    {
      "epoch": 0.5387554867028144,
      "grad_norm": 3.3842999935150146,
      "learning_rate": 4.465570680037008e-06,
      "loss": 0.0617,
      "step": 10433
    },
    {
      "epoch": 0.5388071262587142,
      "grad_norm": 13.50419807434082,
      "learning_rate": 4.464756020427456e-06,
      "loss": 0.1111,
      "step": 10434
    },
    {
      "epoch": 0.538858765814614,
      "grad_norm": 3.947345733642578,
      "learning_rate": 4.463941375191383e-06,
      "loss": 0.058,
      "step": 10435
    },
    {
      "epoch": 0.5389104053705138,
      "grad_norm": 2.5044567584991455,
      "learning_rate": 4.463126744350671e-06,
      "loss": 0.0374,
      "step": 10436
    },
    {
      "epoch": 0.5389620449264136,
      "grad_norm": 5.427083492279053,
      "learning_rate": 4.4623121279271945e-06,
      "loss": 0.0233,
      "step": 10437
    },
    {
      "epoch": 0.5390136844823135,
      "grad_norm": 3.99684739112854,
      "learning_rate": 4.461497525942828e-06,
      "loss": 0.0317,
      "step": 10438
    },
    {
      "epoch": 0.5390653240382133,
      "grad_norm": 4.864384174346924,
      "learning_rate": 4.460682938419447e-06,
      "loss": 0.0855,
      "step": 10439
    },
    {
      "epoch": 0.5391169635941131,
      "grad_norm": 5.78541898727417,
      "learning_rate": 4.459868365378927e-06,
      "loss": 0.0271,
      "step": 10440
    },
    {
      "epoch": 0.5391686031500129,
      "grad_norm": 7.947640895843506,
      "learning_rate": 4.459053806843146e-06,
      "loss": 0.0803,
      "step": 10441
    },
    {
      "epoch": 0.5392202427059127,
      "grad_norm": 5.546284198760986,
      "learning_rate": 4.458239262833971e-06,
      "loss": 0.0514,
      "step": 10442
    },
    {
      "epoch": 0.5392718822618126,
      "grad_norm": 3.877138614654541,
      "learning_rate": 4.457424733373282e-06,
      "loss": 0.0503,
      "step": 10443
    },
    {
      "epoch": 0.5393235218177124,
      "grad_norm": 5.492893695831299,
      "learning_rate": 4.45661021848295e-06,
      "loss": 0.0679,
      "step": 10444
    },
    {
      "epoch": 0.5393751613736122,
      "grad_norm": 2.5764176845550537,
      "learning_rate": 4.4557957181848485e-06,
      "loss": 0.0256,
      "step": 10445
    },
    {
      "epoch": 0.539426800929512,
      "grad_norm": 3.166722536087036,
      "learning_rate": 4.45498123250085e-06,
      "loss": 0.0515,
      "step": 10446
    },
    {
      "epoch": 0.5394784404854118,
      "grad_norm": 1.4951733350753784,
      "learning_rate": 4.454166761452826e-06,
      "loss": 0.029,
      "step": 10447
    },
    {
      "epoch": 0.5395300800413116,
      "grad_norm": 10.800968170166016,
      "learning_rate": 4.453352305062651e-06,
      "loss": 0.0893,
      "step": 10448
    },
    {
      "epoch": 0.5395817195972115,
      "grad_norm": 6.997433185577393,
      "learning_rate": 4.452537863352193e-06,
      "loss": 0.0389,
      "step": 10449
    },
    {
      "epoch": 0.5396333591531113,
      "grad_norm": 7.674325466156006,
      "learning_rate": 4.4517234363433244e-06,
      "loss": 0.0586,
      "step": 10450
    },
    {
      "epoch": 0.5396849987090111,
      "grad_norm": 5.850718975067139,
      "learning_rate": 4.450909024057917e-06,
      "loss": 0.0666,
      "step": 10451
    },
    {
      "epoch": 0.5397366382649109,
      "grad_norm": 3.0474672317504883,
      "learning_rate": 4.450094626517841e-06,
      "loss": 0.0327,
      "step": 10452
    },
    {
      "epoch": 0.5397882778208107,
      "grad_norm": 10.079898834228516,
      "learning_rate": 4.449280243744965e-06,
      "loss": 0.1169,
      "step": 10453
    },
    {
      "epoch": 0.5398399173767106,
      "grad_norm": 4.365137100219727,
      "learning_rate": 4.448465875761158e-06,
      "loss": 0.0823,
      "step": 10454
    },
    {
      "epoch": 0.5398915569326104,
      "grad_norm": 5.159461975097656,
      "learning_rate": 4.447651522588291e-06,
      "loss": 0.0612,
      "step": 10455
    },
    {
      "epoch": 0.5399431964885102,
      "grad_norm": 6.71118688583374,
      "learning_rate": 4.446837184248233e-06,
      "loss": 0.0516,
      "step": 10456
    },
    {
      "epoch": 0.53999483604441,
      "grad_norm": 7.750716686248779,
      "learning_rate": 4.4460228607628495e-06,
      "loss": 0.0745,
      "step": 10457
    },
    {
      "epoch": 0.5400464756003098,
      "grad_norm": 2.475710868835449,
      "learning_rate": 4.445208552154011e-06,
      "loss": 0.0501,
      "step": 10458
    },
    {
      "epoch": 0.5400981151562096,
      "grad_norm": 6.733859539031982,
      "learning_rate": 4.444394258443584e-06,
      "loss": 0.0698,
      "step": 10459
    },
    {
      "epoch": 0.5401497547121095,
      "grad_norm": 3.8220152854919434,
      "learning_rate": 4.4435799796534355e-06,
      "loss": 0.0743,
      "step": 10460
    },
    {
      "epoch": 0.5402013942680093,
      "grad_norm": 4.432066917419434,
      "learning_rate": 4.442765715805431e-06,
      "loss": 0.0492,
      "step": 10461
    },
    {
      "epoch": 0.5402530338239091,
      "grad_norm": 5.394270896911621,
      "learning_rate": 4.44195146692144e-06,
      "loss": 0.055,
      "step": 10462
    },
    {
      "epoch": 0.5403046733798089,
      "grad_norm": 17.33057975769043,
      "learning_rate": 4.441137233023327e-06,
      "loss": 0.142,
      "step": 10463
    },
    {
      "epoch": 0.5403563129357087,
      "grad_norm": 6.26398229598999,
      "learning_rate": 4.440323014132956e-06,
      "loss": 0.0425,
      "step": 10464
    },
    {
      "epoch": 0.5404079524916086,
      "grad_norm": 3.613935947418213,
      "learning_rate": 4.439508810272193e-06,
      "loss": 0.0327,
      "step": 10465
    },
    {
      "epoch": 0.5404595920475084,
      "grad_norm": 3.8068978786468506,
      "learning_rate": 4.438694621462902e-06,
      "loss": 0.0456,
      "step": 10466
    },
    {
      "epoch": 0.5405112316034082,
      "grad_norm": 2.953941822052002,
      "learning_rate": 4.437880447726951e-06,
      "loss": 0.0429,
      "step": 10467
    },
    {
      "epoch": 0.540562871159308,
      "grad_norm": 3.3140885829925537,
      "learning_rate": 4.4370662890861986e-06,
      "loss": 0.0701,
      "step": 10468
    },
    {
      "epoch": 0.5406145107152078,
      "grad_norm": 5.1000871658325195,
      "learning_rate": 4.436252145562511e-06,
      "loss": 0.0439,
      "step": 10469
    },
    {
      "epoch": 0.5406661502711076,
      "grad_norm": 6.263146877288818,
      "learning_rate": 4.4354380171777525e-06,
      "loss": 0.0644,
      "step": 10470
    },
    {
      "epoch": 0.5407177898270075,
      "grad_norm": 3.881574869155884,
      "learning_rate": 4.434623903953783e-06,
      "loss": 0.0608,
      "step": 10471
    },
    {
      "epoch": 0.5407694293829073,
      "grad_norm": 3.855013370513916,
      "learning_rate": 4.433809805912467e-06,
      "loss": 0.0482,
      "step": 10472
    },
    {
      "epoch": 0.5408210689388071,
      "grad_norm": 3.1671664714813232,
      "learning_rate": 4.432995723075663e-06,
      "loss": 0.0534,
      "step": 10473
    },
    {
      "epoch": 0.5408727084947069,
      "grad_norm": 2.736725091934204,
      "learning_rate": 4.4321816554652385e-06,
      "loss": 0.0577,
      "step": 10474
    },
    {
      "epoch": 0.5409243480506067,
      "grad_norm": 3.3945605754852295,
      "learning_rate": 4.4313676031030474e-06,
      "loss": 0.0633,
      "step": 10475
    },
    {
      "epoch": 0.5409759876065066,
      "grad_norm": 2.1347930431365967,
      "learning_rate": 4.430553566010955e-06,
      "loss": 0.0385,
      "step": 10476
    },
    {
      "epoch": 0.5410276271624064,
      "grad_norm": 4.979456424713135,
      "learning_rate": 4.429739544210821e-06,
      "loss": 0.0549,
      "step": 10477
    },
    {
      "epoch": 0.5410792667183062,
      "grad_norm": 6.034342288970947,
      "learning_rate": 4.428925537724505e-06,
      "loss": 0.0557,
      "step": 10478
    },
    {
      "epoch": 0.541130906274206,
      "grad_norm": 3.193112850189209,
      "learning_rate": 4.428111546573865e-06,
      "loss": 0.0522,
      "step": 10479
    },
    {
      "epoch": 0.5411825458301058,
      "grad_norm": 7.545247554779053,
      "learning_rate": 4.42729757078076e-06,
      "loss": 0.0992,
      "step": 10480
    },
    {
      "epoch": 0.5412341853860057,
      "grad_norm": 5.140064239501953,
      "learning_rate": 4.426483610367052e-06,
      "loss": 0.0355,
      "step": 10481
    },
    {
      "epoch": 0.5412858249419055,
      "grad_norm": 9.72088623046875,
      "learning_rate": 4.425669665354594e-06,
      "loss": 0.0735,
      "step": 10482
    },
    {
      "epoch": 0.5413374644978053,
      "grad_norm": 3.3873956203460693,
      "learning_rate": 4.424855735765247e-06,
      "loss": 0.0609,
      "step": 10483
    },
    {
      "epoch": 0.5413891040537051,
      "grad_norm": 4.481218338012695,
      "learning_rate": 4.424041821620867e-06,
      "loss": 0.0471,
      "step": 10484
    },
    {
      "epoch": 0.5414407436096049,
      "grad_norm": 4.899045467376709,
      "learning_rate": 4.423227922943313e-06,
      "loss": 0.0397,
      "step": 10485
    },
    {
      "epoch": 0.5414923831655047,
      "grad_norm": 6.038925647735596,
      "learning_rate": 4.4224140397544395e-06,
      "loss": 0.0734,
      "step": 10486
    },
    {
      "epoch": 0.5415440227214046,
      "grad_norm": 6.026084899902344,
      "learning_rate": 4.421600172076102e-06,
      "loss": 0.0473,
      "step": 10487
    },
    {
      "epoch": 0.5415956622773044,
      "grad_norm": 6.106513977050781,
      "learning_rate": 4.420786319930158e-06,
      "loss": 0.0513,
      "step": 10488
    },
    {
      "epoch": 0.5416473018332042,
      "grad_norm": 3.7477099895477295,
      "learning_rate": 4.419972483338464e-06,
      "loss": 0.0336,
      "step": 10489
    },
    {
      "epoch": 0.541698941389104,
      "grad_norm": 4.519890785217285,
      "learning_rate": 4.419158662322871e-06,
      "loss": 0.0648,
      "step": 10490
    },
    {
      "epoch": 0.5417505809450038,
      "grad_norm": 7.672455310821533,
      "learning_rate": 4.418344856905236e-06,
      "loss": 0.0536,
      "step": 10491
    },
    {
      "epoch": 0.5418022205009037,
      "grad_norm": 7.64317512512207,
      "learning_rate": 4.417531067107412e-06,
      "loss": 0.0687,
      "step": 10492
    },
    {
      "epoch": 0.5418538600568035,
      "grad_norm": 18.818756103515625,
      "learning_rate": 4.416717292951253e-06,
      "loss": 0.156,
      "step": 10493
    },
    {
      "epoch": 0.5419054996127033,
      "grad_norm": 2.6531999111175537,
      "learning_rate": 4.415903534458611e-06,
      "loss": 0.0423,
      "step": 10494
    },
    {
      "epoch": 0.5419571391686031,
      "grad_norm": 5.526663780212402,
      "learning_rate": 4.41508979165134e-06,
      "loss": 0.1357,
      "step": 10495
    },
    {
      "epoch": 0.5420087787245029,
      "grad_norm": 2.9899277687072754,
      "learning_rate": 4.414276064551293e-06,
      "loss": 0.0269,
      "step": 10496
    },
    {
      "epoch": 0.5420604182804027,
      "grad_norm": 3.122546911239624,
      "learning_rate": 4.413462353180321e-06,
      "loss": 0.0336,
      "step": 10497
    },
    {
      "epoch": 0.5421120578363026,
      "grad_norm": 9.226901054382324,
      "learning_rate": 4.412648657560275e-06,
      "loss": 0.0918,
      "step": 10498
    },
    {
      "epoch": 0.5421636973922024,
      "grad_norm": 1.7507399320602417,
      "learning_rate": 4.411834977713005e-06,
      "loss": 0.02,
      "step": 10499
    },
    {
      "epoch": 0.5422153369481022,
      "grad_norm": 4.673086166381836,
      "learning_rate": 4.411021313660367e-06,
      "loss": 0.0694,
      "step": 10500
    },
    {
      "epoch": 0.542266976504002,
      "grad_norm": 6.540033340454102,
      "learning_rate": 4.4102076654242025e-06,
      "loss": 0.0955,
      "step": 10501
    },
    {
      "epoch": 0.5423186160599018,
      "grad_norm": 6.540170669555664,
      "learning_rate": 4.4093940330263685e-06,
      "loss": 0.0571,
      "step": 10502
    },
    {
      "epoch": 0.5423702556158017,
      "grad_norm": 7.461846351623535,
      "learning_rate": 4.408580416488712e-06,
      "loss": 0.126,
      "step": 10503
    },
    {
      "epoch": 0.5424218951717015,
      "grad_norm": 10.31085205078125,
      "learning_rate": 4.4077668158330814e-06,
      "loss": 0.1633,
      "step": 10504
    },
    {
      "epoch": 0.5424735347276013,
      "grad_norm": 4.1097493171691895,
      "learning_rate": 4.406953231081326e-06,
      "loss": 0.0484,
      "step": 10505
    },
    {
      "epoch": 0.5425251742835011,
      "grad_norm": 6.518310070037842,
      "learning_rate": 4.4061396622552925e-06,
      "loss": 0.0706,
      "step": 10506
    },
    {
      "epoch": 0.5425768138394009,
      "grad_norm": 4.058192729949951,
      "learning_rate": 4.405326109376832e-06,
      "loss": 0.0372,
      "step": 10507
    },
    {
      "epoch": 0.5426284533953007,
      "grad_norm": 5.7928619384765625,
      "learning_rate": 4.404512572467786e-06,
      "loss": 0.0801,
      "step": 10508
    },
    {
      "epoch": 0.5426800929512006,
      "grad_norm": 3.9282290935516357,
      "learning_rate": 4.4036990515500055e-06,
      "loss": 0.0278,
      "step": 10509
    },
    {
      "epoch": 0.5427317325071004,
      "grad_norm": 3.8035802841186523,
      "learning_rate": 4.402885546645336e-06,
      "loss": 0.0679,
      "step": 10510
    },
    {
      "epoch": 0.5427833720630003,
      "grad_norm": 1.7425315380096436,
      "learning_rate": 4.402072057775625e-06,
      "loss": 0.0204,
      "step": 10511
    },
    {
      "epoch": 0.5428350116189001,
      "grad_norm": 9.112418174743652,
      "learning_rate": 4.401258584962714e-06,
      "loss": 0.0653,
      "step": 10512
    },
    {
      "epoch": 0.5428866511748,
      "grad_norm": 3.3002493381500244,
      "learning_rate": 4.40044512822845e-06,
      "loss": 0.0262,
      "step": 10513
    },
    {
      "epoch": 0.5429382907306998,
      "grad_norm": 4.809137344360352,
      "learning_rate": 4.399631687594681e-06,
      "loss": 0.0596,
      "step": 10514
    },
    {
      "epoch": 0.5429899302865996,
      "grad_norm": 2.305962562561035,
      "learning_rate": 4.3988182630832455e-06,
      "loss": 0.0265,
      "step": 10515
    },
    {
      "epoch": 0.5430415698424994,
      "grad_norm": 3.9304025173187256,
      "learning_rate": 4.39800485471599e-06,
      "loss": 0.0528,
      "step": 10516
    },
    {
      "epoch": 0.5430932093983992,
      "grad_norm": 9.512884140014648,
      "learning_rate": 4.397191462514759e-06,
      "loss": 0.0652,
      "step": 10517
    },
    {
      "epoch": 0.543144848954299,
      "grad_norm": 7.9552154541015625,
      "learning_rate": 4.396378086501394e-06,
      "loss": 0.0752,
      "step": 10518
    },
    {
      "epoch": 0.5431964885101989,
      "grad_norm": 8.21516227722168,
      "learning_rate": 4.395564726697737e-06,
      "loss": 0.0491,
      "step": 10519
    },
    {
      "epoch": 0.5432481280660987,
      "grad_norm": 6.861050605773926,
      "learning_rate": 4.3947513831256296e-06,
      "loss": 0.0696,
      "step": 10520
    },
    {
      "epoch": 0.5432997676219985,
      "grad_norm": 4.165393352508545,
      "learning_rate": 4.393938055806916e-06,
      "loss": 0.085,
      "step": 10521
    },
    {
      "epoch": 0.5433514071778983,
      "grad_norm": 3.377237319946289,
      "learning_rate": 4.393124744763435e-06,
      "loss": 0.0564,
      "step": 10522
    },
    {
      "epoch": 0.5434030467337981,
      "grad_norm": 3.225958824157715,
      "learning_rate": 4.392311450017029e-06,
      "loss": 0.0593,
      "step": 10523
    },
    {
      "epoch": 0.543454686289698,
      "grad_norm": 3.1462652683258057,
      "learning_rate": 4.391498171589536e-06,
      "loss": 0.043,
      "step": 10524
    },
    {
      "epoch": 0.5435063258455978,
      "grad_norm": 3.27714467048645,
      "learning_rate": 4.3906849095028e-06,
      "loss": 0.0505,
      "step": 10525
    },
    {
      "epoch": 0.5435579654014976,
      "grad_norm": 7.300926208496094,
      "learning_rate": 4.389871663778656e-06,
      "loss": 0.1106,
      "step": 10526
    },
    {
      "epoch": 0.5436096049573974,
      "grad_norm": 4.479794502258301,
      "learning_rate": 4.389058434438942e-06,
      "loss": 0.0667,
      "step": 10527
    },
    {
      "epoch": 0.5436612445132972,
      "grad_norm": 3.161133050918579,
      "learning_rate": 4.388245221505502e-06,
      "loss": 0.0356,
      "step": 10528
    },
    {
      "epoch": 0.543712884069197,
      "grad_norm": 10.638076782226562,
      "learning_rate": 4.38743202500017e-06,
      "loss": 0.1542,
      "step": 10529
    },
    {
      "epoch": 0.5437645236250969,
      "grad_norm": 3.4377779960632324,
      "learning_rate": 4.386618844944786e-06,
      "loss": 0.0462,
      "step": 10530
    },
    {
      "epoch": 0.5438161631809967,
      "grad_norm": 4.28433084487915,
      "learning_rate": 4.385805681361185e-06,
      "loss": 0.0904,
      "step": 10531
    },
    {
      "epoch": 0.5438678027368965,
      "grad_norm": 1.4069538116455078,
      "learning_rate": 4.3849925342712045e-06,
      "loss": 0.0176,
      "step": 10532
    },
    {
      "epoch": 0.5439194422927963,
      "grad_norm": 2.5760064125061035,
      "learning_rate": 4.384179403696683e-06,
      "loss": 0.0548,
      "step": 10533
    },
    {
      "epoch": 0.5439710818486961,
      "grad_norm": 6.727184295654297,
      "learning_rate": 4.383366289659452e-06,
      "loss": 0.0841,
      "step": 10534
    },
    {
      "epoch": 0.544022721404596,
      "grad_norm": 5.351390838623047,
      "learning_rate": 4.38255319218135e-06,
      "loss": 0.066,
      "step": 10535
    },
    {
      "epoch": 0.5440743609604958,
      "grad_norm": 3.510723352432251,
      "learning_rate": 4.381740111284214e-06,
      "loss": 0.0453,
      "step": 10536
    },
    {
      "epoch": 0.5441260005163956,
      "grad_norm": 5.119719505310059,
      "learning_rate": 4.380927046989873e-06,
      "loss": 0.0611,
      "step": 10537
    },
    {
      "epoch": 0.5441776400722954,
      "grad_norm": 5.602949142456055,
      "learning_rate": 4.380113999320164e-06,
      "loss": 0.0719,
      "step": 10538
    },
    {
      "epoch": 0.5442292796281952,
      "grad_norm": 1.814536213874817,
      "learning_rate": 4.3793009682969205e-06,
      "loss": 0.0434,
      "step": 10539
    },
    {
      "epoch": 0.544280919184095,
      "grad_norm": 4.136346340179443,
      "learning_rate": 4.378487953941979e-06,
      "loss": 0.0814,
      "step": 10540
    },
    {
      "epoch": 0.5443325587399949,
      "grad_norm": 12.940081596374512,
      "learning_rate": 4.377674956277166e-06,
      "loss": 0.1689,
      "step": 10541
    },
    {
      "epoch": 0.5443841982958947,
      "grad_norm": 3.7360992431640625,
      "learning_rate": 4.376861975324317e-06,
      "loss": 0.0364,
      "step": 10542
    },
    {
      "epoch": 0.5444358378517945,
      "grad_norm": 4.097688674926758,
      "learning_rate": 4.376049011105264e-06,
      "loss": 0.0381,
      "step": 10543
    },
    {
      "epoch": 0.5444874774076943,
      "grad_norm": 5.061056137084961,
      "learning_rate": 4.37523606364184e-06,
      "loss": 0.0743,
      "step": 10544
    },
    {
      "epoch": 0.5445391169635941,
      "grad_norm": 2.022733688354492,
      "learning_rate": 4.374423132955873e-06,
      "loss": 0.0554,
      "step": 10545
    },
    {
      "epoch": 0.544590756519494,
      "grad_norm": 6.81657600402832,
      "learning_rate": 4.373610219069192e-06,
      "loss": 0.0568,
      "step": 10546
    },
    {
      "epoch": 0.5446423960753938,
      "grad_norm": 2.700662612915039,
      "learning_rate": 4.372797322003635e-06,
      "loss": 0.0355,
      "step": 10547
    },
    {
      "epoch": 0.5446940356312936,
      "grad_norm": 9.451186180114746,
      "learning_rate": 4.3719844417810216e-06,
      "loss": 0.0948,
      "step": 10548
    },
    {
      "epoch": 0.5447456751871934,
      "grad_norm": 8.905359268188477,
      "learning_rate": 4.371171578423187e-06,
      "loss": 0.1713,
      "step": 10549
    },
    {
      "epoch": 0.5447973147430932,
      "grad_norm": 8.801847457885742,
      "learning_rate": 4.370358731951958e-06,
      "loss": 0.0945,
      "step": 10550
    },
    {
      "epoch": 0.544848954298993,
      "grad_norm": 2.9842779636383057,
      "learning_rate": 4.369545902389165e-06,
      "loss": 0.0432,
      "step": 10551
    },
    {
      "epoch": 0.5449005938548929,
      "grad_norm": 1.6876486539840698,
      "learning_rate": 4.368733089756634e-06,
      "loss": 0.0305,
      "step": 10552
    },
    {
      "epoch": 0.5449522334107927,
      "grad_norm": 5.007664203643799,
      "learning_rate": 4.367920294076189e-06,
      "loss": 0.0514,
      "step": 10553
    },
    {
      "epoch": 0.5450038729666925,
      "grad_norm": 2.222515344619751,
      "learning_rate": 4.367107515369663e-06,
      "loss": 0.032,
      "step": 10554
    },
    {
      "epoch": 0.5450555125225923,
      "grad_norm": 4.334994316101074,
      "learning_rate": 4.366294753658881e-06,
      "loss": 0.0436,
      "step": 10555
    },
    {
      "epoch": 0.5451071520784921,
      "grad_norm": 2.1476287841796875,
      "learning_rate": 4.365482008965666e-06,
      "loss": 0.0211,
      "step": 10556
    },
    {
      "epoch": 0.545158791634392,
      "grad_norm": 14.659266471862793,
      "learning_rate": 4.364669281311844e-06,
      "loss": 0.0764,
      "step": 10557
    },
    {
      "epoch": 0.5452104311902918,
      "grad_norm": 3.137458562850952,
      "learning_rate": 4.363856570719244e-06,
      "loss": 0.0414,
      "step": 10558
    },
    {
      "epoch": 0.5452620707461916,
      "grad_norm": 15.711682319641113,
      "learning_rate": 4.3630438772096875e-06,
      "loss": 0.0445,
      "step": 10559
    },
    {
      "epoch": 0.5453137103020914,
      "grad_norm": 1.5300235748291016,
      "learning_rate": 4.362231200804996e-06,
      "loss": 0.0223,
      "step": 10560
    },
    {
      "epoch": 0.5453653498579912,
      "grad_norm": 2.398200511932373,
      "learning_rate": 4.361418541526998e-06,
      "loss": 0.0304,
      "step": 10561
    },
    {
      "epoch": 0.545416989413891,
      "grad_norm": 9.94561767578125,
      "learning_rate": 4.3606058993975165e-06,
      "loss": 0.0838,
      "step": 10562
    },
    {
      "epoch": 0.5454686289697909,
      "grad_norm": 2.331735849380493,
      "learning_rate": 4.359793274438371e-06,
      "loss": 0.0356,
      "step": 10563
    },
    {
      "epoch": 0.5455202685256907,
      "grad_norm": 3.0597054958343506,
      "learning_rate": 4.3589806666713855e-06,
      "loss": 0.0491,
      "step": 10564
    },
    {
      "epoch": 0.5455719080815905,
      "grad_norm": 3.392890691757202,
      "learning_rate": 4.35816807611838e-06,
      "loss": 0.0317,
      "step": 10565
    },
    {
      "epoch": 0.5456235476374903,
      "grad_norm": 4.733499526977539,
      "learning_rate": 4.357355502801182e-06,
      "loss": 0.0745,
      "step": 10566
    },
    {
      "epoch": 0.5456751871933901,
      "grad_norm": 3.0957446098327637,
      "learning_rate": 4.356542946741604e-06,
      "loss": 0.0654,
      "step": 10567
    },
    {
      "epoch": 0.54572682674929,
      "grad_norm": 7.742406845092773,
      "learning_rate": 4.355730407961472e-06,
      "loss": 0.0841,
      "step": 10568
    },
    {
      "epoch": 0.5457784663051898,
      "grad_norm": 3.0347330570220947,
      "learning_rate": 4.354917886482606e-06,
      "loss": 0.0608,
      "step": 10569
    },
    {
      "epoch": 0.5458301058610896,
      "grad_norm": 2.0691964626312256,
      "learning_rate": 4.354105382326822e-06,
      "loss": 0.0358,
      "step": 10570
    },
    {
      "epoch": 0.5458817454169894,
      "grad_norm": 5.038316249847412,
      "learning_rate": 4.353292895515941e-06,
      "loss": 0.0506,
      "step": 10571
    },
    {
      "epoch": 0.5459333849728892,
      "grad_norm": 1.7130969762802124,
      "learning_rate": 4.35248042607178e-06,
      "loss": 0.0228,
      "step": 10572
    },
    {
      "epoch": 0.545985024528789,
      "grad_norm": 12.520923614501953,
      "learning_rate": 4.351667974016164e-06,
      "loss": 0.2258,
      "step": 10573
    },
    {
      "epoch": 0.5460366640846889,
      "grad_norm": 7.882449626922607,
      "learning_rate": 4.3508555393709e-06,
      "loss": 0.0764,
      "step": 10574
    },
    {
      "epoch": 0.5460883036405887,
      "grad_norm": 9.96062183380127,
      "learning_rate": 4.350043122157813e-06,
      "loss": 0.0531,
      "step": 10575
    },
    {
      "epoch": 0.5461399431964885,
      "grad_norm": 4.8100385665893555,
      "learning_rate": 4.349230722398716e-06,
      "loss": 0.0399,
      "step": 10576
    },
    {
      "epoch": 0.5461915827523883,
      "grad_norm": 19.22524070739746,
      "learning_rate": 4.348418340115429e-06,
      "loss": 0.1479,
      "step": 10577
    },
    {
      "epoch": 0.5462432223082881,
      "grad_norm": 7.422255039215088,
      "learning_rate": 4.347605975329763e-06,
      "loss": 0.0664,
      "step": 10578
    },
    {
      "epoch": 0.546294861864188,
      "grad_norm": 16.32981300354004,
      "learning_rate": 4.346793628063536e-06,
      "loss": 0.1107,
      "step": 10579
    },
    {
      "epoch": 0.5463465014200878,
      "grad_norm": 10.751508712768555,
      "learning_rate": 4.345981298338564e-06,
      "loss": 0.1887,
      "step": 10580
    },
    {
      "epoch": 0.5463981409759876,
      "grad_norm": 5.2004499435424805,
      "learning_rate": 4.345168986176657e-06,
      "loss": 0.0402,
      "step": 10581
    },
    {
      "epoch": 0.5464497805318874,
      "grad_norm": 4.86605167388916,
      "learning_rate": 4.344356691599632e-06,
      "loss": 0.0532,
      "step": 10582
    },
    {
      "epoch": 0.5465014200877872,
      "grad_norm": 9.10935115814209,
      "learning_rate": 4.3435444146293035e-06,
      "loss": 0.0872,
      "step": 10583
    },
    {
      "epoch": 0.5465530596436871,
      "grad_norm": 16.998498916625977,
      "learning_rate": 4.342732155287484e-06,
      "loss": 0.0757,
      "step": 10584
    },
    {
      "epoch": 0.5466046991995869,
      "grad_norm": 4.823848247528076,
      "learning_rate": 4.341919913595983e-06,
      "loss": 0.0599,
      "step": 10585
    },
    {
      "epoch": 0.5466563387554867,
      "grad_norm": 1.9454567432403564,
      "learning_rate": 4.341107689576614e-06,
      "loss": 0.0349,
      "step": 10586
    },
    {
      "epoch": 0.5467079783113865,
      "grad_norm": 1.493159532546997,
      "learning_rate": 4.34029548325119e-06,
      "loss": 0.0283,
      "step": 10587
    },
    {
      "epoch": 0.5467596178672863,
      "grad_norm": 4.599574089050293,
      "learning_rate": 4.339483294641522e-06,
      "loss": 0.0676,
      "step": 10588
    },
    {
      "epoch": 0.5468112574231861,
      "grad_norm": 6.4677324295043945,
      "learning_rate": 4.338671123769419e-06,
      "loss": 0.0415,
      "step": 10589
    },
    {
      "epoch": 0.546862896979086,
      "grad_norm": 2.2052528858184814,
      "learning_rate": 4.3378589706566916e-06,
      "loss": 0.0384,
      "step": 10590
    },
    {
      "epoch": 0.5469145365349858,
      "grad_norm": 5.110524654388428,
      "learning_rate": 4.3370468353251504e-06,
      "loss": 0.0874,
      "step": 10591
    },
    {
      "epoch": 0.5469661760908856,
      "grad_norm": 5.391227722167969,
      "learning_rate": 4.336234717796603e-06,
      "loss": 0.0815,
      "step": 10592
    },
    {
      "epoch": 0.5470178156467854,
      "grad_norm": 10.917341232299805,
      "learning_rate": 4.335422618092857e-06,
      "loss": 0.0517,
      "step": 10593
    },
    {
      "epoch": 0.5470694552026852,
      "grad_norm": 1.0631039142608643,
      "learning_rate": 4.334610536235724e-06,
      "loss": 0.0107,
      "step": 10594
    },
    {
      "epoch": 0.5471210947585851,
      "grad_norm": 7.641928672790527,
      "learning_rate": 4.333798472247012e-06,
      "loss": 0.0491,
      "step": 10595
    },
    {
      "epoch": 0.5471727343144849,
      "grad_norm": 2.0175540447235107,
      "learning_rate": 4.332986426148524e-06,
      "loss": 0.0285,
      "step": 10596
    },
    {
      "epoch": 0.5472243738703847,
      "grad_norm": 2.7809317111968994,
      "learning_rate": 4.332174397962069e-06,
      "loss": 0.0271,
      "step": 10597
    },
    {
      "epoch": 0.5472760134262845,
      "grad_norm": 4.281126022338867,
      "learning_rate": 4.3313623877094525e-06,
      "loss": 0.0581,
      "step": 10598
    },
    {
      "epoch": 0.5473276529821843,
      "grad_norm": 6.471827507019043,
      "learning_rate": 4.330550395412485e-06,
      "loss": 0.095,
      "step": 10599
    },
    {
      "epoch": 0.5473792925380842,
      "grad_norm": 3.717750310897827,
      "learning_rate": 4.329738421092964e-06,
      "loss": 0.0816,
      "step": 10600
    },
    {
      "epoch": 0.547430932093984,
      "grad_norm": 3.67044997215271,
      "learning_rate": 4.3289264647726995e-06,
      "loss": 0.0681,
      "step": 10601
    },
    {
      "epoch": 0.5474825716498838,
      "grad_norm": 18.612682342529297,
      "learning_rate": 4.328114526473496e-06,
      "loss": 0.0892,
      "step": 10602
    },
    {
      "epoch": 0.5475342112057836,
      "grad_norm": 6.2975287437438965,
      "learning_rate": 4.327302606217154e-06,
      "loss": 0.1075,
      "step": 10603
    },
    {
      "epoch": 0.5475858507616834,
      "grad_norm": 2.760193347930908,
      "learning_rate": 4.32649070402548e-06,
      "loss": 0.0218,
      "step": 10604
    },
    {
      "epoch": 0.5476374903175832,
      "grad_norm": 4.301696300506592,
      "learning_rate": 4.325678819920273e-06,
      "loss": 0.0369,
      "step": 10605
    },
    {
      "epoch": 0.5476891298734831,
      "grad_norm": 29.760560989379883,
      "learning_rate": 4.324866953923343e-06,
      "loss": 0.1312,
      "step": 10606
    },
    {
      "epoch": 0.5477407694293829,
      "grad_norm": 8.084709167480469,
      "learning_rate": 4.324055106056482e-06,
      "loss": 0.0625,
      "step": 10607
    },
    {
      "epoch": 0.5477924089852827,
      "grad_norm": 3.785426378250122,
      "learning_rate": 4.3232432763414985e-06,
      "loss": 0.0643,
      "step": 10608
    },
    {
      "epoch": 0.5478440485411825,
      "grad_norm": 2.333430528640747,
      "learning_rate": 4.322431464800191e-06,
      "loss": 0.0243,
      "step": 10609
    },
    {
      "epoch": 0.5478956880970823,
      "grad_norm": 5.6993489265441895,
      "learning_rate": 4.321619671454362e-06,
      "loss": 0.0369,
      "step": 10610
    },
    {
      "epoch": 0.5479473276529822,
      "grad_norm": 1.4373286962509155,
      "learning_rate": 4.320807896325808e-06,
      "loss": 0.0289,
      "step": 10611
    },
    {
      "epoch": 0.547998967208882,
      "grad_norm": 8.458471298217773,
      "learning_rate": 4.319996139436329e-06,
      "loss": 0.0348,
      "step": 10612
    },
    {
      "epoch": 0.5480506067647818,
      "grad_norm": 1.5825912952423096,
      "learning_rate": 4.319184400807729e-06,
      "loss": 0.0163,
      "step": 10613
    },
    {
      "epoch": 0.5481022463206816,
      "grad_norm": 5.517801761627197,
      "learning_rate": 4.3183726804618e-06,
      "loss": 0.0563,
      "step": 10614
    },
    {
      "epoch": 0.5481538858765814,
      "grad_norm": 6.79920768737793,
      "learning_rate": 4.317560978420342e-06,
      "loss": 0.0555,
      "step": 10615
    },
    {
      "epoch": 0.5482055254324812,
      "grad_norm": 12.159302711486816,
      "learning_rate": 4.316749294705155e-06,
      "loss": 0.0681,
      "step": 10616
    },
    {
      "epoch": 0.5482571649883811,
      "grad_norm": 3.373290777206421,
      "learning_rate": 4.3159376293380336e-06,
      "loss": 0.0554,
      "step": 10617
    },
    {
      "epoch": 0.5483088045442809,
      "grad_norm": 11.282025337219238,
      "learning_rate": 4.315125982340776e-06,
      "loss": 0.0942,
      "step": 10618
    },
    {
      "epoch": 0.5483604441001807,
      "grad_norm": 9.340785026550293,
      "learning_rate": 4.314314353735174e-06,
      "loss": 0.0713,
      "step": 10619
    },
    {
      "epoch": 0.5484120836560805,
      "grad_norm": 9.097382545471191,
      "learning_rate": 4.313502743543029e-06,
      "loss": 0.0684,
      "step": 10620
    },
    {
      "epoch": 0.5484637232119803,
      "grad_norm": 6.452884197235107,
      "learning_rate": 4.312691151786133e-06,
      "loss": 0.0554,
      "step": 10621
    },
    {
      "epoch": 0.5485153627678802,
      "grad_norm": 4.43087100982666,
      "learning_rate": 4.31187957848628e-06,
      "loss": 0.0554,
      "step": 10622
    },
    {
      "epoch": 0.54856700232378,
      "grad_norm": 3.3100833892822266,
      "learning_rate": 4.3110680236652654e-06,
      "loss": 0.0435,
      "step": 10623
    },
    {
      "epoch": 0.5486186418796798,
      "grad_norm": 2.873181104660034,
      "learning_rate": 4.310256487344884e-06,
      "loss": 0.0508,
      "step": 10624
    },
    {
      "epoch": 0.5486702814355796,
      "grad_norm": 3.767988443374634,
      "learning_rate": 4.3094449695469255e-06,
      "loss": 0.0801,
      "step": 10625
    },
    {
      "epoch": 0.5487219209914794,
      "grad_norm": 4.776011943817139,
      "learning_rate": 4.308633470293183e-06,
      "loss": 0.0497,
      "step": 10626
    },
    {
      "epoch": 0.5487735605473792,
      "grad_norm": 2.9323441982269287,
      "learning_rate": 4.30782198960545e-06,
      "loss": 0.0364,
      "step": 10627
    },
    {
      "epoch": 0.5488252001032791,
      "grad_norm": 2.9455809593200684,
      "learning_rate": 4.3070105275055205e-06,
      "loss": 0.057,
      "step": 10628
    },
    {
      "epoch": 0.5488768396591789,
      "grad_norm": 5.522884845733643,
      "learning_rate": 4.3061990840151805e-06,
      "loss": 0.0415,
      "step": 10629
    },
    {
      "epoch": 0.5489284792150787,
      "grad_norm": 2.6500091552734375,
      "learning_rate": 4.305387659156223e-06,
      "loss": 0.0191,
      "step": 10630
    },
    {
      "epoch": 0.5489801187709785,
      "grad_norm": 4.57493257522583,
      "learning_rate": 4.304576252950437e-06,
      "loss": 0.0255,
      "step": 10631
    },
    {
      "epoch": 0.5490317583268783,
      "grad_norm": 11.097479820251465,
      "learning_rate": 4.303764865419616e-06,
      "loss": 0.1297,
      "step": 10632
    },
    {
      "epoch": 0.5490833978827782,
      "grad_norm": 2.6641204357147217,
      "learning_rate": 4.302953496585544e-06,
      "loss": 0.0265,
      "step": 10633
    },
    {
      "epoch": 0.549135037438678,
      "grad_norm": 5.520297527313232,
      "learning_rate": 4.302142146470011e-06,
      "loss": 0.0601,
      "step": 10634
    },
    {
      "epoch": 0.5491866769945778,
      "grad_norm": 8.64841365814209,
      "learning_rate": 4.301330815094808e-06,
      "loss": 0.1322,
      "step": 10635
    },
    {
      "epoch": 0.5492383165504776,
      "grad_norm": 2.7211976051330566,
      "learning_rate": 4.300519502481719e-06,
      "loss": 0.0581,
      "step": 10636
    },
    {
      "epoch": 0.5492899561063774,
      "grad_norm": 2.7965240478515625,
      "learning_rate": 4.299708208652532e-06,
      "loss": 0.0222,
      "step": 10637
    },
    {
      "epoch": 0.5493415956622772,
      "grad_norm": 7.283077239990234,
      "learning_rate": 4.298896933629032e-06,
      "loss": 0.0775,
      "step": 10638
    },
    {
      "epoch": 0.5493932352181772,
      "grad_norm": 2.7036352157592773,
      "learning_rate": 4.298085677433011e-06,
      "loss": 0.0524,
      "step": 10639
    },
    {
      "epoch": 0.549444874774077,
      "grad_norm": 3.2449352741241455,
      "learning_rate": 4.297274440086247e-06,
      "loss": 0.0258,
      "step": 10640
    },
    {
      "epoch": 0.5494965143299768,
      "grad_norm": 14.359549522399902,
      "learning_rate": 4.296463221610529e-06,
      "loss": 0.104,
      "step": 10641
    },
    {
      "epoch": 0.5495481538858766,
      "grad_norm": 3.4262640476226807,
      "learning_rate": 4.295652022027642e-06,
      "loss": 0.0436,
      "step": 10642
    },
    {
      "epoch": 0.5495997934417765,
      "grad_norm": 3.584709405899048,
      "learning_rate": 4.294840841359369e-06,
      "loss": 0.0889,
      "step": 10643
    },
    {
      "epoch": 0.5496514329976763,
      "grad_norm": 30.71114730834961,
      "learning_rate": 4.2940296796274935e-06,
      "loss": 0.3793,
      "step": 10644
    },
    {
      "epoch": 0.5497030725535761,
      "grad_norm": 2.4560911655426025,
      "learning_rate": 4.2932185368537975e-06,
      "loss": 0.0465,
      "step": 10645
    },
    {
      "epoch": 0.5497547121094759,
      "grad_norm": 3.4296765327453613,
      "learning_rate": 4.292407413060066e-06,
      "loss": 0.064,
      "step": 10646
    },
    {
      "epoch": 0.5498063516653757,
      "grad_norm": 8.04564094543457,
      "learning_rate": 4.291596308268078e-06,
      "loss": 0.0717,
      "step": 10647
    },
    {
      "epoch": 0.5498579912212755,
      "grad_norm": 4.317891597747803,
      "learning_rate": 4.290785222499616e-06,
      "loss": 0.0473,
      "step": 10648
    },
    {
      "epoch": 0.5499096307771754,
      "grad_norm": 5.292735576629639,
      "learning_rate": 4.289974155776462e-06,
      "loss": 0.0546,
      "step": 10649
    },
    {
      "epoch": 0.5499612703330752,
      "grad_norm": 1.8805824518203735,
      "learning_rate": 4.289163108120398e-06,
      "loss": 0.0346,
      "step": 10650
    },
    {
      "epoch": 0.550012909888975,
      "grad_norm": 17.65328025817871,
      "learning_rate": 4.288352079553199e-06,
      "loss": 0.101,
      "step": 10651
    },
    {
      "epoch": 0.5500645494448748,
      "grad_norm": 5.949825763702393,
      "learning_rate": 4.287541070096647e-06,
      "loss": 0.1004,
      "step": 10652
    },
    {
      "epoch": 0.5501161890007746,
      "grad_norm": 6.197657585144043,
      "learning_rate": 4.286730079772522e-06,
      "loss": 0.0446,
      "step": 10653
    },
    {
      "epoch": 0.5501678285566745,
      "grad_norm": 1.3683234453201294,
      "learning_rate": 4.285919108602602e-06,
      "loss": 0.0281,
      "step": 10654
    },
    {
      "epoch": 0.5502194681125743,
      "grad_norm": 6.025450229644775,
      "learning_rate": 4.2851081566086645e-06,
      "loss": 0.0667,
      "step": 10655
    },
    {
      "epoch": 0.5502711076684741,
      "grad_norm": 7.361736297607422,
      "learning_rate": 4.2842972238124854e-06,
      "loss": 0.0898,
      "step": 10656
    },
    {
      "epoch": 0.5503227472243739,
      "grad_norm": 1.3118219375610352,
      "learning_rate": 4.283486310235845e-06,
      "loss": 0.015,
      "step": 10657
    },
    {
      "epoch": 0.5503743867802737,
      "grad_norm": 3.1450276374816895,
      "learning_rate": 4.282675415900515e-06,
      "loss": 0.0725,
      "step": 10658
    },
    {
      "epoch": 0.5504260263361735,
      "grad_norm": 2.091454029083252,
      "learning_rate": 4.281864540828275e-06,
      "loss": 0.0255,
      "step": 10659
    },
    {
      "epoch": 0.5504776658920734,
      "grad_norm": 1.8034228086471558,
      "learning_rate": 4.281053685040898e-06,
      "loss": 0.0163,
      "step": 10660
    },
    {
      "epoch": 0.5505293054479732,
      "grad_norm": 5.402401447296143,
      "learning_rate": 4.280242848560162e-06,
      "loss": 0.116,
      "step": 10661
    },
    {
      "epoch": 0.550580945003873,
      "grad_norm": 3.2233147621154785,
      "learning_rate": 4.279432031407838e-06,
      "loss": 0.0517,
      "step": 10662
    },
    {
      "epoch": 0.5506325845597728,
      "grad_norm": 7.452218532562256,
      "learning_rate": 4.2786212336057e-06,
      "loss": 0.0612,
      "step": 10663
    },
    {
      "epoch": 0.5506842241156726,
      "grad_norm": 0.9734982848167419,
      "learning_rate": 4.277810455175522e-06,
      "loss": 0.0207,
      "step": 10664
    },
    {
      "epoch": 0.5507358636715725,
      "grad_norm": 9.580859184265137,
      "learning_rate": 4.27699969613908e-06,
      "loss": 0.0402,
      "step": 10665
    },
    {
      "epoch": 0.5507875032274723,
      "grad_norm": 12.26071834564209,
      "learning_rate": 4.276188956518139e-06,
      "loss": 0.1196,
      "step": 10666
    },
    {
      "epoch": 0.5508391427833721,
      "grad_norm": 9.645979881286621,
      "learning_rate": 4.2753782363344755e-06,
      "loss": 0.1161,
      "step": 10667
    },
    {
      "epoch": 0.5508907823392719,
      "grad_norm": 4.9891486167907715,
      "learning_rate": 4.274567535609861e-06,
      "loss": 0.0271,
      "step": 10668
    },
    {
      "epoch": 0.5509424218951717,
      "grad_norm": 4.673706531524658,
      "learning_rate": 4.273756854366065e-06,
      "loss": 0.0296,
      "step": 10669
    },
    {
      "epoch": 0.5509940614510715,
      "grad_norm": 2.711820602416992,
      "learning_rate": 4.272946192624856e-06,
      "loss": 0.0332,
      "step": 10670
    },
    {
      "epoch": 0.5510457010069714,
      "grad_norm": 11.609036445617676,
      "learning_rate": 4.2721355504080045e-06,
      "loss": 0.0901,
      "step": 10671
    },
    {
      "epoch": 0.5510973405628712,
      "grad_norm": 3.7139594554901123,
      "learning_rate": 4.271324927737283e-06,
      "loss": 0.0384,
      "step": 10672
    },
    {
      "epoch": 0.551148980118771,
      "grad_norm": 5.950366973876953,
      "learning_rate": 4.270514324634455e-06,
      "loss": 0.0539,
      "step": 10673
    },
    {
      "epoch": 0.5512006196746708,
      "grad_norm": 21.78407859802246,
      "learning_rate": 4.269703741121291e-06,
      "loss": 0.0802,
      "step": 10674
    },
    {
      "epoch": 0.5512522592305706,
      "grad_norm": 9.392613410949707,
      "learning_rate": 4.268893177219558e-06,
      "loss": 0.0721,
      "step": 10675
    },
    {
      "epoch": 0.5513038987864705,
      "grad_norm": 4.63372278213501,
      "learning_rate": 4.2680826329510236e-06,
      "loss": 0.088,
      "step": 10676
    },
    {
      "epoch": 0.5513555383423703,
      "grad_norm": 8.476643562316895,
      "learning_rate": 4.267272108337453e-06,
      "loss": 0.0687,
      "step": 10677
    },
    {
      "epoch": 0.5514071778982701,
      "grad_norm": 4.0037007331848145,
      "learning_rate": 4.266461603400611e-06,
      "loss": 0.0533,
      "step": 10678
    },
    {
      "epoch": 0.5514588174541699,
      "grad_norm": 27.980512619018555,
      "learning_rate": 4.265651118162268e-06,
      "loss": 0.2241,
      "step": 10679
    },
    {
      "epoch": 0.5515104570100697,
      "grad_norm": 3.0781195163726807,
      "learning_rate": 4.264840652644183e-06,
      "loss": 0.0702,
      "step": 10680
    },
    {
      "epoch": 0.5515620965659696,
      "grad_norm": 2.0752077102661133,
      "learning_rate": 4.264030206868124e-06,
      "loss": 0.033,
      "step": 10681
    },
    {
      "epoch": 0.5516137361218694,
      "grad_norm": 6.28902006149292,
      "learning_rate": 4.263219780855853e-06,
      "loss": 0.0621,
      "step": 10682
    },
    {
      "epoch": 0.5516653756777692,
      "grad_norm": 4.386184215545654,
      "learning_rate": 4.262409374629136e-06,
      "loss": 0.0817,
      "step": 10683
    },
    {
      "epoch": 0.551717015233669,
      "grad_norm": 5.040632724761963,
      "learning_rate": 4.261598988209731e-06,
      "loss": 0.0597,
      "step": 10684
    },
    {
      "epoch": 0.5517686547895688,
      "grad_norm": 6.19840669631958,
      "learning_rate": 4.260788621619403e-06,
      "loss": 0.0526,
      "step": 10685
    },
    {
      "epoch": 0.5518202943454686,
      "grad_norm": 4.846629619598389,
      "learning_rate": 4.2599782748799135e-06,
      "loss": 0.0439,
      "step": 10686
    },
    {
      "epoch": 0.5518719339013685,
      "grad_norm": 2.0306687355041504,
      "learning_rate": 4.259167948013025e-06,
      "loss": 0.0283,
      "step": 10687
    },
    {
      "epoch": 0.5519235734572683,
      "grad_norm": 5.027113914489746,
      "learning_rate": 4.258357641040495e-06,
      "loss": 0.0458,
      "step": 10688
    },
    {
      "epoch": 0.5519752130131681,
      "grad_norm": 6.238232135772705,
      "learning_rate": 4.257547353984086e-06,
      "loss": 0.0636,
      "step": 10689
    },
    {
      "epoch": 0.5520268525690679,
      "grad_norm": 13.062620162963867,
      "learning_rate": 4.256737086865557e-06,
      "loss": 0.2134,
      "step": 10690
    },
    {
      "epoch": 0.5520784921249677,
      "grad_norm": 1.7997713088989258,
      "learning_rate": 4.255926839706665e-06,
      "loss": 0.0297,
      "step": 10691
    },
    {
      "epoch": 0.5521301316808676,
      "grad_norm": 5.450239181518555,
      "learning_rate": 4.25511661252917e-06,
      "loss": 0.0857,
      "step": 10692
    },
    {
      "epoch": 0.5521817712367674,
      "grad_norm": 12.323441505432129,
      "learning_rate": 4.25430640535483e-06,
      "loss": 0.1066,
      "step": 10693
    },
    {
      "epoch": 0.5522334107926672,
      "grad_norm": 2.1300790309906006,
      "learning_rate": 4.253496218205404e-06,
      "loss": 0.0258,
      "step": 10694
    },
    {
      "epoch": 0.552285050348567,
      "grad_norm": 7.3883867263793945,
      "learning_rate": 4.252686051102646e-06,
      "loss": 0.113,
      "step": 10695
    },
    {
      "epoch": 0.5523366899044668,
      "grad_norm": 6.986758708953857,
      "learning_rate": 4.251875904068314e-06,
      "loss": 0.0569,
      "step": 10696
    },
    {
      "epoch": 0.5523883294603666,
      "grad_norm": 3.0523486137390137,
      "learning_rate": 4.251065777124161e-06,
      "loss": 0.0511,
      "step": 10697
    },
    {
      "epoch": 0.5524399690162665,
      "grad_norm": 3.105989694595337,
      "learning_rate": 4.250255670291949e-06,
      "loss": 0.0518,
      "step": 10698
    },
    {
      "epoch": 0.5524916085721663,
      "grad_norm": 4.93444299697876,
      "learning_rate": 4.249445583593423e-06,
      "loss": 0.0701,
      "step": 10699
    },
    {
      "epoch": 0.5525432481280661,
      "grad_norm": 3.2519259452819824,
      "learning_rate": 4.2486355170503445e-06,
      "loss": 0.0569,
      "step": 10700
    },
    {
      "epoch": 0.5525948876839659,
      "grad_norm": 6.777836322784424,
      "learning_rate": 4.247825470684465e-06,
      "loss": 0.1186,
      "step": 10701
    },
    {
      "epoch": 0.5526465272398657,
      "grad_norm": 2.382526159286499,
      "learning_rate": 4.247015444517537e-06,
      "loss": 0.0312,
      "step": 10702
    },
    {
      "epoch": 0.5526981667957656,
      "grad_norm": 2.6883487701416016,
      "learning_rate": 4.246205438571313e-06,
      "loss": 0.0248,
      "step": 10703
    },
    {
      "epoch": 0.5527498063516654,
      "grad_norm": 11.642574310302734,
      "learning_rate": 4.245395452867545e-06,
      "loss": 0.0552,
      "step": 10704
    },
    {
      "epoch": 0.5528014459075652,
      "grad_norm": 3.490976333618164,
      "learning_rate": 4.244585487427987e-06,
      "loss": 0.0336,
      "step": 10705
    },
    {
      "epoch": 0.552853085463465,
      "grad_norm": 5.609585285186768,
      "learning_rate": 4.243775542274384e-06,
      "loss": 0.0542,
      "step": 10706
    },
    {
      "epoch": 0.5529047250193648,
      "grad_norm": 3.5338258743286133,
      "learning_rate": 4.242965617428491e-06,
      "loss": 0.0686,
      "step": 10707
    },
    {
      "epoch": 0.5529563645752646,
      "grad_norm": 3.2990148067474365,
      "learning_rate": 4.242155712912057e-06,
      "loss": 0.0546,
      "step": 10708
    },
    {
      "epoch": 0.5530080041311645,
      "grad_norm": 3.037637710571289,
      "learning_rate": 4.2413458287468325e-06,
      "loss": 0.0387,
      "step": 10709
    },
    {
      "epoch": 0.5530596436870643,
      "grad_norm": 22.642398834228516,
      "learning_rate": 4.240535964954563e-06,
      "loss": 0.1032,
      "step": 10710
    },
    {
      "epoch": 0.5531112832429641,
      "grad_norm": 3.982692003250122,
      "learning_rate": 4.2397261215569975e-06,
      "loss": 0.0251,
      "step": 10711
    },
    {
      "epoch": 0.5531629227988639,
      "grad_norm": 2.2405319213867188,
      "learning_rate": 4.238916298575886e-06,
      "loss": 0.0233,
      "step": 10712
    },
    {
      "epoch": 0.5532145623547637,
      "grad_norm": 1.786054015159607,
      "learning_rate": 4.238106496032973e-06,
      "loss": 0.0341,
      "step": 10713
    },
    {
      "epoch": 0.5532662019106636,
      "grad_norm": 3.528350353240967,
      "learning_rate": 4.237296713950006e-06,
      "loss": 0.0962,
      "step": 10714
    },
    {
      "epoch": 0.5533178414665634,
      "grad_norm": 2.0704874992370605,
      "learning_rate": 4.236486952348731e-06,
      "loss": 0.0467,
      "step": 10715
    },
    {
      "epoch": 0.5533694810224632,
      "grad_norm": 4.478702068328857,
      "learning_rate": 4.235677211250894e-06,
      "loss": 0.0809,
      "step": 10716
    },
    {
      "epoch": 0.553421120578363,
      "grad_norm": 4.5547895431518555,
      "learning_rate": 4.234867490678238e-06,
      "loss": 0.0458,
      "step": 10717
    },
    {
      "epoch": 0.5534727601342628,
      "grad_norm": 7.93892240524292,
      "learning_rate": 4.234057790652508e-06,
      "loss": 0.0921,
      "step": 10718
    },
    {
      "epoch": 0.5535243996901626,
      "grad_norm": 6.657670021057129,
      "learning_rate": 4.233248111195448e-06,
      "loss": 0.0598,
      "step": 10719
    },
    {
      "epoch": 0.5535760392460625,
      "grad_norm": 2.5811851024627686,
      "learning_rate": 4.2324384523288034e-06,
      "loss": 0.0628,
      "step": 10720
    },
    {
      "epoch": 0.5536276788019623,
      "grad_norm": 3.5996780395507812,
      "learning_rate": 4.2316288140743136e-06,
      "loss": 0.0467,
      "step": 10721
    },
    {
      "epoch": 0.5536793183578621,
      "grad_norm": 6.2924885749816895,
      "learning_rate": 4.230819196453722e-06,
      "loss": 0.0613,
      "step": 10722
    },
    {
      "epoch": 0.5537309579137619,
      "grad_norm": 5.653545379638672,
      "learning_rate": 4.23000959948877e-06,
      "loss": 0.0664,
      "step": 10723
    },
    {
      "epoch": 0.5537825974696617,
      "grad_norm": 2.4720940589904785,
      "learning_rate": 4.229200023201198e-06,
      "loss": 0.0287,
      "step": 10724
    },
    {
      "epoch": 0.5538342370255616,
      "grad_norm": 7.209152698516846,
      "learning_rate": 4.2283904676127465e-06,
      "loss": 0.0806,
      "step": 10725
    },
    {
      "epoch": 0.5538858765814614,
      "grad_norm": 6.231044292449951,
      "learning_rate": 4.227580932745156e-06,
      "loss": 0.0411,
      "step": 10726
    },
    {
      "epoch": 0.5539375161373612,
      "grad_norm": 2.6738548278808594,
      "learning_rate": 4.226771418620167e-06,
      "loss": 0.0333,
      "step": 10727
    },
    {
      "epoch": 0.553989155693261,
      "grad_norm": 2.826277494430542,
      "learning_rate": 4.225961925259516e-06,
      "loss": 0.0497,
      "step": 10728
    },
    {
      "epoch": 0.5540407952491608,
      "grad_norm": 12.197361946105957,
      "learning_rate": 4.225152452684943e-06,
      "loss": 0.0709,
      "step": 10729
    },
    {
      "epoch": 0.5540924348050607,
      "grad_norm": 7.445424556732178,
      "learning_rate": 4.224343000918184e-06,
      "loss": 0.0863,
      "step": 10730
    },
    {
      "epoch": 0.5541440743609605,
      "grad_norm": 6.065311431884766,
      "learning_rate": 4.223533569980976e-06,
      "loss": 0.0916,
      "step": 10731
    },
    {
      "epoch": 0.5541957139168603,
      "grad_norm": 5.494997978210449,
      "learning_rate": 4.2227241598950575e-06,
      "loss": 0.0579,
      "step": 10732
    },
    {
      "epoch": 0.5542473534727601,
      "grad_norm": 3.1777751445770264,
      "learning_rate": 4.221914770682162e-06,
      "loss": 0.0302,
      "step": 10733
    },
    {
      "epoch": 0.5542989930286599,
      "grad_norm": 14.980722427368164,
      "learning_rate": 4.2211054023640285e-06,
      "loss": 0.0999,
      "step": 10734
    },
    {
      "epoch": 0.5543506325845597,
      "grad_norm": 9.148296356201172,
      "learning_rate": 4.220296054962388e-06,
      "loss": 0.1261,
      "step": 10735
    },
    {
      "epoch": 0.5544022721404596,
      "grad_norm": 3.489102363586426,
      "learning_rate": 4.219486728498977e-06,
      "loss": 0.041,
      "step": 10736
    },
    {
      "epoch": 0.5544539116963594,
      "grad_norm": 5.902113914489746,
      "learning_rate": 4.218677422995527e-06,
      "loss": 0.0627,
      "step": 10737
    },
    {
      "epoch": 0.5545055512522592,
      "grad_norm": 3.0571110248565674,
      "learning_rate": 4.2178681384737745e-06,
      "loss": 0.0524,
      "step": 10738
    },
    {
      "epoch": 0.554557190808159,
      "grad_norm": 1.928216814994812,
      "learning_rate": 4.21705887495545e-06,
      "loss": 0.0409,
      "step": 10739
    },
    {
      "epoch": 0.5546088303640588,
      "grad_norm": 1.7214343547821045,
      "learning_rate": 4.216249632462285e-06,
      "loss": 0.0301,
      "step": 10740
    },
    {
      "epoch": 0.5546604699199587,
      "grad_norm": 4.39156436920166,
      "learning_rate": 4.215440411016013e-06,
      "loss": 0.0505,
      "step": 10741
    },
    {
      "epoch": 0.5547121094758585,
      "grad_norm": 2.2135860919952393,
      "learning_rate": 4.214631210638363e-06,
      "loss": 0.0382,
      "step": 10742
    },
    {
      "epoch": 0.5547637490317583,
      "grad_norm": 2.3036980628967285,
      "learning_rate": 4.213822031351065e-06,
      "loss": 0.0405,
      "step": 10743
    },
    {
      "epoch": 0.5548153885876581,
      "grad_norm": 9.355687141418457,
      "learning_rate": 4.21301287317585e-06,
      "loss": 0.109,
      "step": 10744
    },
    {
      "epoch": 0.5548670281435579,
      "grad_norm": 16.01271629333496,
      "learning_rate": 4.212203736134448e-06,
      "loss": 0.1328,
      "step": 10745
    },
    {
      "epoch": 0.5549186676994577,
      "grad_norm": 2.643326997756958,
      "learning_rate": 4.211394620248586e-06,
      "loss": 0.0423,
      "step": 10746
    },
    {
      "epoch": 0.5549703072553576,
      "grad_norm": 5.150532245635986,
      "learning_rate": 4.210585525539993e-06,
      "loss": 0.0394,
      "step": 10747
    },
    {
      "epoch": 0.5550219468112574,
      "grad_norm": 4.818742752075195,
      "learning_rate": 4.209776452030395e-06,
      "loss": 0.0764,
      "step": 10748
    },
    {
      "epoch": 0.5550735863671572,
      "grad_norm": 10.086525917053223,
      "learning_rate": 4.208967399741522e-06,
      "loss": 0.089,
      "step": 10749
    },
    {
      "epoch": 0.555125225923057,
      "grad_norm": 2.8933377265930176,
      "learning_rate": 4.208158368695097e-06,
      "loss": 0.0496,
      "step": 10750
    },
    {
      "epoch": 0.5551768654789568,
      "grad_norm": 4.806422710418701,
      "learning_rate": 4.207349358912846e-06,
      "loss": 0.0531,
      "step": 10751
    },
    {
      "epoch": 0.5552285050348567,
      "grad_norm": 2.36110258102417,
      "learning_rate": 4.206540370416498e-06,
      "loss": 0.06,
      "step": 10752
    },
    {
      "epoch": 0.5552801445907565,
      "grad_norm": 9.509382247924805,
      "learning_rate": 4.205731403227774e-06,
      "loss": 0.1074,
      "step": 10753
    },
    {
      "epoch": 0.5553317841466563,
      "grad_norm": 4.555253028869629,
      "learning_rate": 4.204922457368398e-06,
      "loss": 0.0276,
      "step": 10754
    },
    {
      "epoch": 0.5553834237025561,
      "grad_norm": 2.456723690032959,
      "learning_rate": 4.2041135328600955e-06,
      "loss": 0.0431,
      "step": 10755
    },
    {
      "epoch": 0.5554350632584559,
      "grad_norm": 5.428714275360107,
      "learning_rate": 4.203304629724589e-06,
      "loss": 0.0458,
      "step": 10756
    },
    {
      "epoch": 0.5554867028143557,
      "grad_norm": 6.56614875793457,
      "learning_rate": 4.202495747983598e-06,
      "loss": 0.0455,
      "step": 10757
    },
    {
      "epoch": 0.5555383423702556,
      "grad_norm": 3.4608240127563477,
      "learning_rate": 4.201686887658849e-06,
      "loss": 0.0499,
      "step": 10758
    },
    {
      "epoch": 0.5555899819261554,
      "grad_norm": 6.29416036605835,
      "learning_rate": 4.20087804877206e-06,
      "loss": 0.0375,
      "step": 10759
    },
    {
      "epoch": 0.5556416214820552,
      "grad_norm": 2.464704751968384,
      "learning_rate": 4.200069231344954e-06,
      "loss": 0.0549,
      "step": 10760
    },
    {
      "epoch": 0.555693261037955,
      "grad_norm": 1.9054561853408813,
      "learning_rate": 4.199260435399248e-06,
      "loss": 0.0238,
      "step": 10761
    },
    {
      "epoch": 0.5557449005938548,
      "grad_norm": 4.528661251068115,
      "learning_rate": 4.198451660956662e-06,
      "loss": 0.0658,
      "step": 10762
    },
    {
      "epoch": 0.5557965401497547,
      "grad_norm": 6.186825275421143,
      "learning_rate": 4.197642908038918e-06,
      "loss": 0.0531,
      "step": 10763
    },
    {
      "epoch": 0.5558481797056545,
      "grad_norm": 13.245595932006836,
      "learning_rate": 4.19683417666773e-06,
      "loss": 0.1252,
      "step": 10764
    },
    {
      "epoch": 0.5558998192615543,
      "grad_norm": 3.5829527378082275,
      "learning_rate": 4.196025466864819e-06,
      "loss": 0.0284,
      "step": 10765
    },
    {
      "epoch": 0.5559514588174542,
      "grad_norm": 5.535786151885986,
      "learning_rate": 4.195216778651901e-06,
      "loss": 0.068,
      "step": 10766
    },
    {
      "epoch": 0.556003098373354,
      "grad_norm": 7.651963233947754,
      "learning_rate": 4.194408112050693e-06,
      "loss": 0.086,
      "step": 10767
    },
    {
      "epoch": 0.5560547379292539,
      "grad_norm": 6.566921710968018,
      "learning_rate": 4.19359946708291e-06,
      "loss": 0.0964,
      "step": 10768
    },
    {
      "epoch": 0.5561063774851537,
      "grad_norm": 3.457742214202881,
      "learning_rate": 4.192790843770268e-06,
      "loss": 0.0553,
      "step": 10769
    },
    {
      "epoch": 0.5561580170410535,
      "grad_norm": 3.5332179069519043,
      "learning_rate": 4.1919822421344805e-06,
      "loss": 0.0501,
      "step": 10770
    },
    {
      "epoch": 0.5562096565969533,
      "grad_norm": 8.898480415344238,
      "learning_rate": 4.191173662197266e-06,
      "loss": 0.0592,
      "step": 10771
    },
    {
      "epoch": 0.5562612961528531,
      "grad_norm": 3.345653772354126,
      "learning_rate": 4.190365103980334e-06,
      "loss": 0.0545,
      "step": 10772
    },
    {
      "epoch": 0.556312935708753,
      "grad_norm": 3.4718332290649414,
      "learning_rate": 4.1895565675053985e-06,
      "loss": 0.0407,
      "step": 10773
    },
    {
      "epoch": 0.5563645752646528,
      "grad_norm": 2.6617166996002197,
      "learning_rate": 4.1887480527941735e-06,
      "loss": 0.0175,
      "step": 10774
    },
    {
      "epoch": 0.5564162148205526,
      "grad_norm": 6.048922538757324,
      "learning_rate": 4.187939559868368e-06,
      "loss": 0.1055,
      "step": 10775
    },
    {
      "epoch": 0.5564678543764524,
      "grad_norm": 2.5828404426574707,
      "learning_rate": 4.1871310887496955e-06,
      "loss": 0.062,
      "step": 10776
    },
    {
      "epoch": 0.5565194939323522,
      "grad_norm": 15.589250564575195,
      "learning_rate": 4.186322639459866e-06,
      "loss": 0.0342,
      "step": 10777
    },
    {
      "epoch": 0.556571133488252,
      "grad_norm": 4.8355841636657715,
      "learning_rate": 4.185514212020592e-06,
      "loss": 0.0955,
      "step": 10778
    },
    {
      "epoch": 0.5566227730441519,
      "grad_norm": 3.0764591693878174,
      "learning_rate": 4.184705806453579e-06,
      "loss": 0.0681,
      "step": 10779
    },
    {
      "epoch": 0.5566744126000517,
      "grad_norm": 2.1389286518096924,
      "learning_rate": 4.183897422780539e-06,
      "loss": 0.0409,
      "step": 10780
    },
    {
      "epoch": 0.5567260521559515,
      "grad_norm": 1.7673639059066772,
      "learning_rate": 4.18308906102318e-06,
      "loss": 0.0219,
      "step": 10781
    },
    {
      "epoch": 0.5567776917118513,
      "grad_norm": 9.132214546203613,
      "learning_rate": 4.182280721203209e-06,
      "loss": 0.127,
      "step": 10782
    },
    {
      "epoch": 0.5568293312677511,
      "grad_norm": 6.675304889678955,
      "learning_rate": 4.181472403342331e-06,
      "loss": 0.0843,
      "step": 10783
    },
    {
      "epoch": 0.556880970823651,
      "grad_norm": 5.280510902404785,
      "learning_rate": 4.180664107462258e-06,
      "loss": 0.0565,
      "step": 10784
    },
    {
      "epoch": 0.5569326103795508,
      "grad_norm": 4.331737041473389,
      "learning_rate": 4.179855833584693e-06,
      "loss": 0.0546,
      "step": 10785
    },
    {
      "epoch": 0.5569842499354506,
      "grad_norm": 2.284088611602783,
      "learning_rate": 4.17904758173134e-06,
      "loss": 0.0288,
      "step": 10786
    },
    {
      "epoch": 0.5570358894913504,
      "grad_norm": 1.6527639627456665,
      "learning_rate": 4.1782393519239065e-06,
      "loss": 0.0211,
      "step": 10787
    },
    {
      "epoch": 0.5570875290472502,
      "grad_norm": 2.042418956756592,
      "learning_rate": 4.177431144184096e-06,
      "loss": 0.0357,
      "step": 10788
    },
    {
      "epoch": 0.55713916860315,
      "grad_norm": 2.9114418029785156,
      "learning_rate": 4.176622958533612e-06,
      "loss": 0.0356,
      "step": 10789
    },
    {
      "epoch": 0.5571908081590499,
      "grad_norm": 5.700042724609375,
      "learning_rate": 4.175814794994156e-06,
      "loss": 0.0303,
      "step": 10790
    },
    {
      "epoch": 0.5572424477149497,
      "grad_norm": 2.4497427940368652,
      "learning_rate": 4.175006653587432e-06,
      "loss": 0.0242,
      "step": 10791
    },
    {
      "epoch": 0.5572940872708495,
      "grad_norm": 9.998087882995605,
      "learning_rate": 4.174198534335144e-06,
      "loss": 0.0551,
      "step": 10792
    },
    {
      "epoch": 0.5573457268267493,
      "grad_norm": 4.867721080780029,
      "learning_rate": 4.173390437258989e-06,
      "loss": 0.0699,
      "step": 10793
    },
    {
      "epoch": 0.5573973663826491,
      "grad_norm": 4.234042167663574,
      "learning_rate": 4.172582362380671e-06,
      "loss": 0.033,
      "step": 10794
    },
    {
      "epoch": 0.557449005938549,
      "grad_norm": 20.141267776489258,
      "learning_rate": 4.171774309721888e-06,
      "loss": 0.0544,
      "step": 10795
    },
    {
      "epoch": 0.5575006454944488,
      "grad_norm": 3.661242723464966,
      "learning_rate": 4.170966279304343e-06,
      "loss": 0.041,
      "step": 10796
    },
    {
      "epoch": 0.5575522850503486,
      "grad_norm": 2.9776265621185303,
      "learning_rate": 4.170158271149729e-06,
      "loss": 0.0516,
      "step": 10797
    },
    {
      "epoch": 0.5576039246062484,
      "grad_norm": 4.518259048461914,
      "learning_rate": 4.169350285279748e-06,
      "loss": 0.0694,
      "step": 10798
    },
    {
      "epoch": 0.5576555641621482,
      "grad_norm": 19.5651798248291,
      "learning_rate": 4.168542321716099e-06,
      "loss": 0.173,
      "step": 10799
    },
    {
      "epoch": 0.557707203718048,
      "grad_norm": 3.421945333480835,
      "learning_rate": 4.1677343804804774e-06,
      "loss": 0.0557,
      "step": 10800
    },
    {
      "epoch": 0.5577588432739479,
      "grad_norm": 2.3102331161499023,
      "learning_rate": 4.16692646159458e-06,
      "loss": 0.0334,
      "step": 10801
    },
    {
      "epoch": 0.5578104828298477,
      "grad_norm": 6.945862770080566,
      "learning_rate": 4.1661185650801e-06,
      "loss": 0.063,
      "step": 10802
    },
    {
      "epoch": 0.5578621223857475,
      "grad_norm": 8.96306037902832,
      "learning_rate": 4.165310690958737e-06,
      "loss": 0.2274,
      "step": 10803
    },
    {
      "epoch": 0.5579137619416473,
      "grad_norm": 5.797304153442383,
      "learning_rate": 4.164502839252185e-06,
      "loss": 0.0444,
      "step": 10804
    },
    {
      "epoch": 0.5579654014975471,
      "grad_norm": 3.894562005996704,
      "learning_rate": 4.163695009982136e-06,
      "loss": 0.0663,
      "step": 10805
    },
    {
      "epoch": 0.558017041053447,
      "grad_norm": 2.330071210861206,
      "learning_rate": 4.162887203170285e-06,
      "loss": 0.0378,
      "step": 10806
    },
    {
      "epoch": 0.5580686806093468,
      "grad_norm": 5.27842378616333,
      "learning_rate": 4.1620794188383255e-06,
      "loss": 0.1039,
      "step": 10807
    },
    {
      "epoch": 0.5581203201652466,
      "grad_norm": 11.82178020477295,
      "learning_rate": 4.161271657007946e-06,
      "loss": 0.0996,
      "step": 10808
    },
    {
      "epoch": 0.5581719597211464,
      "grad_norm": 18.1940860748291,
      "learning_rate": 4.160463917700842e-06,
      "loss": 0.0884,
      "step": 10809
    },
    {
      "epoch": 0.5582235992770462,
      "grad_norm": 9.384925842285156,
      "learning_rate": 4.159656200938704e-06,
      "loss": 0.1056,
      "step": 10810
    },
    {
      "epoch": 0.558275238832946,
      "grad_norm": 7.886077880859375,
      "learning_rate": 4.158848506743223e-06,
      "loss": 0.0864,
      "step": 10811
    },
    {
      "epoch": 0.5583268783888459,
      "grad_norm": 14.186765670776367,
      "learning_rate": 4.158040835136087e-06,
      "loss": 0.0579,
      "step": 10812
    },
    {
      "epoch": 0.5583785179447457,
      "grad_norm": 7.419297218322754,
      "learning_rate": 4.157233186138986e-06,
      "loss": 0.0698,
      "step": 10813
    },
    {
      "epoch": 0.5584301575006455,
      "grad_norm": 2.9329428672790527,
      "learning_rate": 4.1564255597736085e-06,
      "loss": 0.0338,
      "step": 10814
    },
    {
      "epoch": 0.5584817970565453,
      "grad_norm": 5.096760272979736,
      "learning_rate": 4.1556179560616445e-06,
      "loss": 0.0994,
      "step": 10815
    },
    {
      "epoch": 0.5585334366124451,
      "grad_norm": 4.050345420837402,
      "learning_rate": 4.154810375024778e-06,
      "loss": 0.0789,
      "step": 10816
    },
    {
      "epoch": 0.558585076168345,
      "grad_norm": 5.596052169799805,
      "learning_rate": 4.154002816684698e-06,
      "loss": 0.0625,
      "step": 10817
    },
    {
      "epoch": 0.5586367157242448,
      "grad_norm": 14.987353324890137,
      "learning_rate": 4.153195281063092e-06,
      "loss": 0.1669,
      "step": 10818
    },
    {
      "epoch": 0.5586883552801446,
      "grad_norm": 8.607865333557129,
      "learning_rate": 4.152387768181644e-06,
      "loss": 0.1135,
      "step": 10819
    },
    {
      "epoch": 0.5587399948360444,
      "grad_norm": 6.126605987548828,
      "learning_rate": 4.1515802780620376e-06,
      "loss": 0.0445,
      "step": 10820
    },
    {
      "epoch": 0.5587916343919442,
      "grad_norm": 5.250231742858887,
      "learning_rate": 4.1507728107259585e-06,
      "loss": 0.0823,
      "step": 10821
    },
    {
      "epoch": 0.558843273947844,
      "grad_norm": 7.116105556488037,
      "learning_rate": 4.149965366195094e-06,
      "loss": 0.1385,
      "step": 10822
    },
    {
      "epoch": 0.5588949135037439,
      "grad_norm": 1.6077969074249268,
      "learning_rate": 4.14915794449112e-06,
      "loss": 0.0416,
      "step": 10823
    },
    {
      "epoch": 0.5589465530596437,
      "grad_norm": 5.0367655754089355,
      "learning_rate": 4.148350545635724e-06,
      "loss": 0.0373,
      "step": 10824
    },
    {
      "epoch": 0.5589981926155435,
      "grad_norm": 8.213591575622559,
      "learning_rate": 4.147543169650587e-06,
      "loss": 0.1091,
      "step": 10825
    },
    {
      "epoch": 0.5590498321714433,
      "grad_norm": 2.010993480682373,
      "learning_rate": 4.1467358165573925e-06,
      "loss": 0.0487,
      "step": 10826
    },
    {
      "epoch": 0.5591014717273431,
      "grad_norm": 6.218861103057861,
      "learning_rate": 4.145928486377817e-06,
      "loss": 0.0893,
      "step": 10827
    },
    {
      "epoch": 0.559153111283243,
      "grad_norm": 8.615169525146484,
      "learning_rate": 4.145121179133542e-06,
      "loss": 0.1163,
      "step": 10828
    },
    {
      "epoch": 0.5592047508391428,
      "grad_norm": 11.03066635131836,
      "learning_rate": 4.14431389484625e-06,
      "loss": 0.1349,
      "step": 10829
    },
    {
      "epoch": 0.5592563903950426,
      "grad_norm": 4.97226095199585,
      "learning_rate": 4.143506633537615e-06,
      "loss": 0.0624,
      "step": 10830
    },
    {
      "epoch": 0.5593080299509424,
      "grad_norm": 3.5902271270751953,
      "learning_rate": 4.1426993952293195e-06,
      "loss": 0.0693,
      "step": 10831
    },
    {
      "epoch": 0.5593596695068422,
      "grad_norm": 3.576594352722168,
      "learning_rate": 4.141892179943038e-06,
      "loss": 0.0743,
      "step": 10832
    },
    {
      "epoch": 0.5594113090627421,
      "grad_norm": 4.621090888977051,
      "learning_rate": 4.141084987700451e-06,
      "loss": 0.0683,
      "step": 10833
    },
    {
      "epoch": 0.5594629486186419,
      "grad_norm": 4.647607803344727,
      "learning_rate": 4.1402778185232315e-06,
      "loss": 0.0433,
      "step": 10834
    },
    {
      "epoch": 0.5595145881745417,
      "grad_norm": 5.3501296043396,
      "learning_rate": 4.139470672433056e-06,
      "loss": 0.0718,
      "step": 10835
    },
    {
      "epoch": 0.5595662277304415,
      "grad_norm": 2.1826326847076416,
      "learning_rate": 4.1386635494516e-06,
      "loss": 0.0656,
      "step": 10836
    },
    {
      "epoch": 0.5596178672863413,
      "grad_norm": 4.801871299743652,
      "learning_rate": 4.137856449600541e-06,
      "loss": 0.0937,
      "step": 10837
    },
    {
      "epoch": 0.5596695068422411,
      "grad_norm": 2.109934091567993,
      "learning_rate": 4.137049372901548e-06,
      "loss": 0.064,
      "step": 10838
    },
    {
      "epoch": 0.559721146398141,
      "grad_norm": 4.628595352172852,
      "learning_rate": 4.136242319376297e-06,
      "loss": 0.0288,
      "step": 10839
    },
    {
      "epoch": 0.5597727859540408,
      "grad_norm": 2.0371217727661133,
      "learning_rate": 4.135435289046461e-06,
      "loss": 0.0459,
      "step": 10840
    },
    {
      "epoch": 0.5598244255099406,
      "grad_norm": 6.262845039367676,
      "learning_rate": 4.1346282819337104e-06,
      "loss": 0.1686,
      "step": 10841
    },
    {
      "epoch": 0.5598760650658404,
      "grad_norm": 5.456619739532471,
      "learning_rate": 4.133821298059717e-06,
      "loss": 0.1129,
      "step": 10842
    },
    {
      "epoch": 0.5599277046217402,
      "grad_norm": 1.740039587020874,
      "learning_rate": 4.1330143374461525e-06,
      "loss": 0.0314,
      "step": 10843
    },
    {
      "epoch": 0.5599793441776401,
      "grad_norm": 5.296297550201416,
      "learning_rate": 4.132207400114688e-06,
      "loss": 0.0621,
      "step": 10844
    },
    {
      "epoch": 0.5600309837335399,
      "grad_norm": 11.152410507202148,
      "learning_rate": 4.13140048608699e-06,
      "loss": 0.1165,
      "step": 10845
    },
    {
      "epoch": 0.5600826232894397,
      "grad_norm": 6.518157482147217,
      "learning_rate": 4.13059359538473e-06,
      "loss": 0.0758,
      "step": 10846
    },
    {
      "epoch": 0.5601342628453395,
      "grad_norm": 3.8770556449890137,
      "learning_rate": 4.129786728029574e-06,
      "loss": 0.0417,
      "step": 10847
    },
    {
      "epoch": 0.5601859024012393,
      "grad_norm": 5.253688335418701,
      "learning_rate": 4.1289798840431945e-06,
      "loss": 0.1011,
      "step": 10848
    },
    {
      "epoch": 0.5602375419571392,
      "grad_norm": 3.747758150100708,
      "learning_rate": 4.128173063447252e-06,
      "loss": 0.0814,
      "step": 10849
    },
    {
      "epoch": 0.560289181513039,
      "grad_norm": 2.419296979904175,
      "learning_rate": 4.127366266263417e-06,
      "loss": 0.05,
      "step": 10850
    },
    {
      "epoch": 0.5603408210689388,
      "grad_norm": 2.762047529220581,
      "learning_rate": 4.126559492513356e-06,
      "loss": 0.0213,
      "step": 10851
    },
    {
      "epoch": 0.5603924606248386,
      "grad_norm": 3.0259385108947754,
      "learning_rate": 4.1257527422187304e-06,
      "loss": 0.0505,
      "step": 10852
    },
    {
      "epoch": 0.5604441001807384,
      "grad_norm": 5.205542087554932,
      "learning_rate": 4.124946015401208e-06,
      "loss": 0.0701,
      "step": 10853
    },
    {
      "epoch": 0.5604957397366382,
      "grad_norm": 2.9693994522094727,
      "learning_rate": 4.12413931208245e-06,
      "loss": 0.0688,
      "step": 10854
    },
    {
      "epoch": 0.5605473792925381,
      "grad_norm": 3.8142645359039307,
      "learning_rate": 4.123332632284124e-06,
      "loss": 0.0766,
      "step": 10855
    },
    {
      "epoch": 0.5605990188484379,
      "grad_norm": 1.8404655456542969,
      "learning_rate": 4.122525976027887e-06,
      "loss": 0.0175,
      "step": 10856
    },
    {
      "epoch": 0.5606506584043377,
      "grad_norm": 4.41292667388916,
      "learning_rate": 4.121719343335406e-06,
      "loss": 0.1285,
      "step": 10857
    },
    {
      "epoch": 0.5607022979602375,
      "grad_norm": 23.80656623840332,
      "learning_rate": 4.120912734228339e-06,
      "loss": 0.0415,
      "step": 10858
    },
    {
      "epoch": 0.5607539375161373,
      "grad_norm": 17.177066802978516,
      "learning_rate": 4.12010614872835e-06,
      "loss": 0.0867,
      "step": 10859
    },
    {
      "epoch": 0.5608055770720372,
      "grad_norm": 2.5539803504943848,
      "learning_rate": 4.119299586857096e-06,
      "loss": 0.0506,
      "step": 10860
    },
    {
      "epoch": 0.560857216627937,
      "grad_norm": 8.550057411193848,
      "learning_rate": 4.118493048636236e-06,
      "loss": 0.1259,
      "step": 10861
    },
    {
      "epoch": 0.5609088561838368,
      "grad_norm": 6.9461259841918945,
      "learning_rate": 4.117686534087434e-06,
      "loss": 0.0919,
      "step": 10862
    },
    {
      "epoch": 0.5609604957397366,
      "grad_norm": 5.1379523277282715,
      "learning_rate": 4.116880043232342e-06,
      "loss": 0.0518,
      "step": 10863
    },
    {
      "epoch": 0.5610121352956364,
      "grad_norm": 2.33404278755188,
      "learning_rate": 4.116073576092621e-06,
      "loss": 0.0365,
      "step": 10864
    },
    {
      "epoch": 0.5610637748515362,
      "grad_norm": 4.752516746520996,
      "learning_rate": 4.115267132689927e-06,
      "loss": 0.0562,
      "step": 10865
    },
    {
      "epoch": 0.5611154144074361,
      "grad_norm": 2.56294322013855,
      "learning_rate": 4.114460713045917e-06,
      "loss": 0.0273,
      "step": 10866
    },
    {
      "epoch": 0.5611670539633359,
      "grad_norm": 8.355493545532227,
      "learning_rate": 4.113654317182245e-06,
      "loss": 0.111,
      "step": 10867
    },
    {
      "epoch": 0.5612186935192357,
      "grad_norm": 2.7760748863220215,
      "learning_rate": 4.112847945120567e-06,
      "loss": 0.0513,
      "step": 10868
    },
    {
      "epoch": 0.5612703330751355,
      "grad_norm": 8.618376731872559,
      "learning_rate": 4.112041596882539e-06,
      "loss": 0.0722,
      "step": 10869
    },
    {
      "epoch": 0.5613219726310353,
      "grad_norm": 8.501105308532715,
      "learning_rate": 4.111235272489813e-06,
      "loss": 0.0747,
      "step": 10870
    },
    {
      "epoch": 0.5613736121869352,
      "grad_norm": 3.4595351219177246,
      "learning_rate": 4.110428971964042e-06,
      "loss": 0.0454,
      "step": 10871
    },
    {
      "epoch": 0.561425251742835,
      "grad_norm": 6.150484085083008,
      "learning_rate": 4.109622695326879e-06,
      "loss": 0.0796,
      "step": 10872
    },
    {
      "epoch": 0.5614768912987348,
      "grad_norm": 16.398664474487305,
      "learning_rate": 4.108816442599975e-06,
      "loss": 0.0602,
      "step": 10873
    },
    {
      "epoch": 0.5615285308546346,
      "grad_norm": 2.427116870880127,
      "learning_rate": 4.108010213804983e-06,
      "loss": 0.028,
      "step": 10874
    },
    {
      "epoch": 0.5615801704105344,
      "grad_norm": 2.3658535480499268,
      "learning_rate": 4.10720400896355e-06,
      "loss": 0.0297,
      "step": 10875
    },
    {
      "epoch": 0.5616318099664342,
      "grad_norm": 4.8918776512146,
      "learning_rate": 4.1063978280973295e-06,
      "loss": 0.118,
      "step": 10876
    },
    {
      "epoch": 0.5616834495223341,
      "grad_norm": 4.960554122924805,
      "learning_rate": 4.1055916712279705e-06,
      "loss": 0.0545,
      "step": 10877
    },
    {
      "epoch": 0.5617350890782339,
      "grad_norm": 5.435064792633057,
      "learning_rate": 4.104785538377119e-06,
      "loss": 0.095,
      "step": 10878
    },
    {
      "epoch": 0.5617867286341337,
      "grad_norm": 4.110240459442139,
      "learning_rate": 4.103979429566425e-06,
      "loss": 0.0952,
      "step": 10879
    },
    {
      "epoch": 0.5618383681900335,
      "grad_norm": 13.061522483825684,
      "learning_rate": 4.103173344817534e-06,
      "loss": 0.1031,
      "step": 10880
    },
    {
      "epoch": 0.5618900077459333,
      "grad_norm": 3.405125617980957,
      "learning_rate": 4.102367284152098e-06,
      "loss": 0.0335,
      "step": 10881
    },
    {
      "epoch": 0.5619416473018332,
      "grad_norm": 3.967346668243408,
      "learning_rate": 4.1015612475917545e-06,
      "loss": 0.0646,
      "step": 10882
    },
    {
      "epoch": 0.561993286857733,
      "grad_norm": 4.701623916625977,
      "learning_rate": 4.1007552351581556e-06,
      "loss": 0.0717,
      "step": 10883
    },
    {
      "epoch": 0.5620449264136328,
      "grad_norm": 5.210406303405762,
      "learning_rate": 4.099949246872944e-06,
      "loss": 0.0396,
      "step": 10884
    },
    {
      "epoch": 0.5620965659695326,
      "grad_norm": 4.8375244140625,
      "learning_rate": 4.099143282757763e-06,
      "loss": 0.047,
      "step": 10885
    },
    {
      "epoch": 0.5621482055254324,
      "grad_norm": 3.413513422012329,
      "learning_rate": 4.098337342834256e-06,
      "loss": 0.034,
      "step": 10886
    },
    {
      "epoch": 0.5621998450813323,
      "grad_norm": 1.8382179737091064,
      "learning_rate": 4.097531427124065e-06,
      "loss": 0.0302,
      "step": 10887
    },
    {
      "epoch": 0.5622514846372321,
      "grad_norm": 10.23067569732666,
      "learning_rate": 4.0967255356488375e-06,
      "loss": 0.0765,
      "step": 10888
    },
    {
      "epoch": 0.5623031241931319,
      "grad_norm": 10.558683395385742,
      "learning_rate": 4.095919668430207e-06,
      "loss": 0.1066,
      "step": 10889
    },
    {
      "epoch": 0.5623547637490317,
      "grad_norm": 12.393244743347168,
      "learning_rate": 4.095113825489819e-06,
      "loss": 0.0927,
      "step": 10890
    },
    {
      "epoch": 0.5624064033049315,
      "grad_norm": 2.6328742504119873,
      "learning_rate": 4.094308006849314e-06,
      "loss": 0.0628,
      "step": 10891
    },
    {
      "epoch": 0.5624580428608313,
      "grad_norm": 14.832549095153809,
      "learning_rate": 4.0935022125303305e-06,
      "loss": 0.1912,
      "step": 10892
    },
    {
      "epoch": 0.5625096824167313,
      "grad_norm": 11.799562454223633,
      "learning_rate": 4.092696442554507e-06,
      "loss": 0.1208,
      "step": 10893
    },
    {
      "epoch": 0.5625613219726311,
      "grad_norm": 7.727143287658691,
      "learning_rate": 4.091890696943481e-06,
      "loss": 0.0402,
      "step": 10894
    },
    {
      "epoch": 0.5626129615285309,
      "grad_norm": 9.26682186126709,
      "learning_rate": 4.091084975718894e-06,
      "loss": 0.0704,
      "step": 10895
    },
    {
      "epoch": 0.5626646010844307,
      "grad_norm": 2.645397186279297,
      "learning_rate": 4.090279278902376e-06,
      "loss": 0.035,
      "step": 10896
    },
    {
      "epoch": 0.5627162406403305,
      "grad_norm": 6.252890110015869,
      "learning_rate": 4.089473606515569e-06,
      "loss": 0.0669,
      "step": 10897
    },
    {
      "epoch": 0.5627678801962304,
      "grad_norm": 1.8490432500839233,
      "learning_rate": 4.088667958580106e-06,
      "loss": 0.0226,
      "step": 10898
    },
    {
      "epoch": 0.5628195197521302,
      "grad_norm": 9.63810920715332,
      "learning_rate": 4.087862335117625e-06,
      "loss": 0.0624,
      "step": 10899
    },
    {
      "epoch": 0.56287115930803,
      "grad_norm": 15.07397174835205,
      "learning_rate": 4.087056736149756e-06,
      "loss": 0.1041,
      "step": 10900
    },
    {
      "epoch": 0.5629227988639298,
      "grad_norm": 9.940661430358887,
      "learning_rate": 4.086251161698133e-06,
      "loss": 0.0592,
      "step": 10901
    },
    {
      "epoch": 0.5629744384198296,
      "grad_norm": 2.129696846008301,
      "learning_rate": 4.0854456117843934e-06,
      "loss": 0.0226,
      "step": 10902
    },
    {
      "epoch": 0.5630260779757295,
      "grad_norm": 4.043282985687256,
      "learning_rate": 4.084640086430166e-06,
      "loss": 0.04,
      "step": 10903
    },
    {
      "epoch": 0.5630777175316293,
      "grad_norm": 2.6620240211486816,
      "learning_rate": 4.0838345856570834e-06,
      "loss": 0.039,
      "step": 10904
    },
    {
      "epoch": 0.5631293570875291,
      "grad_norm": 3.985761880874634,
      "learning_rate": 4.083029109486776e-06,
      "loss": 0.0742,
      "step": 10905
    },
    {
      "epoch": 0.5631809966434289,
      "grad_norm": 4.1820969581604,
      "learning_rate": 4.082223657940874e-06,
      "loss": 0.0549,
      "step": 10906
    },
    {
      "epoch": 0.5632326361993287,
      "grad_norm": 14.648961067199707,
      "learning_rate": 4.081418231041008e-06,
      "loss": 0.0905,
      "step": 10907
    },
    {
      "epoch": 0.5632842757552285,
      "grad_norm": 6.332015037536621,
      "learning_rate": 4.0806128288088045e-06,
      "loss": 0.1281,
      "step": 10908
    },
    {
      "epoch": 0.5633359153111284,
      "grad_norm": 16.45330047607422,
      "learning_rate": 4.079807451265895e-06,
      "loss": 0.0885,
      "step": 10909
    },
    {
      "epoch": 0.5633875548670282,
      "grad_norm": 3.3917624950408936,
      "learning_rate": 4.079002098433907e-06,
      "loss": 0.0842,
      "step": 10910
    },
    {
      "epoch": 0.563439194422928,
      "grad_norm": 4.602108955383301,
      "learning_rate": 4.0781967703344655e-06,
      "loss": 0.1061,
      "step": 10911
    },
    {
      "epoch": 0.5634908339788278,
      "grad_norm": 9.491086959838867,
      "learning_rate": 4.0773914669891976e-06,
      "loss": 0.1676,
      "step": 10912
    },
    {
      "epoch": 0.5635424735347276,
      "grad_norm": 9.776778221130371,
      "learning_rate": 4.076586188419728e-06,
      "loss": 0.1264,
      "step": 10913
    },
    {
      "epoch": 0.5635941130906275,
      "grad_norm": 6.271432399749756,
      "learning_rate": 4.075780934647686e-06,
      "loss": 0.0877,
      "step": 10914
    },
    {
      "epoch": 0.5636457526465273,
      "grad_norm": 9.032339096069336,
      "learning_rate": 4.07497570569469e-06,
      "loss": 0.0854,
      "step": 10915
    },
    {
      "epoch": 0.5636973922024271,
      "grad_norm": 8.31372356414795,
      "learning_rate": 4.074170501582368e-06,
      "loss": 0.0938,
      "step": 10916
    },
    {
      "epoch": 0.5637490317583269,
      "grad_norm": 2.5179994106292725,
      "learning_rate": 4.073365322332342e-06,
      "loss": 0.0479,
      "step": 10917
    },
    {
      "epoch": 0.5638006713142267,
      "grad_norm": 10.802301406860352,
      "learning_rate": 4.072560167966232e-06,
      "loss": 0.1267,
      "step": 10918
    },
    {
      "epoch": 0.5638523108701265,
      "grad_norm": 2.6703317165374756,
      "learning_rate": 4.071755038505663e-06,
      "loss": 0.041,
      "step": 10919
    },
    {
      "epoch": 0.5639039504260264,
      "grad_norm": 11.135527610778809,
      "learning_rate": 4.070949933972252e-06,
      "loss": 0.0643,
      "step": 10920
    },
    {
      "epoch": 0.5639555899819262,
      "grad_norm": 4.2720160484313965,
      "learning_rate": 4.0701448543876265e-06,
      "loss": 0.0678,
      "step": 10921
    },
    {
      "epoch": 0.564007229537826,
      "grad_norm": 5.2649827003479,
      "learning_rate": 4.069339799773397e-06,
      "loss": 0.0655,
      "step": 10922
    },
    {
      "epoch": 0.5640588690937258,
      "grad_norm": 4.577552318572998,
      "learning_rate": 4.068534770151189e-06,
      "loss": 0.0506,
      "step": 10923
    },
    {
      "epoch": 0.5641105086496256,
      "grad_norm": 2.3704185485839844,
      "learning_rate": 4.067729765542619e-06,
      "loss": 0.0456,
      "step": 10924
    },
    {
      "epoch": 0.5641621482055255,
      "grad_norm": 1.6091099977493286,
      "learning_rate": 4.066924785969305e-06,
      "loss": 0.0363,
      "step": 10925
    },
    {
      "epoch": 0.5642137877614253,
      "grad_norm": 3.372728109359741,
      "learning_rate": 4.066119831452862e-06,
      "loss": 0.0441,
      "step": 10926
    },
    {
      "epoch": 0.5642654273173251,
      "grad_norm": 8.38576889038086,
      "learning_rate": 4.0653149020149085e-06,
      "loss": 0.1651,
      "step": 10927
    },
    {
      "epoch": 0.5643170668732249,
      "grad_norm": 2.2447290420532227,
      "learning_rate": 4.064509997677061e-06,
      "loss": 0.0567,
      "step": 10928
    },
    {
      "epoch": 0.5643687064291247,
      "grad_norm": 7.248225212097168,
      "learning_rate": 4.063705118460931e-06,
      "loss": 0.0773,
      "step": 10929
    },
    {
      "epoch": 0.5644203459850246,
      "grad_norm": 3.128265142440796,
      "learning_rate": 4.062900264388135e-06,
      "loss": 0.0352,
      "step": 10930
    },
    {
      "epoch": 0.5644719855409244,
      "grad_norm": 5.261227607727051,
      "learning_rate": 4.062095435480286e-06,
      "loss": 0.0613,
      "step": 10931
    },
    {
      "epoch": 0.5645236250968242,
      "grad_norm": 1.8691465854644775,
      "learning_rate": 4.061290631758998e-06,
      "loss": 0.0483,
      "step": 10932
    },
    {
      "epoch": 0.564575264652724,
      "grad_norm": 3.5767102241516113,
      "learning_rate": 4.060485853245883e-06,
      "loss": 0.0441,
      "step": 10933
    },
    {
      "epoch": 0.5646269042086238,
      "grad_norm": 4.846078395843506,
      "learning_rate": 4.05968109996255e-06,
      "loss": 0.0367,
      "step": 10934
    },
    {
      "epoch": 0.5646785437645236,
      "grad_norm": 4.932077407836914,
      "learning_rate": 4.058876371930613e-06,
      "loss": 0.0403,
      "step": 10935
    },
    {
      "epoch": 0.5647301833204235,
      "grad_norm": 2.866225242614746,
      "learning_rate": 4.058071669171682e-06,
      "loss": 0.0735,
      "step": 10936
    },
    {
      "epoch": 0.5647818228763233,
      "grad_norm": 12.689428329467773,
      "learning_rate": 4.057266991707366e-06,
      "loss": 0.1484,
      "step": 10937
    },
    {
      "epoch": 0.5648334624322231,
      "grad_norm": 3.9753713607788086,
      "learning_rate": 4.056462339559273e-06,
      "loss": 0.0499,
      "step": 10938
    },
    {
      "epoch": 0.5648851019881229,
      "grad_norm": 5.676607131958008,
      "learning_rate": 4.0556577127490125e-06,
      "loss": 0.1451,
      "step": 10939
    },
    {
      "epoch": 0.5649367415440227,
      "grad_norm": 3.6908681392669678,
      "learning_rate": 4.05485311129819e-06,
      "loss": 0.0754,
      "step": 10940
    },
    {
      "epoch": 0.5649883810999226,
      "grad_norm": 4.9291276931762695,
      "learning_rate": 4.0540485352284135e-06,
      "loss": 0.0668,
      "step": 10941
    },
    {
      "epoch": 0.5650400206558224,
      "grad_norm": 5.395501136779785,
      "learning_rate": 4.05324398456129e-06,
      "loss": 0.0311,
      "step": 10942
    },
    {
      "epoch": 0.5650916602117222,
      "grad_norm": 2.8044474124908447,
      "learning_rate": 4.0524394593184256e-06,
      "loss": 0.0536,
      "step": 10943
    },
    {
      "epoch": 0.565143299767622,
      "grad_norm": 4.297303199768066,
      "learning_rate": 4.051634959521421e-06,
      "loss": 0.0705,
      "step": 10944
    },
    {
      "epoch": 0.5651949393235218,
      "grad_norm": 4.918227672576904,
      "learning_rate": 4.050830485191885e-06,
      "loss": 0.0708,
      "step": 10945
    },
    {
      "epoch": 0.5652465788794216,
      "grad_norm": 5.640282154083252,
      "learning_rate": 4.050026036351418e-06,
      "loss": 0.0593,
      "step": 10946
    },
    {
      "epoch": 0.5652982184353215,
      "grad_norm": 7.922787666320801,
      "learning_rate": 4.049221613021626e-06,
      "loss": 0.0604,
      "step": 10947
    },
    {
      "epoch": 0.5653498579912213,
      "grad_norm": 4.548568248748779,
      "learning_rate": 4.048417215224107e-06,
      "loss": 0.0408,
      "step": 10948
    },
    {
      "epoch": 0.5654014975471211,
      "grad_norm": 2.2340619564056396,
      "learning_rate": 4.047612842980464e-06,
      "loss": 0.0204,
      "step": 10949
    },
    {
      "epoch": 0.5654531371030209,
      "grad_norm": 2.5904836654663086,
      "learning_rate": 4.046808496312299e-06,
      "loss": 0.0379,
      "step": 10950
    },
    {
      "epoch": 0.5655047766589207,
      "grad_norm": 5.976500034332275,
      "learning_rate": 4.04600417524121e-06,
      "loss": 0.0645,
      "step": 10951
    },
    {
      "epoch": 0.5655564162148206,
      "grad_norm": 1.8105168342590332,
      "learning_rate": 4.045199879788798e-06,
      "loss": 0.0657,
      "step": 10952
    },
    {
      "epoch": 0.5656080557707204,
      "grad_norm": 5.1426801681518555,
      "learning_rate": 4.044395609976659e-06,
      "loss": 0.0561,
      "step": 10953
    },
    {
      "epoch": 0.5656596953266202,
      "grad_norm": 2.3335354328155518,
      "learning_rate": 4.0435913658263955e-06,
      "loss": 0.0238,
      "step": 10954
    },
    {
      "epoch": 0.56571133488252,
      "grad_norm": 7.7309489250183105,
      "learning_rate": 4.042787147359599e-06,
      "loss": 0.0572,
      "step": 10955
    },
    {
      "epoch": 0.5657629744384198,
      "grad_norm": 4.644815444946289,
      "learning_rate": 4.04198295459787e-06,
      "loss": 0.0594,
      "step": 10956
    },
    {
      "epoch": 0.5658146139943196,
      "grad_norm": 6.637940883636475,
      "learning_rate": 4.041178787562804e-06,
      "loss": 0.0813,
      "step": 10957
    },
    {
      "epoch": 0.5658662535502195,
      "grad_norm": 16.28322982788086,
      "learning_rate": 4.0403746462759956e-06,
      "loss": 0.1522,
      "step": 10958
    },
    {
      "epoch": 0.5659178931061193,
      "grad_norm": 7.831544399261475,
      "learning_rate": 4.039570530759039e-06,
      "loss": 0.0868,
      "step": 10959
    },
    {
      "epoch": 0.5659695326620191,
      "grad_norm": 2.2831013202667236,
      "learning_rate": 4.038766441033526e-06,
      "loss": 0.0252,
      "step": 10960
    },
    {
      "epoch": 0.5660211722179189,
      "grad_norm": 6.4619622230529785,
      "learning_rate": 4.037962377121057e-06,
      "loss": 0.0567,
      "step": 10961
    },
    {
      "epoch": 0.5660728117738187,
      "grad_norm": 5.478593349456787,
      "learning_rate": 4.037158339043214e-06,
      "loss": 0.0586,
      "step": 10962
    },
    {
      "epoch": 0.5661244513297186,
      "grad_norm": 7.560959339141846,
      "learning_rate": 4.036354326821595e-06,
      "loss": 0.0804,
      "step": 10963
    },
    {
      "epoch": 0.5661760908856184,
      "grad_norm": 11.801752090454102,
      "learning_rate": 4.035550340477792e-06,
      "loss": 0.0914,
      "step": 10964
    },
    {
      "epoch": 0.5662277304415182,
      "grad_norm": 4.941536903381348,
      "learning_rate": 4.0347463800333915e-06,
      "loss": 0.1037,
      "step": 10965
    },
    {
      "epoch": 0.566279369997418,
      "grad_norm": 3.096592426300049,
      "learning_rate": 4.033942445509986e-06,
      "loss": 0.0456,
      "step": 10966
    },
    {
      "epoch": 0.5663310095533178,
      "grad_norm": 5.030693531036377,
      "learning_rate": 4.033138536929161e-06,
      "loss": 0.0334,
      "step": 10967
    },
    {
      "epoch": 0.5663826491092177,
      "grad_norm": 2.854541540145874,
      "learning_rate": 4.032334654312509e-06,
      "loss": 0.0382,
      "step": 10968
    },
    {
      "epoch": 0.5664342886651175,
      "grad_norm": 1.5970274209976196,
      "learning_rate": 4.031530797681615e-06,
      "loss": 0.0135,
      "step": 10969
    },
    {
      "epoch": 0.5664859282210173,
      "grad_norm": 6.626004219055176,
      "learning_rate": 4.030726967058067e-06,
      "loss": 0.0919,
      "step": 10970
    },
    {
      "epoch": 0.5665375677769171,
      "grad_norm": 15.261619567871094,
      "learning_rate": 4.02992316246345e-06,
      "loss": 0.1227,
      "step": 10971
    },
    {
      "epoch": 0.5665892073328169,
      "grad_norm": 5.002558708190918,
      "learning_rate": 4.029119383919351e-06,
      "loss": 0.0461,
      "step": 10972
    },
    {
      "epoch": 0.5666408468887167,
      "grad_norm": 1.6160740852355957,
      "learning_rate": 4.028315631447352e-06,
      "loss": 0.016,
      "step": 10973
    },
    {
      "epoch": 0.5666924864446166,
      "grad_norm": 2.612661838531494,
      "learning_rate": 4.027511905069038e-06,
      "loss": 0.0451,
      "step": 10974
    },
    {
      "epoch": 0.5667441260005164,
      "grad_norm": 4.234030723571777,
      "learning_rate": 4.026708204805994e-06,
      "loss": 0.0329,
      "step": 10975
    },
    {
      "epoch": 0.5667957655564162,
      "grad_norm": 4.433470726013184,
      "learning_rate": 4.025904530679803e-06,
      "loss": 0.0809,
      "step": 10976
    },
    {
      "epoch": 0.566847405112316,
      "grad_norm": 2.6651246547698975,
      "learning_rate": 4.025100882712043e-06,
      "loss": 0.018,
      "step": 10977
    },
    {
      "epoch": 0.5668990446682158,
      "grad_norm": 2.2311525344848633,
      "learning_rate": 4.0242972609243e-06,
      "loss": 0.0146,
      "step": 10978
    },
    {
      "epoch": 0.5669506842241157,
      "grad_norm": 3.7695834636688232,
      "learning_rate": 4.02349366533815e-06,
      "loss": 0.0555,
      "step": 10979
    },
    {
      "epoch": 0.5670023237800155,
      "grad_norm": 3.024468421936035,
      "learning_rate": 4.022690095975178e-06,
      "loss": 0.0445,
      "step": 10980
    },
    {
      "epoch": 0.5670539633359153,
      "grad_norm": 10.03271198272705,
      "learning_rate": 4.0218865528569575e-06,
      "loss": 0.2054,
      "step": 10981
    },
    {
      "epoch": 0.5671056028918151,
      "grad_norm": 1.261991262435913,
      "learning_rate": 4.0210830360050706e-06,
      "loss": 0.0173,
      "step": 10982
    },
    {
      "epoch": 0.5671572424477149,
      "grad_norm": 2.7315683364868164,
      "learning_rate": 4.020279545441095e-06,
      "loss": 0.0409,
      "step": 10983
    },
    {
      "epoch": 0.5672088820036147,
      "grad_norm": 4.53544807434082,
      "learning_rate": 4.019476081186607e-06,
      "loss": 0.0747,
      "step": 10984
    },
    {
      "epoch": 0.5672605215595146,
      "grad_norm": 39.072914123535156,
      "learning_rate": 4.018672643263181e-06,
      "loss": 0.095,
      "step": 10985
    },
    {
      "epoch": 0.5673121611154144,
      "grad_norm": 8.055767059326172,
      "learning_rate": 4.017869231692393e-06,
      "loss": 0.0343,
      "step": 10986
    },
    {
      "epoch": 0.5673638006713142,
      "grad_norm": 2.99442195892334,
      "learning_rate": 4.017065846495823e-06,
      "loss": 0.0402,
      "step": 10987
    },
    {
      "epoch": 0.567415440227214,
      "grad_norm": 5.350686550140381,
      "learning_rate": 4.016262487695038e-06,
      "loss": 0.0792,
      "step": 10988
    },
    {
      "epoch": 0.5674670797831138,
      "grad_norm": 2.547053813934326,
      "learning_rate": 4.015459155311616e-06,
      "loss": 0.0193,
      "step": 10989
    },
    {
      "epoch": 0.5675187193390137,
      "grad_norm": 8.990994453430176,
      "learning_rate": 4.014655849367127e-06,
      "loss": 0.0373,
      "step": 10990
    },
    {
      "epoch": 0.5675703588949135,
      "grad_norm": 2.877718925476074,
      "learning_rate": 4.013852569883147e-06,
      "loss": 0.0384,
      "step": 10991
    },
    {
      "epoch": 0.5676219984508133,
      "grad_norm": 3.042971611022949,
      "learning_rate": 4.013049316881243e-06,
      "loss": 0.0565,
      "step": 10992
    },
    {
      "epoch": 0.5676736380067131,
      "grad_norm": 4.548878192901611,
      "learning_rate": 4.012246090382987e-06,
      "loss": 0.0398,
      "step": 10993
    },
    {
      "epoch": 0.5677252775626129,
      "grad_norm": 6.035217761993408,
      "learning_rate": 4.011442890409953e-06,
      "loss": 0.0954,
      "step": 10994
    },
    {
      "epoch": 0.5677769171185127,
      "grad_norm": 15.412093162536621,
      "learning_rate": 4.010639716983702e-06,
      "loss": 0.1803,
      "step": 10995
    },
    {
      "epoch": 0.5678285566744126,
      "grad_norm": 2.025624990463257,
      "learning_rate": 4.009836570125809e-06,
      "loss": 0.0283,
      "step": 10996
    },
    {
      "epoch": 0.5678801962303124,
      "grad_norm": 5.841176986694336,
      "learning_rate": 4.009033449857839e-06,
      "loss": 0.0785,
      "step": 10997
    },
    {
      "epoch": 0.5679318357862122,
      "grad_norm": 4.929833889007568,
      "learning_rate": 4.008230356201362e-06,
      "loss": 0.061,
      "step": 10998
    },
    {
      "epoch": 0.567983475342112,
      "grad_norm": 2.3134725093841553,
      "learning_rate": 4.007427289177941e-06,
      "loss": 0.0291,
      "step": 10999
    },
    {
      "epoch": 0.5680351148980118,
      "grad_norm": 3.284177303314209,
      "learning_rate": 4.006624248809141e-06,
      "loss": 0.0363,
      "step": 11000
    },
    {
      "epoch": 0.5680867544539117,
      "grad_norm": 3.4500393867492676,
      "learning_rate": 4.00582123511653e-06,
      "loss": 0.0391,
      "step": 11001
    },
    {
      "epoch": 0.5681383940098115,
      "grad_norm": 1.9448312520980835,
      "learning_rate": 4.005018248121671e-06,
      "loss": 0.032,
      "step": 11002
    },
    {
      "epoch": 0.5681900335657113,
      "grad_norm": 2.9579005241394043,
      "learning_rate": 4.004215287846127e-06,
      "loss": 0.0248,
      "step": 11003
    },
    {
      "epoch": 0.5682416731216111,
      "grad_norm": 2.246645212173462,
      "learning_rate": 4.003412354311462e-06,
      "loss": 0.0182,
      "step": 11004
    },
    {
      "epoch": 0.5682933126775109,
      "grad_norm": 2.852443218231201,
      "learning_rate": 4.002609447539238e-06,
      "loss": 0.0284,
      "step": 11005
    },
    {
      "epoch": 0.5683449522334107,
      "grad_norm": 1.7839133739471436,
      "learning_rate": 4.001806567551013e-06,
      "loss": 0.0257,
      "step": 11006
    },
    {
      "epoch": 0.5683965917893106,
      "grad_norm": 7.894222259521484,
      "learning_rate": 4.00100371436835e-06,
      "loss": 0.0438,
      "step": 11007
    },
    {
      "epoch": 0.5684482313452104,
      "grad_norm": 5.558276176452637,
      "learning_rate": 4.00020088801281e-06,
      "loss": 0.0696,
      "step": 11008
    },
    {
      "epoch": 0.5684998709011102,
      "grad_norm": 13.678121566772461,
      "learning_rate": 3.999398088505952e-06,
      "loss": 0.1503,
      "step": 11009
    },
    {
      "epoch": 0.56855151045701,
      "grad_norm": 13.722709655761719,
      "learning_rate": 3.998595315869332e-06,
      "loss": 0.0813,
      "step": 11010
    },
    {
      "epoch": 0.5686031500129098,
      "grad_norm": 5.142416000366211,
      "learning_rate": 3.9977925701245105e-06,
      "loss": 0.0977,
      "step": 11011
    },
    {
      "epoch": 0.5686547895688097,
      "grad_norm": 15.295707702636719,
      "learning_rate": 3.996989851293042e-06,
      "loss": 0.1298,
      "step": 11012
    },
    {
      "epoch": 0.5687064291247095,
      "grad_norm": 4.364101886749268,
      "learning_rate": 3.9961871593964865e-06,
      "loss": 0.0383,
      "step": 11013
    },
    {
      "epoch": 0.5687580686806093,
      "grad_norm": 4.00634241104126,
      "learning_rate": 3.995384494456394e-06,
      "loss": 0.0826,
      "step": 11014
    },
    {
      "epoch": 0.5688097082365091,
      "grad_norm": 4.679533958435059,
      "learning_rate": 3.994581856494324e-06,
      "loss": 0.0439,
      "step": 11015
    },
    {
      "epoch": 0.5688613477924089,
      "grad_norm": 3.992490530014038,
      "learning_rate": 3.99377924553183e-06,
      "loss": 0.0728,
      "step": 11016
    },
    {
      "epoch": 0.5689129873483088,
      "grad_norm": 10.085114479064941,
      "learning_rate": 3.992976661590463e-06,
      "loss": 0.0948,
      "step": 11017
    },
    {
      "epoch": 0.5689646269042086,
      "grad_norm": 5.106435298919678,
      "learning_rate": 3.992174104691777e-06,
      "loss": 0.0368,
      "step": 11018
    },
    {
      "epoch": 0.5690162664601084,
      "grad_norm": 2.640735387802124,
      "learning_rate": 3.991371574857323e-06,
      "loss": 0.0509,
      "step": 11019
    },
    {
      "epoch": 0.5690679060160082,
      "grad_norm": 3.5454459190368652,
      "learning_rate": 3.9905690721086565e-06,
      "loss": 0.0157,
      "step": 11020
    },
    {
      "epoch": 0.5691195455719081,
      "grad_norm": 1.6824924945831299,
      "learning_rate": 3.989766596467321e-06,
      "loss": 0.0329,
      "step": 11021
    },
    {
      "epoch": 0.569171185127808,
      "grad_norm": 2.5095643997192383,
      "learning_rate": 3.988964147954871e-06,
      "loss": 0.0374,
      "step": 11022
    },
    {
      "epoch": 0.5692228246837078,
      "grad_norm": 1.9502862691879272,
      "learning_rate": 3.988161726592855e-06,
      "loss": 0.0409,
      "step": 11023
    },
    {
      "epoch": 0.5692744642396076,
      "grad_norm": 1.6267305612564087,
      "learning_rate": 3.9873593324028205e-06,
      "loss": 0.028,
      "step": 11024
    },
    {
      "epoch": 0.5693261037955074,
      "grad_norm": 4.892776012420654,
      "learning_rate": 3.9865569654063155e-06,
      "loss": 0.0419,
      "step": 11025
    },
    {
      "epoch": 0.5693777433514072,
      "grad_norm": 17.690637588500977,
      "learning_rate": 3.985754625624885e-06,
      "loss": 0.077,
      "step": 11026
    },
    {
      "epoch": 0.569429382907307,
      "grad_norm": 7.990448951721191,
      "learning_rate": 3.98495231308008e-06,
      "loss": 0.0887,
      "step": 11027
    },
    {
      "epoch": 0.5694810224632069,
      "grad_norm": 17.89295768737793,
      "learning_rate": 3.984150027793439e-06,
      "loss": 0.1004,
      "step": 11028
    },
    {
      "epoch": 0.5695326620191067,
      "grad_norm": 3.2472150325775146,
      "learning_rate": 3.983347769786511e-06,
      "loss": 0.0296,
      "step": 11029
    },
    {
      "epoch": 0.5695843015750065,
      "grad_norm": 3.1155924797058105,
      "learning_rate": 3.98254553908084e-06,
      "loss": 0.061,
      "step": 11030
    },
    {
      "epoch": 0.5696359411309063,
      "grad_norm": 2.2757041454315186,
      "learning_rate": 3.981743335697968e-06,
      "loss": 0.0474,
      "step": 11031
    },
    {
      "epoch": 0.5696875806868061,
      "grad_norm": 10.181480407714844,
      "learning_rate": 3.980941159659437e-06,
      "loss": 0.1856,
      "step": 11032
    },
    {
      "epoch": 0.569739220242706,
      "grad_norm": 8.489794731140137,
      "learning_rate": 3.980139010986788e-06,
      "loss": 0.0658,
      "step": 11033
    },
    {
      "epoch": 0.5697908597986058,
      "grad_norm": 3.8195695877075195,
      "learning_rate": 3.979336889701566e-06,
      "loss": 0.0656,
      "step": 11034
    },
    {
      "epoch": 0.5698424993545056,
      "grad_norm": 1.6570978164672852,
      "learning_rate": 3.978534795825308e-06,
      "loss": 0.0151,
      "step": 11035
    },
    {
      "epoch": 0.5698941389104054,
      "grad_norm": 6.467540264129639,
      "learning_rate": 3.9777327293795534e-06,
      "loss": 0.056,
      "step": 11036
    },
    {
      "epoch": 0.5699457784663052,
      "grad_norm": 5.8188605308532715,
      "learning_rate": 3.976930690385841e-06,
      "loss": 0.068,
      "step": 11037
    },
    {
      "epoch": 0.569997418022205,
      "grad_norm": 14.228700637817383,
      "learning_rate": 3.976128678865711e-06,
      "loss": 0.1132,
      "step": 11038
    },
    {
      "epoch": 0.5700490575781049,
      "grad_norm": 3.684112310409546,
      "learning_rate": 3.9753266948406975e-06,
      "loss": 0.0589,
      "step": 11039
    },
    {
      "epoch": 0.5701006971340047,
      "grad_norm": 3.2455861568450928,
      "learning_rate": 3.974524738332338e-06,
      "loss": 0.0456,
      "step": 11040
    },
    {
      "epoch": 0.5701523366899045,
      "grad_norm": 7.125722408294678,
      "learning_rate": 3.973722809362169e-06,
      "loss": 0.0808,
      "step": 11041
    },
    {
      "epoch": 0.5702039762458043,
      "grad_norm": 2.5664751529693604,
      "learning_rate": 3.972920907951728e-06,
      "loss": 0.0539,
      "step": 11042
    },
    {
      "epoch": 0.5702556158017041,
      "grad_norm": 2.532911539077759,
      "learning_rate": 3.972119034122544e-06,
      "loss": 0.0242,
      "step": 11043
    },
    {
      "epoch": 0.570307255357604,
      "grad_norm": 3.872544527053833,
      "learning_rate": 3.971317187896155e-06,
      "loss": 0.0284,
      "step": 11044
    },
    {
      "epoch": 0.5703588949135038,
      "grad_norm": 12.661962509155273,
      "learning_rate": 3.970515369294092e-06,
      "loss": 0.0763,
      "step": 11045
    },
    {
      "epoch": 0.5704105344694036,
      "grad_norm": 11.971898078918457,
      "learning_rate": 3.969713578337887e-06,
      "loss": 0.1802,
      "step": 11046
    },
    {
      "epoch": 0.5704621740253034,
      "grad_norm": 7.259903430938721,
      "learning_rate": 3.968911815049071e-06,
      "loss": 0.0449,
      "step": 11047
    },
    {
      "epoch": 0.5705138135812032,
      "grad_norm": 8.041376113891602,
      "learning_rate": 3.968110079449176e-06,
      "loss": 0.1155,
      "step": 11048
    },
    {
      "epoch": 0.570565453137103,
      "grad_norm": 1.231109380722046,
      "learning_rate": 3.967308371559731e-06,
      "loss": 0.0318,
      "step": 11049
    },
    {
      "epoch": 0.5706170926930029,
      "grad_norm": 2.8512754440307617,
      "learning_rate": 3.966506691402265e-06,
      "loss": 0.0601,
      "step": 11050
    },
    {
      "epoch": 0.5706687322489027,
      "grad_norm": 4.099743843078613,
      "learning_rate": 3.965705038998306e-06,
      "loss": 0.0602,
      "step": 11051
    },
    {
      "epoch": 0.5707203718048025,
      "grad_norm": 7.373632907867432,
      "learning_rate": 3.964903414369382e-06,
      "loss": 0.1313,
      "step": 11052
    },
    {
      "epoch": 0.5707720113607023,
      "grad_norm": 3.8768744468688965,
      "learning_rate": 3.964101817537024e-06,
      "loss": 0.0327,
      "step": 11053
    },
    {
      "epoch": 0.5708236509166021,
      "grad_norm": 2.7340996265411377,
      "learning_rate": 3.9633002485227505e-06,
      "loss": 0.0269,
      "step": 11054
    },
    {
      "epoch": 0.570875290472502,
      "grad_norm": 4.595797538757324,
      "learning_rate": 3.962498707348091e-06,
      "loss": 0.0461,
      "step": 11055
    },
    {
      "epoch": 0.5709269300284018,
      "grad_norm": 4.1020379066467285,
      "learning_rate": 3.961697194034572e-06,
      "loss": 0.0313,
      "step": 11056
    },
    {
      "epoch": 0.5709785695843016,
      "grad_norm": 8.269216537475586,
      "learning_rate": 3.960895708603713e-06,
      "loss": 0.0601,
      "step": 11057
    },
    {
      "epoch": 0.5710302091402014,
      "grad_norm": 6.498102188110352,
      "learning_rate": 3.96009425107704e-06,
      "loss": 0.0233,
      "step": 11058
    },
    {
      "epoch": 0.5710818486961012,
      "grad_norm": 3.4905755519866943,
      "learning_rate": 3.959292821476074e-06,
      "loss": 0.0498,
      "step": 11059
    },
    {
      "epoch": 0.571133488252001,
      "grad_norm": 3.0130348205566406,
      "learning_rate": 3.95849141982234e-06,
      "loss": 0.0539,
      "step": 11060
    },
    {
      "epoch": 0.5711851278079009,
      "grad_norm": 6.727479934692383,
      "learning_rate": 3.957690046137353e-06,
      "loss": 0.0649,
      "step": 11061
    },
    {
      "epoch": 0.5712367673638007,
      "grad_norm": 3.004549980163574,
      "learning_rate": 3.9568887004426385e-06,
      "loss": 0.0375,
      "step": 11062
    },
    {
      "epoch": 0.5712884069197005,
      "grad_norm": 3.0516417026519775,
      "learning_rate": 3.956087382759713e-06,
      "loss": 0.0334,
      "step": 11063
    },
    {
      "epoch": 0.5713400464756003,
      "grad_norm": 10.579197883605957,
      "learning_rate": 3.955286093110098e-06,
      "loss": 0.0582,
      "step": 11064
    },
    {
      "epoch": 0.5713916860315001,
      "grad_norm": 14.656830787658691,
      "learning_rate": 3.9544848315153075e-06,
      "loss": 0.0835,
      "step": 11065
    },
    {
      "epoch": 0.5714433255874,
      "grad_norm": 0.7912313938140869,
      "learning_rate": 3.953683597996859e-06,
      "loss": 0.0097,
      "step": 11066
    },
    {
      "epoch": 0.5714949651432998,
      "grad_norm": 2.997136354446411,
      "learning_rate": 3.952882392576275e-06,
      "loss": 0.0213,
      "step": 11067
    },
    {
      "epoch": 0.5715466046991996,
      "grad_norm": 5.824460983276367,
      "learning_rate": 3.952081215275063e-06,
      "loss": 0.0405,
      "step": 11068
    },
    {
      "epoch": 0.5715982442550994,
      "grad_norm": 2.408402919769287,
      "learning_rate": 3.951280066114742e-06,
      "loss": 0.0418,
      "step": 11069
    },
    {
      "epoch": 0.5716498838109992,
      "grad_norm": 11.856775283813477,
      "learning_rate": 3.950478945116826e-06,
      "loss": 0.1319,
      "step": 11070
    },
    {
      "epoch": 0.5717015233668991,
      "grad_norm": 18.76457405090332,
      "learning_rate": 3.949677852302829e-06,
      "loss": 0.149,
      "step": 11071
    },
    {
      "epoch": 0.5717531629227989,
      "grad_norm": 6.1285176277160645,
      "learning_rate": 3.948876787694261e-06,
      "loss": 0.0777,
      "step": 11072
    },
    {
      "epoch": 0.5718048024786987,
      "grad_norm": 4.824357032775879,
      "learning_rate": 3.948075751312634e-06,
      "loss": 0.0502,
      "step": 11073
    },
    {
      "epoch": 0.5718564420345985,
      "grad_norm": 3.4890522956848145,
      "learning_rate": 3.947274743179462e-06,
      "loss": 0.0232,
      "step": 11074
    },
    {
      "epoch": 0.5719080815904983,
      "grad_norm": 5.498516082763672,
      "learning_rate": 3.946473763316256e-06,
      "loss": 0.1244,
      "step": 11075
    },
    {
      "epoch": 0.5719597211463981,
      "grad_norm": 16.195796966552734,
      "learning_rate": 3.9456728117445216e-06,
      "loss": 0.0986,
      "step": 11076
    },
    {
      "epoch": 0.572011360702298,
      "grad_norm": 8.26898193359375,
      "learning_rate": 3.944871888485769e-06,
      "loss": 0.096,
      "step": 11077
    },
    {
      "epoch": 0.5720630002581978,
      "grad_norm": 8.853918075561523,
      "learning_rate": 3.944070993561508e-06,
      "loss": 0.07,
      "step": 11078
    },
    {
      "epoch": 0.5721146398140976,
      "grad_norm": 2.6399097442626953,
      "learning_rate": 3.943270126993244e-06,
      "loss": 0.0313,
      "step": 11079
    },
    {
      "epoch": 0.5721662793699974,
      "grad_norm": 13.92055892944336,
      "learning_rate": 3.942469288802482e-06,
      "loss": 0.0966,
      "step": 11080
    },
    {
      "epoch": 0.5722179189258972,
      "grad_norm": 3.0992658138275146,
      "learning_rate": 3.9416684790107315e-06,
      "loss": 0.0594,
      "step": 11081
    },
    {
      "epoch": 0.5722695584817971,
      "grad_norm": 7.601799964904785,
      "learning_rate": 3.940867697639497e-06,
      "loss": 0.0359,
      "step": 11082
    },
    {
      "epoch": 0.5723211980376969,
      "grad_norm": 11.839635848999023,
      "learning_rate": 3.9400669447102805e-06,
      "loss": 0.0706,
      "step": 11083
    },
    {
      "epoch": 0.5723728375935967,
      "grad_norm": 2.22711443901062,
      "learning_rate": 3.939266220244585e-06,
      "loss": 0.0328,
      "step": 11084
    },
    {
      "epoch": 0.5724244771494965,
      "grad_norm": 10.923615455627441,
      "learning_rate": 3.9384655242639155e-06,
      "loss": 0.089,
      "step": 11085
    },
    {
      "epoch": 0.5724761167053963,
      "grad_norm": 12.488141059875488,
      "learning_rate": 3.937664856789777e-06,
      "loss": 0.1226,
      "step": 11086
    },
    {
      "epoch": 0.5725277562612962,
      "grad_norm": 4.149901390075684,
      "learning_rate": 3.936864217843662e-06,
      "loss": 0.0423,
      "step": 11087
    },
    {
      "epoch": 0.572579395817196,
      "grad_norm": 19.794803619384766,
      "learning_rate": 3.9360636074470785e-06,
      "loss": 0.1365,
      "step": 11088
    },
    {
      "epoch": 0.5726310353730958,
      "grad_norm": 2.7392406463623047,
      "learning_rate": 3.935263025621523e-06,
      "loss": 0.0507,
      "step": 11089
    },
    {
      "epoch": 0.5726826749289956,
      "grad_norm": 6.320991516113281,
      "learning_rate": 3.934462472388495e-06,
      "loss": 0.0779,
      "step": 11090
    },
    {
      "epoch": 0.5727343144848954,
      "grad_norm": 2.955618381500244,
      "learning_rate": 3.9336619477694924e-06,
      "loss": 0.0339,
      "step": 11091
    },
    {
      "epoch": 0.5727859540407952,
      "grad_norm": 2.2993648052215576,
      "learning_rate": 3.932861451786011e-06,
      "loss": 0.0276,
      "step": 11092
    },
    {
      "epoch": 0.5728375935966951,
      "grad_norm": 6.078196048736572,
      "learning_rate": 3.9320609844595534e-06,
      "loss": 0.0711,
      "step": 11093
    },
    {
      "epoch": 0.5728892331525949,
      "grad_norm": 3.378786563873291,
      "learning_rate": 3.931260545811607e-06,
      "loss": 0.0366,
      "step": 11094
    },
    {
      "epoch": 0.5729408727084947,
      "grad_norm": 14.47591495513916,
      "learning_rate": 3.930460135863672e-06,
      "loss": 0.1913,
      "step": 11095
    },
    {
      "epoch": 0.5729925122643945,
      "grad_norm": 5.652558326721191,
      "learning_rate": 3.929659754637241e-06,
      "loss": 0.0567,
      "step": 11096
    },
    {
      "epoch": 0.5730441518202943,
      "grad_norm": 2.7468605041503906,
      "learning_rate": 3.92885940215381e-06,
      "loss": 0.0394,
      "step": 11097
    },
    {
      "epoch": 0.5730957913761942,
      "grad_norm": 1.7393299341201782,
      "learning_rate": 3.928059078434868e-06,
      "loss": 0.0173,
      "step": 11098
    },
    {
      "epoch": 0.573147430932094,
      "grad_norm": 6.766208648681641,
      "learning_rate": 3.927258783501908e-06,
      "loss": 0.0754,
      "step": 11099
    },
    {
      "epoch": 0.5731990704879938,
      "grad_norm": 18.45243263244629,
      "learning_rate": 3.9264585173764255e-06,
      "loss": 0.1949,
      "step": 11100
    },
    {
      "epoch": 0.5732507100438936,
      "grad_norm": 3.1846823692321777,
      "learning_rate": 3.925658280079904e-06,
      "loss": 0.0596,
      "step": 11101
    },
    {
      "epoch": 0.5733023495997934,
      "grad_norm": 2.7686080932617188,
      "learning_rate": 3.924858071633837e-06,
      "loss": 0.0527,
      "step": 11102
    },
    {
      "epoch": 0.5733539891556932,
      "grad_norm": 1.6839276552200317,
      "learning_rate": 3.924057892059713e-06,
      "loss": 0.0295,
      "step": 11103
    },
    {
      "epoch": 0.5734056287115931,
      "grad_norm": 4.862895965576172,
      "learning_rate": 3.923257741379021e-06,
      "loss": 0.0625,
      "step": 11104
    },
    {
      "epoch": 0.5734572682674929,
      "grad_norm": 15.119518280029297,
      "learning_rate": 3.922457619613247e-06,
      "loss": 0.081,
      "step": 11105
    },
    {
      "epoch": 0.5735089078233927,
      "grad_norm": 2.529276132583618,
      "learning_rate": 3.921657526783875e-06,
      "loss": 0.0261,
      "step": 11106
    },
    {
      "epoch": 0.5735605473792925,
      "grad_norm": 4.2559285163879395,
      "learning_rate": 3.920857462912396e-06,
      "loss": 0.0598,
      "step": 11107
    },
    {
      "epoch": 0.5736121869351923,
      "grad_norm": 3.230227470397949,
      "learning_rate": 3.920057428020293e-06,
      "loss": 0.0489,
      "step": 11108
    },
    {
      "epoch": 0.5736638264910922,
      "grad_norm": 10.33010196685791,
      "learning_rate": 3.919257422129048e-06,
      "loss": 0.0819,
      "step": 11109
    },
    {
      "epoch": 0.573715466046992,
      "grad_norm": 3.533099412918091,
      "learning_rate": 3.918457445260146e-06,
      "loss": 0.0378,
      "step": 11110
    },
    {
      "epoch": 0.5737671056028918,
      "grad_norm": 3.843013286590576,
      "learning_rate": 3.917657497435073e-06,
      "loss": 0.0283,
      "step": 11111
    },
    {
      "epoch": 0.5738187451587916,
      "grad_norm": 2.3212387561798096,
      "learning_rate": 3.916857578675304e-06,
      "loss": 0.0244,
      "step": 11112
    },
    {
      "epoch": 0.5738703847146914,
      "grad_norm": 3.1800537109375,
      "learning_rate": 3.916057689002324e-06,
      "loss": 0.0484,
      "step": 11113
    },
    {
      "epoch": 0.5739220242705912,
      "grad_norm": 3.5500869750976562,
      "learning_rate": 3.915257828437613e-06,
      "loss": 0.0661,
      "step": 11114
    },
    {
      "epoch": 0.5739736638264911,
      "grad_norm": 4.20070219039917,
      "learning_rate": 3.9144579970026524e-06,
      "loss": 0.0755,
      "step": 11115
    },
    {
      "epoch": 0.5740253033823909,
      "grad_norm": 9.980060577392578,
      "learning_rate": 3.913658194718918e-06,
      "loss": 0.0562,
      "step": 11116
    },
    {
      "epoch": 0.5740769429382907,
      "grad_norm": 2.5432071685791016,
      "learning_rate": 3.912858421607888e-06,
      "loss": 0.0596,
      "step": 11117
    },
    {
      "epoch": 0.5741285824941905,
      "grad_norm": 7.448259353637695,
      "learning_rate": 3.912058677691039e-06,
      "loss": 0.0494,
      "step": 11118
    },
    {
      "epoch": 0.5741802220500903,
      "grad_norm": 2.7118191719055176,
      "learning_rate": 3.911258962989853e-06,
      "loss": 0.0443,
      "step": 11119
    },
    {
      "epoch": 0.5742318616059902,
      "grad_norm": 5.516829013824463,
      "learning_rate": 3.910459277525798e-06,
      "loss": 0.0403,
      "step": 11120
    },
    {
      "epoch": 0.57428350116189,
      "grad_norm": 2.669914960861206,
      "learning_rate": 3.909659621320352e-06,
      "loss": 0.0445,
      "step": 11121
    },
    {
      "epoch": 0.5743351407177898,
      "grad_norm": 8.3796968460083,
      "learning_rate": 3.908859994394992e-06,
      "loss": 0.0434,
      "step": 11122
    },
    {
      "epoch": 0.5743867802736896,
      "grad_norm": 5.090461254119873,
      "learning_rate": 3.908060396771186e-06,
      "loss": 0.0266,
      "step": 11123
    },
    {
      "epoch": 0.5744384198295894,
      "grad_norm": 11.125322341918945,
      "learning_rate": 3.9072608284704095e-06,
      "loss": 0.1197,
      "step": 11124
    },
    {
      "epoch": 0.5744900593854892,
      "grad_norm": 6.577829837799072,
      "learning_rate": 3.9064612895141315e-06,
      "loss": 0.103,
      "step": 11125
    },
    {
      "epoch": 0.5745416989413891,
      "grad_norm": 2.6135637760162354,
      "learning_rate": 3.905661779923829e-06,
      "loss": 0.0537,
      "step": 11126
    },
    {
      "epoch": 0.5745933384972889,
      "grad_norm": 2.7769477367401123,
      "learning_rate": 3.904862299720964e-06,
      "loss": 0.0518,
      "step": 11127
    },
    {
      "epoch": 0.5746449780531887,
      "grad_norm": 2.740772247314453,
      "learning_rate": 3.904062848927012e-06,
      "loss": 0.019,
      "step": 11128
    },
    {
      "epoch": 0.5746966176090885,
      "grad_norm": 8.769685745239258,
      "learning_rate": 3.903263427563439e-06,
      "loss": 0.0609,
      "step": 11129
    },
    {
      "epoch": 0.5747482571649883,
      "grad_norm": 5.677814483642578,
      "learning_rate": 3.9024640356517135e-06,
      "loss": 0.0475,
      "step": 11130
    },
    {
      "epoch": 0.5747998967208882,
      "grad_norm": 9.874110221862793,
      "learning_rate": 3.901664673213301e-06,
      "loss": 0.1049,
      "step": 11131
    },
    {
      "epoch": 0.574851536276788,
      "grad_norm": 4.31019401550293,
      "learning_rate": 3.900865340269668e-06,
      "loss": 0.0662,
      "step": 11132
    },
    {
      "epoch": 0.5749031758326878,
      "grad_norm": 5.782042026519775,
      "learning_rate": 3.900066036842283e-06,
      "loss": 0.0724,
      "step": 11133
    },
    {
      "epoch": 0.5749548153885876,
      "grad_norm": 2.291386365890503,
      "learning_rate": 3.899266762952605e-06,
      "loss": 0.0251,
      "step": 11134
    },
    {
      "epoch": 0.5750064549444874,
      "grad_norm": 5.628931045532227,
      "learning_rate": 3.898467518622102e-06,
      "loss": 0.0679,
      "step": 11135
    },
    {
      "epoch": 0.5750580945003873,
      "grad_norm": 5.234676361083984,
      "learning_rate": 3.8976683038722354e-06,
      "loss": 0.0631,
      "step": 11136
    },
    {
      "epoch": 0.5751097340562871,
      "grad_norm": 0.7673627138137817,
      "learning_rate": 3.89686911872447e-06,
      "loss": 0.008,
      "step": 11137
    },
    {
      "epoch": 0.5751613736121869,
      "grad_norm": 1.822398066520691,
      "learning_rate": 3.8960699632002626e-06,
      "loss": 0.0286,
      "step": 11138
    },
    {
      "epoch": 0.5752130131680867,
      "grad_norm": 7.719532012939453,
      "learning_rate": 3.895270837321075e-06,
      "loss": 0.0606,
      "step": 11139
    },
    {
      "epoch": 0.5752646527239865,
      "grad_norm": 4.170701503753662,
      "learning_rate": 3.894471741108369e-06,
      "loss": 0.0645,
      "step": 11140
    },
    {
      "epoch": 0.5753162922798863,
      "grad_norm": 4.299149990081787,
      "learning_rate": 3.893672674583604e-06,
      "loss": 0.0553,
      "step": 11141
    },
    {
      "epoch": 0.5753679318357862,
      "grad_norm": 2.5019984245300293,
      "learning_rate": 3.8928736377682364e-06,
      "loss": 0.0439,
      "step": 11142
    },
    {
      "epoch": 0.575419571391686,
      "grad_norm": 4.9428300857543945,
      "learning_rate": 3.892074630683723e-06,
      "loss": 0.0259,
      "step": 11143
    },
    {
      "epoch": 0.5754712109475858,
      "grad_norm": 4.555307865142822,
      "learning_rate": 3.891275653351523e-06,
      "loss": 0.0973,
      "step": 11144
    },
    {
      "epoch": 0.5755228505034856,
      "grad_norm": 2.516684055328369,
      "learning_rate": 3.89047670579309e-06,
      "loss": 0.0303,
      "step": 11145
    },
    {
      "epoch": 0.5755744900593854,
      "grad_norm": 7.667434215545654,
      "learning_rate": 3.889677788029877e-06,
      "loss": 0.0842,
      "step": 11146
    },
    {
      "epoch": 0.5756261296152853,
      "grad_norm": 2.2262868881225586,
      "learning_rate": 3.888878900083344e-06,
      "loss": 0.0298,
      "step": 11147
    },
    {
      "epoch": 0.5756777691711852,
      "grad_norm": 2.3388772010803223,
      "learning_rate": 3.888080041974941e-06,
      "loss": 0.0531,
      "step": 11148
    },
    {
      "epoch": 0.575729408727085,
      "grad_norm": 2.5966382026672363,
      "learning_rate": 3.887281213726119e-06,
      "loss": 0.0353,
      "step": 11149
    },
    {
      "epoch": 0.5757810482829848,
      "grad_norm": 4.049068927764893,
      "learning_rate": 3.886482415358333e-06,
      "loss": 0.061,
      "step": 11150
    },
    {
      "epoch": 0.5758326878388846,
      "grad_norm": 6.988111972808838,
      "learning_rate": 3.885683646893031e-06,
      "loss": 0.1024,
      "step": 11151
    },
    {
      "epoch": 0.5758843273947845,
      "grad_norm": 2.9257888793945312,
      "learning_rate": 3.884884908351668e-06,
      "loss": 0.038,
      "step": 11152
    },
    {
      "epoch": 0.5759359669506843,
      "grad_norm": 7.23843240737915,
      "learning_rate": 3.8840861997556875e-06,
      "loss": 0.0756,
      "step": 11153
    },
    {
      "epoch": 0.5759876065065841,
      "grad_norm": 5.133107662200928,
      "learning_rate": 3.8832875211265405e-06,
      "loss": 0.1051,
      "step": 11154
    },
    {
      "epoch": 0.5760392460624839,
      "grad_norm": 2.7382278442382812,
      "learning_rate": 3.882488872485678e-06,
      "loss": 0.0296,
      "step": 11155
    },
    {
      "epoch": 0.5760908856183837,
      "grad_norm": 2.377133369445801,
      "learning_rate": 3.881690253854541e-06,
      "loss": 0.0373,
      "step": 11156
    },
    {
      "epoch": 0.5761425251742835,
      "grad_norm": 2.985677719116211,
      "learning_rate": 3.88089166525458e-06,
      "loss": 0.0534,
      "step": 11157
    },
    {
      "epoch": 0.5761941647301834,
      "grad_norm": 3.934102773666382,
      "learning_rate": 3.8800931067072375e-06,
      "loss": 0.0411,
      "step": 11158
    },
    {
      "epoch": 0.5762458042860832,
      "grad_norm": 3.963939905166626,
      "learning_rate": 3.879294578233962e-06,
      "loss": 0.0398,
      "step": 11159
    },
    {
      "epoch": 0.576297443841983,
      "grad_norm": 5.050247669219971,
      "learning_rate": 3.878496079856193e-06,
      "loss": 0.064,
      "step": 11160
    },
    {
      "epoch": 0.5763490833978828,
      "grad_norm": 7.555764198303223,
      "learning_rate": 3.877697611595376e-06,
      "loss": 0.0752,
      "step": 11161
    },
    {
      "epoch": 0.5764007229537826,
      "grad_norm": 1.6588548421859741,
      "learning_rate": 3.876899173472951e-06,
      "loss": 0.0315,
      "step": 11162
    },
    {
      "epoch": 0.5764523625096825,
      "grad_norm": 2.1770999431610107,
      "learning_rate": 3.876100765510363e-06,
      "loss": 0.0176,
      "step": 11163
    },
    {
      "epoch": 0.5765040020655823,
      "grad_norm": 3.344083786010742,
      "learning_rate": 3.875302387729049e-06,
      "loss": 0.0448,
      "step": 11164
    },
    {
      "epoch": 0.5765556416214821,
      "grad_norm": 4.4390339851379395,
      "learning_rate": 3.874504040150449e-06,
      "loss": 0.0869,
      "step": 11165
    },
    {
      "epoch": 0.5766072811773819,
      "grad_norm": 2.6471667289733887,
      "learning_rate": 3.873705722796006e-06,
      "loss": 0.0292,
      "step": 11166
    },
    {
      "epoch": 0.5766589207332817,
      "grad_norm": 12.145041465759277,
      "learning_rate": 3.872907435687151e-06,
      "loss": 0.0692,
      "step": 11167
    },
    {
      "epoch": 0.5767105602891816,
      "grad_norm": 3.1950759887695312,
      "learning_rate": 3.872109178845327e-06,
      "loss": 0.0436,
      "step": 11168
    },
    {
      "epoch": 0.5767621998450814,
      "grad_norm": 7.655698299407959,
      "learning_rate": 3.871310952291968e-06,
      "loss": 0.1264,
      "step": 11169
    },
    {
      "epoch": 0.5768138394009812,
      "grad_norm": 4.986310958862305,
      "learning_rate": 3.870512756048511e-06,
      "loss": 0.0447,
      "step": 11170
    },
    {
      "epoch": 0.576865478956881,
      "grad_norm": 5.588435173034668,
      "learning_rate": 3.8697145901363894e-06,
      "loss": 0.0646,
      "step": 11171
    },
    {
      "epoch": 0.5769171185127808,
      "grad_norm": 3.2323803901672363,
      "learning_rate": 3.868916454577037e-06,
      "loss": 0.057,
      "step": 11172
    },
    {
      "epoch": 0.5769687580686806,
      "grad_norm": 3.260643243789673,
      "learning_rate": 3.86811834939189e-06,
      "loss": 0.0406,
      "step": 11173
    },
    {
      "epoch": 0.5770203976245805,
      "grad_norm": 4.35840368270874,
      "learning_rate": 3.8673202746023775e-06,
      "loss": 0.0475,
      "step": 11174
    },
    {
      "epoch": 0.5770720371804803,
      "grad_norm": 3.2187492847442627,
      "learning_rate": 3.866522230229932e-06,
      "loss": 0.0563,
      "step": 11175
    },
    {
      "epoch": 0.5771236767363801,
      "grad_norm": 7.741909503936768,
      "learning_rate": 3.8657242162959845e-06,
      "loss": 0.0994,
      "step": 11176
    },
    {
      "epoch": 0.5771753162922799,
      "grad_norm": 3.5148375034332275,
      "learning_rate": 3.864926232821966e-06,
      "loss": 0.0634,
      "step": 11177
    },
    {
      "epoch": 0.5772269558481797,
      "grad_norm": 7.05987548828125,
      "learning_rate": 3.8641282798293035e-06,
      "loss": 0.0712,
      "step": 11178
    },
    {
      "epoch": 0.5772785954040796,
      "grad_norm": 8.414325714111328,
      "learning_rate": 3.8633303573394246e-06,
      "loss": 0.0794,
      "step": 11179
    },
    {
      "epoch": 0.5773302349599794,
      "grad_norm": 2.662283420562744,
      "learning_rate": 3.8625324653737594e-06,
      "loss": 0.026,
      "step": 11180
    },
    {
      "epoch": 0.5773818745158792,
      "grad_norm": 4.074368953704834,
      "learning_rate": 3.8617346039537355e-06,
      "loss": 0.0485,
      "step": 11181
    },
    {
      "epoch": 0.577433514071779,
      "grad_norm": 1.9704419374465942,
      "learning_rate": 3.860936773100775e-06,
      "loss": 0.0111,
      "step": 11182
    },
    {
      "epoch": 0.5774851536276788,
      "grad_norm": 7.444137096405029,
      "learning_rate": 3.8601389728363045e-06,
      "loss": 0.0808,
      "step": 11183
    },
    {
      "epoch": 0.5775367931835786,
      "grad_norm": 4.79498815536499,
      "learning_rate": 3.859341203181749e-06,
      "loss": 0.0789,
      "step": 11184
    },
    {
      "epoch": 0.5775884327394785,
      "grad_norm": 12.922369956970215,
      "learning_rate": 3.858543464158533e-06,
      "loss": 0.0758,
      "step": 11185
    },
    {
      "epoch": 0.5776400722953783,
      "grad_norm": 3.692077875137329,
      "learning_rate": 3.857745755788074e-06,
      "loss": 0.0631,
      "step": 11186
    },
    {
      "epoch": 0.5776917118512781,
      "grad_norm": 5.677352428436279,
      "learning_rate": 3.8569480780917985e-06,
      "loss": 0.073,
      "step": 11187
    },
    {
      "epoch": 0.5777433514071779,
      "grad_norm": 2.351456880569458,
      "learning_rate": 3.856150431091127e-06,
      "loss": 0.0402,
      "step": 11188
    },
    {
      "epoch": 0.5777949909630777,
      "grad_norm": 8.183959007263184,
      "learning_rate": 3.855352814807477e-06,
      "loss": 0.0809,
      "step": 11189
    },
    {
      "epoch": 0.5778466305189776,
      "grad_norm": 4.358241081237793,
      "learning_rate": 3.85455522926227e-06,
      "loss": 0.0738,
      "step": 11190
    },
    {
      "epoch": 0.5778982700748774,
      "grad_norm": 3.704857587814331,
      "learning_rate": 3.853757674476923e-06,
      "loss": 0.0467,
      "step": 11191
    },
    {
      "epoch": 0.5779499096307772,
      "grad_norm": 15.281079292297363,
      "learning_rate": 3.852960150472856e-06,
      "loss": 0.1134,
      "step": 11192
    },
    {
      "epoch": 0.578001549186677,
      "grad_norm": 3.198152780532837,
      "learning_rate": 3.852162657271481e-06,
      "loss": 0.0529,
      "step": 11193
    },
    {
      "epoch": 0.5780531887425768,
      "grad_norm": 2.4487898349761963,
      "learning_rate": 3.8513651948942184e-06,
      "loss": 0.0423,
      "step": 11194
    },
    {
      "epoch": 0.5781048282984766,
      "grad_norm": 7.034802436828613,
      "learning_rate": 3.850567763362483e-06,
      "loss": 0.0671,
      "step": 11195
    },
    {
      "epoch": 0.5781564678543765,
      "grad_norm": 1.8790817260742188,
      "learning_rate": 3.849770362697688e-06,
      "loss": 0.0258,
      "step": 11196
    },
    {
      "epoch": 0.5782081074102763,
      "grad_norm": 7.357606410980225,
      "learning_rate": 3.848972992921245e-06,
      "loss": 0.0963,
      "step": 11197
    },
    {
      "epoch": 0.5782597469661761,
      "grad_norm": 3.565680742263794,
      "learning_rate": 3.848175654054569e-06,
      "loss": 0.0547,
      "step": 11198
    },
    {
      "epoch": 0.5783113865220759,
      "grad_norm": 9.006486892700195,
      "learning_rate": 3.847378346119073e-06,
      "loss": 0.0377,
      "step": 11199
    },
    {
      "epoch": 0.5783630260779757,
      "grad_norm": 3.730607748031616,
      "learning_rate": 3.846581069136164e-06,
      "loss": 0.0412,
      "step": 11200
    },
    {
      "epoch": 0.5784146656338756,
      "grad_norm": 6.657228946685791,
      "learning_rate": 3.845783823127256e-06,
      "loss": 0.0797,
      "step": 11201
    },
    {
      "epoch": 0.5784663051897754,
      "grad_norm": 10.429649353027344,
      "learning_rate": 3.844986608113756e-06,
      "loss": 0.1289,
      "step": 11202
    },
    {
      "epoch": 0.5785179447456752,
      "grad_norm": 2.386704206466675,
      "learning_rate": 3.844189424117074e-06,
      "loss": 0.0336,
      "step": 11203
    },
    {
      "epoch": 0.578569584301575,
      "grad_norm": 4.332094669342041,
      "learning_rate": 3.843392271158616e-06,
      "loss": 0.0356,
      "step": 11204
    },
    {
      "epoch": 0.5786212238574748,
      "grad_norm": 3.0240366458892822,
      "learning_rate": 3.842595149259788e-06,
      "loss": 0.0435,
      "step": 11205
    },
    {
      "epoch": 0.5786728634133746,
      "grad_norm": 3.4464898109436035,
      "learning_rate": 3.841798058441999e-06,
      "loss": 0.0405,
      "step": 11206
    },
    {
      "epoch": 0.5787245029692745,
      "grad_norm": 1.5501046180725098,
      "learning_rate": 3.841000998726654e-06,
      "loss": 0.024,
      "step": 11207
    },
    {
      "epoch": 0.5787761425251743,
      "grad_norm": 2.9390969276428223,
      "learning_rate": 3.840203970135154e-06,
      "loss": 0.05,
      "step": 11208
    },
    {
      "epoch": 0.5788277820810741,
      "grad_norm": 3.9138071537017822,
      "learning_rate": 3.839406972688906e-06,
      "loss": 0.0802,
      "step": 11209
    },
    {
      "epoch": 0.5788794216369739,
      "grad_norm": 8.86423110961914,
      "learning_rate": 3.838610006409311e-06,
      "loss": 0.0892,
      "step": 11210
    },
    {
      "epoch": 0.5789310611928737,
      "grad_norm": 2.7400600910186768,
      "learning_rate": 3.83781307131777e-06,
      "loss": 0.0498,
      "step": 11211
    },
    {
      "epoch": 0.5789827007487736,
      "grad_norm": 4.155879974365234,
      "learning_rate": 3.837016167435684e-06,
      "loss": 0.0404,
      "step": 11212
    },
    {
      "epoch": 0.5790343403046734,
      "grad_norm": 3.231323480606079,
      "learning_rate": 3.836219294784455e-06,
      "loss": 0.0356,
      "step": 11213
    },
    {
      "epoch": 0.5790859798605732,
      "grad_norm": 18.113510131835938,
      "learning_rate": 3.835422453385482e-06,
      "loss": 0.0689,
      "step": 11214
    },
    {
      "epoch": 0.579137619416473,
      "grad_norm": 9.800552368164062,
      "learning_rate": 3.8346256432601616e-06,
      "loss": 0.1272,
      "step": 11215
    },
    {
      "epoch": 0.5791892589723728,
      "grad_norm": 5.941273212432861,
      "learning_rate": 3.833828864429893e-06,
      "loss": 0.0485,
      "step": 11216
    },
    {
      "epoch": 0.5792408985282727,
      "grad_norm": 2.246011734008789,
      "learning_rate": 3.833032116916071e-06,
      "loss": 0.0362,
      "step": 11217
    },
    {
      "epoch": 0.5792925380841725,
      "grad_norm": 25.599557876586914,
      "learning_rate": 3.832235400740095e-06,
      "loss": 0.1152,
      "step": 11218
    },
    {
      "epoch": 0.5793441776400723,
      "grad_norm": 6.2794694900512695,
      "learning_rate": 3.831438715923358e-06,
      "loss": 0.0394,
      "step": 11219
    },
    {
      "epoch": 0.5793958171959721,
      "grad_norm": 4.126479148864746,
      "learning_rate": 3.830642062487253e-06,
      "loss": 0.0505,
      "step": 11220
    },
    {
      "epoch": 0.5794474567518719,
      "grad_norm": 2.747063636779785,
      "learning_rate": 3.829845440453177e-06,
      "loss": 0.0507,
      "step": 11221
    },
    {
      "epoch": 0.5794990963077717,
      "grad_norm": 3.3718338012695312,
      "learning_rate": 3.829048849842518e-06,
      "loss": 0.0331,
      "step": 11222
    },
    {
      "epoch": 0.5795507358636716,
      "grad_norm": 4.742069721221924,
      "learning_rate": 3.82825229067667e-06,
      "loss": 0.0386,
      "step": 11223
    },
    {
      "epoch": 0.5796023754195714,
      "grad_norm": 4.596510410308838,
      "learning_rate": 3.8274557629770225e-06,
      "loss": 0.0721,
      "step": 11224
    },
    {
      "epoch": 0.5796540149754712,
      "grad_norm": 12.590519905090332,
      "learning_rate": 3.82665926676497e-06,
      "loss": 0.097,
      "step": 11225
    },
    {
      "epoch": 0.579705654531371,
      "grad_norm": 3.2965071201324463,
      "learning_rate": 3.8258628020618965e-06,
      "loss": 0.0442,
      "step": 11226
    },
    {
      "epoch": 0.5797572940872708,
      "grad_norm": 4.946275234222412,
      "learning_rate": 3.825066368889192e-06,
      "loss": 0.0727,
      "step": 11227
    },
    {
      "epoch": 0.5798089336431707,
      "grad_norm": 6.469065189361572,
      "learning_rate": 3.824269967268245e-06,
      "loss": 0.028,
      "step": 11228
    },
    {
      "epoch": 0.5798605731990705,
      "grad_norm": 9.995840072631836,
      "learning_rate": 3.823473597220443e-06,
      "loss": 0.0565,
      "step": 11229
    },
    {
      "epoch": 0.5799122127549703,
      "grad_norm": 4.792623996734619,
      "learning_rate": 3.822677258767168e-06,
      "loss": 0.0765,
      "step": 11230
    },
    {
      "epoch": 0.5799638523108701,
      "grad_norm": 4.056150436401367,
      "learning_rate": 3.8218809519298065e-06,
      "loss": 0.056,
      "step": 11231
    },
    {
      "epoch": 0.5800154918667699,
      "grad_norm": 8.94653034210205,
      "learning_rate": 3.821084676729746e-06,
      "loss": 0.1231,
      "step": 11232
    },
    {
      "epoch": 0.5800671314226697,
      "grad_norm": 2.779862642288208,
      "learning_rate": 3.820288433188364e-06,
      "loss": 0.0354,
      "step": 11233
    },
    {
      "epoch": 0.5801187709785696,
      "grad_norm": 4.995369911193848,
      "learning_rate": 3.819492221327047e-06,
      "loss": 0.0382,
      "step": 11234
    },
    {
      "epoch": 0.5801704105344694,
      "grad_norm": 4.1797871589660645,
      "learning_rate": 3.818696041167176e-06,
      "loss": 0.0556,
      "step": 11235
    },
    {
      "epoch": 0.5802220500903692,
      "grad_norm": 26.355030059814453,
      "learning_rate": 3.817899892730131e-06,
      "loss": 0.0752,
      "step": 11236
    },
    {
      "epoch": 0.580273689646269,
      "grad_norm": 2.5491297245025635,
      "learning_rate": 3.817103776037291e-06,
      "loss": 0.0376,
      "step": 11237
    },
    {
      "epoch": 0.5803253292021688,
      "grad_norm": 1.8025641441345215,
      "learning_rate": 3.816307691110035e-06,
      "loss": 0.0368,
      "step": 11238
    },
    {
      "epoch": 0.5803769687580687,
      "grad_norm": 10.306572914123535,
      "learning_rate": 3.8155116379697425e-06,
      "loss": 0.0574,
      "step": 11239
    },
    {
      "epoch": 0.5804286083139685,
      "grad_norm": 2.501926898956299,
      "learning_rate": 3.814715616637792e-06,
      "loss": 0.0366,
      "step": 11240
    },
    {
      "epoch": 0.5804802478698683,
      "grad_norm": 5.4202351570129395,
      "learning_rate": 3.813919627135557e-06,
      "loss": 0.0205,
      "step": 11241
    },
    {
      "epoch": 0.5805318874257681,
      "grad_norm": 3.9522130489349365,
      "learning_rate": 3.8131236694844137e-06,
      "loss": 0.0426,
      "step": 11242
    },
    {
      "epoch": 0.5805835269816679,
      "grad_norm": 2.759641170501709,
      "learning_rate": 3.812327743705739e-06,
      "loss": 0.0605,
      "step": 11243
    },
    {
      "epoch": 0.5806351665375677,
      "grad_norm": 3.8832898139953613,
      "learning_rate": 3.8115318498209025e-06,
      "loss": 0.0696,
      "step": 11244
    },
    {
      "epoch": 0.5806868060934676,
      "grad_norm": 5.93319845199585,
      "learning_rate": 3.810735987851281e-06,
      "loss": 0.0268,
      "step": 11245
    },
    {
      "epoch": 0.5807384456493674,
      "grad_norm": 9.905107498168945,
      "learning_rate": 3.809940157818245e-06,
      "loss": 0.0557,
      "step": 11246
    },
    {
      "epoch": 0.5807900852052672,
      "grad_norm": 7.695933818817139,
      "learning_rate": 3.809144359743168e-06,
      "loss": 0.0998,
      "step": 11247
    },
    {
      "epoch": 0.580841724761167,
      "grad_norm": 1.7435274124145508,
      "learning_rate": 3.808348593647417e-06,
      "loss": 0.0262,
      "step": 11248
    },
    {
      "epoch": 0.5808933643170668,
      "grad_norm": 5.3026533126831055,
      "learning_rate": 3.8075528595523633e-06,
      "loss": 0.0616,
      "step": 11249
    },
    {
      "epoch": 0.5809450038729667,
      "grad_norm": 3.6934499740600586,
      "learning_rate": 3.8067571574793737e-06,
      "loss": 0.0399,
      "step": 11250
    },
    {
      "epoch": 0.5809966434288665,
      "grad_norm": 2.640491485595703,
      "learning_rate": 3.805961487449821e-06,
      "loss": 0.0295,
      "step": 11251
    },
    {
      "epoch": 0.5810482829847663,
      "grad_norm": 10.379626274108887,
      "learning_rate": 3.8051658494850675e-06,
      "loss": 0.1057,
      "step": 11252
    },
    {
      "epoch": 0.5810999225406661,
      "grad_norm": 3.224012851715088,
      "learning_rate": 3.804370243606481e-06,
      "loss": 0.0392,
      "step": 11253
    },
    {
      "epoch": 0.5811515620965659,
      "grad_norm": 10.336311340332031,
      "learning_rate": 3.8035746698354275e-06,
      "loss": 0.1099,
      "step": 11254
    },
    {
      "epoch": 0.5812032016524658,
      "grad_norm": 4.512557029724121,
      "learning_rate": 3.802779128193269e-06,
      "loss": 0.0889,
      "step": 11255
    },
    {
      "epoch": 0.5812548412083656,
      "grad_norm": 6.317713737487793,
      "learning_rate": 3.8019836187013715e-06,
      "loss": 0.1582,
      "step": 11256
    },
    {
      "epoch": 0.5813064807642654,
      "grad_norm": 7.257702350616455,
      "learning_rate": 3.8011881413810947e-06,
      "loss": 0.0559,
      "step": 11257
    },
    {
      "epoch": 0.5813581203201652,
      "grad_norm": 19.921524047851562,
      "learning_rate": 3.800392696253805e-06,
      "loss": 0.0955,
      "step": 11258
    },
    {
      "epoch": 0.581409759876065,
      "grad_norm": 2.5032012462615967,
      "learning_rate": 3.7995972833408596e-06,
      "loss": 0.0502,
      "step": 11259
    },
    {
      "epoch": 0.5814613994319648,
      "grad_norm": 2.9153013229370117,
      "learning_rate": 3.7988019026636197e-06,
      "loss": 0.0287,
      "step": 11260
    },
    {
      "epoch": 0.5815130389878647,
      "grad_norm": 6.229789733886719,
      "learning_rate": 3.798006554243444e-06,
      "loss": 0.0719,
      "step": 11261
    },
    {
      "epoch": 0.5815646785437645,
      "grad_norm": 4.571768760681152,
      "learning_rate": 3.7972112381016937e-06,
      "loss": 0.0272,
      "step": 11262
    },
    {
      "epoch": 0.5816163180996643,
      "grad_norm": 13.2096586227417,
      "learning_rate": 3.7964159542597205e-06,
      "loss": 0.0884,
      "step": 11263
    },
    {
      "epoch": 0.5816679576555641,
      "grad_norm": 4.897929668426514,
      "learning_rate": 3.795620702738886e-06,
      "loss": 0.0762,
      "step": 11264
    },
    {
      "epoch": 0.5817195972114639,
      "grad_norm": 4.570412635803223,
      "learning_rate": 3.794825483560546e-06,
      "loss": 0.0666,
      "step": 11265
    },
    {
      "epoch": 0.5817712367673638,
      "grad_norm": 5.817856788635254,
      "learning_rate": 3.794030296746052e-06,
      "loss": 0.047,
      "step": 11266
    },
    {
      "epoch": 0.5818228763232636,
      "grad_norm": 39.16423034667969,
      "learning_rate": 3.79323514231676e-06,
      "loss": 0.1014,
      "step": 11267
    },
    {
      "epoch": 0.5818745158791634,
      "grad_norm": 9.259040832519531,
      "learning_rate": 3.7924400202940227e-06,
      "loss": 0.0921,
      "step": 11268
    },
    {
      "epoch": 0.5819261554350632,
      "grad_norm": 1.8464385271072388,
      "learning_rate": 3.7916449306991943e-06,
      "loss": 0.0283,
      "step": 11269
    },
    {
      "epoch": 0.581977794990963,
      "grad_norm": 3.011427164077759,
      "learning_rate": 3.790849873553622e-06,
      "loss": 0.0541,
      "step": 11270
    },
    {
      "epoch": 0.5820294345468628,
      "grad_norm": 3.191032648086548,
      "learning_rate": 3.79005484887866e-06,
      "loss": 0.0388,
      "step": 11271
    },
    {
      "epoch": 0.5820810741027627,
      "grad_norm": 4.808909893035889,
      "learning_rate": 3.7892598566956562e-06,
      "loss": 0.0692,
      "step": 11272
    },
    {
      "epoch": 0.5821327136586625,
      "grad_norm": 1.6119788885116577,
      "learning_rate": 3.788464897025962e-06,
      "loss": 0.0335,
      "step": 11273
    },
    {
      "epoch": 0.5821843532145623,
      "grad_norm": 3.081813335418701,
      "learning_rate": 3.7876699698909215e-06,
      "loss": 0.0504,
      "step": 11274
    },
    {
      "epoch": 0.5822359927704621,
      "grad_norm": 6.421976566314697,
      "learning_rate": 3.786875075311883e-06,
      "loss": 0.0755,
      "step": 11275
    },
    {
      "epoch": 0.582287632326362,
      "grad_norm": 2.8409981727600098,
      "learning_rate": 3.7860802133101955e-06,
      "loss": 0.1019,
      "step": 11276
    },
    {
      "epoch": 0.5823392718822619,
      "grad_norm": 5.155486106872559,
      "learning_rate": 3.785285383907199e-06,
      "loss": 0.0614,
      "step": 11277
    },
    {
      "epoch": 0.5823909114381617,
      "grad_norm": 3.01609468460083,
      "learning_rate": 3.784490587124242e-06,
      "loss": 0.0505,
      "step": 11278
    },
    {
      "epoch": 0.5824425509940615,
      "grad_norm": 2.3626742362976074,
      "learning_rate": 3.783695822982667e-06,
      "loss": 0.0413,
      "step": 11279
    },
    {
      "epoch": 0.5824941905499613,
      "grad_norm": 1.7197331190109253,
      "learning_rate": 3.782901091503818e-06,
      "loss": 0.0328,
      "step": 11280
    },
    {
      "epoch": 0.5825458301058611,
      "grad_norm": 2.4429614543914795,
      "learning_rate": 3.7821063927090336e-06,
      "loss": 0.0463,
      "step": 11281
    },
    {
      "epoch": 0.582597469661761,
      "grad_norm": 2.6684646606445312,
      "learning_rate": 3.7813117266196564e-06,
      "loss": 0.0628,
      "step": 11282
    },
    {
      "epoch": 0.5826491092176608,
      "grad_norm": 9.272198677062988,
      "learning_rate": 3.7805170932570255e-06,
      "loss": 0.0851,
      "step": 11283
    },
    {
      "epoch": 0.5827007487735606,
      "grad_norm": 3.1981468200683594,
      "learning_rate": 3.7797224926424837e-06,
      "loss": 0.0691,
      "step": 11284
    },
    {
      "epoch": 0.5827523883294604,
      "grad_norm": 2.698625326156616,
      "learning_rate": 3.7789279247973646e-06,
      "loss": 0.0272,
      "step": 11285
    },
    {
      "epoch": 0.5828040278853602,
      "grad_norm": 7.857949256896973,
      "learning_rate": 3.778133389743008e-06,
      "loss": 0.0678,
      "step": 11286
    },
    {
      "epoch": 0.58285566744126,
      "grad_norm": 6.358802318572998,
      "learning_rate": 3.777338887500751e-06,
      "loss": 0.0775,
      "step": 11287
    },
    {
      "epoch": 0.5829073069971599,
      "grad_norm": 1.7476799488067627,
      "learning_rate": 3.776544418091927e-06,
      "loss": 0.0448,
      "step": 11288
    },
    {
      "epoch": 0.5829589465530597,
      "grad_norm": 5.920764446258545,
      "learning_rate": 3.7757499815378706e-06,
      "loss": 0.055,
      "step": 11289
    },
    {
      "epoch": 0.5830105861089595,
      "grad_norm": 2.1064400672912598,
      "learning_rate": 3.774955577859919e-06,
      "loss": 0.0477,
      "step": 11290
    },
    {
      "epoch": 0.5830622256648593,
      "grad_norm": 9.898158073425293,
      "learning_rate": 3.774161207079403e-06,
      "loss": 0.1185,
      "step": 11291
    },
    {
      "epoch": 0.5831138652207591,
      "grad_norm": 3.104647636413574,
      "learning_rate": 3.7733668692176546e-06,
      "loss": 0.0405,
      "step": 11292
    },
    {
      "epoch": 0.583165504776659,
      "grad_norm": 4.9685444831848145,
      "learning_rate": 3.7725725642960047e-06,
      "loss": 0.0629,
      "step": 11293
    },
    {
      "epoch": 0.5832171443325588,
      "grad_norm": 6.976975917816162,
      "learning_rate": 3.7717782923357843e-06,
      "loss": 0.1079,
      "step": 11294
    },
    {
      "epoch": 0.5832687838884586,
      "grad_norm": 5.838769912719727,
      "learning_rate": 3.770984053358324e-06,
      "loss": 0.0745,
      "step": 11295
    },
    {
      "epoch": 0.5833204234443584,
      "grad_norm": 2.183239459991455,
      "learning_rate": 3.7701898473849486e-06,
      "loss": 0.0221,
      "step": 11296
    },
    {
      "epoch": 0.5833720630002582,
      "grad_norm": 2.552243232727051,
      "learning_rate": 3.76939567443699e-06,
      "loss": 0.0559,
      "step": 11297
    },
    {
      "epoch": 0.583423702556158,
      "grad_norm": 3.992802619934082,
      "learning_rate": 3.7686015345357734e-06,
      "loss": 0.0445,
      "step": 11298
    },
    {
      "epoch": 0.5834753421120579,
      "grad_norm": 5.167900562286377,
      "learning_rate": 3.7678074277026237e-06,
      "loss": 0.0248,
      "step": 11299
    },
    {
      "epoch": 0.5835269816679577,
      "grad_norm": 8.56865119934082,
      "learning_rate": 3.7670133539588673e-06,
      "loss": 0.095,
      "step": 11300
    },
    {
      "epoch": 0.5835786212238575,
      "grad_norm": 6.090936183929443,
      "learning_rate": 3.766219313325827e-06,
      "loss": 0.0269,
      "step": 11301
    },
    {
      "epoch": 0.5836302607797573,
      "grad_norm": 2.179647922515869,
      "learning_rate": 3.765425305824828e-06,
      "loss": 0.0369,
      "step": 11302
    },
    {
      "epoch": 0.5836819003356571,
      "grad_norm": 2.844165802001953,
      "learning_rate": 3.764631331477189e-06,
      "loss": 0.0513,
      "step": 11303
    },
    {
      "epoch": 0.583733539891557,
      "grad_norm": 5.097165107727051,
      "learning_rate": 3.7638373903042354e-06,
      "loss": 0.075,
      "step": 11304
    },
    {
      "epoch": 0.5837851794474568,
      "grad_norm": 6.951509475708008,
      "learning_rate": 3.7630434823272853e-06,
      "loss": 0.1117,
      "step": 11305
    },
    {
      "epoch": 0.5838368190033566,
      "grad_norm": 6.079875946044922,
      "learning_rate": 3.762249607567662e-06,
      "loss": 0.0791,
      "step": 11306
    },
    {
      "epoch": 0.5838884585592564,
      "grad_norm": 5.67171573638916,
      "learning_rate": 3.761455766046679e-06,
      "loss": 0.0656,
      "step": 11307
    },
    {
      "epoch": 0.5839400981151562,
      "grad_norm": 2.9252562522888184,
      "learning_rate": 3.7606619577856563e-06,
      "loss": 0.0465,
      "step": 11308
    },
    {
      "epoch": 0.583991737671056,
      "grad_norm": 9.594783782958984,
      "learning_rate": 3.7598681828059143e-06,
      "loss": 0.0943,
      "step": 11309
    },
    {
      "epoch": 0.5840433772269559,
      "grad_norm": 3.8439674377441406,
      "learning_rate": 3.7590744411287626e-06,
      "loss": 0.0616,
      "step": 11310
    },
    {
      "epoch": 0.5840950167828557,
      "grad_norm": 2.6963164806365967,
      "learning_rate": 3.758280732775521e-06,
      "loss": 0.0348,
      "step": 11311
    },
    {
      "epoch": 0.5841466563387555,
      "grad_norm": 5.471311569213867,
      "learning_rate": 3.7574870577675033e-06,
      "loss": 0.1333,
      "step": 11312
    },
    {
      "epoch": 0.5841982958946553,
      "grad_norm": 4.199850082397461,
      "learning_rate": 3.7566934161260226e-06,
      "loss": 0.0496,
      "step": 11313
    },
    {
      "epoch": 0.5842499354505551,
      "grad_norm": 4.008106231689453,
      "learning_rate": 3.7558998078723907e-06,
      "loss": 0.0395,
      "step": 11314
    },
    {
      "epoch": 0.584301575006455,
      "grad_norm": 2.1618425846099854,
      "learning_rate": 3.755106233027918e-06,
      "loss": 0.0353,
      "step": 11315
    },
    {
      "epoch": 0.5843532145623548,
      "grad_norm": 3.8902392387390137,
      "learning_rate": 3.7543126916139182e-06,
      "loss": 0.0691,
      "step": 11316
    },
    {
      "epoch": 0.5844048541182546,
      "grad_norm": 11.27973461151123,
      "learning_rate": 3.753519183651701e-06,
      "loss": 0.2437,
      "step": 11317
    },
    {
      "epoch": 0.5844564936741544,
      "grad_norm": 2.6802937984466553,
      "learning_rate": 3.752725709162573e-06,
      "loss": 0.0336,
      "step": 11318
    },
    {
      "epoch": 0.5845081332300542,
      "grad_norm": 31.847888946533203,
      "learning_rate": 3.751932268167843e-06,
      "loss": 0.178,
      "step": 11319
    },
    {
      "epoch": 0.5845597727859541,
      "grad_norm": 8.415777206420898,
      "learning_rate": 3.75113886068882e-06,
      "loss": 0.0795,
      "step": 11320
    },
    {
      "epoch": 0.5846114123418539,
      "grad_norm": 5.040017604827881,
      "learning_rate": 3.750345486746807e-06,
      "loss": 0.0753,
      "step": 11321
    },
    {
      "epoch": 0.5846630518977537,
      "grad_norm": 2.5647470951080322,
      "learning_rate": 3.7495521463631102e-06,
      "loss": 0.0491,
      "step": 11322
    },
    {
      "epoch": 0.5847146914536535,
      "grad_norm": 2.467352867126465,
      "learning_rate": 3.748758839559036e-06,
      "loss": 0.0724,
      "step": 11323
    },
    {
      "epoch": 0.5847663310095533,
      "grad_norm": 2.2588796615600586,
      "learning_rate": 3.747965566355888e-06,
      "loss": 0.0371,
      "step": 11324
    },
    {
      "epoch": 0.5848179705654531,
      "grad_norm": 3.030583620071411,
      "learning_rate": 3.747172326774966e-06,
      "loss": 0.0361,
      "step": 11325
    },
    {
      "epoch": 0.584869610121353,
      "grad_norm": 1.9888622760772705,
      "learning_rate": 3.746379120837573e-06,
      "loss": 0.0374,
      "step": 11326
    },
    {
      "epoch": 0.5849212496772528,
      "grad_norm": 2.352658987045288,
      "learning_rate": 3.7455859485650104e-06,
      "loss": 0.0428,
      "step": 11327
    },
    {
      "epoch": 0.5849728892331526,
      "grad_norm": 6.409963130950928,
      "learning_rate": 3.7447928099785803e-06,
      "loss": 0.0984,
      "step": 11328
    },
    {
      "epoch": 0.5850245287890524,
      "grad_norm": 2.409914970397949,
      "learning_rate": 3.7439997050995752e-06,
      "loss": 0.0373,
      "step": 11329
    },
    {
      "epoch": 0.5850761683449522,
      "grad_norm": 2.0099265575408936,
      "learning_rate": 3.743206633949299e-06,
      "loss": 0.0208,
      "step": 11330
    },
    {
      "epoch": 0.5851278079008521,
      "grad_norm": 3.314845085144043,
      "learning_rate": 3.742413596549048e-06,
      "loss": 0.0518,
      "step": 11331
    },
    {
      "epoch": 0.5851794474567519,
      "grad_norm": 3.7936534881591797,
      "learning_rate": 3.7416205929201167e-06,
      "loss": 0.0334,
      "step": 11332
    },
    {
      "epoch": 0.5852310870126517,
      "grad_norm": 2.9197044372558594,
      "learning_rate": 3.740827623083801e-06,
      "loss": 0.0404,
      "step": 11333
    },
    {
      "epoch": 0.5852827265685515,
      "grad_norm": 2.9874088764190674,
      "learning_rate": 3.7400346870613946e-06,
      "loss": 0.0328,
      "step": 11334
    },
    {
      "epoch": 0.5853343661244513,
      "grad_norm": 2.42094349861145,
      "learning_rate": 3.7392417848741957e-06,
      "loss": 0.0389,
      "step": 11335
    },
    {
      "epoch": 0.5853860056803512,
      "grad_norm": 3.0771613121032715,
      "learning_rate": 3.7384489165434903e-06,
      "loss": 0.062,
      "step": 11336
    },
    {
      "epoch": 0.585437645236251,
      "grad_norm": 3.547558069229126,
      "learning_rate": 3.737656082090574e-06,
      "loss": 0.0548,
      "step": 11337
    },
    {
      "epoch": 0.5854892847921508,
      "grad_norm": 2.4451236724853516,
      "learning_rate": 3.7368632815367367e-06,
      "loss": 0.0322,
      "step": 11338
    },
    {
      "epoch": 0.5855409243480506,
      "grad_norm": 2.4411122798919678,
      "learning_rate": 3.7360705149032705e-06,
      "loss": 0.0424,
      "step": 11339
    },
    {
      "epoch": 0.5855925639039504,
      "grad_norm": 2.403329610824585,
      "learning_rate": 3.7352777822114606e-06,
      "loss": 0.0117,
      "step": 11340
    },
    {
      "epoch": 0.5856442034598502,
      "grad_norm": 9.698739051818848,
      "learning_rate": 3.7344850834825964e-06,
      "loss": 0.179,
      "step": 11341
    },
    {
      "epoch": 0.5856958430157501,
      "grad_norm": 3.392132520675659,
      "learning_rate": 3.7336924187379687e-06,
      "loss": 0.0487,
      "step": 11342
    },
    {
      "epoch": 0.5857474825716499,
      "grad_norm": 3.7034332752227783,
      "learning_rate": 3.7328997879988573e-06,
      "loss": 0.0683,
      "step": 11343
    },
    {
      "epoch": 0.5857991221275497,
      "grad_norm": 3.4193503856658936,
      "learning_rate": 3.7321071912865527e-06,
      "loss": 0.0793,
      "step": 11344
    },
    {
      "epoch": 0.5858507616834495,
      "grad_norm": 4.8868184089660645,
      "learning_rate": 3.7313146286223366e-06,
      "loss": 0.0392,
      "step": 11345
    },
    {
      "epoch": 0.5859024012393493,
      "grad_norm": 2.7148451805114746,
      "learning_rate": 3.730522100027496e-06,
      "loss": 0.0382,
      "step": 11346
    },
    {
      "epoch": 0.5859540407952492,
      "grad_norm": 1.9924501180648804,
      "learning_rate": 3.729729605523309e-06,
      "loss": 0.035,
      "step": 11347
    },
    {
      "epoch": 0.586005680351149,
      "grad_norm": 1.6569879055023193,
      "learning_rate": 3.7289371451310584e-06,
      "loss": 0.0129,
      "step": 11348
    },
    {
      "epoch": 0.5860573199070488,
      "grad_norm": 8.16529369354248,
      "learning_rate": 3.7281447188720278e-06,
      "loss": 0.0652,
      "step": 11349
    },
    {
      "epoch": 0.5861089594629486,
      "grad_norm": 2.9217026233673096,
      "learning_rate": 3.727352326767496e-06,
      "loss": 0.0126,
      "step": 11350
    },
    {
      "epoch": 0.5861605990188484,
      "grad_norm": 3.508958339691162,
      "learning_rate": 3.7265599688387397e-06,
      "loss": 0.0477,
      "step": 11351
    },
    {
      "epoch": 0.5862122385747482,
      "grad_norm": 2.706690788269043,
      "learning_rate": 3.72576764510704e-06,
      "loss": 0.0369,
      "step": 11352
    },
    {
      "epoch": 0.5862638781306481,
      "grad_norm": 2.710596799850464,
      "learning_rate": 3.724975355593672e-06,
      "loss": 0.0388,
      "step": 11353
    },
    {
      "epoch": 0.5863155176865479,
      "grad_norm": 8.527763366699219,
      "learning_rate": 3.724183100319912e-06,
      "loss": 0.0933,
      "step": 11354
    },
    {
      "epoch": 0.5863671572424477,
      "grad_norm": 12.65064525604248,
      "learning_rate": 3.7233908793070345e-06,
      "loss": 0.1196,
      "step": 11355
    },
    {
      "epoch": 0.5864187967983475,
      "grad_norm": 5.268025875091553,
      "learning_rate": 3.7225986925763158e-06,
      "loss": 0.0309,
      "step": 11356
    },
    {
      "epoch": 0.5864704363542473,
      "grad_norm": 5.910369873046875,
      "learning_rate": 3.72180654014903e-06,
      "loss": 0.0734,
      "step": 11357
    },
    {
      "epoch": 0.5865220759101472,
      "grad_norm": 3.7204549312591553,
      "learning_rate": 3.721014422046447e-06,
      "loss": 0.0365,
      "step": 11358
    },
    {
      "epoch": 0.586573715466047,
      "grad_norm": 2.2930989265441895,
      "learning_rate": 3.7202223382898406e-06,
      "loss": 0.0457,
      "step": 11359
    },
    {
      "epoch": 0.5866253550219468,
      "grad_norm": 2.766362190246582,
      "learning_rate": 3.719430288900481e-06,
      "loss": 0.0388,
      "step": 11360
    },
    {
      "epoch": 0.5866769945778466,
      "grad_norm": 15.595471382141113,
      "learning_rate": 3.718638273899636e-06,
      "loss": 0.0341,
      "step": 11361
    },
    {
      "epoch": 0.5867286341337464,
      "grad_norm": 3.2151241302490234,
      "learning_rate": 3.7178462933085744e-06,
      "loss": 0.0284,
      "step": 11362
    },
    {
      "epoch": 0.5867802736896462,
      "grad_norm": 2.417850971221924,
      "learning_rate": 3.7170543471485675e-06,
      "loss": 0.0472,
      "step": 11363
    },
    {
      "epoch": 0.5868319132455461,
      "grad_norm": 14.113678932189941,
      "learning_rate": 3.716262435440881e-06,
      "loss": 0.0899,
      "step": 11364
    },
    {
      "epoch": 0.5868835528014459,
      "grad_norm": 3.30176043510437,
      "learning_rate": 3.7154705582067798e-06,
      "loss": 0.0408,
      "step": 11365
    },
    {
      "epoch": 0.5869351923573457,
      "grad_norm": 4.075703144073486,
      "learning_rate": 3.714678715467529e-06,
      "loss": 0.0731,
      "step": 11366
    },
    {
      "epoch": 0.5869868319132455,
      "grad_norm": 7.817165374755859,
      "learning_rate": 3.713886907244393e-06,
      "loss": 0.0788,
      "step": 11367
    },
    {
      "epoch": 0.5870384714691453,
      "grad_norm": 4.104890823364258,
      "learning_rate": 3.7130951335586374e-06,
      "loss": 0.0507,
      "step": 11368
    },
    {
      "epoch": 0.5870901110250452,
      "grad_norm": 1.754713773727417,
      "learning_rate": 3.71230339443152e-06,
      "loss": 0.0249,
      "step": 11369
    },
    {
      "epoch": 0.587141750580945,
      "grad_norm": 1.3855910301208496,
      "learning_rate": 3.711511689884306e-06,
      "loss": 0.0152,
      "step": 11370
    },
    {
      "epoch": 0.5871933901368448,
      "grad_norm": 2.7153759002685547,
      "learning_rate": 3.7107200199382565e-06,
      "loss": 0.0692,
      "step": 11371
    },
    {
      "epoch": 0.5872450296927446,
      "grad_norm": 2.1386892795562744,
      "learning_rate": 3.7099283846146273e-06,
      "loss": 0.0446,
      "step": 11372
    },
    {
      "epoch": 0.5872966692486444,
      "grad_norm": 5.187109470367432,
      "learning_rate": 3.709136783934679e-06,
      "loss": 0.14,
      "step": 11373
    },
    {
      "epoch": 0.5873483088045443,
      "grad_norm": 6.390959739685059,
      "learning_rate": 3.7083452179196687e-06,
      "loss": 0.0207,
      "step": 11374
    },
    {
      "epoch": 0.5873999483604441,
      "grad_norm": 11.2365140914917,
      "learning_rate": 3.707553686590857e-06,
      "loss": 0.0917,
      "step": 11375
    },
    {
      "epoch": 0.5874515879163439,
      "grad_norm": 1.94650399684906,
      "learning_rate": 3.706762189969493e-06,
      "loss": 0.0332,
      "step": 11376
    },
    {
      "epoch": 0.5875032274722437,
      "grad_norm": 2.6381704807281494,
      "learning_rate": 3.7059707280768363e-06,
      "loss": 0.0511,
      "step": 11377
    },
    {
      "epoch": 0.5875548670281435,
      "grad_norm": 13.001819610595703,
      "learning_rate": 3.7051793009341397e-06,
      "loss": 0.0916,
      "step": 11378
    },
    {
      "epoch": 0.5876065065840433,
      "grad_norm": 11.982989311218262,
      "learning_rate": 3.7043879085626575e-06,
      "loss": 0.2268,
      "step": 11379
    },
    {
      "epoch": 0.5876581461399432,
      "grad_norm": 7.175168514251709,
      "learning_rate": 3.703596550983639e-06,
      "loss": 0.0621,
      "step": 11380
    },
    {
      "epoch": 0.587709785695843,
      "grad_norm": 4.346513271331787,
      "learning_rate": 3.702805228218337e-06,
      "loss": 0.0496,
      "step": 11381
    },
    {
      "epoch": 0.5877614252517428,
      "grad_norm": 3.48637318611145,
      "learning_rate": 3.702013940288004e-06,
      "loss": 0.0254,
      "step": 11382
    },
    {
      "epoch": 0.5878130648076426,
      "grad_norm": 3.873312473297119,
      "learning_rate": 3.701222687213884e-06,
      "loss": 0.0662,
      "step": 11383
    },
    {
      "epoch": 0.5878647043635424,
      "grad_norm": 3.4008264541625977,
      "learning_rate": 3.7004314690172294e-06,
      "loss": 0.0546,
      "step": 11384
    },
    {
      "epoch": 0.5879163439194423,
      "grad_norm": 6.939237594604492,
      "learning_rate": 3.699640285719287e-06,
      "loss": 0.0485,
      "step": 11385
    },
    {
      "epoch": 0.5879679834753421,
      "grad_norm": 2.2180159091949463,
      "learning_rate": 3.6988491373413034e-06,
      "loss": 0.0233,
      "step": 11386
    },
    {
      "epoch": 0.5880196230312419,
      "grad_norm": 20.466796875,
      "learning_rate": 3.698058023904523e-06,
      "loss": 0.0589,
      "step": 11387
    },
    {
      "epoch": 0.5880712625871417,
      "grad_norm": 0.8335243463516235,
      "learning_rate": 3.6972669454301894e-06,
      "loss": 0.0141,
      "step": 11388
    },
    {
      "epoch": 0.5881229021430415,
      "grad_norm": 5.070894241333008,
      "learning_rate": 3.6964759019395495e-06,
      "loss": 0.0565,
      "step": 11389
    },
    {
      "epoch": 0.5881745416989413,
      "grad_norm": 5.291976451873779,
      "learning_rate": 3.695684893453846e-06,
      "loss": 0.082,
      "step": 11390
    },
    {
      "epoch": 0.5882261812548412,
      "grad_norm": 2.836517810821533,
      "learning_rate": 3.694893919994317e-06,
      "loss": 0.0519,
      "step": 11391
    },
    {
      "epoch": 0.588277820810741,
      "grad_norm": 2.7175168991088867,
      "learning_rate": 3.6941029815822064e-06,
      "loss": 0.0204,
      "step": 11392
    },
    {
      "epoch": 0.5883294603666408,
      "grad_norm": 8.032014846801758,
      "learning_rate": 3.6933120782387543e-06,
      "loss": 0.0939,
      "step": 11393
    },
    {
      "epoch": 0.5883810999225406,
      "grad_norm": 3.839350938796997,
      "learning_rate": 3.692521209985197e-06,
      "loss": 0.0416,
      "step": 11394
    },
    {
      "epoch": 0.5884327394784404,
      "grad_norm": 7.7076921463012695,
      "learning_rate": 3.691730376842774e-06,
      "loss": 0.0902,
      "step": 11395
    },
    {
      "epoch": 0.5884843790343403,
      "grad_norm": 6.812474250793457,
      "learning_rate": 3.690939578832723e-06,
      "loss": 0.0674,
      "step": 11396
    },
    {
      "epoch": 0.5885360185902401,
      "grad_norm": 2.2652339935302734,
      "learning_rate": 3.690148815976281e-06,
      "loss": 0.0369,
      "step": 11397
    },
    {
      "epoch": 0.5885876581461399,
      "grad_norm": 3.72770357131958,
      "learning_rate": 3.6893580882946813e-06,
      "loss": 0.0517,
      "step": 11398
    },
    {
      "epoch": 0.5886392977020397,
      "grad_norm": 2.5135512351989746,
      "learning_rate": 3.6885673958091584e-06,
      "loss": 0.0329,
      "step": 11399
    },
    {
      "epoch": 0.5886909372579395,
      "grad_norm": 1.3292210102081299,
      "learning_rate": 3.687776738540945e-06,
      "loss": 0.0332,
      "step": 11400
    },
    {
      "epoch": 0.5887425768138393,
      "grad_norm": 3.3231325149536133,
      "learning_rate": 3.6869861165112776e-06,
      "loss": 0.0376,
      "step": 11401
    },
    {
      "epoch": 0.5887942163697392,
      "grad_norm": 10.597342491149902,
      "learning_rate": 3.6861955297413823e-06,
      "loss": 0.0634,
      "step": 11402
    },
    {
      "epoch": 0.5888458559256391,
      "grad_norm": 2.582472324371338,
      "learning_rate": 3.6854049782524916e-06,
      "loss": 0.0275,
      "step": 11403
    },
    {
      "epoch": 0.5888974954815389,
      "grad_norm": 23.43663787841797,
      "learning_rate": 3.6846144620658365e-06,
      "loss": 0.2234,
      "step": 11404
    },
    {
      "epoch": 0.5889491350374387,
      "grad_norm": 3.689568042755127,
      "learning_rate": 3.683823981202643e-06,
      "loss": 0.0269,
      "step": 11405
    },
    {
      "epoch": 0.5890007745933385,
      "grad_norm": 2.175269603729248,
      "learning_rate": 3.683033535684141e-06,
      "loss": 0.0224,
      "step": 11406
    },
    {
      "epoch": 0.5890524141492384,
      "grad_norm": 11.41054916381836,
      "learning_rate": 3.682243125531555e-06,
      "loss": 0.1001,
      "step": 11407
    },
    {
      "epoch": 0.5891040537051382,
      "grad_norm": 7.502758502960205,
      "learning_rate": 3.6814527507661148e-06,
      "loss": 0.0767,
      "step": 11408
    },
    {
      "epoch": 0.589155693261038,
      "grad_norm": 5.1942853927612305,
      "learning_rate": 3.6806624114090384e-06,
      "loss": 0.0546,
      "step": 11409
    },
    {
      "epoch": 0.5892073328169378,
      "grad_norm": 4.8604936599731445,
      "learning_rate": 3.679872107481556e-06,
      "loss": 0.0202,
      "step": 11410
    },
    {
      "epoch": 0.5892589723728376,
      "grad_norm": 11.698371887207031,
      "learning_rate": 3.6790818390048873e-06,
      "loss": 0.1055,
      "step": 11411
    },
    {
      "epoch": 0.5893106119287375,
      "grad_norm": 8.163782119750977,
      "learning_rate": 3.678291606000256e-06,
      "loss": 0.106,
      "step": 11412
    },
    {
      "epoch": 0.5893622514846373,
      "grad_norm": 14.157074928283691,
      "learning_rate": 3.6775014084888816e-06,
      "loss": 0.0942,
      "step": 11413
    },
    {
      "epoch": 0.5894138910405371,
      "grad_norm": 11.106053352355957,
      "learning_rate": 3.676711246491983e-06,
      "loss": 0.1875,
      "step": 11414
    },
    {
      "epoch": 0.5894655305964369,
      "grad_norm": 36.79030990600586,
      "learning_rate": 3.6759211200307843e-06,
      "loss": 0.182,
      "step": 11415
    },
    {
      "epoch": 0.5895171701523367,
      "grad_norm": 1.9198158979415894,
      "learning_rate": 3.6751310291264963e-06,
      "loss": 0.0188,
      "step": 11416
    },
    {
      "epoch": 0.5895688097082366,
      "grad_norm": 4.398242473602295,
      "learning_rate": 3.6743409738003424e-06,
      "loss": 0.0641,
      "step": 11417
    },
    {
      "epoch": 0.5896204492641364,
      "grad_norm": 3.4539573192596436,
      "learning_rate": 3.6735509540735348e-06,
      "loss": 0.0479,
      "step": 11418
    },
    {
      "epoch": 0.5896720888200362,
      "grad_norm": 14.056687355041504,
      "learning_rate": 3.672760969967293e-06,
      "loss": 0.1485,
      "step": 11419
    },
    {
      "epoch": 0.589723728375936,
      "grad_norm": 21.89202880859375,
      "learning_rate": 3.671971021502827e-06,
      "loss": 0.1158,
      "step": 11420
    },
    {
      "epoch": 0.5897753679318358,
      "grad_norm": 6.5581841468811035,
      "learning_rate": 3.6711811087013507e-06,
      "loss": 0.0586,
      "step": 11421
    },
    {
      "epoch": 0.5898270074877356,
      "grad_norm": 6.283742904663086,
      "learning_rate": 3.6703912315840784e-06,
      "loss": 0.0597,
      "step": 11422
    },
    {
      "epoch": 0.5898786470436355,
      "grad_norm": 14.090797424316406,
      "learning_rate": 3.6696013901722223e-06,
      "loss": 0.1319,
      "step": 11423
    },
    {
      "epoch": 0.5899302865995353,
      "grad_norm": 4.512355327606201,
      "learning_rate": 3.6688115844869898e-06,
      "loss": 0.0369,
      "step": 11424
    },
    {
      "epoch": 0.5899819261554351,
      "grad_norm": 4.045856475830078,
      "learning_rate": 3.6680218145495914e-06,
      "loss": 0.0545,
      "step": 11425
    },
    {
      "epoch": 0.5900335657113349,
      "grad_norm": 1.415835976600647,
      "learning_rate": 3.6672320803812382e-06,
      "loss": 0.0227,
      "step": 11426
    },
    {
      "epoch": 0.5900852052672347,
      "grad_norm": 4.57244873046875,
      "learning_rate": 3.666442382003135e-06,
      "loss": 0.0389,
      "step": 11427
    },
    {
      "epoch": 0.5901368448231346,
      "grad_norm": 1.0192934274673462,
      "learning_rate": 3.6656527194364867e-06,
      "loss": 0.0068,
      "step": 11428
    },
    {
      "epoch": 0.5901884843790344,
      "grad_norm": 3.2478842735290527,
      "learning_rate": 3.664863092702503e-06,
      "loss": 0.0286,
      "step": 11429
    },
    {
      "epoch": 0.5902401239349342,
      "grad_norm": 4.322948455810547,
      "learning_rate": 3.6640735018223882e-06,
      "loss": 0.0597,
      "step": 11430
    },
    {
      "epoch": 0.590291763490834,
      "grad_norm": 3.8159759044647217,
      "learning_rate": 3.663283946817344e-06,
      "loss": 0.0436,
      "step": 11431
    },
    {
      "epoch": 0.5903434030467338,
      "grad_norm": 6.139216423034668,
      "learning_rate": 3.662494427708574e-06,
      "loss": 0.0845,
      "step": 11432
    },
    {
      "epoch": 0.5903950426026336,
      "grad_norm": 3.3078153133392334,
      "learning_rate": 3.661704944517278e-06,
      "loss": 0.0413,
      "step": 11433
    },
    {
      "epoch": 0.5904466821585335,
      "grad_norm": 7.558433532714844,
      "learning_rate": 3.660915497264663e-06,
      "loss": 0.0893,
      "step": 11434
    },
    {
      "epoch": 0.5904983217144333,
      "grad_norm": 17.971506118774414,
      "learning_rate": 3.660126085971921e-06,
      "loss": 0.0777,
      "step": 11435
    },
    {
      "epoch": 0.5905499612703331,
      "grad_norm": 9.916054725646973,
      "learning_rate": 3.6593367106602556e-06,
      "loss": 0.1334,
      "step": 11436
    },
    {
      "epoch": 0.5906016008262329,
      "grad_norm": 3.5769503116607666,
      "learning_rate": 3.658547371350865e-06,
      "loss": 0.0591,
      "step": 11437
    },
    {
      "epoch": 0.5906532403821327,
      "grad_norm": 4.361085414886475,
      "learning_rate": 3.657758068064944e-06,
      "loss": 0.0683,
      "step": 11438
    },
    {
      "epoch": 0.5907048799380326,
      "grad_norm": 8.587952613830566,
      "learning_rate": 3.6569688008236894e-06,
      "loss": 0.0798,
      "step": 11439
    },
    {
      "epoch": 0.5907565194939324,
      "grad_norm": 3.874985933303833,
      "learning_rate": 3.6561795696482954e-06,
      "loss": 0.0504,
      "step": 11440
    },
    {
      "epoch": 0.5908081590498322,
      "grad_norm": 4.544193267822266,
      "learning_rate": 3.6553903745599606e-06,
      "loss": 0.0821,
      "step": 11441
    },
    {
      "epoch": 0.590859798605732,
      "grad_norm": 4.4888811111450195,
      "learning_rate": 3.6546012155798703e-06,
      "loss": 0.0447,
      "step": 11442
    },
    {
      "epoch": 0.5909114381616318,
      "grad_norm": 2.9161672592163086,
      "learning_rate": 3.6538120927292233e-06,
      "loss": 0.0404,
      "step": 11443
    },
    {
      "epoch": 0.5909630777175316,
      "grad_norm": 3.695713996887207,
      "learning_rate": 3.653023006029207e-06,
      "loss": 0.0617,
      "step": 11444
    },
    {
      "epoch": 0.5910147172734315,
      "grad_norm": 4.09121036529541,
      "learning_rate": 3.6522339555010157e-06,
      "loss": 0.0355,
      "step": 11445
    },
    {
      "epoch": 0.5910663568293313,
      "grad_norm": 3.960810422897339,
      "learning_rate": 3.6514449411658336e-06,
      "loss": 0.0703,
      "step": 11446
    },
    {
      "epoch": 0.5911179963852311,
      "grad_norm": 1.9656206369400024,
      "learning_rate": 3.65065596304485e-06,
      "loss": 0.033,
      "step": 11447
    },
    {
      "epoch": 0.5911696359411309,
      "grad_norm": 6.140805244445801,
      "learning_rate": 3.6498670211592575e-06,
      "loss": 0.0483,
      "step": 11448
    },
    {
      "epoch": 0.5912212754970307,
      "grad_norm": 4.488682746887207,
      "learning_rate": 3.6490781155302344e-06,
      "loss": 0.048,
      "step": 11449
    },
    {
      "epoch": 0.5912729150529306,
      "grad_norm": 1.8698810338974,
      "learning_rate": 3.6482892461789713e-06,
      "loss": 0.0546,
      "step": 11450
    },
    {
      "epoch": 0.5913245546088304,
      "grad_norm": 2.9777607917785645,
      "learning_rate": 3.6475004131266513e-06,
      "loss": 0.0489,
      "step": 11451
    },
    {
      "epoch": 0.5913761941647302,
      "grad_norm": 4.840794086456299,
      "learning_rate": 3.646711616394459e-06,
      "loss": 0.0311,
      "step": 11452
    },
    {
      "epoch": 0.59142783372063,
      "grad_norm": 2.352839946746826,
      "learning_rate": 3.645922856003574e-06,
      "loss": 0.0254,
      "step": 11453
    },
    {
      "epoch": 0.5914794732765298,
      "grad_norm": 5.746642112731934,
      "learning_rate": 3.6451341319751783e-06,
      "loss": 0.0762,
      "step": 11454
    },
    {
      "epoch": 0.5915311128324297,
      "grad_norm": 6.34208869934082,
      "learning_rate": 3.644345444330454e-06,
      "loss": 0.0595,
      "step": 11455
    },
    {
      "epoch": 0.5915827523883295,
      "grad_norm": 6.588391304016113,
      "learning_rate": 3.6435567930905813e-06,
      "loss": 0.0853,
      "step": 11456
    },
    {
      "epoch": 0.5916343919442293,
      "grad_norm": 3.4757988452911377,
      "learning_rate": 3.6427681782767367e-06,
      "loss": 0.0488,
      "step": 11457
    },
    {
      "epoch": 0.5916860315001291,
      "grad_norm": 12.594568252563477,
      "learning_rate": 3.641979599910097e-06,
      "loss": 0.1071,
      "step": 11458
    },
    {
      "epoch": 0.5917376710560289,
      "grad_norm": 3.072374105453491,
      "learning_rate": 3.641191058011842e-06,
      "loss": 0.0363,
      "step": 11459
    },
    {
      "epoch": 0.5917893106119287,
      "grad_norm": 6.962220191955566,
      "learning_rate": 3.640402552603144e-06,
      "loss": 0.0367,
      "step": 11460
    },
    {
      "epoch": 0.5918409501678286,
      "grad_norm": 6.772146224975586,
      "learning_rate": 3.639614083705178e-06,
      "loss": 0.0494,
      "step": 11461
    },
    {
      "epoch": 0.5918925897237284,
      "grad_norm": 2.9653217792510986,
      "learning_rate": 3.6388256513391196e-06,
      "loss": 0.036,
      "step": 11462
    },
    {
      "epoch": 0.5919442292796282,
      "grad_norm": 2.5672767162323,
      "learning_rate": 3.638037255526141e-06,
      "loss": 0.0433,
      "step": 11463
    },
    {
      "epoch": 0.591995868835528,
      "grad_norm": 4.394668102264404,
      "learning_rate": 3.637248896287412e-06,
      "loss": 0.0533,
      "step": 11464
    },
    {
      "epoch": 0.5920475083914278,
      "grad_norm": 3.207838535308838,
      "learning_rate": 3.636460573644105e-06,
      "loss": 0.0388,
      "step": 11465
    },
    {
      "epoch": 0.5920991479473277,
      "grad_norm": 2.981294631958008,
      "learning_rate": 3.6356722876173876e-06,
      "loss": 0.0392,
      "step": 11466
    },
    {
      "epoch": 0.5921507875032275,
      "grad_norm": 5.389118194580078,
      "learning_rate": 3.6348840382284334e-06,
      "loss": 0.0947,
      "step": 11467
    },
    {
      "epoch": 0.5922024270591273,
      "grad_norm": 2.8718841075897217,
      "learning_rate": 3.6340958254984027e-06,
      "loss": 0.024,
      "step": 11468
    },
    {
      "epoch": 0.5922540666150271,
      "grad_norm": 6.259592056274414,
      "learning_rate": 3.633307649448468e-06,
      "loss": 0.0714,
      "step": 11469
    },
    {
      "epoch": 0.5923057061709269,
      "grad_norm": 8.854859352111816,
      "learning_rate": 3.6325195100997947e-06,
      "loss": 0.1045,
      "step": 11470
    },
    {
      "epoch": 0.5923573457268267,
      "grad_norm": 2.828406572341919,
      "learning_rate": 3.6317314074735444e-06,
      "loss": 0.0362,
      "step": 11471
    },
    {
      "epoch": 0.5924089852827266,
      "grad_norm": 4.633342742919922,
      "learning_rate": 3.6309433415908824e-06,
      "loss": 0.0624,
      "step": 11472
    },
    {
      "epoch": 0.5924606248386264,
      "grad_norm": 11.781526565551758,
      "learning_rate": 3.6301553124729706e-06,
      "loss": 0.0769,
      "step": 11473
    },
    {
      "epoch": 0.5925122643945262,
      "grad_norm": 20.269502639770508,
      "learning_rate": 3.629367320140975e-06,
      "loss": 0.1765,
      "step": 11474
    },
    {
      "epoch": 0.592563903950426,
      "grad_norm": 2.68074369430542,
      "learning_rate": 3.6285793646160504e-06,
      "loss": 0.0398,
      "step": 11475
    },
    {
      "epoch": 0.5926155435063258,
      "grad_norm": 3.1582064628601074,
      "learning_rate": 3.6277914459193607e-06,
      "loss": 0.0468,
      "step": 11476
    },
    {
      "epoch": 0.5926671830622257,
      "grad_norm": 11.897098541259766,
      "learning_rate": 3.6270035640720634e-06,
      "loss": 0.1183,
      "step": 11477
    },
    {
      "epoch": 0.5927188226181255,
      "grad_norm": 8.715653419494629,
      "learning_rate": 3.6262157190953175e-06,
      "loss": 0.1368,
      "step": 11478
    },
    {
      "epoch": 0.5927704621740253,
      "grad_norm": 3.6870245933532715,
      "learning_rate": 3.6254279110102777e-06,
      "loss": 0.05,
      "step": 11479
    },
    {
      "epoch": 0.5928221017299251,
      "grad_norm": 13.227896690368652,
      "learning_rate": 3.6246401398380993e-06,
      "loss": 0.1091,
      "step": 11480
    },
    {
      "epoch": 0.5928737412858249,
      "grad_norm": 14.219395637512207,
      "learning_rate": 3.6238524055999426e-06,
      "loss": 0.0912,
      "step": 11481
    },
    {
      "epoch": 0.5929253808417247,
      "grad_norm": 4.811700820922852,
      "learning_rate": 3.6230647083169546e-06,
      "loss": 0.0369,
      "step": 11482
    },
    {
      "epoch": 0.5929770203976246,
      "grad_norm": 9.040460586547852,
      "learning_rate": 3.622277048010292e-06,
      "loss": 0.1236,
      "step": 11483
    },
    {
      "epoch": 0.5930286599535244,
      "grad_norm": 3.082779884338379,
      "learning_rate": 3.621489424701106e-06,
      "loss": 0.0324,
      "step": 11484
    },
    {
      "epoch": 0.5930802995094242,
      "grad_norm": 6.306744575500488,
      "learning_rate": 3.6207018384105484e-06,
      "loss": 0.0535,
      "step": 11485
    },
    {
      "epoch": 0.593131939065324,
      "grad_norm": 3.541118621826172,
      "learning_rate": 3.6199142891597673e-06,
      "loss": 0.0448,
      "step": 11486
    },
    {
      "epoch": 0.5931835786212238,
      "grad_norm": 3.2993762493133545,
      "learning_rate": 3.619126776969911e-06,
      "loss": 0.0662,
      "step": 11487
    },
    {
      "epoch": 0.5932352181771237,
      "grad_norm": 4.0389323234558105,
      "learning_rate": 3.6183393018621303e-06,
      "loss": 0.0684,
      "step": 11488
    },
    {
      "epoch": 0.5932868577330235,
      "grad_norm": 2.25408935546875,
      "learning_rate": 3.617551863857571e-06,
      "loss": 0.0308,
      "step": 11489
    },
    {
      "epoch": 0.5933384972889233,
      "grad_norm": 29.534027099609375,
      "learning_rate": 3.616764462977379e-06,
      "loss": 0.3502,
      "step": 11490
    },
    {
      "epoch": 0.5933901368448231,
      "grad_norm": 12.67044448852539,
      "learning_rate": 3.615977099242698e-06,
      "loss": 0.1173,
      "step": 11491
    },
    {
      "epoch": 0.5934417764007229,
      "grad_norm": 15.74967098236084,
      "learning_rate": 3.6151897726746743e-06,
      "loss": 0.1169,
      "step": 11492
    },
    {
      "epoch": 0.5934934159566227,
      "grad_norm": 2.8250672817230225,
      "learning_rate": 3.614402483294448e-06,
      "loss": 0.0561,
      "step": 11493
    },
    {
      "epoch": 0.5935450555125226,
      "grad_norm": 6.576835632324219,
      "learning_rate": 3.6136152311231614e-06,
      "loss": 0.0744,
      "step": 11494
    },
    {
      "epoch": 0.5935966950684224,
      "grad_norm": 12.871258735656738,
      "learning_rate": 3.6128280161819574e-06,
      "loss": 0.1065,
      "step": 11495
    },
    {
      "epoch": 0.5936483346243222,
      "grad_norm": 3.824711322784424,
      "learning_rate": 3.6120408384919762e-06,
      "loss": 0.059,
      "step": 11496
    },
    {
      "epoch": 0.593699974180222,
      "grad_norm": 2.3186116218566895,
      "learning_rate": 3.6112536980743547e-06,
      "loss": 0.0268,
      "step": 11497
    },
    {
      "epoch": 0.5937516137361218,
      "grad_norm": 4.912849426269531,
      "learning_rate": 3.6104665949502314e-06,
      "loss": 0.0791,
      "step": 11498
    },
    {
      "epoch": 0.5938032532920217,
      "grad_norm": 1.6931780576705933,
      "learning_rate": 3.6096795291407423e-06,
      "loss": 0.032,
      "step": 11499
    },
    {
      "epoch": 0.5938548928479215,
      "grad_norm": 6.667208194732666,
      "learning_rate": 3.608892500667027e-06,
      "loss": 0.0647,
      "step": 11500
    },
    {
      "epoch": 0.5939065324038213,
      "grad_norm": 17.43433952331543,
      "learning_rate": 3.608105509550216e-06,
      "loss": 0.1112,
      "step": 11501
    },
    {
      "epoch": 0.5939581719597211,
      "grad_norm": 4.31650972366333,
      "learning_rate": 3.6073185558114455e-06,
      "loss": 0.0601,
      "step": 11502
    },
    {
      "epoch": 0.5940098115156209,
      "grad_norm": 3.288914203643799,
      "learning_rate": 3.6065316394718497e-06,
      "loss": 0.0737,
      "step": 11503
    },
    {
      "epoch": 0.5940614510715208,
      "grad_norm": 8.271084785461426,
      "learning_rate": 3.605744760552557e-06,
      "loss": 0.0768,
      "step": 11504
    },
    {
      "epoch": 0.5941130906274206,
      "grad_norm": 5.491982460021973,
      "learning_rate": 3.6049579190747004e-06,
      "loss": 0.0804,
      "step": 11505
    },
    {
      "epoch": 0.5941647301833204,
      "grad_norm": 1.9612329006195068,
      "learning_rate": 3.6041711150594084e-06,
      "loss": 0.0318,
      "step": 11506
    },
    {
      "epoch": 0.5942163697392202,
      "grad_norm": 13.863758087158203,
      "learning_rate": 3.6033843485278142e-06,
      "loss": 0.0619,
      "step": 11507
    },
    {
      "epoch": 0.59426800929512,
      "grad_norm": 6.53391695022583,
      "learning_rate": 3.6025976195010394e-06,
      "loss": 0.0697,
      "step": 11508
    },
    {
      "epoch": 0.5943196488510198,
      "grad_norm": 17.197463989257812,
      "learning_rate": 3.601810928000216e-06,
      "loss": 0.1181,
      "step": 11509
    },
    {
      "epoch": 0.5943712884069197,
      "grad_norm": 6.767115116119385,
      "learning_rate": 3.601024274046466e-06,
      "loss": 0.0623,
      "step": 11510
    },
    {
      "epoch": 0.5944229279628195,
      "grad_norm": 15.351476669311523,
      "learning_rate": 3.6002376576609186e-06,
      "loss": 0.0706,
      "step": 11511
    },
    {
      "epoch": 0.5944745675187193,
      "grad_norm": 9.019513130187988,
      "learning_rate": 3.5994510788646942e-06,
      "loss": 0.0381,
      "step": 11512
    },
    {
      "epoch": 0.5945262070746191,
      "grad_norm": 2.7440595626831055,
      "learning_rate": 3.5986645376789146e-06,
      "loss": 0.0234,
      "step": 11513
    },
    {
      "epoch": 0.5945778466305189,
      "grad_norm": 2.13801908493042,
      "learning_rate": 3.597878034124708e-06,
      "loss": 0.052,
      "step": 11514
    },
    {
      "epoch": 0.5946294861864188,
      "grad_norm": 1.3082818984985352,
      "learning_rate": 3.5970915682231866e-06,
      "loss": 0.0209,
      "step": 11515
    },
    {
      "epoch": 0.5946811257423186,
      "grad_norm": 1.7701480388641357,
      "learning_rate": 3.596305139995476e-06,
      "loss": 0.0276,
      "step": 11516
    },
    {
      "epoch": 0.5947327652982184,
      "grad_norm": 3.107874631881714,
      "learning_rate": 3.595518749462694e-06,
      "loss": 0.0209,
      "step": 11517
    },
    {
      "epoch": 0.5947844048541182,
      "grad_norm": 10.13880443572998,
      "learning_rate": 3.594732396645959e-06,
      "loss": 0.0607,
      "step": 11518
    },
    {
      "epoch": 0.594836044410018,
      "grad_norm": 5.024535655975342,
      "learning_rate": 3.5939460815663853e-06,
      "loss": 0.0847,
      "step": 11519
    },
    {
      "epoch": 0.5948876839659178,
      "grad_norm": 2.9980671405792236,
      "learning_rate": 3.5931598042450886e-06,
      "loss": 0.0395,
      "step": 11520
    },
    {
      "epoch": 0.5949393235218177,
      "grad_norm": 5.355659484863281,
      "learning_rate": 3.592373564703187e-06,
      "loss": 0.0617,
      "step": 11521
    },
    {
      "epoch": 0.5949909630777175,
      "grad_norm": 2.5424582958221436,
      "learning_rate": 3.591587362961794e-06,
      "loss": 0.0236,
      "step": 11522
    },
    {
      "epoch": 0.5950426026336173,
      "grad_norm": 10.007575988769531,
      "learning_rate": 3.5908011990420177e-06,
      "loss": 0.2055,
      "step": 11523
    },
    {
      "epoch": 0.5950942421895171,
      "grad_norm": 10.939432144165039,
      "learning_rate": 3.590015072964974e-06,
      "loss": 0.0887,
      "step": 11524
    },
    {
      "epoch": 0.5951458817454169,
      "grad_norm": 4.633469104766846,
      "learning_rate": 3.5892289847517735e-06,
      "loss": 0.0451,
      "step": 11525
    },
    {
      "epoch": 0.5951975213013168,
      "grad_norm": 11.08211898803711,
      "learning_rate": 3.5884429344235227e-06,
      "loss": 0.1275,
      "step": 11526
    },
    {
      "epoch": 0.5952491608572166,
      "grad_norm": 2.6620218753814697,
      "learning_rate": 3.5876569220013326e-06,
      "loss": 0.063,
      "step": 11527
    },
    {
      "epoch": 0.5953008004131164,
      "grad_norm": 5.293104648590088,
      "learning_rate": 3.5868709475063106e-06,
      "loss": 0.0704,
      "step": 11528
    },
    {
      "epoch": 0.5953524399690162,
      "grad_norm": 4.338155746459961,
      "learning_rate": 3.5860850109595646e-06,
      "loss": 0.0953,
      "step": 11529
    },
    {
      "epoch": 0.5954040795249161,
      "grad_norm": 11.31679916381836,
      "learning_rate": 3.5852991123821977e-06,
      "loss": 0.0328,
      "step": 11530
    },
    {
      "epoch": 0.595455719080816,
      "grad_norm": 7.051494598388672,
      "learning_rate": 3.584513251795316e-06,
      "loss": 0.0826,
      "step": 11531
    },
    {
      "epoch": 0.5955073586367158,
      "grad_norm": 3.2486984729766846,
      "learning_rate": 3.5837274292200214e-06,
      "loss": 0.0592,
      "step": 11532
    },
    {
      "epoch": 0.5955589981926156,
      "grad_norm": 4.6461005210876465,
      "learning_rate": 3.5829416446774213e-06,
      "loss": 0.0815,
      "step": 11533
    },
    {
      "epoch": 0.5956106377485154,
      "grad_norm": 3.9336795806884766,
      "learning_rate": 3.5821558981886093e-06,
      "loss": 0.0392,
      "step": 11534
    },
    {
      "epoch": 0.5956622773044152,
      "grad_norm": 6.097018241882324,
      "learning_rate": 3.5813701897746924e-06,
      "loss": 0.0483,
      "step": 11535
    },
    {
      "epoch": 0.595713916860315,
      "grad_norm": 6.788157939910889,
      "learning_rate": 3.580584519456769e-06,
      "loss": 0.0705,
      "step": 11536
    },
    {
      "epoch": 0.5957655564162149,
      "grad_norm": 4.47444486618042,
      "learning_rate": 3.5797988872559347e-06,
      "loss": 0.0586,
      "step": 11537
    },
    {
      "epoch": 0.5958171959721147,
      "grad_norm": 3.869873285293579,
      "learning_rate": 3.5790132931932893e-06,
      "loss": 0.0416,
      "step": 11538
    },
    {
      "epoch": 0.5958688355280145,
      "grad_norm": 5.140684127807617,
      "learning_rate": 3.578227737289927e-06,
      "loss": 0.0487,
      "step": 11539
    },
    {
      "epoch": 0.5959204750839143,
      "grad_norm": 3.3918933868408203,
      "learning_rate": 3.5774422195669486e-06,
      "loss": 0.0751,
      "step": 11540
    },
    {
      "epoch": 0.5959721146398141,
      "grad_norm": 7.646668910980225,
      "learning_rate": 3.576656740045441e-06,
      "loss": 0.068,
      "step": 11541
    },
    {
      "epoch": 0.596023754195714,
      "grad_norm": 3.9362246990203857,
      "learning_rate": 3.575871298746502e-06,
      "loss": 0.0557,
      "step": 11542
    },
    {
      "epoch": 0.5960753937516138,
      "grad_norm": 4.135868549346924,
      "learning_rate": 3.575085895691224e-06,
      "loss": 0.069,
      "step": 11543
    },
    {
      "epoch": 0.5961270333075136,
      "grad_norm": 3.5256853103637695,
      "learning_rate": 3.5743005309006974e-06,
      "loss": 0.0524,
      "step": 11544
    },
    {
      "epoch": 0.5961786728634134,
      "grad_norm": 4.618027210235596,
      "learning_rate": 3.573515204396012e-06,
      "loss": 0.0481,
      "step": 11545
    },
    {
      "epoch": 0.5962303124193132,
      "grad_norm": 1.6978092193603516,
      "learning_rate": 3.5727299161982555e-06,
      "loss": 0.0223,
      "step": 11546
    },
    {
      "epoch": 0.596281951975213,
      "grad_norm": 2.1320242881774902,
      "learning_rate": 3.571944666328522e-06,
      "loss": 0.0517,
      "step": 11547
    },
    {
      "epoch": 0.5963335915311129,
      "grad_norm": 14.97817611694336,
      "learning_rate": 3.571159454807891e-06,
      "loss": 0.1734,
      "step": 11548
    },
    {
      "epoch": 0.5963852310870127,
      "grad_norm": 2.7433512210845947,
      "learning_rate": 3.5703742816574538e-06,
      "loss": 0.0435,
      "step": 11549
    },
    {
      "epoch": 0.5964368706429125,
      "grad_norm": 4.404533863067627,
      "learning_rate": 3.5695891468982936e-06,
      "loss": 0.0609,
      "step": 11550
    },
    {
      "epoch": 0.5964885101988123,
      "grad_norm": 14.764697074890137,
      "learning_rate": 3.5688040505514952e-06,
      "loss": 0.0815,
      "step": 11551
    },
    {
      "epoch": 0.5965401497547121,
      "grad_norm": 6.57779598236084,
      "learning_rate": 3.5680189926381413e-06,
      "loss": 0.082,
      "step": 11552
    },
    {
      "epoch": 0.596591789310612,
      "grad_norm": 1.7133777141571045,
      "learning_rate": 3.5672339731793124e-06,
      "loss": 0.0249,
      "step": 11553
    },
    {
      "epoch": 0.5966434288665118,
      "grad_norm": 4.164237022399902,
      "learning_rate": 3.566448992196092e-06,
      "loss": 0.0505,
      "step": 11554
    },
    {
      "epoch": 0.5966950684224116,
      "grad_norm": 7.857363700866699,
      "learning_rate": 3.5656640497095607e-06,
      "loss": 0.0973,
      "step": 11555
    },
    {
      "epoch": 0.5967467079783114,
      "grad_norm": 1.2459405660629272,
      "learning_rate": 3.564879145740794e-06,
      "loss": 0.0137,
      "step": 11556
    },
    {
      "epoch": 0.5967983475342112,
      "grad_norm": 2.8634870052337646,
      "learning_rate": 3.564094280310871e-06,
      "loss": 0.0421,
      "step": 11557
    },
    {
      "epoch": 0.5968499870901111,
      "grad_norm": 4.078132629394531,
      "learning_rate": 3.5633094534408706e-06,
      "loss": 0.0475,
      "step": 11558
    },
    {
      "epoch": 0.5969016266460109,
      "grad_norm": 2.1289026737213135,
      "learning_rate": 3.5625246651518664e-06,
      "loss": 0.0274,
      "step": 11559
    },
    {
      "epoch": 0.5969532662019107,
      "grad_norm": 4.355618953704834,
      "learning_rate": 3.5617399154649323e-06,
      "loss": 0.0599,
      "step": 11560
    },
    {
      "epoch": 0.5970049057578105,
      "grad_norm": 3.7905123233795166,
      "learning_rate": 3.5609552044011442e-06,
      "loss": 0.0486,
      "step": 11561
    },
    {
      "epoch": 0.5970565453137103,
      "grad_norm": 6.74152135848999,
      "learning_rate": 3.560170531981576e-06,
      "loss": 0.0408,
      "step": 11562
    },
    {
      "epoch": 0.5971081848696101,
      "grad_norm": 2.4173083305358887,
      "learning_rate": 3.5593858982272954e-06,
      "loss": 0.022,
      "step": 11563
    },
    {
      "epoch": 0.59715982442551,
      "grad_norm": 3.6854875087738037,
      "learning_rate": 3.558601303159376e-06,
      "loss": 0.0727,
      "step": 11564
    },
    {
      "epoch": 0.5972114639814098,
      "grad_norm": 12.054100036621094,
      "learning_rate": 3.5578167467988847e-06,
      "loss": 0.1162,
      "step": 11565
    },
    {
      "epoch": 0.5972631035373096,
      "grad_norm": 4.309013843536377,
      "learning_rate": 3.5570322291668946e-06,
      "loss": 0.0607,
      "step": 11566
    },
    {
      "epoch": 0.5973147430932094,
      "grad_norm": 2.806809663772583,
      "learning_rate": 3.5562477502844677e-06,
      "loss": 0.0674,
      "step": 11567
    },
    {
      "epoch": 0.5973663826491092,
      "grad_norm": 15.759173393249512,
      "learning_rate": 3.5554633101726742e-06,
      "loss": 0.1779,
      "step": 11568
    },
    {
      "epoch": 0.5974180222050091,
      "grad_norm": 7.385586738586426,
      "learning_rate": 3.5546789088525795e-06,
      "loss": 0.0661,
      "step": 11569
    },
    {
      "epoch": 0.5974696617609089,
      "grad_norm": 2.123793363571167,
      "learning_rate": 3.553894546345246e-06,
      "loss": 0.0514,
      "step": 11570
    },
    {
      "epoch": 0.5975213013168087,
      "grad_norm": 6.880616188049316,
      "learning_rate": 3.5531102226717373e-06,
      "loss": 0.1062,
      "step": 11571
    },
    {
      "epoch": 0.5975729408727085,
      "grad_norm": 3.8998475074768066,
      "learning_rate": 3.552325937853116e-06,
      "loss": 0.0896,
      "step": 11572
    },
    {
      "epoch": 0.5976245804286083,
      "grad_norm": 4.96627950668335,
      "learning_rate": 3.551541691910446e-06,
      "loss": 0.0799,
      "step": 11573
    },
    {
      "epoch": 0.5976762199845082,
      "grad_norm": 2.261035203933716,
      "learning_rate": 3.550757484864783e-06,
      "loss": 0.037,
      "step": 11574
    },
    {
      "epoch": 0.597727859540408,
      "grad_norm": 5.232062816619873,
      "learning_rate": 3.5499733167371885e-06,
      "loss": 0.0453,
      "step": 11575
    },
    {
      "epoch": 0.5977794990963078,
      "grad_norm": 8.920319557189941,
      "learning_rate": 3.5491891875487207e-06,
      "loss": 0.082,
      "step": 11576
    },
    {
      "epoch": 0.5978311386522076,
      "grad_norm": 3.4317939281463623,
      "learning_rate": 3.5484050973204376e-06,
      "loss": 0.0508,
      "step": 11577
    },
    {
      "epoch": 0.5978827782081074,
      "grad_norm": 1.7727857828140259,
      "learning_rate": 3.547621046073393e-06,
      "loss": 0.0203,
      "step": 11578
    },
    {
      "epoch": 0.5979344177640072,
      "grad_norm": 3.511082172393799,
      "learning_rate": 3.546837033828642e-06,
      "loss": 0.0375,
      "step": 11579
    },
    {
      "epoch": 0.5979860573199071,
      "grad_norm": 2.6924479007720947,
      "learning_rate": 3.546053060607242e-06,
      "loss": 0.0347,
      "step": 11580
    },
    {
      "epoch": 0.5980376968758069,
      "grad_norm": 2.7127737998962402,
      "learning_rate": 3.5452691264302404e-06,
      "loss": 0.0366,
      "step": 11581
    },
    {
      "epoch": 0.5980893364317067,
      "grad_norm": 8.473963737487793,
      "learning_rate": 3.5444852313186928e-06,
      "loss": 0.0559,
      "step": 11582
    },
    {
      "epoch": 0.5981409759876065,
      "grad_norm": 5.106199264526367,
      "learning_rate": 3.5437013752936494e-06,
      "loss": 0.0538,
      "step": 11583
    },
    {
      "epoch": 0.5981926155435063,
      "grad_norm": 6.816015243530273,
      "learning_rate": 3.5429175583761605e-06,
      "loss": 0.0508,
      "step": 11584
    },
    {
      "epoch": 0.5982442550994062,
      "grad_norm": 3.714362144470215,
      "learning_rate": 3.542133780587273e-06,
      "loss": 0.0702,
      "step": 11585
    },
    {
      "epoch": 0.598295894655306,
      "grad_norm": 2.1306159496307373,
      "learning_rate": 3.541350041948034e-06,
      "loss": 0.0346,
      "step": 11586
    },
    {
      "epoch": 0.5983475342112058,
      "grad_norm": 3.563809633255005,
      "learning_rate": 3.5405663424794928e-06,
      "loss": 0.053,
      "step": 11587
    },
    {
      "epoch": 0.5983991737671056,
      "grad_norm": 1.9909400939941406,
      "learning_rate": 3.539782682202695e-06,
      "loss": 0.0484,
      "step": 11588
    },
    {
      "epoch": 0.5984508133230054,
      "grad_norm": 2.1136903762817383,
      "learning_rate": 3.538999061138682e-06,
      "loss": 0.035,
      "step": 11589
    },
    {
      "epoch": 0.5985024528789052,
      "grad_norm": 2.4494729042053223,
      "learning_rate": 3.5382154793084993e-06,
      "loss": 0.0475,
      "step": 11590
    },
    {
      "epoch": 0.5985540924348051,
      "grad_norm": 8.346284866333008,
      "learning_rate": 3.53743193673319e-06,
      "loss": 0.0538,
      "step": 11591
    },
    {
      "epoch": 0.5986057319907049,
      "grad_norm": 9.589308738708496,
      "learning_rate": 3.536648433433793e-06,
      "loss": 0.0724,
      "step": 11592
    },
    {
      "epoch": 0.5986573715466047,
      "grad_norm": 3.7418389320373535,
      "learning_rate": 3.535864969431349e-06,
      "loss": 0.0311,
      "step": 11593
    },
    {
      "epoch": 0.5987090111025045,
      "grad_norm": 3.405039072036743,
      "learning_rate": 3.535081544746899e-06,
      "loss": 0.0457,
      "step": 11594
    },
    {
      "epoch": 0.5987606506584043,
      "grad_norm": 18.178333282470703,
      "learning_rate": 3.5342981594014813e-06,
      "loss": 0.0947,
      "step": 11595
    },
    {
      "epoch": 0.5988122902143042,
      "grad_norm": 8.733964920043945,
      "learning_rate": 3.5335148134161307e-06,
      "loss": 0.0589,
      "step": 11596
    },
    {
      "epoch": 0.598863929770204,
      "grad_norm": 1.8037697076797485,
      "learning_rate": 3.5327315068118843e-06,
      "loss": 0.0224,
      "step": 11597
    },
    {
      "epoch": 0.5989155693261038,
      "grad_norm": 2.6654293537139893,
      "learning_rate": 3.531948239609777e-06,
      "loss": 0.0322,
      "step": 11598
    },
    {
      "epoch": 0.5989672088820036,
      "grad_norm": 13.13000774383545,
      "learning_rate": 3.531165011830845e-06,
      "loss": 0.0566,
      "step": 11599
    },
    {
      "epoch": 0.5990188484379034,
      "grad_norm": 3.7076566219329834,
      "learning_rate": 3.5303818234961164e-06,
      "loss": 0.0221,
      "step": 11600
    },
    {
      "epoch": 0.5990704879938032,
      "grad_norm": 3.408747434616089,
      "learning_rate": 3.5295986746266266e-06,
      "loss": 0.0516,
      "step": 11601
    },
    {
      "epoch": 0.5991221275497031,
      "grad_norm": 3.7074451446533203,
      "learning_rate": 3.5288155652434064e-06,
      "loss": 0.0505,
      "step": 11602
    },
    {
      "epoch": 0.5991737671056029,
      "grad_norm": 4.201208114624023,
      "learning_rate": 3.5280324953674837e-06,
      "loss": 0.0241,
      "step": 11603
    },
    {
      "epoch": 0.5992254066615027,
      "grad_norm": 2.262575626373291,
      "learning_rate": 3.527249465019888e-06,
      "loss": 0.0255,
      "step": 11604
    },
    {
      "epoch": 0.5992770462174025,
      "grad_norm": 5.103691577911377,
      "learning_rate": 3.526466474221646e-06,
      "loss": 0.0736,
      "step": 11605
    },
    {
      "epoch": 0.5993286857733023,
      "grad_norm": 2.2850406169891357,
      "learning_rate": 3.525683522993787e-06,
      "loss": 0.032,
      "step": 11606
    },
    {
      "epoch": 0.5993803253292022,
      "grad_norm": 6.584433555603027,
      "learning_rate": 3.524900611357332e-06,
      "loss": 0.0572,
      "step": 11607
    },
    {
      "epoch": 0.599431964885102,
      "grad_norm": 2.3393383026123047,
      "learning_rate": 3.5241177393333087e-06,
      "loss": 0.0315,
      "step": 11608
    },
    {
      "epoch": 0.5994836044410018,
      "grad_norm": 5.847134113311768,
      "learning_rate": 3.523334906942739e-06,
      "loss": 0.0897,
      "step": 11609
    },
    {
      "epoch": 0.5995352439969016,
      "grad_norm": 2.2866570949554443,
      "learning_rate": 3.5225521142066478e-06,
      "loss": 0.0323,
      "step": 11610
    },
    {
      "epoch": 0.5995868835528014,
      "grad_norm": 9.585341453552246,
      "learning_rate": 3.5217693611460522e-06,
      "loss": 0.1079,
      "step": 11611
    },
    {
      "epoch": 0.5996385231087012,
      "grad_norm": 3.0996248722076416,
      "learning_rate": 3.5209866477819734e-06,
      "loss": 0.0431,
      "step": 11612
    },
    {
      "epoch": 0.5996901626646011,
      "grad_norm": 3.0332560539245605,
      "learning_rate": 3.5202039741354342e-06,
      "loss": 0.038,
      "step": 11613
    },
    {
      "epoch": 0.5997418022205009,
      "grad_norm": 7.409884929656982,
      "learning_rate": 3.519421340227446e-06,
      "loss": 0.0929,
      "step": 11614
    },
    {
      "epoch": 0.5997934417764007,
      "grad_norm": 8.988779067993164,
      "learning_rate": 3.518638746079031e-06,
      "loss": 0.0379,
      "step": 11615
    },
    {
      "epoch": 0.5998450813323005,
      "grad_norm": 6.419554710388184,
      "learning_rate": 3.517856191711203e-06,
      "loss": 0.0681,
      "step": 11616
    },
    {
      "epoch": 0.5998967208882003,
      "grad_norm": 2.4551498889923096,
      "learning_rate": 3.517073677144979e-06,
      "loss": 0.033,
      "step": 11617
    },
    {
      "epoch": 0.5999483604441002,
      "grad_norm": 3.3513593673706055,
      "learning_rate": 3.516291202401369e-06,
      "loss": 0.0785,
      "step": 11618
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6508549451828003,
      "learning_rate": 3.5155087675013865e-06,
      "loss": 0.0406,
      "step": 11619
    },
    {
      "epoch": 0.6000516395558998,
      "grad_norm": 3.9291820526123047,
      "learning_rate": 3.514726372466045e-06,
      "loss": 0.0453,
      "step": 11620
    },
    {
      "epoch": 0.6001032791117996,
      "grad_norm": 3.3719394207000732,
      "learning_rate": 3.513944017316355e-06,
      "loss": 0.0478,
      "step": 11621
    },
    {
      "epoch": 0.6001549186676994,
      "grad_norm": 3.4417757987976074,
      "learning_rate": 3.513161702073325e-06,
      "loss": 0.05,
      "step": 11622
    },
    {
      "epoch": 0.6001549186676994,
      "eval_loss": 0.05985761061310768,
      "eval_runtime": 34.1695,
      "eval_samples_per_second": 45.801,
      "eval_steps_per_second": 5.736,
      "step": 11622
    },
    {
      "epoch": 0.6002065582235993,
      "grad_norm": 3.4763007164001465,
      "learning_rate": 3.5123794267579625e-06,
      "loss": 0.0309,
      "step": 11623
    },
    {
      "epoch": 0.6002581977794991,
      "grad_norm": 7.139785289764404,
      "learning_rate": 3.5115971913912768e-06,
      "loss": 0.0596,
      "step": 11624
    },
    {
      "epoch": 0.6003098373353989,
      "grad_norm": 2.593587875366211,
      "learning_rate": 3.5108149959942716e-06,
      "loss": 0.0594,
      "step": 11625
    },
    {
      "epoch": 0.6003614768912987,
      "grad_norm": 3.9516594409942627,
      "learning_rate": 3.5100328405879526e-06,
      "loss": 0.0521,
      "step": 11626
    },
    {
      "epoch": 0.6004131164471985,
      "grad_norm": 2.9663517475128174,
      "learning_rate": 3.509250725193325e-06,
      "loss": 0.0675,
      "step": 11627
    },
    {
      "epoch": 0.6004647560030983,
      "grad_norm": 2.8697056770324707,
      "learning_rate": 3.5084686498313934e-06,
      "loss": 0.0278,
      "step": 11628
    },
    {
      "epoch": 0.6005163955589982,
      "grad_norm": 5.517875671386719,
      "learning_rate": 3.5076866145231557e-06,
      "loss": 0.0567,
      "step": 11629
    },
    {
      "epoch": 0.600568035114898,
      "grad_norm": 4.887753009796143,
      "learning_rate": 3.5069046192896156e-06,
      "loss": 0.0634,
      "step": 11630
    },
    {
      "epoch": 0.6006196746707978,
      "grad_norm": 5.960012912750244,
      "learning_rate": 3.5061226641517703e-06,
      "loss": 0.0643,
      "step": 11631
    },
    {
      "epoch": 0.6006713142266976,
      "grad_norm": 5.359626770019531,
      "learning_rate": 3.5053407491306235e-06,
      "loss": 0.0517,
      "step": 11632
    },
    {
      "epoch": 0.6007229537825974,
      "grad_norm": 14.074085235595703,
      "learning_rate": 3.504558874247167e-06,
      "loss": 0.0845,
      "step": 11633
    },
    {
      "epoch": 0.6007745933384973,
      "grad_norm": 3.1502137184143066,
      "learning_rate": 3.5037770395224004e-06,
      "loss": 0.0292,
      "step": 11634
    },
    {
      "epoch": 0.6008262328943971,
      "grad_norm": 1.8970935344696045,
      "learning_rate": 3.5029952449773188e-06,
      "loss": 0.0466,
      "step": 11635
    },
    {
      "epoch": 0.6008778724502969,
      "grad_norm": 6.841381549835205,
      "learning_rate": 3.5022134906329163e-06,
      "loss": 0.0963,
      "step": 11636
    },
    {
      "epoch": 0.6009295120061967,
      "grad_norm": 4.848183631896973,
      "learning_rate": 3.501431776510186e-06,
      "loss": 0.0517,
      "step": 11637
    },
    {
      "epoch": 0.6009811515620965,
      "grad_norm": 2.6511542797088623,
      "learning_rate": 3.5006501026301187e-06,
      "loss": 0.0503,
      "step": 11638
    },
    {
      "epoch": 0.6010327911179963,
      "grad_norm": 3.772038698196411,
      "learning_rate": 3.499868469013711e-06,
      "loss": 0.0434,
      "step": 11639
    },
    {
      "epoch": 0.6010844306738962,
      "grad_norm": 3.4718775749206543,
      "learning_rate": 3.499086875681945e-06,
      "loss": 0.0389,
      "step": 11640
    },
    {
      "epoch": 0.601136070229796,
      "grad_norm": 2.014981746673584,
      "learning_rate": 3.498305322655816e-06,
      "loss": 0.0357,
      "step": 11641
    },
    {
      "epoch": 0.6011877097856958,
      "grad_norm": 6.491550445556641,
      "learning_rate": 3.497523809956309e-06,
      "loss": 0.1343,
      "step": 11642
    },
    {
      "epoch": 0.6012393493415956,
      "grad_norm": 3.3638103008270264,
      "learning_rate": 3.496742337604413e-06,
      "loss": 0.061,
      "step": 11643
    },
    {
      "epoch": 0.6012909888974954,
      "grad_norm": 3.038710594177246,
      "learning_rate": 3.4959609056211107e-06,
      "loss": 0.0548,
      "step": 11644
    },
    {
      "epoch": 0.6013426284533953,
      "grad_norm": 3.5689258575439453,
      "learning_rate": 3.495179514027387e-06,
      "loss": 0.0826,
      "step": 11645
    },
    {
      "epoch": 0.6013942680092951,
      "grad_norm": 2.070680618286133,
      "learning_rate": 3.4943981628442304e-06,
      "loss": 0.0459,
      "step": 11646
    },
    {
      "epoch": 0.6014459075651949,
      "grad_norm": 8.160440444946289,
      "learning_rate": 3.4936168520926163e-06,
      "loss": 0.0562,
      "step": 11647
    },
    {
      "epoch": 0.6014975471210947,
      "grad_norm": 3.8600010871887207,
      "learning_rate": 3.49283558179353e-06,
      "loss": 0.0456,
      "step": 11648
    },
    {
      "epoch": 0.6015491866769945,
      "grad_norm": 4.31606388092041,
      "learning_rate": 3.4920543519679517e-06,
      "loss": 0.0502,
      "step": 11649
    },
    {
      "epoch": 0.6016008262328943,
      "grad_norm": 10.177774429321289,
      "learning_rate": 3.4912731626368606e-06,
      "loss": 0.0744,
      "step": 11650
    },
    {
      "epoch": 0.6016524657887942,
      "grad_norm": 4.021449565887451,
      "learning_rate": 3.490492013821234e-06,
      "loss": 0.0888,
      "step": 11651
    },
    {
      "epoch": 0.601704105344694,
      "grad_norm": 20.06867218017578,
      "learning_rate": 3.4897109055420476e-06,
      "loss": 0.2744,
      "step": 11652
    },
    {
      "epoch": 0.6017557449005938,
      "grad_norm": 4.388668060302734,
      "learning_rate": 3.48892983782028e-06,
      "loss": 0.0408,
      "step": 11653
    },
    {
      "epoch": 0.6018073844564936,
      "grad_norm": 3.6569247245788574,
      "learning_rate": 3.4881488106769067e-06,
      "loss": 0.0883,
      "step": 11654
    },
    {
      "epoch": 0.6018590240123934,
      "grad_norm": 1.7793798446655273,
      "learning_rate": 3.487367824132898e-06,
      "loss": 0.0263,
      "step": 11655
    },
    {
      "epoch": 0.6019106635682933,
      "grad_norm": 7.182848930358887,
      "learning_rate": 3.486586878209229e-06,
      "loss": 0.0597,
      "step": 11656
    },
    {
      "epoch": 0.6019623031241931,
      "grad_norm": 1.9796205759048462,
      "learning_rate": 3.485805972926871e-06,
      "loss": 0.0317,
      "step": 11657
    },
    {
      "epoch": 0.602013942680093,
      "grad_norm": 3.212841510772705,
      "learning_rate": 3.4850251083067938e-06,
      "loss": 0.0449,
      "step": 11658
    },
    {
      "epoch": 0.6020655822359928,
      "grad_norm": 10.115202903747559,
      "learning_rate": 3.484244284369966e-06,
      "loss": 0.0656,
      "step": 11659
    },
    {
      "epoch": 0.6021172217918926,
      "grad_norm": 3.066527843475342,
      "learning_rate": 3.4834635011373572e-06,
      "loss": 0.0568,
      "step": 11660
    },
    {
      "epoch": 0.6021688613477925,
      "grad_norm": 9.631052017211914,
      "learning_rate": 3.4826827586299368e-06,
      "loss": 0.1254,
      "step": 11661
    },
    {
      "epoch": 0.6022205009036923,
      "grad_norm": 2.1975719928741455,
      "learning_rate": 3.481902056868667e-06,
      "loss": 0.0385,
      "step": 11662
    },
    {
      "epoch": 0.6022721404595921,
      "grad_norm": 12.686437606811523,
      "learning_rate": 3.481121395874514e-06,
      "loss": 0.1152,
      "step": 11663
    },
    {
      "epoch": 0.6023237800154919,
      "grad_norm": 5.2828874588012695,
      "learning_rate": 3.4803407756684417e-06,
      "loss": 0.0602,
      "step": 11664
    },
    {
      "epoch": 0.6023754195713917,
      "grad_norm": 2.491072654724121,
      "learning_rate": 3.4795601962714158e-06,
      "loss": 0.0482,
      "step": 11665
    },
    {
      "epoch": 0.6024270591272916,
      "grad_norm": 1.9719592332839966,
      "learning_rate": 3.4787796577043933e-06,
      "loss": 0.0258,
      "step": 11666
    },
    {
      "epoch": 0.6024786986831914,
      "grad_norm": 0.9179092645645142,
      "learning_rate": 3.477999159988338e-06,
      "loss": 0.0065,
      "step": 11667
    },
    {
      "epoch": 0.6025303382390912,
      "grad_norm": 8.741328239440918,
      "learning_rate": 3.4772187031442087e-06,
      "loss": 0.1622,
      "step": 11668
    },
    {
      "epoch": 0.602581977794991,
      "grad_norm": 9.505453109741211,
      "learning_rate": 3.476438287192964e-06,
      "loss": 0.1053,
      "step": 11669
    },
    {
      "epoch": 0.6026336173508908,
      "grad_norm": 2.6239981651306152,
      "learning_rate": 3.47565791215556e-06,
      "loss": 0.0272,
      "step": 11670
    },
    {
      "epoch": 0.6026852569067906,
      "grad_norm": 8.224369049072266,
      "learning_rate": 3.4748775780529533e-06,
      "loss": 0.0868,
      "step": 11671
    },
    {
      "epoch": 0.6027368964626905,
      "grad_norm": 1.3787134885787964,
      "learning_rate": 3.4740972849061027e-06,
      "loss": 0.0131,
      "step": 11672
    },
    {
      "epoch": 0.6027885360185903,
      "grad_norm": 4.548517227172852,
      "learning_rate": 3.4733170327359562e-06,
      "loss": 0.0869,
      "step": 11673
    },
    {
      "epoch": 0.6028401755744901,
      "grad_norm": 10.95710563659668,
      "learning_rate": 3.4725368215634707e-06,
      "loss": 0.0497,
      "step": 11674
    },
    {
      "epoch": 0.6028918151303899,
      "grad_norm": 18.0744686126709,
      "learning_rate": 3.471756651409598e-06,
      "loss": 0.0756,
      "step": 11675
    },
    {
      "epoch": 0.6029434546862897,
      "grad_norm": 4.609593868255615,
      "learning_rate": 3.470976522295287e-06,
      "loss": 0.0484,
      "step": 11676
    },
    {
      "epoch": 0.6029950942421896,
      "grad_norm": 2.127546548843384,
      "learning_rate": 3.470196434241489e-06,
      "loss": 0.0252,
      "step": 11677
    },
    {
      "epoch": 0.6030467337980894,
      "grad_norm": 5.928811073303223,
      "learning_rate": 3.4694163872691507e-06,
      "loss": 0.0825,
      "step": 11678
    },
    {
      "epoch": 0.6030983733539892,
      "grad_norm": 7.588301181793213,
      "learning_rate": 3.468636381399223e-06,
      "loss": 0.0883,
      "step": 11679
    },
    {
      "epoch": 0.603150012909889,
      "grad_norm": 2.8765673637390137,
      "learning_rate": 3.4678564166526475e-06,
      "loss": 0.0271,
      "step": 11680
    },
    {
      "epoch": 0.6032016524657888,
      "grad_norm": 8.21338939666748,
      "learning_rate": 3.467076493050373e-06,
      "loss": 0.06,
      "step": 11681
    },
    {
      "epoch": 0.6032532920216886,
      "grad_norm": 7.0065460205078125,
      "learning_rate": 3.4662966106133424e-06,
      "loss": 0.0732,
      "step": 11682
    },
    {
      "epoch": 0.6033049315775885,
      "grad_norm": 16.59554672241211,
      "learning_rate": 3.4655167693625e-06,
      "loss": 0.1073,
      "step": 11683
    },
    {
      "epoch": 0.6033565711334883,
      "grad_norm": 2.6628637313842773,
      "learning_rate": 3.464736969318785e-06,
      "loss": 0.0484,
      "step": 11684
    },
    {
      "epoch": 0.6034082106893881,
      "grad_norm": 3.317413091659546,
      "learning_rate": 3.4639572105031386e-06,
      "loss": 0.0429,
      "step": 11685
    },
    {
      "epoch": 0.6034598502452879,
      "grad_norm": 2.704880714416504,
      "learning_rate": 3.463177492936505e-06,
      "loss": 0.0536,
      "step": 11686
    },
    {
      "epoch": 0.6035114898011877,
      "grad_norm": 1.762701392173767,
      "learning_rate": 3.462397816639817e-06,
      "loss": 0.0278,
      "step": 11687
    },
    {
      "epoch": 0.6035631293570876,
      "grad_norm": 8.594497680664062,
      "learning_rate": 3.461618181634015e-06,
      "loss": 0.0624,
      "step": 11688
    },
    {
      "epoch": 0.6036147689129874,
      "grad_norm": 4.6382060050964355,
      "learning_rate": 3.4608385879400353e-06,
      "loss": 0.0412,
      "step": 11689
    },
    {
      "epoch": 0.6036664084688872,
      "grad_norm": 2.640287399291992,
      "learning_rate": 3.4600590355788135e-06,
      "loss": 0.0398,
      "step": 11690
    },
    {
      "epoch": 0.603718048024787,
      "grad_norm": 1.8508851528167725,
      "learning_rate": 3.4592795245712817e-06,
      "loss": 0.0165,
      "step": 11691
    },
    {
      "epoch": 0.6037696875806868,
      "grad_norm": 10.895705223083496,
      "learning_rate": 3.458500054938374e-06,
      "loss": 0.1028,
      "step": 11692
    },
    {
      "epoch": 0.6038213271365866,
      "grad_norm": 5.203815937042236,
      "learning_rate": 3.457720626701023e-06,
      "loss": 0.1192,
      "step": 11693
    },
    {
      "epoch": 0.6038729666924865,
      "grad_norm": 4.417368412017822,
      "learning_rate": 3.4569412398801605e-06,
      "loss": 0.0481,
      "step": 11694
    },
    {
      "epoch": 0.6039246062483863,
      "grad_norm": 7.818565845489502,
      "learning_rate": 3.4561618944967133e-06,
      "loss": 0.0904,
      "step": 11695
    },
    {
      "epoch": 0.6039762458042861,
      "grad_norm": 7.8960161209106445,
      "learning_rate": 3.4553825905716127e-06,
      "loss": 0.0808,
      "step": 11696
    },
    {
      "epoch": 0.6040278853601859,
      "grad_norm": 3.7234740257263184,
      "learning_rate": 3.454603328125785e-06,
      "loss": 0.0508,
      "step": 11697
    },
    {
      "epoch": 0.6040795249160857,
      "grad_norm": 2.5210726261138916,
      "learning_rate": 3.4538241071801565e-06,
      "loss": 0.0436,
      "step": 11698
    },
    {
      "epoch": 0.6041311644719856,
      "grad_norm": 4.033976078033447,
      "learning_rate": 3.453044927755651e-06,
      "loss": 0.074,
      "step": 11699
    },
    {
      "epoch": 0.6041828040278854,
      "grad_norm": 1.590124249458313,
      "learning_rate": 3.4522657898731955e-06,
      "loss": 0.0243,
      "step": 11700
    },
    {
      "epoch": 0.6042344435837852,
      "grad_norm": 3.0335965156555176,
      "learning_rate": 3.451486693553713e-06,
      "loss": 0.0667,
      "step": 11701
    },
    {
      "epoch": 0.604286083139685,
      "grad_norm": 5.145448207855225,
      "learning_rate": 3.4507076388181227e-06,
      "loss": 0.0359,
      "step": 11702
    },
    {
      "epoch": 0.6043377226955848,
      "grad_norm": 6.301932334899902,
      "learning_rate": 3.4499286256873467e-06,
      "loss": 0.0959,
      "step": 11703
    },
    {
      "epoch": 0.6043893622514847,
      "grad_norm": 5.026033878326416,
      "learning_rate": 3.449149654182304e-06,
      "loss": 0.0587,
      "step": 11704
    },
    {
      "epoch": 0.6044410018073845,
      "grad_norm": 2.7362687587738037,
      "learning_rate": 3.448370724323916e-06,
      "loss": 0.0346,
      "step": 11705
    },
    {
      "epoch": 0.6044926413632843,
      "grad_norm": 1.1646004915237427,
      "learning_rate": 3.4475918361330975e-06,
      "loss": 0.021,
      "step": 11706
    },
    {
      "epoch": 0.6045442809191841,
      "grad_norm": 1.259385347366333,
      "learning_rate": 3.446812989630765e-06,
      "loss": 0.016,
      "step": 11707
    },
    {
      "epoch": 0.6045959204750839,
      "grad_norm": 2.7097723484039307,
      "learning_rate": 3.4460341848378355e-06,
      "loss": 0.0315,
      "step": 11708
    },
    {
      "epoch": 0.6046475600309837,
      "grad_norm": 7.081597328186035,
      "learning_rate": 3.4452554217752205e-06,
      "loss": 0.0596,
      "step": 11709
    },
    {
      "epoch": 0.6046991995868836,
      "grad_norm": 2.8649227619171143,
      "learning_rate": 3.444476700463835e-06,
      "loss": 0.044,
      "step": 11710
    },
    {
      "epoch": 0.6047508391427834,
      "grad_norm": 3.1057779788970947,
      "learning_rate": 3.4436980209245875e-06,
      "loss": 0.0608,
      "step": 11711
    },
    {
      "epoch": 0.6048024786986832,
      "grad_norm": 4.410979270935059,
      "learning_rate": 3.4429193831783947e-06,
      "loss": 0.0557,
      "step": 11712
    },
    {
      "epoch": 0.604854118254583,
      "grad_norm": 9.751058578491211,
      "learning_rate": 3.4421407872461608e-06,
      "loss": 0.0938,
      "step": 11713
    },
    {
      "epoch": 0.6049057578104828,
      "grad_norm": 4.170066833496094,
      "learning_rate": 3.4413622331487973e-06,
      "loss": 0.0585,
      "step": 11714
    },
    {
      "epoch": 0.6049573973663827,
      "grad_norm": 4.229996681213379,
      "learning_rate": 3.4405837209072095e-06,
      "loss": 0.0525,
      "step": 11715
    },
    {
      "epoch": 0.6050090369222825,
      "grad_norm": 14.755990982055664,
      "learning_rate": 3.4398052505423063e-06,
      "loss": 0.1175,
      "step": 11716
    },
    {
      "epoch": 0.6050606764781823,
      "grad_norm": 5.239312648773193,
      "learning_rate": 3.4390268220749896e-06,
      "loss": 0.0781,
      "step": 11717
    },
    {
      "epoch": 0.6051123160340821,
      "grad_norm": 10.502850532531738,
      "learning_rate": 3.438248435526164e-06,
      "loss": 0.1046,
      "step": 11718
    },
    {
      "epoch": 0.6051639555899819,
      "grad_norm": 5.213775157928467,
      "learning_rate": 3.437470090916736e-06,
      "loss": 0.0696,
      "step": 11719
    },
    {
      "epoch": 0.6052155951458817,
      "grad_norm": 4.313795566558838,
      "learning_rate": 3.436691788267602e-06,
      "loss": 0.0994,
      "step": 11720
    },
    {
      "epoch": 0.6052672347017816,
      "grad_norm": 2.2954487800598145,
      "learning_rate": 3.4359135275996668e-06,
      "loss": 0.0332,
      "step": 11721
    },
    {
      "epoch": 0.6053188742576814,
      "grad_norm": 15.742526054382324,
      "learning_rate": 3.4351353089338273e-06,
      "loss": 0.0902,
      "step": 11722
    },
    {
      "epoch": 0.6053705138135812,
      "grad_norm": 6.722144603729248,
      "learning_rate": 3.4343571322909842e-06,
      "loss": 0.0458,
      "step": 11723
    },
    {
      "epoch": 0.605422153369481,
      "grad_norm": 1.7355279922485352,
      "learning_rate": 3.433578997692032e-06,
      "loss": 0.0267,
      "step": 11724
    },
    {
      "epoch": 0.6054737929253808,
      "grad_norm": 3.3168036937713623,
      "learning_rate": 3.4328009051578677e-06,
      "loss": 0.0434,
      "step": 11725
    },
    {
      "epoch": 0.6055254324812807,
      "grad_norm": 1.8943657875061035,
      "learning_rate": 3.432022854709387e-06,
      "loss": 0.0362,
      "step": 11726
    },
    {
      "epoch": 0.6055770720371805,
      "grad_norm": 17.432924270629883,
      "learning_rate": 3.4312448463674845e-06,
      "loss": 0.0681,
      "step": 11727
    },
    {
      "epoch": 0.6056287115930803,
      "grad_norm": 4.687534809112549,
      "learning_rate": 3.430466880153051e-06,
      "loss": 0.0295,
      "step": 11728
    },
    {
      "epoch": 0.6056803511489801,
      "grad_norm": 4.828396797180176,
      "learning_rate": 3.4296889560869786e-06,
      "loss": 0.0537,
      "step": 11729
    },
    {
      "epoch": 0.6057319907048799,
      "grad_norm": 2.9427249431610107,
      "learning_rate": 3.428911074190159e-06,
      "loss": 0.03,
      "step": 11730
    },
    {
      "epoch": 0.6057836302607797,
      "grad_norm": 14.114060401916504,
      "learning_rate": 3.4281332344834783e-06,
      "loss": 0.0617,
      "step": 11731
    },
    {
      "epoch": 0.6058352698166796,
      "grad_norm": 1.9413254261016846,
      "learning_rate": 3.427355436987828e-06,
      "loss": 0.0363,
      "step": 11732
    },
    {
      "epoch": 0.6058869093725794,
      "grad_norm": 10.284218788146973,
      "learning_rate": 3.4265776817240948e-06,
      "loss": 0.0654,
      "step": 11733
    },
    {
      "epoch": 0.6059385489284792,
      "grad_norm": 8.928373336791992,
      "learning_rate": 3.425799968713164e-06,
      "loss": 0.1254,
      "step": 11734
    },
    {
      "epoch": 0.605990188484379,
      "grad_norm": 1.7256157398223877,
      "learning_rate": 3.4250222979759185e-06,
      "loss": 0.0233,
      "step": 11735
    },
    {
      "epoch": 0.6060418280402788,
      "grad_norm": 2.2483901977539062,
      "learning_rate": 3.4242446695332444e-06,
      "loss": 0.0455,
      "step": 11736
    },
    {
      "epoch": 0.6060934675961787,
      "grad_norm": 4.413260459899902,
      "learning_rate": 3.423467083406022e-06,
      "loss": 0.0621,
      "step": 11737
    },
    {
      "epoch": 0.6061451071520785,
      "grad_norm": 1.8059773445129395,
      "learning_rate": 3.422689539615136e-06,
      "loss": 0.027,
      "step": 11738
    },
    {
      "epoch": 0.6061967467079783,
      "grad_norm": 6.556427955627441,
      "learning_rate": 3.421912038181464e-06,
      "loss": 0.0555,
      "step": 11739
    },
    {
      "epoch": 0.6062483862638781,
      "grad_norm": 5.037285804748535,
      "learning_rate": 3.4211345791258855e-06,
      "loss": 0.1275,
      "step": 11740
    },
    {
      "epoch": 0.6063000258197779,
      "grad_norm": 4.2907586097717285,
      "learning_rate": 3.4203571624692793e-06,
      "loss": 0.0827,
      "step": 11741
    },
    {
      "epoch": 0.6063516653756778,
      "grad_norm": 5.044006824493408,
      "learning_rate": 3.419579788232521e-06,
      "loss": 0.067,
      "step": 11742
    },
    {
      "epoch": 0.6064033049315776,
      "grad_norm": 2.242168426513672,
      "learning_rate": 3.4188024564364865e-06,
      "loss": 0.0259,
      "step": 11743
    },
    {
      "epoch": 0.6064549444874774,
      "grad_norm": 4.115336894989014,
      "learning_rate": 3.4180251671020497e-06,
      "loss": 0.1008,
      "step": 11744
    },
    {
      "epoch": 0.6065065840433772,
      "grad_norm": 2.0630321502685547,
      "learning_rate": 3.4172479202500876e-06,
      "loss": 0.0287,
      "step": 11745
    },
    {
      "epoch": 0.606558223599277,
      "grad_norm": 3.4904561042785645,
      "learning_rate": 3.4164707159014675e-06,
      "loss": 0.031,
      "step": 11746
    },
    {
      "epoch": 0.6066098631551768,
      "grad_norm": 2.3651416301727295,
      "learning_rate": 3.415693554077064e-06,
      "loss": 0.043,
      "step": 11747
    },
    {
      "epoch": 0.6066615027110767,
      "grad_norm": 4.825507640838623,
      "learning_rate": 3.4149164347977448e-06,
      "loss": 0.0701,
      "step": 11748
    },
    {
      "epoch": 0.6067131422669765,
      "grad_norm": 2.249844551086426,
      "learning_rate": 3.4141393580843817e-06,
      "loss": 0.0477,
      "step": 11749
    },
    {
      "epoch": 0.6067647818228763,
      "grad_norm": 32.223716735839844,
      "learning_rate": 3.4133623239578384e-06,
      "loss": 0.0538,
      "step": 11750
    },
    {
      "epoch": 0.6068164213787761,
      "grad_norm": 7.561460494995117,
      "learning_rate": 3.4125853324389836e-06,
      "loss": 0.0449,
      "step": 11751
    },
    {
      "epoch": 0.6068680609346759,
      "grad_norm": 9.649168968200684,
      "learning_rate": 3.4118083835486848e-06,
      "loss": 0.0983,
      "step": 11752
    },
    {
      "epoch": 0.6069197004905758,
      "grad_norm": 7.186933517456055,
      "learning_rate": 3.411031477307802e-06,
      "loss": 0.0561,
      "step": 11753
    },
    {
      "epoch": 0.6069713400464756,
      "grad_norm": 7.59083366394043,
      "learning_rate": 3.4102546137372005e-06,
      "loss": 0.0615,
      "step": 11754
    },
    {
      "epoch": 0.6070229796023754,
      "grad_norm": 3.689035177230835,
      "learning_rate": 3.409477792857742e-06,
      "loss": 0.0365,
      "step": 11755
    },
    {
      "epoch": 0.6070746191582752,
      "grad_norm": 5.857674598693848,
      "learning_rate": 3.408701014690289e-06,
      "loss": 0.0751,
      "step": 11756
    },
    {
      "epoch": 0.607126258714175,
      "grad_norm": 4.144607067108154,
      "learning_rate": 3.4079242792556965e-06,
      "loss": 0.0359,
      "step": 11757
    },
    {
      "epoch": 0.6071778982700748,
      "grad_norm": 3.3206050395965576,
      "learning_rate": 3.4071475865748283e-06,
      "loss": 0.0506,
      "step": 11758
    },
    {
      "epoch": 0.6072295378259747,
      "grad_norm": 18.782258987426758,
      "learning_rate": 3.4063709366685387e-06,
      "loss": 0.1355,
      "step": 11759
    },
    {
      "epoch": 0.6072811773818745,
      "grad_norm": 15.368948936462402,
      "learning_rate": 3.4055943295576858e-06,
      "loss": 0.0747,
      "step": 11760
    },
    {
      "epoch": 0.6073328169377743,
      "grad_norm": 9.947954177856445,
      "learning_rate": 3.404817765263123e-06,
      "loss": 0.1599,
      "step": 11761
    },
    {
      "epoch": 0.6073844564936741,
      "grad_norm": 3.255465030670166,
      "learning_rate": 3.4040412438057047e-06,
      "loss": 0.0631,
      "step": 11762
    },
    {
      "epoch": 0.6074360960495739,
      "grad_norm": 17.857942581176758,
      "learning_rate": 3.4032647652062844e-06,
      "loss": 0.1409,
      "step": 11763
    },
    {
      "epoch": 0.6074877356054738,
      "grad_norm": 1.5478488206863403,
      "learning_rate": 3.4024883294857107e-06,
      "loss": 0.0285,
      "step": 11764
    },
    {
      "epoch": 0.6075393751613736,
      "grad_norm": 3.7282114028930664,
      "learning_rate": 3.4017119366648384e-06,
      "loss": 0.0526,
      "step": 11765
    },
    {
      "epoch": 0.6075910147172734,
      "grad_norm": 4.908134937286377,
      "learning_rate": 3.4009355867645146e-06,
      "loss": 0.0343,
      "step": 11766
    },
    {
      "epoch": 0.6076426542731732,
      "grad_norm": 3.3545658588409424,
      "learning_rate": 3.4001592798055883e-06,
      "loss": 0.0375,
      "step": 11767
    },
    {
      "epoch": 0.607694293829073,
      "grad_norm": 2.871894598007202,
      "learning_rate": 3.3993830158089054e-06,
      "loss": 0.0392,
      "step": 11768
    },
    {
      "epoch": 0.6077459333849728,
      "grad_norm": 2.5072426795959473,
      "learning_rate": 3.3986067947953117e-06,
      "loss": 0.0333,
      "step": 11769
    },
    {
      "epoch": 0.6077975729408727,
      "grad_norm": 1.6293385028839111,
      "learning_rate": 3.3978306167856523e-06,
      "loss": 0.0156,
      "step": 11770
    },
    {
      "epoch": 0.6078492124967725,
      "grad_norm": 4.251851558685303,
      "learning_rate": 3.3970544818007722e-06,
      "loss": 0.0597,
      "step": 11771
    },
    {
      "epoch": 0.6079008520526723,
      "grad_norm": 2.454683303833008,
      "learning_rate": 3.3962783898615115e-06,
      "loss": 0.0306,
      "step": 11772
    },
    {
      "epoch": 0.6079524916085721,
      "grad_norm": 3.6161575317382812,
      "learning_rate": 3.395502340988713e-06,
      "loss": 0.0995,
      "step": 11773
    },
    {
      "epoch": 0.6080041311644719,
      "grad_norm": 4.9303436279296875,
      "learning_rate": 3.3947263352032165e-06,
      "loss": 0.0666,
      "step": 11774
    },
    {
      "epoch": 0.6080557707203718,
      "grad_norm": 2.1109673976898193,
      "learning_rate": 3.3939503725258605e-06,
      "loss": 0.0311,
      "step": 11775
    },
    {
      "epoch": 0.6081074102762716,
      "grad_norm": 1.5712193250656128,
      "learning_rate": 3.393174452977481e-06,
      "loss": 0.0236,
      "step": 11776
    },
    {
      "epoch": 0.6081590498321714,
      "grad_norm": 4.067266464233398,
      "learning_rate": 3.3923985765789185e-06,
      "loss": 0.0618,
      "step": 11777
    },
    {
      "epoch": 0.6082106893880712,
      "grad_norm": 13.619823455810547,
      "learning_rate": 3.391622743351007e-06,
      "loss": 0.0684,
      "step": 11778
    },
    {
      "epoch": 0.608262328943971,
      "grad_norm": 2.9928903579711914,
      "learning_rate": 3.3908469533145793e-06,
      "loss": 0.0467,
      "step": 11779
    },
    {
      "epoch": 0.6083139684998708,
      "grad_norm": 3.292694330215454,
      "learning_rate": 3.390071206490469e-06,
      "loss": 0.0414,
      "step": 11780
    },
    {
      "epoch": 0.6083656080557707,
      "grad_norm": 2.658207654953003,
      "learning_rate": 3.389295502899509e-06,
      "loss": 0.035,
      "step": 11781
    },
    {
      "epoch": 0.6084172476116705,
      "grad_norm": 5.254611015319824,
      "learning_rate": 3.38851984256253e-06,
      "loss": 0.0215,
      "step": 11782
    },
    {
      "epoch": 0.6084688871675703,
      "grad_norm": 2.971296548843384,
      "learning_rate": 3.38774422550036e-06,
      "loss": 0.0521,
      "step": 11783
    },
    {
      "epoch": 0.6085205267234701,
      "grad_norm": 7.101728916168213,
      "learning_rate": 3.3869686517338298e-06,
      "loss": 0.0964,
      "step": 11784
    },
    {
      "epoch": 0.60857216627937,
      "grad_norm": 20.269628524780273,
      "learning_rate": 3.3861931212837656e-06,
      "loss": 0.2033,
      "step": 11785
    },
    {
      "epoch": 0.6086238058352699,
      "grad_norm": 1.8764469623565674,
      "learning_rate": 3.385417634170993e-06,
      "loss": 0.0285,
      "step": 11786
    },
    {
      "epoch": 0.6086754453911697,
      "grad_norm": 4.1144280433654785,
      "learning_rate": 3.3846421904163384e-06,
      "loss": 0.0322,
      "step": 11787
    },
    {
      "epoch": 0.6087270849470695,
      "grad_norm": 7.339650630950928,
      "learning_rate": 3.383866790040624e-06,
      "loss": 0.1222,
      "step": 11788
    },
    {
      "epoch": 0.6087787245029693,
      "grad_norm": 6.410027503967285,
      "learning_rate": 3.3830914330646748e-06,
      "loss": 0.0721,
      "step": 11789
    },
    {
      "epoch": 0.6088303640588691,
      "grad_norm": 9.736802101135254,
      "learning_rate": 3.3823161195093085e-06,
      "loss": 0.1262,
      "step": 11790
    },
    {
      "epoch": 0.608882003614769,
      "grad_norm": 2.4092276096343994,
      "learning_rate": 3.3815408493953484e-06,
      "loss": 0.0317,
      "step": 11791
    },
    {
      "epoch": 0.6089336431706688,
      "grad_norm": 4.64654541015625,
      "learning_rate": 3.3807656227436135e-06,
      "loss": 0.031,
      "step": 11792
    },
    {
      "epoch": 0.6089852827265686,
      "grad_norm": 6.326040744781494,
      "learning_rate": 3.3799904395749223e-06,
      "loss": 0.0765,
      "step": 11793
    },
    {
      "epoch": 0.6090369222824684,
      "grad_norm": 13.866130828857422,
      "learning_rate": 3.379215299910089e-06,
      "loss": 0.0952,
      "step": 11794
    },
    {
      "epoch": 0.6090885618383682,
      "grad_norm": 4.127470970153809,
      "learning_rate": 3.3784402037699305e-06,
      "loss": 0.0605,
      "step": 11795
    },
    {
      "epoch": 0.609140201394268,
      "grad_norm": 4.719638347625732,
      "learning_rate": 3.3776651511752644e-06,
      "loss": 0.0664,
      "step": 11796
    },
    {
      "epoch": 0.6091918409501679,
      "grad_norm": 3.885575771331787,
      "learning_rate": 3.3768901421468982e-06,
      "loss": 0.0542,
      "step": 11797
    },
    {
      "epoch": 0.6092434805060677,
      "grad_norm": 1.5261183977127075,
      "learning_rate": 3.3761151767056476e-06,
      "loss": 0.0274,
      "step": 11798
    },
    {
      "epoch": 0.6092951200619675,
      "grad_norm": 3.912205457687378,
      "learning_rate": 3.375340254872324e-06,
      "loss": 0.0249,
      "step": 11799
    },
    {
      "epoch": 0.6093467596178673,
      "grad_norm": 2.0951156616210938,
      "learning_rate": 3.374565376667737e-06,
      "loss": 0.0314,
      "step": 11800
    },
    {
      "epoch": 0.6093983991737671,
      "grad_norm": 3.666445016860962,
      "learning_rate": 3.373790542112694e-06,
      "loss": 0.0583,
      "step": 11801
    },
    {
      "epoch": 0.609450038729667,
      "grad_norm": 6.978396415710449,
      "learning_rate": 3.373015751228001e-06,
      "loss": 0.0412,
      "step": 11802
    },
    {
      "epoch": 0.6095016782855668,
      "grad_norm": 9.823441505432129,
      "learning_rate": 3.3722410040344676e-06,
      "loss": 0.0594,
      "step": 11803
    },
    {
      "epoch": 0.6095533178414666,
      "grad_norm": 5.031275272369385,
      "learning_rate": 3.371466300552899e-06,
      "loss": 0.0586,
      "step": 11804
    },
    {
      "epoch": 0.6096049573973664,
      "grad_norm": 7.941634178161621,
      "learning_rate": 3.3706916408040968e-06,
      "loss": 0.114,
      "step": 11805
    },
    {
      "epoch": 0.6096565969532662,
      "grad_norm": 3.332585096359253,
      "learning_rate": 3.369917024808864e-06,
      "loss": 0.0323,
      "step": 11806
    },
    {
      "epoch": 0.6097082365091661,
      "grad_norm": 4.680325508117676,
      "learning_rate": 3.3691424525880047e-06,
      "loss": 0.0581,
      "step": 11807
    },
    {
      "epoch": 0.6097598760650659,
      "grad_norm": 5.375843048095703,
      "learning_rate": 3.3683679241623164e-06,
      "loss": 0.067,
      "step": 11808
    },
    {
      "epoch": 0.6098115156209657,
      "grad_norm": 7.919137477874756,
      "learning_rate": 3.367593439552598e-06,
      "loss": 0.0398,
      "step": 11809
    },
    {
      "epoch": 0.6098631551768655,
      "grad_norm": 44.33608627319336,
      "learning_rate": 3.366818998779651e-06,
      "loss": 0.1321,
      "step": 11810
    },
    {
      "epoch": 0.6099147947327653,
      "grad_norm": 5.193764686584473,
      "learning_rate": 3.3660446018642704e-06,
      "loss": 0.0631,
      "step": 11811
    },
    {
      "epoch": 0.6099664342886651,
      "grad_norm": 2.230181932449341,
      "learning_rate": 3.3652702488272516e-06,
      "loss": 0.0147,
      "step": 11812
    },
    {
      "epoch": 0.610018073844565,
      "grad_norm": 22.942659378051758,
      "learning_rate": 3.364495939689389e-06,
      "loss": 0.2895,
      "step": 11813
    },
    {
      "epoch": 0.6100697134004648,
      "grad_norm": 2.2510437965393066,
      "learning_rate": 3.363721674471475e-06,
      "loss": 0.0383,
      "step": 11814
    },
    {
      "epoch": 0.6101213529563646,
      "grad_norm": 5.569010257720947,
      "learning_rate": 3.3629474531943073e-06,
      "loss": 0.0724,
      "step": 11815
    },
    {
      "epoch": 0.6101729925122644,
      "grad_norm": 6.004213333129883,
      "learning_rate": 3.3621732758786687e-06,
      "loss": 0.0243,
      "step": 11816
    },
    {
      "epoch": 0.6102246320681642,
      "grad_norm": 3.0408103466033936,
      "learning_rate": 3.361399142545354e-06,
      "loss": 0.0476,
      "step": 11817
    },
    {
      "epoch": 0.6102762716240641,
      "grad_norm": 2.763394594192505,
      "learning_rate": 3.360625053215153e-06,
      "loss": 0.0325,
      "step": 11818
    },
    {
      "epoch": 0.6103279111799639,
      "grad_norm": 5.790143013000488,
      "learning_rate": 3.359851007908849e-06,
      "loss": 0.06,
      "step": 11819
    },
    {
      "epoch": 0.6103795507358637,
      "grad_norm": 6.287694454193115,
      "learning_rate": 3.359077006647231e-06,
      "loss": 0.0361,
      "step": 11820
    },
    {
      "epoch": 0.6104311902917635,
      "grad_norm": 14.051570892333984,
      "learning_rate": 3.3583030494510815e-06,
      "loss": 0.0607,
      "step": 11821
    },
    {
      "epoch": 0.6104828298476633,
      "grad_norm": 3.922025680541992,
      "learning_rate": 3.35752913634119e-06,
      "loss": 0.0605,
      "step": 11822
    },
    {
      "epoch": 0.6105344694035632,
      "grad_norm": 5.320842266082764,
      "learning_rate": 3.3567552673383323e-06,
      "loss": 0.0357,
      "step": 11823
    },
    {
      "epoch": 0.610586108959463,
      "grad_norm": 3.458550453186035,
      "learning_rate": 3.3559814424632944e-06,
      "loss": 0.0455,
      "step": 11824
    },
    {
      "epoch": 0.6106377485153628,
      "grad_norm": 4.792598247528076,
      "learning_rate": 3.355207661736856e-06,
      "loss": 0.0737,
      "step": 11825
    },
    {
      "epoch": 0.6106893880712626,
      "grad_norm": 2.9417078495025635,
      "learning_rate": 3.354433925179795e-06,
      "loss": 0.042,
      "step": 11826
    },
    {
      "epoch": 0.6107410276271624,
      "grad_norm": 2.4911105632781982,
      "learning_rate": 3.3536602328128902e-06,
      "loss": 0.0359,
      "step": 11827
    },
    {
      "epoch": 0.6107926671830622,
      "grad_norm": 10.8975191116333,
      "learning_rate": 3.352886584656917e-06,
      "loss": 0.145,
      "step": 11828
    },
    {
      "epoch": 0.6108443067389621,
      "grad_norm": 15.040268898010254,
      "learning_rate": 3.3521129807326557e-06,
      "loss": 0.0354,
      "step": 11829
    },
    {
      "epoch": 0.6108959462948619,
      "grad_norm": 1.634170651435852,
      "learning_rate": 3.351339421060874e-06,
      "loss": 0.0229,
      "step": 11830
    },
    {
      "epoch": 0.6109475858507617,
      "grad_norm": 2.985563278198242,
      "learning_rate": 3.3505659056623497e-06,
      "loss": 0.0282,
      "step": 11831
    },
    {
      "epoch": 0.6109992254066615,
      "grad_norm": 3.296308994293213,
      "learning_rate": 3.3497924345578525e-06,
      "loss": 0.0471,
      "step": 11832
    },
    {
      "epoch": 0.6110508649625613,
      "grad_norm": 12.237857818603516,
      "learning_rate": 3.349019007768156e-06,
      "loss": 0.1623,
      "step": 11833
    },
    {
      "epoch": 0.6111025045184612,
      "grad_norm": 2.102043867111206,
      "learning_rate": 3.348245625314027e-06,
      "loss": 0.0181,
      "step": 11834
    },
    {
      "epoch": 0.611154144074361,
      "grad_norm": 3.0246200561523438,
      "learning_rate": 3.3474722872162345e-06,
      "loss": 0.0394,
      "step": 11835
    },
    {
      "epoch": 0.6112057836302608,
      "grad_norm": 3.0370535850524902,
      "learning_rate": 3.3466989934955474e-06,
      "loss": 0.0575,
      "step": 11836
    },
    {
      "epoch": 0.6112574231861606,
      "grad_norm": 12.454582214355469,
      "learning_rate": 3.345925744172732e-06,
      "loss": 0.145,
      "step": 11837
    },
    {
      "epoch": 0.6113090627420604,
      "grad_norm": 2.894158363342285,
      "learning_rate": 3.34515253926855e-06,
      "loss": 0.0537,
      "step": 11838
    },
    {
      "epoch": 0.6113607022979602,
      "grad_norm": 1.5911412239074707,
      "learning_rate": 3.3443793788037683e-06,
      "loss": 0.0221,
      "step": 11839
    },
    {
      "epoch": 0.6114123418538601,
      "grad_norm": 1.92158842086792,
      "learning_rate": 3.343606262799148e-06,
      "loss": 0.025,
      "step": 11840
    },
    {
      "epoch": 0.6114639814097599,
      "grad_norm": 6.7991743087768555,
      "learning_rate": 3.3428331912754507e-06,
      "loss": 0.0843,
      "step": 11841
    },
    {
      "epoch": 0.6115156209656597,
      "grad_norm": 9.01163101196289,
      "learning_rate": 3.342060164253435e-06,
      "loss": 0.0429,
      "step": 11842
    },
    {
      "epoch": 0.6115672605215595,
      "grad_norm": 4.970338344573975,
      "learning_rate": 3.3412871817538628e-06,
      "loss": 0.0486,
      "step": 11843
    },
    {
      "epoch": 0.6116189000774593,
      "grad_norm": 4.637857437133789,
      "learning_rate": 3.34051424379749e-06,
      "loss": 0.1029,
      "step": 11844
    },
    {
      "epoch": 0.6116705396333592,
      "grad_norm": 5.732492923736572,
      "learning_rate": 3.3397413504050734e-06,
      "loss": 0.0416,
      "step": 11845
    },
    {
      "epoch": 0.611722179189259,
      "grad_norm": 3.854210138320923,
      "learning_rate": 3.3389685015973684e-06,
      "loss": 0.0456,
      "step": 11846
    },
    {
      "epoch": 0.6117738187451588,
      "grad_norm": 7.113239765167236,
      "learning_rate": 3.3381956973951278e-06,
      "loss": 0.0835,
      "step": 11847
    },
    {
      "epoch": 0.6118254583010586,
      "grad_norm": 4.427857398986816,
      "learning_rate": 3.3374229378191093e-06,
      "loss": 0.0756,
      "step": 11848
    },
    {
      "epoch": 0.6118770978569584,
      "grad_norm": 2.580638885498047,
      "learning_rate": 3.3366502228900582e-06,
      "loss": 0.0493,
      "step": 11849
    },
    {
      "epoch": 0.6119287374128582,
      "grad_norm": 2.9497292041778564,
      "learning_rate": 3.335877552628729e-06,
      "loss": 0.0565,
      "step": 11850
    },
    {
      "epoch": 0.6119803769687581,
      "grad_norm": 13.269761085510254,
      "learning_rate": 3.335104927055871e-06,
      "loss": 0.1017,
      "step": 11851
    },
    {
      "epoch": 0.6120320165246579,
      "grad_norm": 5.142080783843994,
      "learning_rate": 3.33433234619223e-06,
      "loss": 0.0499,
      "step": 11852
    },
    {
      "epoch": 0.6120836560805577,
      "grad_norm": 6.591442108154297,
      "learning_rate": 3.333559810058555e-06,
      "loss": 0.0627,
      "step": 11853
    },
    {
      "epoch": 0.6121352956364575,
      "grad_norm": 2.2240607738494873,
      "learning_rate": 3.3327873186755897e-06,
      "loss": 0.0361,
      "step": 11854
    },
    {
      "epoch": 0.6121869351923573,
      "grad_norm": 1.9858970642089844,
      "learning_rate": 3.3320148720640833e-06,
      "loss": 0.023,
      "step": 11855
    },
    {
      "epoch": 0.6122385747482572,
      "grad_norm": 3.53617000579834,
      "learning_rate": 3.331242470244773e-06,
      "loss": 0.0483,
      "step": 11856
    },
    {
      "epoch": 0.612290214304157,
      "grad_norm": 15.101688385009766,
      "learning_rate": 3.3304701132384042e-06,
      "loss": 0.0971,
      "step": 11857
    },
    {
      "epoch": 0.6123418538600568,
      "grad_norm": 9.658061027526855,
      "learning_rate": 3.3296978010657184e-06,
      "loss": 0.1077,
      "step": 11858
    },
    {
      "epoch": 0.6123934934159566,
      "grad_norm": 2.979865074157715,
      "learning_rate": 3.3289255337474547e-06,
      "loss": 0.0288,
      "step": 11859
    },
    {
      "epoch": 0.6124451329718564,
      "grad_norm": 10.20047378540039,
      "learning_rate": 3.3281533113043507e-06,
      "loss": 0.1976,
      "step": 11860
    },
    {
      "epoch": 0.6124967725277563,
      "grad_norm": 3.531428575515747,
      "learning_rate": 3.327381133757144e-06,
      "loss": 0.0859,
      "step": 11861
    },
    {
      "epoch": 0.6125484120836561,
      "grad_norm": 3.0378613471984863,
      "learning_rate": 3.326609001126573e-06,
      "loss": 0.0542,
      "step": 11862
    },
    {
      "epoch": 0.6126000516395559,
      "grad_norm": 3.0671982765197754,
      "learning_rate": 3.325836913433368e-06,
      "loss": 0.0431,
      "step": 11863
    },
    {
      "epoch": 0.6126516911954557,
      "grad_norm": 5.024641513824463,
      "learning_rate": 3.325064870698267e-06,
      "loss": 0.071,
      "step": 11864
    },
    {
      "epoch": 0.6127033307513555,
      "grad_norm": 3.1081042289733887,
      "learning_rate": 3.324292872942e-06,
      "loss": 0.0462,
      "step": 11865
    },
    {
      "epoch": 0.6127549703072553,
      "grad_norm": 1.7327247858047485,
      "learning_rate": 3.3235209201853004e-06,
      "loss": 0.0186,
      "step": 11866
    },
    {
      "epoch": 0.6128066098631552,
      "grad_norm": 7.6099653244018555,
      "learning_rate": 3.322749012448896e-06,
      "loss": 0.0555,
      "step": 11867
    },
    {
      "epoch": 0.612858249419055,
      "grad_norm": 4.361402988433838,
      "learning_rate": 3.321977149753516e-06,
      "loss": 0.0677,
      "step": 11868
    },
    {
      "epoch": 0.6129098889749548,
      "grad_norm": 3.9738101959228516,
      "learning_rate": 3.3212053321198902e-06,
      "loss": 0.1167,
      "step": 11869
    },
    {
      "epoch": 0.6129615285308546,
      "grad_norm": 16.986265182495117,
      "learning_rate": 3.3204335595687435e-06,
      "loss": 0.074,
      "step": 11870
    },
    {
      "epoch": 0.6130131680867544,
      "grad_norm": 8.612709045410156,
      "learning_rate": 3.3196618321208007e-06,
      "loss": 0.0266,
      "step": 11871
    },
    {
      "epoch": 0.6130648076426543,
      "grad_norm": 6.338871002197266,
      "learning_rate": 3.318890149796787e-06,
      "loss": 0.0531,
      "step": 11872
    },
    {
      "epoch": 0.6131164471985541,
      "grad_norm": 4.130552768707275,
      "learning_rate": 3.3181185126174255e-06,
      "loss": 0.0757,
      "step": 11873
    },
    {
      "epoch": 0.6131680867544539,
      "grad_norm": 4.84971809387207,
      "learning_rate": 3.3173469206034347e-06,
      "loss": 0.0562,
      "step": 11874
    },
    {
      "epoch": 0.6132197263103537,
      "grad_norm": 3.433310031890869,
      "learning_rate": 3.3165753737755375e-06,
      "loss": 0.0365,
      "step": 11875
    },
    {
      "epoch": 0.6132713658662535,
      "grad_norm": 1.698527455329895,
      "learning_rate": 3.315803872154453e-06,
      "loss": 0.0161,
      "step": 11876
    },
    {
      "epoch": 0.6133230054221533,
      "grad_norm": 1.7401736974716187,
      "learning_rate": 3.3150324157609003e-06,
      "loss": 0.0306,
      "step": 11877
    },
    {
      "epoch": 0.6133746449780532,
      "grad_norm": 1.9231849908828735,
      "learning_rate": 3.3142610046155937e-06,
      "loss": 0.0301,
      "step": 11878
    },
    {
      "epoch": 0.613426284533953,
      "grad_norm": 4.160882949829102,
      "learning_rate": 3.31348963873925e-06,
      "loss": 0.0993,
      "step": 11879
    },
    {
      "epoch": 0.6134779240898528,
      "grad_norm": 5.331837177276611,
      "learning_rate": 3.312718318152582e-06,
      "loss": 0.0629,
      "step": 11880
    },
    {
      "epoch": 0.6135295636457526,
      "grad_norm": 3.1969802379608154,
      "learning_rate": 3.311947042876308e-06,
      "loss": 0.0264,
      "step": 11881
    },
    {
      "epoch": 0.6135812032016524,
      "grad_norm": 7.772464275360107,
      "learning_rate": 3.3111758129311315e-06,
      "loss": 0.1207,
      "step": 11882
    },
    {
      "epoch": 0.6136328427575523,
      "grad_norm": 2.489821672439575,
      "learning_rate": 3.31040462833777e-06,
      "loss": 0.058,
      "step": 11883
    },
    {
      "epoch": 0.6136844823134521,
      "grad_norm": 3.1574900150299072,
      "learning_rate": 3.309633489116931e-06,
      "loss": 0.0481,
      "step": 11884
    },
    {
      "epoch": 0.6137361218693519,
      "grad_norm": 6.040851593017578,
      "learning_rate": 3.3088623952893217e-06,
      "loss": 0.0926,
      "step": 11885
    },
    {
      "epoch": 0.6137877614252517,
      "grad_norm": 11.132185935974121,
      "learning_rate": 3.3080913468756494e-06,
      "loss": 0.0687,
      "step": 11886
    },
    {
      "epoch": 0.6138394009811515,
      "grad_norm": 3.8681440353393555,
      "learning_rate": 3.3073203438966194e-06,
      "loss": 0.0482,
      "step": 11887
    },
    {
      "epoch": 0.6138910405370513,
      "grad_norm": 6.762217044830322,
      "learning_rate": 3.3065493863729404e-06,
      "loss": 0.0749,
      "step": 11888
    },
    {
      "epoch": 0.6139426800929512,
      "grad_norm": 5.129919528961182,
      "learning_rate": 3.3057784743253096e-06,
      "loss": 0.0989,
      "step": 11889
    },
    {
      "epoch": 0.613994319648851,
      "grad_norm": 14.215719223022461,
      "learning_rate": 3.3050076077744324e-06,
      "loss": 0.0709,
      "step": 11890
    },
    {
      "epoch": 0.6140459592047508,
      "grad_norm": 2.377804756164551,
      "learning_rate": 3.3042367867410097e-06,
      "loss": 0.0273,
      "step": 11891
    },
    {
      "epoch": 0.6140975987606506,
      "grad_norm": 3.8813581466674805,
      "learning_rate": 3.303466011245742e-06,
      "loss": 0.0635,
      "step": 11892
    },
    {
      "epoch": 0.6141492383165504,
      "grad_norm": 2.2318859100341797,
      "learning_rate": 3.302695281309325e-06,
      "loss": 0.0269,
      "step": 11893
    },
    {
      "epoch": 0.6142008778724503,
      "grad_norm": 4.676445484161377,
      "learning_rate": 3.301924596952457e-06,
      "loss": 0.0404,
      "step": 11894
    },
    {
      "epoch": 0.6142525174283501,
      "grad_norm": 2.465423107147217,
      "learning_rate": 3.3011539581958373e-06,
      "loss": 0.0384,
      "step": 11895
    },
    {
      "epoch": 0.6143041569842499,
      "grad_norm": 2.7675225734710693,
      "learning_rate": 3.3003833650601557e-06,
      "loss": 0.0377,
      "step": 11896
    },
    {
      "epoch": 0.6143557965401497,
      "grad_norm": 5.488786697387695,
      "learning_rate": 3.299612817566108e-06,
      "loss": 0.0304,
      "step": 11897
    },
    {
      "epoch": 0.6144074360960495,
      "grad_norm": 7.867591857910156,
      "learning_rate": 3.2988423157343873e-06,
      "loss": 0.0961,
      "step": 11898
    },
    {
      "epoch": 0.6144590756519493,
      "grad_norm": 2.0589230060577393,
      "learning_rate": 3.2980718595856842e-06,
      "loss": 0.0356,
      "step": 11899
    },
    {
      "epoch": 0.6145107152078492,
      "grad_norm": 6.660339832305908,
      "learning_rate": 3.2973014491406886e-06,
      "loss": 0.069,
      "step": 11900
    },
    {
      "epoch": 0.614562354763749,
      "grad_norm": 2.243924617767334,
      "learning_rate": 3.2965310844200866e-06,
      "loss": 0.0392,
      "step": 11901
    },
    {
      "epoch": 0.6146139943196488,
      "grad_norm": 9.279814720153809,
      "learning_rate": 3.2957607654445696e-06,
      "loss": 0.1533,
      "step": 11902
    },
    {
      "epoch": 0.6146656338755486,
      "grad_norm": 3.894543170928955,
      "learning_rate": 3.294990492234823e-06,
      "loss": 0.0664,
      "step": 11903
    },
    {
      "epoch": 0.6147172734314484,
      "grad_norm": 11.984320640563965,
      "learning_rate": 3.2942202648115295e-06,
      "loss": 0.0454,
      "step": 11904
    },
    {
      "epoch": 0.6147689129873483,
      "grad_norm": 8.746020317077637,
      "learning_rate": 3.2934500831953747e-06,
      "loss": 0.0688,
      "step": 11905
    },
    {
      "epoch": 0.6148205525432481,
      "grad_norm": 2.0131189823150635,
      "learning_rate": 3.2926799474070413e-06,
      "loss": 0.0164,
      "step": 11906
    },
    {
      "epoch": 0.6148721920991479,
      "grad_norm": 6.0524067878723145,
      "learning_rate": 3.2919098574672097e-06,
      "loss": 0.0491,
      "step": 11907
    },
    {
      "epoch": 0.6149238316550477,
      "grad_norm": 3.023470640182495,
      "learning_rate": 3.291139813396559e-06,
      "loss": 0.0446,
      "step": 11908
    },
    {
      "epoch": 0.6149754712109475,
      "grad_norm": 2.229295253753662,
      "learning_rate": 3.29036981521577e-06,
      "loss": 0.0257,
      "step": 11909
    },
    {
      "epoch": 0.6150271107668474,
      "grad_norm": 4.6873393058776855,
      "learning_rate": 3.2895998629455204e-06,
      "loss": 0.0725,
      "step": 11910
    },
    {
      "epoch": 0.6150787503227472,
      "grad_norm": 3.6792261600494385,
      "learning_rate": 3.2888299566064853e-06,
      "loss": 0.0609,
      "step": 11911
    },
    {
      "epoch": 0.615130389878647,
      "grad_norm": 8.253216743469238,
      "learning_rate": 3.288060096219339e-06,
      "loss": 0.075,
      "step": 11912
    },
    {
      "epoch": 0.6151820294345469,
      "grad_norm": 2.195758581161499,
      "learning_rate": 3.2872902818047567e-06,
      "loss": 0.0395,
      "step": 11913
    },
    {
      "epoch": 0.6152336689904467,
      "grad_norm": 5.362803936004639,
      "learning_rate": 3.2865205133834133e-06,
      "loss": 0.0555,
      "step": 11914
    },
    {
      "epoch": 0.6152853085463466,
      "grad_norm": 2.209979295730591,
      "learning_rate": 3.2857507909759755e-06,
      "loss": 0.0369,
      "step": 11915
    },
    {
      "epoch": 0.6153369481022464,
      "grad_norm": 3.119157075881958,
      "learning_rate": 3.2849811146031165e-06,
      "loss": 0.0612,
      "step": 11916
    },
    {
      "epoch": 0.6153885876581462,
      "grad_norm": 4.408115863800049,
      "learning_rate": 3.2842114842855056e-06,
      "loss": 0.0345,
      "step": 11917
    },
    {
      "epoch": 0.615440227214046,
      "grad_norm": 3.5845234394073486,
      "learning_rate": 3.283441900043809e-06,
      "loss": 0.059,
      "step": 11918
    },
    {
      "epoch": 0.6154918667699458,
      "grad_norm": 10.660920143127441,
      "learning_rate": 3.2826723618986926e-06,
      "loss": 0.028,
      "step": 11919
    },
    {
      "epoch": 0.6155435063258456,
      "grad_norm": 7.2024126052856445,
      "learning_rate": 3.2819028698708223e-06,
      "loss": 0.0573,
      "step": 11920
    },
    {
      "epoch": 0.6155951458817455,
      "grad_norm": 16.86775779724121,
      "learning_rate": 3.2811334239808655e-06,
      "loss": 0.0639,
      "step": 11921
    },
    {
      "epoch": 0.6156467854376453,
      "grad_norm": 4.087299346923828,
      "learning_rate": 3.2803640242494795e-06,
      "loss": 0.0538,
      "step": 11922
    },
    {
      "epoch": 0.6156984249935451,
      "grad_norm": 3.1104443073272705,
      "learning_rate": 3.279594670697329e-06,
      "loss": 0.0556,
      "step": 11923
    },
    {
      "epoch": 0.6157500645494449,
      "grad_norm": 1.857970952987671,
      "learning_rate": 3.278825363345073e-06,
      "loss": 0.0284,
      "step": 11924
    },
    {
      "epoch": 0.6158017041053447,
      "grad_norm": 5.513825416564941,
      "learning_rate": 3.278056102213373e-06,
      "loss": 0.0315,
      "step": 11925
    },
    {
      "epoch": 0.6158533436612446,
      "grad_norm": 1.9483108520507812,
      "learning_rate": 3.2772868873228834e-06,
      "loss": 0.0318,
      "step": 11926
    },
    {
      "epoch": 0.6159049832171444,
      "grad_norm": 6.727470874786377,
      "learning_rate": 3.276517718694261e-06,
      "loss": 0.0794,
      "step": 11927
    },
    {
      "epoch": 0.6159566227730442,
      "grad_norm": 5.660664081573486,
      "learning_rate": 3.275748596348166e-06,
      "loss": 0.0656,
      "step": 11928
    },
    {
      "epoch": 0.616008262328944,
      "grad_norm": 2.1540029048919678,
      "learning_rate": 3.2749795203052455e-06,
      "loss": 0.0416,
      "step": 11929
    },
    {
      "epoch": 0.6160599018848438,
      "grad_norm": 7.537657737731934,
      "learning_rate": 3.2742104905861566e-06,
      "loss": 0.1245,
      "step": 11930
    },
    {
      "epoch": 0.6161115414407436,
      "grad_norm": 8.295329093933105,
      "learning_rate": 3.27344150721155e-06,
      "loss": 0.1187,
      "step": 11931
    },
    {
      "epoch": 0.6161631809966435,
      "grad_norm": 9.172406196594238,
      "learning_rate": 3.2726725702020773e-06,
      "loss": 0.0854,
      "step": 11932
    },
    {
      "epoch": 0.6162148205525433,
      "grad_norm": 6.868453025817871,
      "learning_rate": 3.2719036795783853e-06,
      "loss": 0.0319,
      "step": 11933
    },
    {
      "epoch": 0.6162664601084431,
      "grad_norm": 18.271177291870117,
      "learning_rate": 3.2711348353611217e-06,
      "loss": 0.0654,
      "step": 11934
    },
    {
      "epoch": 0.6163180996643429,
      "grad_norm": 6.907104015350342,
      "learning_rate": 3.2703660375709345e-06,
      "loss": 0.0762,
      "step": 11935
    },
    {
      "epoch": 0.6163697392202427,
      "grad_norm": 11.743776321411133,
      "learning_rate": 3.2695972862284707e-06,
      "loss": 0.0555,
      "step": 11936
    },
    {
      "epoch": 0.6164213787761426,
      "grad_norm": 3.1837427616119385,
      "learning_rate": 3.2688285813543712e-06,
      "loss": 0.0703,
      "step": 11937
    },
    {
      "epoch": 0.6164730183320424,
      "grad_norm": 8.588716506958008,
      "learning_rate": 3.26805992296928e-06,
      "loss": 0.0426,
      "step": 11938
    },
    {
      "epoch": 0.6165246578879422,
      "grad_norm": 1.4901281595230103,
      "learning_rate": 3.26729131109384e-06,
      "loss": 0.0256,
      "step": 11939
    },
    {
      "epoch": 0.616576297443842,
      "grad_norm": 7.327631950378418,
      "learning_rate": 3.2665227457486897e-06,
      "loss": 0.0896,
      "step": 11940
    },
    {
      "epoch": 0.6166279369997418,
      "grad_norm": 5.932357311248779,
      "learning_rate": 3.265754226954467e-06,
      "loss": 0.1004,
      "step": 11941
    },
    {
      "epoch": 0.6166795765556417,
      "grad_norm": 5.8975443840026855,
      "learning_rate": 3.2649857547318143e-06,
      "loss": 0.0967,
      "step": 11942
    },
    {
      "epoch": 0.6167312161115415,
      "grad_norm": 1.4912748336791992,
      "learning_rate": 3.264217329101366e-06,
      "loss": 0.0211,
      "step": 11943
    },
    {
      "epoch": 0.6167828556674413,
      "grad_norm": 5.983984470367432,
      "learning_rate": 3.2634489500837553e-06,
      "loss": 0.0426,
      "step": 11944
    },
    {
      "epoch": 0.6168344952233411,
      "grad_norm": 2.5612175464630127,
      "learning_rate": 3.2626806176996183e-06,
      "loss": 0.0538,
      "step": 11945
    },
    {
      "epoch": 0.6168861347792409,
      "grad_norm": 2.5502383708953857,
      "learning_rate": 3.261912331969587e-06,
      "loss": 0.0353,
      "step": 11946
    },
    {
      "epoch": 0.6169377743351407,
      "grad_norm": 10.27586555480957,
      "learning_rate": 3.261144092914297e-06,
      "loss": 0.1032,
      "step": 11947
    },
    {
      "epoch": 0.6169894138910406,
      "grad_norm": 6.57877779006958,
      "learning_rate": 3.2603759005543716e-06,
      "loss": 0.0414,
      "step": 11948
    },
    {
      "epoch": 0.6170410534469404,
      "grad_norm": 5.0048747062683105,
      "learning_rate": 3.259607754910445e-06,
      "loss": 0.0765,
      "step": 11949
    },
    {
      "epoch": 0.6170926930028402,
      "grad_norm": 4.759245872497559,
      "learning_rate": 3.2588396560031445e-06,
      "loss": 0.071,
      "step": 11950
    },
    {
      "epoch": 0.61714433255874,
      "grad_norm": 2.369486093521118,
      "learning_rate": 3.2580716038530945e-06,
      "loss": 0.0407,
      "step": 11951
    },
    {
      "epoch": 0.6171959721146398,
      "grad_norm": 3.824868679046631,
      "learning_rate": 3.257303598480922e-06,
      "loss": 0.109,
      "step": 11952
    },
    {
      "epoch": 0.6172476116705397,
      "grad_norm": 10.390612602233887,
      "learning_rate": 3.2565356399072497e-06,
      "loss": 0.1207,
      "step": 11953
    },
    {
      "epoch": 0.6172992512264395,
      "grad_norm": 3.7387430667877197,
      "learning_rate": 3.2557677281527044e-06,
      "loss": 0.0702,
      "step": 11954
    },
    {
      "epoch": 0.6173508907823393,
      "grad_norm": 3.915032386779785,
      "learning_rate": 3.2549998632379017e-06,
      "loss": 0.0398,
      "step": 11955
    },
    {
      "epoch": 0.6174025303382391,
      "grad_norm": 4.405710220336914,
      "learning_rate": 3.2542320451834654e-06,
      "loss": 0.0835,
      "step": 11956
    },
    {
      "epoch": 0.6174541698941389,
      "grad_norm": 3.2038233280181885,
      "learning_rate": 3.2534642740100145e-06,
      "loss": 0.0615,
      "step": 11957
    },
    {
      "epoch": 0.6175058094500387,
      "grad_norm": 8.67119312286377,
      "learning_rate": 3.252696549738168e-06,
      "loss": 0.1473,
      "step": 11958
    },
    {
      "epoch": 0.6175574490059386,
      "grad_norm": 3.2631149291992188,
      "learning_rate": 3.251928872388539e-06,
      "loss": 0.0602,
      "step": 11959
    },
    {
      "epoch": 0.6176090885618384,
      "grad_norm": 8.353286743164062,
      "learning_rate": 3.2511612419817427e-06,
      "loss": 0.0584,
      "step": 11960
    },
    {
      "epoch": 0.6176607281177382,
      "grad_norm": 37.15969467163086,
      "learning_rate": 3.2503936585383995e-06,
      "loss": 0.0943,
      "step": 11961
    },
    {
      "epoch": 0.617712367673638,
      "grad_norm": 7.452663898468018,
      "learning_rate": 3.2496261220791136e-06,
      "loss": 0.053,
      "step": 11962
    },
    {
      "epoch": 0.6177640072295378,
      "grad_norm": 1.4775478839874268,
      "learning_rate": 3.248858632624502e-06,
      "loss": 0.0273,
      "step": 11963
    },
    {
      "epoch": 0.6178156467854377,
      "grad_norm": 5.275636196136475,
      "learning_rate": 3.248091190195173e-06,
      "loss": 0.0532,
      "step": 11964
    },
    {
      "epoch": 0.6178672863413375,
      "grad_norm": 9.259055137634277,
      "learning_rate": 3.2473237948117374e-06,
      "loss": 0.0788,
      "step": 11965
    },
    {
      "epoch": 0.6179189258972373,
      "grad_norm": 3.2238967418670654,
      "learning_rate": 3.2465564464947995e-06,
      "loss": 0.0721,
      "step": 11966
    },
    {
      "epoch": 0.6179705654531371,
      "grad_norm": 13.951435089111328,
      "learning_rate": 3.2457891452649674e-06,
      "loss": 0.1247,
      "step": 11967
    },
    {
      "epoch": 0.6180222050090369,
      "grad_norm": 3.8954858779907227,
      "learning_rate": 3.2450218911428476e-06,
      "loss": 0.0537,
      "step": 11968
    },
    {
      "epoch": 0.6180738445649367,
      "grad_norm": 1.5160813331604004,
      "learning_rate": 3.2442546841490434e-06,
      "loss": 0.038,
      "step": 11969
    },
    {
      "epoch": 0.6181254841208366,
      "grad_norm": 3.1713790893554688,
      "learning_rate": 3.2434875243041564e-06,
      "loss": 0.0497,
      "step": 11970
    },
    {
      "epoch": 0.6181771236767364,
      "grad_norm": 1.8510042428970337,
      "learning_rate": 3.242720411628788e-06,
      "loss": 0.035,
      "step": 11971
    },
    {
      "epoch": 0.6182287632326362,
      "grad_norm": 2.458780527114868,
      "learning_rate": 3.2419533461435403e-06,
      "loss": 0.0218,
      "step": 11972
    },
    {
      "epoch": 0.618280402788536,
      "grad_norm": 16.00887680053711,
      "learning_rate": 3.2411863278690097e-06,
      "loss": 0.2021,
      "step": 11973
    },
    {
      "epoch": 0.6183320423444358,
      "grad_norm": 9.154698371887207,
      "learning_rate": 3.2404193568257934e-06,
      "loss": 0.0623,
      "step": 11974
    },
    {
      "epoch": 0.6183836819003357,
      "grad_norm": 5.2548370361328125,
      "learning_rate": 3.23965243303449e-06,
      "loss": 0.0385,
      "step": 11975
    },
    {
      "epoch": 0.6184353214562355,
      "grad_norm": 6.128093719482422,
      "learning_rate": 3.238885556515695e-06,
      "loss": 0.0701,
      "step": 11976
    },
    {
      "epoch": 0.6184869610121353,
      "grad_norm": 3.1107239723205566,
      "learning_rate": 3.2381187272899993e-06,
      "loss": 0.0468,
      "step": 11977
    },
    {
      "epoch": 0.6185386005680351,
      "grad_norm": 2.6097030639648438,
      "learning_rate": 3.2373519453779967e-06,
      "loss": 0.0285,
      "step": 11978
    },
    {
      "epoch": 0.6185902401239349,
      "grad_norm": 6.842384338378906,
      "learning_rate": 3.2365852108002784e-06,
      "loss": 0.0445,
      "step": 11979
    },
    {
      "epoch": 0.6186418796798347,
      "grad_norm": 13.133550643920898,
      "learning_rate": 3.2358185235774363e-06,
      "loss": 0.1289,
      "step": 11980
    },
    {
      "epoch": 0.6186935192357346,
      "grad_norm": 5.595849514007568,
      "learning_rate": 3.235051883730055e-06,
      "loss": 0.083,
      "step": 11981
    },
    {
      "epoch": 0.6187451587916344,
      "grad_norm": 3.888911008834839,
      "learning_rate": 3.234285291278725e-06,
      "loss": 0.0522,
      "step": 11982
    },
    {
      "epoch": 0.6187967983475342,
      "grad_norm": 1.6679432392120361,
      "learning_rate": 3.233518746244033e-06,
      "loss": 0.016,
      "step": 11983
    },
    {
      "epoch": 0.618848437903434,
      "grad_norm": 5.459802627563477,
      "learning_rate": 3.23275224864656e-06,
      "loss": 0.0722,
      "step": 11984
    },
    {
      "epoch": 0.6189000774593338,
      "grad_norm": 3.47750186920166,
      "learning_rate": 3.231985798506894e-06,
      "loss": 0.0674,
      "step": 11985
    },
    {
      "epoch": 0.6189517170152337,
      "grad_norm": 2.9235570430755615,
      "learning_rate": 3.2312193958456127e-06,
      "loss": 0.0385,
      "step": 11986
    },
    {
      "epoch": 0.6190033565711335,
      "grad_norm": 2.8256638050079346,
      "learning_rate": 3.2304530406833033e-06,
      "loss": 0.0379,
      "step": 11987
    },
    {
      "epoch": 0.6190549961270333,
      "grad_norm": 11.731746673583984,
      "learning_rate": 3.2296867330405392e-06,
      "loss": 0.142,
      "step": 11988
    },
    {
      "epoch": 0.6191066356829331,
      "grad_norm": 3.0632410049438477,
      "learning_rate": 3.2289204729379025e-06,
      "loss": 0.0451,
      "step": 11989
    },
    {
      "epoch": 0.6191582752388329,
      "grad_norm": 10.203215599060059,
      "learning_rate": 3.228154260395971e-06,
      "loss": 0.0807,
      "step": 11990
    },
    {
      "epoch": 0.6192099147947328,
      "grad_norm": 11.331507682800293,
      "learning_rate": 3.2273880954353177e-06,
      "loss": 0.0982,
      "step": 11991
    },
    {
      "epoch": 0.6192615543506326,
      "grad_norm": 6.817446708679199,
      "learning_rate": 3.2266219780765197e-06,
      "loss": 0.0844,
      "step": 11992
    },
    {
      "epoch": 0.6193131939065324,
      "grad_norm": 3.336796522140503,
      "learning_rate": 3.2258559083401473e-06,
      "loss": 0.0642,
      "step": 11993
    },
    {
      "epoch": 0.6193648334624322,
      "grad_norm": 7.611915588378906,
      "learning_rate": 3.225089886246778e-06,
      "loss": 0.1329,
      "step": 11994
    },
    {
      "epoch": 0.619416473018332,
      "grad_norm": 2.329575777053833,
      "learning_rate": 3.224323911816977e-06,
      "loss": 0.0312,
      "step": 11995
    },
    {
      "epoch": 0.6194681125742318,
      "grad_norm": 3.470057487487793,
      "learning_rate": 3.223557985071317e-06,
      "loss": 0.0644,
      "step": 11996
    },
    {
      "epoch": 0.6195197521301317,
      "grad_norm": 8.149574279785156,
      "learning_rate": 3.2227921060303655e-06,
      "loss": 0.1526,
      "step": 11997
    },
    {
      "epoch": 0.6195713916860315,
      "grad_norm": 4.57371187210083,
      "learning_rate": 3.222026274714691e-06,
      "loss": 0.0545,
      "step": 11998
    },
    {
      "epoch": 0.6196230312419313,
      "grad_norm": 3.3925156593322754,
      "learning_rate": 3.2212604911448557e-06,
      "loss": 0.0416,
      "step": 11999
    },
    {
      "epoch": 0.6196746707978311,
      "grad_norm": 6.758366584777832,
      "learning_rate": 3.2204947553414255e-06,
      "loss": 0.0551,
      "step": 12000
    },
    {
      "epoch": 0.6197263103537309,
      "grad_norm": 3.02199649810791,
      "learning_rate": 3.219729067324967e-06,
      "loss": 0.0539,
      "step": 12001
    },
    {
      "epoch": 0.6197779499096308,
      "grad_norm": 11.510737419128418,
      "learning_rate": 3.2189634271160355e-06,
      "loss": 0.113,
      "step": 12002
    },
    {
      "epoch": 0.6198295894655306,
      "grad_norm": 7.303288459777832,
      "learning_rate": 3.218197834735196e-06,
      "loss": 0.0469,
      "step": 12003
    },
    {
      "epoch": 0.6198812290214304,
      "grad_norm": 2.7042746543884277,
      "learning_rate": 3.2174322902030075e-06,
      "loss": 0.0528,
      "step": 12004
    },
    {
      "epoch": 0.6199328685773302,
      "grad_norm": 9.17089557647705,
      "learning_rate": 3.216666793540028e-06,
      "loss": 0.0813,
      "step": 12005
    },
    {
      "epoch": 0.61998450813323,
      "grad_norm": 2.431196451187134,
      "learning_rate": 3.215901344766813e-06,
      "loss": 0.0293,
      "step": 12006
    },
    {
      "epoch": 0.6200361476891298,
      "grad_norm": 6.002686023712158,
      "learning_rate": 3.215135943903917e-06,
      "loss": 0.0767,
      "step": 12007
    },
    {
      "epoch": 0.6200877872450297,
      "grad_norm": 1.6990251541137695,
      "learning_rate": 3.214370590971897e-06,
      "loss": 0.0289,
      "step": 12008
    },
    {
      "epoch": 0.6201394268009295,
      "grad_norm": 12.512818336486816,
      "learning_rate": 3.2136052859913047e-06,
      "loss": 0.0994,
      "step": 12009
    },
    {
      "epoch": 0.6201910663568293,
      "grad_norm": 11.069459915161133,
      "learning_rate": 3.212840028982691e-06,
      "loss": 0.043,
      "step": 12010
    },
    {
      "epoch": 0.6202427059127291,
      "grad_norm": 3.2022740840911865,
      "learning_rate": 3.2120748199666064e-06,
      "loss": 0.0314,
      "step": 12011
    },
    {
      "epoch": 0.6202943454686289,
      "grad_norm": 2.8046140670776367,
      "learning_rate": 3.2113096589636018e-06,
      "loss": 0.0264,
      "step": 12012
    },
    {
      "epoch": 0.6203459850245288,
      "grad_norm": 7.801157474517822,
      "learning_rate": 3.2105445459942207e-06,
      "loss": 0.0797,
      "step": 12013
    },
    {
      "epoch": 0.6203976245804286,
      "grad_norm": 6.294312000274658,
      "learning_rate": 3.2097794810790122e-06,
      "loss": 0.0452,
      "step": 12014
    },
    {
      "epoch": 0.6204492641363284,
      "grad_norm": 1.8578864336013794,
      "learning_rate": 3.209014464238522e-06,
      "loss": 0.0315,
      "step": 12015
    },
    {
      "epoch": 0.6205009036922282,
      "grad_norm": 3.2807750701904297,
      "learning_rate": 3.2082494954932935e-06,
      "loss": 0.0327,
      "step": 12016
    },
    {
      "epoch": 0.620552543248128,
      "grad_norm": 10.384690284729004,
      "learning_rate": 3.2074845748638685e-06,
      "loss": 0.0769,
      "step": 12017
    },
    {
      "epoch": 0.6206041828040278,
      "grad_norm": 6.865361213684082,
      "learning_rate": 3.206719702370788e-06,
      "loss": 0.059,
      "step": 12018
    },
    {
      "epoch": 0.6206558223599277,
      "grad_norm": 9.88557243347168,
      "learning_rate": 3.205954878034592e-06,
      "loss": 0.1434,
      "step": 12019
    },
    {
      "epoch": 0.6207074619158275,
      "grad_norm": 9.813817024230957,
      "learning_rate": 3.2051901018758226e-06,
      "loss": 0.1157,
      "step": 12020
    },
    {
      "epoch": 0.6207591014717273,
      "grad_norm": 11.277828216552734,
      "learning_rate": 3.2044253739150105e-06,
      "loss": 0.1282,
      "step": 12021
    },
    {
      "epoch": 0.6208107410276271,
      "grad_norm": 3.4154841899871826,
      "learning_rate": 3.2036606941726967e-06,
      "loss": 0.044,
      "step": 12022
    },
    {
      "epoch": 0.6208623805835269,
      "grad_norm": 8.477184295654297,
      "learning_rate": 3.2028960626694163e-06,
      "loss": 0.0456,
      "step": 12023
    },
    {
      "epoch": 0.6209140201394268,
      "grad_norm": 2.7762789726257324,
      "learning_rate": 3.2021314794257e-06,
      "loss": 0.0536,
      "step": 12024
    },
    {
      "epoch": 0.6209656596953266,
      "grad_norm": 2.7467615604400635,
      "learning_rate": 3.201366944462081e-06,
      "loss": 0.0379,
      "step": 12025
    },
    {
      "epoch": 0.6210172992512264,
      "grad_norm": 5.663909435272217,
      "learning_rate": 3.200602457799089e-06,
      "loss": 0.0811,
      "step": 12026
    },
    {
      "epoch": 0.6210689388071262,
      "grad_norm": 4.812783718109131,
      "learning_rate": 3.1998380194572587e-06,
      "loss": 0.08,
      "step": 12027
    },
    {
      "epoch": 0.621120578363026,
      "grad_norm": 11.569259643554688,
      "learning_rate": 3.199073629457111e-06,
      "loss": 0.1155,
      "step": 12028
    },
    {
      "epoch": 0.6211722179189259,
      "grad_norm": 2.8732354640960693,
      "learning_rate": 3.198309287819178e-06,
      "loss": 0.0482,
      "step": 12029
    },
    {
      "epoch": 0.6212238574748257,
      "grad_norm": 3.6502411365509033,
      "learning_rate": 3.1975449945639836e-06,
      "loss": 0.0477,
      "step": 12030
    },
    {
      "epoch": 0.6212754970307255,
      "grad_norm": 33.567596435546875,
      "learning_rate": 3.196780749712054e-06,
      "loss": 0.0957,
      "step": 12031
    },
    {
      "epoch": 0.6213271365866253,
      "grad_norm": 8.026968002319336,
      "learning_rate": 3.1960165532839096e-06,
      "loss": 0.0297,
      "step": 12032
    },
    {
      "epoch": 0.6213787761425251,
      "grad_norm": 1.8668028116226196,
      "learning_rate": 3.1952524053000718e-06,
      "loss": 0.0149,
      "step": 12033
    },
    {
      "epoch": 0.6214304156984249,
      "grad_norm": 7.531372547149658,
      "learning_rate": 3.194488305781066e-06,
      "loss": 0.093,
      "step": 12034
    },
    {
      "epoch": 0.6214820552543248,
      "grad_norm": 2.8226661682128906,
      "learning_rate": 3.1937242547474046e-06,
      "loss": 0.0447,
      "step": 12035
    },
    {
      "epoch": 0.6215336948102246,
      "grad_norm": 9.578965187072754,
      "learning_rate": 3.1929602522196106e-06,
      "loss": 0.0414,
      "step": 12036
    },
    {
      "epoch": 0.6215853343661244,
      "grad_norm": 3.48384690284729,
      "learning_rate": 3.1921962982181987e-06,
      "loss": 0.0472,
      "step": 12037
    },
    {
      "epoch": 0.6216369739220242,
      "grad_norm": 3.449730634689331,
      "learning_rate": 3.191432392763686e-06,
      "loss": 0.0613,
      "step": 12038
    },
    {
      "epoch": 0.621688613477924,
      "grad_norm": 3.805140495300293,
      "learning_rate": 3.190668535876583e-06,
      "loss": 0.0476,
      "step": 12039
    },
    {
      "epoch": 0.621740253033824,
      "grad_norm": 3.3484673500061035,
      "learning_rate": 3.189904727577403e-06,
      "loss": 0.051,
      "step": 12040
    },
    {
      "epoch": 0.6217918925897238,
      "grad_norm": 2.964439630508423,
      "learning_rate": 3.1891409678866603e-06,
      "loss": 0.0515,
      "step": 12041
    },
    {
      "epoch": 0.6218435321456236,
      "grad_norm": 2.564377546310425,
      "learning_rate": 3.1883772568248643e-06,
      "loss": 0.0229,
      "step": 12042
    },
    {
      "epoch": 0.6218951717015234,
      "grad_norm": 3.136652708053589,
      "learning_rate": 3.1876135944125217e-06,
      "loss": 0.0164,
      "step": 12043
    },
    {
      "epoch": 0.6219468112574232,
      "grad_norm": 1.5839052200317383,
      "learning_rate": 3.1868499806701405e-06,
      "loss": 0.0134,
      "step": 12044
    },
    {
      "epoch": 0.6219984508133231,
      "grad_norm": 1.7180161476135254,
      "learning_rate": 3.1860864156182293e-06,
      "loss": 0.0242,
      "step": 12045
    },
    {
      "epoch": 0.6220500903692229,
      "grad_norm": 3.9883909225463867,
      "learning_rate": 3.185322899277289e-06,
      "loss": 0.0441,
      "step": 12046
    },
    {
      "epoch": 0.6221017299251227,
      "grad_norm": 9.434224128723145,
      "learning_rate": 3.184559431667824e-06,
      "loss": 0.054,
      "step": 12047
    },
    {
      "epoch": 0.6221533694810225,
      "grad_norm": 12.33787727355957,
      "learning_rate": 3.183796012810339e-06,
      "loss": 0.1478,
      "step": 12048
    },
    {
      "epoch": 0.6222050090369223,
      "grad_norm": 17.85774040222168,
      "learning_rate": 3.1830326427253345e-06,
      "loss": 0.1509,
      "step": 12049
    },
    {
      "epoch": 0.6222566485928221,
      "grad_norm": 5.888818264007568,
      "learning_rate": 3.1822693214333077e-06,
      "loss": 0.0782,
      "step": 12050
    },
    {
      "epoch": 0.622308288148722,
      "grad_norm": 4.086420059204102,
      "learning_rate": 3.181506048954759e-06,
      "loss": 0.0316,
      "step": 12051
    },
    {
      "epoch": 0.6223599277046218,
      "grad_norm": 19.619009017944336,
      "learning_rate": 3.180742825310183e-06,
      "loss": 0.1247,
      "step": 12052
    },
    {
      "epoch": 0.6224115672605216,
      "grad_norm": 5.647104740142822,
      "learning_rate": 3.1799796505200805e-06,
      "loss": 0.0525,
      "step": 12053
    },
    {
      "epoch": 0.6224632068164214,
      "grad_norm": 10.711792945861816,
      "learning_rate": 3.179216524604939e-06,
      "loss": 0.1413,
      "step": 12054
    },
    {
      "epoch": 0.6225148463723212,
      "grad_norm": 2.837167739868164,
      "learning_rate": 3.1784534475852557e-06,
      "loss": 0.0509,
      "step": 12055
    },
    {
      "epoch": 0.6225664859282211,
      "grad_norm": 8.378273963928223,
      "learning_rate": 3.1776904194815227e-06,
      "loss": 0.1094,
      "step": 12056
    },
    {
      "epoch": 0.6226181254841209,
      "grad_norm": 5.428055763244629,
      "learning_rate": 3.1769274403142285e-06,
      "loss": 0.0661,
      "step": 12057
    },
    {
      "epoch": 0.6226697650400207,
      "grad_norm": 10.650221824645996,
      "learning_rate": 3.176164510103863e-06,
      "loss": 0.0927,
      "step": 12058
    },
    {
      "epoch": 0.6227214045959205,
      "grad_norm": 10.842232704162598,
      "learning_rate": 3.1754016288709134e-06,
      "loss": 0.1267,
      "step": 12059
    },
    {
      "epoch": 0.6227730441518203,
      "grad_norm": 6.906634330749512,
      "learning_rate": 3.1746387966358687e-06,
      "loss": 0.0561,
      "step": 12060
    },
    {
      "epoch": 0.6228246837077202,
      "grad_norm": 9.202470779418945,
      "learning_rate": 3.1738760134192092e-06,
      "loss": 0.0944,
      "step": 12061
    },
    {
      "epoch": 0.62287632326362,
      "grad_norm": 1.489656686782837,
      "learning_rate": 3.173113279241423e-06,
      "loss": 0.0288,
      "step": 12062
    },
    {
      "epoch": 0.6229279628195198,
      "grad_norm": 2.526993751525879,
      "learning_rate": 3.1723505941229913e-06,
      "loss": 0.0361,
      "step": 12063
    },
    {
      "epoch": 0.6229796023754196,
      "grad_norm": 4.523404121398926,
      "learning_rate": 3.1715879580843967e-06,
      "loss": 0.0442,
      "step": 12064
    },
    {
      "epoch": 0.6230312419313194,
      "grad_norm": 6.526827335357666,
      "learning_rate": 3.1708253711461157e-06,
      "loss": 0.0465,
      "step": 12065
    },
    {
      "epoch": 0.6230828814872192,
      "grad_norm": 16.141555786132812,
      "learning_rate": 3.1700628333286277e-06,
      "loss": 0.1031,
      "step": 12066
    },
    {
      "epoch": 0.6231345210431191,
      "grad_norm": 3.766939401626587,
      "learning_rate": 3.169300344652415e-06,
      "loss": 0.0384,
      "step": 12067
    },
    {
      "epoch": 0.6231861605990189,
      "grad_norm": 3.968998432159424,
      "learning_rate": 3.168537905137946e-06,
      "loss": 0.0359,
      "step": 12068
    },
    {
      "epoch": 0.6232378001549187,
      "grad_norm": 11.915452003479004,
      "learning_rate": 3.1677755148056997e-06,
      "loss": 0.1609,
      "step": 12069
    },
    {
      "epoch": 0.6232894397108185,
      "grad_norm": 2.687617301940918,
      "learning_rate": 3.167013173676149e-06,
      "loss": 0.0376,
      "step": 12070
    },
    {
      "epoch": 0.6233410792667183,
      "grad_norm": 2.8894665241241455,
      "learning_rate": 3.166250881769766e-06,
      "loss": 0.0331,
      "step": 12071
    },
    {
      "epoch": 0.6233927188226182,
      "grad_norm": 2.9355523586273193,
      "learning_rate": 3.16548863910702e-06,
      "loss": 0.0353,
      "step": 12072
    },
    {
      "epoch": 0.623444358378518,
      "grad_norm": 1.6111923456192017,
      "learning_rate": 3.1647264457083796e-06,
      "loss": 0.0157,
      "step": 12073
    },
    {
      "epoch": 0.6234959979344178,
      "grad_norm": 1.741870641708374,
      "learning_rate": 3.163964301594316e-06,
      "loss": 0.0227,
      "step": 12074
    },
    {
      "epoch": 0.6235476374903176,
      "grad_norm": 4.090499401092529,
      "learning_rate": 3.163202206785295e-06,
      "loss": 0.0489,
      "step": 12075
    },
    {
      "epoch": 0.6235992770462174,
      "grad_norm": 11.561288833618164,
      "learning_rate": 3.1624401613017795e-06,
      "loss": 0.1418,
      "step": 12076
    },
    {
      "epoch": 0.6236509166021172,
      "grad_norm": 5.179390907287598,
      "learning_rate": 3.161678165164236e-06,
      "loss": 0.0508,
      "step": 12077
    },
    {
      "epoch": 0.6237025561580171,
      "grad_norm": 3.491694927215576,
      "learning_rate": 3.1609162183931273e-06,
      "loss": 0.0281,
      "step": 12078
    },
    {
      "epoch": 0.6237541957139169,
      "grad_norm": 4.2244720458984375,
      "learning_rate": 3.1601543210089125e-06,
      "loss": 0.0428,
      "step": 12079
    },
    {
      "epoch": 0.6238058352698167,
      "grad_norm": 19.54926109313965,
      "learning_rate": 3.1593924730320525e-06,
      "loss": 0.2958,
      "step": 12080
    },
    {
      "epoch": 0.6238574748257165,
      "grad_norm": 8.975648880004883,
      "learning_rate": 3.1586306744830064e-06,
      "loss": 0.062,
      "step": 12081
    },
    {
      "epoch": 0.6239091143816163,
      "grad_norm": 2.612438917160034,
      "learning_rate": 3.1578689253822335e-06,
      "loss": 0.0446,
      "step": 12082
    },
    {
      "epoch": 0.6239607539375162,
      "grad_norm": 2.444092273712158,
      "learning_rate": 3.1571072257501867e-06,
      "loss": 0.0446,
      "step": 12083
    },
    {
      "epoch": 0.624012393493416,
      "grad_norm": 1.6211057901382446,
      "learning_rate": 3.1563455756073226e-06,
      "loss": 0.0191,
      "step": 12084
    },
    {
      "epoch": 0.6240640330493158,
      "grad_norm": 1.845308780670166,
      "learning_rate": 3.1555839749740926e-06,
      "loss": 0.0314,
      "step": 12085
    },
    {
      "epoch": 0.6241156726052156,
      "grad_norm": 4.131995677947998,
      "learning_rate": 3.1548224238709536e-06,
      "loss": 0.0767,
      "step": 12086
    },
    {
      "epoch": 0.6241673121611154,
      "grad_norm": 12.738625526428223,
      "learning_rate": 3.15406092231835e-06,
      "loss": 0.1072,
      "step": 12087
    },
    {
      "epoch": 0.6242189517170152,
      "grad_norm": 2.46655011177063,
      "learning_rate": 3.1532994703367353e-06,
      "loss": 0.0195,
      "step": 12088
    },
    {
      "epoch": 0.6242705912729151,
      "grad_norm": 2.7085065841674805,
      "learning_rate": 3.1525380679465576e-06,
      "loss": 0.0359,
      "step": 12089
    },
    {
      "epoch": 0.6243222308288149,
      "grad_norm": 24.30638313293457,
      "learning_rate": 3.151776715168261e-06,
      "loss": 0.0639,
      "step": 12090
    },
    {
      "epoch": 0.6243738703847147,
      "grad_norm": 6.813990592956543,
      "learning_rate": 3.1510154120222935e-06,
      "loss": 0.0751,
      "step": 12091
    },
    {
      "epoch": 0.6244255099406145,
      "grad_norm": 12.298781394958496,
      "learning_rate": 3.1502541585290967e-06,
      "loss": 0.1699,
      "step": 12092
    },
    {
      "epoch": 0.6244771494965143,
      "grad_norm": 3.3494746685028076,
      "learning_rate": 3.149492954709118e-06,
      "loss": 0.0326,
      "step": 12093
    },
    {
      "epoch": 0.6245287890524142,
      "grad_norm": 1.6948846578598022,
      "learning_rate": 3.1487318005827927e-06,
      "loss": 0.0163,
      "step": 12094
    },
    {
      "epoch": 0.624580428608314,
      "grad_norm": 16.769432067871094,
      "learning_rate": 3.1479706961705647e-06,
      "loss": 0.1446,
      "step": 12095
    },
    {
      "epoch": 0.6246320681642138,
      "grad_norm": 3.1083672046661377,
      "learning_rate": 3.1472096414928722e-06,
      "loss": 0.0593,
      "step": 12096
    },
    {
      "epoch": 0.6246837077201136,
      "grad_norm": 2.342179298400879,
      "learning_rate": 3.146448636570153e-06,
      "loss": 0.0481,
      "step": 12097
    },
    {
      "epoch": 0.6247353472760134,
      "grad_norm": 7.041646957397461,
      "learning_rate": 3.145687681422842e-06,
      "loss": 0.0606,
      "step": 12098
    },
    {
      "epoch": 0.6247869868319132,
      "grad_norm": 2.2244153022766113,
      "learning_rate": 3.144926776071374e-06,
      "loss": 0.0161,
      "step": 12099
    },
    {
      "epoch": 0.6248386263878131,
      "grad_norm": 9.08401107788086,
      "learning_rate": 3.1441659205361853e-06,
      "loss": 0.0347,
      "step": 12100
    },
    {
      "epoch": 0.6248902659437129,
      "grad_norm": 7.342604637145996,
      "learning_rate": 3.143405114837702e-06,
      "loss": 0.0829,
      "step": 12101
    },
    {
      "epoch": 0.6249419054996127,
      "grad_norm": 6.516989231109619,
      "learning_rate": 3.1426443589963606e-06,
      "loss": 0.0948,
      "step": 12102
    },
    {
      "epoch": 0.6249935450555125,
      "grad_norm": 2.971315860748291,
      "learning_rate": 3.1418836530325873e-06,
      "loss": 0.0272,
      "step": 12103
    },
    {
      "epoch": 0.6250451846114123,
      "grad_norm": 3.455596685409546,
      "learning_rate": 3.1411229969668126e-06,
      "loss": 0.0344,
      "step": 12104
    },
    {
      "epoch": 0.6250968241673122,
      "grad_norm": 3.623988389968872,
      "learning_rate": 3.1403623908194614e-06,
      "loss": 0.0401,
      "step": 12105
    },
    {
      "epoch": 0.625148463723212,
      "grad_norm": 39.56899642944336,
      "learning_rate": 3.1396018346109573e-06,
      "loss": 0.0949,
      "step": 12106
    },
    {
      "epoch": 0.6252001032791118,
      "grad_norm": 7.351990699768066,
      "learning_rate": 3.138841328361728e-06,
      "loss": 0.1117,
      "step": 12107
    },
    {
      "epoch": 0.6252517428350116,
      "grad_norm": 4.264903545379639,
      "learning_rate": 3.1380808720921963e-06,
      "loss": 0.0327,
      "step": 12108
    },
    {
      "epoch": 0.6253033823909114,
      "grad_norm": 4.195711612701416,
      "learning_rate": 3.1373204658227807e-06,
      "loss": 0.0647,
      "step": 12109
    },
    {
      "epoch": 0.6253550219468113,
      "grad_norm": 7.0044121742248535,
      "learning_rate": 3.136560109573902e-06,
      "loss": 0.1202,
      "step": 12110
    },
    {
      "epoch": 0.6254066615027111,
      "grad_norm": 4.505059242248535,
      "learning_rate": 3.1357998033659804e-06,
      "loss": 0.062,
      "step": 12111
    },
    {
      "epoch": 0.6254583010586109,
      "grad_norm": 5.940981864929199,
      "learning_rate": 3.1350395472194316e-06,
      "loss": 0.0729,
      "step": 12112
    },
    {
      "epoch": 0.6255099406145107,
      "grad_norm": 10.70734691619873,
      "learning_rate": 3.13427934115467e-06,
      "loss": 0.0616,
      "step": 12113
    },
    {
      "epoch": 0.6255615801704105,
      "grad_norm": 6.794731140136719,
      "learning_rate": 3.1335191851921145e-06,
      "loss": 0.0571,
      "step": 12114
    },
    {
      "epoch": 0.6256132197263103,
      "grad_norm": 3.1072442531585693,
      "learning_rate": 3.1327590793521766e-06,
      "loss": 0.0326,
      "step": 12115
    },
    {
      "epoch": 0.6256648592822102,
      "grad_norm": 4.848915100097656,
      "learning_rate": 3.131999023655268e-06,
      "loss": 0.0356,
      "step": 12116
    },
    {
      "epoch": 0.62571649883811,
      "grad_norm": 9.456729888916016,
      "learning_rate": 3.131239018121798e-06,
      "loss": 0.0963,
      "step": 12117
    },
    {
      "epoch": 0.6257681383940098,
      "grad_norm": 3.4727773666381836,
      "learning_rate": 3.130479062772176e-06,
      "loss": 0.0627,
      "step": 12118
    },
    {
      "epoch": 0.6258197779499096,
      "grad_norm": 4.452741622924805,
      "learning_rate": 3.1297191576268143e-06,
      "loss": 0.0483,
      "step": 12119
    },
    {
      "epoch": 0.6258714175058094,
      "grad_norm": 5.2438812255859375,
      "learning_rate": 3.1289593027061125e-06,
      "loss": 0.0597,
      "step": 12120
    },
    {
      "epoch": 0.6259230570617093,
      "grad_norm": 7.467198848724365,
      "learning_rate": 3.1281994980304814e-06,
      "loss": 0.121,
      "step": 12121
    },
    {
      "epoch": 0.6259746966176091,
      "grad_norm": 4.27384090423584,
      "learning_rate": 3.127439743620323e-06,
      "loss": 0.0351,
      "step": 12122
    },
    {
      "epoch": 0.6260263361735089,
      "grad_norm": 5.983897686004639,
      "learning_rate": 3.1266800394960396e-06,
      "loss": 0.1006,
      "step": 12123
    },
    {
      "epoch": 0.6260779757294087,
      "grad_norm": 8.215129852294922,
      "learning_rate": 3.1259203856780314e-06,
      "loss": 0.0892,
      "step": 12124
    },
    {
      "epoch": 0.6261296152853085,
      "grad_norm": 3.192274332046509,
      "learning_rate": 3.125160782186698e-06,
      "loss": 0.0435,
      "step": 12125
    },
    {
      "epoch": 0.6261812548412083,
      "grad_norm": 5.018156051635742,
      "learning_rate": 3.124401229042443e-06,
      "loss": 0.0562,
      "step": 12126
    },
    {
      "epoch": 0.6262328943971082,
      "grad_norm": 12.292742729187012,
      "learning_rate": 3.123641726265656e-06,
      "loss": 0.1159,
      "step": 12127
    },
    {
      "epoch": 0.626284533953008,
      "grad_norm": 1.700503945350647,
      "learning_rate": 3.1228822738767373e-06,
      "loss": 0.025,
      "step": 12128
    },
    {
      "epoch": 0.6263361735089078,
      "grad_norm": 4.583454608917236,
      "learning_rate": 3.1221228718960793e-06,
      "loss": 0.0542,
      "step": 12129
    },
    {
      "epoch": 0.6263878130648076,
      "grad_norm": 1.389399766921997,
      "learning_rate": 3.1213635203440784e-06,
      "loss": 0.0194,
      "step": 12130
    },
    {
      "epoch": 0.6264394526207074,
      "grad_norm": 21.86770248413086,
      "learning_rate": 3.1206042192411223e-06,
      "loss": 0.1288,
      "step": 12131
    },
    {
      "epoch": 0.6264910921766073,
      "grad_norm": 3.8546700477600098,
      "learning_rate": 3.1198449686076015e-06,
      "loss": 0.0644,
      "step": 12132
    },
    {
      "epoch": 0.6265427317325071,
      "grad_norm": 3.9984805583953857,
      "learning_rate": 3.11908576846391e-06,
      "loss": 0.0538,
      "step": 12133
    },
    {
      "epoch": 0.6265943712884069,
      "grad_norm": 10.113866806030273,
      "learning_rate": 3.118326618830429e-06,
      "loss": 0.0505,
      "step": 12134
    },
    {
      "epoch": 0.6266460108443067,
      "grad_norm": 6.643614768981934,
      "learning_rate": 3.117567519727548e-06,
      "loss": 0.0758,
      "step": 12135
    },
    {
      "epoch": 0.6266976504002065,
      "grad_norm": 5.167250156402588,
      "learning_rate": 3.116808471175652e-06,
      "loss": 0.0332,
      "step": 12136
    },
    {
      "epoch": 0.6267492899561063,
      "grad_norm": 18.239219665527344,
      "learning_rate": 3.116049473195124e-06,
      "loss": 0.1075,
      "step": 12137
    },
    {
      "epoch": 0.6268009295120062,
      "grad_norm": 3.019200086593628,
      "learning_rate": 3.115290525806346e-06,
      "loss": 0.0657,
      "step": 12138
    },
    {
      "epoch": 0.626852569067906,
      "grad_norm": 2.6683883666992188,
      "learning_rate": 3.1145316290296977e-06,
      "loss": 0.0261,
      "step": 12139
    },
    {
      "epoch": 0.6269042086238058,
      "grad_norm": 2.524651527404785,
      "learning_rate": 3.1137727828855614e-06,
      "loss": 0.03,
      "step": 12140
    },
    {
      "epoch": 0.6269558481797056,
      "grad_norm": 3.875659465789795,
      "learning_rate": 3.1130139873943143e-06,
      "loss": 0.0628,
      "step": 12141
    },
    {
      "epoch": 0.6270074877356054,
      "grad_norm": 7.733884334564209,
      "learning_rate": 3.112255242576332e-06,
      "loss": 0.09,
      "step": 12142
    },
    {
      "epoch": 0.6270591272915053,
      "grad_norm": 1.7540462017059326,
      "learning_rate": 3.11149654845199e-06,
      "loss": 0.0357,
      "step": 12143
    },
    {
      "epoch": 0.6271107668474051,
      "grad_norm": 3.662105083465576,
      "learning_rate": 3.110737905041664e-06,
      "loss": 0.0452,
      "step": 12144
    },
    {
      "epoch": 0.6271624064033049,
      "grad_norm": 6.913881301879883,
      "learning_rate": 3.1099793123657245e-06,
      "loss": 0.043,
      "step": 12145
    },
    {
      "epoch": 0.6272140459592047,
      "grad_norm": 5.9357991218566895,
      "learning_rate": 3.1092207704445428e-06,
      "loss": 0.0348,
      "step": 12146
    },
    {
      "epoch": 0.6272656855151045,
      "grad_norm": 29.089256286621094,
      "learning_rate": 3.108462279298491e-06,
      "loss": 0.2942,
      "step": 12147
    },
    {
      "epoch": 0.6273173250710044,
      "grad_norm": 8.496036529541016,
      "learning_rate": 3.1077038389479362e-06,
      "loss": 0.0593,
      "step": 12148
    },
    {
      "epoch": 0.6273689646269042,
      "grad_norm": 6.697878360748291,
      "learning_rate": 3.106945449413247e-06,
      "loss": 0.0537,
      "step": 12149
    },
    {
      "epoch": 0.627420604182804,
      "grad_norm": 2.98496150970459,
      "learning_rate": 3.1061871107147866e-06,
      "loss": 0.0326,
      "step": 12150
    },
    {
      "epoch": 0.6274722437387038,
      "grad_norm": 4.333249568939209,
      "learning_rate": 3.10542882287292e-06,
      "loss": 0.0928,
      "step": 12151
    },
    {
      "epoch": 0.6275238832946036,
      "grad_norm": 3.8273890018463135,
      "learning_rate": 3.104670585908015e-06,
      "loss": 0.0461,
      "step": 12152
    },
    {
      "epoch": 0.6275755228505034,
      "grad_norm": 5.297685623168945,
      "learning_rate": 3.1039123998404263e-06,
      "loss": 0.0481,
      "step": 12153
    },
    {
      "epoch": 0.6276271624064033,
      "grad_norm": 5.349859237670898,
      "learning_rate": 3.103154264690519e-06,
      "loss": 0.0452,
      "step": 12154
    },
    {
      "epoch": 0.6276788019623031,
      "grad_norm": 3.6938605308532715,
      "learning_rate": 3.102396180478652e-06,
      "loss": 0.0481,
      "step": 12155
    },
    {
      "epoch": 0.6277304415182029,
      "grad_norm": 2.73162579536438,
      "learning_rate": 3.10163814722518e-06,
      "loss": 0.0342,
      "step": 12156
    },
    {
      "epoch": 0.6277820810741027,
      "grad_norm": 6.586240768432617,
      "learning_rate": 3.1008801649504615e-06,
      "loss": 0.0196,
      "step": 12157
    },
    {
      "epoch": 0.6278337206300025,
      "grad_norm": 4.737054347991943,
      "learning_rate": 3.10012223367485e-06,
      "loss": 0.0611,
      "step": 12158
    },
    {
      "epoch": 0.6278853601859024,
      "grad_norm": 3.1259772777557373,
      "learning_rate": 3.099364353418703e-06,
      "loss": 0.0603,
      "step": 12159
    },
    {
      "epoch": 0.6279369997418022,
      "grad_norm": 10.046931266784668,
      "learning_rate": 3.0986065242023666e-06,
      "loss": 0.0829,
      "step": 12160
    },
    {
      "epoch": 0.627988639297702,
      "grad_norm": 5.76084566116333,
      "learning_rate": 3.0978487460461957e-06,
      "loss": 0.0457,
      "step": 12161
    },
    {
      "epoch": 0.6280402788536018,
      "grad_norm": 13.896099090576172,
      "learning_rate": 3.097091018970538e-06,
      "loss": 0.1016,
      "step": 12162
    },
    {
      "epoch": 0.6280919184095016,
      "grad_norm": 2.862642765045166,
      "learning_rate": 3.0963333429957443e-06,
      "loss": 0.0195,
      "step": 12163
    },
    {
      "epoch": 0.6281435579654014,
      "grad_norm": 2.7959609031677246,
      "learning_rate": 3.0955757181421587e-06,
      "loss": 0.028,
      "step": 12164
    },
    {
      "epoch": 0.6281951975213013,
      "grad_norm": 4.031534671783447,
      "learning_rate": 3.094818144430125e-06,
      "loss": 0.0335,
      "step": 12165
    },
    {
      "epoch": 0.6282468370772011,
      "grad_norm": 3.31980299949646,
      "learning_rate": 3.094060621879992e-06,
      "loss": 0.0225,
      "step": 12166
    },
    {
      "epoch": 0.628298476633101,
      "grad_norm": 5.2676496505737305,
      "learning_rate": 3.093303150512097e-06,
      "loss": 0.0874,
      "step": 12167
    },
    {
      "epoch": 0.6283501161890008,
      "grad_norm": 3.943744421005249,
      "learning_rate": 3.0925457303467847e-06,
      "loss": 0.0426,
      "step": 12168
    },
    {
      "epoch": 0.6284017557449006,
      "grad_norm": 6.1519904136657715,
      "learning_rate": 3.091788361404394e-06,
      "loss": 0.0628,
      "step": 12169
    },
    {
      "epoch": 0.6284533953008005,
      "grad_norm": 9.096879005432129,
      "learning_rate": 3.0910310437052644e-06,
      "loss": 0.0951,
      "step": 12170
    },
    {
      "epoch": 0.6285050348567003,
      "grad_norm": 4.4686737060546875,
      "learning_rate": 3.09027377726973e-06,
      "loss": 0.0766,
      "step": 12171
    },
    {
      "epoch": 0.6285566744126001,
      "grad_norm": 5.534108638763428,
      "learning_rate": 3.0895165621181276e-06,
      "loss": 0.0519,
      "step": 12172
    },
    {
      "epoch": 0.6286083139684999,
      "grad_norm": 1.6116870641708374,
      "learning_rate": 3.088759398270794e-06,
      "loss": 0.0273,
      "step": 12173
    },
    {
      "epoch": 0.6286599535243997,
      "grad_norm": 4.938299655914307,
      "learning_rate": 3.0880022857480606e-06,
      "loss": 0.0655,
      "step": 12174
    },
    {
      "epoch": 0.6287115930802996,
      "grad_norm": 5.547278881072998,
      "learning_rate": 3.0872452245702577e-06,
      "loss": 0.0261,
      "step": 12175
    },
    {
      "epoch": 0.6287632326361994,
      "grad_norm": 4.4957380294799805,
      "learning_rate": 3.086488214757717e-06,
      "loss": 0.0284,
      "step": 12176
    },
    {
      "epoch": 0.6288148721920992,
      "grad_norm": 3.6808273792266846,
      "learning_rate": 3.0857312563307675e-06,
      "loss": 0.0596,
      "step": 12177
    },
    {
      "epoch": 0.628866511747999,
      "grad_norm": 11.326131820678711,
      "learning_rate": 3.0849743493097347e-06,
      "loss": 0.1729,
      "step": 12178
    },
    {
      "epoch": 0.6289181513038988,
      "grad_norm": 22.824207305908203,
      "learning_rate": 3.084217493714945e-06,
      "loss": 0.2656,
      "step": 12179
    },
    {
      "epoch": 0.6289697908597986,
      "grad_norm": 9.585591316223145,
      "learning_rate": 3.083460689566724e-06,
      "loss": 0.0783,
      "step": 12180
    },
    {
      "epoch": 0.6290214304156985,
      "grad_norm": 12.07768440246582,
      "learning_rate": 3.0827039368853972e-06,
      "loss": 0.1555,
      "step": 12181
    },
    {
      "epoch": 0.6290730699715983,
      "grad_norm": 1.6084353923797607,
      "learning_rate": 3.0819472356912827e-06,
      "loss": 0.02,
      "step": 12182
    },
    {
      "epoch": 0.6291247095274981,
      "grad_norm": 5.018582344055176,
      "learning_rate": 3.081190586004702e-06,
      "loss": 0.0634,
      "step": 12183
    },
    {
      "epoch": 0.6291763490833979,
      "grad_norm": 15.04322624206543,
      "learning_rate": 3.0804339878459745e-06,
      "loss": 0.0649,
      "step": 12184
    },
    {
      "epoch": 0.6292279886392977,
      "grad_norm": 4.809981822967529,
      "learning_rate": 3.0796774412354202e-06,
      "loss": 0.0971,
      "step": 12185
    },
    {
      "epoch": 0.6292796281951976,
      "grad_norm": 4.18874979019165,
      "learning_rate": 3.078920946193351e-06,
      "loss": 0.0637,
      "step": 12186
    },
    {
      "epoch": 0.6293312677510974,
      "grad_norm": 4.918069362640381,
      "learning_rate": 3.078164502740085e-06,
      "loss": 0.0431,
      "step": 12187
    },
    {
      "epoch": 0.6293829073069972,
      "grad_norm": 3.2067770957946777,
      "learning_rate": 3.0774081108959368e-06,
      "loss": 0.0482,
      "step": 12188
    },
    {
      "epoch": 0.629434546862897,
      "grad_norm": 4.499867916107178,
      "learning_rate": 3.0766517706812155e-06,
      "loss": 0.0577,
      "step": 12189
    },
    {
      "epoch": 0.6294861864187968,
      "grad_norm": 7.420556545257568,
      "learning_rate": 3.0758954821162336e-06,
      "loss": 0.0206,
      "step": 12190
    },
    {
      "epoch": 0.6295378259746967,
      "grad_norm": 4.078341007232666,
      "learning_rate": 3.075139245221298e-06,
      "loss": 0.0607,
      "step": 12191
    },
    {
      "epoch": 0.6295894655305965,
      "grad_norm": 11.87093448638916,
      "learning_rate": 3.0743830600167222e-06,
      "loss": 0.1061,
      "step": 12192
    },
    {
      "epoch": 0.6296411050864963,
      "grad_norm": 5.6031341552734375,
      "learning_rate": 3.073626926522808e-06,
      "loss": 0.0873,
      "step": 12193
    },
    {
      "epoch": 0.6296927446423961,
      "grad_norm": 8.883302688598633,
      "learning_rate": 3.072870844759862e-06,
      "loss": 0.0926,
      "step": 12194
    },
    {
      "epoch": 0.6297443841982959,
      "grad_norm": 10.830636024475098,
      "learning_rate": 3.0721148147481892e-06,
      "loss": 0.0861,
      "step": 12195
    },
    {
      "epoch": 0.6297960237541957,
      "grad_norm": 2.887648105621338,
      "learning_rate": 3.071358836508092e-06,
      "loss": 0.029,
      "step": 12196
    },
    {
      "epoch": 0.6298476633100956,
      "grad_norm": 1.5951558351516724,
      "learning_rate": 3.07060291005987e-06,
      "loss": 0.0143,
      "step": 12197
    },
    {
      "epoch": 0.6298993028659954,
      "grad_norm": 2.398500442504883,
      "learning_rate": 3.0698470354238217e-06,
      "loss": 0.0334,
      "step": 12198
    },
    {
      "epoch": 0.6299509424218952,
      "grad_norm": 3.0299086570739746,
      "learning_rate": 3.06909121262025e-06,
      "loss": 0.05,
      "step": 12199
    },
    {
      "epoch": 0.630002581977795,
      "grad_norm": 2.148231267929077,
      "learning_rate": 3.0683354416694483e-06,
      "loss": 0.0241,
      "step": 12200
    },
    {
      "epoch": 0.6300542215336948,
      "grad_norm": 3.1891653537750244,
      "learning_rate": 3.0675797225917126e-06,
      "loss": 0.0522,
      "step": 12201
    },
    {
      "epoch": 0.6301058610895947,
      "grad_norm": 3.749124050140381,
      "learning_rate": 3.0668240554073383e-06,
      "loss": 0.0628,
      "step": 12202
    },
    {
      "epoch": 0.6301575006454945,
      "grad_norm": 1.2573750019073486,
      "learning_rate": 3.066068440136617e-06,
      "loss": 0.0119,
      "step": 12203
    },
    {
      "epoch": 0.6302091402013943,
      "grad_norm": 9.189210891723633,
      "learning_rate": 3.0653128767998403e-06,
      "loss": 0.0492,
      "step": 12204
    },
    {
      "epoch": 0.6302607797572941,
      "grad_norm": 3.3972837924957275,
      "learning_rate": 3.064557365417297e-06,
      "loss": 0.0431,
      "step": 12205
    },
    {
      "epoch": 0.6303124193131939,
      "grad_norm": 7.546717166900635,
      "learning_rate": 3.0638019060092777e-06,
      "loss": 0.0489,
      "step": 12206
    },
    {
      "epoch": 0.6303640588690937,
      "grad_norm": 1.592820644378662,
      "learning_rate": 3.0630464985960703e-06,
      "loss": 0.0244,
      "step": 12207
    },
    {
      "epoch": 0.6304156984249936,
      "grad_norm": 3.2142255306243896,
      "learning_rate": 3.0622911431979576e-06,
      "loss": 0.0554,
      "step": 12208
    },
    {
      "epoch": 0.6304673379808934,
      "grad_norm": 5.587131500244141,
      "learning_rate": 3.061535839835225e-06,
      "loss": 0.0731,
      "step": 12209
    },
    {
      "epoch": 0.6305189775367932,
      "grad_norm": 4.910318374633789,
      "learning_rate": 3.060780588528157e-06,
      "loss": 0.0558,
      "step": 12210
    },
    {
      "epoch": 0.630570617092693,
      "grad_norm": 10.140238761901855,
      "learning_rate": 3.0600253892970333e-06,
      "loss": 0.052,
      "step": 12211
    },
    {
      "epoch": 0.6306222566485928,
      "grad_norm": 2.4577949047088623,
      "learning_rate": 3.0592702421621335e-06,
      "loss": 0.0276,
      "step": 12212
    },
    {
      "epoch": 0.6306738962044927,
      "grad_norm": 9.080822944641113,
      "learning_rate": 3.058515147143739e-06,
      "loss": 0.0655,
      "step": 12213
    },
    {
      "epoch": 0.6307255357603925,
      "grad_norm": 2.62825608253479,
      "learning_rate": 3.057760104262127e-06,
      "loss": 0.0223,
      "step": 12214
    },
    {
      "epoch": 0.6307771753162923,
      "grad_norm": 2.128244638442993,
      "learning_rate": 3.057005113537571e-06,
      "loss": 0.0196,
      "step": 12215
    },
    {
      "epoch": 0.6308288148721921,
      "grad_norm": 2.009467124938965,
      "learning_rate": 3.056250174990347e-06,
      "loss": 0.017,
      "step": 12216
    },
    {
      "epoch": 0.6308804544280919,
      "grad_norm": 5.480401992797852,
      "learning_rate": 3.0554952886407267e-06,
      "loss": 0.0406,
      "step": 12217
    },
    {
      "epoch": 0.6309320939839917,
      "grad_norm": 2.1045033931732178,
      "learning_rate": 3.0547404545089852e-06,
      "loss": 0.0324,
      "step": 12218
    },
    {
      "epoch": 0.6309837335398916,
      "grad_norm": 4.774508953094482,
      "learning_rate": 3.0539856726153906e-06,
      "loss": 0.0587,
      "step": 12219
    },
    {
      "epoch": 0.6310353730957914,
      "grad_norm": 2.2542917728424072,
      "learning_rate": 3.053230942980212e-06,
      "loss": 0.0301,
      "step": 12220
    },
    {
      "epoch": 0.6310870126516912,
      "grad_norm": 5.292320251464844,
      "learning_rate": 3.0524762656237184e-06,
      "loss": 0.0759,
      "step": 12221
    },
    {
      "epoch": 0.631138652207591,
      "grad_norm": 4.320824146270752,
      "learning_rate": 3.051721640566173e-06,
      "loss": 0.0461,
      "step": 12222
    },
    {
      "epoch": 0.6311902917634908,
      "grad_norm": 6.354045867919922,
      "learning_rate": 3.050967067827843e-06,
      "loss": 0.0767,
      "step": 12223
    },
    {
      "epoch": 0.6312419313193907,
      "grad_norm": 11.335355758666992,
      "learning_rate": 3.05021254742899e-06,
      "loss": 0.1838,
      "step": 12224
    },
    {
      "epoch": 0.6312935708752905,
      "grad_norm": 13.509820938110352,
      "learning_rate": 3.0494580793898785e-06,
      "loss": 0.0945,
      "step": 12225
    },
    {
      "epoch": 0.6313452104311903,
      "grad_norm": 1.1937280893325806,
      "learning_rate": 3.048703663730767e-06,
      "loss": 0.0097,
      "step": 12226
    },
    {
      "epoch": 0.6313968499870901,
      "grad_norm": 7.217153549194336,
      "learning_rate": 3.0479493004719153e-06,
      "loss": 0.072,
      "step": 12227
    },
    {
      "epoch": 0.6314484895429899,
      "grad_norm": 1.0211222171783447,
      "learning_rate": 3.047194989633581e-06,
      "loss": 0.0117,
      "step": 12228
    },
    {
      "epoch": 0.6315001290988898,
      "grad_norm": 4.04384183883667,
      "learning_rate": 3.0464407312360204e-06,
      "loss": 0.0499,
      "step": 12229
    },
    {
      "epoch": 0.6315517686547896,
      "grad_norm": 5.372827529907227,
      "learning_rate": 3.0456865252994885e-06,
      "loss": 0.0432,
      "step": 12230
    },
    {
      "epoch": 0.6316034082106894,
      "grad_norm": 4.177854537963867,
      "learning_rate": 3.044932371844237e-06,
      "loss": 0.0375,
      "step": 12231
    },
    {
      "epoch": 0.6316550477665892,
      "grad_norm": 6.9434709548950195,
      "learning_rate": 3.044178270890522e-06,
      "loss": 0.0725,
      "step": 12232
    },
    {
      "epoch": 0.631706687322489,
      "grad_norm": 2.790839910507202,
      "learning_rate": 3.043424222458591e-06,
      "loss": 0.0155,
      "step": 12233
    },
    {
      "epoch": 0.6317583268783888,
      "grad_norm": 7.751299858093262,
      "learning_rate": 3.042670226568695e-06,
      "loss": 0.0611,
      "step": 12234
    },
    {
      "epoch": 0.6318099664342887,
      "grad_norm": 1.2194184064865112,
      "learning_rate": 3.04191628324108e-06,
      "loss": 0.0094,
      "step": 12235
    },
    {
      "epoch": 0.6318616059901885,
      "grad_norm": 5.327554225921631,
      "learning_rate": 3.0411623924959943e-06,
      "loss": 0.0496,
      "step": 12236
    },
    {
      "epoch": 0.6319132455460883,
      "grad_norm": 2.294854164123535,
      "learning_rate": 3.040408554353681e-06,
      "loss": 0.0101,
      "step": 12237
    },
    {
      "epoch": 0.6319648851019881,
      "grad_norm": 2.536081075668335,
      "learning_rate": 3.039654768834386e-06,
      "loss": 0.0268,
      "step": 12238
    },
    {
      "epoch": 0.6320165246578879,
      "grad_norm": 12.39173698425293,
      "learning_rate": 3.0389010359583495e-06,
      "loss": 0.0943,
      "step": 12239
    },
    {
      "epoch": 0.6320681642137878,
      "grad_norm": 1.9061007499694824,
      "learning_rate": 3.038147355745814e-06,
      "loss": 0.0205,
      "step": 12240
    },
    {
      "epoch": 0.6321198037696876,
      "grad_norm": 3.8335349559783936,
      "learning_rate": 3.037393728217018e-06,
      "loss": 0.0154,
      "step": 12241
    },
    {
      "epoch": 0.6321714433255874,
      "grad_norm": 8.22493839263916,
      "learning_rate": 3.0366401533921997e-06,
      "loss": 0.0484,
      "step": 12242
    },
    {
      "epoch": 0.6322230828814872,
      "grad_norm": 4.974142074584961,
      "learning_rate": 3.0358866312915957e-06,
      "loss": 0.0717,
      "step": 12243
    },
    {
      "epoch": 0.632274722437387,
      "grad_norm": 3.052281141281128,
      "learning_rate": 3.035133161935439e-06,
      "loss": 0.0438,
      "step": 12244
    },
    {
      "epoch": 0.6323263619932868,
      "grad_norm": 1.9948229789733887,
      "learning_rate": 3.034379745343966e-06,
      "loss": 0.0246,
      "step": 12245
    },
    {
      "epoch": 0.6323780015491867,
      "grad_norm": 1.982298731803894,
      "learning_rate": 3.033626381537409e-06,
      "loss": 0.017,
      "step": 12246
    },
    {
      "epoch": 0.6324296411050865,
      "grad_norm": 5.309435844421387,
      "learning_rate": 3.0328730705359988e-06,
      "loss": 0.0573,
      "step": 12247
    },
    {
      "epoch": 0.6324812806609863,
      "grad_norm": 6.540038585662842,
      "learning_rate": 3.0321198123599636e-06,
      "loss": 0.0414,
      "step": 12248
    },
    {
      "epoch": 0.6325329202168861,
      "grad_norm": 3.5378730297088623,
      "learning_rate": 3.031366607029532e-06,
      "loss": 0.0643,
      "step": 12249
    },
    {
      "epoch": 0.6325845597727859,
      "grad_norm": 2.575282335281372,
      "learning_rate": 3.030613454564929e-06,
      "loss": 0.0169,
      "step": 12250
    },
    {
      "epoch": 0.6326361993286858,
      "grad_norm": 5.055859088897705,
      "learning_rate": 3.0298603549863842e-06,
      "loss": 0.0784,
      "step": 12251
    },
    {
      "epoch": 0.6326878388845856,
      "grad_norm": 6.073066234588623,
      "learning_rate": 3.0291073083141177e-06,
      "loss": 0.0321,
      "step": 12252
    },
    {
      "epoch": 0.6327394784404854,
      "grad_norm": 6.485296249389648,
      "learning_rate": 3.0283543145683525e-06,
      "loss": 0.0544,
      "step": 12253
    },
    {
      "epoch": 0.6327911179963852,
      "grad_norm": 2.748805284500122,
      "learning_rate": 3.027601373769311e-06,
      "loss": 0.0291,
      "step": 12254
    },
    {
      "epoch": 0.632842757552285,
      "grad_norm": 1.4160300493240356,
      "learning_rate": 3.0268484859372106e-06,
      "loss": 0.017,
      "step": 12255
    },
    {
      "epoch": 0.6328943971081848,
      "grad_norm": 5.961915493011475,
      "learning_rate": 3.02609565109227e-06,
      "loss": 0.0465,
      "step": 12256
    },
    {
      "epoch": 0.6329460366640847,
      "grad_norm": 3.010951042175293,
      "learning_rate": 3.025342869254707e-06,
      "loss": 0.0306,
      "step": 12257
    },
    {
      "epoch": 0.6329976762199845,
      "grad_norm": 3.560636043548584,
      "learning_rate": 3.024590140444737e-06,
      "loss": 0.0321,
      "step": 12258
    },
    {
      "epoch": 0.6330493157758843,
      "grad_norm": 3.050971508026123,
      "learning_rate": 3.0238374646825723e-06,
      "loss": 0.0197,
      "step": 12259
    },
    {
      "epoch": 0.6331009553317841,
      "grad_norm": 3.0623035430908203,
      "learning_rate": 3.0230848419884263e-06,
      "loss": 0.0402,
      "step": 12260
    },
    {
      "epoch": 0.6331525948876839,
      "grad_norm": 7.588770389556885,
      "learning_rate": 3.0223322723825095e-06,
      "loss": 0.0567,
      "step": 12261
    },
    {
      "epoch": 0.6332042344435838,
      "grad_norm": 15.017911911010742,
      "learning_rate": 3.0215797558850325e-06,
      "loss": 0.0643,
      "step": 12262
    },
    {
      "epoch": 0.6332558739994836,
      "grad_norm": 2.3359007835388184,
      "learning_rate": 3.0208272925162008e-06,
      "loss": 0.0266,
      "step": 12263
    },
    {
      "epoch": 0.6333075135553834,
      "grad_norm": 7.589531421661377,
      "learning_rate": 3.0200748822962246e-06,
      "loss": 0.0522,
      "step": 12264
    },
    {
      "epoch": 0.6333591531112832,
      "grad_norm": 17.2775936126709,
      "learning_rate": 3.0193225252453074e-06,
      "loss": 0.1345,
      "step": 12265
    },
    {
      "epoch": 0.633410792667183,
      "grad_norm": 3.4428937435150146,
      "learning_rate": 3.018570221383653e-06,
      "loss": 0.0539,
      "step": 12266
    },
    {
      "epoch": 0.6334624322230828,
      "grad_norm": 4.540652275085449,
      "learning_rate": 3.0178179707314637e-06,
      "loss": 0.0406,
      "step": 12267
    },
    {
      "epoch": 0.6335140717789827,
      "grad_norm": 3.3585331439971924,
      "learning_rate": 3.017065773308941e-06,
      "loss": 0.0306,
      "step": 12268
    },
    {
      "epoch": 0.6335657113348825,
      "grad_norm": 9.608258247375488,
      "learning_rate": 3.016313629136285e-06,
      "loss": 0.1319,
      "step": 12269
    },
    {
      "epoch": 0.6336173508907823,
      "grad_norm": 20.19443130493164,
      "learning_rate": 3.0155615382336913e-06,
      "loss": 0.1871,
      "step": 12270
    },
    {
      "epoch": 0.6336689904466821,
      "grad_norm": 8.463577270507812,
      "learning_rate": 3.0148095006213597e-06,
      "loss": 0.1044,
      "step": 12271
    },
    {
      "epoch": 0.6337206300025819,
      "grad_norm": 3.3071913719177246,
      "learning_rate": 3.0140575163194836e-06,
      "loss": 0.031,
      "step": 12272
    },
    {
      "epoch": 0.6337722695584818,
      "grad_norm": 11.509284019470215,
      "learning_rate": 3.0133055853482583e-06,
      "loss": 0.1016,
      "step": 12273
    },
    {
      "epoch": 0.6338239091143816,
      "grad_norm": 3.126922369003296,
      "learning_rate": 3.0125537077278754e-06,
      "loss": 0.0374,
      "step": 12274
    },
    {
      "epoch": 0.6338755486702814,
      "grad_norm": 3.561108350753784,
      "learning_rate": 3.011801883478525e-06,
      "loss": 0.0177,
      "step": 12275
    },
    {
      "epoch": 0.6339271882261812,
      "grad_norm": 18.064775466918945,
      "learning_rate": 3.0110501126203984e-06,
      "loss": 0.1233,
      "step": 12276
    },
    {
      "epoch": 0.633978827782081,
      "grad_norm": 7.12279748916626,
      "learning_rate": 3.0102983951736816e-06,
      "loss": 0.0387,
      "step": 12277
    },
    {
      "epoch": 0.6340304673379809,
      "grad_norm": 13.900460243225098,
      "learning_rate": 3.0095467311585624e-06,
      "loss": 0.1932,
      "step": 12278
    },
    {
      "epoch": 0.6340821068938807,
      "grad_norm": 5.708982944488525,
      "learning_rate": 3.008795120595227e-06,
      "loss": 0.1079,
      "step": 12279
    },
    {
      "epoch": 0.6341337464497805,
      "grad_norm": 5.924612522125244,
      "learning_rate": 3.0080435635038592e-06,
      "loss": 0.0525,
      "step": 12280
    },
    {
      "epoch": 0.6341853860056803,
      "grad_norm": 5.0825724601745605,
      "learning_rate": 3.007292059904639e-06,
      "loss": 0.0555,
      "step": 12281
    },
    {
      "epoch": 0.6342370255615801,
      "grad_norm": 2.1171159744262695,
      "learning_rate": 3.0065406098177486e-06,
      "loss": 0.0201,
      "step": 12282
    },
    {
      "epoch": 0.6342886651174799,
      "grad_norm": 12.853099822998047,
      "learning_rate": 3.0057892132633678e-06,
      "loss": 0.1102,
      "step": 12283
    },
    {
      "epoch": 0.6343403046733798,
      "grad_norm": 3.977418899536133,
      "learning_rate": 3.0050378702616756e-06,
      "loss": 0.0409,
      "step": 12284
    },
    {
      "epoch": 0.6343919442292796,
      "grad_norm": 7.6528801918029785,
      "learning_rate": 3.004286580832847e-06,
      "loss": 0.0476,
      "step": 12285
    },
    {
      "epoch": 0.6344435837851794,
      "grad_norm": 2.539405345916748,
      "learning_rate": 3.003535344997058e-06,
      "loss": 0.0427,
      "step": 12286
    },
    {
      "epoch": 0.6344952233410792,
      "grad_norm": 2.321396827697754,
      "learning_rate": 3.002784162774483e-06,
      "loss": 0.0533,
      "step": 12287
    },
    {
      "epoch": 0.634546862896979,
      "grad_norm": 11.694437980651855,
      "learning_rate": 3.0020330341852923e-06,
      "loss": 0.0804,
      "step": 12288
    },
    {
      "epoch": 0.6345985024528789,
      "grad_norm": 4.300792217254639,
      "learning_rate": 3.001281959249657e-06,
      "loss": 0.0567,
      "step": 12289
    },
    {
      "epoch": 0.6346501420087787,
      "grad_norm": 3.7006375789642334,
      "learning_rate": 3.0005309379877478e-06,
      "loss": 0.0596,
      "step": 12290
    },
    {
      "epoch": 0.6347017815646785,
      "grad_norm": 2.398914098739624,
      "learning_rate": 2.999779970419734e-06,
      "loss": 0.0326,
      "step": 12291
    },
    {
      "epoch": 0.6347534211205783,
      "grad_norm": 2.925659418106079,
      "learning_rate": 2.99902905656578e-06,
      "loss": 0.0434,
      "step": 12292
    },
    {
      "epoch": 0.6348050606764781,
      "grad_norm": 6.1498870849609375,
      "learning_rate": 2.99827819644605e-06,
      "loss": 0.0486,
      "step": 12293
    },
    {
      "epoch": 0.634856700232378,
      "grad_norm": 15.14373779296875,
      "learning_rate": 2.9975273900807105e-06,
      "loss": 0.2255,
      "step": 12294
    },
    {
      "epoch": 0.6349083397882779,
      "grad_norm": 5.20753288269043,
      "learning_rate": 2.9967766374899214e-06,
      "loss": 0.0797,
      "step": 12295
    },
    {
      "epoch": 0.6349599793441777,
      "grad_norm": 5.4602580070495605,
      "learning_rate": 2.996025938693843e-06,
      "loss": 0.0738,
      "step": 12296
    },
    {
      "epoch": 0.6350116189000775,
      "grad_norm": 5.380947589874268,
      "learning_rate": 2.995275293712637e-06,
      "loss": 0.0465,
      "step": 12297
    },
    {
      "epoch": 0.6350632584559773,
      "grad_norm": 1.7357592582702637,
      "learning_rate": 2.9945247025664612e-06,
      "loss": 0.0295,
      "step": 12298
    },
    {
      "epoch": 0.6351148980118771,
      "grad_norm": 3.3627710342407227,
      "learning_rate": 2.99377416527547e-06,
      "loss": 0.0714,
      "step": 12299
    },
    {
      "epoch": 0.635166537567777,
      "grad_norm": 12.9249267578125,
      "learning_rate": 2.9930236818598196e-06,
      "loss": 0.109,
      "step": 12300
    },
    {
      "epoch": 0.6352181771236768,
      "grad_norm": 7.354869365692139,
      "learning_rate": 2.9922732523396624e-06,
      "loss": 0.0791,
      "step": 12301
    },
    {
      "epoch": 0.6352698166795766,
      "grad_norm": 11.924341201782227,
      "learning_rate": 2.991522876735154e-06,
      "loss": 0.1515,
      "step": 12302
    },
    {
      "epoch": 0.6353214562354764,
      "grad_norm": 1.5937923192977905,
      "learning_rate": 2.9907725550664395e-06,
      "loss": 0.0251,
      "step": 12303
    },
    {
      "epoch": 0.6353730957913762,
      "grad_norm": 3.759459972381592,
      "learning_rate": 2.990022287353672e-06,
      "loss": 0.0347,
      "step": 12304
    },
    {
      "epoch": 0.6354247353472761,
      "grad_norm": 4.000202655792236,
      "learning_rate": 2.989272073617e-06,
      "loss": 0.0429,
      "step": 12305
    },
    {
      "epoch": 0.6354763749031759,
      "grad_norm": 2.6872761249542236,
      "learning_rate": 2.9885219138765665e-06,
      "loss": 0.0374,
      "step": 12306
    },
    {
      "epoch": 0.6355280144590757,
      "grad_norm": 4.718688488006592,
      "learning_rate": 2.9877718081525174e-06,
      "loss": 0.0567,
      "step": 12307
    },
    {
      "epoch": 0.6355796540149755,
      "grad_norm": 7.524980068206787,
      "learning_rate": 2.9870217564649966e-06,
      "loss": 0.1323,
      "step": 12308
    },
    {
      "epoch": 0.6356312935708753,
      "grad_norm": 1.2199901342391968,
      "learning_rate": 2.986271758834149e-06,
      "loss": 0.0234,
      "step": 12309
    },
    {
      "epoch": 0.6356829331267752,
      "grad_norm": 10.058489799499512,
      "learning_rate": 2.9855218152801092e-06,
      "loss": 0.1341,
      "step": 12310
    },
    {
      "epoch": 0.635734572682675,
      "grad_norm": 0.7305796146392822,
      "learning_rate": 2.9847719258230195e-06,
      "loss": 0.0038,
      "step": 12311
    },
    {
      "epoch": 0.6357862122385748,
      "grad_norm": 2.9911856651306152,
      "learning_rate": 2.9840220904830176e-06,
      "loss": 0.0429,
      "step": 12312
    },
    {
      "epoch": 0.6358378517944746,
      "grad_norm": 6.261666774749756,
      "learning_rate": 2.9832723092802405e-06,
      "loss": 0.0275,
      "step": 12313
    },
    {
      "epoch": 0.6358894913503744,
      "grad_norm": 5.712886810302734,
      "learning_rate": 2.98252258223482e-06,
      "loss": 0.0719,
      "step": 12314
    },
    {
      "epoch": 0.6359411309062742,
      "grad_norm": 6.12447452545166,
      "learning_rate": 2.9817729093668905e-06,
      "loss": 0.0894,
      "step": 12315
    },
    {
      "epoch": 0.6359927704621741,
      "grad_norm": 5.419749736785889,
      "learning_rate": 2.9810232906965875e-06,
      "loss": 0.0538,
      "step": 12316
    },
    {
      "epoch": 0.6360444100180739,
      "grad_norm": 3.5306012630462646,
      "learning_rate": 2.980273726244034e-06,
      "loss": 0.0718,
      "step": 12317
    },
    {
      "epoch": 0.6360960495739737,
      "grad_norm": 8.459571838378906,
      "learning_rate": 2.9795242160293647e-06,
      "loss": 0.075,
      "step": 12318
    },
    {
      "epoch": 0.6361476891298735,
      "grad_norm": 3.7670488357543945,
      "learning_rate": 2.978774760072705e-06,
      "loss": 0.0519,
      "step": 12319
    },
    {
      "epoch": 0.6361993286857733,
      "grad_norm": 3.2197353839874268,
      "learning_rate": 2.978025358394182e-06,
      "loss": 0.0304,
      "step": 12320
    },
    {
      "epoch": 0.6362509682416732,
      "grad_norm": 2.715548515319824,
      "learning_rate": 2.9772760110139186e-06,
      "loss": 0.0556,
      "step": 12321
    },
    {
      "epoch": 0.636302607797573,
      "grad_norm": 2.8622140884399414,
      "learning_rate": 2.9765267179520375e-06,
      "loss": 0.0528,
      "step": 12322
    },
    {
      "epoch": 0.6363542473534728,
      "grad_norm": 4.330197334289551,
      "learning_rate": 2.9757774792286613e-06,
      "loss": 0.0731,
      "step": 12323
    },
    {
      "epoch": 0.6364058869093726,
      "grad_norm": 4.645443439483643,
      "learning_rate": 2.975028294863912e-06,
      "loss": 0.0479,
      "step": 12324
    },
    {
      "epoch": 0.6364575264652724,
      "grad_norm": 3.126258373260498,
      "learning_rate": 2.9742791648779046e-06,
      "loss": 0.0382,
      "step": 12325
    },
    {
      "epoch": 0.6365091660211722,
      "grad_norm": 2.1982908248901367,
      "learning_rate": 2.9735300892907586e-06,
      "loss": 0.0236,
      "step": 12326
    },
    {
      "epoch": 0.6365608055770721,
      "grad_norm": 7.716640949249268,
      "learning_rate": 2.97278106812259e-06,
      "loss": 0.0507,
      "step": 12327
    },
    {
      "epoch": 0.6366124451329719,
      "grad_norm": 2.3053853511810303,
      "learning_rate": 2.972032101393511e-06,
      "loss": 0.0284,
      "step": 12328
    },
    {
      "epoch": 0.6366640846888717,
      "grad_norm": 10.724300384521484,
      "learning_rate": 2.9712831891236348e-06,
      "loss": 0.1645,
      "step": 12329
    },
    {
      "epoch": 0.6367157242447715,
      "grad_norm": 5.905523777008057,
      "learning_rate": 2.970534331333075e-06,
      "loss": 0.0581,
      "step": 12330
    },
    {
      "epoch": 0.6367673638006713,
      "grad_norm": 2.377410411834717,
      "learning_rate": 2.9697855280419404e-06,
      "loss": 0.0292,
      "step": 12331
    },
    {
      "epoch": 0.6368190033565712,
      "grad_norm": 2.7719364166259766,
      "learning_rate": 2.9690367792703377e-06,
      "loss": 0.0449,
      "step": 12332
    },
    {
      "epoch": 0.636870642912471,
      "grad_norm": 3.7220187187194824,
      "learning_rate": 2.968288085038376e-06,
      "loss": 0.0453,
      "step": 12333
    },
    {
      "epoch": 0.6369222824683708,
      "grad_norm": 2.8383445739746094,
      "learning_rate": 2.9675394453661586e-06,
      "loss": 0.0297,
      "step": 12334
    },
    {
      "epoch": 0.6369739220242706,
      "grad_norm": 10.657941818237305,
      "learning_rate": 2.9667908602737945e-06,
      "loss": 0.1576,
      "step": 12335
    },
    {
      "epoch": 0.6370255615801704,
      "grad_norm": 7.386448383331299,
      "learning_rate": 2.9660423297813788e-06,
      "loss": 0.0827,
      "step": 12336
    },
    {
      "epoch": 0.6370772011360702,
      "grad_norm": 1.5849976539611816,
      "learning_rate": 2.9652938539090182e-06,
      "loss": 0.0275,
      "step": 12337
    },
    {
      "epoch": 0.6371288406919701,
      "grad_norm": 3.1713314056396484,
      "learning_rate": 2.9645454326768117e-06,
      "loss": 0.0616,
      "step": 12338
    },
    {
      "epoch": 0.6371804802478699,
      "grad_norm": 15.094063758850098,
      "learning_rate": 2.9637970661048555e-06,
      "loss": 0.1398,
      "step": 12339
    },
    {
      "epoch": 0.6372321198037697,
      "grad_norm": 2.405853271484375,
      "learning_rate": 2.9630487542132463e-06,
      "loss": 0.0242,
      "step": 12340
    },
    {
      "epoch": 0.6372837593596695,
      "grad_norm": 5.0771708488464355,
      "learning_rate": 2.9623004970220796e-06,
      "loss": 0.0504,
      "step": 12341
    },
    {
      "epoch": 0.6373353989155693,
      "grad_norm": 2.998136043548584,
      "learning_rate": 2.961552294551453e-06,
      "loss": 0.0171,
      "step": 12342
    },
    {
      "epoch": 0.6373870384714692,
      "grad_norm": 4.15360164642334,
      "learning_rate": 2.9608041468214527e-06,
      "loss": 0.0518,
      "step": 12343
    },
    {
      "epoch": 0.637438678027369,
      "grad_norm": 2.9075541496276855,
      "learning_rate": 2.9600560538521727e-06,
      "loss": 0.0372,
      "step": 12344
    },
    {
      "epoch": 0.6374903175832688,
      "grad_norm": 5.034331321716309,
      "learning_rate": 2.9593080156637026e-06,
      "loss": 0.0393,
      "step": 12345
    },
    {
      "epoch": 0.6375419571391686,
      "grad_norm": 16.830116271972656,
      "learning_rate": 2.9585600322761304e-06,
      "loss": 0.0722,
      "step": 12346
    },
    {
      "epoch": 0.6375935966950684,
      "grad_norm": 4.04472541809082,
      "learning_rate": 2.9578121037095407e-06,
      "loss": 0.0744,
      "step": 12347
    },
    {
      "epoch": 0.6376452362509683,
      "grad_norm": 4.068057537078857,
      "learning_rate": 2.957064229984019e-06,
      "loss": 0.0331,
      "step": 12348
    },
    {
      "epoch": 0.6376968758068681,
      "grad_norm": 3.8106961250305176,
      "learning_rate": 2.9563164111196518e-06,
      "loss": 0.0515,
      "step": 12349
    },
    {
      "epoch": 0.6377485153627679,
      "grad_norm": 6.873981952667236,
      "learning_rate": 2.955568647136516e-06,
      "loss": 0.0524,
      "step": 12350
    },
    {
      "epoch": 0.6378001549186677,
      "grad_norm": 3.900984287261963,
      "learning_rate": 2.9548209380546955e-06,
      "loss": 0.0433,
      "step": 12351
    },
    {
      "epoch": 0.6378517944745675,
      "grad_norm": 2.3585708141326904,
      "learning_rate": 2.9540732838942677e-06,
      "loss": 0.0292,
      "step": 12352
    },
    {
      "epoch": 0.6379034340304673,
      "grad_norm": 7.490009307861328,
      "learning_rate": 2.953325684675313e-06,
      "loss": 0.057,
      "step": 12353
    },
    {
      "epoch": 0.6379550735863672,
      "grad_norm": 1.5986741781234741,
      "learning_rate": 2.9525781404179043e-06,
      "loss": 0.0127,
      "step": 12354
    },
    {
      "epoch": 0.638006713142267,
      "grad_norm": 6.087902069091797,
      "learning_rate": 2.9518306511421158e-06,
      "loss": 0.0443,
      "step": 12355
    },
    {
      "epoch": 0.6380583526981668,
      "grad_norm": 2.6724038124084473,
      "learning_rate": 2.951083216868024e-06,
      "loss": 0.0545,
      "step": 12356
    },
    {
      "epoch": 0.6381099922540666,
      "grad_norm": 3.4822418689727783,
      "learning_rate": 2.9503358376156987e-06,
      "loss": 0.0467,
      "step": 12357
    },
    {
      "epoch": 0.6381616318099664,
      "grad_norm": 4.376955032348633,
      "learning_rate": 2.9495885134052094e-06,
      "loss": 0.0439,
      "step": 12358
    },
    {
      "epoch": 0.6382132713658663,
      "grad_norm": 3.906958818435669,
      "learning_rate": 2.948841244256626e-06,
      "loss": 0.0711,
      "step": 12359
    },
    {
      "epoch": 0.6382649109217661,
      "grad_norm": 5.61776876449585,
      "learning_rate": 2.948094030190016e-06,
      "loss": 0.0562,
      "step": 12360
    },
    {
      "epoch": 0.6383165504776659,
      "grad_norm": 5.650981903076172,
      "learning_rate": 2.9473468712254438e-06,
      "loss": 0.0361,
      "step": 12361
    },
    {
      "epoch": 0.6383681900335657,
      "grad_norm": 10.182510375976562,
      "learning_rate": 2.946599767382973e-06,
      "loss": 0.1063,
      "step": 12362
    },
    {
      "epoch": 0.6384198295894655,
      "grad_norm": 6.738576412200928,
      "learning_rate": 2.9458527186826686e-06,
      "loss": 0.0759,
      "step": 12363
    },
    {
      "epoch": 0.6384714691453653,
      "grad_norm": 10.167048454284668,
      "learning_rate": 2.945105725144592e-06,
      "loss": 0.0871,
      "step": 12364
    },
    {
      "epoch": 0.6385231087012652,
      "grad_norm": 2.761223554611206,
      "learning_rate": 2.9443587867888006e-06,
      "loss": 0.0438,
      "step": 12365
    },
    {
      "epoch": 0.638574748257165,
      "grad_norm": 2.8631577491760254,
      "learning_rate": 2.9436119036353545e-06,
      "loss": 0.0237,
      "step": 12366
    },
    {
      "epoch": 0.6386263878130648,
      "grad_norm": 3.682485580444336,
      "learning_rate": 2.9428650757043086e-06,
      "loss": 0.063,
      "step": 12367
    },
    {
      "epoch": 0.6386780273689646,
      "grad_norm": 6.253869533538818,
      "learning_rate": 2.942118303015724e-06,
      "loss": 0.1076,
      "step": 12368
    },
    {
      "epoch": 0.6387296669248644,
      "grad_norm": 3.2721705436706543,
      "learning_rate": 2.9413715855896464e-06,
      "loss": 0.0713,
      "step": 12369
    },
    {
      "epoch": 0.6387813064807643,
      "grad_norm": 2.610896110534668,
      "learning_rate": 2.9406249234461337e-06,
      "loss": 0.0334,
      "step": 12370
    },
    {
      "epoch": 0.6388329460366641,
      "grad_norm": 4.485632419586182,
      "learning_rate": 2.9398783166052357e-06,
      "loss": 0.0267,
      "step": 12371
    },
    {
      "epoch": 0.6388845855925639,
      "grad_norm": 9.928370475769043,
      "learning_rate": 2.939131765087001e-06,
      "loss": 0.0447,
      "step": 12372
    },
    {
      "epoch": 0.6389362251484637,
      "grad_norm": 2.50465989112854,
      "learning_rate": 2.938385268911478e-06,
      "loss": 0.0333,
      "step": 12373
    },
    {
      "epoch": 0.6389878647043635,
      "grad_norm": 1.6622047424316406,
      "learning_rate": 2.9376388280987122e-06,
      "loss": 0.02,
      "step": 12374
    },
    {
      "epoch": 0.6390395042602633,
      "grad_norm": 4.75626802444458,
      "learning_rate": 2.936892442668753e-06,
      "loss": 0.0684,
      "step": 12375
    },
    {
      "epoch": 0.6390911438161632,
      "grad_norm": 8.785540580749512,
      "learning_rate": 2.936146112641637e-06,
      "loss": 0.1313,
      "step": 12376
    },
    {
      "epoch": 0.639142783372063,
      "grad_norm": 2.8172519207000732,
      "learning_rate": 2.935399838037411e-06,
      "loss": 0.0482,
      "step": 12377
    },
    {
      "epoch": 0.6391944229279628,
      "grad_norm": 23.147655487060547,
      "learning_rate": 2.934653618876114e-06,
      "loss": 0.1549,
      "step": 12378
    },
    {
      "epoch": 0.6392460624838626,
      "grad_norm": 15.412216186523438,
      "learning_rate": 2.933907455177786e-06,
      "loss": 0.0916,
      "step": 12379
    },
    {
      "epoch": 0.6392977020397624,
      "grad_norm": 6.934510231018066,
      "learning_rate": 2.933161346962463e-06,
      "loss": 0.054,
      "step": 12380
    },
    {
      "epoch": 0.6393493415956623,
      "grad_norm": 9.35827922821045,
      "learning_rate": 2.9324152942501804e-06,
      "loss": 0.0414,
      "step": 12381
    },
    {
      "epoch": 0.6394009811515621,
      "grad_norm": 3.2501556873321533,
      "learning_rate": 2.9316692970609776e-06,
      "loss": 0.0332,
      "step": 12382
    },
    {
      "epoch": 0.6394526207074619,
      "grad_norm": 2.192112445831299,
      "learning_rate": 2.9309233554148813e-06,
      "loss": 0.0292,
      "step": 12383
    },
    {
      "epoch": 0.6395042602633617,
      "grad_norm": 8.258064270019531,
      "learning_rate": 2.9301774693319275e-06,
      "loss": 0.1167,
      "step": 12384
    },
    {
      "epoch": 0.6395558998192615,
      "grad_norm": 7.1631083488464355,
      "learning_rate": 2.929431638832144e-06,
      "loss": 0.0695,
      "step": 12385
    },
    {
      "epoch": 0.6396075393751613,
      "grad_norm": 4.53215217590332,
      "learning_rate": 2.9286858639355607e-06,
      "loss": 0.0406,
      "step": 12386
    },
    {
      "epoch": 0.6396591789310612,
      "grad_norm": 14.836722373962402,
      "learning_rate": 2.9279401446622035e-06,
      "loss": 0.0995,
      "step": 12387
    },
    {
      "epoch": 0.639710818486961,
      "grad_norm": 17.19609832763672,
      "learning_rate": 2.9271944810320986e-06,
      "loss": 0.1596,
      "step": 12388
    },
    {
      "epoch": 0.6397624580428608,
      "grad_norm": 3.223116397857666,
      "learning_rate": 2.92644887306527e-06,
      "loss": 0.0408,
      "step": 12389
    },
    {
      "epoch": 0.6398140975987606,
      "grad_norm": 13.766746520996094,
      "learning_rate": 2.925703320781742e-06,
      "loss": 0.0572,
      "step": 12390
    },
    {
      "epoch": 0.6398657371546604,
      "grad_norm": 4.299533367156982,
      "learning_rate": 2.924957824201533e-06,
      "loss": 0.0376,
      "step": 12391
    },
    {
      "epoch": 0.6399173767105603,
      "grad_norm": 2.8734984397888184,
      "learning_rate": 2.9242123833446645e-06,
      "loss": 0.0511,
      "step": 12392
    },
    {
      "epoch": 0.6399690162664601,
      "grad_norm": 8.176648139953613,
      "learning_rate": 2.923466998231154e-06,
      "loss": 0.0851,
      "step": 12393
    },
    {
      "epoch": 0.6400206558223599,
      "grad_norm": 4.2865777015686035,
      "learning_rate": 2.922721668881018e-06,
      "loss": 0.0239,
      "step": 12394
    },
    {
      "epoch": 0.6400722953782597,
      "grad_norm": 2.5724706649780273,
      "learning_rate": 2.9219763953142697e-06,
      "loss": 0.0386,
      "step": 12395
    },
    {
      "epoch": 0.6401239349341595,
      "grad_norm": 3.1176271438598633,
      "learning_rate": 2.9212311775509263e-06,
      "loss": 0.0242,
      "step": 12396
    },
    {
      "epoch": 0.6401755744900594,
      "grad_norm": 5.10321569442749,
      "learning_rate": 2.920486015610999e-06,
      "loss": 0.0523,
      "step": 12397
    },
    {
      "epoch": 0.6402272140459592,
      "grad_norm": 3.796454429626465,
      "learning_rate": 2.919740909514498e-06,
      "loss": 0.0778,
      "step": 12398
    },
    {
      "epoch": 0.640278853601859,
      "grad_norm": 4.554867267608643,
      "learning_rate": 2.91899585928143e-06,
      "loss": 0.0814,
      "step": 12399
    },
    {
      "epoch": 0.6403304931577588,
      "grad_norm": 1.7349932193756104,
      "learning_rate": 2.9182508649318063e-06,
      "loss": 0.0299,
      "step": 12400
    },
    {
      "epoch": 0.6403821327136586,
      "grad_norm": 6.786896228790283,
      "learning_rate": 2.9175059264856338e-06,
      "loss": 0.0917,
      "step": 12401
    },
    {
      "epoch": 0.6404337722695584,
      "grad_norm": 1.3641557693481445,
      "learning_rate": 2.9167610439629126e-06,
      "loss": 0.0201,
      "step": 12402
    },
    {
      "epoch": 0.6404854118254583,
      "grad_norm": 2.3581066131591797,
      "learning_rate": 2.9160162173836483e-06,
      "loss": 0.049,
      "step": 12403
    },
    {
      "epoch": 0.6405370513813581,
      "grad_norm": 4.435232162475586,
      "learning_rate": 2.9152714467678424e-06,
      "loss": 0.0719,
      "step": 12404
    },
    {
      "epoch": 0.6405886909372579,
      "grad_norm": 5.435920238494873,
      "learning_rate": 2.9145267321354954e-06,
      "loss": 0.0459,
      "step": 12405
    },
    {
      "epoch": 0.6406403304931577,
      "grad_norm": 9.55729866027832,
      "learning_rate": 2.9137820735066045e-06,
      "loss": 0.1071,
      "step": 12406
    },
    {
      "epoch": 0.6406919700490575,
      "grad_norm": 16.62078094482422,
      "learning_rate": 2.913037470901169e-06,
      "loss": 0.0795,
      "step": 12407
    },
    {
      "epoch": 0.6407436096049574,
      "grad_norm": 2.3594915866851807,
      "learning_rate": 2.9122929243391827e-06,
      "loss": 0.0481,
      "step": 12408
    },
    {
      "epoch": 0.6407952491608572,
      "grad_norm": 3.294724702835083,
      "learning_rate": 2.9115484338406406e-06,
      "loss": 0.047,
      "step": 12409
    },
    {
      "epoch": 0.640846888716757,
      "grad_norm": 7.092392921447754,
      "learning_rate": 2.9108039994255353e-06,
      "loss": 0.0806,
      "step": 12410
    },
    {
      "epoch": 0.6408985282726568,
      "grad_norm": 12.148018836975098,
      "learning_rate": 2.9100596211138576e-06,
      "loss": 0.107,
      "step": 12411
    },
    {
      "epoch": 0.6409501678285566,
      "grad_norm": 4.787096977233887,
      "learning_rate": 2.9093152989255995e-06,
      "loss": 0.0389,
      "step": 12412
    },
    {
      "epoch": 0.6410018073844564,
      "grad_norm": 10.502293586730957,
      "learning_rate": 2.9085710328807427e-06,
      "loss": 0.0727,
      "step": 12413
    },
    {
      "epoch": 0.6410534469403563,
      "grad_norm": 9.796573638916016,
      "learning_rate": 2.9078268229992795e-06,
      "loss": 0.0325,
      "step": 12414
    },
    {
      "epoch": 0.6411050864962561,
      "grad_norm": 3.834418773651123,
      "learning_rate": 2.9070826693011954e-06,
      "loss": 0.0602,
      "step": 12415
    },
    {
      "epoch": 0.6411567260521559,
      "grad_norm": 1.3370620012283325,
      "learning_rate": 2.90633857180647e-06,
      "loss": 0.0272,
      "step": 12416
    },
    {
      "epoch": 0.6412083656080557,
      "grad_norm": 7.528456687927246,
      "learning_rate": 2.9055945305350873e-06,
      "loss": 0.1017,
      "step": 12417
    },
    {
      "epoch": 0.6412600051639555,
      "grad_norm": 4.274728298187256,
      "learning_rate": 2.904850545507026e-06,
      "loss": 0.0422,
      "step": 12418
    },
    {
      "epoch": 0.6413116447198554,
      "grad_norm": 4.7762064933776855,
      "learning_rate": 2.9041066167422704e-06,
      "loss": 0.0686,
      "step": 12419
    },
    {
      "epoch": 0.6413632842757552,
      "grad_norm": 5.071956157684326,
      "learning_rate": 2.9033627442607925e-06,
      "loss": 0.0801,
      "step": 12420
    },
    {
      "epoch": 0.641414923831655,
      "grad_norm": 2.9407124519348145,
      "learning_rate": 2.9026189280825713e-06,
      "loss": 0.0483,
      "step": 12421
    },
    {
      "epoch": 0.6414665633875549,
      "grad_norm": 8.727302551269531,
      "learning_rate": 2.9018751682275794e-06,
      "loss": 0.083,
      "step": 12422
    },
    {
      "epoch": 0.6415182029434547,
      "grad_norm": 4.470276832580566,
      "learning_rate": 2.9011314647157905e-06,
      "loss": 0.0779,
      "step": 12423
    },
    {
      "epoch": 0.6415698424993546,
      "grad_norm": 5.03700065612793,
      "learning_rate": 2.9003878175671775e-06,
      "loss": 0.0805,
      "step": 12424
    },
    {
      "epoch": 0.6416214820552544,
      "grad_norm": 3.931124448776245,
      "learning_rate": 2.8996442268017077e-06,
      "loss": 0.044,
      "step": 12425
    },
    {
      "epoch": 0.6416731216111542,
      "grad_norm": 2.8569045066833496,
      "learning_rate": 2.8989006924393516e-06,
      "loss": 0.0665,
      "step": 12426
    },
    {
      "epoch": 0.641724761167054,
      "grad_norm": 3.966498613357544,
      "learning_rate": 2.8981572145000756e-06,
      "loss": 0.0612,
      "step": 12427
    },
    {
      "epoch": 0.6417764007229538,
      "grad_norm": 2.7018465995788574,
      "learning_rate": 2.897413793003845e-06,
      "loss": 0.0568,
      "step": 12428
    },
    {
      "epoch": 0.6418280402788537,
      "grad_norm": 8.061681747436523,
      "learning_rate": 2.8966704279706237e-06,
      "loss": 0.0893,
      "step": 12429
    },
    {
      "epoch": 0.6418796798347535,
      "grad_norm": 5.302611351013184,
      "learning_rate": 2.8959271194203764e-06,
      "loss": 0.0596,
      "step": 12430
    },
    {
      "epoch": 0.6419313193906533,
      "grad_norm": 4.005281925201416,
      "learning_rate": 2.8951838673730594e-06,
      "loss": 0.0337,
      "step": 12431
    },
    {
      "epoch": 0.6419829589465531,
      "grad_norm": 1.4221675395965576,
      "learning_rate": 2.894440671848632e-06,
      "loss": 0.02,
      "step": 12432
    },
    {
      "epoch": 0.6420345985024529,
      "grad_norm": 14.640641212463379,
      "learning_rate": 2.893697532867057e-06,
      "loss": 0.1403,
      "step": 12433
    },
    {
      "epoch": 0.6420862380583527,
      "grad_norm": 3.148710250854492,
      "learning_rate": 2.89295445044829e-06,
      "loss": 0.0474,
      "step": 12434
    },
    {
      "epoch": 0.6421378776142526,
      "grad_norm": 4.0603251457214355,
      "learning_rate": 2.892211424612281e-06,
      "loss": 0.0493,
      "step": 12435
    },
    {
      "epoch": 0.6421895171701524,
      "grad_norm": 4.221436500549316,
      "learning_rate": 2.891468455378987e-06,
      "loss": 0.0492,
      "step": 12436
    },
    {
      "epoch": 0.6422411567260522,
      "grad_norm": 4.398338794708252,
      "learning_rate": 2.8907255427683582e-06,
      "loss": 0.088,
      "step": 12437
    },
    {
      "epoch": 0.642292796281952,
      "grad_norm": 14.295393943786621,
      "learning_rate": 2.889982686800346e-06,
      "loss": 0.0562,
      "step": 12438
    },
    {
      "epoch": 0.6423444358378518,
      "grad_norm": 6.8151326179504395,
      "learning_rate": 2.8892398874948987e-06,
      "loss": 0.0837,
      "step": 12439
    },
    {
      "epoch": 0.6423960753937517,
      "grad_norm": 1.5719807147979736,
      "learning_rate": 2.8884971448719634e-06,
      "loss": 0.0271,
      "step": 12440
    },
    {
      "epoch": 0.6424477149496515,
      "grad_norm": 8.550725936889648,
      "learning_rate": 2.887754458951485e-06,
      "loss": 0.0722,
      "step": 12441
    },
    {
      "epoch": 0.6424993545055513,
      "grad_norm": 3.271307945251465,
      "learning_rate": 2.8870118297534088e-06,
      "loss": 0.0366,
      "step": 12442
    },
    {
      "epoch": 0.6425509940614511,
      "grad_norm": 3.0518457889556885,
      "learning_rate": 2.8862692572976774e-06,
      "loss": 0.0351,
      "step": 12443
    },
    {
      "epoch": 0.6426026336173509,
      "grad_norm": 2.6046628952026367,
      "learning_rate": 2.8855267416042314e-06,
      "loss": 0.0433,
      "step": 12444
    },
    {
      "epoch": 0.6426542731732507,
      "grad_norm": 2.3913938999176025,
      "learning_rate": 2.8847842826930104e-06,
      "loss": 0.0298,
      "step": 12445
    },
    {
      "epoch": 0.6427059127291506,
      "grad_norm": 2.014664888381958,
      "learning_rate": 2.8840418805839527e-06,
      "loss": 0.0363,
      "step": 12446
    },
    {
      "epoch": 0.6427575522850504,
      "grad_norm": 2.6742873191833496,
      "learning_rate": 2.8832995352969944e-06,
      "loss": 0.0341,
      "step": 12447
    },
    {
      "epoch": 0.6428091918409502,
      "grad_norm": 5.408083915710449,
      "learning_rate": 2.882557246852073e-06,
      "loss": 0.0508,
      "step": 12448
    },
    {
      "epoch": 0.64286083139685,
      "grad_norm": 6.640805244445801,
      "learning_rate": 2.8818150152691176e-06,
      "loss": 0.0526,
      "step": 12449
    },
    {
      "epoch": 0.6429124709527498,
      "grad_norm": 2.9146790504455566,
      "learning_rate": 2.881072840568062e-06,
      "loss": 0.0303,
      "step": 12450
    },
    {
      "epoch": 0.6429641105086497,
      "grad_norm": 2.103519916534424,
      "learning_rate": 2.880330722768836e-06,
      "loss": 0.0264,
      "step": 12451
    },
    {
      "epoch": 0.6430157500645495,
      "grad_norm": 3.353503704071045,
      "learning_rate": 2.8795886618913728e-06,
      "loss": 0.0252,
      "step": 12452
    },
    {
      "epoch": 0.6430673896204493,
      "grad_norm": 2.7695531845092773,
      "learning_rate": 2.8788466579555942e-06,
      "loss": 0.0577,
      "step": 12453
    },
    {
      "epoch": 0.6431190291763491,
      "grad_norm": 6.5872697830200195,
      "learning_rate": 2.8781047109814274e-06,
      "loss": 0.0508,
      "step": 12454
    },
    {
      "epoch": 0.6431706687322489,
      "grad_norm": 3.0319128036499023,
      "learning_rate": 2.8773628209887983e-06,
      "loss": 0.0472,
      "step": 12455
    },
    {
      "epoch": 0.6432223082881487,
      "grad_norm": 3.111309289932251,
      "learning_rate": 2.876620987997629e-06,
      "loss": 0.0476,
      "step": 12456
    },
    {
      "epoch": 0.6432739478440486,
      "grad_norm": 3.021766424179077,
      "learning_rate": 2.87587921202784e-06,
      "loss": 0.0388,
      "step": 12457
    },
    {
      "epoch": 0.6433255873999484,
      "grad_norm": 9.471821784973145,
      "learning_rate": 2.8751374930993525e-06,
      "loss": 0.2426,
      "step": 12458
    },
    {
      "epoch": 0.6433772269558482,
      "grad_norm": 11.481602668762207,
      "learning_rate": 2.874395831232083e-06,
      "loss": 0.0429,
      "step": 12459
    },
    {
      "epoch": 0.643428866511748,
      "grad_norm": 2.869590997695923,
      "learning_rate": 2.873654226445949e-06,
      "loss": 0.0419,
      "step": 12460
    },
    {
      "epoch": 0.6434805060676478,
      "grad_norm": 3.924034357070923,
      "learning_rate": 2.8729126787608665e-06,
      "loss": 0.0403,
      "step": 12461
    },
    {
      "epoch": 0.6435321456235477,
      "grad_norm": 9.515241622924805,
      "learning_rate": 2.8721711881967466e-06,
      "loss": 0.124,
      "step": 12462
    },
    {
      "epoch": 0.6435837851794475,
      "grad_norm": 2.5900495052337646,
      "learning_rate": 2.871429754773506e-06,
      "loss": 0.0483,
      "step": 12463
    },
    {
      "epoch": 0.6436354247353473,
      "grad_norm": 3.0821166038513184,
      "learning_rate": 2.8706883785110496e-06,
      "loss": 0.0615,
      "step": 12464
    },
    {
      "epoch": 0.6436870642912471,
      "grad_norm": 6.5362138748168945,
      "learning_rate": 2.869947059429287e-06,
      "loss": 0.0525,
      "step": 12465
    },
    {
      "epoch": 0.6437387038471469,
      "grad_norm": 3.951205015182495,
      "learning_rate": 2.869205797548129e-06,
      "loss": 0.0639,
      "step": 12466
    },
    {
      "epoch": 0.6437903434030467,
      "grad_norm": 10.990612983703613,
      "learning_rate": 2.868464592887482e-06,
      "loss": 0.185,
      "step": 12467
    },
    {
      "epoch": 0.6438419829589466,
      "grad_norm": 1.9667339324951172,
      "learning_rate": 2.8677234454672466e-06,
      "loss": 0.0277,
      "step": 12468
    },
    {
      "epoch": 0.6438936225148464,
      "grad_norm": 2.9396533966064453,
      "learning_rate": 2.866982355307327e-06,
      "loss": 0.0365,
      "step": 12469
    },
    {
      "epoch": 0.6439452620707462,
      "grad_norm": 4.5610857009887695,
      "learning_rate": 2.8662413224276254e-06,
      "loss": 0.0602,
      "step": 12470
    },
    {
      "epoch": 0.643996901626646,
      "grad_norm": 3.245605945587158,
      "learning_rate": 2.8655003468480415e-06,
      "loss": 0.0455,
      "step": 12471
    },
    {
      "epoch": 0.6440485411825458,
      "grad_norm": 4.026700496673584,
      "learning_rate": 2.8647594285884718e-06,
      "loss": 0.0387,
      "step": 12472
    },
    {
      "epoch": 0.6441001807384457,
      "grad_norm": 4.934284210205078,
      "learning_rate": 2.864018567668815e-06,
      "loss": 0.0869,
      "step": 12473
    },
    {
      "epoch": 0.6441518202943455,
      "grad_norm": 6.816290855407715,
      "learning_rate": 2.863277764108965e-06,
      "loss": 0.048,
      "step": 12474
    },
    {
      "epoch": 0.6442034598502453,
      "grad_norm": 2.8332061767578125,
      "learning_rate": 2.8625370179288157e-06,
      "loss": 0.0295,
      "step": 12475
    },
    {
      "epoch": 0.6442550994061451,
      "grad_norm": 10.544422149658203,
      "learning_rate": 2.86179632914826e-06,
      "loss": 0.0415,
      "step": 12476
    },
    {
      "epoch": 0.6443067389620449,
      "grad_norm": 20.242076873779297,
      "learning_rate": 2.8610556977871872e-06,
      "loss": 0.0565,
      "step": 12477
    },
    {
      "epoch": 0.6443583785179448,
      "grad_norm": 2.3256614208221436,
      "learning_rate": 2.860315123865487e-06,
      "loss": 0.0329,
      "step": 12478
    },
    {
      "epoch": 0.6444100180738446,
      "grad_norm": 3.902959108352661,
      "learning_rate": 2.8595746074030462e-06,
      "loss": 0.0491,
      "step": 12479
    },
    {
      "epoch": 0.6444616576297444,
      "grad_norm": 4.894342422485352,
      "learning_rate": 2.8588341484197512e-06,
      "loss": 0.0419,
      "step": 12480
    },
    {
      "epoch": 0.6445132971856442,
      "grad_norm": 2.633596420288086,
      "learning_rate": 2.8580937469354888e-06,
      "loss": 0.0348,
      "step": 12481
    },
    {
      "epoch": 0.644564936741544,
      "grad_norm": 4.741799831390381,
      "learning_rate": 2.8573534029701366e-06,
      "loss": 0.0805,
      "step": 12482
    },
    {
      "epoch": 0.6446165762974438,
      "grad_norm": 4.984810829162598,
      "learning_rate": 2.8566131165435788e-06,
      "loss": 0.0588,
      "step": 12483
    },
    {
      "epoch": 0.6446682158533437,
      "grad_norm": 2.23140549659729,
      "learning_rate": 2.8558728876756926e-06,
      "loss": 0.0261,
      "step": 12484
    },
    {
      "epoch": 0.6447198554092435,
      "grad_norm": 4.13513708114624,
      "learning_rate": 2.8551327163863627e-06,
      "loss": 0.0464,
      "step": 12485
    },
    {
      "epoch": 0.6447714949651433,
      "grad_norm": 6.324662685394287,
      "learning_rate": 2.8543926026954596e-06,
      "loss": 0.0959,
      "step": 12486
    },
    {
      "epoch": 0.6448231345210431,
      "grad_norm": 1.0637524127960205,
      "learning_rate": 2.8536525466228592e-06,
      "loss": 0.0095,
      "step": 12487
    },
    {
      "epoch": 0.6448747740769429,
      "grad_norm": 1.2884621620178223,
      "learning_rate": 2.852912548188436e-06,
      "loss": 0.018,
      "step": 12488
    },
    {
      "epoch": 0.6449264136328428,
      "grad_norm": 1.42567777633667,
      "learning_rate": 2.8521726074120627e-06,
      "loss": 0.0132,
      "step": 12489
    },
    {
      "epoch": 0.6449780531887426,
      "grad_norm": 4.32729959487915,
      "learning_rate": 2.8514327243136085e-06,
      "loss": 0.0753,
      "step": 12490
    },
    {
      "epoch": 0.6450296927446424,
      "grad_norm": 3.6092629432678223,
      "learning_rate": 2.8506928989129422e-06,
      "loss": 0.0566,
      "step": 12491
    },
    {
      "epoch": 0.6450813323005422,
      "grad_norm": 10.73443603515625,
      "learning_rate": 2.8499531312299323e-06,
      "loss": 0.0619,
      "step": 12492
    },
    {
      "epoch": 0.645132971856442,
      "grad_norm": 10.82341480255127,
      "learning_rate": 2.8492134212844443e-06,
      "loss": 0.116,
      "step": 12493
    },
    {
      "epoch": 0.6451846114123418,
      "grad_norm": 4.7486572265625,
      "learning_rate": 2.8484737690963415e-06,
      "loss": 0.031,
      "step": 12494
    },
    {
      "epoch": 0.6452362509682417,
      "grad_norm": 3.5749969482421875,
      "learning_rate": 2.8477341746854868e-06,
      "loss": 0.0651,
      "step": 12495
    },
    {
      "epoch": 0.6452878905241415,
      "grad_norm": 42.00044631958008,
      "learning_rate": 2.8469946380717444e-06,
      "loss": 0.0953,
      "step": 12496
    },
    {
      "epoch": 0.6453395300800413,
      "grad_norm": 3.8230080604553223,
      "learning_rate": 2.8462551592749666e-06,
      "loss": 0.0625,
      "step": 12497
    },
    {
      "epoch": 0.6453911696359411,
      "grad_norm": 9.11643123626709,
      "learning_rate": 2.8455157383150186e-06,
      "loss": 0.1109,
      "step": 12498
    },
    {
      "epoch": 0.6454428091918409,
      "grad_norm": 3.688450336456299,
      "learning_rate": 2.8447763752117534e-06,
      "loss": 0.067,
      "step": 12499
    },
    {
      "epoch": 0.6454944487477408,
      "grad_norm": 3.8545870780944824,
      "learning_rate": 2.8440370699850293e-06,
      "loss": 0.0794,
      "step": 12500
    },
    {
      "epoch": 0.6455460883036406,
      "grad_norm": 4.4317851066589355,
      "learning_rate": 2.843297822654695e-06,
      "loss": 0.0518,
      "step": 12501
    },
    {
      "epoch": 0.6455977278595404,
      "grad_norm": 9.906557083129883,
      "learning_rate": 2.8425586332406042e-06,
      "loss": 0.0437,
      "step": 12502
    },
    {
      "epoch": 0.6456493674154402,
      "grad_norm": 2.3554494380950928,
      "learning_rate": 2.841819501762608e-06,
      "loss": 0.0528,
      "step": 12503
    },
    {
      "epoch": 0.64570100697134,
      "grad_norm": 3.894406318664551,
      "learning_rate": 2.8410804282405535e-06,
      "loss": 0.0404,
      "step": 12504
    },
    {
      "epoch": 0.6457526465272398,
      "grad_norm": 7.670564651489258,
      "learning_rate": 2.840341412694289e-06,
      "loss": 0.1324,
      "step": 12505
    },
    {
      "epoch": 0.6458042860831397,
      "grad_norm": 10.320457458496094,
      "learning_rate": 2.83960245514366e-06,
      "loss": 0.1262,
      "step": 12506
    },
    {
      "epoch": 0.6458559256390395,
      "grad_norm": 2.8559091091156006,
      "learning_rate": 2.838863555608511e-06,
      "loss": 0.0467,
      "step": 12507
    },
    {
      "epoch": 0.6459075651949393,
      "grad_norm": 4.129083156585693,
      "learning_rate": 2.8381247141086833e-06,
      "loss": 0.0445,
      "step": 12508
    },
    {
      "epoch": 0.6459592047508391,
      "grad_norm": 9.103227615356445,
      "learning_rate": 2.8373859306640176e-06,
      "loss": 0.0875,
      "step": 12509
    },
    {
      "epoch": 0.6460108443067389,
      "grad_norm": 3.316537618637085,
      "learning_rate": 2.836647205294355e-06,
      "loss": 0.0329,
      "step": 12510
    },
    {
      "epoch": 0.6460624838626388,
      "grad_norm": 9.233914375305176,
      "learning_rate": 2.835908538019532e-06,
      "loss": 0.0496,
      "step": 12511
    },
    {
      "epoch": 0.6461141234185386,
      "grad_norm": 4.693800926208496,
      "learning_rate": 2.835169928859385e-06,
      "loss": 0.0702,
      "step": 12512
    },
    {
      "epoch": 0.6461657629744384,
      "grad_norm": 2.3104732036590576,
      "learning_rate": 2.8344313778337486e-06,
      "loss": 0.0259,
      "step": 12513
    },
    {
      "epoch": 0.6462174025303382,
      "grad_norm": 4.741734027862549,
      "learning_rate": 2.833692884962458e-06,
      "loss": 0.0469,
      "step": 12514
    },
    {
      "epoch": 0.646269042086238,
      "grad_norm": 7.764102935791016,
      "learning_rate": 2.8329544502653415e-06,
      "loss": 0.0651,
      "step": 12515
    },
    {
      "epoch": 0.6463206816421379,
      "grad_norm": 5.642404079437256,
      "learning_rate": 2.832216073762228e-06,
      "loss": 0.0413,
      "step": 12516
    },
    {
      "epoch": 0.6463723211980377,
      "grad_norm": 11.180177688598633,
      "learning_rate": 2.83147775547295e-06,
      "loss": 0.0726,
      "step": 12517
    },
    {
      "epoch": 0.6464239607539375,
      "grad_norm": 4.294575214385986,
      "learning_rate": 2.830739495417335e-06,
      "loss": 0.0452,
      "step": 12518
    },
    {
      "epoch": 0.6464756003098373,
      "grad_norm": 5.494909763336182,
      "learning_rate": 2.830001293615204e-06,
      "loss": 0.0889,
      "step": 12519
    },
    {
      "epoch": 0.6465272398657371,
      "grad_norm": 2.08843731880188,
      "learning_rate": 2.8292631500863827e-06,
      "loss": 0.0345,
      "step": 12520
    },
    {
      "epoch": 0.6465788794216369,
      "grad_norm": 3.9814062118530273,
      "learning_rate": 2.8285250648506934e-06,
      "loss": 0.0247,
      "step": 12521
    },
    {
      "epoch": 0.6466305189775368,
      "grad_norm": 2.8946075439453125,
      "learning_rate": 2.827787037927956e-06,
      "loss": 0.0195,
      "step": 12522
    },
    {
      "epoch": 0.6466821585334366,
      "grad_norm": 2.768812417984009,
      "learning_rate": 2.8270490693379905e-06,
      "loss": 0.061,
      "step": 12523
    },
    {
      "epoch": 0.6467337980893364,
      "grad_norm": 2.836608648300171,
      "learning_rate": 2.8263111591006143e-06,
      "loss": 0.0397,
      "step": 12524
    },
    {
      "epoch": 0.6467854376452362,
      "grad_norm": 3.9705965518951416,
      "learning_rate": 2.8255733072356424e-06,
      "loss": 0.0495,
      "step": 12525
    },
    {
      "epoch": 0.646837077201136,
      "grad_norm": 2.733151912689209,
      "learning_rate": 2.8248355137628896e-06,
      "loss": 0.0385,
      "step": 12526
    },
    {
      "epoch": 0.6468887167570359,
      "grad_norm": 3.0576565265655518,
      "learning_rate": 2.824097778702169e-06,
      "loss": 0.0339,
      "step": 12527
    },
    {
      "epoch": 0.6469403563129357,
      "grad_norm": 14.462777137756348,
      "learning_rate": 2.8233601020732915e-06,
      "loss": 0.0887,
      "step": 12528
    },
    {
      "epoch": 0.6469919958688355,
      "grad_norm": 2.3974153995513916,
      "learning_rate": 2.8226224838960688e-06,
      "loss": 0.0326,
      "step": 12529
    },
    {
      "epoch": 0.6470436354247353,
      "grad_norm": 2.3406224250793457,
      "learning_rate": 2.8218849241903025e-06,
      "loss": 0.0288,
      "step": 12530
    },
    {
      "epoch": 0.6470952749806351,
      "grad_norm": 8.332925796508789,
      "learning_rate": 2.8211474229758064e-06,
      "loss": 0.045,
      "step": 12531
    },
    {
      "epoch": 0.647146914536535,
      "grad_norm": 5.53086519241333,
      "learning_rate": 2.8204099802723816e-06,
      "loss": 0.0539,
      "step": 12532
    },
    {
      "epoch": 0.6471985540924348,
      "grad_norm": 20.48463249206543,
      "learning_rate": 2.8196725960998343e-06,
      "loss": 0.0807,
      "step": 12533
    },
    {
      "epoch": 0.6472501936483346,
      "grad_norm": 15.311437606811523,
      "learning_rate": 2.8189352704779625e-06,
      "loss": 0.0562,
      "step": 12534
    },
    {
      "epoch": 0.6473018332042344,
      "grad_norm": 6.388208389282227,
      "learning_rate": 2.8181980034265667e-06,
      "loss": 0.1001,
      "step": 12535
    },
    {
      "epoch": 0.6473534727601342,
      "grad_norm": 3.031709671020508,
      "learning_rate": 2.8174607949654507e-06,
      "loss": 0.0404,
      "step": 12536
    },
    {
      "epoch": 0.647405112316034,
      "grad_norm": 3.5609781742095947,
      "learning_rate": 2.8167236451144064e-06,
      "loss": 0.0397,
      "step": 12537
    },
    {
      "epoch": 0.6474567518719339,
      "grad_norm": 4.994254112243652,
      "learning_rate": 2.8159865538932297e-06,
      "loss": 0.0293,
      "step": 12538
    },
    {
      "epoch": 0.6475083914278337,
      "grad_norm": 6.029503345489502,
      "learning_rate": 2.815249521321717e-06,
      "loss": 0.053,
      "step": 12539
    },
    {
      "epoch": 0.6475600309837335,
      "grad_norm": 2.2238550186157227,
      "learning_rate": 2.8145125474196576e-06,
      "loss": 0.0463,
      "step": 12540
    },
    {
      "epoch": 0.6476116705396333,
      "grad_norm": 20.13422966003418,
      "learning_rate": 2.8137756322068448e-06,
      "loss": 0.1153,
      "step": 12541
    },
    {
      "epoch": 0.6476633100955331,
      "grad_norm": 3.1357648372650146,
      "learning_rate": 2.8130387757030663e-06,
      "loss": 0.0201,
      "step": 12542
    },
    {
      "epoch": 0.647714949651433,
      "grad_norm": 1.4606504440307617,
      "learning_rate": 2.8123019779281114e-06,
      "loss": 0.0184,
      "step": 12543
    },
    {
      "epoch": 0.6477665892073328,
      "grad_norm": 5.2678985595703125,
      "learning_rate": 2.8115652389017634e-06,
      "loss": 0.048,
      "step": 12544
    },
    {
      "epoch": 0.6478182287632326,
      "grad_norm": 3.8213391304016113,
      "learning_rate": 2.8108285586438093e-06,
      "loss": 0.0663,
      "step": 12545
    },
    {
      "epoch": 0.6478698683191324,
      "grad_norm": 4.842336177825928,
      "learning_rate": 2.810091937174031e-06,
      "loss": 0.0615,
      "step": 12546
    },
    {
      "epoch": 0.6479215078750322,
      "grad_norm": 2.3324646949768066,
      "learning_rate": 2.8093553745122117e-06,
      "loss": 0.026,
      "step": 12547
    },
    {
      "epoch": 0.647973147430932,
      "grad_norm": 3.5873489379882812,
      "learning_rate": 2.8086188706781263e-06,
      "loss": 0.0372,
      "step": 12548
    },
    {
      "epoch": 0.6480247869868319,
      "grad_norm": 6.828932285308838,
      "learning_rate": 2.807882425691555e-06,
      "loss": 0.041,
      "step": 12549
    },
    {
      "epoch": 0.6480764265427318,
      "grad_norm": 3.948425054550171,
      "learning_rate": 2.8071460395722762e-06,
      "loss": 0.0299,
      "step": 12550
    },
    {
      "epoch": 0.6481280660986316,
      "grad_norm": 11.651561737060547,
      "learning_rate": 2.8064097123400664e-06,
      "loss": 0.0929,
      "step": 12551
    },
    {
      "epoch": 0.6481797056545314,
      "grad_norm": 2.8749680519104004,
      "learning_rate": 2.805673444014694e-06,
      "loss": 0.0541,
      "step": 12552
    },
    {
      "epoch": 0.6482313452104312,
      "grad_norm": 19.291730880737305,
      "learning_rate": 2.8049372346159338e-06,
      "loss": 0.1039,
      "step": 12553
    },
    {
      "epoch": 0.6482829847663311,
      "grad_norm": 3.5732975006103516,
      "learning_rate": 2.8042010841635526e-06,
      "loss": 0.0726,
      "step": 12554
    },
    {
      "epoch": 0.6483346243222309,
      "grad_norm": 1.589890480041504,
      "learning_rate": 2.8034649926773265e-06,
      "loss": 0.017,
      "step": 12555
    },
    {
      "epoch": 0.6483862638781307,
      "grad_norm": 5.042576789855957,
      "learning_rate": 2.802728960177016e-06,
      "loss": 0.07,
      "step": 12556
    },
    {
      "epoch": 0.6484379034340305,
      "grad_norm": 1.819970726966858,
      "learning_rate": 2.801992986682388e-06,
      "loss": 0.0109,
      "step": 12557
    },
    {
      "epoch": 0.6484895429899303,
      "grad_norm": 4.606026649475098,
      "learning_rate": 2.801257072213207e-06,
      "loss": 0.0364,
      "step": 12558
    },
    {
      "epoch": 0.6485411825458302,
      "grad_norm": 2.653735876083374,
      "learning_rate": 2.800521216789236e-06,
      "loss": 0.0197,
      "step": 12559
    },
    {
      "epoch": 0.64859282210173,
      "grad_norm": 2.226956605911255,
      "learning_rate": 2.7997854204302338e-06,
      "loss": 0.0302,
      "step": 12560
    },
    {
      "epoch": 0.6486444616576298,
      "grad_norm": 10.037842750549316,
      "learning_rate": 2.799049683155961e-06,
      "loss": 0.0307,
      "step": 12561
    },
    {
      "epoch": 0.6486961012135296,
      "grad_norm": 2.758732795715332,
      "learning_rate": 2.7983140049861767e-06,
      "loss": 0.0301,
      "step": 12562
    },
    {
      "epoch": 0.6487477407694294,
      "grad_norm": 6.809382915496826,
      "learning_rate": 2.7975783859406313e-06,
      "loss": 0.0781,
      "step": 12563
    },
    {
      "epoch": 0.6487993803253292,
      "grad_norm": 3.405395269393921,
      "learning_rate": 2.7968428260390844e-06,
      "loss": 0.0512,
      "step": 12564
    },
    {
      "epoch": 0.6488510198812291,
      "grad_norm": 2.7375848293304443,
      "learning_rate": 2.7961073253012875e-06,
      "loss": 0.0332,
      "step": 12565
    },
    {
      "epoch": 0.6489026594371289,
      "grad_norm": 3.9163827896118164,
      "learning_rate": 2.7953718837469934e-06,
      "loss": 0.0478,
      "step": 12566
    },
    {
      "epoch": 0.6489542989930287,
      "grad_norm": 9.780037879943848,
      "learning_rate": 2.7946365013959477e-06,
      "loss": 0.1678,
      "step": 12567
    },
    {
      "epoch": 0.6490059385489285,
      "grad_norm": 3.871857166290283,
      "learning_rate": 2.7939011782678986e-06,
      "loss": 0.0609,
      "step": 12568
    },
    {
      "epoch": 0.6490575781048283,
      "grad_norm": 2.418056011199951,
      "learning_rate": 2.7931659143825984e-06,
      "loss": 0.0259,
      "step": 12569
    },
    {
      "epoch": 0.6491092176607282,
      "grad_norm": 4.301186561584473,
      "learning_rate": 2.7924307097597865e-06,
      "loss": 0.0522,
      "step": 12570
    },
    {
      "epoch": 0.649160857216628,
      "grad_norm": 3.571669816970825,
      "learning_rate": 2.7916955644192067e-06,
      "loss": 0.0467,
      "step": 12571
    },
    {
      "epoch": 0.6492124967725278,
      "grad_norm": 1.3542230129241943,
      "learning_rate": 2.790960478380602e-06,
      "loss": 0.0174,
      "step": 12572
    },
    {
      "epoch": 0.6492641363284276,
      "grad_norm": 4.233685493469238,
      "learning_rate": 2.790225451663712e-06,
      "loss": 0.0479,
      "step": 12573
    },
    {
      "epoch": 0.6493157758843274,
      "grad_norm": 7.603804111480713,
      "learning_rate": 2.7894904842882753e-06,
      "loss": 0.102,
      "step": 12574
    },
    {
      "epoch": 0.6493674154402272,
      "grad_norm": 3.8144338130950928,
      "learning_rate": 2.788755576274028e-06,
      "loss": 0.0334,
      "step": 12575
    },
    {
      "epoch": 0.6494190549961271,
      "grad_norm": 4.033970832824707,
      "learning_rate": 2.788020727640707e-06,
      "loss": 0.0354,
      "step": 12576
    },
    {
      "epoch": 0.6494706945520269,
      "grad_norm": 10.609709739685059,
      "learning_rate": 2.787285938408044e-06,
      "loss": 0.1154,
      "step": 12577
    },
    {
      "epoch": 0.6495223341079267,
      "grad_norm": 3.866792917251587,
      "learning_rate": 2.7865512085957722e-06,
      "loss": 0.0498,
      "step": 12578
    },
    {
      "epoch": 0.6495739736638265,
      "grad_norm": 2.750051736831665,
      "learning_rate": 2.7858165382236223e-06,
      "loss": 0.0374,
      "step": 12579
    },
    {
      "epoch": 0.6496256132197263,
      "grad_norm": 2.464158535003662,
      "learning_rate": 2.785081927311325e-06,
      "loss": 0.0381,
      "step": 12580
    },
    {
      "epoch": 0.6496772527756262,
      "grad_norm": 4.71734619140625,
      "learning_rate": 2.7843473758786033e-06,
      "loss": 0.0601,
      "step": 12581
    },
    {
      "epoch": 0.649728892331526,
      "grad_norm": 4.071814060211182,
      "learning_rate": 2.783612883945183e-06,
      "loss": 0.1126,
      "step": 12582
    },
    {
      "epoch": 0.6497805318874258,
      "grad_norm": 11.197559356689453,
      "learning_rate": 2.7828784515307927e-06,
      "loss": 0.0639,
      "step": 12583
    },
    {
      "epoch": 0.6498321714433256,
      "grad_norm": 4.314314365386963,
      "learning_rate": 2.7821440786551533e-06,
      "loss": 0.0441,
      "step": 12584
    },
    {
      "epoch": 0.6498838109992254,
      "grad_norm": 6.112794399261475,
      "learning_rate": 2.781409765337984e-06,
      "loss": 0.0596,
      "step": 12585
    },
    {
      "epoch": 0.6499354505551252,
      "grad_norm": 6.255725383758545,
      "learning_rate": 2.780675511599005e-06,
      "loss": 0.0395,
      "step": 12586
    },
    {
      "epoch": 0.6499870901110251,
      "grad_norm": 6.82603645324707,
      "learning_rate": 2.7799413174579315e-06,
      "loss": 0.0447,
      "step": 12587
    },
    {
      "epoch": 0.6500387296669249,
      "grad_norm": 4.396553039550781,
      "learning_rate": 2.7792071829344857e-06,
      "loss": 0.0587,
      "step": 12588
    },
    {
      "epoch": 0.6500903692228247,
      "grad_norm": 5.637723445892334,
      "learning_rate": 2.7784731080483775e-06,
      "loss": 0.1019,
      "step": 12589
    },
    {
      "epoch": 0.6501420087787245,
      "grad_norm": 3.8719654083251953,
      "learning_rate": 2.7777390928193198e-06,
      "loss": 0.0732,
      "step": 12590
    },
    {
      "epoch": 0.6501936483346243,
      "grad_norm": 9.379411697387695,
      "learning_rate": 2.777005137267026e-06,
      "loss": 0.157,
      "step": 12591
    },
    {
      "epoch": 0.6502452878905242,
      "grad_norm": 2.5924031734466553,
      "learning_rate": 2.7762712414112038e-06,
      "loss": 0.0369,
      "step": 12592
    },
    {
      "epoch": 0.650296927446424,
      "grad_norm": 10.360581398010254,
      "learning_rate": 2.775537405271562e-06,
      "loss": 0.069,
      "step": 12593
    },
    {
      "epoch": 0.6503485670023238,
      "grad_norm": 2.6808815002441406,
      "learning_rate": 2.7748036288678083e-06,
      "loss": 0.0311,
      "step": 12594
    },
    {
      "epoch": 0.6504002065582236,
      "grad_norm": 3.0079896450042725,
      "learning_rate": 2.7740699122196475e-06,
      "loss": 0.0383,
      "step": 12595
    },
    {
      "epoch": 0.6504518461141234,
      "grad_norm": 1.8827393054962158,
      "learning_rate": 2.7733362553467784e-06,
      "loss": 0.0189,
      "step": 12596
    },
    {
      "epoch": 0.6505034856700233,
      "grad_norm": 15.526881217956543,
      "learning_rate": 2.772602658268908e-06,
      "loss": 0.1406,
      "step": 12597
    },
    {
      "epoch": 0.6505551252259231,
      "grad_norm": 12.237553596496582,
      "learning_rate": 2.7718691210057346e-06,
      "loss": 0.1059,
      "step": 12598
    },
    {
      "epoch": 0.6506067647818229,
      "grad_norm": 5.4453959465026855,
      "learning_rate": 2.7711356435769593e-06,
      "loss": 0.0412,
      "step": 12599
    },
    {
      "epoch": 0.6506584043377227,
      "grad_norm": 2.538275957107544,
      "learning_rate": 2.770402226002274e-06,
      "loss": 0.0186,
      "step": 12600
    },
    {
      "epoch": 0.6507100438936225,
      "grad_norm": 5.195626735687256,
      "learning_rate": 2.769668868301374e-06,
      "loss": 0.0601,
      "step": 12601
    },
    {
      "epoch": 0.6507616834495223,
      "grad_norm": 3.4037997722625732,
      "learning_rate": 2.7689355704939595e-06,
      "loss": 0.0191,
      "step": 12602
    },
    {
      "epoch": 0.6508133230054222,
      "grad_norm": 6.255314350128174,
      "learning_rate": 2.7682023325997163e-06,
      "loss": 0.0524,
      "step": 12603
    },
    {
      "epoch": 0.650864962561322,
      "grad_norm": 2.248440742492676,
      "learning_rate": 2.7674691546383366e-06,
      "loss": 0.0169,
      "step": 12604
    },
    {
      "epoch": 0.6509166021172218,
      "grad_norm": 9.297323226928711,
      "learning_rate": 2.7667360366295094e-06,
      "loss": 0.1227,
      "step": 12605
    },
    {
      "epoch": 0.6509682416731216,
      "grad_norm": 4.472349166870117,
      "learning_rate": 2.766002978592922e-06,
      "loss": 0.0439,
      "step": 12606
    },
    {
      "epoch": 0.6510198812290214,
      "grad_norm": 2.027316093444824,
      "learning_rate": 2.76526998054826e-06,
      "loss": 0.0307,
      "step": 12607
    },
    {
      "epoch": 0.6510715207849213,
      "grad_norm": 2.9478495121002197,
      "learning_rate": 2.7645370425152064e-06,
      "loss": 0.0378,
      "step": 12608
    },
    {
      "epoch": 0.6511231603408211,
      "grad_norm": 5.416861057281494,
      "learning_rate": 2.763804164513447e-06,
      "loss": 0.0312,
      "step": 12609
    },
    {
      "epoch": 0.6511747998967209,
      "grad_norm": 3.2928450107574463,
      "learning_rate": 2.7630713465626568e-06,
      "loss": 0.0527,
      "step": 12610
    },
    {
      "epoch": 0.6512264394526207,
      "grad_norm": 8.687313079833984,
      "learning_rate": 2.7623385886825193e-06,
      "loss": 0.0654,
      "step": 12611
    },
    {
      "epoch": 0.6512780790085205,
      "grad_norm": 3.09938383102417,
      "learning_rate": 2.7616058908927112e-06,
      "loss": 0.0551,
      "step": 12612
    },
    {
      "epoch": 0.6513297185644203,
      "grad_norm": 3.627441167831421,
      "learning_rate": 2.76087325321291e-06,
      "loss": 0.0474,
      "step": 12613
    },
    {
      "epoch": 0.6513813581203202,
      "grad_norm": 5.906589508056641,
      "learning_rate": 2.760140675662786e-06,
      "loss": 0.0664,
      "step": 12614
    },
    {
      "epoch": 0.65143299767622,
      "grad_norm": 17.674617767333984,
      "learning_rate": 2.759408158262012e-06,
      "loss": 0.046,
      "step": 12615
    },
    {
      "epoch": 0.6514846372321198,
      "grad_norm": 11.835156440734863,
      "learning_rate": 2.7586757010302634e-06,
      "loss": 0.0664,
      "step": 12616
    },
    {
      "epoch": 0.6515362767880196,
      "grad_norm": 6.484145164489746,
      "learning_rate": 2.7579433039872096e-06,
      "loss": 0.086,
      "step": 12617
    },
    {
      "epoch": 0.6515879163439194,
      "grad_norm": 11.111468315124512,
      "learning_rate": 2.757210967152513e-06,
      "loss": 0.0524,
      "step": 12618
    },
    {
      "epoch": 0.6516395558998193,
      "grad_norm": 2.496515989303589,
      "learning_rate": 2.756478690545844e-06,
      "loss": 0.0387,
      "step": 12619
    },
    {
      "epoch": 0.6516911954557191,
      "grad_norm": 3.4385757446289062,
      "learning_rate": 2.755746474186866e-06,
      "loss": 0.0221,
      "step": 12620
    },
    {
      "epoch": 0.6517428350116189,
      "grad_norm": 2.830535888671875,
      "learning_rate": 2.755014318095243e-06,
      "loss": 0.0376,
      "step": 12621
    },
    {
      "epoch": 0.6517944745675187,
      "grad_norm": 2.537318706512451,
      "learning_rate": 2.7542822222906345e-06,
      "loss": 0.0378,
      "step": 12622
    },
    {
      "epoch": 0.6518461141234185,
      "grad_norm": 7.614969253540039,
      "learning_rate": 2.753550186792702e-06,
      "loss": 0.0558,
      "step": 12623
    },
    {
      "epoch": 0.6518977536793183,
      "grad_norm": 6.178466796875,
      "learning_rate": 2.752818211621103e-06,
      "loss": 0.0676,
      "step": 12624
    },
    {
      "epoch": 0.6519493932352182,
      "grad_norm": 5.39486026763916,
      "learning_rate": 2.7520862967954937e-06,
      "loss": 0.0541,
      "step": 12625
    },
    {
      "epoch": 0.652001032791118,
      "grad_norm": 7.185129642486572,
      "learning_rate": 2.7513544423355294e-06,
      "loss": 0.043,
      "step": 12626
    },
    {
      "epoch": 0.6520526723470178,
      "grad_norm": 2.293764114379883,
      "learning_rate": 2.7506226482608632e-06,
      "loss": 0.0437,
      "step": 12627
    },
    {
      "epoch": 0.6521043119029176,
      "grad_norm": 6.181435585021973,
      "learning_rate": 2.749890914591149e-06,
      "loss": 0.036,
      "step": 12628
    },
    {
      "epoch": 0.6521559514588174,
      "grad_norm": 3.813369035720825,
      "learning_rate": 2.749159241346031e-06,
      "loss": 0.0427,
      "step": 12629
    },
    {
      "epoch": 0.6522075910147173,
      "grad_norm": 3.026762008666992,
      "learning_rate": 2.7484276285451628e-06,
      "loss": 0.031,
      "step": 12630
    },
    {
      "epoch": 0.6522592305706171,
      "grad_norm": 12.016717910766602,
      "learning_rate": 2.7476960762081915e-06,
      "loss": 0.137,
      "step": 12631
    },
    {
      "epoch": 0.6523108701265169,
      "grad_norm": 3.4052064418792725,
      "learning_rate": 2.746964584354759e-06,
      "loss": 0.0507,
      "step": 12632
    },
    {
      "epoch": 0.6523625096824167,
      "grad_norm": 2.8557822704315186,
      "learning_rate": 2.7462331530045104e-06,
      "loss": 0.0511,
      "step": 12633
    },
    {
      "epoch": 0.6524141492383165,
      "grad_norm": 8.013261795043945,
      "learning_rate": 2.7455017821770846e-06,
      "loss": 0.102,
      "step": 12634
    },
    {
      "epoch": 0.6524657887942164,
      "grad_norm": 2.6769232749938965,
      "learning_rate": 2.74477047189213e-06,
      "loss": 0.0338,
      "step": 12635
    },
    {
      "epoch": 0.6525174283501162,
      "grad_norm": 5.660281181335449,
      "learning_rate": 2.744039222169277e-06,
      "loss": 0.0807,
      "step": 12636
    },
    {
      "epoch": 0.652569067906016,
      "grad_norm": 3.060302495956421,
      "learning_rate": 2.743308033028166e-06,
      "loss": 0.0362,
      "step": 12637
    },
    {
      "epoch": 0.6526207074619158,
      "grad_norm": 18.66330909729004,
      "learning_rate": 2.742576904488432e-06,
      "loss": 0.1209,
      "step": 12638
    },
    {
      "epoch": 0.6526723470178156,
      "grad_norm": 3.1134231090545654,
      "learning_rate": 2.7418458365697093e-06,
      "loss": 0.0543,
      "step": 12639
    },
    {
      "epoch": 0.6527239865737154,
      "grad_norm": 8.950819969177246,
      "learning_rate": 2.741114829291629e-06,
      "loss": 0.078,
      "step": 12640
    },
    {
      "epoch": 0.6527756261296153,
      "grad_norm": 8.919194221496582,
      "learning_rate": 2.7403838826738226e-06,
      "loss": 0.0902,
      "step": 12641
    },
    {
      "epoch": 0.6528272656855151,
      "grad_norm": 5.137451171875,
      "learning_rate": 2.739652996735921e-06,
      "loss": 0.0608,
      "step": 12642
    },
    {
      "epoch": 0.6528789052414149,
      "grad_norm": 3.11952805519104,
      "learning_rate": 2.7389221714975444e-06,
      "loss": 0.0653,
      "step": 12643
    },
    {
      "epoch": 0.6529305447973147,
      "grad_norm": 1.420857548713684,
      "learning_rate": 2.7381914069783254e-06,
      "loss": 0.019,
      "step": 12644
    },
    {
      "epoch": 0.6529821843532145,
      "grad_norm": 17.999996185302734,
      "learning_rate": 2.7374607031978863e-06,
      "loss": 0.1159,
      "step": 12645
    },
    {
      "epoch": 0.6530338239091144,
      "grad_norm": 1.9605858325958252,
      "learning_rate": 2.7367300601758497e-06,
      "loss": 0.0327,
      "step": 12646
    },
    {
      "epoch": 0.6530854634650142,
      "grad_norm": 2.6327600479125977,
      "learning_rate": 2.7359994779318343e-06,
      "loss": 0.0345,
      "step": 12647
    },
    {
      "epoch": 0.653137103020914,
      "grad_norm": 19.62729263305664,
      "learning_rate": 2.7352689564854582e-06,
      "loss": 0.1296,
      "step": 12648
    },
    {
      "epoch": 0.6531887425768138,
      "grad_norm": 13.445870399475098,
      "learning_rate": 2.7345384958563438e-06,
      "loss": 0.2109,
      "step": 12649
    },
    {
      "epoch": 0.6532403821327136,
      "grad_norm": 1.5898380279541016,
      "learning_rate": 2.7338080960641057e-06,
      "loss": 0.0063,
      "step": 12650
    },
    {
      "epoch": 0.6532920216886134,
      "grad_norm": 3.756241798400879,
      "learning_rate": 2.733077757128355e-06,
      "loss": 0.095,
      "step": 12651
    },
    {
      "epoch": 0.6533436612445133,
      "grad_norm": 2.724780321121216,
      "learning_rate": 2.7323474790687054e-06,
      "loss": 0.0279,
      "step": 12652
    },
    {
      "epoch": 0.6533953008004131,
      "grad_norm": 4.166650295257568,
      "learning_rate": 2.7316172619047693e-06,
      "loss": 0.0771,
      "step": 12653
    },
    {
      "epoch": 0.6534469403563129,
      "grad_norm": 6.324096202850342,
      "learning_rate": 2.7308871056561546e-06,
      "loss": 0.0469,
      "step": 12654
    },
    {
      "epoch": 0.6534985799122127,
      "grad_norm": 4.914486885070801,
      "learning_rate": 2.73015701034247e-06,
      "loss": 0.0541,
      "step": 12655
    },
    {
      "epoch": 0.6535502194681125,
      "grad_norm": 20.363149642944336,
      "learning_rate": 2.7294269759833203e-06,
      "loss": 0.0878,
      "step": 12656
    },
    {
      "epoch": 0.6536018590240124,
      "grad_norm": 5.850318908691406,
      "learning_rate": 2.728697002598311e-06,
      "loss": 0.0699,
      "step": 12657
    },
    {
      "epoch": 0.6536534985799122,
      "grad_norm": 7.116106986999512,
      "learning_rate": 2.7279670902070443e-06,
      "loss": 0.0838,
      "step": 12658
    },
    {
      "epoch": 0.653705138135812,
      "grad_norm": 10.735383033752441,
      "learning_rate": 2.7272372388291225e-06,
      "loss": 0.1244,
      "step": 12659
    },
    {
      "epoch": 0.6537567776917118,
      "grad_norm": 3.1376001834869385,
      "learning_rate": 2.7265074484841424e-06,
      "loss": 0.0575,
      "step": 12660
    },
    {
      "epoch": 0.6538084172476116,
      "grad_norm": 2.457338571548462,
      "learning_rate": 2.725777719191707e-06,
      "loss": 0.0229,
      "step": 12661
    },
    {
      "epoch": 0.6538600568035114,
      "grad_norm": 1.708425760269165,
      "learning_rate": 2.7250480509714043e-06,
      "loss": 0.0258,
      "step": 12662
    },
    {
      "epoch": 0.6539116963594113,
      "grad_norm": 6.7078962326049805,
      "learning_rate": 2.724318443842836e-06,
      "loss": 0.0743,
      "step": 12663
    },
    {
      "epoch": 0.6539633359153111,
      "grad_norm": 2.2017123699188232,
      "learning_rate": 2.723588897825594e-06,
      "loss": 0.0158,
      "step": 12664
    },
    {
      "epoch": 0.6540149754712109,
      "grad_norm": 3.2830376625061035,
      "learning_rate": 2.722859412939267e-06,
      "loss": 0.0511,
      "step": 12665
    },
    {
      "epoch": 0.6540666150271107,
      "grad_norm": 2.7904036045074463,
      "learning_rate": 2.722129989203446e-06,
      "loss": 0.0552,
      "step": 12666
    },
    {
      "epoch": 0.6541182545830105,
      "grad_norm": 3.434236764907837,
      "learning_rate": 2.721400626637716e-06,
      "loss": 0.0473,
      "step": 12667
    },
    {
      "epoch": 0.6541698941389104,
      "grad_norm": 12.416927337646484,
      "learning_rate": 2.720671325261671e-06,
      "loss": 0.1572,
      "step": 12668
    },
    {
      "epoch": 0.6542215336948102,
      "grad_norm": 10.547676086425781,
      "learning_rate": 2.719942085094889e-06,
      "loss": 0.099,
      "step": 12669
    },
    {
      "epoch": 0.65427317325071,
      "grad_norm": 2.8618600368499756,
      "learning_rate": 2.7192129061569545e-06,
      "loss": 0.0417,
      "step": 12670
    },
    {
      "epoch": 0.6543248128066098,
      "grad_norm": 2.5111212730407715,
      "learning_rate": 2.71848378846745e-06,
      "loss": 0.0492,
      "step": 12671
    },
    {
      "epoch": 0.6543764523625096,
      "grad_norm": 2.802178382873535,
      "learning_rate": 2.7177547320459542e-06,
      "loss": 0.0262,
      "step": 12672
    },
    {
      "epoch": 0.6544280919184094,
      "grad_norm": 2.517641544342041,
      "learning_rate": 2.7170257369120454e-06,
      "loss": 0.0497,
      "step": 12673
    },
    {
      "epoch": 0.6544797314743093,
      "grad_norm": 2.1791553497314453,
      "learning_rate": 2.716296803085301e-06,
      "loss": 0.0372,
      "step": 12674
    },
    {
      "epoch": 0.6545313710302091,
      "grad_norm": 3.3757200241088867,
      "learning_rate": 2.715567930585298e-06,
      "loss": 0.031,
      "step": 12675
    },
    {
      "epoch": 0.6545830105861089,
      "grad_norm": 3.1067659854888916,
      "learning_rate": 2.7148391194316014e-06,
      "loss": 0.0243,
      "step": 12676
    },
    {
      "epoch": 0.6546346501420088,
      "grad_norm": 8.475635528564453,
      "learning_rate": 2.714110369643791e-06,
      "loss": 0.0786,
      "step": 12677
    },
    {
      "epoch": 0.6546862896979087,
      "grad_norm": 2.6647560596466064,
      "learning_rate": 2.713381681241434e-06,
      "loss": 0.02,
      "step": 12678
    },
    {
      "epoch": 0.6547379292538085,
      "grad_norm": 3.5335419178009033,
      "learning_rate": 2.712653054244101e-06,
      "loss": 0.048,
      "step": 12679
    },
    {
      "epoch": 0.6547895688097083,
      "grad_norm": 9.369400024414062,
      "learning_rate": 2.711924488671354e-06,
      "loss": 0.0696,
      "step": 12680
    },
    {
      "epoch": 0.6548412083656081,
      "grad_norm": 23.525104522705078,
      "learning_rate": 2.711195984542757e-06,
      "loss": 0.0838,
      "step": 12681
    },
    {
      "epoch": 0.6548928479215079,
      "grad_norm": 2.1825196743011475,
      "learning_rate": 2.71046754187788e-06,
      "loss": 0.0477,
      "step": 12682
    },
    {
      "epoch": 0.6549444874774077,
      "grad_norm": 2.6400821208953857,
      "learning_rate": 2.709739160696282e-06,
      "loss": 0.0346,
      "step": 12683
    },
    {
      "epoch": 0.6549961270333076,
      "grad_norm": 2.398519515991211,
      "learning_rate": 2.709010841017521e-06,
      "loss": 0.0218,
      "step": 12684
    },
    {
      "epoch": 0.6550477665892074,
      "grad_norm": 5.458869457244873,
      "learning_rate": 2.708282582861156e-06,
      "loss": 0.0583,
      "step": 12685
    },
    {
      "epoch": 0.6550994061451072,
      "grad_norm": 3.494992971420288,
      "learning_rate": 2.7075543862467446e-06,
      "loss": 0.0517,
      "step": 12686
    },
    {
      "epoch": 0.655151045701007,
      "grad_norm": 7.926393032073975,
      "learning_rate": 2.7068262511938402e-06,
      "loss": 0.0673,
      "step": 12687
    },
    {
      "epoch": 0.6552026852569068,
      "grad_norm": 4.420928001403809,
      "learning_rate": 2.7060981777219985e-06,
      "loss": 0.0603,
      "step": 12688
    },
    {
      "epoch": 0.6552543248128067,
      "grad_norm": 5.1408562660217285,
      "learning_rate": 2.70537016585077e-06,
      "loss": 0.0452,
      "step": 12689
    },
    {
      "epoch": 0.6553059643687065,
      "grad_norm": 5.57029390335083,
      "learning_rate": 2.7046422155997055e-06,
      "loss": 0.0773,
      "step": 12690
    },
    {
      "epoch": 0.6553576039246063,
      "grad_norm": 27.518335342407227,
      "learning_rate": 2.7039143269883517e-06,
      "loss": 0.189,
      "step": 12691
    },
    {
      "epoch": 0.6554092434805061,
      "grad_norm": 15.66183853149414,
      "learning_rate": 2.703186500036257e-06,
      "loss": 0.0991,
      "step": 12692
    },
    {
      "epoch": 0.6554608830364059,
      "grad_norm": 2.988036870956421,
      "learning_rate": 2.7024587347629666e-06,
      "loss": 0.0186,
      "step": 12693
    },
    {
      "epoch": 0.6555125225923057,
      "grad_norm": 13.64883804321289,
      "learning_rate": 2.701731031188025e-06,
      "loss": 0.1763,
      "step": 12694
    },
    {
      "epoch": 0.6555641621482056,
      "grad_norm": 32.02070617675781,
      "learning_rate": 2.7010033893309683e-06,
      "loss": 0.1726,
      "step": 12695
    },
    {
      "epoch": 0.6556158017041054,
      "grad_norm": 2.8480021953582764,
      "learning_rate": 2.700275809211343e-06,
      "loss": 0.0285,
      "step": 12696
    },
    {
      "epoch": 0.6556674412600052,
      "grad_norm": 5.931643486022949,
      "learning_rate": 2.699548290848687e-06,
      "loss": 0.0368,
      "step": 12697
    },
    {
      "epoch": 0.655719080815905,
      "grad_norm": 4.51618766784668,
      "learning_rate": 2.6988208342625343e-06,
      "loss": 0.0794,
      "step": 12698
    },
    {
      "epoch": 0.6557707203718048,
      "grad_norm": 3.0220465660095215,
      "learning_rate": 2.698093439472421e-06,
      "loss": 0.0539,
      "step": 12699
    },
    {
      "epoch": 0.6558223599277047,
      "grad_norm": 5.857562065124512,
      "learning_rate": 2.6973661064978786e-06,
      "loss": 0.0855,
      "step": 12700
    },
    {
      "epoch": 0.6558739994836045,
      "grad_norm": 6.573150157928467,
      "learning_rate": 2.696638835358445e-06,
      "loss": 0.0622,
      "step": 12701
    },
    {
      "epoch": 0.6559256390395043,
      "grad_norm": 9.570236206054688,
      "learning_rate": 2.695911626073645e-06,
      "loss": 0.1152,
      "step": 12702
    },
    {
      "epoch": 0.6559772785954041,
      "grad_norm": 4.3379693031311035,
      "learning_rate": 2.695184478663009e-06,
      "loss": 0.051,
      "step": 12703
    },
    {
      "epoch": 0.6560289181513039,
      "grad_norm": 3.2588934898376465,
      "learning_rate": 2.6944573931460638e-06,
      "loss": 0.025,
      "step": 12704
    },
    {
      "epoch": 0.6560805577072037,
      "grad_norm": 7.624474048614502,
      "learning_rate": 2.693730369542334e-06,
      "loss": 0.0578,
      "step": 12705
    },
    {
      "epoch": 0.6561321972631036,
      "grad_norm": 3.9553935527801514,
      "learning_rate": 2.6930034078713437e-06,
      "loss": 0.0474,
      "step": 12706
    },
    {
      "epoch": 0.6561838368190034,
      "grad_norm": 3.487436294555664,
      "learning_rate": 2.692276508152615e-06,
      "loss": 0.0261,
      "step": 12707
    },
    {
      "epoch": 0.6562354763749032,
      "grad_norm": 5.1297926902771,
      "learning_rate": 2.6915496704056694e-06,
      "loss": 0.0791,
      "step": 12708
    },
    {
      "epoch": 0.656287115930803,
      "grad_norm": 14.531330108642578,
      "learning_rate": 2.69082289465002e-06,
      "loss": 0.1731,
      "step": 12709
    },
    {
      "epoch": 0.6563387554867028,
      "grad_norm": 3.07755184173584,
      "learning_rate": 2.6900961809051894e-06,
      "loss": 0.0206,
      "step": 12710
    },
    {
      "epoch": 0.6563903950426027,
      "grad_norm": 5.16549825668335,
      "learning_rate": 2.689369529190691e-06,
      "loss": 0.0788,
      "step": 12711
    },
    {
      "epoch": 0.6564420345985025,
      "grad_norm": 7.500154495239258,
      "learning_rate": 2.6886429395260405e-06,
      "loss": 0.0935,
      "step": 12712
    },
    {
      "epoch": 0.6564936741544023,
      "grad_norm": 2.202723979949951,
      "learning_rate": 2.687916411930746e-06,
      "loss": 0.0234,
      "step": 12713
    },
    {
      "epoch": 0.6565453137103021,
      "grad_norm": 4.306081295013428,
      "learning_rate": 2.687189946424317e-06,
      "loss": 0.0747,
      "step": 12714
    },
    {
      "epoch": 0.6565969532662019,
      "grad_norm": 2.8118784427642822,
      "learning_rate": 2.6864635430262665e-06,
      "loss": 0.0348,
      "step": 12715
    },
    {
      "epoch": 0.6566485928221018,
      "grad_norm": 4.745942115783691,
      "learning_rate": 2.685737201756101e-06,
      "loss": 0.0586,
      "step": 12716
    },
    {
      "epoch": 0.6567002323780016,
      "grad_norm": 2.8297736644744873,
      "learning_rate": 2.6850109226333224e-06,
      "loss": 0.0524,
      "step": 12717
    },
    {
      "epoch": 0.6567518719339014,
      "grad_norm": 7.625026702880859,
      "learning_rate": 2.6842847056774355e-06,
      "loss": 0.0712,
      "step": 12718
    },
    {
      "epoch": 0.6568035114898012,
      "grad_norm": 1.675257921218872,
      "learning_rate": 2.6835585509079433e-06,
      "loss": 0.0194,
      "step": 12719
    },
    {
      "epoch": 0.656855151045701,
      "grad_norm": 2.404731273651123,
      "learning_rate": 2.6828324583443453e-06,
      "loss": 0.0536,
      "step": 12720
    },
    {
      "epoch": 0.6569067906016008,
      "grad_norm": 5.236094951629639,
      "learning_rate": 2.6821064280061394e-06,
      "loss": 0.1156,
      "step": 12721
    },
    {
      "epoch": 0.6569584301575007,
      "grad_norm": 2.0631344318389893,
      "learning_rate": 2.6813804599128235e-06,
      "loss": 0.0294,
      "step": 12722
    },
    {
      "epoch": 0.6570100697134005,
      "grad_norm": 3.701624631881714,
      "learning_rate": 2.6806545540838924e-06,
      "loss": 0.0376,
      "step": 12723
    },
    {
      "epoch": 0.6570617092693003,
      "grad_norm": 14.26345157623291,
      "learning_rate": 2.6799287105388393e-06,
      "loss": 0.0729,
      "step": 12724
    },
    {
      "epoch": 0.6571133488252001,
      "grad_norm": 4.7186598777771,
      "learning_rate": 2.6792029292971566e-06,
      "loss": 0.0811,
      "step": 12725
    },
    {
      "epoch": 0.6571649883810999,
      "grad_norm": 2.299649953842163,
      "learning_rate": 2.6784772103783344e-06,
      "loss": 0.0234,
      "step": 12726
    },
    {
      "epoch": 0.6572166279369998,
      "grad_norm": 1.8695613145828247,
      "learning_rate": 2.6777515538018636e-06,
      "loss": 0.024,
      "step": 12727
    },
    {
      "epoch": 0.6572682674928996,
      "grad_norm": 2.88781476020813,
      "learning_rate": 2.677025959587224e-06,
      "loss": 0.0653,
      "step": 12728
    },
    {
      "epoch": 0.6573199070487994,
      "grad_norm": 7.121158599853516,
      "learning_rate": 2.6763004277539073e-06,
      "loss": 0.0533,
      "step": 12729
    },
    {
      "epoch": 0.6573715466046992,
      "grad_norm": 2.9074442386627197,
      "learning_rate": 2.6755749583213968e-06,
      "loss": 0.0538,
      "step": 12730
    },
    {
      "epoch": 0.657423186160599,
      "grad_norm": 5.322697639465332,
      "learning_rate": 2.674849551309171e-06,
      "loss": 0.0862,
      "step": 12731
    },
    {
      "epoch": 0.6574748257164988,
      "grad_norm": 1.597469449043274,
      "learning_rate": 2.6741242067367114e-06,
      "loss": 0.0325,
      "step": 12732
    },
    {
      "epoch": 0.6575264652723987,
      "grad_norm": 9.56027889251709,
      "learning_rate": 2.6733989246234938e-06,
      "loss": 0.048,
      "step": 12733
    },
    {
      "epoch": 0.6575781048282985,
      "grad_norm": 2.9408016204833984,
      "learning_rate": 2.6726737049890026e-06,
      "loss": 0.0305,
      "step": 12734
    },
    {
      "epoch": 0.6576297443841983,
      "grad_norm": 1.9155508279800415,
      "learning_rate": 2.671948547852706e-06,
      "loss": 0.0241,
      "step": 12735
    },
    {
      "epoch": 0.6576813839400981,
      "grad_norm": 4.302579402923584,
      "learning_rate": 2.67122345323408e-06,
      "loss": 0.04,
      "step": 12736
    },
    {
      "epoch": 0.6577330234959979,
      "grad_norm": 9.079405784606934,
      "learning_rate": 2.670498421152595e-06,
      "loss": 0.0962,
      "step": 12737
    },
    {
      "epoch": 0.6577846630518978,
      "grad_norm": 15.623397827148438,
      "learning_rate": 2.669773451627722e-06,
      "loss": 0.1624,
      "step": 12738
    },
    {
      "epoch": 0.6578363026077976,
      "grad_norm": 3.72482967376709,
      "learning_rate": 2.669048544678929e-06,
      "loss": 0.0794,
      "step": 12739
    },
    {
      "epoch": 0.6578879421636974,
      "grad_norm": 4.518167018890381,
      "learning_rate": 2.6683237003256833e-06,
      "loss": 0.0662,
      "step": 12740
    },
    {
      "epoch": 0.6579395817195972,
      "grad_norm": 1.9863823652267456,
      "learning_rate": 2.6675989185874523e-06,
      "loss": 0.0298,
      "step": 12741
    },
    {
      "epoch": 0.657991221275497,
      "grad_norm": 13.754284858703613,
      "learning_rate": 2.666874199483692e-06,
      "loss": 0.1283,
      "step": 12742
    },
    {
      "epoch": 0.6580428608313968,
      "grad_norm": 4.0980072021484375,
      "learning_rate": 2.666149543033871e-06,
      "loss": 0.0446,
      "step": 12743
    },
    {
      "epoch": 0.6580945003872967,
      "grad_norm": 8.051972389221191,
      "learning_rate": 2.665424949257447e-06,
      "loss": 0.089,
      "step": 12744
    },
    {
      "epoch": 0.6581461399431965,
      "grad_norm": 3.3004238605499268,
      "learning_rate": 2.66470041817388e-06,
      "loss": 0.0384,
      "step": 12745
    },
    {
      "epoch": 0.6581977794990963,
      "grad_norm": 6.596912384033203,
      "learning_rate": 2.6639759498026236e-06,
      "loss": 0.0948,
      "step": 12746
    },
    {
      "epoch": 0.6582494190549961,
      "grad_norm": 7.524192810058594,
      "learning_rate": 2.6632515441631313e-06,
      "loss": 0.0701,
      "step": 12747
    },
    {
      "epoch": 0.6583010586108959,
      "grad_norm": 1.8983440399169922,
      "learning_rate": 2.6625272012748616e-06,
      "loss": 0.0323,
      "step": 12748
    },
    {
      "epoch": 0.6583526981667958,
      "grad_norm": 7.643192768096924,
      "learning_rate": 2.6618029211572656e-06,
      "loss": 0.0702,
      "step": 12749
    },
    {
      "epoch": 0.6584043377226956,
      "grad_norm": 1.6766738891601562,
      "learning_rate": 2.6610787038297892e-06,
      "loss": 0.0172,
      "step": 12750
    },
    {
      "epoch": 0.6584559772785954,
      "grad_norm": 1.2909780740737915,
      "learning_rate": 2.660354549311882e-06,
      "loss": 0.0194,
      "step": 12751
    },
    {
      "epoch": 0.6585076168344952,
      "grad_norm": 5.40613317489624,
      "learning_rate": 2.6596304576229916e-06,
      "loss": 0.0544,
      "step": 12752
    },
    {
      "epoch": 0.658559256390395,
      "grad_norm": 18.40021324157715,
      "learning_rate": 2.6589064287825616e-06,
      "loss": 0.1218,
      "step": 12753
    },
    {
      "epoch": 0.6586108959462948,
      "grad_norm": 9.051777839660645,
      "learning_rate": 2.658182462810036e-06,
      "loss": 0.0482,
      "step": 12754
    },
    {
      "epoch": 0.6586625355021947,
      "grad_norm": 4.845804691314697,
      "learning_rate": 2.6574585597248557e-06,
      "loss": 0.0615,
      "step": 12755
    },
    {
      "epoch": 0.6587141750580945,
      "grad_norm": 5.579860687255859,
      "learning_rate": 2.656734719546461e-06,
      "loss": 0.0813,
      "step": 12756
    },
    {
      "epoch": 0.6587658146139943,
      "grad_norm": 2.5900793075561523,
      "learning_rate": 2.6560109422942903e-06,
      "loss": 0.0534,
      "step": 12757
    },
    {
      "epoch": 0.6588174541698941,
      "grad_norm": 2.518270969390869,
      "learning_rate": 2.6552872279877782e-06,
      "loss": 0.0664,
      "step": 12758
    },
    {
      "epoch": 0.6588690937257939,
      "grad_norm": 4.141894340515137,
      "learning_rate": 2.654563576646362e-06,
      "loss": 0.038,
      "step": 12759
    },
    {
      "epoch": 0.6589207332816938,
      "grad_norm": 7.294877529144287,
      "learning_rate": 2.653839988289474e-06,
      "loss": 0.0842,
      "step": 12760
    },
    {
      "epoch": 0.6589723728375936,
      "grad_norm": 4.979546070098877,
      "learning_rate": 2.6531164629365413e-06,
      "loss": 0.0557,
      "step": 12761
    },
    {
      "epoch": 0.6590240123934934,
      "grad_norm": 5.083327293395996,
      "learning_rate": 2.652393000606999e-06,
      "loss": 0.0316,
      "step": 12762
    },
    {
      "epoch": 0.6590756519493932,
      "grad_norm": 2.6932480335235596,
      "learning_rate": 2.6516696013202747e-06,
      "loss": 0.0447,
      "step": 12763
    },
    {
      "epoch": 0.659127291505293,
      "grad_norm": 3.237840175628662,
      "learning_rate": 2.650946265095791e-06,
      "loss": 0.0402,
      "step": 12764
    },
    {
      "epoch": 0.6591789310611929,
      "grad_norm": 2.6783511638641357,
      "learning_rate": 2.6502229919529744e-06,
      "loss": 0.0555,
      "step": 12765
    },
    {
      "epoch": 0.6592305706170927,
      "grad_norm": 10.40979290008545,
      "learning_rate": 2.6494997819112454e-06,
      "loss": 0.0982,
      "step": 12766
    },
    {
      "epoch": 0.6592822101729925,
      "grad_norm": 5.433568954467773,
      "learning_rate": 2.6487766349900313e-06,
      "loss": 0.0714,
      "step": 12767
    },
    {
      "epoch": 0.6593338497288923,
      "grad_norm": 2.366837501525879,
      "learning_rate": 2.6480535512087456e-06,
      "loss": 0.0351,
      "step": 12768
    },
    {
      "epoch": 0.6593854892847921,
      "grad_norm": 3.4238827228546143,
      "learning_rate": 2.647330530586808e-06,
      "loss": 0.056,
      "step": 12769
    },
    {
      "epoch": 0.6594371288406919,
      "grad_norm": 2.3108301162719727,
      "learning_rate": 2.646607573143635e-06,
      "loss": 0.0246,
      "step": 12770
    },
    {
      "epoch": 0.6594887683965918,
      "grad_norm": 3.334160089492798,
      "learning_rate": 2.64588467889864e-06,
      "loss": 0.0321,
      "step": 12771
    },
    {
      "epoch": 0.6595404079524916,
      "grad_norm": 2.482010841369629,
      "learning_rate": 2.645161847871237e-06,
      "loss": 0.0353,
      "step": 12772
    },
    {
      "epoch": 0.6595920475083914,
      "grad_norm": 11.726509094238281,
      "learning_rate": 2.6444390800808357e-06,
      "loss": 0.0935,
      "step": 12773
    },
    {
      "epoch": 0.6596436870642912,
      "grad_norm": 5.021522045135498,
      "learning_rate": 2.6437163755468474e-06,
      "loss": 0.0311,
      "step": 12774
    },
    {
      "epoch": 0.659695326620191,
      "grad_norm": 7.890905857086182,
      "learning_rate": 2.6429937342886746e-06,
      "loss": 0.0452,
      "step": 12775
    },
    {
      "epoch": 0.6597469661760909,
      "grad_norm": 3.672342300415039,
      "learning_rate": 2.642271156325729e-06,
      "loss": 0.0579,
      "step": 12776
    },
    {
      "epoch": 0.6597986057319907,
      "grad_norm": 3.5448389053344727,
      "learning_rate": 2.6415486416774123e-06,
      "loss": 0.057,
      "step": 12777
    },
    {
      "epoch": 0.6598502452878905,
      "grad_norm": 2.528825044631958,
      "learning_rate": 2.6408261903631282e-06,
      "loss": 0.0302,
      "step": 12778
    },
    {
      "epoch": 0.6599018848437903,
      "grad_norm": 3.2803702354431152,
      "learning_rate": 2.640103802402275e-06,
      "loss": 0.0414,
      "step": 12779
    },
    {
      "epoch": 0.6599535243996901,
      "grad_norm": 2.0684256553649902,
      "learning_rate": 2.639381477814251e-06,
      "loss": 0.0312,
      "step": 12780
    },
    {
      "epoch": 0.66000516395559,
      "grad_norm": 2.6800873279571533,
      "learning_rate": 2.6386592166184583e-06,
      "loss": 0.0252,
      "step": 12781
    },
    {
      "epoch": 0.6600568035114898,
      "grad_norm": 3.161264419555664,
      "learning_rate": 2.637937018834291e-06,
      "loss": 0.0619,
      "step": 12782
    },
    {
      "epoch": 0.6601084430673896,
      "grad_norm": 7.828662395477295,
      "learning_rate": 2.6372148844811407e-06,
      "loss": 0.0736,
      "step": 12783
    },
    {
      "epoch": 0.6601600826232894,
      "grad_norm": 3.7543468475341797,
      "learning_rate": 2.636492813578401e-06,
      "loss": 0.0532,
      "step": 12784
    },
    {
      "epoch": 0.6602117221791892,
      "grad_norm": 14.988953590393066,
      "learning_rate": 2.635770806145462e-06,
      "loss": 0.069,
      "step": 12785
    },
    {
      "epoch": 0.660263361735089,
      "grad_norm": 5.184351921081543,
      "learning_rate": 2.635048862201712e-06,
      "loss": 0.0614,
      "step": 12786
    },
    {
      "epoch": 0.6603150012909889,
      "grad_norm": 7.936893939971924,
      "learning_rate": 2.6343269817665405e-06,
      "loss": 0.1099,
      "step": 12787
    },
    {
      "epoch": 0.6603666408468887,
      "grad_norm": 6.926980972290039,
      "learning_rate": 2.6336051648593308e-06,
      "loss": 0.0747,
      "step": 12788
    },
    {
      "epoch": 0.6604182804027885,
      "grad_norm": 2.2415828704833984,
      "learning_rate": 2.6328834114994677e-06,
      "loss": 0.0296,
      "step": 12789
    },
    {
      "epoch": 0.6604699199586883,
      "grad_norm": 7.428380966186523,
      "learning_rate": 2.6321617217063322e-06,
      "loss": 0.0913,
      "step": 12790
    },
    {
      "epoch": 0.6605215595145881,
      "grad_norm": 13.345431327819824,
      "learning_rate": 2.631440095499306e-06,
      "loss": 0.0759,
      "step": 12791
    },
    {
      "epoch": 0.660573199070488,
      "grad_norm": 4.181146144866943,
      "learning_rate": 2.630718532897767e-06,
      "loss": 0.0258,
      "step": 12792
    },
    {
      "epoch": 0.6606248386263878,
      "grad_norm": 10.698216438293457,
      "learning_rate": 2.6299970339210932e-06,
      "loss": 0.0798,
      "step": 12793
    },
    {
      "epoch": 0.6606764781822876,
      "grad_norm": 6.451892375946045,
      "learning_rate": 2.6292755985886555e-06,
      "loss": 0.079,
      "step": 12794
    },
    {
      "epoch": 0.6607281177381874,
      "grad_norm": 1.3233217000961304,
      "learning_rate": 2.6285542269198315e-06,
      "loss": 0.0198,
      "step": 12795
    },
    {
      "epoch": 0.6607797572940872,
      "grad_norm": 2.21951961517334,
      "learning_rate": 2.6278329189339947e-06,
      "loss": 0.0224,
      "step": 12796
    },
    {
      "epoch": 0.660831396849987,
      "grad_norm": 5.21697473526001,
      "learning_rate": 2.6271116746505103e-06,
      "loss": 0.0492,
      "step": 12797
    },
    {
      "epoch": 0.6608830364058869,
      "grad_norm": 4.806826114654541,
      "learning_rate": 2.6263904940887487e-06,
      "loss": 0.0607,
      "step": 12798
    },
    {
      "epoch": 0.6609346759617867,
      "grad_norm": 3.043674945831299,
      "learning_rate": 2.6256693772680752e-06,
      "loss": 0.0357,
      "step": 12799
    },
    {
      "epoch": 0.6609863155176865,
      "grad_norm": 11.235980033874512,
      "learning_rate": 2.6249483242078598e-06,
      "loss": 0.1815,
      "step": 12800
    },
    {
      "epoch": 0.6610379550735863,
      "grad_norm": 3.430821418762207,
      "learning_rate": 2.624227334927459e-06,
      "loss": 0.0339,
      "step": 12801
    },
    {
      "epoch": 0.6610895946294861,
      "grad_norm": 2.2815358638763428,
      "learning_rate": 2.623506409446239e-06,
      "loss": 0.0296,
      "step": 12802
    },
    {
      "epoch": 0.661141234185386,
      "grad_norm": 1.2867308855056763,
      "learning_rate": 2.6227855477835575e-06,
      "loss": 0.0183,
      "step": 12803
    },
    {
      "epoch": 0.6611928737412859,
      "grad_norm": 22.137880325317383,
      "learning_rate": 2.6220647499587725e-06,
      "loss": 0.1907,
      "step": 12804
    },
    {
      "epoch": 0.6612445132971857,
      "grad_norm": 2.521660327911377,
      "learning_rate": 2.621344015991242e-06,
      "loss": 0.0271,
      "step": 12805
    },
    {
      "epoch": 0.6612961528530855,
      "grad_norm": 6.34006929397583,
      "learning_rate": 2.6206233459003183e-06,
      "loss": 0.0493,
      "step": 12806
    },
    {
      "epoch": 0.6613477924089853,
      "grad_norm": 3.523862361907959,
      "learning_rate": 2.6199027397053578e-06,
      "loss": 0.0386,
      "step": 12807
    },
    {
      "epoch": 0.6613994319648852,
      "grad_norm": 4.123152732849121,
      "learning_rate": 2.619182197425706e-06,
      "loss": 0.0526,
      "step": 12808
    },
    {
      "epoch": 0.661451071520785,
      "grad_norm": 4.128074645996094,
      "learning_rate": 2.6184617190807177e-06,
      "loss": 0.0615,
      "step": 12809
    },
    {
      "epoch": 0.6615027110766848,
      "grad_norm": 4.984338283538818,
      "learning_rate": 2.6177413046897384e-06,
      "loss": 0.0861,
      "step": 12810
    },
    {
      "epoch": 0.6615543506325846,
      "grad_norm": 4.322892189025879,
      "learning_rate": 2.617020954272117e-06,
      "loss": 0.0196,
      "step": 12811
    },
    {
      "epoch": 0.6616059901884844,
      "grad_norm": 2.475424289703369,
      "learning_rate": 2.6163006678471937e-06,
      "loss": 0.0358,
      "step": 12812
    },
    {
      "epoch": 0.6616576297443842,
      "grad_norm": 3.667241096496582,
      "learning_rate": 2.6155804454343103e-06,
      "loss": 0.0639,
      "step": 12813
    },
    {
      "epoch": 0.6617092693002841,
      "grad_norm": 4.145332336425781,
      "learning_rate": 2.6148602870528127e-06,
      "loss": 0.0309,
      "step": 12814
    },
    {
      "epoch": 0.6617609088561839,
      "grad_norm": 13.532744407653809,
      "learning_rate": 2.61414019272204e-06,
      "loss": 0.0568,
      "step": 12815
    },
    {
      "epoch": 0.6618125484120837,
      "grad_norm": 4.738780498504639,
      "learning_rate": 2.613420162461324e-06,
      "loss": 0.0528,
      "step": 12816
    },
    {
      "epoch": 0.6618641879679835,
      "grad_norm": 16.624723434448242,
      "learning_rate": 2.6127001962900057e-06,
      "loss": 0.0619,
      "step": 12817
    },
    {
      "epoch": 0.6619158275238833,
      "grad_norm": 3.1211109161376953,
      "learning_rate": 2.6119802942274157e-06,
      "loss": 0.0347,
      "step": 12818
    },
    {
      "epoch": 0.6619674670797832,
      "grad_norm": 5.315061092376709,
      "learning_rate": 2.6112604562928888e-06,
      "loss": 0.0482,
      "step": 12819
    },
    {
      "epoch": 0.662019106635683,
      "grad_norm": 2.47029972076416,
      "learning_rate": 2.6105406825057543e-06,
      "loss": 0.0473,
      "step": 12820
    },
    {
      "epoch": 0.6620707461915828,
      "grad_norm": 19.596670150756836,
      "learning_rate": 2.6098209728853407e-06,
      "loss": 0.0894,
      "step": 12821
    },
    {
      "epoch": 0.6621223857474826,
      "grad_norm": 6.445183753967285,
      "learning_rate": 2.609101327450977e-06,
      "loss": 0.0533,
      "step": 12822
    },
    {
      "epoch": 0.6621740253033824,
      "grad_norm": 15.548309326171875,
      "learning_rate": 2.608381746221986e-06,
      "loss": 0.0557,
      "step": 12823
    },
    {
      "epoch": 0.6622256648592822,
      "grad_norm": 3.724905490875244,
      "learning_rate": 2.6076622292176944e-06,
      "loss": 0.0476,
      "step": 12824
    },
    {
      "epoch": 0.6622773044151821,
      "grad_norm": 6.67092227935791,
      "learning_rate": 2.6069427764574216e-06,
      "loss": 0.063,
      "step": 12825
    },
    {
      "epoch": 0.6623289439710819,
      "grad_norm": 3.676300048828125,
      "learning_rate": 2.606223387960491e-06,
      "loss": 0.0598,
      "step": 12826
    },
    {
      "epoch": 0.6623805835269817,
      "grad_norm": 3.2378408908843994,
      "learning_rate": 2.6055040637462147e-06,
      "loss": 0.0537,
      "step": 12827
    },
    {
      "epoch": 0.6624322230828815,
      "grad_norm": 3.0976786613464355,
      "learning_rate": 2.6047848038339163e-06,
      "loss": 0.0426,
      "step": 12828
    },
    {
      "epoch": 0.6624838626387813,
      "grad_norm": 4.579397678375244,
      "learning_rate": 2.6040656082429095e-06,
      "loss": 0.0625,
      "step": 12829
    },
    {
      "epoch": 0.6625355021946812,
      "grad_norm": 3.2573482990264893,
      "learning_rate": 2.6033464769925055e-06,
      "loss": 0.067,
      "step": 12830
    },
    {
      "epoch": 0.662587141750581,
      "grad_norm": 4.014169692993164,
      "learning_rate": 2.6026274101020154e-06,
      "loss": 0.0322,
      "step": 12831
    },
    {
      "epoch": 0.6626387813064808,
      "grad_norm": 11.934799194335938,
      "learning_rate": 2.6019084075907497e-06,
      "loss": 0.2258,
      "step": 12832
    },
    {
      "epoch": 0.6626904208623806,
      "grad_norm": 5.535560607910156,
      "learning_rate": 2.601189469478021e-06,
      "loss": 0.0265,
      "step": 12833
    },
    {
      "epoch": 0.6627420604182804,
      "grad_norm": 5.583761692047119,
      "learning_rate": 2.60047059578313e-06,
      "loss": 0.0321,
      "step": 12834
    },
    {
      "epoch": 0.6627936999741803,
      "grad_norm": 2.674820899963379,
      "learning_rate": 2.599751786525383e-06,
      "loss": 0.0415,
      "step": 12835
    },
    {
      "epoch": 0.6628453395300801,
      "grad_norm": 6.671516418457031,
      "learning_rate": 2.599033041724084e-06,
      "loss": 0.0885,
      "step": 12836
    },
    {
      "epoch": 0.6628969790859799,
      "grad_norm": 3.2565784454345703,
      "learning_rate": 2.5983143613985333e-06,
      "loss": 0.051,
      "step": 12837
    },
    {
      "epoch": 0.6629486186418797,
      "grad_norm": 2.2025396823883057,
      "learning_rate": 2.5975957455680313e-06,
      "loss": 0.0225,
      "step": 12838
    },
    {
      "epoch": 0.6630002581977795,
      "grad_norm": 3.0805001258850098,
      "learning_rate": 2.5968771942518743e-06,
      "loss": 0.0345,
      "step": 12839
    },
    {
      "epoch": 0.6630518977536793,
      "grad_norm": 4.845678329467773,
      "learning_rate": 2.5961587074693618e-06,
      "loss": 0.0596,
      "step": 12840
    },
    {
      "epoch": 0.6631035373095792,
      "grad_norm": 5.387184143066406,
      "learning_rate": 2.5954402852397807e-06,
      "loss": 0.0562,
      "step": 12841
    },
    {
      "epoch": 0.663155176865479,
      "grad_norm": 2.7122690677642822,
      "learning_rate": 2.5947219275824316e-06,
      "loss": 0.0344,
      "step": 12842
    },
    {
      "epoch": 0.6632068164213788,
      "grad_norm": 1.2205414772033691,
      "learning_rate": 2.594003634516602e-06,
      "loss": 0.0251,
      "step": 12843
    },
    {
      "epoch": 0.6632584559772786,
      "grad_norm": 3.4250426292419434,
      "learning_rate": 2.593285406061582e-06,
      "loss": 0.0612,
      "step": 12844
    },
    {
      "epoch": 0.6633100955331784,
      "grad_norm": 5.934640407562256,
      "learning_rate": 2.5925672422366567e-06,
      "loss": 0.0337,
      "step": 12845
    },
    {
      "epoch": 0.6633617350890783,
      "grad_norm": 4.767578601837158,
      "learning_rate": 2.5918491430611113e-06,
      "loss": 0.0571,
      "step": 12846
    },
    {
      "epoch": 0.6634133746449781,
      "grad_norm": 5.126905918121338,
      "learning_rate": 2.5911311085542334e-06,
      "loss": 0.0716,
      "step": 12847
    },
    {
      "epoch": 0.6634650142008779,
      "grad_norm": 26.275466918945312,
      "learning_rate": 2.590413138735305e-06,
      "loss": 0.26,
      "step": 12848
    },
    {
      "epoch": 0.6635166537567777,
      "grad_norm": 4.369559288024902,
      "learning_rate": 2.589695233623603e-06,
      "loss": 0.0485,
      "step": 12849
    },
    {
      "epoch": 0.6635682933126775,
      "grad_norm": 9.853241920471191,
      "learning_rate": 2.5889773932384078e-06,
      "loss": 0.1167,
      "step": 12850
    },
    {
      "epoch": 0.6636199328685773,
      "grad_norm": 4.088782787322998,
      "learning_rate": 2.5882596175989965e-06,
      "loss": 0.0636,
      "step": 12851
    },
    {
      "epoch": 0.6636715724244772,
      "grad_norm": 11.629018783569336,
      "learning_rate": 2.5875419067246443e-06,
      "loss": 0.0718,
      "step": 12852
    },
    {
      "epoch": 0.663723211980377,
      "grad_norm": 4.379209995269775,
      "learning_rate": 2.586824260634624e-06,
      "loss": 0.0566,
      "step": 12853
    },
    {
      "epoch": 0.6637748515362768,
      "grad_norm": 3.9726784229278564,
      "learning_rate": 2.5861066793482083e-06,
      "loss": 0.0569,
      "step": 12854
    },
    {
      "epoch": 0.6638264910921766,
      "grad_norm": 22.144746780395508,
      "learning_rate": 2.5853891628846663e-06,
      "loss": 0.2167,
      "step": 12855
    },
    {
      "epoch": 0.6638781306480764,
      "grad_norm": 7.607303619384766,
      "learning_rate": 2.5846717112632667e-06,
      "loss": 0.1528,
      "step": 12856
    },
    {
      "epoch": 0.6639297702039763,
      "grad_norm": 4.702268600463867,
      "learning_rate": 2.5839543245032766e-06,
      "loss": 0.0732,
      "step": 12857
    },
    {
      "epoch": 0.6639814097598761,
      "grad_norm": 2.4878077507019043,
      "learning_rate": 2.5832370026239595e-06,
      "loss": 0.0213,
      "step": 12858
    },
    {
      "epoch": 0.6640330493157759,
      "grad_norm": 2.9783589839935303,
      "learning_rate": 2.5825197456445817e-06,
      "loss": 0.0224,
      "step": 12859
    },
    {
      "epoch": 0.6640846888716757,
      "grad_norm": 1.116297960281372,
      "learning_rate": 2.581802553584397e-06,
      "loss": 0.01,
      "step": 12860
    },
    {
      "epoch": 0.6641363284275755,
      "grad_norm": 3.0015645027160645,
      "learning_rate": 2.5810854264626727e-06,
      "loss": 0.0537,
      "step": 12861
    },
    {
      "epoch": 0.6641879679834753,
      "grad_norm": 10.078815460205078,
      "learning_rate": 2.5803683642986654e-06,
      "loss": 0.0403,
      "step": 12862
    },
    {
      "epoch": 0.6642396075393752,
      "grad_norm": 2.857759952545166,
      "learning_rate": 2.579651367111627e-06,
      "loss": 0.028,
      "step": 12863
    },
    {
      "epoch": 0.664291247095275,
      "grad_norm": 3.5821456909179688,
      "learning_rate": 2.5789344349208144e-06,
      "loss": 0.0261,
      "step": 12864
    },
    {
      "epoch": 0.6643428866511748,
      "grad_norm": 3.2717185020446777,
      "learning_rate": 2.578217567745478e-06,
      "loss": 0.0488,
      "step": 12865
    },
    {
      "epoch": 0.6643945262070746,
      "grad_norm": 14.092599868774414,
      "learning_rate": 2.5775007656048743e-06,
      "loss": 0.1937,
      "step": 12866
    },
    {
      "epoch": 0.6644461657629744,
      "grad_norm": 2.7608542442321777,
      "learning_rate": 2.5767840285182466e-06,
      "loss": 0.0279,
      "step": 12867
    },
    {
      "epoch": 0.6644978053188743,
      "grad_norm": 4.870190143585205,
      "learning_rate": 2.5760673565048443e-06,
      "loss": 0.0525,
      "step": 12868
    },
    {
      "epoch": 0.6645494448747741,
      "grad_norm": 17.22003173828125,
      "learning_rate": 2.575350749583913e-06,
      "loss": 0.132,
      "step": 12869
    },
    {
      "epoch": 0.6646010844306739,
      "grad_norm": 5.057433605194092,
      "learning_rate": 2.5746342077746967e-06,
      "loss": 0.0619,
      "step": 12870
    },
    {
      "epoch": 0.6646527239865737,
      "grad_norm": 2.111298084259033,
      "learning_rate": 2.573917731096437e-06,
      "loss": 0.0214,
      "step": 12871
    },
    {
      "epoch": 0.6647043635424735,
      "grad_norm": 4.909757614135742,
      "learning_rate": 2.5732013195683735e-06,
      "loss": 0.0643,
      "step": 12872
    },
    {
      "epoch": 0.6647560030983733,
      "grad_norm": 3.6331753730773926,
      "learning_rate": 2.572484973209749e-06,
      "loss": 0.039,
      "step": 12873
    },
    {
      "epoch": 0.6648076426542732,
      "grad_norm": 4.855466365814209,
      "learning_rate": 2.5717686920397924e-06,
      "loss": 0.0472,
      "step": 12874
    },
    {
      "epoch": 0.664859282210173,
      "grad_norm": 3.8031303882598877,
      "learning_rate": 2.571052476077745e-06,
      "loss": 0.068,
      "step": 12875
    },
    {
      "epoch": 0.6649109217660728,
      "grad_norm": 4.1892900466918945,
      "learning_rate": 2.570336325342839e-06,
      "loss": 0.0504,
      "step": 12876
    },
    {
      "epoch": 0.6649625613219726,
      "grad_norm": 4.698145866394043,
      "learning_rate": 2.569620239854307e-06,
      "loss": 0.0236,
      "step": 12877
    },
    {
      "epoch": 0.6650142008778724,
      "grad_norm": 2.736868381500244,
      "learning_rate": 2.5689042196313752e-06,
      "loss": 0.0299,
      "step": 12878
    },
    {
      "epoch": 0.6650658404337723,
      "grad_norm": 3.372561454772949,
      "learning_rate": 2.568188264693272e-06,
      "loss": 0.0662,
      "step": 12879
    },
    {
      "epoch": 0.6651174799896721,
      "grad_norm": 3.4623122215270996,
      "learning_rate": 2.5674723750592273e-06,
      "loss": 0.0481,
      "step": 12880
    },
    {
      "epoch": 0.6651691195455719,
      "grad_norm": 4.704540729522705,
      "learning_rate": 2.566756550748466e-06,
      "loss": 0.0471,
      "step": 12881
    },
    {
      "epoch": 0.6652207591014717,
      "grad_norm": 2.8839824199676514,
      "learning_rate": 2.566040791780206e-06,
      "loss": 0.0295,
      "step": 12882
    },
    {
      "epoch": 0.6652723986573715,
      "grad_norm": 3.3537209033966064,
      "learning_rate": 2.5653250981736714e-06,
      "loss": 0.0609,
      "step": 12883
    },
    {
      "epoch": 0.6653240382132714,
      "grad_norm": 2.948195457458496,
      "learning_rate": 2.5646094699480818e-06,
      "loss": 0.0465,
      "step": 12884
    },
    {
      "epoch": 0.6653756777691712,
      "grad_norm": 7.842380046844482,
      "learning_rate": 2.563893907122653e-06,
      "loss": 0.0386,
      "step": 12885
    },
    {
      "epoch": 0.665427317325071,
      "grad_norm": 5.988517761230469,
      "learning_rate": 2.5631784097166024e-06,
      "loss": 0.0929,
      "step": 12886
    },
    {
      "epoch": 0.6654789568809708,
      "grad_norm": 6.776972770690918,
      "learning_rate": 2.562462977749143e-06,
      "loss": 0.0611,
      "step": 12887
    },
    {
      "epoch": 0.6655305964368706,
      "grad_norm": 4.7020344734191895,
      "learning_rate": 2.561747611239488e-06,
      "loss": 0.0851,
      "step": 12888
    },
    {
      "epoch": 0.6655822359927704,
      "grad_norm": 3.1774511337280273,
      "learning_rate": 2.561032310206847e-06,
      "loss": 0.0528,
      "step": 12889
    },
    {
      "epoch": 0.6656338755486703,
      "grad_norm": 2.331756830215454,
      "learning_rate": 2.5603170746704293e-06,
      "loss": 0.0349,
      "step": 12890
    },
    {
      "epoch": 0.6656855151045701,
      "grad_norm": 1.623792290687561,
      "learning_rate": 2.5596019046494415e-06,
      "loss": 0.0263,
      "step": 12891
    },
    {
      "epoch": 0.6657371546604699,
      "grad_norm": 7.774179935455322,
      "learning_rate": 2.5588868001630913e-06,
      "loss": 0.0781,
      "step": 12892
    },
    {
      "epoch": 0.6657887942163697,
      "grad_norm": 3.120657444000244,
      "learning_rate": 2.5581717612305758e-06,
      "loss": 0.0486,
      "step": 12893
    },
    {
      "epoch": 0.6658404337722695,
      "grad_norm": 8.111640930175781,
      "learning_rate": 2.5574567878711032e-06,
      "loss": 0.0562,
      "step": 12894
    },
    {
      "epoch": 0.6658920733281694,
      "grad_norm": 11.324129104614258,
      "learning_rate": 2.556741880103873e-06,
      "loss": 0.1296,
      "step": 12895
    },
    {
      "epoch": 0.6659437128840692,
      "grad_norm": 2.45238995552063,
      "learning_rate": 2.5560270379480788e-06,
      "loss": 0.0402,
      "step": 12896
    },
    {
      "epoch": 0.665995352439969,
      "grad_norm": 2.83681321144104,
      "learning_rate": 2.555312261422921e-06,
      "loss": 0.0168,
      "step": 12897
    },
    {
      "epoch": 0.6660469919958688,
      "grad_norm": 3.5093939304351807,
      "learning_rate": 2.55459755054759e-06,
      "loss": 0.0431,
      "step": 12898
    },
    {
      "epoch": 0.6660986315517686,
      "grad_norm": 2.3094425201416016,
      "learning_rate": 2.553882905341286e-06,
      "loss": 0.0364,
      "step": 12899
    },
    {
      "epoch": 0.6661502711076684,
      "grad_norm": 2.2365734577178955,
      "learning_rate": 2.5531683258231938e-06,
      "loss": 0.0251,
      "step": 12900
    },
    {
      "epoch": 0.6662019106635683,
      "grad_norm": 11.728551864624023,
      "learning_rate": 2.5524538120125046e-06,
      "loss": 0.1258,
      "step": 12901
    },
    {
      "epoch": 0.6662535502194681,
      "grad_norm": 15.274405479431152,
      "learning_rate": 2.551739363928406e-06,
      "loss": 0.1889,
      "step": 12902
    },
    {
      "epoch": 0.6663051897753679,
      "grad_norm": 6.723742485046387,
      "learning_rate": 2.5510249815900843e-06,
      "loss": 0.07,
      "step": 12903
    },
    {
      "epoch": 0.6663568293312677,
      "grad_norm": 4.391429424285889,
      "learning_rate": 2.5503106650167242e-06,
      "loss": 0.0788,
      "step": 12904
    },
    {
      "epoch": 0.6664084688871675,
      "grad_norm": 20.63717269897461,
      "learning_rate": 2.549596414227506e-06,
      "loss": 0.3175,
      "step": 12905
    },
    {
      "epoch": 0.6664601084430674,
      "grad_norm": 2.2751264572143555,
      "learning_rate": 2.5488822292416137e-06,
      "loss": 0.0122,
      "step": 12906
    },
    {
      "epoch": 0.6665117479989672,
      "grad_norm": 5.539568901062012,
      "learning_rate": 2.5481681100782194e-06,
      "loss": 0.0402,
      "step": 12907
    },
    {
      "epoch": 0.666563387554867,
      "grad_norm": 6.229928493499756,
      "learning_rate": 2.5474540567565075e-06,
      "loss": 0.0978,
      "step": 12908
    },
    {
      "epoch": 0.6666150271107668,
      "grad_norm": 4.439349174499512,
      "learning_rate": 2.546740069295649e-06,
      "loss": 0.0565,
      "step": 12909
    },
    {
      "epoch": 0.6666666666666666,
      "grad_norm": 1.55403733253479,
      "learning_rate": 2.546026147714821e-06,
      "loss": 0.0197,
      "step": 12910
    },
    {
      "epoch": 0.6667183062225664,
      "grad_norm": 2.4968409538269043,
      "learning_rate": 2.5453122920331907e-06,
      "loss": 0.0346,
      "step": 12911
    },
    {
      "epoch": 0.6667699457784663,
      "grad_norm": 5.285383701324463,
      "learning_rate": 2.5445985022699287e-06,
      "loss": 0.0851,
      "step": 12912
    },
    {
      "epoch": 0.6668215853343661,
      "grad_norm": 4.017685890197754,
      "learning_rate": 2.543884778444206e-06,
      "loss": 0.06,
      "step": 12913
    },
    {
      "epoch": 0.6668732248902659,
      "grad_norm": 6.033015727996826,
      "learning_rate": 2.5431711205751895e-06,
      "loss": 0.0686,
      "step": 12914
    },
    {
      "epoch": 0.6669248644461657,
      "grad_norm": 3.8123974800109863,
      "learning_rate": 2.5424575286820406e-06,
      "loss": 0.0345,
      "step": 12915
    },
    {
      "epoch": 0.6669765040020655,
      "grad_norm": 3.6067564487457275,
      "learning_rate": 2.5417440027839225e-06,
      "loss": 0.0309,
      "step": 12916
    },
    {
      "epoch": 0.6670281435579654,
      "grad_norm": 3.0835578441619873,
      "learning_rate": 2.5410305428999983e-06,
      "loss": 0.0695,
      "step": 12917
    },
    {
      "epoch": 0.6670797831138652,
      "grad_norm": 4.425593852996826,
      "learning_rate": 2.540317149049425e-06,
      "loss": 0.0498,
      "step": 12918
    },
    {
      "epoch": 0.667131422669765,
      "grad_norm": 9.133813858032227,
      "learning_rate": 2.5396038212513617e-06,
      "loss": 0.091,
      "step": 12919
    },
    {
      "epoch": 0.6671830622256648,
      "grad_norm": 5.726680278778076,
      "learning_rate": 2.5388905595249645e-06,
      "loss": 0.0411,
      "step": 12920
    },
    {
      "epoch": 0.6672347017815646,
      "grad_norm": 2.964268207550049,
      "learning_rate": 2.5381773638893866e-06,
      "loss": 0.0337,
      "step": 12921
    },
    {
      "epoch": 0.6672863413374645,
      "grad_norm": 2.5861196517944336,
      "learning_rate": 2.5374642343637802e-06,
      "loss": 0.0362,
      "step": 12922
    },
    {
      "epoch": 0.6673379808933643,
      "grad_norm": 3.3981308937072754,
      "learning_rate": 2.5367511709672952e-06,
      "loss": 0.068,
      "step": 12923
    },
    {
      "epoch": 0.6673896204492641,
      "grad_norm": 14.573487281799316,
      "learning_rate": 2.536038173719083e-06,
      "loss": 0.142,
      "step": 12924
    },
    {
      "epoch": 0.6674412600051639,
      "grad_norm": 2.439941167831421,
      "learning_rate": 2.5353252426382867e-06,
      "loss": 0.0394,
      "step": 12925
    },
    {
      "epoch": 0.6674928995610637,
      "grad_norm": 3.0746946334838867,
      "learning_rate": 2.5346123777440513e-06,
      "loss": 0.0635,
      "step": 12926
    },
    {
      "epoch": 0.6675445391169635,
      "grad_norm": 8.213950157165527,
      "learning_rate": 2.533899579055524e-06,
      "loss": 0.0719,
      "step": 12927
    },
    {
      "epoch": 0.6675961786728634,
      "grad_norm": 3.692363977432251,
      "learning_rate": 2.533186846591846e-06,
      "loss": 0.0405,
      "step": 12928
    },
    {
      "epoch": 0.6676478182287632,
      "grad_norm": 3.2368204593658447,
      "learning_rate": 2.532474180372153e-06,
      "loss": 0.0392,
      "step": 12929
    },
    {
      "epoch": 0.667699457784663,
      "grad_norm": 11.39273738861084,
      "learning_rate": 2.531761580415586e-06,
      "loss": 0.1064,
      "step": 12930
    },
    {
      "epoch": 0.6677510973405628,
      "grad_norm": 18.197429656982422,
      "learning_rate": 2.5310490467412786e-06,
      "loss": 0.123,
      "step": 12931
    },
    {
      "epoch": 0.6678027368964627,
      "grad_norm": 4.600941181182861,
      "learning_rate": 2.5303365793683704e-06,
      "loss": 0.1023,
      "step": 12932
    },
    {
      "epoch": 0.6678543764523626,
      "grad_norm": 1.5378564596176147,
      "learning_rate": 2.529624178315989e-06,
      "loss": 0.022,
      "step": 12933
    },
    {
      "epoch": 0.6679060160082624,
      "grad_norm": 13.795268058776855,
      "learning_rate": 2.528911843603268e-06,
      "loss": 0.0745,
      "step": 12934
    },
    {
      "epoch": 0.6679576555641622,
      "grad_norm": 6.031709671020508,
      "learning_rate": 2.5281995752493348e-06,
      "loss": 0.0277,
      "step": 12935
    },
    {
      "epoch": 0.668009295120062,
      "grad_norm": 3.5068769454956055,
      "learning_rate": 2.5274873732733174e-06,
      "loss": 0.0681,
      "step": 12936
    },
    {
      "epoch": 0.6680609346759618,
      "grad_norm": 8.995674133300781,
      "learning_rate": 2.526775237694342e-06,
      "loss": 0.0582,
      "step": 12937
    },
    {
      "epoch": 0.6681125742318617,
      "grad_norm": 4.412352085113525,
      "learning_rate": 2.526063168531532e-06,
      "loss": 0.0474,
      "step": 12938
    },
    {
      "epoch": 0.6681642137877615,
      "grad_norm": 1.6649821996688843,
      "learning_rate": 2.525351165804008e-06,
      "loss": 0.0269,
      "step": 12939
    },
    {
      "epoch": 0.6682158533436613,
      "grad_norm": 6.744621753692627,
      "learning_rate": 2.5246392295308917e-06,
      "loss": 0.0554,
      "step": 12940
    },
    {
      "epoch": 0.6682674928995611,
      "grad_norm": 2.6595969200134277,
      "learning_rate": 2.523927359731302e-06,
      "loss": 0.0351,
      "step": 12941
    },
    {
      "epoch": 0.6683191324554609,
      "grad_norm": 8.026155471801758,
      "learning_rate": 2.5232155564243533e-06,
      "loss": 0.099,
      "step": 12942
    },
    {
      "epoch": 0.6683707720113607,
      "grad_norm": 7.337757110595703,
      "learning_rate": 2.5225038196291646e-06,
      "loss": 0.082,
      "step": 12943
    },
    {
      "epoch": 0.6684224115672606,
      "grad_norm": 2.273899555206299,
      "learning_rate": 2.5217921493648433e-06,
      "loss": 0.0254,
      "step": 12944
    },
    {
      "epoch": 0.6684740511231604,
      "grad_norm": 16.80782127380371,
      "learning_rate": 2.5210805456505026e-06,
      "loss": 0.0924,
      "step": 12945
    },
    {
      "epoch": 0.6685256906790602,
      "grad_norm": 2.5161073207855225,
      "learning_rate": 2.520369008505256e-06,
      "loss": 0.0244,
      "step": 12946
    },
    {
      "epoch": 0.66857733023496,
      "grad_norm": 3.2824459075927734,
      "learning_rate": 2.5196575379482057e-06,
      "loss": 0.0308,
      "step": 12947
    },
    {
      "epoch": 0.6686289697908598,
      "grad_norm": 2.2333085536956787,
      "learning_rate": 2.51894613399846e-06,
      "loss": 0.0292,
      "step": 12948
    },
    {
      "epoch": 0.6686806093467597,
      "grad_norm": 2.6918890476226807,
      "learning_rate": 2.518234796675123e-06,
      "loss": 0.0395,
      "step": 12949
    },
    {
      "epoch": 0.6687322489026595,
      "grad_norm": 8.18360424041748,
      "learning_rate": 2.5175235259972976e-06,
      "loss": 0.0404,
      "step": 12950
    },
    {
      "epoch": 0.6687838884585593,
      "grad_norm": 3.408018112182617,
      "learning_rate": 2.5168123219840823e-06,
      "loss": 0.0366,
      "step": 12951
    },
    {
      "epoch": 0.6688355280144591,
      "grad_norm": 6.992154598236084,
      "learning_rate": 2.5161011846545773e-06,
      "loss": 0.1045,
      "step": 12952
    },
    {
      "epoch": 0.6688871675703589,
      "grad_norm": 2.199437141418457,
      "learning_rate": 2.5153901140278797e-06,
      "loss": 0.0221,
      "step": 12953
    },
    {
      "epoch": 0.6689388071262587,
      "grad_norm": 13.832008361816406,
      "learning_rate": 2.5146791101230837e-06,
      "loss": 0.1461,
      "step": 12954
    },
    {
      "epoch": 0.6689904466821586,
      "grad_norm": 4.491519451141357,
      "learning_rate": 2.5139681729592836e-06,
      "loss": 0.0614,
      "step": 12955
    },
    {
      "epoch": 0.6690420862380584,
      "grad_norm": 8.886078834533691,
      "learning_rate": 2.5132573025555707e-06,
      "loss": 0.1202,
      "step": 12956
    },
    {
      "epoch": 0.6690937257939582,
      "grad_norm": 8.875821113586426,
      "learning_rate": 2.5125464989310366e-06,
      "loss": 0.1072,
      "step": 12957
    },
    {
      "epoch": 0.669145365349858,
      "grad_norm": 6.545002460479736,
      "learning_rate": 2.511835762104763e-06,
      "loss": 0.0357,
      "step": 12958
    },
    {
      "epoch": 0.6691970049057578,
      "grad_norm": 7.27475643157959,
      "learning_rate": 2.511125092095843e-06,
      "loss": 0.0537,
      "step": 12959
    },
    {
      "epoch": 0.6692486444616577,
      "grad_norm": 24.19255828857422,
      "learning_rate": 2.510414488923357e-06,
      "loss": 0.1443,
      "step": 12960
    },
    {
      "epoch": 0.6693002840175575,
      "grad_norm": 5.9114885330200195,
      "learning_rate": 2.509703952606393e-06,
      "loss": 0.0489,
      "step": 12961
    },
    {
      "epoch": 0.6693519235734573,
      "grad_norm": 2.8124210834503174,
      "learning_rate": 2.508993483164024e-06,
      "loss": 0.0518,
      "step": 12962
    },
    {
      "epoch": 0.6694035631293571,
      "grad_norm": 4.470129489898682,
      "learning_rate": 2.5082830806153337e-06,
      "loss": 0.0875,
      "step": 12963
    },
    {
      "epoch": 0.6694552026852569,
      "grad_norm": 3.753633499145508,
      "learning_rate": 2.5075727449793956e-06,
      "loss": 0.0363,
      "step": 12964
    },
    {
      "epoch": 0.6695068422411568,
      "grad_norm": 2.467106342315674,
      "learning_rate": 2.5068624762752924e-06,
      "loss": 0.0175,
      "step": 12965
    },
    {
      "epoch": 0.6695584817970566,
      "grad_norm": 5.088837146759033,
      "learning_rate": 2.5061522745220913e-06,
      "loss": 0.0627,
      "step": 12966
    },
    {
      "epoch": 0.6696101213529564,
      "grad_norm": 6.608964443206787,
      "learning_rate": 2.505442139738866e-06,
      "loss": 0.0576,
      "step": 12967
    },
    {
      "epoch": 0.6696617609088562,
      "grad_norm": 3.8776142597198486,
      "learning_rate": 2.5047320719446856e-06,
      "loss": 0.0238,
      "step": 12968
    },
    {
      "epoch": 0.669713400464756,
      "grad_norm": 6.706936836242676,
      "learning_rate": 2.5040220711586207e-06,
      "loss": 0.041,
      "step": 12969
    },
    {
      "epoch": 0.6697650400206558,
      "grad_norm": 7.190527439117432,
      "learning_rate": 2.503312137399735e-06,
      "loss": 0.0689,
      "step": 12970
    },
    {
      "epoch": 0.6698166795765557,
      "grad_norm": 8.729595184326172,
      "learning_rate": 2.5026022706870944e-06,
      "loss": 0.0939,
      "step": 12971
    },
    {
      "epoch": 0.6698683191324555,
      "grad_norm": 4.036057472229004,
      "learning_rate": 2.5018924710397623e-06,
      "loss": 0.0332,
      "step": 12972
    },
    {
      "epoch": 0.6699199586883553,
      "grad_norm": 4.637510776519775,
      "learning_rate": 2.501182738476799e-06,
      "loss": 0.0831,
      "step": 12973
    },
    {
      "epoch": 0.6699715982442551,
      "grad_norm": 4.169833660125732,
      "learning_rate": 2.5004730730172635e-06,
      "loss": 0.0325,
      "step": 12974
    },
    {
      "epoch": 0.6700232378001549,
      "grad_norm": 8.133556365966797,
      "learning_rate": 2.4997634746802137e-06,
      "loss": 0.0547,
      "step": 12975
    },
    {
      "epoch": 0.6700748773560548,
      "grad_norm": 2.8671629428863525,
      "learning_rate": 2.499053943484707e-06,
      "loss": 0.025,
      "step": 12976
    },
    {
      "epoch": 0.6701265169119546,
      "grad_norm": 8.752161979675293,
      "learning_rate": 2.4983444794497905e-06,
      "loss": 0.1382,
      "step": 12977
    },
    {
      "epoch": 0.6701781564678544,
      "grad_norm": 6.028860092163086,
      "learning_rate": 2.497635082594524e-06,
      "loss": 0.0553,
      "step": 12978
    },
    {
      "epoch": 0.6702297960237542,
      "grad_norm": 2.5780227184295654,
      "learning_rate": 2.496925752937956e-06,
      "loss": 0.0289,
      "step": 12979
    },
    {
      "epoch": 0.670281435579654,
      "grad_norm": 3.3859028816223145,
      "learning_rate": 2.4962164904991313e-06,
      "loss": 0.0244,
      "step": 12980
    },
    {
      "epoch": 0.6703330751355538,
      "grad_norm": 4.011892318725586,
      "learning_rate": 2.4955072952970993e-06,
      "loss": 0.0381,
      "step": 12981
    },
    {
      "epoch": 0.6703847146914537,
      "grad_norm": 4.033030986785889,
      "learning_rate": 2.4947981673509038e-06,
      "loss": 0.0474,
      "step": 12982
    },
    {
      "epoch": 0.6704363542473535,
      "grad_norm": 2.563612222671509,
      "learning_rate": 2.4940891066795886e-06,
      "loss": 0.0414,
      "step": 12983
    },
    {
      "epoch": 0.6704879938032533,
      "grad_norm": 5.099480152130127,
      "learning_rate": 2.493380113302194e-06,
      "loss": 0.0351,
      "step": 12984
    },
    {
      "epoch": 0.6705396333591531,
      "grad_norm": 3.4962081909179688,
      "learning_rate": 2.4926711872377597e-06,
      "loss": 0.0381,
      "step": 12985
    },
    {
      "epoch": 0.6705912729150529,
      "grad_norm": 3.509113311767578,
      "learning_rate": 2.4919623285053235e-06,
      "loss": 0.0312,
      "step": 12986
    },
    {
      "epoch": 0.6706429124709528,
      "grad_norm": 4.110750675201416,
      "learning_rate": 2.491253537123921e-06,
      "loss": 0.0554,
      "step": 12987
    },
    {
      "epoch": 0.6706945520268526,
      "grad_norm": 5.120530605316162,
      "learning_rate": 2.490544813112586e-06,
      "loss": 0.0851,
      "step": 12988
    },
    {
      "epoch": 0.6707461915827524,
      "grad_norm": 3.417825698852539,
      "learning_rate": 2.4898361564903505e-06,
      "loss": 0.043,
      "step": 12989
    },
    {
      "epoch": 0.6707978311386522,
      "grad_norm": 1.8351973295211792,
      "learning_rate": 2.4891275672762473e-06,
      "loss": 0.0163,
      "step": 12990
    },
    {
      "epoch": 0.670849470694552,
      "grad_norm": 2.9565176963806152,
      "learning_rate": 2.488419045489299e-06,
      "loss": 0.0315,
      "step": 12991
    },
    {
      "epoch": 0.6709011102504518,
      "grad_norm": 3.8084917068481445,
      "learning_rate": 2.4877105911485377e-06,
      "loss": 0.0776,
      "step": 12992
    },
    {
      "epoch": 0.6709527498063517,
      "grad_norm": 3.875654697418213,
      "learning_rate": 2.4870022042729865e-06,
      "loss": 0.0651,
      "step": 12993
    },
    {
      "epoch": 0.6710043893622515,
      "grad_norm": 3.228867530822754,
      "learning_rate": 2.4862938848816703e-06,
      "loss": 0.0544,
      "step": 12994
    },
    {
      "epoch": 0.6710560289181513,
      "grad_norm": 2.4828827381134033,
      "learning_rate": 2.485585632993607e-06,
      "loss": 0.0421,
      "step": 12995
    },
    {
      "epoch": 0.6711076684740511,
      "grad_norm": 4.2178168296813965,
      "learning_rate": 2.4848774486278175e-06,
      "loss": 0.0728,
      "step": 12996
    },
    {
      "epoch": 0.6711593080299509,
      "grad_norm": 5.118000030517578,
      "learning_rate": 2.4841693318033177e-06,
      "loss": 0.0527,
      "step": 12997
    },
    {
      "epoch": 0.6712109475858508,
      "grad_norm": 3.9121532440185547,
      "learning_rate": 2.4834612825391292e-06,
      "loss": 0.0632,
      "step": 12998
    },
    {
      "epoch": 0.6712625871417506,
      "grad_norm": 2.955549955368042,
      "learning_rate": 2.48275330085426e-06,
      "loss": 0.0401,
      "step": 12999
    },
    {
      "epoch": 0.6713142266976504,
      "grad_norm": 2.1814935207366943,
      "learning_rate": 2.482045386767724e-06,
      "loss": 0.0282,
      "step": 13000
    },
    {
      "epoch": 0.6713658662535502,
      "grad_norm": 2.0936646461486816,
      "learning_rate": 2.481337540298533e-06,
      "loss": 0.0224,
      "step": 13001
    },
    {
      "epoch": 0.67141750580945,
      "grad_norm": 3.404218912124634,
      "learning_rate": 2.480629761465694e-06,
      "loss": 0.0546,
      "step": 13002
    },
    {
      "epoch": 0.6714691453653499,
      "grad_norm": 4.7304301261901855,
      "learning_rate": 2.4799220502882143e-06,
      "loss": 0.0714,
      "step": 13003
    },
    {
      "epoch": 0.6715207849212497,
      "grad_norm": 17.56608009338379,
      "learning_rate": 2.4792144067850986e-06,
      "loss": 0.0598,
      "step": 13004
    },
    {
      "epoch": 0.6715724244771495,
      "grad_norm": 3.903858184814453,
      "learning_rate": 2.47850683097535e-06,
      "loss": 0.0418,
      "step": 13005
    },
    {
      "epoch": 0.6716240640330493,
      "grad_norm": 2.9618122577667236,
      "learning_rate": 2.4777993228779705e-06,
      "loss": 0.0548,
      "step": 13006
    },
    {
      "epoch": 0.6716757035889491,
      "grad_norm": 14.77334976196289,
      "learning_rate": 2.4770918825119594e-06,
      "loss": 0.1477,
      "step": 13007
    },
    {
      "epoch": 0.6717273431448489,
      "grad_norm": 2.9791438579559326,
      "learning_rate": 2.4763845098963135e-06,
      "loss": 0.0369,
      "step": 13008
    },
    {
      "epoch": 0.6717789827007488,
      "grad_norm": 4.473401069641113,
      "learning_rate": 2.4756772050500318e-06,
      "loss": 0.053,
      "step": 13009
    },
    {
      "epoch": 0.6718306222566486,
      "grad_norm": 14.886780738830566,
      "learning_rate": 2.4749699679921013e-06,
      "loss": 0.2169,
      "step": 13010
    },
    {
      "epoch": 0.6718822618125484,
      "grad_norm": 1.490741491317749,
      "learning_rate": 2.4742627987415207e-06,
      "loss": 0.0222,
      "step": 13011
    },
    {
      "epoch": 0.6719339013684482,
      "grad_norm": 5.6050848960876465,
      "learning_rate": 2.47355569731728e-06,
      "loss": 0.1224,
      "step": 13012
    },
    {
      "epoch": 0.671985540924348,
      "grad_norm": 4.171627521514893,
      "learning_rate": 2.472848663738364e-06,
      "loss": 0.0396,
      "step": 13013
    },
    {
      "epoch": 0.6720371804802479,
      "grad_norm": 9.56411075592041,
      "learning_rate": 2.472141698023762e-06,
      "loss": 0.091,
      "step": 13014
    },
    {
      "epoch": 0.6720888200361477,
      "grad_norm": 7.019248962402344,
      "learning_rate": 2.4714348001924584e-06,
      "loss": 0.074,
      "step": 13015
    },
    {
      "epoch": 0.6721404595920475,
      "grad_norm": 4.678829669952393,
      "learning_rate": 2.4707279702634357e-06,
      "loss": 0.0303,
      "step": 13016
    },
    {
      "epoch": 0.6721920991479473,
      "grad_norm": 3.4566309452056885,
      "learning_rate": 2.4700212082556764e-06,
      "loss": 0.0402,
      "step": 13017
    },
    {
      "epoch": 0.6722437387038471,
      "grad_norm": 3.23567795753479,
      "learning_rate": 2.4693145141881588e-06,
      "loss": 0.0542,
      "step": 13018
    },
    {
      "epoch": 0.672295378259747,
      "grad_norm": 3.614107370376587,
      "learning_rate": 2.468607888079861e-06,
      "loss": 0.0572,
      "step": 13019
    },
    {
      "epoch": 0.6723470178156468,
      "grad_norm": 3.0921261310577393,
      "learning_rate": 2.4679013299497587e-06,
      "loss": 0.0567,
      "step": 13020
    },
    {
      "epoch": 0.6723986573715466,
      "grad_norm": 4.848458290100098,
      "learning_rate": 2.4671948398168268e-06,
      "loss": 0.0769,
      "step": 13021
    },
    {
      "epoch": 0.6724502969274464,
      "grad_norm": 2.4687254428863525,
      "learning_rate": 2.4664884177000357e-06,
      "loss": 0.0382,
      "step": 13022
    },
    {
      "epoch": 0.6725019364833462,
      "grad_norm": 11.412627220153809,
      "learning_rate": 2.4657820636183586e-06,
      "loss": 0.0381,
      "step": 13023
    },
    {
      "epoch": 0.672553576039246,
      "grad_norm": 2.702340841293335,
      "learning_rate": 2.4650757775907585e-06,
      "loss": 0.0284,
      "step": 13024
    },
    {
      "epoch": 0.6726052155951459,
      "grad_norm": 1.7377815246582031,
      "learning_rate": 2.464369559636207e-06,
      "loss": 0.03,
      "step": 13025
    },
    {
      "epoch": 0.6726568551510457,
      "grad_norm": 3.1714799404144287,
      "learning_rate": 2.4636634097736684e-06,
      "loss": 0.0454,
      "step": 13026
    },
    {
      "epoch": 0.6727084947069455,
      "grad_norm": 2.9773712158203125,
      "learning_rate": 2.462957328022106e-06,
      "loss": 0.0442,
      "step": 13027
    },
    {
      "epoch": 0.6727601342628453,
      "grad_norm": 4.412643909454346,
      "learning_rate": 2.462251314400478e-06,
      "loss": 0.0457,
      "step": 13028
    },
    {
      "epoch": 0.6728117738187451,
      "grad_norm": 4.408515453338623,
      "learning_rate": 2.4615453689277436e-06,
      "loss": 0.0581,
      "step": 13029
    },
    {
      "epoch": 0.672863413374645,
      "grad_norm": 4.844653129577637,
      "learning_rate": 2.460839491622864e-06,
      "loss": 0.0949,
      "step": 13030
    },
    {
      "epoch": 0.6729150529305448,
      "grad_norm": 2.8953514099121094,
      "learning_rate": 2.460133682504795e-06,
      "loss": 0.0265,
      "step": 13031
    },
    {
      "epoch": 0.6729666924864446,
      "grad_norm": 2.4435105323791504,
      "learning_rate": 2.4594279415924866e-06,
      "loss": 0.034,
      "step": 13032
    },
    {
      "epoch": 0.6730183320423444,
      "grad_norm": 3.8331401348114014,
      "learning_rate": 2.4587222689048927e-06,
      "loss": 0.0421,
      "step": 13033
    },
    {
      "epoch": 0.6730699715982442,
      "grad_norm": 4.954230308532715,
      "learning_rate": 2.4580166644609637e-06,
      "loss": 0.0791,
      "step": 13034
    },
    {
      "epoch": 0.673121611154144,
      "grad_norm": 2.266693592071533,
      "learning_rate": 2.457311128279648e-06,
      "loss": 0.0296,
      "step": 13035
    },
    {
      "epoch": 0.6731732507100439,
      "grad_norm": 2.810157299041748,
      "learning_rate": 2.4566056603798923e-06,
      "loss": 0.0446,
      "step": 13036
    },
    {
      "epoch": 0.6732248902659437,
      "grad_norm": 7.987293243408203,
      "learning_rate": 2.4559002607806397e-06,
      "loss": 0.0719,
      "step": 13037
    },
    {
      "epoch": 0.6732765298218435,
      "grad_norm": 3.62688946723938,
      "learning_rate": 2.4551949295008353e-06,
      "loss": 0.0531,
      "step": 13038
    },
    {
      "epoch": 0.6733281693777433,
      "grad_norm": 2.57342267036438,
      "learning_rate": 2.4544896665594186e-06,
      "loss": 0.0603,
      "step": 13039
    },
    {
      "epoch": 0.6733798089336431,
      "grad_norm": 2.8701369762420654,
      "learning_rate": 2.4537844719753294e-06,
      "loss": 0.0471,
      "step": 13040
    },
    {
      "epoch": 0.673431448489543,
      "grad_norm": 7.3397216796875,
      "learning_rate": 2.4530793457675052e-06,
      "loss": 0.057,
      "step": 13041
    },
    {
      "epoch": 0.6734830880454428,
      "grad_norm": 3.7688417434692383,
      "learning_rate": 2.4523742879548827e-06,
      "loss": 0.0399,
      "step": 13042
    },
    {
      "epoch": 0.6735347276013426,
      "grad_norm": 1.5469270944595337,
      "learning_rate": 2.4516692985563905e-06,
      "loss": 0.0222,
      "step": 13043
    },
    {
      "epoch": 0.6735863671572424,
      "grad_norm": 4.297579288482666,
      "learning_rate": 2.4509643775909666e-06,
      "loss": 0.0525,
      "step": 13044
    },
    {
      "epoch": 0.6736380067131422,
      "grad_norm": 2.5368664264678955,
      "learning_rate": 2.450259525077539e-06,
      "loss": 0.0299,
      "step": 13045
    },
    {
      "epoch": 0.673689646269042,
      "grad_norm": 3.8824267387390137,
      "learning_rate": 2.4495547410350344e-06,
      "loss": 0.0316,
      "step": 13046
    },
    {
      "epoch": 0.6737412858249419,
      "grad_norm": 11.207176208496094,
      "learning_rate": 2.44885002548238e-06,
      "loss": 0.1223,
      "step": 13047
    },
    {
      "epoch": 0.6737929253808417,
      "grad_norm": 3.312213897705078,
      "learning_rate": 2.448145378438498e-06,
      "loss": 0.0534,
      "step": 13048
    },
    {
      "epoch": 0.6738445649367415,
      "grad_norm": 6.603152751922607,
      "learning_rate": 2.4474407999223176e-06,
      "loss": 0.0717,
      "step": 13049
    },
    {
      "epoch": 0.6738962044926413,
      "grad_norm": 6.513789176940918,
      "learning_rate": 2.446736289952753e-06,
      "loss": 0.0534,
      "step": 13050
    },
    {
      "epoch": 0.6739478440485411,
      "grad_norm": 14.848703384399414,
      "learning_rate": 2.446031848548726e-06,
      "loss": 0.058,
      "step": 13051
    },
    {
      "epoch": 0.673999483604441,
      "grad_norm": 2.5969061851501465,
      "learning_rate": 2.445327475729153e-06,
      "loss": 0.0479,
      "step": 13052
    },
    {
      "epoch": 0.6740511231603408,
      "grad_norm": 2.596951723098755,
      "learning_rate": 2.4446231715129498e-06,
      "loss": 0.02,
      "step": 13053
    },
    {
      "epoch": 0.6741027627162406,
      "grad_norm": 15.68122386932373,
      "learning_rate": 2.4439189359190292e-06,
      "loss": 0.1484,
      "step": 13054
    },
    {
      "epoch": 0.6741544022721404,
      "grad_norm": 13.72541332244873,
      "learning_rate": 2.443214768966303e-06,
      "loss": 0.1582,
      "step": 13055
    },
    {
      "epoch": 0.6742060418280402,
      "grad_norm": 15.80510425567627,
      "learning_rate": 2.442510670673684e-06,
      "loss": 0.2469,
      "step": 13056
    },
    {
      "epoch": 0.67425768138394,
      "grad_norm": 7.639889240264893,
      "learning_rate": 2.4418066410600728e-06,
      "loss": 0.066,
      "step": 13057
    },
    {
      "epoch": 0.6743093209398399,
      "grad_norm": 2.225391387939453,
      "learning_rate": 2.441102680144382e-06,
      "loss": 0.0421,
      "step": 13058
    },
    {
      "epoch": 0.6743609604957398,
      "grad_norm": 15.27952766418457,
      "learning_rate": 2.4403987879455136e-06,
      "loss": 0.0918,
      "step": 13059
    },
    {
      "epoch": 0.6744126000516396,
      "grad_norm": 2.838897705078125,
      "learning_rate": 2.4396949644823718e-06,
      "loss": 0.0541,
      "step": 13060
    },
    {
      "epoch": 0.6744642396075394,
      "grad_norm": 4.432607650756836,
      "learning_rate": 2.438991209773854e-06,
      "loss": 0.0675,
      "step": 13061
    },
    {
      "epoch": 0.6745158791634392,
      "grad_norm": 2.1591250896453857,
      "learning_rate": 2.438287523838858e-06,
      "loss": 0.035,
      "step": 13062
    },
    {
      "epoch": 0.6745675187193391,
      "grad_norm": 23.716224670410156,
      "learning_rate": 2.437583906696285e-06,
      "loss": 0.2664,
      "step": 13063
    },
    {
      "epoch": 0.6746191582752389,
      "grad_norm": 2.5113377571105957,
      "learning_rate": 2.4368803583650297e-06,
      "loss": 0.0458,
      "step": 13064
    },
    {
      "epoch": 0.6746707978311387,
      "grad_norm": 2.925093650817871,
      "learning_rate": 2.4361768788639815e-06,
      "loss": 0.0398,
      "step": 13065
    },
    {
      "epoch": 0.6747224373870385,
      "grad_norm": 4.504459857940674,
      "learning_rate": 2.435473468212034e-06,
      "loss": 0.0351,
      "step": 13066
    },
    {
      "epoch": 0.6747740769429383,
      "grad_norm": 3.9534003734588623,
      "learning_rate": 2.4347701264280767e-06,
      "loss": 0.0196,
      "step": 13067
    },
    {
      "epoch": 0.6748257164988382,
      "grad_norm": 5.361576080322266,
      "learning_rate": 2.4340668535309957e-06,
      "loss": 0.0368,
      "step": 13068
    },
    {
      "epoch": 0.674877356054738,
      "grad_norm": 6.826662063598633,
      "learning_rate": 2.433363649539679e-06,
      "loss": 0.0416,
      "step": 13069
    },
    {
      "epoch": 0.6749289956106378,
      "grad_norm": 16.32998275756836,
      "learning_rate": 2.4326605144730088e-06,
      "loss": 0.0701,
      "step": 13070
    },
    {
      "epoch": 0.6749806351665376,
      "grad_norm": 4.58186674118042,
      "learning_rate": 2.4319574483498678e-06,
      "loss": 0.0584,
      "step": 13071
    },
    {
      "epoch": 0.6750322747224374,
      "grad_norm": 19.867395401000977,
      "learning_rate": 2.4312544511891356e-06,
      "loss": 0.05,
      "step": 13072
    },
    {
      "epoch": 0.6750839142783372,
      "grad_norm": 3.1180076599121094,
      "learning_rate": 2.430551523009691e-06,
      "loss": 0.0389,
      "step": 13073
    },
    {
      "epoch": 0.6751355538342371,
      "grad_norm": 2.8559043407440186,
      "learning_rate": 2.429848663830411e-06,
      "loss": 0.0269,
      "step": 13074
    },
    {
      "epoch": 0.6751871933901369,
      "grad_norm": 11.320276260375977,
      "learning_rate": 2.4291458736701716e-06,
      "loss": 0.1705,
      "step": 13075
    },
    {
      "epoch": 0.6752388329460367,
      "grad_norm": 11.348836898803711,
      "learning_rate": 2.42844315254784e-06,
      "loss": 0.0838,
      "step": 13076
    },
    {
      "epoch": 0.6752904725019365,
      "grad_norm": 2.916085720062256,
      "learning_rate": 2.4277405004822925e-06,
      "loss": 0.0666,
      "step": 13077
    },
    {
      "epoch": 0.6753421120578363,
      "grad_norm": 2.536719799041748,
      "learning_rate": 2.427037917492398e-06,
      "loss": 0.0462,
      "step": 13078
    },
    {
      "epoch": 0.6753937516137362,
      "grad_norm": 3.6657068729400635,
      "learning_rate": 2.426335403597021e-06,
      "loss": 0.0553,
      "step": 13079
    },
    {
      "epoch": 0.675445391169636,
      "grad_norm": 3.7080273628234863,
      "learning_rate": 2.425632958815028e-06,
      "loss": 0.0342,
      "step": 13080
    },
    {
      "epoch": 0.6754970307255358,
      "grad_norm": 3.271716833114624,
      "learning_rate": 2.42493058316528e-06,
      "loss": 0.0445,
      "step": 13081
    },
    {
      "epoch": 0.6755486702814356,
      "grad_norm": 2.8736698627471924,
      "learning_rate": 2.424228276666646e-06,
      "loss": 0.0539,
      "step": 13082
    },
    {
      "epoch": 0.6756003098373354,
      "grad_norm": 1.4135016202926636,
      "learning_rate": 2.4235260393379787e-06,
      "loss": 0.0189,
      "step": 13083
    },
    {
      "epoch": 0.6756519493932353,
      "grad_norm": 1.8832508325576782,
      "learning_rate": 2.422823871198138e-06,
      "loss": 0.0296,
      "step": 13084
    },
    {
      "epoch": 0.6757035889491351,
      "grad_norm": 21.00322723388672,
      "learning_rate": 2.4221217722659813e-06,
      "loss": 0.0631,
      "step": 13085
    },
    {
      "epoch": 0.6757552285050349,
      "grad_norm": 8.535290718078613,
      "learning_rate": 2.4214197425603615e-06,
      "loss": 0.0889,
      "step": 13086
    },
    {
      "epoch": 0.6758068680609347,
      "grad_norm": 3.68654465675354,
      "learning_rate": 2.420717782100131e-06,
      "loss": 0.0412,
      "step": 13087
    },
    {
      "epoch": 0.6758585076168345,
      "grad_norm": 1.1058918237686157,
      "learning_rate": 2.4200158909041403e-06,
      "loss": 0.0142,
      "step": 13088
    },
    {
      "epoch": 0.6759101471727343,
      "grad_norm": 2.5891506671905518,
      "learning_rate": 2.419314068991241e-06,
      "loss": 0.0427,
      "step": 13089
    },
    {
      "epoch": 0.6759617867286342,
      "grad_norm": 6.390981674194336,
      "learning_rate": 2.418612316380273e-06,
      "loss": 0.0898,
      "step": 13090
    },
    {
      "epoch": 0.676013426284534,
      "grad_norm": 29.242422103881836,
      "learning_rate": 2.4179106330900876e-06,
      "loss": 0.1681,
      "step": 13091
    },
    {
      "epoch": 0.6760650658404338,
      "grad_norm": 8.01323127746582,
      "learning_rate": 2.417209019139526e-06,
      "loss": 0.1227,
      "step": 13092
    },
    {
      "epoch": 0.6761167053963336,
      "grad_norm": 4.422117710113525,
      "learning_rate": 2.4165074745474305e-06,
      "loss": 0.0885,
      "step": 13093
    },
    {
      "epoch": 0.6761683449522334,
      "grad_norm": 3.3589773178100586,
      "learning_rate": 2.415805999332637e-06,
      "loss": 0.0485,
      "step": 13094
    },
    {
      "epoch": 0.6762199845081333,
      "grad_norm": 5.621755599975586,
      "learning_rate": 2.415104593513984e-06,
      "loss": 0.0901,
      "step": 13095
    },
    {
      "epoch": 0.6762716240640331,
      "grad_norm": 4.151707649230957,
      "learning_rate": 2.4144032571103095e-06,
      "loss": 0.0529,
      "step": 13096
    },
    {
      "epoch": 0.6763232636199329,
      "grad_norm": 4.7275824546813965,
      "learning_rate": 2.4137019901404475e-06,
      "loss": 0.0645,
      "step": 13097
    },
    {
      "epoch": 0.6763749031758327,
      "grad_norm": 2.0362277030944824,
      "learning_rate": 2.4130007926232264e-06,
      "loss": 0.0269,
      "step": 13098
    },
    {
      "epoch": 0.6764265427317325,
      "grad_norm": 2.8492393493652344,
      "learning_rate": 2.412299664577478e-06,
      "loss": 0.0302,
      "step": 13099
    },
    {
      "epoch": 0.6764781822876323,
      "grad_norm": 60.38947677612305,
      "learning_rate": 2.411598606022031e-06,
      "loss": 0.0886,
      "step": 13100
    },
    {
      "epoch": 0.6765298218435322,
      "grad_norm": 3.6380834579467773,
      "learning_rate": 2.4108976169757103e-06,
      "loss": 0.0535,
      "step": 13101
    },
    {
      "epoch": 0.676581461399432,
      "grad_norm": 5.885711193084717,
      "learning_rate": 2.410196697457342e-06,
      "loss": 0.0216,
      "step": 13102
    },
    {
      "epoch": 0.6766331009553318,
      "grad_norm": 14.314859390258789,
      "learning_rate": 2.409495847485748e-06,
      "loss": 0.229,
      "step": 13103
    },
    {
      "epoch": 0.6766847405112316,
      "grad_norm": 4.001273155212402,
      "learning_rate": 2.408795067079748e-06,
      "loss": 0.0484,
      "step": 13104
    },
    {
      "epoch": 0.6767363800671314,
      "grad_norm": 39.047481536865234,
      "learning_rate": 2.4080943562581614e-06,
      "loss": 0.111,
      "step": 13105
    },
    {
      "epoch": 0.6767880196230313,
      "grad_norm": 2.303830623626709,
      "learning_rate": 2.4073937150398065e-06,
      "loss": 0.0429,
      "step": 13106
    },
    {
      "epoch": 0.6768396591789311,
      "grad_norm": 2.600696086883545,
      "learning_rate": 2.4066931434434963e-06,
      "loss": 0.0531,
      "step": 13107
    },
    {
      "epoch": 0.6768912987348309,
      "grad_norm": 3.2074620723724365,
      "learning_rate": 2.4059926414880475e-06,
      "loss": 0.0299,
      "step": 13108
    },
    {
      "epoch": 0.6769429382907307,
      "grad_norm": 4.473392009735107,
      "learning_rate": 2.4052922091922643e-06,
      "loss": 0.0866,
      "step": 13109
    },
    {
      "epoch": 0.6769945778466305,
      "grad_norm": 5.976121425628662,
      "learning_rate": 2.404591846574963e-06,
      "loss": 0.0746,
      "step": 13110
    },
    {
      "epoch": 0.6770462174025303,
      "grad_norm": 10.044296264648438,
      "learning_rate": 2.403891553654951e-06,
      "loss": 0.0957,
      "step": 13111
    },
    {
      "epoch": 0.6770978569584302,
      "grad_norm": 7.102682590484619,
      "learning_rate": 2.40319133045103e-06,
      "loss": 0.0835,
      "step": 13112
    },
    {
      "epoch": 0.67714949651433,
      "grad_norm": 5.176320552825928,
      "learning_rate": 2.402491176982006e-06,
      "loss": 0.1441,
      "step": 13113
    },
    {
      "epoch": 0.6772011360702298,
      "grad_norm": 8.161581039428711,
      "learning_rate": 2.4017910932666784e-06,
      "loss": 0.066,
      "step": 13114
    },
    {
      "epoch": 0.6772527756261296,
      "grad_norm": 5.593484878540039,
      "learning_rate": 2.4010910793238534e-06,
      "loss": 0.0573,
      "step": 13115
    },
    {
      "epoch": 0.6773044151820294,
      "grad_norm": 6.170487403869629,
      "learning_rate": 2.4003911351723234e-06,
      "loss": 0.034,
      "step": 13116
    },
    {
      "epoch": 0.6773560547379293,
      "grad_norm": 9.023218154907227,
      "learning_rate": 2.3996912608308874e-06,
      "loss": 0.0875,
      "step": 13117
    },
    {
      "epoch": 0.6774076942938291,
      "grad_norm": 7.57507848739624,
      "learning_rate": 2.3989914563183393e-06,
      "loss": 0.0171,
      "step": 13118
    },
    {
      "epoch": 0.6774593338497289,
      "grad_norm": 2.9093358516693115,
      "learning_rate": 2.398291721653472e-06,
      "loss": 0.0438,
      "step": 13119
    },
    {
      "epoch": 0.6775109734056287,
      "grad_norm": 1.3284776210784912,
      "learning_rate": 2.3975920568550752e-06,
      "loss": 0.0223,
      "step": 13120
    },
    {
      "epoch": 0.6775626129615285,
      "grad_norm": 2.7749674320220947,
      "learning_rate": 2.3968924619419387e-06,
      "loss": 0.0494,
      "step": 13121
    },
    {
      "epoch": 0.6776142525174284,
      "grad_norm": 14.996905326843262,
      "learning_rate": 2.396192936932851e-06,
      "loss": 0.1478,
      "step": 13122
    },
    {
      "epoch": 0.6776658920733282,
      "grad_norm": 7.31981086730957,
      "learning_rate": 2.3954934818465926e-06,
      "loss": 0.1007,
      "step": 13123
    },
    {
      "epoch": 0.677717531629228,
      "grad_norm": 2.2749993801116943,
      "learning_rate": 2.3947940967019507e-06,
      "loss": 0.0296,
      "step": 13124
    },
    {
      "epoch": 0.6777691711851278,
      "grad_norm": 2.7053632736206055,
      "learning_rate": 2.394094781517706e-06,
      "loss": 0.0284,
      "step": 13125
    },
    {
      "epoch": 0.6778208107410276,
      "grad_norm": 2.900383710861206,
      "learning_rate": 2.3933955363126387e-06,
      "loss": 0.037,
      "step": 13126
    },
    {
      "epoch": 0.6778724502969274,
      "grad_norm": 6.615448474884033,
      "learning_rate": 2.3926963611055236e-06,
      "loss": 0.0634,
      "step": 13127
    },
    {
      "epoch": 0.6779240898528273,
      "grad_norm": 1.5601204633712769,
      "learning_rate": 2.3919972559151365e-06,
      "loss": 0.0335,
      "step": 13128
    },
    {
      "epoch": 0.6779757294087271,
      "grad_norm": 2.5840566158294678,
      "learning_rate": 2.391298220760254e-06,
      "loss": 0.0388,
      "step": 13129
    },
    {
      "epoch": 0.6780273689646269,
      "grad_norm": 3.523895502090454,
      "learning_rate": 2.3905992556596493e-06,
      "loss": 0.0467,
      "step": 13130
    },
    {
      "epoch": 0.6780790085205267,
      "grad_norm": 11.664867401123047,
      "learning_rate": 2.3899003606320876e-06,
      "loss": 0.0569,
      "step": 13131
    },
    {
      "epoch": 0.6781306480764265,
      "grad_norm": 8.893267631530762,
      "learning_rate": 2.3892015356963395e-06,
      "loss": 0.0912,
      "step": 13132
    },
    {
      "epoch": 0.6781822876323264,
      "grad_norm": 3.762282609939575,
      "learning_rate": 2.388502780871171e-06,
      "loss": 0.0277,
      "step": 13133
    },
    {
      "epoch": 0.6782339271882262,
      "grad_norm": 5.066977500915527,
      "learning_rate": 2.3878040961753467e-06,
      "loss": 0.0585,
      "step": 13134
    },
    {
      "epoch": 0.678285566744126,
      "grad_norm": 11.257970809936523,
      "learning_rate": 2.3871054816276297e-06,
      "loss": 0.0337,
      "step": 13135
    },
    {
      "epoch": 0.6783372063000258,
      "grad_norm": 2.2854185104370117,
      "learning_rate": 2.3864069372467794e-06,
      "loss": 0.021,
      "step": 13136
    },
    {
      "epoch": 0.6783888458559256,
      "grad_norm": 2.895324945449829,
      "learning_rate": 2.3857084630515553e-06,
      "loss": 0.0302,
      "step": 13137
    },
    {
      "epoch": 0.6784404854118254,
      "grad_norm": 11.640578269958496,
      "learning_rate": 2.3850100590607147e-06,
      "loss": 0.0757,
      "step": 13138
    },
    {
      "epoch": 0.6784921249677253,
      "grad_norm": 5.148433685302734,
      "learning_rate": 2.384311725293012e-06,
      "loss": 0.1451,
      "step": 13139
    },
    {
      "epoch": 0.6785437645236251,
      "grad_norm": 3.9637951850891113,
      "learning_rate": 2.3836134617672e-06,
      "loss": 0.0505,
      "step": 13140
    },
    {
      "epoch": 0.6785954040795249,
      "grad_norm": 6.043554306030273,
      "learning_rate": 2.382915268502033e-06,
      "loss": 0.0505,
      "step": 13141
    },
    {
      "epoch": 0.6786470436354247,
      "grad_norm": 3.8180384635925293,
      "learning_rate": 2.382217145516253e-06,
      "loss": 0.0569,
      "step": 13142
    },
    {
      "epoch": 0.6786986831913245,
      "grad_norm": 6.169572353363037,
      "learning_rate": 2.381519092828615e-06,
      "loss": 0.0522,
      "step": 13143
    },
    {
      "epoch": 0.6787503227472244,
      "grad_norm": 2.46952748298645,
      "learning_rate": 2.3808211104578627e-06,
      "loss": 0.0349,
      "step": 13144
    },
    {
      "epoch": 0.6788019623031242,
      "grad_norm": 4.887331485748291,
      "learning_rate": 2.3801231984227376e-06,
      "loss": 0.0646,
      "step": 13145
    },
    {
      "epoch": 0.678853601859024,
      "grad_norm": 2.5005440711975098,
      "learning_rate": 2.379425356741982e-06,
      "loss": 0.0476,
      "step": 13146
    },
    {
      "epoch": 0.6789052414149238,
      "grad_norm": 2.1475837230682373,
      "learning_rate": 2.378727585434335e-06,
      "loss": 0.0305,
      "step": 13147
    },
    {
      "epoch": 0.6789568809708236,
      "grad_norm": 9.855497360229492,
      "learning_rate": 2.3780298845185395e-06,
      "loss": 0.1177,
      "step": 13148
    },
    {
      "epoch": 0.6790085205267234,
      "grad_norm": 2.9740242958068848,
      "learning_rate": 2.3773322540133264e-06,
      "loss": 0.0521,
      "step": 13149
    },
    {
      "epoch": 0.6790601600826233,
      "grad_norm": 6.284585475921631,
      "learning_rate": 2.376634693937432e-06,
      "loss": 0.0755,
      "step": 13150
    },
    {
      "epoch": 0.6791117996385231,
      "grad_norm": 1.4464069604873657,
      "learning_rate": 2.375937204309588e-06,
      "loss": 0.024,
      "step": 13151
    },
    {
      "epoch": 0.6791634391944229,
      "grad_norm": 3.913058042526245,
      "learning_rate": 2.3752397851485246e-06,
      "loss": 0.077,
      "step": 13152
    },
    {
      "epoch": 0.6792150787503227,
      "grad_norm": 2.178349256515503,
      "learning_rate": 2.374542436472972e-06,
      "loss": 0.0316,
      "step": 13153
    },
    {
      "epoch": 0.6792667183062225,
      "grad_norm": 13.662489891052246,
      "learning_rate": 2.3738451583016553e-06,
      "loss": 0.058,
      "step": 13154
    },
    {
      "epoch": 0.6793183578621224,
      "grad_norm": 2.678828239440918,
      "learning_rate": 2.373147950653301e-06,
      "loss": 0.0428,
      "step": 13155
    },
    {
      "epoch": 0.6793699974180222,
      "grad_norm": 8.830041885375977,
      "learning_rate": 2.372450813546628e-06,
      "loss": 0.0722,
      "step": 13156
    },
    {
      "epoch": 0.679421636973922,
      "grad_norm": 2.806596517562866,
      "learning_rate": 2.371753747000361e-06,
      "loss": 0.0426,
      "step": 13157
    },
    {
      "epoch": 0.6794732765298218,
      "grad_norm": 2.3659729957580566,
      "learning_rate": 2.3710567510332185e-06,
      "loss": 0.0621,
      "step": 13158
    },
    {
      "epoch": 0.6795249160857216,
      "grad_norm": 2.7660882472991943,
      "learning_rate": 2.370359825663919e-06,
      "loss": 0.0421,
      "step": 13159
    },
    {
      "epoch": 0.6795765556416214,
      "grad_norm": 5.75144100189209,
      "learning_rate": 2.369662970911174e-06,
      "loss": 0.0497,
      "step": 13160
    },
    {
      "epoch": 0.6796281951975213,
      "grad_norm": 13.639877319335938,
      "learning_rate": 2.368966186793697e-06,
      "loss": 0.0627,
      "step": 13161
    },
    {
      "epoch": 0.6796798347534211,
      "grad_norm": 8.584572792053223,
      "learning_rate": 2.368269473330203e-06,
      "loss": 0.0704,
      "step": 13162
    },
    {
      "epoch": 0.6797314743093209,
      "grad_norm": 3.2901883125305176,
      "learning_rate": 2.3675728305394023e-06,
      "loss": 0.0309,
      "step": 13163
    },
    {
      "epoch": 0.6797831138652207,
      "grad_norm": 1.6006600856781006,
      "learning_rate": 2.3668762584399983e-06,
      "loss": 0.0303,
      "step": 13164
    },
    {
      "epoch": 0.6798347534211205,
      "grad_norm": 2.2247118949890137,
      "learning_rate": 2.3661797570506988e-06,
      "loss": 0.0588,
      "step": 13165
    },
    {
      "epoch": 0.6798863929770204,
      "grad_norm": 1.9891494512557983,
      "learning_rate": 2.3654833263902077e-06,
      "loss": 0.0575,
      "step": 13166
    },
    {
      "epoch": 0.6799380325329202,
      "grad_norm": 5.441124439239502,
      "learning_rate": 2.364786966477227e-06,
      "loss": 0.0804,
      "step": 13167
    },
    {
      "epoch": 0.67998967208882,
      "grad_norm": 5.371921539306641,
      "learning_rate": 2.3640906773304567e-06,
      "loss": 0.0536,
      "step": 13168
    },
    {
      "epoch": 0.6800413116447198,
      "grad_norm": 3.8195483684539795,
      "learning_rate": 2.3633944589685954e-06,
      "loss": 0.0465,
      "step": 13169
    },
    {
      "epoch": 0.6800929512006196,
      "grad_norm": 3.1297647953033447,
      "learning_rate": 2.3626983114103385e-06,
      "loss": 0.0503,
      "step": 13170
    },
    {
      "epoch": 0.6801445907565195,
      "grad_norm": 4.823495388031006,
      "learning_rate": 2.3620022346743816e-06,
      "loss": 0.0591,
      "step": 13171
    },
    {
      "epoch": 0.6801962303124193,
      "grad_norm": 3.506803512573242,
      "learning_rate": 2.361306228779416e-06,
      "loss": 0.0438,
      "step": 13172
    },
    {
      "epoch": 0.6802478698683191,
      "grad_norm": 8.787314414978027,
      "learning_rate": 2.3606102937441327e-06,
      "loss": 0.0914,
      "step": 13173
    },
    {
      "epoch": 0.6802995094242189,
      "grad_norm": 2.2494423389434814,
      "learning_rate": 2.3599144295872236e-06,
      "loss": 0.0448,
      "step": 13174
    },
    {
      "epoch": 0.6803511489801187,
      "grad_norm": 3.880748748779297,
      "learning_rate": 2.3592186363273677e-06,
      "loss": 0.0631,
      "step": 13175
    },
    {
      "epoch": 0.6804027885360185,
      "grad_norm": 2.495084047317505,
      "learning_rate": 2.3585229139832565e-06,
      "loss": 0.0381,
      "step": 13176
    },
    {
      "epoch": 0.6804544280919184,
      "grad_norm": 13.281698226928711,
      "learning_rate": 2.3578272625735736e-06,
      "loss": 0.1008,
      "step": 13177
    },
    {
      "epoch": 0.6805060676478182,
      "grad_norm": 4.3344950675964355,
      "learning_rate": 2.3571316821169954e-06,
      "loss": 0.0451,
      "step": 13178
    },
    {
      "epoch": 0.680557707203718,
      "grad_norm": 4.070594787597656,
      "learning_rate": 2.3564361726322033e-06,
      "loss": 0.0511,
      "step": 13179
    },
    {
      "epoch": 0.6806093467596178,
      "grad_norm": 5.276401996612549,
      "learning_rate": 2.3557407341378723e-06,
      "loss": 0.068,
      "step": 13180
    },
    {
      "epoch": 0.6806609863155176,
      "grad_norm": 7.896429538726807,
      "learning_rate": 2.3550453666526846e-06,
      "loss": 0.1166,
      "step": 13181
    },
    {
      "epoch": 0.6807126258714175,
      "grad_norm": 7.458094596862793,
      "learning_rate": 2.3543500701953065e-06,
      "loss": 0.0274,
      "step": 13182
    },
    {
      "epoch": 0.6807642654273173,
      "grad_norm": 2.9073917865753174,
      "learning_rate": 2.353654844784412e-06,
      "loss": 0.038,
      "step": 13183
    },
    {
      "epoch": 0.6808159049832171,
      "grad_norm": 3.279090166091919,
      "learning_rate": 2.3529596904386714e-06,
      "loss": 0.0379,
      "step": 13184
    },
    {
      "epoch": 0.6808675445391169,
      "grad_norm": 8.843628883361816,
      "learning_rate": 2.352264607176751e-06,
      "loss": 0.1175,
      "step": 13185
    },
    {
      "epoch": 0.6809191840950167,
      "grad_norm": 5.669408798217773,
      "learning_rate": 2.3515695950173184e-06,
      "loss": 0.0564,
      "step": 13186
    },
    {
      "epoch": 0.6809708236509167,
      "grad_norm": 2.222714900970459,
      "learning_rate": 2.3508746539790363e-06,
      "loss": 0.0319,
      "step": 13187
    },
    {
      "epoch": 0.6810224632068165,
      "grad_norm": 5.615285396575928,
      "learning_rate": 2.3501797840805684e-06,
      "loss": 0.0598,
      "step": 13188
    },
    {
      "epoch": 0.6810741027627163,
      "grad_norm": 5.537501811981201,
      "learning_rate": 2.34948498534057e-06,
      "loss": 0.0294,
      "step": 13189
    },
    {
      "epoch": 0.6811257423186161,
      "grad_norm": 2.5568182468414307,
      "learning_rate": 2.3487902577777046e-06,
      "loss": 0.0743,
      "step": 13190
    },
    {
      "epoch": 0.6811773818745159,
      "grad_norm": 4.325739860534668,
      "learning_rate": 2.3480956014106264e-06,
      "loss": 0.0447,
      "step": 13191
    },
    {
      "epoch": 0.6812290214304157,
      "grad_norm": 5.33252477645874,
      "learning_rate": 2.347401016257992e-06,
      "loss": 0.0501,
      "step": 13192
    },
    {
      "epoch": 0.6812806609863156,
      "grad_norm": 3.1712374687194824,
      "learning_rate": 2.3467065023384494e-06,
      "loss": 0.0615,
      "step": 13193
    },
    {
      "epoch": 0.6813323005422154,
      "grad_norm": 1.2550088167190552,
      "learning_rate": 2.3460120596706496e-06,
      "loss": 0.0097,
      "step": 13194
    },
    {
      "epoch": 0.6813839400981152,
      "grad_norm": 2.5837888717651367,
      "learning_rate": 2.345317688273245e-06,
      "loss": 0.0438,
      "step": 13195
    },
    {
      "epoch": 0.681435579654015,
      "grad_norm": 1.8800837993621826,
      "learning_rate": 2.3446233881648823e-06,
      "loss": 0.0211,
      "step": 13196
    },
    {
      "epoch": 0.6814872192099148,
      "grad_norm": 5.515478610992432,
      "learning_rate": 2.3439291593642023e-06,
      "loss": 0.0959,
      "step": 13197
    },
    {
      "epoch": 0.6815388587658147,
      "grad_norm": 3.2714247703552246,
      "learning_rate": 2.34323500188985e-06,
      "loss": 0.0436,
      "step": 13198
    },
    {
      "epoch": 0.6815904983217145,
      "grad_norm": 2.86063289642334,
      "learning_rate": 2.342540915760467e-06,
      "loss": 0.0596,
      "step": 13199
    },
    {
      "epoch": 0.6816421378776143,
      "grad_norm": 2.6324265003204346,
      "learning_rate": 2.3418469009946903e-06,
      "loss": 0.0504,
      "step": 13200
    },
    {
      "epoch": 0.6816937774335141,
      "grad_norm": 5.459235191345215,
      "learning_rate": 2.3411529576111593e-06,
      "loss": 0.0446,
      "step": 13201
    },
    {
      "epoch": 0.6817454169894139,
      "grad_norm": 8.02857494354248,
      "learning_rate": 2.340459085628508e-06,
      "loss": 0.1054,
      "step": 13202
    },
    {
      "epoch": 0.6817970565453138,
      "grad_norm": 2.6741743087768555,
      "learning_rate": 2.3397652850653695e-06,
      "loss": 0.043,
      "step": 13203
    },
    {
      "epoch": 0.6818486961012136,
      "grad_norm": 4.502798080444336,
      "learning_rate": 2.3390715559403764e-06,
      "loss": 0.0999,
      "step": 13204
    },
    {
      "epoch": 0.6819003356571134,
      "grad_norm": 6.106907367706299,
      "learning_rate": 2.3383778982721568e-06,
      "loss": 0.0944,
      "step": 13205
    },
    {
      "epoch": 0.6819519752130132,
      "grad_norm": 4.257429599761963,
      "learning_rate": 2.337684312079339e-06,
      "loss": 0.0819,
      "step": 13206
    },
    {
      "epoch": 0.682003614768913,
      "grad_norm": 3.1654226779937744,
      "learning_rate": 2.33699079738055e-06,
      "loss": 0.0344,
      "step": 13207
    },
    {
      "epoch": 0.6820552543248128,
      "grad_norm": 2.6425108909606934,
      "learning_rate": 2.3362973541944084e-06,
      "loss": 0.0383,
      "step": 13208
    },
    {
      "epoch": 0.6821068938807127,
      "grad_norm": 16.229780197143555,
      "learning_rate": 2.3356039825395417e-06,
      "loss": 0.0737,
      "step": 13209
    },
    {
      "epoch": 0.6821585334366125,
      "grad_norm": 4.473243713378906,
      "learning_rate": 2.3349106824345693e-06,
      "loss": 0.0605,
      "step": 13210
    },
    {
      "epoch": 0.6822101729925123,
      "grad_norm": 3.026059150695801,
      "learning_rate": 2.334217453898105e-06,
      "loss": 0.0512,
      "step": 13211
    },
    {
      "epoch": 0.6822618125484121,
      "grad_norm": 8.508502006530762,
      "learning_rate": 2.3335242969487678e-06,
      "loss": 0.1106,
      "step": 13212
    },
    {
      "epoch": 0.6823134521043119,
      "grad_norm": 2.754365921020508,
      "learning_rate": 2.3328312116051688e-06,
      "loss": 0.0444,
      "step": 13213
    },
    {
      "epoch": 0.6823650916602118,
      "grad_norm": 6.878031253814697,
      "learning_rate": 2.3321381978859266e-06,
      "loss": 0.0995,
      "step": 13214
    },
    {
      "epoch": 0.6824167312161116,
      "grad_norm": 4.17535400390625,
      "learning_rate": 2.331445255809645e-06,
      "loss": 0.0329,
      "step": 13215
    },
    {
      "epoch": 0.6824683707720114,
      "grad_norm": 2.9709503650665283,
      "learning_rate": 2.3307523853949356e-06,
      "loss": 0.0089,
      "step": 13216
    },
    {
      "epoch": 0.6825200103279112,
      "grad_norm": 3.3249213695526123,
      "learning_rate": 2.3300595866604033e-06,
      "loss": 0.0714,
      "step": 13217
    },
    {
      "epoch": 0.682571649883811,
      "grad_norm": 3.435560464859009,
      "learning_rate": 2.329366859624653e-06,
      "loss": 0.0948,
      "step": 13218
    },
    {
      "epoch": 0.6826232894397108,
      "grad_norm": 2.7903003692626953,
      "learning_rate": 2.3286742043062882e-06,
      "loss": 0.0496,
      "step": 13219
    },
    {
      "epoch": 0.6826749289956107,
      "grad_norm": 3.9629666805267334,
      "learning_rate": 2.327981620723908e-06,
      "loss": 0.0396,
      "step": 13220
    },
    {
      "epoch": 0.6827265685515105,
      "grad_norm": 3.4515151977539062,
      "learning_rate": 2.3272891088961143e-06,
      "loss": 0.0384,
      "step": 13221
    },
    {
      "epoch": 0.6827782081074103,
      "grad_norm": 2.648343801498413,
      "learning_rate": 2.3265966688414972e-06,
      "loss": 0.0383,
      "step": 13222
    },
    {
      "epoch": 0.6828298476633101,
      "grad_norm": 3.2964019775390625,
      "learning_rate": 2.3259043005786573e-06,
      "loss": 0.0478,
      "step": 13223
    },
    {
      "epoch": 0.6828814872192099,
      "grad_norm": 7.532867431640625,
      "learning_rate": 2.3252120041261864e-06,
      "loss": 0.0425,
      "step": 13224
    },
    {
      "epoch": 0.6829331267751098,
      "grad_norm": 8.954176902770996,
      "learning_rate": 2.324519779502677e-06,
      "loss": 0.0842,
      "step": 13225
    },
    {
      "epoch": 0.6829847663310096,
      "grad_norm": 4.7594685554504395,
      "learning_rate": 2.3238276267267146e-06,
      "loss": 0.0484,
      "step": 13226
    },
    {
      "epoch": 0.6830364058869094,
      "grad_norm": 15.389161109924316,
      "learning_rate": 2.3231355458168853e-06,
      "loss": 0.074,
      "step": 13227
    },
    {
      "epoch": 0.6830880454428092,
      "grad_norm": 4.5147881507873535,
      "learning_rate": 2.3224435367917792e-06,
      "loss": 0.0484,
      "step": 13228
    },
    {
      "epoch": 0.683139684998709,
      "grad_norm": 2.2272143363952637,
      "learning_rate": 2.3217515996699787e-06,
      "loss": 0.0393,
      "step": 13229
    },
    {
      "epoch": 0.6831913245546088,
      "grad_norm": 2.7227518558502197,
      "learning_rate": 2.321059734470062e-06,
      "loss": 0.0275,
      "step": 13230
    },
    {
      "epoch": 0.6832429641105087,
      "grad_norm": 1.8548296689987183,
      "learning_rate": 2.3203679412106104e-06,
      "loss": 0.0394,
      "step": 13231
    },
    {
      "epoch": 0.6832946036664085,
      "grad_norm": 5.841651439666748,
      "learning_rate": 2.3196762199102015e-06,
      "loss": 0.0491,
      "step": 13232
    },
    {
      "epoch": 0.6833462432223083,
      "grad_norm": 5.356567859649658,
      "learning_rate": 2.3189845705874096e-06,
      "loss": 0.0898,
      "step": 13233
    },
    {
      "epoch": 0.6833978827782081,
      "grad_norm": 2.346104145050049,
      "learning_rate": 2.3182929932608093e-06,
      "loss": 0.0348,
      "step": 13234
    },
    {
      "epoch": 0.6834495223341079,
      "grad_norm": 2.85080623626709,
      "learning_rate": 2.3176014879489726e-06,
      "loss": 0.0589,
      "step": 13235
    },
    {
      "epoch": 0.6835011618900078,
      "grad_norm": 2.25333309173584,
      "learning_rate": 2.3169100546704686e-06,
      "loss": 0.0225,
      "step": 13236
    },
    {
      "epoch": 0.6835528014459076,
      "grad_norm": 4.195980072021484,
      "learning_rate": 2.316218693443866e-06,
      "loss": 0.049,
      "step": 13237
    },
    {
      "epoch": 0.6836044410018074,
      "grad_norm": 3.4068844318389893,
      "learning_rate": 2.315527404287729e-06,
      "loss": 0.0263,
      "step": 13238
    },
    {
      "epoch": 0.6836560805577072,
      "grad_norm": 4.7078776359558105,
      "learning_rate": 2.314836187220625e-06,
      "loss": 0.0234,
      "step": 13239
    },
    {
      "epoch": 0.683707720113607,
      "grad_norm": 2.6708407402038574,
      "learning_rate": 2.314145042261111e-06,
      "loss": 0.0396,
      "step": 13240
    },
    {
      "epoch": 0.6837593596695068,
      "grad_norm": 1.7792751789093018,
      "learning_rate": 2.313453969427748e-06,
      "loss": 0.0221,
      "step": 13241
    },
    {
      "epoch": 0.6838109992254067,
      "grad_norm": 6.844933986663818,
      "learning_rate": 2.3127629687390973e-06,
      "loss": 0.0981,
      "step": 13242
    },
    {
      "epoch": 0.6838626387813065,
      "grad_norm": 4.11875581741333,
      "learning_rate": 2.312072040213716e-06,
      "loss": 0.0854,
      "step": 13243
    },
    {
      "epoch": 0.6839142783372063,
      "grad_norm": 2.227949619293213,
      "learning_rate": 2.311381183870153e-06,
      "loss": 0.0359,
      "step": 13244
    },
    {
      "epoch": 0.6839659178931061,
      "grad_norm": 4.234358310699463,
      "learning_rate": 2.310690399726964e-06,
      "loss": 0.0725,
      "step": 13245
    },
    {
      "epoch": 0.6840175574490059,
      "grad_norm": 1.2064430713653564,
      "learning_rate": 2.309999687802697e-06,
      "loss": 0.0127,
      "step": 13246
    },
    {
      "epoch": 0.6840691970049058,
      "grad_norm": 3.718940258026123,
      "learning_rate": 2.3093090481159057e-06,
      "loss": 0.027,
      "step": 13247
    },
    {
      "epoch": 0.6841208365608056,
      "grad_norm": 3.6503450870513916,
      "learning_rate": 2.308618480685132e-06,
      "loss": 0.0441,
      "step": 13248
    },
    {
      "epoch": 0.6841724761167054,
      "grad_norm": 3.389364004135132,
      "learning_rate": 2.3079279855289206e-06,
      "loss": 0.0496,
      "step": 13249
    },
    {
      "epoch": 0.6842241156726052,
      "grad_norm": 2.5871119499206543,
      "learning_rate": 2.3072375626658155e-06,
      "loss": 0.0406,
      "step": 13250
    },
    {
      "epoch": 0.684275755228505,
      "grad_norm": 6.066843509674072,
      "learning_rate": 2.3065472121143567e-06,
      "loss": 0.0697,
      "step": 13251
    },
    {
      "epoch": 0.6843273947844049,
      "grad_norm": 2.8936374187469482,
      "learning_rate": 2.3058569338930838e-06,
      "loss": 0.0425,
      "step": 13252
    },
    {
      "epoch": 0.6843790343403047,
      "grad_norm": 24.233604431152344,
      "learning_rate": 2.305166728020532e-06,
      "loss": 0.1427,
      "step": 13253
    },
    {
      "epoch": 0.6844306738962045,
      "grad_norm": 7.528492450714111,
      "learning_rate": 2.3044765945152393e-06,
      "loss": 0.0894,
      "step": 13254
    },
    {
      "epoch": 0.6844823134521043,
      "grad_norm": 3.9527664184570312,
      "learning_rate": 2.3037865333957326e-06,
      "loss": 0.0579,
      "step": 13255
    },
    {
      "epoch": 0.6845339530080041,
      "grad_norm": 8.104798316955566,
      "learning_rate": 2.303096544680549e-06,
      "loss": 0.0533,
      "step": 13256
    },
    {
      "epoch": 0.6845855925639039,
      "grad_norm": 3.733168601989746,
      "learning_rate": 2.3024066283882148e-06,
      "loss": 0.0466,
      "step": 13257
    },
    {
      "epoch": 0.6846372321198038,
      "grad_norm": 2.8667757511138916,
      "learning_rate": 2.301716784537259e-06,
      "loss": 0.0674,
      "step": 13258
    },
    {
      "epoch": 0.6846888716757036,
      "grad_norm": 2.268709659576416,
      "learning_rate": 2.3010270131462035e-06,
      "loss": 0.0401,
      "step": 13259
    },
    {
      "epoch": 0.6847405112316034,
      "grad_norm": 14.673824310302734,
      "learning_rate": 2.3003373142335717e-06,
      "loss": 0.0713,
      "step": 13260
    },
    {
      "epoch": 0.6847921507875032,
      "grad_norm": 18.437759399414062,
      "learning_rate": 2.2996476878178894e-06,
      "loss": 0.0679,
      "step": 13261
    },
    {
      "epoch": 0.684843790343403,
      "grad_norm": 11.4677095413208,
      "learning_rate": 2.298958133917671e-06,
      "loss": 0.0979,
      "step": 13262
    },
    {
      "epoch": 0.6848954298993029,
      "grad_norm": 4.855597496032715,
      "learning_rate": 2.2982686525514353e-06,
      "loss": 0.0336,
      "step": 13263
    },
    {
      "epoch": 0.6849470694552027,
      "grad_norm": 4.409307956695557,
      "learning_rate": 2.297579243737698e-06,
      "loss": 0.0415,
      "step": 13264
    },
    {
      "epoch": 0.6849987090111025,
      "grad_norm": 15.23831558227539,
      "learning_rate": 2.2968899074949725e-06,
      "loss": 0.1349,
      "step": 13265
    },
    {
      "epoch": 0.6850503485670023,
      "grad_norm": 3.55464768409729,
      "learning_rate": 2.2962006438417704e-06,
      "loss": 0.0421,
      "step": 13266
    },
    {
      "epoch": 0.6851019881229021,
      "grad_norm": 3.5942063331604004,
      "learning_rate": 2.295511452796601e-06,
      "loss": 0.0547,
      "step": 13267
    },
    {
      "epoch": 0.685153627678802,
      "grad_norm": 6.119871139526367,
      "learning_rate": 2.2948223343779723e-06,
      "loss": 0.0973,
      "step": 13268
    },
    {
      "epoch": 0.6852052672347018,
      "grad_norm": 3.113281488418579,
      "learning_rate": 2.294133288604389e-06,
      "loss": 0.0431,
      "step": 13269
    },
    {
      "epoch": 0.6852569067906016,
      "grad_norm": 18.76163101196289,
      "learning_rate": 2.293444315494356e-06,
      "loss": 0.1463,
      "step": 13270
    },
    {
      "epoch": 0.6853085463465014,
      "grad_norm": 5.173757553100586,
      "learning_rate": 2.292755415066374e-06,
      "loss": 0.0434,
      "step": 13271
    },
    {
      "epoch": 0.6853601859024012,
      "grad_norm": 5.764746189117432,
      "learning_rate": 2.2920665873389454e-06,
      "loss": 0.028,
      "step": 13272
    },
    {
      "epoch": 0.685411825458301,
      "grad_norm": 3.3376433849334717,
      "learning_rate": 2.2913778323305633e-06,
      "loss": 0.0478,
      "step": 13273
    },
    {
      "epoch": 0.6854634650142009,
      "grad_norm": 4.276121139526367,
      "learning_rate": 2.2906891500597257e-06,
      "loss": 0.0986,
      "step": 13274
    },
    {
      "epoch": 0.6855151045701007,
      "grad_norm": 3.9466264247894287,
      "learning_rate": 2.2900005405449276e-06,
      "loss": 0.0431,
      "step": 13275
    },
    {
      "epoch": 0.6855667441260005,
      "grad_norm": 2.6246824264526367,
      "learning_rate": 2.2893120038046625e-06,
      "loss": 0.0271,
      "step": 13276
    },
    {
      "epoch": 0.6856183836819003,
      "grad_norm": 2.7964577674865723,
      "learning_rate": 2.2886235398574162e-06,
      "loss": 0.0356,
      "step": 13277
    },
    {
      "epoch": 0.6856700232378001,
      "grad_norm": 3.1799566745758057,
      "learning_rate": 2.2879351487216794e-06,
      "loss": 0.0428,
      "step": 13278
    },
    {
      "epoch": 0.6857216627937,
      "grad_norm": 11.053889274597168,
      "learning_rate": 2.2872468304159352e-06,
      "loss": 0.0731,
      "step": 13279
    },
    {
      "epoch": 0.6857733023495998,
      "grad_norm": 20.186809539794922,
      "learning_rate": 2.286558584958674e-06,
      "loss": 0.1068,
      "step": 13280
    },
    {
      "epoch": 0.6858249419054996,
      "grad_norm": 8.344745635986328,
      "learning_rate": 2.2858704123683728e-06,
      "loss": 0.122,
      "step": 13281
    },
    {
      "epoch": 0.6858765814613994,
      "grad_norm": 3.004446506500244,
      "learning_rate": 2.2851823126635126e-06,
      "loss": 0.049,
      "step": 13282
    },
    {
      "epoch": 0.6859282210172992,
      "grad_norm": 2.1565985679626465,
      "learning_rate": 2.2844942858625724e-06,
      "loss": 0.0384,
      "step": 13283
    },
    {
      "epoch": 0.685979860573199,
      "grad_norm": 5.4731011390686035,
      "learning_rate": 2.2838063319840285e-06,
      "loss": 0.0768,
      "step": 13284
    },
    {
      "epoch": 0.6860315001290989,
      "grad_norm": 1.5901538133621216,
      "learning_rate": 2.2831184510463555e-06,
      "loss": 0.0415,
      "step": 13285
    },
    {
      "epoch": 0.6860831396849987,
      "grad_norm": 2.94661283493042,
      "learning_rate": 2.282430643068025e-06,
      "loss": 0.0697,
      "step": 13286
    },
    {
      "epoch": 0.6861347792408985,
      "grad_norm": 2.2876267433166504,
      "learning_rate": 2.28174290806751e-06,
      "loss": 0.025,
      "step": 13287
    },
    {
      "epoch": 0.6861864187967983,
      "grad_norm": 10.32313346862793,
      "learning_rate": 2.2810552460632735e-06,
      "loss": 0.1384,
      "step": 13288
    },
    {
      "epoch": 0.6862380583526981,
      "grad_norm": 3.3673248291015625,
      "learning_rate": 2.2803676570737864e-06,
      "loss": 0.0272,
      "step": 13289
    },
    {
      "epoch": 0.686289697908598,
      "grad_norm": 1.7769649028778076,
      "learning_rate": 2.279680141117513e-06,
      "loss": 0.0274,
      "step": 13290
    },
    {
      "epoch": 0.6863413374644978,
      "grad_norm": 2.7709739208221436,
      "learning_rate": 2.2789926982129175e-06,
      "loss": 0.0251,
      "step": 13291
    },
    {
      "epoch": 0.6863929770203976,
      "grad_norm": 2.0258471965789795,
      "learning_rate": 2.2783053283784557e-06,
      "loss": 0.0337,
      "step": 13292
    },
    {
      "epoch": 0.6864446165762974,
      "grad_norm": 2.4454426765441895,
      "learning_rate": 2.277618031632587e-06,
      "loss": 0.0304,
      "step": 13293
    },
    {
      "epoch": 0.6864962561321972,
      "grad_norm": 3.2475950717926025,
      "learning_rate": 2.276930807993774e-06,
      "loss": 0.0673,
      "step": 13294
    },
    {
      "epoch": 0.686547895688097,
      "grad_norm": 8.150235176086426,
      "learning_rate": 2.2762436574804652e-06,
      "loss": 0.0627,
      "step": 13295
    },
    {
      "epoch": 0.6865995352439969,
      "grad_norm": 1.8489775657653809,
      "learning_rate": 2.275556580111116e-06,
      "loss": 0.0275,
      "step": 13296
    },
    {
      "epoch": 0.6866511747998967,
      "grad_norm": 6.025032997131348,
      "learning_rate": 2.2748695759041766e-06,
      "loss": 0.038,
      "step": 13297
    },
    {
      "epoch": 0.6867028143557965,
      "grad_norm": 6.2120466232299805,
      "learning_rate": 2.274182644878096e-06,
      "loss": 0.0656,
      "step": 13298
    },
    {
      "epoch": 0.6867544539116963,
      "grad_norm": 3.1495742797851562,
      "learning_rate": 2.2734957870513213e-06,
      "loss": 0.0355,
      "step": 13299
    },
    {
      "epoch": 0.6868060934675961,
      "grad_norm": 5.285636901855469,
      "learning_rate": 2.2728090024422976e-06,
      "loss": 0.0526,
      "step": 13300
    },
    {
      "epoch": 0.686857733023496,
      "grad_norm": 4.140035152435303,
      "learning_rate": 2.272122291069467e-06,
      "loss": 0.051,
      "step": 13301
    },
    {
      "epoch": 0.6869093725793958,
      "grad_norm": 5.074998378753662,
      "learning_rate": 2.271435652951272e-06,
      "loss": 0.0427,
      "step": 13302
    },
    {
      "epoch": 0.6869610121352956,
      "grad_norm": 3.5527877807617188,
      "learning_rate": 2.27074908810615e-06,
      "loss": 0.0505,
      "step": 13303
    },
    {
      "epoch": 0.6870126516911954,
      "grad_norm": 9.040656089782715,
      "learning_rate": 2.270062596552539e-06,
      "loss": 0.0455,
      "step": 13304
    },
    {
      "epoch": 0.6870642912470952,
      "grad_norm": 3.795771360397339,
      "learning_rate": 2.2693761783088765e-06,
      "loss": 0.0548,
      "step": 13305
    },
    {
      "epoch": 0.687115930802995,
      "grad_norm": 2.7585158348083496,
      "learning_rate": 2.2686898333935902e-06,
      "loss": 0.0212,
      "step": 13306
    },
    {
      "epoch": 0.6871675703588949,
      "grad_norm": 3.0483150482177734,
      "learning_rate": 2.2680035618251124e-06,
      "loss": 0.0293,
      "step": 13307
    },
    {
      "epoch": 0.6872192099147947,
      "grad_norm": 2.0719966888427734,
      "learning_rate": 2.2673173636218758e-06,
      "loss": 0.0367,
      "step": 13308
    },
    {
      "epoch": 0.6872708494706945,
      "grad_norm": 10.183104515075684,
      "learning_rate": 2.2666312388023075e-06,
      "loss": 0.0839,
      "step": 13309
    },
    {
      "epoch": 0.6873224890265943,
      "grad_norm": 5.621847152709961,
      "learning_rate": 2.26594518738483e-06,
      "loss": 0.0364,
      "step": 13310
    },
    {
      "epoch": 0.6873741285824941,
      "grad_norm": 4.636148929595947,
      "learning_rate": 2.265259209387867e-06,
      "loss": 0.0437,
      "step": 13311
    },
    {
      "epoch": 0.687425768138394,
      "grad_norm": 2.76182222366333,
      "learning_rate": 2.2645733048298386e-06,
      "loss": 0.0193,
      "step": 13312
    },
    {
      "epoch": 0.6874774076942938,
      "grad_norm": 6.817203044891357,
      "learning_rate": 2.2638874737291694e-06,
      "loss": 0.0941,
      "step": 13313
    },
    {
      "epoch": 0.6875290472501937,
      "grad_norm": 2.9043080806732178,
      "learning_rate": 2.263201716104272e-06,
      "loss": 0.0654,
      "step": 13314
    },
    {
      "epoch": 0.6875806868060935,
      "grad_norm": 2.1077585220336914,
      "learning_rate": 2.2625160319735625e-06,
      "loss": 0.0467,
      "step": 13315
    },
    {
      "epoch": 0.6876323263619933,
      "grad_norm": 3.23294734954834,
      "learning_rate": 2.2618304213554555e-06,
      "loss": 0.0344,
      "step": 13316
    },
    {
      "epoch": 0.6876839659178932,
      "grad_norm": 8.695328712463379,
      "learning_rate": 2.2611448842683612e-06,
      "loss": 0.1104,
      "step": 13317
    },
    {
      "epoch": 0.687735605473793,
      "grad_norm": 4.164798736572266,
      "learning_rate": 2.26045942073069e-06,
      "loss": 0.0276,
      "step": 13318
    },
    {
      "epoch": 0.6877872450296928,
      "grad_norm": 3.774817943572998,
      "learning_rate": 2.259774030760849e-06,
      "loss": 0.058,
      "step": 13319
    },
    {
      "epoch": 0.6878388845855926,
      "grad_norm": 4.158964157104492,
      "learning_rate": 2.259088714377246e-06,
      "loss": 0.1096,
      "step": 13320
    },
    {
      "epoch": 0.6878905241414924,
      "grad_norm": 1.9977833032608032,
      "learning_rate": 2.2584034715982783e-06,
      "loss": 0.0322,
      "step": 13321
    },
    {
      "epoch": 0.6879421636973923,
      "grad_norm": 12.779746055603027,
      "learning_rate": 2.2577183024423537e-06,
      "loss": 0.0532,
      "step": 13322
    },
    {
      "epoch": 0.6879938032532921,
      "grad_norm": 4.162701606750488,
      "learning_rate": 2.2570332069278692e-06,
      "loss": 0.0663,
      "step": 13323
    },
    {
      "epoch": 0.6880454428091919,
      "grad_norm": 9.508678436279297,
      "learning_rate": 2.2563481850732257e-06,
      "loss": 0.0434,
      "step": 13324
    },
    {
      "epoch": 0.6880970823650917,
      "grad_norm": 3.1539885997772217,
      "learning_rate": 2.255663236896813e-06,
      "loss": 0.0549,
      "step": 13325
    },
    {
      "epoch": 0.6881487219209915,
      "grad_norm": 6.561854839324951,
      "learning_rate": 2.254978362417026e-06,
      "loss": 0.0493,
      "step": 13326
    },
    {
      "epoch": 0.6882003614768913,
      "grad_norm": 3.2889082431793213,
      "learning_rate": 2.254293561652262e-06,
      "loss": 0.0435,
      "step": 13327
    },
    {
      "epoch": 0.6882520010327912,
      "grad_norm": 9.121264457702637,
      "learning_rate": 2.2536088346209043e-06,
      "loss": 0.1517,
      "step": 13328
    },
    {
      "epoch": 0.688303640588691,
      "grad_norm": 4.327681064605713,
      "learning_rate": 2.2529241813413437e-06,
      "loss": 0.0879,
      "step": 13329
    },
    {
      "epoch": 0.6883552801445908,
      "grad_norm": 2.876878023147583,
      "learning_rate": 2.2522396018319653e-06,
      "loss": 0.0457,
      "step": 13330
    },
    {
      "epoch": 0.6884069197004906,
      "grad_norm": 1.5210957527160645,
      "learning_rate": 2.2515550961111527e-06,
      "loss": 0.0284,
      "step": 13331
    },
    {
      "epoch": 0.6884585592563904,
      "grad_norm": 12.730542182922363,
      "learning_rate": 2.2508706641972877e-06,
      "loss": 0.1149,
      "step": 13332
    },
    {
      "epoch": 0.6885101988122903,
      "grad_norm": 2.789372444152832,
      "learning_rate": 2.2501863061087503e-06,
      "loss": 0.0506,
      "step": 13333
    },
    {
      "epoch": 0.6885618383681901,
      "grad_norm": 6.702371597290039,
      "learning_rate": 2.2495020218639184e-06,
      "loss": 0.115,
      "step": 13334
    },
    {
      "epoch": 0.6886134779240899,
      "grad_norm": 8.715108871459961,
      "learning_rate": 2.2488178114811675e-06,
      "loss": 0.0987,
      "step": 13335
    },
    {
      "epoch": 0.6886651174799897,
      "grad_norm": 3.160663604736328,
      "learning_rate": 2.248133674978871e-06,
      "loss": 0.0618,
      "step": 13336
    },
    {
      "epoch": 0.6887167570358895,
      "grad_norm": 10.077156066894531,
      "learning_rate": 2.2474496123754023e-06,
      "loss": 0.0649,
      "step": 13337
    },
    {
      "epoch": 0.6887683965917893,
      "grad_norm": 14.386580467224121,
      "learning_rate": 2.2467656236891318e-06,
      "loss": 0.0805,
      "step": 13338
    },
    {
      "epoch": 0.6888200361476892,
      "grad_norm": 1.7272720336914062,
      "learning_rate": 2.2460817089384237e-06,
      "loss": 0.0115,
      "step": 13339
    },
    {
      "epoch": 0.688871675703589,
      "grad_norm": 1.4940478801727295,
      "learning_rate": 2.2453978681416443e-06,
      "loss": 0.0097,
      "step": 13340
    },
    {
      "epoch": 0.6889233152594888,
      "grad_norm": 10.886966705322266,
      "learning_rate": 2.2447141013171614e-06,
      "loss": 0.1591,
      "step": 13341
    },
    {
      "epoch": 0.6889749548153886,
      "grad_norm": 3.4753756523132324,
      "learning_rate": 2.2440304084833363e-06,
      "loss": 0.0353,
      "step": 13342
    },
    {
      "epoch": 0.6890265943712884,
      "grad_norm": 4.0732526779174805,
      "learning_rate": 2.2433467896585257e-06,
      "loss": 0.0686,
      "step": 13343
    },
    {
      "epoch": 0.6890782339271883,
      "grad_norm": 3.563063621520996,
      "learning_rate": 2.2426632448610893e-06,
      "loss": 0.0474,
      "step": 13344
    },
    {
      "epoch": 0.6891298734830881,
      "grad_norm": 1.3504654169082642,
      "learning_rate": 2.241979774109381e-06,
      "loss": 0.0359,
      "step": 13345
    },
    {
      "epoch": 0.6891815130389879,
      "grad_norm": 2.023728132247925,
      "learning_rate": 2.241296377421761e-06,
      "loss": 0.0265,
      "step": 13346
    },
    {
      "epoch": 0.6892331525948877,
      "grad_norm": 3.1013593673706055,
      "learning_rate": 2.240613054816575e-06,
      "loss": 0.0328,
      "step": 13347
    },
    {
      "epoch": 0.6892847921507875,
      "grad_norm": 9.401655197143555,
      "learning_rate": 2.239929806312175e-06,
      "loss": 0.0715,
      "step": 13348
    },
    {
      "epoch": 0.6893364317066873,
      "grad_norm": 3.7186148166656494,
      "learning_rate": 2.23924663192691e-06,
      "loss": 0.064,
      "step": 13349
    },
    {
      "epoch": 0.6893880712625872,
      "grad_norm": 11.02357006072998,
      "learning_rate": 2.238563531679124e-06,
      "loss": 0.1454,
      "step": 13350
    },
    {
      "epoch": 0.689439710818487,
      "grad_norm": 1.8263236284255981,
      "learning_rate": 2.237880505587163e-06,
      "loss": 0.0139,
      "step": 13351
    },
    {
      "epoch": 0.6894913503743868,
      "grad_norm": 2.148937702178955,
      "learning_rate": 2.2371975536693672e-06,
      "loss": 0.0441,
      "step": 13352
    },
    {
      "epoch": 0.6895429899302866,
      "grad_norm": 9.472305297851562,
      "learning_rate": 2.23651467594408e-06,
      "loss": 0.0977,
      "step": 13353
    },
    {
      "epoch": 0.6895946294861864,
      "grad_norm": 2.7792420387268066,
      "learning_rate": 2.2358318724296336e-06,
      "loss": 0.0482,
      "step": 13354
    },
    {
      "epoch": 0.6896462690420863,
      "grad_norm": 5.953433036804199,
      "learning_rate": 2.2351491431443696e-06,
      "loss": 0.0152,
      "step": 13355
    },
    {
      "epoch": 0.6896979085979861,
      "grad_norm": 1.5485326051712036,
      "learning_rate": 2.2344664881066198e-06,
      "loss": 0.0424,
      "step": 13356
    },
    {
      "epoch": 0.6897495481538859,
      "grad_norm": 4.166126728057861,
      "learning_rate": 2.2337839073347176e-06,
      "loss": 0.0524,
      "step": 13357
    },
    {
      "epoch": 0.6898011877097857,
      "grad_norm": 4.769894123077393,
      "learning_rate": 2.233101400846991e-06,
      "loss": 0.0507,
      "step": 13358
    },
    {
      "epoch": 0.6898528272656855,
      "grad_norm": 2.8615834712982178,
      "learning_rate": 2.232418968661767e-06,
      "loss": 0.0167,
      "step": 13359
    },
    {
      "epoch": 0.6899044668215853,
      "grad_norm": 6.59104061126709,
      "learning_rate": 2.2317366107973775e-06,
      "loss": 0.0487,
      "step": 13360
    },
    {
      "epoch": 0.6899561063774852,
      "grad_norm": 1.0863187313079834,
      "learning_rate": 2.231054327272141e-06,
      "loss": 0.0136,
      "step": 13361
    },
    {
      "epoch": 0.690007745933385,
      "grad_norm": 23.687772750854492,
      "learning_rate": 2.230372118104381e-06,
      "loss": 0.0255,
      "step": 13362
    },
    {
      "epoch": 0.6900593854892848,
      "grad_norm": 20.16495704650879,
      "learning_rate": 2.229689983312418e-06,
      "loss": 0.0962,
      "step": 13363
    },
    {
      "epoch": 0.6901110250451846,
      "grad_norm": 2.7199418544769287,
      "learning_rate": 2.2290079229145706e-06,
      "loss": 0.0306,
      "step": 13364
    },
    {
      "epoch": 0.6901626646010844,
      "grad_norm": 4.688424587249756,
      "learning_rate": 2.2283259369291545e-06,
      "loss": 0.0419,
      "step": 13365
    },
    {
      "epoch": 0.6902143041569843,
      "grad_norm": 3.5861032009124756,
      "learning_rate": 2.227644025374483e-06,
      "loss": 0.057,
      "step": 13366
    },
    {
      "epoch": 0.6902659437128841,
      "grad_norm": 2.480902910232544,
      "learning_rate": 2.226962188268869e-06,
      "loss": 0.0429,
      "step": 13367
    },
    {
      "epoch": 0.6903175832687839,
      "grad_norm": 10.422347068786621,
      "learning_rate": 2.226280425630623e-06,
      "loss": 0.0569,
      "step": 13368
    },
    {
      "epoch": 0.6903692228246837,
      "grad_norm": 1.5668933391571045,
      "learning_rate": 2.2255987374780526e-06,
      "loss": 0.0169,
      "step": 13369
    },
    {
      "epoch": 0.6904208623805835,
      "grad_norm": 2.1634275913238525,
      "learning_rate": 2.224917123829464e-06,
      "loss": 0.023,
      "step": 13370
    },
    {
      "epoch": 0.6904725019364834,
      "grad_norm": 3.0259158611297607,
      "learning_rate": 2.224235584703163e-06,
      "loss": 0.0348,
      "step": 13371
    },
    {
      "epoch": 0.6905241414923832,
      "grad_norm": 5.68220853805542,
      "learning_rate": 2.2235541201174477e-06,
      "loss": 0.0776,
      "step": 13372
    },
    {
      "epoch": 0.690575781048283,
      "grad_norm": 6.637631416320801,
      "learning_rate": 2.222872730090619e-06,
      "loss": 0.0569,
      "step": 13373
    },
    {
      "epoch": 0.6906274206041828,
      "grad_norm": 2.9405441284179688,
      "learning_rate": 2.222191414640978e-06,
      "loss": 0.0503,
      "step": 13374
    },
    {
      "epoch": 0.6906790601600826,
      "grad_norm": 13.3320894241333,
      "learning_rate": 2.221510173786821e-06,
      "loss": 0.0402,
      "step": 13375
    },
    {
      "epoch": 0.6907306997159824,
      "grad_norm": 6.235198974609375,
      "learning_rate": 2.220829007546439e-06,
      "loss": 0.0894,
      "step": 13376
    },
    {
      "epoch": 0.6907823392718823,
      "grad_norm": 2.3970799446105957,
      "learning_rate": 2.220147915938125e-06,
      "loss": 0.0227,
      "step": 13377
    },
    {
      "epoch": 0.6908339788277821,
      "grad_norm": 3.9366536140441895,
      "learning_rate": 2.2194668989801676e-06,
      "loss": 0.0468,
      "step": 13378
    },
    {
      "epoch": 0.6908856183836819,
      "grad_norm": 2.274873971939087,
      "learning_rate": 2.2187859566908605e-06,
      "loss": 0.0258,
      "step": 13379
    },
    {
      "epoch": 0.6909372579395817,
      "grad_norm": 3.8625969886779785,
      "learning_rate": 2.2181050890884846e-06,
      "loss": 0.0522,
      "step": 13380
    },
    {
      "epoch": 0.6909888974954815,
      "grad_norm": 3.9387366771698,
      "learning_rate": 2.217424296191325e-06,
      "loss": 0.0987,
      "step": 13381
    },
    {
      "epoch": 0.6910405370513814,
      "grad_norm": 6.422934532165527,
      "learning_rate": 2.2167435780176642e-06,
      "loss": 0.0669,
      "step": 13382
    },
    {
      "epoch": 0.6910921766072812,
      "grad_norm": 5.8542304039001465,
      "learning_rate": 2.2160629345857824e-06,
      "loss": 0.0794,
      "step": 13383
    },
    {
      "epoch": 0.691143816163181,
      "grad_norm": 1.3023817539215088,
      "learning_rate": 2.2153823659139577e-06,
      "loss": 0.0213,
      "step": 13384
    },
    {
      "epoch": 0.6911954557190808,
      "grad_norm": 7.88633394241333,
      "learning_rate": 2.214701872020466e-06,
      "loss": 0.0647,
      "step": 13385
    },
    {
      "epoch": 0.6912470952749806,
      "grad_norm": 3.2536866664886475,
      "learning_rate": 2.2140214529235833e-06,
      "loss": 0.0392,
      "step": 13386
    },
    {
      "epoch": 0.6912987348308804,
      "grad_norm": 3.713080406188965,
      "learning_rate": 2.213341108641575e-06,
      "loss": 0.0524,
      "step": 13387
    },
    {
      "epoch": 0.6913503743867803,
      "grad_norm": 5.500785827636719,
      "learning_rate": 2.2126608391927186e-06,
      "loss": 0.0305,
      "step": 13388
    },
    {
      "epoch": 0.6914020139426801,
      "grad_norm": 6.072653770446777,
      "learning_rate": 2.2119806445952785e-06,
      "loss": 0.0788,
      "step": 13389
    },
    {
      "epoch": 0.6914536534985799,
      "grad_norm": 3.523679733276367,
      "learning_rate": 2.211300524867524e-06,
      "loss": 0.0404,
      "step": 13390
    },
    {
      "epoch": 0.6915052930544797,
      "grad_norm": 2.3930375576019287,
      "learning_rate": 2.2106204800277143e-06,
      "loss": 0.0285,
      "step": 13391
    },
    {
      "epoch": 0.6915569326103795,
      "grad_norm": 21.748220443725586,
      "learning_rate": 2.2099405100941113e-06,
      "loss": 0.0955,
      "step": 13392
    },
    {
      "epoch": 0.6916085721662794,
      "grad_norm": 9.06838607788086,
      "learning_rate": 2.2092606150849817e-06,
      "loss": 0.0828,
      "step": 13393
    },
    {
      "epoch": 0.6916602117221792,
      "grad_norm": 3.3585543632507324,
      "learning_rate": 2.2085807950185765e-06,
      "loss": 0.0593,
      "step": 13394
    },
    {
      "epoch": 0.691711851278079,
      "grad_norm": 2.4500489234924316,
      "learning_rate": 2.2079010499131544e-06,
      "loss": 0.0356,
      "step": 13395
    },
    {
      "epoch": 0.6917634908339788,
      "grad_norm": 9.1322660446167,
      "learning_rate": 2.207221379786969e-06,
      "loss": 0.0664,
      "step": 13396
    },
    {
      "epoch": 0.6918151303898786,
      "grad_norm": 3.5138816833496094,
      "learning_rate": 2.2065417846582724e-06,
      "loss": 0.0433,
      "step": 13397
    },
    {
      "epoch": 0.6918667699457784,
      "grad_norm": 1.8856873512268066,
      "learning_rate": 2.205862264545315e-06,
      "loss": 0.0232,
      "step": 13398
    },
    {
      "epoch": 0.6919184095016783,
      "grad_norm": 2.3759632110595703,
      "learning_rate": 2.2051828194663434e-06,
      "loss": 0.0347,
      "step": 13399
    },
    {
      "epoch": 0.6919700490575781,
      "grad_norm": 14.84041976928711,
      "learning_rate": 2.204503449439604e-06,
      "loss": 0.07,
      "step": 13400
    },
    {
      "epoch": 0.6920216886134779,
      "grad_norm": 8.724696159362793,
      "learning_rate": 2.2038241544833413e-06,
      "loss": 0.2339,
      "step": 13401
    },
    {
      "epoch": 0.6920733281693777,
      "grad_norm": 12.865487098693848,
      "learning_rate": 2.2031449346157958e-06,
      "loss": 0.1063,
      "step": 13402
    },
    {
      "epoch": 0.6921249677252775,
      "grad_norm": 6.352489948272705,
      "learning_rate": 2.202465789855209e-06,
      "loss": 0.0845,
      "step": 13403
    },
    {
      "epoch": 0.6921766072811774,
      "grad_norm": 1.8529664278030396,
      "learning_rate": 2.201786720219819e-06,
      "loss": 0.0294,
      "step": 13404
    },
    {
      "epoch": 0.6922282468370772,
      "grad_norm": 6.110254287719727,
      "learning_rate": 2.201107725727859e-06,
      "loss": 0.0616,
      "step": 13405
    },
    {
      "epoch": 0.692279886392977,
      "grad_norm": 2.140312433242798,
      "learning_rate": 2.2004288063975628e-06,
      "loss": 0.0343,
      "step": 13406
    },
    {
      "epoch": 0.6923315259488768,
      "grad_norm": 3.1853721141815186,
      "learning_rate": 2.1997499622471647e-06,
      "loss": 0.0381,
      "step": 13407
    },
    {
      "epoch": 0.6923831655047766,
      "grad_norm": 2.9998233318328857,
      "learning_rate": 2.1990711932948956e-06,
      "loss": 0.0401,
      "step": 13408
    },
    {
      "epoch": 0.6924348050606765,
      "grad_norm": 3.8581554889678955,
      "learning_rate": 2.198392499558979e-06,
      "loss": 0.0436,
      "step": 13409
    },
    {
      "epoch": 0.6924864446165763,
      "grad_norm": 9.562256813049316,
      "learning_rate": 2.197713881057642e-06,
      "loss": 0.0926,
      "step": 13410
    },
    {
      "epoch": 0.6925380841724761,
      "grad_norm": 8.755595207214355,
      "learning_rate": 2.1970353378091075e-06,
      "loss": 0.0356,
      "step": 13411
    },
    {
      "epoch": 0.6925897237283759,
      "grad_norm": 2.344639778137207,
      "learning_rate": 2.1963568698316017e-06,
      "loss": 0.0479,
      "step": 13412
    },
    {
      "epoch": 0.6926413632842757,
      "grad_norm": 3.535555124282837,
      "learning_rate": 2.1956784771433397e-06,
      "loss": 0.0722,
      "step": 13413
    },
    {
      "epoch": 0.6926930028401755,
      "grad_norm": 3.367074728012085,
      "learning_rate": 2.1950001597625397e-06,
      "loss": 0.0338,
      "step": 13414
    },
    {
      "epoch": 0.6927446423960754,
      "grad_norm": 2.424454689025879,
      "learning_rate": 2.1943219177074187e-06,
      "loss": 0.025,
      "step": 13415
    },
    {
      "epoch": 0.6927962819519752,
      "grad_norm": 4.654897689819336,
      "learning_rate": 2.193643750996189e-06,
      "loss": 0.0261,
      "step": 13416
    },
    {
      "epoch": 0.692847921507875,
      "grad_norm": 1.7408325672149658,
      "learning_rate": 2.192965659647063e-06,
      "loss": 0.0156,
      "step": 13417
    },
    {
      "epoch": 0.6928995610637748,
      "grad_norm": 3.9729957580566406,
      "learning_rate": 2.1922876436782497e-06,
      "loss": 0.0296,
      "step": 13418
    },
    {
      "epoch": 0.6929512006196746,
      "grad_norm": 1.6303510665893555,
      "learning_rate": 2.1916097031079565e-06,
      "loss": 0.0179,
      "step": 13419
    },
    {
      "epoch": 0.6930028401755745,
      "grad_norm": 2.300462007522583,
      "learning_rate": 2.190931837954389e-06,
      "loss": 0.0217,
      "step": 13420
    },
    {
      "epoch": 0.6930544797314743,
      "grad_norm": 7.266059398651123,
      "learning_rate": 2.1902540482357515e-06,
      "loss": 0.0613,
      "step": 13421
    },
    {
      "epoch": 0.6931061192873741,
      "grad_norm": 3.319443941116333,
      "learning_rate": 2.1895763339702447e-06,
      "loss": 0.0329,
      "step": 13422
    },
    {
      "epoch": 0.6931577588432739,
      "grad_norm": 2.4500434398651123,
      "learning_rate": 2.1888986951760693e-06,
      "loss": 0.0408,
      "step": 13423
    },
    {
      "epoch": 0.6932093983991737,
      "grad_norm": 3.702317237854004,
      "learning_rate": 2.1882211318714196e-06,
      "loss": 0.052,
      "step": 13424
    },
    {
      "epoch": 0.6932610379550735,
      "grad_norm": 10.806971549987793,
      "learning_rate": 2.1875436440744908e-06,
      "loss": 0.0906,
      "step": 13425
    },
    {
      "epoch": 0.6933126775109734,
      "grad_norm": 8.742377281188965,
      "learning_rate": 2.186866231803482e-06,
      "loss": 0.0168,
      "step": 13426
    },
    {
      "epoch": 0.6933643170668732,
      "grad_norm": 4.979975700378418,
      "learning_rate": 2.1861888950765785e-06,
      "loss": 0.0618,
      "step": 13427
    },
    {
      "epoch": 0.693415956622773,
      "grad_norm": 0.97891765832901,
      "learning_rate": 2.185511633911972e-06,
      "loss": 0.0094,
      "step": 13428
    },
    {
      "epoch": 0.6934675961786728,
      "grad_norm": 10.647327423095703,
      "learning_rate": 2.184834448327849e-06,
      "loss": 0.1448,
      "step": 13429
    },
    {
      "epoch": 0.6935192357345726,
      "grad_norm": 5.207602500915527,
      "learning_rate": 2.184157338342395e-06,
      "loss": 0.0543,
      "step": 13430
    },
    {
      "epoch": 0.6935708752904725,
      "grad_norm": 3.483778715133667,
      "learning_rate": 2.1834803039737933e-06,
      "loss": 0.0228,
      "step": 13431
    },
    {
      "epoch": 0.6936225148463723,
      "grad_norm": 4.281791687011719,
      "learning_rate": 2.182803345240225e-06,
      "loss": 0.0382,
      "step": 13432
    },
    {
      "epoch": 0.6936741544022721,
      "grad_norm": 6.0189008712768555,
      "learning_rate": 2.1821264621598693e-06,
      "loss": 0.0917,
      "step": 13433
    },
    {
      "epoch": 0.6937257939581719,
      "grad_norm": 18.59571075439453,
      "learning_rate": 2.1814496547509035e-06,
      "loss": 0.069,
      "step": 13434
    },
    {
      "epoch": 0.6937774335140717,
      "grad_norm": 2.6884872913360596,
      "learning_rate": 2.1807729230315012e-06,
      "loss": 0.0458,
      "step": 13435
    },
    {
      "epoch": 0.6938290730699715,
      "grad_norm": 6.881412029266357,
      "learning_rate": 2.1800962670198373e-06,
      "loss": 0.1024,
      "step": 13436
    },
    {
      "epoch": 0.6938807126258714,
      "grad_norm": 2.57269549369812,
      "learning_rate": 2.1794196867340835e-06,
      "loss": 0.0303,
      "step": 13437
    },
    {
      "epoch": 0.6939323521817712,
      "grad_norm": 2.6999189853668213,
      "learning_rate": 2.178743182192405e-06,
      "loss": 0.0254,
      "step": 13438
    },
    {
      "epoch": 0.693983991737671,
      "grad_norm": 5.347397327423096,
      "learning_rate": 2.178066753412969e-06,
      "loss": 0.0797,
      "step": 13439
    },
    {
      "epoch": 0.6940356312935708,
      "grad_norm": 2.648314952850342,
      "learning_rate": 2.1773904004139446e-06,
      "loss": 0.037,
      "step": 13440
    },
    {
      "epoch": 0.6940872708494707,
      "grad_norm": 2.5334300994873047,
      "learning_rate": 2.1767141232134937e-06,
      "loss": 0.0292,
      "step": 13441
    },
    {
      "epoch": 0.6941389104053706,
      "grad_norm": 10.117839813232422,
      "learning_rate": 2.176037921829774e-06,
      "loss": 0.2032,
      "step": 13442
    },
    {
      "epoch": 0.6941905499612704,
      "grad_norm": 3.0450870990753174,
      "learning_rate": 2.175361796280946e-06,
      "loss": 0.0468,
      "step": 13443
    },
    {
      "epoch": 0.6942421895171702,
      "grad_norm": 6.941258907318115,
      "learning_rate": 2.1746857465851645e-06,
      "loss": 0.0498,
      "step": 13444
    },
    {
      "epoch": 0.69429382907307,
      "grad_norm": 28.844758987426758,
      "learning_rate": 2.1740097727605896e-06,
      "loss": 0.1839,
      "step": 13445
    },
    {
      "epoch": 0.6943454686289698,
      "grad_norm": 2.962198257446289,
      "learning_rate": 2.1733338748253687e-06,
      "loss": 0.0492,
      "step": 13446
    },
    {
      "epoch": 0.6943971081848697,
      "grad_norm": 11.655594825744629,
      "learning_rate": 2.1726580527976537e-06,
      "loss": 0.1388,
      "step": 13447
    },
    {
      "epoch": 0.6944487477407695,
      "grad_norm": 1.9838064908981323,
      "learning_rate": 2.171982306695594e-06,
      "loss": 0.0269,
      "step": 13448
    },
    {
      "epoch": 0.6945003872966693,
      "grad_norm": 35.01442337036133,
      "learning_rate": 2.171306636537336e-06,
      "loss": 0.0919,
      "step": 13449
    },
    {
      "epoch": 0.6945520268525691,
      "grad_norm": 14.195931434631348,
      "learning_rate": 2.1706310423410233e-06,
      "loss": 0.0922,
      "step": 13450
    },
    {
      "epoch": 0.6946036664084689,
      "grad_norm": 1.384447455406189,
      "learning_rate": 2.1699555241247993e-06,
      "loss": 0.0113,
      "step": 13451
    },
    {
      "epoch": 0.6946553059643688,
      "grad_norm": 3.15437912940979,
      "learning_rate": 2.169280081906804e-06,
      "loss": 0.0299,
      "step": 13452
    },
    {
      "epoch": 0.6947069455202686,
      "grad_norm": 7.395329475402832,
      "learning_rate": 2.168604715705176e-06,
      "loss": 0.0523,
      "step": 13453
    },
    {
      "epoch": 0.6947585850761684,
      "grad_norm": 3.810046434402466,
      "learning_rate": 2.1679294255380514e-06,
      "loss": 0.0452,
      "step": 13454
    },
    {
      "epoch": 0.6948102246320682,
      "grad_norm": 2.489778518676758,
      "learning_rate": 2.1672542114235643e-06,
      "loss": 0.0362,
      "step": 13455
    },
    {
      "epoch": 0.694861864187968,
      "grad_norm": 1.9025155305862427,
      "learning_rate": 2.1665790733798497e-06,
      "loss": 0.0232,
      "step": 13456
    },
    {
      "epoch": 0.6949135037438678,
      "grad_norm": 2.7153892517089844,
      "learning_rate": 2.1659040114250336e-06,
      "loss": 0.0261,
      "step": 13457
    },
    {
      "epoch": 0.6949651432997677,
      "grad_norm": 2.5371909141540527,
      "learning_rate": 2.1652290255772434e-06,
      "loss": 0.056,
      "step": 13458
    },
    {
      "epoch": 0.6950167828556675,
      "grad_norm": 4.10399055480957,
      "learning_rate": 2.164554115854612e-06,
      "loss": 0.0342,
      "step": 13459
    },
    {
      "epoch": 0.6950684224115673,
      "grad_norm": 12.399497032165527,
      "learning_rate": 2.163879282275257e-06,
      "loss": 0.0558,
      "step": 13460
    },
    {
      "epoch": 0.6951200619674671,
      "grad_norm": 5.546214580535889,
      "learning_rate": 2.1632045248573026e-06,
      "loss": 0.0636,
      "step": 13461
    },
    {
      "epoch": 0.6951717015233669,
      "grad_norm": 2.5316922664642334,
      "learning_rate": 2.162529843618869e-06,
      "loss": 0.0348,
      "step": 13462
    },
    {
      "epoch": 0.6952233410792668,
      "grad_norm": 2.644365072250366,
      "learning_rate": 2.1618552385780746e-06,
      "loss": 0.0352,
      "step": 13463
    },
    {
      "epoch": 0.6952749806351666,
      "grad_norm": 3.3150103092193604,
      "learning_rate": 2.1611807097530343e-06,
      "loss": 0.0434,
      "step": 13464
    },
    {
      "epoch": 0.6953266201910664,
      "grad_norm": 5.545908451080322,
      "learning_rate": 2.1605062571618617e-06,
      "loss": 0.0599,
      "step": 13465
    },
    {
      "epoch": 0.6953782597469662,
      "grad_norm": 4.47021484375,
      "learning_rate": 2.15983188082267e-06,
      "loss": 0.047,
      "step": 13466
    },
    {
      "epoch": 0.695429899302866,
      "grad_norm": 3.3535468578338623,
      "learning_rate": 2.1591575807535674e-06,
      "loss": 0.0237,
      "step": 13467
    },
    {
      "epoch": 0.6954815388587658,
      "grad_norm": 4.397764682769775,
      "learning_rate": 2.158483356972663e-06,
      "loss": 0.0387,
      "step": 13468
    },
    {
      "epoch": 0.6955331784146657,
      "grad_norm": 5.906664848327637,
      "learning_rate": 2.157809209498062e-06,
      "loss": 0.0549,
      "step": 13469
    },
    {
      "epoch": 0.6955848179705655,
      "grad_norm": 5.801663398742676,
      "learning_rate": 2.1571351383478695e-06,
      "loss": 0.0636,
      "step": 13470
    },
    {
      "epoch": 0.6956364575264653,
      "grad_norm": 6.016435623168945,
      "learning_rate": 2.156461143540181e-06,
      "loss": 0.1356,
      "step": 13471
    },
    {
      "epoch": 0.6956880970823651,
      "grad_norm": 12.921041488647461,
      "learning_rate": 2.1557872250931032e-06,
      "loss": 0.0438,
      "step": 13472
    },
    {
      "epoch": 0.6957397366382649,
      "grad_norm": 2.3450748920440674,
      "learning_rate": 2.1551133830247307e-06,
      "loss": 0.0314,
      "step": 13473
    },
    {
      "epoch": 0.6957913761941648,
      "grad_norm": 4.269041538238525,
      "learning_rate": 2.1544396173531608e-06,
      "loss": 0.0564,
      "step": 13474
    },
    {
      "epoch": 0.6958430157500646,
      "grad_norm": 3.290346622467041,
      "learning_rate": 2.153765928096483e-06,
      "loss": 0.0569,
      "step": 13475
    },
    {
      "epoch": 0.6958946553059644,
      "grad_norm": 1.3095277547836304,
      "learning_rate": 2.15309231527279e-06,
      "loss": 0.0052,
      "step": 13476
    },
    {
      "epoch": 0.6959462948618642,
      "grad_norm": 1.987243890762329,
      "learning_rate": 2.152418778900171e-06,
      "loss": 0.0466,
      "step": 13477
    },
    {
      "epoch": 0.695997934417764,
      "grad_norm": 2.0978779792785645,
      "learning_rate": 2.1517453189967167e-06,
      "loss": 0.0291,
      "step": 13478
    },
    {
      "epoch": 0.6960495739736638,
      "grad_norm": 2.7948496341705322,
      "learning_rate": 2.151071935580508e-06,
      "loss": 0.0417,
      "step": 13479
    },
    {
      "epoch": 0.6961012135295637,
      "grad_norm": 7.3051066398620605,
      "learning_rate": 2.1503986286696294e-06,
      "loss": 0.046,
      "step": 13480
    },
    {
      "epoch": 0.6961528530854635,
      "grad_norm": 3.0340213775634766,
      "learning_rate": 2.149725398282162e-06,
      "loss": 0.0468,
      "step": 13481
    },
    {
      "epoch": 0.6962044926413633,
      "grad_norm": 7.552919387817383,
      "learning_rate": 2.1490522444361846e-06,
      "loss": 0.1034,
      "step": 13482
    },
    {
      "epoch": 0.6962561321972631,
      "grad_norm": 28.81458854675293,
      "learning_rate": 2.1483791671497745e-06,
      "loss": 0.0541,
      "step": 13483
    },
    {
      "epoch": 0.6963077717531629,
      "grad_norm": 2.4478232860565186,
      "learning_rate": 2.147706166441006e-06,
      "loss": 0.0264,
      "step": 13484
    },
    {
      "epoch": 0.6963594113090628,
      "grad_norm": 14.992273330688477,
      "learning_rate": 2.147033242327952e-06,
      "loss": 0.0944,
      "step": 13485
    },
    {
      "epoch": 0.6964110508649626,
      "grad_norm": 9.81857967376709,
      "learning_rate": 2.146360394828684e-06,
      "loss": 0.1262,
      "step": 13486
    },
    {
      "epoch": 0.6964626904208624,
      "grad_norm": 3.847689390182495,
      "learning_rate": 2.14568762396127e-06,
      "loss": 0.0399,
      "step": 13487
    },
    {
      "epoch": 0.6965143299767622,
      "grad_norm": 5.053318023681641,
      "learning_rate": 2.145014929743777e-06,
      "loss": 0.0624,
      "step": 13488
    },
    {
      "epoch": 0.696565969532662,
      "grad_norm": 9.04240894317627,
      "learning_rate": 2.1443423121942714e-06,
      "loss": 0.04,
      "step": 13489
    },
    {
      "epoch": 0.6966176090885619,
      "grad_norm": 8.503633499145508,
      "learning_rate": 2.1436697713308096e-06,
      "loss": 0.0781,
      "step": 13490
    },
    {
      "epoch": 0.6966692486444617,
      "grad_norm": 6.092586994171143,
      "learning_rate": 2.1429973071714582e-06,
      "loss": 0.0305,
      "step": 13491
    },
    {
      "epoch": 0.6967208882003615,
      "grad_norm": 2.5476090908050537,
      "learning_rate": 2.1423249197342756e-06,
      "loss": 0.0285,
      "step": 13492
    },
    {
      "epoch": 0.6967725277562613,
      "grad_norm": 1.6690845489501953,
      "learning_rate": 2.141652609037314e-06,
      "loss": 0.0165,
      "step": 13493
    },
    {
      "epoch": 0.6968241673121611,
      "grad_norm": 2.457848072052002,
      "learning_rate": 2.14098037509863e-06,
      "loss": 0.0208,
      "step": 13494
    },
    {
      "epoch": 0.6968758068680609,
      "grad_norm": 3.189767599105835,
      "learning_rate": 2.1403082179362756e-06,
      "loss": 0.0418,
      "step": 13495
    },
    {
      "epoch": 0.6969274464239608,
      "grad_norm": 4.6321563720703125,
      "learning_rate": 2.139636137568301e-06,
      "loss": 0.0566,
      "step": 13496
    },
    {
      "epoch": 0.6969790859798606,
      "grad_norm": 6.633030414581299,
      "learning_rate": 2.138964134012755e-06,
      "loss": 0.033,
      "step": 13497
    },
    {
      "epoch": 0.6970307255357604,
      "grad_norm": 2.1115784645080566,
      "learning_rate": 2.138292207287682e-06,
      "loss": 0.0371,
      "step": 13498
    },
    {
      "epoch": 0.6970823650916602,
      "grad_norm": 3.9084670543670654,
      "learning_rate": 2.137620357411127e-06,
      "loss": 0.0611,
      "step": 13499
    },
    {
      "epoch": 0.69713400464756,
      "grad_norm": 9.278324127197266,
      "learning_rate": 2.1369485844011322e-06,
      "loss": 0.1835,
      "step": 13500
    },
    {
      "epoch": 0.6971856442034599,
      "grad_norm": 11.154550552368164,
      "learning_rate": 2.136276888275737e-06,
      "loss": 0.0556,
      "step": 13501
    },
    {
      "epoch": 0.6972372837593597,
      "grad_norm": 8.945772171020508,
      "learning_rate": 2.135605269052979e-06,
      "loss": 0.0603,
      "step": 13502
    },
    {
      "epoch": 0.6972889233152595,
      "grad_norm": 7.548126220703125,
      "learning_rate": 2.1349337267508962e-06,
      "loss": 0.0515,
      "step": 13503
    },
    {
      "epoch": 0.6973405628711593,
      "grad_norm": 13.06295394897461,
      "learning_rate": 2.1342622613875166e-06,
      "loss": 0.0727,
      "step": 13504
    },
    {
      "epoch": 0.6973922024270591,
      "grad_norm": 2.697695016860962,
      "learning_rate": 2.133590872980877e-06,
      "loss": 0.0393,
      "step": 13505
    },
    {
      "epoch": 0.697443841982959,
      "grad_norm": 5.378254413604736,
      "learning_rate": 2.132919561549005e-06,
      "loss": 0.0588,
      "step": 13506
    },
    {
      "epoch": 0.6974954815388588,
      "grad_norm": 5.791621208190918,
      "learning_rate": 2.132248327109931e-06,
      "loss": 0.0684,
      "step": 13507
    },
    {
      "epoch": 0.6975471210947586,
      "grad_norm": 18.194978713989258,
      "learning_rate": 2.131577169681675e-06,
      "loss": 0.1199,
      "step": 13508
    },
    {
      "epoch": 0.6975987606506584,
      "grad_norm": 2.3105592727661133,
      "learning_rate": 2.130906089282262e-06,
      "loss": 0.0533,
      "step": 13509
    },
    {
      "epoch": 0.6976504002065582,
      "grad_norm": 7.139570713043213,
      "learning_rate": 2.1302350859297157e-06,
      "loss": 0.1054,
      "step": 13510
    },
    {
      "epoch": 0.697702039762458,
      "grad_norm": 4.3792405128479,
      "learning_rate": 2.1295641596420557e-06,
      "loss": 0.081,
      "step": 13511
    },
    {
      "epoch": 0.6977536793183579,
      "grad_norm": 6.385483264923096,
      "learning_rate": 2.1288933104372957e-06,
      "loss": 0.0432,
      "step": 13512
    },
    {
      "epoch": 0.6978053188742577,
      "grad_norm": 8.51550579071045,
      "learning_rate": 2.1282225383334515e-06,
      "loss": 0.0616,
      "step": 13513
    },
    {
      "epoch": 0.6978569584301575,
      "grad_norm": 14.408575057983398,
      "learning_rate": 2.1275518433485376e-06,
      "loss": 0.0743,
      "step": 13514
    },
    {
      "epoch": 0.6979085979860573,
      "grad_norm": 10.912947654724121,
      "learning_rate": 2.1268812255005637e-06,
      "loss": 0.1036,
      "step": 13515
    },
    {
      "epoch": 0.6979602375419571,
      "grad_norm": 3.8264949321746826,
      "learning_rate": 2.12621068480754e-06,
      "loss": 0.0435,
      "step": 13516
    },
    {
      "epoch": 0.698011877097857,
      "grad_norm": 3.176243782043457,
      "learning_rate": 2.1255402212874715e-06,
      "loss": 0.0484,
      "step": 13517
    },
    {
      "epoch": 0.6980635166537568,
      "grad_norm": 3.0434443950653076,
      "learning_rate": 2.1248698349583634e-06,
      "loss": 0.0386,
      "step": 13518
    },
    {
      "epoch": 0.6981151562096566,
      "grad_norm": 3.6795098781585693,
      "learning_rate": 2.1241995258382198e-06,
      "loss": 0.0396,
      "step": 13519
    },
    {
      "epoch": 0.6981667957655564,
      "grad_norm": 4.850698947906494,
      "learning_rate": 2.1235292939450393e-06,
      "loss": 0.0621,
      "step": 13520
    },
    {
      "epoch": 0.6982184353214562,
      "grad_norm": 3.1513285636901855,
      "learning_rate": 2.1228591392968212e-06,
      "loss": 0.0443,
      "step": 13521
    },
    {
      "epoch": 0.698270074877356,
      "grad_norm": 5.821083068847656,
      "learning_rate": 2.122189061911563e-06,
      "loss": 0.064,
      "step": 13522
    },
    {
      "epoch": 0.6983217144332559,
      "grad_norm": 3.112612724304199,
      "learning_rate": 2.121519061807255e-06,
      "loss": 0.0373,
      "step": 13523
    },
    {
      "epoch": 0.6983733539891557,
      "grad_norm": 3.0796940326690674,
      "learning_rate": 2.1208491390018937e-06,
      "loss": 0.0398,
      "step": 13524
    },
    {
      "epoch": 0.6984249935450555,
      "grad_norm": 2.205451250076294,
      "learning_rate": 2.1201792935134695e-06,
      "loss": 0.0307,
      "step": 13525
    },
    {
      "epoch": 0.6984766331009553,
      "grad_norm": 1.5856363773345947,
      "learning_rate": 2.1195095253599673e-06,
      "loss": 0.0252,
      "step": 13526
    },
    {
      "epoch": 0.6985282726568551,
      "grad_norm": 8.524869918823242,
      "learning_rate": 2.118839834559375e-06,
      "loss": 0.1665,
      "step": 13527
    },
    {
      "epoch": 0.698579912212755,
      "grad_norm": 2.879950761795044,
      "learning_rate": 2.1181702211296733e-06,
      "loss": 0.0452,
      "step": 13528
    },
    {
      "epoch": 0.6986315517686548,
      "grad_norm": 11.834121704101562,
      "learning_rate": 2.1175006850888507e-06,
      "loss": 0.1546,
      "step": 13529
    },
    {
      "epoch": 0.6986831913245546,
      "grad_norm": 1.869556188583374,
      "learning_rate": 2.116831226454882e-06,
      "loss": 0.019,
      "step": 13530
    },
    {
      "epoch": 0.6987348308804544,
      "grad_norm": 12.734801292419434,
      "learning_rate": 2.116161845245745e-06,
      "loss": 0.1352,
      "step": 13531
    },
    {
      "epoch": 0.6987864704363542,
      "grad_norm": 3.4878392219543457,
      "learning_rate": 2.115492541479417e-06,
      "loss": 0.0686,
      "step": 13532
    },
    {
      "epoch": 0.698838109992254,
      "grad_norm": 9.324979782104492,
      "learning_rate": 2.1148233151738705e-06,
      "loss": 0.1055,
      "step": 13533
    },
    {
      "epoch": 0.6988897495481539,
      "grad_norm": 4.640315055847168,
      "learning_rate": 2.1141541663470777e-06,
      "loss": 0.0677,
      "step": 13534
    },
    {
      "epoch": 0.6989413891040537,
      "grad_norm": 3.527759075164795,
      "learning_rate": 2.113485095017007e-06,
      "loss": 0.0646,
      "step": 13535
    },
    {
      "epoch": 0.6989930286599535,
      "grad_norm": 4.284962177276611,
      "learning_rate": 2.112816101201629e-06,
      "loss": 0.0528,
      "step": 13536
    },
    {
      "epoch": 0.6990446682158533,
      "grad_norm": 6.873495101928711,
      "learning_rate": 2.1121471849189017e-06,
      "loss": 0.0472,
      "step": 13537
    },
    {
      "epoch": 0.6990963077717531,
      "grad_norm": 6.134288311004639,
      "learning_rate": 2.111478346186795e-06,
      "loss": 0.0705,
      "step": 13538
    },
    {
      "epoch": 0.699147947327653,
      "grad_norm": 2.3531289100646973,
      "learning_rate": 2.110809585023268e-06,
      "loss": 0.043,
      "step": 13539
    },
    {
      "epoch": 0.6991995868835528,
      "grad_norm": 15.980062484741211,
      "learning_rate": 2.1101409014462804e-06,
      "loss": 0.1642,
      "step": 13540
    },
    {
      "epoch": 0.6992512264394526,
      "grad_norm": 5.270541667938232,
      "learning_rate": 2.109472295473787e-06,
      "loss": 0.0643,
      "step": 13541
    },
    {
      "epoch": 0.6993028659953524,
      "grad_norm": 3.776124954223633,
      "learning_rate": 2.108803767123741e-06,
      "loss": 0.0467,
      "step": 13542
    },
    {
      "epoch": 0.6993545055512522,
      "grad_norm": 4.2480573654174805,
      "learning_rate": 2.1081353164141e-06,
      "loss": 0.0367,
      "step": 13543
    },
    {
      "epoch": 0.699406145107152,
      "grad_norm": 4.852509498596191,
      "learning_rate": 2.1074669433628136e-06,
      "loss": 0.0489,
      "step": 13544
    },
    {
      "epoch": 0.6994577846630519,
      "grad_norm": 1.8984178304672241,
      "learning_rate": 2.1067986479878278e-06,
      "loss": 0.0268,
      "step": 13545
    },
    {
      "epoch": 0.6995094242189517,
      "grad_norm": 4.12035608291626,
      "learning_rate": 2.10613043030709e-06,
      "loss": 0.0297,
      "step": 13546
    },
    {
      "epoch": 0.6995610637748515,
      "grad_norm": 3.311574935913086,
      "learning_rate": 2.1054622903385453e-06,
      "loss": 0.0393,
      "step": 13547
    },
    {
      "epoch": 0.6996127033307513,
      "grad_norm": 3.762514352798462,
      "learning_rate": 2.1047942281001354e-06,
      "loss": 0.0563,
      "step": 13548
    },
    {
      "epoch": 0.6996643428866511,
      "grad_norm": 2.2307450771331787,
      "learning_rate": 2.1041262436098005e-06,
      "loss": 0.0264,
      "step": 13549
    },
    {
      "epoch": 0.699715982442551,
      "grad_norm": 1.777919054031372,
      "learning_rate": 2.1034583368854793e-06,
      "loss": 0.0287,
      "step": 13550
    },
    {
      "epoch": 0.6997676219984508,
      "grad_norm": 4.6642069816589355,
      "learning_rate": 2.102790507945107e-06,
      "loss": 0.0367,
      "step": 13551
    },
    {
      "epoch": 0.6998192615543506,
      "grad_norm": 5.545788288116455,
      "learning_rate": 2.102122756806618e-06,
      "loss": 0.0778,
      "step": 13552
    },
    {
      "epoch": 0.6998709011102504,
      "grad_norm": 4.196952819824219,
      "learning_rate": 2.1014550834879443e-06,
      "loss": 0.0331,
      "step": 13553
    },
    {
      "epoch": 0.6999225406661502,
      "grad_norm": 2.3256194591522217,
      "learning_rate": 2.1007874880070175e-06,
      "loss": 0.0272,
      "step": 13554
    },
    {
      "epoch": 0.69997418022205,
      "grad_norm": 3.18228816986084,
      "learning_rate": 2.1001199703817605e-06,
      "loss": 0.0328,
      "step": 13555
    },
    {
      "epoch": 0.7000258197779499,
      "grad_norm": 10.998749732971191,
      "learning_rate": 2.0994525306301005e-06,
      "loss": 0.0637,
      "step": 13556
    },
    {
      "epoch": 0.7000774593338497,
      "grad_norm": 12.628999710083008,
      "learning_rate": 2.0987851687699632e-06,
      "loss": 0.0715,
      "step": 13557
    },
    {
      "epoch": 0.7001290988897495,
      "grad_norm": 1.628541111946106,
      "learning_rate": 2.0981178848192714e-06,
      "loss": 0.0351,
      "step": 13558
    },
    {
      "epoch": 0.7001807384456493,
      "grad_norm": 4.542937278747559,
      "learning_rate": 2.0974506787959393e-06,
      "loss": 0.0951,
      "step": 13559
    },
    {
      "epoch": 0.7001807384456493,
      "eval_loss": 0.056631121784448624,
      "eval_runtime": 34.1831,
      "eval_samples_per_second": 45.783,
      "eval_steps_per_second": 5.734,
      "step": 13559
    },
    {
      "epoch": 0.7002323780015491,
      "grad_norm": 6.09835958480835,
      "learning_rate": 2.096783550717887e-06,
      "loss": 0.0626,
      "step": 13560
    },
    {
      "epoch": 0.700284017557449,
      "grad_norm": 3.891915798187256,
      "learning_rate": 2.096116500603027e-06,
      "loss": 0.0609,
      "step": 13561
    },
    {
      "epoch": 0.7003356571133488,
      "grad_norm": 13.709524154663086,
      "learning_rate": 2.0954495284692787e-06,
      "loss": 0.1138,
      "step": 13562
    },
    {
      "epoch": 0.7003872966692486,
      "grad_norm": 2.8756496906280518,
      "learning_rate": 2.094782634334547e-06,
      "loss": 0.0402,
      "step": 13563
    },
    {
      "epoch": 0.7004389362251484,
      "grad_norm": 5.198162078857422,
      "learning_rate": 2.094115818216742e-06,
      "loss": 0.0274,
      "step": 13564
    },
    {
      "epoch": 0.7004905757810482,
      "grad_norm": 3.094038724899292,
      "learning_rate": 2.0934490801337714e-06,
      "loss": 0.0541,
      "step": 13565
    },
    {
      "epoch": 0.700542215336948,
      "grad_norm": 2.9264276027679443,
      "learning_rate": 2.0927824201035395e-06,
      "loss": 0.026,
      "step": 13566
    },
    {
      "epoch": 0.7005938548928479,
      "grad_norm": 3.5738301277160645,
      "learning_rate": 2.092115838143949e-06,
      "loss": 0.0325,
      "step": 13567
    },
    {
      "epoch": 0.7006454944487477,
      "grad_norm": 4.352476596832275,
      "learning_rate": 2.0914493342728996e-06,
      "loss": 0.0547,
      "step": 13568
    },
    {
      "epoch": 0.7006971340046476,
      "grad_norm": 3.281144380569458,
      "learning_rate": 2.0907829085082924e-06,
      "loss": 0.0246,
      "step": 13569
    },
    {
      "epoch": 0.7007487735605474,
      "grad_norm": 4.178084373474121,
      "learning_rate": 2.0901165608680174e-06,
      "loss": 0.0515,
      "step": 13570
    },
    {
      "epoch": 0.7008004131164473,
      "grad_norm": 7.206217288970947,
      "learning_rate": 2.089450291369975e-06,
      "loss": 0.0645,
      "step": 13571
    },
    {
      "epoch": 0.7008520526723471,
      "grad_norm": 4.816487789154053,
      "learning_rate": 2.088784100032055e-06,
      "loss": 0.0636,
      "step": 13572
    },
    {
      "epoch": 0.7009036922282469,
      "grad_norm": 21.552318572998047,
      "learning_rate": 2.088117986872149e-06,
      "loss": 0.1024,
      "step": 13573
    },
    {
      "epoch": 0.7009553317841467,
      "grad_norm": 3.1266417503356934,
      "learning_rate": 2.087451951908141e-06,
      "loss": 0.0158,
      "step": 13574
    },
    {
      "epoch": 0.7010069713400465,
      "grad_norm": 3.922400712966919,
      "learning_rate": 2.0867859951579176e-06,
      "loss": 0.0722,
      "step": 13575
    },
    {
      "epoch": 0.7010586108959463,
      "grad_norm": 2.2138113975524902,
      "learning_rate": 2.0861201166393675e-06,
      "loss": 0.0262,
      "step": 13576
    },
    {
      "epoch": 0.7011102504518462,
      "grad_norm": 4.046550750732422,
      "learning_rate": 2.0854543163703665e-06,
      "loss": 0.0141,
      "step": 13577
    },
    {
      "epoch": 0.701161890007746,
      "grad_norm": 2.5350890159606934,
      "learning_rate": 2.0847885943687967e-06,
      "loss": 0.0317,
      "step": 13578
    },
    {
      "epoch": 0.7012135295636458,
      "grad_norm": 1.946623682975769,
      "learning_rate": 2.0841229506525344e-06,
      "loss": 0.0178,
      "step": 13579
    },
    {
      "epoch": 0.7012651691195456,
      "grad_norm": 2.6917724609375,
      "learning_rate": 2.0834573852394557e-06,
      "loss": 0.0434,
      "step": 13580
    },
    {
      "epoch": 0.7013168086754454,
      "grad_norm": 15.616056442260742,
      "learning_rate": 2.082791898147433e-06,
      "loss": 0.1246,
      "step": 13581
    },
    {
      "epoch": 0.7013684482313453,
      "grad_norm": 2.527012348175049,
      "learning_rate": 2.082126489394338e-06,
      "loss": 0.0528,
      "step": 13582
    },
    {
      "epoch": 0.7014200877872451,
      "grad_norm": 2.0932509899139404,
      "learning_rate": 2.0814611589980393e-06,
      "loss": 0.0191,
      "step": 13583
    },
    {
      "epoch": 0.7014717273431449,
      "grad_norm": 7.033574104309082,
      "learning_rate": 2.080795906976404e-06,
      "loss": 0.1293,
      "step": 13584
    },
    {
      "epoch": 0.7015233668990447,
      "grad_norm": 4.800602436065674,
      "learning_rate": 2.080130733347296e-06,
      "loss": 0.0787,
      "step": 13585
    },
    {
      "epoch": 0.7015750064549445,
      "grad_norm": 2.8117525577545166,
      "learning_rate": 2.079465638128579e-06,
      "loss": 0.0385,
      "step": 13586
    },
    {
      "epoch": 0.7016266460108443,
      "grad_norm": 2.837271213531494,
      "learning_rate": 2.0788006213381147e-06,
      "loss": 0.0342,
      "step": 13587
    },
    {
      "epoch": 0.7016782855667442,
      "grad_norm": 2.7490320205688477,
      "learning_rate": 2.0781356829937582e-06,
      "loss": 0.0476,
      "step": 13588
    },
    {
      "epoch": 0.701729925122644,
      "grad_norm": 4.805720329284668,
      "learning_rate": 2.0774708231133654e-06,
      "loss": 0.0535,
      "step": 13589
    },
    {
      "epoch": 0.7017815646785438,
      "grad_norm": 2.7814276218414307,
      "learning_rate": 2.0768060417147945e-06,
      "loss": 0.0522,
      "step": 13590
    },
    {
      "epoch": 0.7018332042344436,
      "grad_norm": 2.9123198986053467,
      "learning_rate": 2.0761413388158974e-06,
      "loss": 0.0433,
      "step": 13591
    },
    {
      "epoch": 0.7018848437903434,
      "grad_norm": 2.2156224250793457,
      "learning_rate": 2.0754767144345206e-06,
      "loss": 0.0227,
      "step": 13592
    },
    {
      "epoch": 0.7019364833462433,
      "grad_norm": 3.0687732696533203,
      "learning_rate": 2.0748121685885135e-06,
      "loss": 0.0343,
      "step": 13593
    },
    {
      "epoch": 0.7019881229021431,
      "grad_norm": 9.816119194030762,
      "learning_rate": 2.0741477012957207e-06,
      "loss": 0.0642,
      "step": 13594
    },
    {
      "epoch": 0.7020397624580429,
      "grad_norm": 6.119958400726318,
      "learning_rate": 2.0734833125739908e-06,
      "loss": 0.0835,
      "step": 13595
    },
    {
      "epoch": 0.7020914020139427,
      "grad_norm": 3.3334662914276123,
      "learning_rate": 2.0728190024411597e-06,
      "loss": 0.0615,
      "step": 13596
    },
    {
      "epoch": 0.7021430415698425,
      "grad_norm": 2.0053462982177734,
      "learning_rate": 2.072154770915069e-06,
      "loss": 0.0302,
      "step": 13597
    },
    {
      "epoch": 0.7021946811257423,
      "grad_norm": 2.2033448219299316,
      "learning_rate": 2.0714906180135564e-06,
      "loss": 0.0207,
      "step": 13598
    },
    {
      "epoch": 0.7022463206816422,
      "grad_norm": 4.043063163757324,
      "learning_rate": 2.070826543754456e-06,
      "loss": 0.0409,
      "step": 13599
    },
    {
      "epoch": 0.702297960237542,
      "grad_norm": 4.510144233703613,
      "learning_rate": 2.070162548155602e-06,
      "loss": 0.0631,
      "step": 13600
    },
    {
      "epoch": 0.7023495997934418,
      "grad_norm": 3.1207003593444824,
      "learning_rate": 2.0694986312348247e-06,
      "loss": 0.053,
      "step": 13601
    },
    {
      "epoch": 0.7024012393493416,
      "grad_norm": 9.771223068237305,
      "learning_rate": 2.068834793009955e-06,
      "loss": 0.0837,
      "step": 13602
    },
    {
      "epoch": 0.7024528789052414,
      "grad_norm": 6.045623302459717,
      "learning_rate": 2.068171033498814e-06,
      "loss": 0.082,
      "step": 13603
    },
    {
      "epoch": 0.7025045184611413,
      "grad_norm": 3.3472588062286377,
      "learning_rate": 2.0675073527192323e-06,
      "loss": 0.0542,
      "step": 13604
    },
    {
      "epoch": 0.7025561580170411,
      "grad_norm": 3.080207586288452,
      "learning_rate": 2.066843750689031e-06,
      "loss": 0.0296,
      "step": 13605
    },
    {
      "epoch": 0.7026077975729409,
      "grad_norm": 10.321681022644043,
      "learning_rate": 2.0661802274260306e-06,
      "loss": 0.0917,
      "step": 13606
    },
    {
      "epoch": 0.7026594371288407,
      "grad_norm": 3.2445805072784424,
      "learning_rate": 2.0655167829480476e-06,
      "loss": 0.0361,
      "step": 13607
    },
    {
      "epoch": 0.7027110766847405,
      "grad_norm": 5.2933759689331055,
      "learning_rate": 2.0648534172728973e-06,
      "loss": 0.0355,
      "step": 13608
    },
    {
      "epoch": 0.7027627162406404,
      "grad_norm": 2.030409097671509,
      "learning_rate": 2.064190130418399e-06,
      "loss": 0.0264,
      "step": 13609
    },
    {
      "epoch": 0.7028143557965402,
      "grad_norm": 10.943961143493652,
      "learning_rate": 2.06352692240236e-06,
      "loss": 0.1615,
      "step": 13610
    },
    {
      "epoch": 0.70286599535244,
      "grad_norm": 6.329029083251953,
      "learning_rate": 2.0628637932425915e-06,
      "loss": 0.085,
      "step": 13611
    },
    {
      "epoch": 0.7029176349083398,
      "grad_norm": 2.731304407119751,
      "learning_rate": 2.0622007429569015e-06,
      "loss": 0.0338,
      "step": 13612
    },
    {
      "epoch": 0.7029692744642396,
      "grad_norm": 10.845376014709473,
      "learning_rate": 2.0615377715630948e-06,
      "loss": 0.0428,
      "step": 13613
    },
    {
      "epoch": 0.7030209140201394,
      "grad_norm": 3.4180409908294678,
      "learning_rate": 2.0608748790789757e-06,
      "loss": 0.0637,
      "step": 13614
    },
    {
      "epoch": 0.7030725535760393,
      "grad_norm": 3.1894445419311523,
      "learning_rate": 2.060212065522345e-06,
      "loss": 0.0517,
      "step": 13615
    },
    {
      "epoch": 0.7031241931319391,
      "grad_norm": 3.1150858402252197,
      "learning_rate": 2.0595493309110025e-06,
      "loss": 0.0416,
      "step": 13616
    },
    {
      "epoch": 0.7031758326878389,
      "grad_norm": 5.857375144958496,
      "learning_rate": 2.058886675262745e-06,
      "loss": 0.0344,
      "step": 13617
    },
    {
      "epoch": 0.7032274722437387,
      "grad_norm": 4.123198986053467,
      "learning_rate": 2.0582240985953676e-06,
      "loss": 0.0763,
      "step": 13618
    },
    {
      "epoch": 0.7032791117996385,
      "grad_norm": 5.12405252456665,
      "learning_rate": 2.0575616009266634e-06,
      "loss": 0.0426,
      "step": 13619
    },
    {
      "epoch": 0.7033307513555384,
      "grad_norm": 4.362744331359863,
      "learning_rate": 2.056899182274425e-06,
      "loss": 0.051,
      "step": 13620
    },
    {
      "epoch": 0.7033823909114382,
      "grad_norm": 2.537060499191284,
      "learning_rate": 2.0562368426564365e-06,
      "loss": 0.0284,
      "step": 13621
    },
    {
      "epoch": 0.703434030467338,
      "grad_norm": 3.8366222381591797,
      "learning_rate": 2.0555745820904853e-06,
      "loss": 0.0212,
      "step": 13622
    },
    {
      "epoch": 0.7034856700232378,
      "grad_norm": 16.68157386779785,
      "learning_rate": 2.0549124005943593e-06,
      "loss": 0.0524,
      "step": 13623
    },
    {
      "epoch": 0.7035373095791376,
      "grad_norm": 3.7747862339019775,
      "learning_rate": 2.054250298185841e-06,
      "loss": 0.0551,
      "step": 13624
    },
    {
      "epoch": 0.7035889491350374,
      "grad_norm": 13.895265579223633,
      "learning_rate": 2.053588274882706e-06,
      "loss": 0.0887,
      "step": 13625
    },
    {
      "epoch": 0.7036405886909373,
      "grad_norm": 6.5883564949035645,
      "learning_rate": 2.052926330702735e-06,
      "loss": 0.0372,
      "step": 13626
    },
    {
      "epoch": 0.7036922282468371,
      "grad_norm": 7.33372688293457,
      "learning_rate": 2.0522644656637015e-06,
      "loss": 0.1179,
      "step": 13627
    },
    {
      "epoch": 0.7037438678027369,
      "grad_norm": 2.7926971912384033,
      "learning_rate": 2.0516026797833853e-06,
      "loss": 0.0547,
      "step": 13628
    },
    {
      "epoch": 0.7037955073586367,
      "grad_norm": 1.5770466327667236,
      "learning_rate": 2.050940973079552e-06,
      "loss": 0.0158,
      "step": 13629
    },
    {
      "epoch": 0.7038471469145365,
      "grad_norm": 2.3574328422546387,
      "learning_rate": 2.0502793455699736e-06,
      "loss": 0.0421,
      "step": 13630
    },
    {
      "epoch": 0.7038987864704364,
      "grad_norm": 16.09488868713379,
      "learning_rate": 2.0496177972724176e-06,
      "loss": 0.1133,
      "step": 13631
    },
    {
      "epoch": 0.7039504260263362,
      "grad_norm": 4.929728031158447,
      "learning_rate": 2.048956328204648e-06,
      "loss": 0.046,
      "step": 13632
    },
    {
      "epoch": 0.704002065582236,
      "grad_norm": 5.996077537536621,
      "learning_rate": 2.0482949383844293e-06,
      "loss": 0.0531,
      "step": 13633
    },
    {
      "epoch": 0.7040537051381358,
      "grad_norm": 10.226825714111328,
      "learning_rate": 2.0476336278295216e-06,
      "loss": 0.0657,
      "step": 13634
    },
    {
      "epoch": 0.7041053446940356,
      "grad_norm": 4.903057098388672,
      "learning_rate": 2.0469723965576864e-06,
      "loss": 0.0566,
      "step": 13635
    },
    {
      "epoch": 0.7041569842499354,
      "grad_norm": 2.41310977935791,
      "learning_rate": 2.0463112445866744e-06,
      "loss": 0.0352,
      "step": 13636
    },
    {
      "epoch": 0.7042086238058353,
      "grad_norm": 3.668764591217041,
      "learning_rate": 2.045650171934246e-06,
      "loss": 0.069,
      "step": 13637
    },
    {
      "epoch": 0.7042602633617351,
      "grad_norm": 3.8545494079589844,
      "learning_rate": 2.0449891786181518e-06,
      "loss": 0.0571,
      "step": 13638
    },
    {
      "epoch": 0.7043119029176349,
      "grad_norm": 3.494389057159424,
      "learning_rate": 2.0443282646561436e-06,
      "loss": 0.0515,
      "step": 13639
    },
    {
      "epoch": 0.7043635424735347,
      "grad_norm": 7.75425910949707,
      "learning_rate": 2.0436674300659663e-06,
      "loss": 0.0261,
      "step": 13640
    },
    {
      "epoch": 0.7044151820294345,
      "grad_norm": 2.824897289276123,
      "learning_rate": 2.0430066748653666e-06,
      "loss": 0.032,
      "step": 13641
    },
    {
      "epoch": 0.7044668215853344,
      "grad_norm": 4.048281192779541,
      "learning_rate": 2.0423459990720927e-06,
      "loss": 0.0645,
      "step": 13642
    },
    {
      "epoch": 0.7045184611412342,
      "grad_norm": 5.872314929962158,
      "learning_rate": 2.041685402703882e-06,
      "loss": 0.0657,
      "step": 13643
    },
    {
      "epoch": 0.704570100697134,
      "grad_norm": 2.6756622791290283,
      "learning_rate": 2.0410248857784758e-06,
      "loss": 0.0259,
      "step": 13644
    },
    {
      "epoch": 0.7046217402530338,
      "grad_norm": 5.0187859535217285,
      "learning_rate": 2.0403644483136107e-06,
      "loss": 0.037,
      "step": 13645
    },
    {
      "epoch": 0.7046733798089336,
      "grad_norm": 3.2187728881835938,
      "learning_rate": 2.039704090327024e-06,
      "loss": 0.0214,
      "step": 13646
    },
    {
      "epoch": 0.7047250193648334,
      "grad_norm": 1.9888978004455566,
      "learning_rate": 2.0390438118364477e-06,
      "loss": 0.027,
      "step": 13647
    },
    {
      "epoch": 0.7047766589207333,
      "grad_norm": 6.535171985626221,
      "learning_rate": 2.0383836128596128e-06,
      "loss": 0.0958,
      "step": 13648
    },
    {
      "epoch": 0.7048282984766331,
      "grad_norm": 8.752528190612793,
      "learning_rate": 2.037723493414249e-06,
      "loss": 0.0562,
      "step": 13649
    },
    {
      "epoch": 0.7048799380325329,
      "grad_norm": 6.653469085693359,
      "learning_rate": 2.0370634535180833e-06,
      "loss": 0.0319,
      "step": 13650
    },
    {
      "epoch": 0.7049315775884327,
      "grad_norm": 5.519618988037109,
      "learning_rate": 2.03640349318884e-06,
      "loss": 0.0734,
      "step": 13651
    },
    {
      "epoch": 0.7049832171443325,
      "grad_norm": 3.5101640224456787,
      "learning_rate": 2.0357436124442413e-06,
      "loss": 0.0452,
      "step": 13652
    },
    {
      "epoch": 0.7050348567002324,
      "grad_norm": 2.9855825901031494,
      "learning_rate": 2.035083811302011e-06,
      "loss": 0.047,
      "step": 13653
    },
    {
      "epoch": 0.7050864962561322,
      "grad_norm": 2.3133106231689453,
      "learning_rate": 2.0344240897798624e-06,
      "loss": 0.0317,
      "step": 13654
    },
    {
      "epoch": 0.705138135812032,
      "grad_norm": 6.328242301940918,
      "learning_rate": 2.0337644478955122e-06,
      "loss": 0.0385,
      "step": 13655
    },
    {
      "epoch": 0.7051897753679318,
      "grad_norm": 7.1317925453186035,
      "learning_rate": 2.033104885666678e-06,
      "loss": 0.0827,
      "step": 13656
    },
    {
      "epoch": 0.7052414149238316,
      "grad_norm": 1.6850874423980713,
      "learning_rate": 2.0324454031110717e-06,
      "loss": 0.034,
      "step": 13657
    },
    {
      "epoch": 0.7052930544797315,
      "grad_norm": 5.290034294128418,
      "learning_rate": 2.0317860002464e-06,
      "loss": 0.0458,
      "step": 13658
    },
    {
      "epoch": 0.7053446940356313,
      "grad_norm": 11.567654609680176,
      "learning_rate": 2.0311266770903717e-06,
      "loss": 0.0566,
      "step": 13659
    },
    {
      "epoch": 0.7053963335915311,
      "grad_norm": 7.632626533508301,
      "learning_rate": 2.030467433660691e-06,
      "loss": 0.036,
      "step": 13660
    },
    {
      "epoch": 0.7054479731474309,
      "grad_norm": 4.002381324768066,
      "learning_rate": 2.029808269975066e-06,
      "loss": 0.0518,
      "step": 13661
    },
    {
      "epoch": 0.7054996127033307,
      "grad_norm": 18.212068557739258,
      "learning_rate": 2.0291491860511935e-06,
      "loss": 0.0747,
      "step": 13662
    },
    {
      "epoch": 0.7055512522592305,
      "grad_norm": 11.811059951782227,
      "learning_rate": 2.0284901819067738e-06,
      "loss": 0.127,
      "step": 13663
    },
    {
      "epoch": 0.7056028918151304,
      "grad_norm": 3.955150842666626,
      "learning_rate": 2.0278312575595035e-06,
      "loss": 0.0511,
      "step": 13664
    },
    {
      "epoch": 0.7056545313710302,
      "grad_norm": 14.80383586883545,
      "learning_rate": 2.027172413027078e-06,
      "loss": 0.2508,
      "step": 13665
    },
    {
      "epoch": 0.70570617092693,
      "grad_norm": 3.636312246322632,
      "learning_rate": 2.02651364832719e-06,
      "loss": 0.0314,
      "step": 13666
    },
    {
      "epoch": 0.7057578104828298,
      "grad_norm": 12.578056335449219,
      "learning_rate": 2.02585496347753e-06,
      "loss": 0.1096,
      "step": 13667
    },
    {
      "epoch": 0.7058094500387296,
      "grad_norm": 7.062102794647217,
      "learning_rate": 2.0251963584957873e-06,
      "loss": 0.0865,
      "step": 13668
    },
    {
      "epoch": 0.7058610895946295,
      "grad_norm": 2.4762187004089355,
      "learning_rate": 2.024537833399644e-06,
      "loss": 0.0382,
      "step": 13669
    },
    {
      "epoch": 0.7059127291505293,
      "grad_norm": 2.770996332168579,
      "learning_rate": 2.0238793882067887e-06,
      "loss": 0.0416,
      "step": 13670
    },
    {
      "epoch": 0.7059643687064291,
      "grad_norm": 2.2869324684143066,
      "learning_rate": 2.023221022934902e-06,
      "loss": 0.0297,
      "step": 13671
    },
    {
      "epoch": 0.7060160082623289,
      "grad_norm": 3.486518144607544,
      "learning_rate": 2.022562737601665e-06,
      "loss": 0.0678,
      "step": 13672
    },
    {
      "epoch": 0.7060676478182287,
      "grad_norm": 15.189983367919922,
      "learning_rate": 2.021904532224752e-06,
      "loss": 0.1293,
      "step": 13673
    },
    {
      "epoch": 0.7061192873741285,
      "grad_norm": 6.534488201141357,
      "learning_rate": 2.021246406821839e-06,
      "loss": 0.0356,
      "step": 13674
    },
    {
      "epoch": 0.7061709269300284,
      "grad_norm": 3.6936981678009033,
      "learning_rate": 2.020588361410604e-06,
      "loss": 0.0382,
      "step": 13675
    },
    {
      "epoch": 0.7062225664859282,
      "grad_norm": 2.703415632247925,
      "learning_rate": 2.019930396008713e-06,
      "loss": 0.0384,
      "step": 13676
    },
    {
      "epoch": 0.706274206041828,
      "grad_norm": 4.493269920349121,
      "learning_rate": 2.019272510633837e-06,
      "loss": 0.047,
      "step": 13677
    },
    {
      "epoch": 0.7063258455977278,
      "grad_norm": 7.499509334564209,
      "learning_rate": 2.018614705303644e-06,
      "loss": 0.0647,
      "step": 13678
    },
    {
      "epoch": 0.7063774851536276,
      "grad_norm": 25.047504425048828,
      "learning_rate": 2.0179569800357962e-06,
      "loss": 0.2386,
      "step": 13679
    },
    {
      "epoch": 0.7064291247095275,
      "grad_norm": 18.12742805480957,
      "learning_rate": 2.017299334847958e-06,
      "loss": 0.2559,
      "step": 13680
    },
    {
      "epoch": 0.7064807642654273,
      "grad_norm": 14.892402648925781,
      "learning_rate": 2.0166417697577897e-06,
      "loss": 0.1059,
      "step": 13681
    },
    {
      "epoch": 0.7065324038213271,
      "grad_norm": 5.690623760223389,
      "learning_rate": 2.015984284782949e-06,
      "loss": 0.0351,
      "step": 13682
    },
    {
      "epoch": 0.7065840433772269,
      "grad_norm": 5.6135334968566895,
      "learning_rate": 2.0153268799410923e-06,
      "loss": 0.1074,
      "step": 13683
    },
    {
      "epoch": 0.7066356829331267,
      "grad_norm": 1.7501188516616821,
      "learning_rate": 2.0146695552498745e-06,
      "loss": 0.0316,
      "step": 13684
    },
    {
      "epoch": 0.7066873224890265,
      "grad_norm": 3.8808135986328125,
      "learning_rate": 2.014012310726946e-06,
      "loss": 0.0647,
      "step": 13685
    },
    {
      "epoch": 0.7067389620449264,
      "grad_norm": 2.9620089530944824,
      "learning_rate": 2.0133551463899593e-06,
      "loss": 0.0331,
      "step": 13686
    },
    {
      "epoch": 0.7067906016008262,
      "grad_norm": 5.221902847290039,
      "learning_rate": 2.012698062256558e-06,
      "loss": 0.0425,
      "step": 13687
    },
    {
      "epoch": 0.706842241156726,
      "grad_norm": 9.145194053649902,
      "learning_rate": 2.012041058344387e-06,
      "loss": 0.0581,
      "step": 13688
    },
    {
      "epoch": 0.7068938807126258,
      "grad_norm": 5.103292465209961,
      "learning_rate": 2.0113841346710942e-06,
      "loss": 0.0578,
      "step": 13689
    },
    {
      "epoch": 0.7069455202685256,
      "grad_norm": 1.5619001388549805,
      "learning_rate": 2.01072729125432e-06,
      "loss": 0.0211,
      "step": 13690
    },
    {
      "epoch": 0.7069971598244255,
      "grad_norm": 13.566744804382324,
      "learning_rate": 2.0100705281117e-06,
      "loss": 0.0548,
      "step": 13691
    },
    {
      "epoch": 0.7070487993803253,
      "grad_norm": 17.648252487182617,
      "learning_rate": 2.0094138452608725e-06,
      "loss": 0.1082,
      "step": 13692
    },
    {
      "epoch": 0.7071004389362251,
      "grad_norm": 16.55352210998535,
      "learning_rate": 2.00875724271947e-06,
      "loss": 0.0752,
      "step": 13693
    },
    {
      "epoch": 0.7071520784921249,
      "grad_norm": 4.533754348754883,
      "learning_rate": 2.008100720505131e-06,
      "loss": 0.0647,
      "step": 13694
    },
    {
      "epoch": 0.7072037180480247,
      "grad_norm": 2.969466209411621,
      "learning_rate": 2.0074442786354805e-06,
      "loss": 0.028,
      "step": 13695
    },
    {
      "epoch": 0.7072553576039247,
      "grad_norm": 2.0127344131469727,
      "learning_rate": 2.0067879171281477e-06,
      "loss": 0.0505,
      "step": 13696
    },
    {
      "epoch": 0.7073069971598245,
      "grad_norm": 4.477894306182861,
      "learning_rate": 2.0061316360007595e-06,
      "loss": 0.0704,
      "step": 13697
    },
    {
      "epoch": 0.7073586367157243,
      "grad_norm": 1.9269590377807617,
      "learning_rate": 2.0054754352709388e-06,
      "loss": 0.0436,
      "step": 13698
    },
    {
      "epoch": 0.7074102762716241,
      "grad_norm": 5.398134231567383,
      "learning_rate": 2.0048193149563076e-06,
      "loss": 0.0529,
      "step": 13699
    },
    {
      "epoch": 0.7074619158275239,
      "grad_norm": 4.278831958770752,
      "learning_rate": 2.0041632750744856e-06,
      "loss": 0.053,
      "step": 13700
    },
    {
      "epoch": 0.7075135553834238,
      "grad_norm": 8.017248153686523,
      "learning_rate": 2.003507315643092e-06,
      "loss": 0.1099,
      "step": 13701
    },
    {
      "epoch": 0.7075651949393236,
      "grad_norm": 3.792670488357544,
      "learning_rate": 2.002851436679736e-06,
      "loss": 0.0357,
      "step": 13702
    },
    {
      "epoch": 0.7076168344952234,
      "grad_norm": 3.2052576541900635,
      "learning_rate": 2.002195638202036e-06,
      "loss": 0.0469,
      "step": 13703
    },
    {
      "epoch": 0.7076684740511232,
      "grad_norm": 2.6070868968963623,
      "learning_rate": 2.0015399202276026e-06,
      "loss": 0.0205,
      "step": 13704
    },
    {
      "epoch": 0.707720113607023,
      "grad_norm": 4.298387050628662,
      "learning_rate": 2.000884282774045e-06,
      "loss": 0.0328,
      "step": 13705
    },
    {
      "epoch": 0.7077717531629228,
      "grad_norm": 2.2306768894195557,
      "learning_rate": 2.0002287258589658e-06,
      "loss": 0.0367,
      "step": 13706
    },
    {
      "epoch": 0.7078233927188227,
      "grad_norm": 1.462233543395996,
      "learning_rate": 1.99957324949997e-06,
      "loss": 0.0142,
      "step": 13707
    },
    {
      "epoch": 0.7078750322747225,
      "grad_norm": 4.442813396453857,
      "learning_rate": 1.9989178537146653e-06,
      "loss": 0.0552,
      "step": 13708
    },
    {
      "epoch": 0.7079266718306223,
      "grad_norm": 7.084499359130859,
      "learning_rate": 1.998262538520646e-06,
      "loss": 0.0314,
      "step": 13709
    },
    {
      "epoch": 0.7079783113865221,
      "grad_norm": 3.303330898284912,
      "learning_rate": 1.997607303935512e-06,
      "loss": 0.0543,
      "step": 13710
    },
    {
      "epoch": 0.7080299509424219,
      "grad_norm": 8.162323951721191,
      "learning_rate": 1.996952149976859e-06,
      "loss": 0.0931,
      "step": 13711
    },
    {
      "epoch": 0.7080815904983218,
      "grad_norm": 10.292730331420898,
      "learning_rate": 1.996297076662281e-06,
      "loss": 0.0798,
      "step": 13712
    },
    {
      "epoch": 0.7081332300542216,
      "grad_norm": 4.989542007446289,
      "learning_rate": 1.9956420840093684e-06,
      "loss": 0.049,
      "step": 13713
    },
    {
      "epoch": 0.7081848696101214,
      "grad_norm": 2.2967147827148438,
      "learning_rate": 1.9949871720357105e-06,
      "loss": 0.0328,
      "step": 13714
    },
    {
      "epoch": 0.7082365091660212,
      "grad_norm": 6.964964866638184,
      "learning_rate": 1.9943323407588955e-06,
      "loss": 0.0326,
      "step": 13715
    },
    {
      "epoch": 0.708288148721921,
      "grad_norm": 5.932208061218262,
      "learning_rate": 1.9936775901965073e-06,
      "loss": 0.0829,
      "step": 13716
    },
    {
      "epoch": 0.7083397882778208,
      "grad_norm": 8.951225280761719,
      "learning_rate": 1.993022920366129e-06,
      "loss": 0.0516,
      "step": 13717
    },
    {
      "epoch": 0.7083914278337207,
      "grad_norm": 53.50362014770508,
      "learning_rate": 1.992368331285341e-06,
      "loss": 0.4989,
      "step": 13718
    },
    {
      "epoch": 0.7084430673896205,
      "grad_norm": 2.4734926223754883,
      "learning_rate": 1.9917138229717233e-06,
      "loss": 0.0306,
      "step": 13719
    },
    {
      "epoch": 0.7084947069455203,
      "grad_norm": 11.260939598083496,
      "learning_rate": 1.991059395442849e-06,
      "loss": 0.0794,
      "step": 13720
    },
    {
      "epoch": 0.7085463465014201,
      "grad_norm": 2.527959108352661,
      "learning_rate": 1.990405048716292e-06,
      "loss": 0.0228,
      "step": 13721
    },
    {
      "epoch": 0.7085979860573199,
      "grad_norm": 2.8324711322784424,
      "learning_rate": 1.9897507828096275e-06,
      "loss": 0.0302,
      "step": 13722
    },
    {
      "epoch": 0.7086496256132198,
      "grad_norm": 1.6234067678451538,
      "learning_rate": 1.9890965977404254e-06,
      "loss": 0.0256,
      "step": 13723
    },
    {
      "epoch": 0.7087012651691196,
      "grad_norm": 5.608086585998535,
      "learning_rate": 1.988442493526249e-06,
      "loss": 0.0328,
      "step": 13724
    },
    {
      "epoch": 0.7087529047250194,
      "grad_norm": 2.1186649799346924,
      "learning_rate": 1.9877884701846673e-06,
      "loss": 0.0591,
      "step": 13725
    },
    {
      "epoch": 0.7088045442809192,
      "grad_norm": 2.195462465286255,
      "learning_rate": 1.987134527733239e-06,
      "loss": 0.0338,
      "step": 13726
    },
    {
      "epoch": 0.708856183836819,
      "grad_norm": 1.8200898170471191,
      "learning_rate": 1.9864806661895337e-06,
      "loss": 0.0257,
      "step": 13727
    },
    {
      "epoch": 0.7089078233927188,
      "grad_norm": 30.619953155517578,
      "learning_rate": 1.9858268855711023e-06,
      "loss": 0.0957,
      "step": 13728
    },
    {
      "epoch": 0.7089594629486187,
      "grad_norm": 2.1885054111480713,
      "learning_rate": 1.9851731858955037e-06,
      "loss": 0.0545,
      "step": 13729
    },
    {
      "epoch": 0.7090111025045185,
      "grad_norm": 3.8935842514038086,
      "learning_rate": 1.9845195671802932e-06,
      "loss": 0.0372,
      "step": 13730
    },
    {
      "epoch": 0.7090627420604183,
      "grad_norm": 4.730315208435059,
      "learning_rate": 1.983866029443023e-06,
      "loss": 0.0679,
      "step": 13731
    },
    {
      "epoch": 0.7091143816163181,
      "grad_norm": 12.696858406066895,
      "learning_rate": 1.9832125727012423e-06,
      "loss": 0.1335,
      "step": 13732
    },
    {
      "epoch": 0.7091660211722179,
      "grad_norm": 5.523862361907959,
      "learning_rate": 1.9825591969725006e-06,
      "loss": 0.0588,
      "step": 13733
    },
    {
      "epoch": 0.7092176607281178,
      "grad_norm": 10.676239013671875,
      "learning_rate": 1.9819059022743443e-06,
      "loss": 0.1341,
      "step": 13734
    },
    {
      "epoch": 0.7092693002840176,
      "grad_norm": 4.05470085144043,
      "learning_rate": 1.9812526886243115e-06,
      "loss": 0.0494,
      "step": 13735
    },
    {
      "epoch": 0.7093209398399174,
      "grad_norm": 3.0863654613494873,
      "learning_rate": 1.9805995560399505e-06,
      "loss": 0.0519,
      "step": 13736
    },
    {
      "epoch": 0.7093725793958172,
      "grad_norm": 18.545873641967773,
      "learning_rate": 1.9799465045387972e-06,
      "loss": 0.12,
      "step": 13737
    },
    {
      "epoch": 0.709424218951717,
      "grad_norm": 8.813986778259277,
      "learning_rate": 1.979293534138391e-06,
      "loss": 0.1021,
      "step": 13738
    },
    {
      "epoch": 0.7094758585076169,
      "grad_norm": 4.686181545257568,
      "learning_rate": 1.978640644856264e-06,
      "loss": 0.0724,
      "step": 13739
    },
    {
      "epoch": 0.7095274980635167,
      "grad_norm": 1.6493593454360962,
      "learning_rate": 1.9779878367099477e-06,
      "loss": 0.0215,
      "step": 13740
    },
    {
      "epoch": 0.7095791376194165,
      "grad_norm": 6.5721025466918945,
      "learning_rate": 1.9773351097169785e-06,
      "loss": 0.0619,
      "step": 13741
    },
    {
      "epoch": 0.7096307771753163,
      "grad_norm": 3.8691704273223877,
      "learning_rate": 1.9766824638948802e-06,
      "loss": 0.0375,
      "step": 13742
    },
    {
      "epoch": 0.7096824167312161,
      "grad_norm": 5.782525062561035,
      "learning_rate": 1.976029899261179e-06,
      "loss": 0.0459,
      "step": 13743
    },
    {
      "epoch": 0.7097340562871159,
      "grad_norm": 3.562318801879883,
      "learning_rate": 1.975377415833401e-06,
      "loss": 0.0541,
      "step": 13744
    },
    {
      "epoch": 0.7097856958430158,
      "grad_norm": 3.859095573425293,
      "learning_rate": 1.9747250136290667e-06,
      "loss": 0.0792,
      "step": 13745
    },
    {
      "epoch": 0.7098373353989156,
      "grad_norm": 2.5713934898376465,
      "learning_rate": 1.9740726926656962e-06,
      "loss": 0.0423,
      "step": 13746
    },
    {
      "epoch": 0.7098889749548154,
      "grad_norm": 5.817602634429932,
      "learning_rate": 1.973420452960807e-06,
      "loss": 0.0857,
      "step": 13747
    },
    {
      "epoch": 0.7099406145107152,
      "grad_norm": 17.46750259399414,
      "learning_rate": 1.972768294531914e-06,
      "loss": 0.179,
      "step": 13748
    },
    {
      "epoch": 0.709992254066615,
      "grad_norm": 4.862061023712158,
      "learning_rate": 1.972116217396531e-06,
      "loss": 0.0769,
      "step": 13749
    },
    {
      "epoch": 0.7100438936225149,
      "grad_norm": 2.913259267807007,
      "learning_rate": 1.9714642215721684e-06,
      "loss": 0.0447,
      "step": 13750
    },
    {
      "epoch": 0.7100955331784147,
      "grad_norm": 3.400441884994507,
      "learning_rate": 1.970812307076335e-06,
      "loss": 0.0516,
      "step": 13751
    },
    {
      "epoch": 0.7101471727343145,
      "grad_norm": 2.4079384803771973,
      "learning_rate": 1.970160473926539e-06,
      "loss": 0.0376,
      "step": 13752
    },
    {
      "epoch": 0.7101988122902143,
      "grad_norm": 2.339087963104248,
      "learning_rate": 1.969508722140282e-06,
      "loss": 0.0297,
      "step": 13753
    },
    {
      "epoch": 0.7102504518461141,
      "grad_norm": 2.2859110832214355,
      "learning_rate": 1.9688570517350654e-06,
      "loss": 0.0295,
      "step": 13754
    },
    {
      "epoch": 0.710302091402014,
      "grad_norm": 1.8333015441894531,
      "learning_rate": 1.9682054627283926e-06,
      "loss": 0.0327,
      "step": 13755
    },
    {
      "epoch": 0.7103537309579138,
      "grad_norm": 8.20678997039795,
      "learning_rate": 1.9675539551377622e-06,
      "loss": 0.0674,
      "step": 13756
    },
    {
      "epoch": 0.7104053705138136,
      "grad_norm": 6.6934309005737305,
      "learning_rate": 1.9669025289806656e-06,
      "loss": 0.068,
      "step": 13757
    },
    {
      "epoch": 0.7104570100697134,
      "grad_norm": 4.01378870010376,
      "learning_rate": 1.9662511842745984e-06,
      "loss": 0.0588,
      "step": 13758
    },
    {
      "epoch": 0.7105086496256132,
      "grad_norm": 2.515359401702881,
      "learning_rate": 1.9655999210370504e-06,
      "loss": 0.0227,
      "step": 13759
    },
    {
      "epoch": 0.710560289181513,
      "grad_norm": 14.88364028930664,
      "learning_rate": 1.964948739285515e-06,
      "loss": 0.1168,
      "step": 13760
    },
    {
      "epoch": 0.7106119287374129,
      "grad_norm": 15.18450927734375,
      "learning_rate": 1.9642976390374744e-06,
      "loss": 0.0927,
      "step": 13761
    },
    {
      "epoch": 0.7106635682933127,
      "grad_norm": 16.10773277282715,
      "learning_rate": 1.963646620310416e-06,
      "loss": 0.0338,
      "step": 13762
    },
    {
      "epoch": 0.7107152078492125,
      "grad_norm": 3.2807798385620117,
      "learning_rate": 1.9629956831218205e-06,
      "loss": 0.0252,
      "step": 13763
    },
    {
      "epoch": 0.7107668474051123,
      "grad_norm": 3.1879615783691406,
      "learning_rate": 1.9623448274891686e-06,
      "loss": 0.0492,
      "step": 13764
    },
    {
      "epoch": 0.7108184869610121,
      "grad_norm": 3.347635507583618,
      "learning_rate": 1.96169405342994e-06,
      "loss": 0.0506,
      "step": 13765
    },
    {
      "epoch": 0.710870126516912,
      "grad_norm": 9.168074607849121,
      "learning_rate": 1.9610433609616087e-06,
      "loss": 0.0739,
      "step": 13766
    },
    {
      "epoch": 0.7109217660728118,
      "grad_norm": 1.7047358751296997,
      "learning_rate": 1.9603927501016516e-06,
      "loss": 0.0337,
      "step": 13767
    },
    {
      "epoch": 0.7109734056287116,
      "grad_norm": 5.7313551902771,
      "learning_rate": 1.9597422208675336e-06,
      "loss": 0.0476,
      "step": 13768
    },
    {
      "epoch": 0.7110250451846114,
      "grad_norm": 9.83671760559082,
      "learning_rate": 1.9590917732767313e-06,
      "loss": 0.141,
      "step": 13769
    },
    {
      "epoch": 0.7110766847405112,
      "grad_norm": 1.3245553970336914,
      "learning_rate": 1.9584414073467077e-06,
      "loss": 0.0098,
      "step": 13770
    },
    {
      "epoch": 0.711128324296411,
      "grad_norm": 2.8511955738067627,
      "learning_rate": 1.9577911230949314e-06,
      "loss": 0.0556,
      "step": 13771
    },
    {
      "epoch": 0.7111799638523109,
      "grad_norm": 1.2601901292800903,
      "learning_rate": 1.9571409205388613e-06,
      "loss": 0.0209,
      "step": 13772
    },
    {
      "epoch": 0.7112316034082107,
      "grad_norm": 1.9393523931503296,
      "learning_rate": 1.9564907996959574e-06,
      "loss": 0.0299,
      "step": 13773
    },
    {
      "epoch": 0.7112832429641105,
      "grad_norm": 3.575983762741089,
      "learning_rate": 1.9558407605836837e-06,
      "loss": 0.0311,
      "step": 13774
    },
    {
      "epoch": 0.7113348825200103,
      "grad_norm": 7.630002975463867,
      "learning_rate": 1.955190803219491e-06,
      "loss": 0.0687,
      "step": 13775
    },
    {
      "epoch": 0.7113865220759101,
      "grad_norm": 4.769417762756348,
      "learning_rate": 1.9545409276208353e-06,
      "loss": 0.0507,
      "step": 13776
    },
    {
      "epoch": 0.71143816163181,
      "grad_norm": 5.682544708251953,
      "learning_rate": 1.9538911338051687e-06,
      "loss": 0.0569,
      "step": 13777
    },
    {
      "epoch": 0.7114898011877098,
      "grad_norm": 2.4052927494049072,
      "learning_rate": 1.95324142178994e-06,
      "loss": 0.0399,
      "step": 13778
    },
    {
      "epoch": 0.7115414407436096,
      "grad_norm": 1.614892601966858,
      "learning_rate": 1.9525917915925967e-06,
      "loss": 0.0159,
      "step": 13779
    },
    {
      "epoch": 0.7115930802995094,
      "grad_norm": 16.66523551940918,
      "learning_rate": 1.951942243230585e-06,
      "loss": 0.0898,
      "step": 13780
    },
    {
      "epoch": 0.7116447198554092,
      "grad_norm": 3.405778169631958,
      "learning_rate": 1.9512927767213467e-06,
      "loss": 0.0394,
      "step": 13781
    },
    {
      "epoch": 0.711696359411309,
      "grad_norm": 2.3304028511047363,
      "learning_rate": 1.950643392082324e-06,
      "loss": 0.0338,
      "step": 13782
    },
    {
      "epoch": 0.7117479989672089,
      "grad_norm": 2.2284789085388184,
      "learning_rate": 1.949994089330954e-06,
      "loss": 0.023,
      "step": 13783
    },
    {
      "epoch": 0.7117996385231087,
      "grad_norm": 5.726614952087402,
      "learning_rate": 1.9493448684846738e-06,
      "loss": 0.0645,
      "step": 13784
    },
    {
      "epoch": 0.7118512780790085,
      "grad_norm": 16.54052734375,
      "learning_rate": 1.9486957295609198e-06,
      "loss": 0.0582,
      "step": 13785
    },
    {
      "epoch": 0.7119029176349083,
      "grad_norm": 8.512991905212402,
      "learning_rate": 1.94804667257712e-06,
      "loss": 0.0719,
      "step": 13786
    },
    {
      "epoch": 0.7119545571908081,
      "grad_norm": 5.331706523895264,
      "learning_rate": 1.947397697550704e-06,
      "loss": 0.0526,
      "step": 13787
    },
    {
      "epoch": 0.712006196746708,
      "grad_norm": 11.14366626739502,
      "learning_rate": 1.946748804499104e-06,
      "loss": 0.1546,
      "step": 13788
    },
    {
      "epoch": 0.7120578363026078,
      "grad_norm": 9.787412643432617,
      "learning_rate": 1.9460999934397445e-06,
      "loss": 0.1025,
      "step": 13789
    },
    {
      "epoch": 0.7121094758585076,
      "grad_norm": 7.007328510284424,
      "learning_rate": 1.945451264390045e-06,
      "loss": 0.0539,
      "step": 13790
    },
    {
      "epoch": 0.7121611154144074,
      "grad_norm": 3.493576765060425,
      "learning_rate": 1.944802617367429e-06,
      "loss": 0.0601,
      "step": 13791
    },
    {
      "epoch": 0.7122127549703072,
      "grad_norm": 13.1928071975708,
      "learning_rate": 1.944154052389313e-06,
      "loss": 0.157,
      "step": 13792
    },
    {
      "epoch": 0.712264394526207,
      "grad_norm": 2.676164388656616,
      "learning_rate": 1.94350556947312e-06,
      "loss": 0.0445,
      "step": 13793
    },
    {
      "epoch": 0.7123160340821069,
      "grad_norm": 3.928980588912964,
      "learning_rate": 1.9428571686362574e-06,
      "loss": 0.0667,
      "step": 13794
    },
    {
      "epoch": 0.7123676736380067,
      "grad_norm": 9.926896095275879,
      "learning_rate": 1.9422088498961406e-06,
      "loss": 0.0848,
      "step": 13795
    },
    {
      "epoch": 0.7124193131939065,
      "grad_norm": 2.3673112392425537,
      "learning_rate": 1.9415606132701784e-06,
      "loss": 0.038,
      "step": 13796
    },
    {
      "epoch": 0.7124709527498063,
      "grad_norm": 5.042214393615723,
      "learning_rate": 1.94091245877578e-06,
      "loss": 0.0939,
      "step": 13797
    },
    {
      "epoch": 0.7125225923057061,
      "grad_norm": 0.9184242486953735,
      "learning_rate": 1.9402643864303496e-06,
      "loss": 0.0135,
      "step": 13798
    },
    {
      "epoch": 0.712574231861606,
      "grad_norm": 3.1509759426116943,
      "learning_rate": 1.9396163962512916e-06,
      "loss": 0.047,
      "step": 13799
    },
    {
      "epoch": 0.7126258714175058,
      "grad_norm": 2.399247884750366,
      "learning_rate": 1.9389684882560085e-06,
      "loss": 0.0447,
      "step": 13800
    },
    {
      "epoch": 0.7126775109734056,
      "grad_norm": 8.774128913879395,
      "learning_rate": 1.9383206624618938e-06,
      "loss": 0.162,
      "step": 13801
    },
    {
      "epoch": 0.7127291505293054,
      "grad_norm": 2.7784361839294434,
      "learning_rate": 1.9376729188863496e-06,
      "loss": 0.0346,
      "step": 13802
    },
    {
      "epoch": 0.7127807900852052,
      "grad_norm": 3.5665359497070312,
      "learning_rate": 1.93702525754677e-06,
      "loss": 0.0391,
      "step": 13803
    },
    {
      "epoch": 0.712832429641105,
      "grad_norm": 2.0509374141693115,
      "learning_rate": 1.936377678460547e-06,
      "loss": 0.0449,
      "step": 13804
    },
    {
      "epoch": 0.7128840691970049,
      "grad_norm": 9.793618202209473,
      "learning_rate": 1.935730181645068e-06,
      "loss": 0.1486,
      "step": 13805
    },
    {
      "epoch": 0.7129357087529047,
      "grad_norm": 3.2710893154144287,
      "learning_rate": 1.935082767117722e-06,
      "loss": 0.051,
      "step": 13806
    },
    {
      "epoch": 0.7129873483088045,
      "grad_norm": 7.876561164855957,
      "learning_rate": 1.9344354348958983e-06,
      "loss": 0.1525,
      "step": 13807
    },
    {
      "epoch": 0.7130389878647043,
      "grad_norm": 5.998064994812012,
      "learning_rate": 1.9337881849969763e-06,
      "loss": 0.0317,
      "step": 13808
    },
    {
      "epoch": 0.7130906274206041,
      "grad_norm": 32.1985969543457,
      "learning_rate": 1.9331410174383393e-06,
      "loss": 0.2379,
      "step": 13809
    },
    {
      "epoch": 0.713142266976504,
      "grad_norm": 7.188408851623535,
      "learning_rate": 1.9324939322373646e-06,
      "loss": 0.0915,
      "step": 13810
    },
    {
      "epoch": 0.7131939065324038,
      "grad_norm": 4.563446998596191,
      "learning_rate": 1.931846929411431e-06,
      "loss": 0.05,
      "step": 13811
    },
    {
      "epoch": 0.7132455460883036,
      "grad_norm": 3.831027030944824,
      "learning_rate": 1.9312000089779125e-06,
      "loss": 0.0889,
      "step": 13812
    },
    {
      "epoch": 0.7132971856442034,
      "grad_norm": 13.753164291381836,
      "learning_rate": 1.930553170954182e-06,
      "loss": 0.1544,
      "step": 13813
    },
    {
      "epoch": 0.7133488252001032,
      "grad_norm": 17.307498931884766,
      "learning_rate": 1.9299064153576084e-06,
      "loss": 0.0872,
      "step": 13814
    },
    {
      "epoch": 0.713400464756003,
      "grad_norm": 8.716706275939941,
      "learning_rate": 1.929259742205561e-06,
      "loss": 0.1692,
      "step": 13815
    },
    {
      "epoch": 0.7134521043119029,
      "grad_norm": 2.2990777492523193,
      "learning_rate": 1.928613151515405e-06,
      "loss": 0.0447,
      "step": 13816
    },
    {
      "epoch": 0.7135037438678027,
      "grad_norm": 18.806758880615234,
      "learning_rate": 1.927966643304504e-06,
      "loss": 0.218,
      "step": 13817
    },
    {
      "epoch": 0.7135553834237025,
      "grad_norm": 2.584820032119751,
      "learning_rate": 1.927320217590221e-06,
      "loss": 0.0435,
      "step": 13818
    },
    {
      "epoch": 0.7136070229796023,
      "grad_norm": 4.039050102233887,
      "learning_rate": 1.9266738743899126e-06,
      "loss": 0.0251,
      "step": 13819
    },
    {
      "epoch": 0.7136586625355021,
      "grad_norm": 2.888599395751953,
      "learning_rate": 1.926027613720935e-06,
      "loss": 0.0391,
      "step": 13820
    },
    {
      "epoch": 0.713710302091402,
      "grad_norm": 1.3722590208053589,
      "learning_rate": 1.9253814356006462e-06,
      "loss": 0.0249,
      "step": 13821
    },
    {
      "epoch": 0.7137619416473018,
      "grad_norm": 5.012323379516602,
      "learning_rate": 1.9247353400464e-06,
      "loss": 0.0767,
      "step": 13822
    },
    {
      "epoch": 0.7138135812032016,
      "grad_norm": 2.210949182510376,
      "learning_rate": 1.9240893270755412e-06,
      "loss": 0.0512,
      "step": 13823
    },
    {
      "epoch": 0.7138652207591015,
      "grad_norm": 1.7190321683883667,
      "learning_rate": 1.9234433967054213e-06,
      "loss": 0.0306,
      "step": 13824
    },
    {
      "epoch": 0.7139168603150013,
      "grad_norm": 2.313260078430176,
      "learning_rate": 1.922797548953383e-06,
      "loss": 0.0287,
      "step": 13825
    },
    {
      "epoch": 0.7139684998709012,
      "grad_norm": 4.712308883666992,
      "learning_rate": 1.9221517838367763e-06,
      "loss": 0.0367,
      "step": 13826
    },
    {
      "epoch": 0.714020139426801,
      "grad_norm": 7.898391246795654,
      "learning_rate": 1.9215061013729373e-06,
      "loss": 0.0287,
      "step": 13827
    },
    {
      "epoch": 0.7140717789827008,
      "grad_norm": 4.020408630371094,
      "learning_rate": 1.920860501579207e-06,
      "loss": 0.0587,
      "step": 13828
    },
    {
      "epoch": 0.7141234185386006,
      "grad_norm": 3.273886203765869,
      "learning_rate": 1.9202149844729218e-06,
      "loss": 0.058,
      "step": 13829
    },
    {
      "epoch": 0.7141750580945004,
      "grad_norm": 1.5587974786758423,
      "learning_rate": 1.919569550071417e-06,
      "loss": 0.032,
      "step": 13830
    },
    {
      "epoch": 0.7142266976504003,
      "grad_norm": 9.67599105834961,
      "learning_rate": 1.918924198392025e-06,
      "loss": 0.1724,
      "step": 13831
    },
    {
      "epoch": 0.7142783372063001,
      "grad_norm": 2.9304347038269043,
      "learning_rate": 1.9182789294520755e-06,
      "loss": 0.0645,
      "step": 13832
    },
    {
      "epoch": 0.7143299767621999,
      "grad_norm": 3.2245430946350098,
      "learning_rate": 1.9176337432688993e-06,
      "loss": 0.0362,
      "step": 13833
    },
    {
      "epoch": 0.7143816163180997,
      "grad_norm": 6.0686798095703125,
      "learning_rate": 1.9169886398598164e-06,
      "loss": 0.0731,
      "step": 13834
    },
    {
      "epoch": 0.7144332558739995,
      "grad_norm": 16.867630004882812,
      "learning_rate": 1.916343619242156e-06,
      "loss": 0.141,
      "step": 13835
    },
    {
      "epoch": 0.7144848954298993,
      "grad_norm": 3.48079252243042,
      "learning_rate": 1.9156986814332374e-06,
      "loss": 0.0972,
      "step": 13836
    },
    {
      "epoch": 0.7145365349857992,
      "grad_norm": 3.1285316944122314,
      "learning_rate": 1.915053826450382e-06,
      "loss": 0.057,
      "step": 13837
    },
    {
      "epoch": 0.714588174541699,
      "grad_norm": 3.9746317863464355,
      "learning_rate": 1.9144090543109033e-06,
      "loss": 0.0594,
      "step": 13838
    },
    {
      "epoch": 0.7146398140975988,
      "grad_norm": 2.2081923484802246,
      "learning_rate": 1.9137643650321165e-06,
      "loss": 0.0303,
      "step": 13839
    },
    {
      "epoch": 0.7146914536534986,
      "grad_norm": 7.330829620361328,
      "learning_rate": 1.9131197586313384e-06,
      "loss": 0.0426,
      "step": 13840
    },
    {
      "epoch": 0.7147430932093984,
      "grad_norm": 1.190846562385559,
      "learning_rate": 1.912475235125874e-06,
      "loss": 0.0256,
      "step": 13841
    },
    {
      "epoch": 0.7147947327652983,
      "grad_norm": 1.6923589706420898,
      "learning_rate": 1.9118307945330338e-06,
      "loss": 0.0307,
      "step": 13842
    },
    {
      "epoch": 0.7148463723211981,
      "grad_norm": 2.3140971660614014,
      "learning_rate": 1.911186436870124e-06,
      "loss": 0.03,
      "step": 13843
    },
    {
      "epoch": 0.7148980118770979,
      "grad_norm": 11.328039169311523,
      "learning_rate": 1.9105421621544467e-06,
      "loss": 0.1859,
      "step": 13844
    },
    {
      "epoch": 0.7149496514329977,
      "grad_norm": 2.6405298709869385,
      "learning_rate": 1.9098979704033047e-06,
      "loss": 0.0369,
      "step": 13845
    },
    {
      "epoch": 0.7150012909888975,
      "grad_norm": 7.929543972015381,
      "learning_rate": 1.9092538616339973e-06,
      "loss": 0.0925,
      "step": 13846
    },
    {
      "epoch": 0.7150529305447973,
      "grad_norm": 4.447042942047119,
      "learning_rate": 1.9086098358638207e-06,
      "loss": 0.0428,
      "step": 13847
    },
    {
      "epoch": 0.7151045701006972,
      "grad_norm": 12.903636932373047,
      "learning_rate": 1.907965893110069e-06,
      "loss": 0.1023,
      "step": 13848
    },
    {
      "epoch": 0.715156209656597,
      "grad_norm": 2.9849658012390137,
      "learning_rate": 1.9073220333900361e-06,
      "loss": 0.0408,
      "step": 13849
    },
    {
      "epoch": 0.7152078492124968,
      "grad_norm": 2.591735601425171,
      "learning_rate": 1.9066782567210118e-06,
      "loss": 0.0414,
      "step": 13850
    },
    {
      "epoch": 0.7152594887683966,
      "grad_norm": 2.368577718734741,
      "learning_rate": 1.9060345631202858e-06,
      "loss": 0.0531,
      "step": 13851
    },
    {
      "epoch": 0.7153111283242964,
      "grad_norm": 3.9456660747528076,
      "learning_rate": 1.90539095260514e-06,
      "loss": 0.0489,
      "step": 13852
    },
    {
      "epoch": 0.7153627678801963,
      "grad_norm": 3.078747272491455,
      "learning_rate": 1.9047474251928583e-06,
      "loss": 0.0727,
      "step": 13853
    },
    {
      "epoch": 0.7154144074360961,
      "grad_norm": 2.241201400756836,
      "learning_rate": 1.9041039809007257e-06,
      "loss": 0.0202,
      "step": 13854
    },
    {
      "epoch": 0.7154660469919959,
      "grad_norm": 4.9263153076171875,
      "learning_rate": 1.9034606197460216e-06,
      "loss": 0.0303,
      "step": 13855
    },
    {
      "epoch": 0.7155176865478957,
      "grad_norm": 3.1221907138824463,
      "learning_rate": 1.9028173417460176e-06,
      "loss": 0.0498,
      "step": 13856
    },
    {
      "epoch": 0.7155693261037955,
      "grad_norm": 2.4636476039886475,
      "learning_rate": 1.9021741469179927e-06,
      "loss": 0.0384,
      "step": 13857
    },
    {
      "epoch": 0.7156209656596954,
      "grad_norm": 21.74239730834961,
      "learning_rate": 1.9015310352792155e-06,
      "loss": 0.0911,
      "step": 13858
    },
    {
      "epoch": 0.7156726052155952,
      "grad_norm": 16.627174377441406,
      "learning_rate": 1.9008880068469626e-06,
      "loss": 0.127,
      "step": 13859
    },
    {
      "epoch": 0.715724244771495,
      "grad_norm": 2.5555999279022217,
      "learning_rate": 1.9002450616384959e-06,
      "loss": 0.0391,
      "step": 13860
    },
    {
      "epoch": 0.7157758843273948,
      "grad_norm": 2.652273654937744,
      "learning_rate": 1.8996021996710835e-06,
      "loss": 0.0424,
      "step": 13861
    },
    {
      "epoch": 0.7158275238832946,
      "grad_norm": 7.840039253234863,
      "learning_rate": 1.8989594209619884e-06,
      "loss": 0.0595,
      "step": 13862
    },
    {
      "epoch": 0.7158791634391944,
      "grad_norm": 18.505874633789062,
      "learning_rate": 1.8983167255284718e-06,
      "loss": 0.0916,
      "step": 13863
    },
    {
      "epoch": 0.7159308029950943,
      "grad_norm": 2.7715978622436523,
      "learning_rate": 1.8976741133877936e-06,
      "loss": 0.044,
      "step": 13864
    },
    {
      "epoch": 0.7159824425509941,
      "grad_norm": 2.440866708755493,
      "learning_rate": 1.897031584557209e-06,
      "loss": 0.0201,
      "step": 13865
    },
    {
      "epoch": 0.7160340821068939,
      "grad_norm": 42.845088958740234,
      "learning_rate": 1.8963891390539751e-06,
      "loss": 0.1207,
      "step": 13866
    },
    {
      "epoch": 0.7160857216627937,
      "grad_norm": 5.633686542510986,
      "learning_rate": 1.8957467768953396e-06,
      "loss": 0.0454,
      "step": 13867
    },
    {
      "epoch": 0.7161373612186935,
      "grad_norm": 4.272457599639893,
      "learning_rate": 1.8951044980985566e-06,
      "loss": 0.0402,
      "step": 13868
    },
    {
      "epoch": 0.7161890007745934,
      "grad_norm": 3.9831690788269043,
      "learning_rate": 1.8944623026808746e-06,
      "loss": 0.0718,
      "step": 13869
    },
    {
      "epoch": 0.7162406403304932,
      "grad_norm": 17.741600036621094,
      "learning_rate": 1.8938201906595356e-06,
      "loss": 0.0912,
      "step": 13870
    },
    {
      "epoch": 0.716292279886393,
      "grad_norm": 8.897831916809082,
      "learning_rate": 1.8931781620517848e-06,
      "loss": 0.0746,
      "step": 13871
    },
    {
      "epoch": 0.7163439194422928,
      "grad_norm": 2.540708541870117,
      "learning_rate": 1.8925362168748607e-06,
      "loss": 0.0412,
      "step": 13872
    },
    {
      "epoch": 0.7163955589981926,
      "grad_norm": 3.5334603786468506,
      "learning_rate": 1.8918943551460084e-06,
      "loss": 0.0287,
      "step": 13873
    },
    {
      "epoch": 0.7164471985540924,
      "grad_norm": 1.5109165906906128,
      "learning_rate": 1.8912525768824585e-06,
      "loss": 0.0172,
      "step": 13874
    },
    {
      "epoch": 0.7164988381099923,
      "grad_norm": 2.0948596000671387,
      "learning_rate": 1.890610882101448e-06,
      "loss": 0.0427,
      "step": 13875
    },
    {
      "epoch": 0.7165504776658921,
      "grad_norm": 5.515292644500732,
      "learning_rate": 1.8899692708202077e-06,
      "loss": 0.0619,
      "step": 13876
    },
    {
      "epoch": 0.7166021172217919,
      "grad_norm": 4.318607330322266,
      "learning_rate": 1.8893277430559692e-06,
      "loss": 0.0785,
      "step": 13877
    },
    {
      "epoch": 0.7166537567776917,
      "grad_norm": 8.290339469909668,
      "learning_rate": 1.8886862988259586e-06,
      "loss": 0.0893,
      "step": 13878
    },
    {
      "epoch": 0.7167053963335915,
      "grad_norm": 6.8754754066467285,
      "learning_rate": 1.8880449381474019e-06,
      "loss": 0.0422,
      "step": 13879
    },
    {
      "epoch": 0.7167570358894914,
      "grad_norm": 3.3310844898223877,
      "learning_rate": 1.887403661037524e-06,
      "loss": 0.0773,
      "step": 13880
    },
    {
      "epoch": 0.7168086754453912,
      "grad_norm": 3.6336376667022705,
      "learning_rate": 1.886762467513541e-06,
      "loss": 0.0363,
      "step": 13881
    },
    {
      "epoch": 0.716860315001291,
      "grad_norm": 3.1410586833953857,
      "learning_rate": 1.8861213575926756e-06,
      "loss": 0.0517,
      "step": 13882
    },
    {
      "epoch": 0.7169119545571908,
      "grad_norm": 12.240683555603027,
      "learning_rate": 1.8854803312921439e-06,
      "loss": 0.0406,
      "step": 13883
    },
    {
      "epoch": 0.7169635941130906,
      "grad_norm": 17.336793899536133,
      "learning_rate": 1.8848393886291611e-06,
      "loss": 0.0869,
      "step": 13884
    },
    {
      "epoch": 0.7170152336689904,
      "grad_norm": 10.18488883972168,
      "learning_rate": 1.8841985296209352e-06,
      "loss": 0.0788,
      "step": 13885
    },
    {
      "epoch": 0.7170668732248903,
      "grad_norm": 4.2667107582092285,
      "learning_rate": 1.8835577542846767e-06,
      "loss": 0.0459,
      "step": 13886
    },
    {
      "epoch": 0.7171185127807901,
      "grad_norm": 2.0314245223999023,
      "learning_rate": 1.8829170626375959e-06,
      "loss": 0.0442,
      "step": 13887
    },
    {
      "epoch": 0.7171701523366899,
      "grad_norm": 3.214254379272461,
      "learning_rate": 1.8822764546968986e-06,
      "loss": 0.041,
      "step": 13888
    },
    {
      "epoch": 0.7172217918925897,
      "grad_norm": 2.7439804077148438,
      "learning_rate": 1.8816359304797832e-06,
      "loss": 0.0602,
      "step": 13889
    },
    {
      "epoch": 0.7172734314484895,
      "grad_norm": 2.665957450866699,
      "learning_rate": 1.8809954900034534e-06,
      "loss": 0.0207,
      "step": 13890
    },
    {
      "epoch": 0.7173250710043894,
      "grad_norm": 1.8874784708023071,
      "learning_rate": 1.8803551332851066e-06,
      "loss": 0.0305,
      "step": 13891
    },
    {
      "epoch": 0.7173767105602892,
      "grad_norm": 3.7174835205078125,
      "learning_rate": 1.87971486034194e-06,
      "loss": 0.0362,
      "step": 13892
    },
    {
      "epoch": 0.717428350116189,
      "grad_norm": 2.842590570449829,
      "learning_rate": 1.8790746711911468e-06,
      "loss": 0.0399,
      "step": 13893
    },
    {
      "epoch": 0.7174799896720888,
      "grad_norm": 3.4884839057922363,
      "learning_rate": 1.8784345658499186e-06,
      "loss": 0.0376,
      "step": 13894
    },
    {
      "epoch": 0.7175316292279886,
      "grad_norm": 2.498598098754883,
      "learning_rate": 1.8777945443354455e-06,
      "loss": 0.0408,
      "step": 13895
    },
    {
      "epoch": 0.7175832687838885,
      "grad_norm": 5.955401420593262,
      "learning_rate": 1.8771546066649144e-06,
      "loss": 0.053,
      "step": 13896
    },
    {
      "epoch": 0.7176349083397883,
      "grad_norm": 5.966044902801514,
      "learning_rate": 1.8765147528555095e-06,
      "loss": 0.0623,
      "step": 13897
    },
    {
      "epoch": 0.7176865478956881,
      "grad_norm": 10.193347930908203,
      "learning_rate": 1.8758749829244145e-06,
      "loss": 0.0842,
      "step": 13898
    },
    {
      "epoch": 0.7177381874515879,
      "grad_norm": 2.355532646179199,
      "learning_rate": 1.875235296888811e-06,
      "loss": 0.026,
      "step": 13899
    },
    {
      "epoch": 0.7177898270074877,
      "grad_norm": 2.959784746170044,
      "learning_rate": 1.8745956947658722e-06,
      "loss": 0.0541,
      "step": 13900
    },
    {
      "epoch": 0.7178414665633875,
      "grad_norm": 2.923488140106201,
      "learning_rate": 1.8739561765727794e-06,
      "loss": 0.0341,
      "step": 13901
    },
    {
      "epoch": 0.7178931061192874,
      "grad_norm": 3.581263780593872,
      "learning_rate": 1.8733167423267058e-06,
      "loss": 0.0453,
      "step": 13902
    },
    {
      "epoch": 0.7179447456751872,
      "grad_norm": 4.620807647705078,
      "learning_rate": 1.8726773920448193e-06,
      "loss": 0.0369,
      "step": 13903
    },
    {
      "epoch": 0.717996385231087,
      "grad_norm": 3.847121477127075,
      "learning_rate": 1.8720381257442916e-06,
      "loss": 0.0491,
      "step": 13904
    },
    {
      "epoch": 0.7180480247869868,
      "grad_norm": 3.9189391136169434,
      "learning_rate": 1.8713989434422874e-06,
      "loss": 0.0513,
      "step": 13905
    },
    {
      "epoch": 0.7180996643428866,
      "grad_norm": 3.831117630004883,
      "learning_rate": 1.8707598451559761e-06,
      "loss": 0.0489,
      "step": 13906
    },
    {
      "epoch": 0.7181513038987865,
      "grad_norm": 13.597660064697266,
      "learning_rate": 1.8701208309025154e-06,
      "loss": 0.1193,
      "step": 13907
    },
    {
      "epoch": 0.7182029434546863,
      "grad_norm": 2.2481982707977295,
      "learning_rate": 1.8694819006990671e-06,
      "loss": 0.0327,
      "step": 13908
    },
    {
      "epoch": 0.7182545830105861,
      "grad_norm": 2.9039969444274902,
      "learning_rate": 1.8688430545627896e-06,
      "loss": 0.0277,
      "step": 13909
    },
    {
      "epoch": 0.7183062225664859,
      "grad_norm": 7.8444085121154785,
      "learning_rate": 1.8682042925108374e-06,
      "loss": 0.0579,
      "step": 13910
    },
    {
      "epoch": 0.7183578621223857,
      "grad_norm": 5.740543842315674,
      "learning_rate": 1.8675656145603643e-06,
      "loss": 0.0445,
      "step": 13911
    },
    {
      "epoch": 0.7184095016782855,
      "grad_norm": 16.751081466674805,
      "learning_rate": 1.8669270207285216e-06,
      "loss": 0.1239,
      "step": 13912
    },
    {
      "epoch": 0.7184611412341854,
      "grad_norm": 11.383505821228027,
      "learning_rate": 1.8662885110324585e-06,
      "loss": 0.1193,
      "step": 13913
    },
    {
      "epoch": 0.7185127807900852,
      "grad_norm": 11.925025939941406,
      "learning_rate": 1.865650085489321e-06,
      "loss": 0.0592,
      "step": 13914
    },
    {
      "epoch": 0.718564420345985,
      "grad_norm": 2.1472818851470947,
      "learning_rate": 1.8650117441162535e-06,
      "loss": 0.0433,
      "step": 13915
    },
    {
      "epoch": 0.7186160599018848,
      "grad_norm": 4.560026168823242,
      "learning_rate": 1.8643734869303982e-06,
      "loss": 0.0366,
      "step": 13916
    },
    {
      "epoch": 0.7186676994577846,
      "grad_norm": 7.458511829376221,
      "learning_rate": 1.8637353139488966e-06,
      "loss": 0.0729,
      "step": 13917
    },
    {
      "epoch": 0.7187193390136845,
      "grad_norm": 7.612422943115234,
      "learning_rate": 1.8630972251888828e-06,
      "loss": 0.0485,
      "step": 13918
    },
    {
      "epoch": 0.7187709785695843,
      "grad_norm": 2.4956254959106445,
      "learning_rate": 1.8624592206674925e-06,
      "loss": 0.0279,
      "step": 13919
    },
    {
      "epoch": 0.7188226181254841,
      "grad_norm": 3.7073545455932617,
      "learning_rate": 1.861821300401861e-06,
      "loss": 0.0375,
      "step": 13920
    },
    {
      "epoch": 0.7188742576813839,
      "grad_norm": 14.03593921661377,
      "learning_rate": 1.8611834644091203e-06,
      "loss": 0.2666,
      "step": 13921
    },
    {
      "epoch": 0.7189258972372837,
      "grad_norm": 5.577965259552002,
      "learning_rate": 1.8605457127063948e-06,
      "loss": 0.0397,
      "step": 13922
    },
    {
      "epoch": 0.7189775367931835,
      "grad_norm": 2.163569211959839,
      "learning_rate": 1.8599080453108126e-06,
      "loss": 0.0245,
      "step": 13923
    },
    {
      "epoch": 0.7190291763490834,
      "grad_norm": 16.631263732910156,
      "learning_rate": 1.859270462239498e-06,
      "loss": 0.1058,
      "step": 13924
    },
    {
      "epoch": 0.7190808159049832,
      "grad_norm": 2.474665403366089,
      "learning_rate": 1.8586329635095724e-06,
      "loss": 0.037,
      "step": 13925
    },
    {
      "epoch": 0.719132455460883,
      "grad_norm": 7.33547306060791,
      "learning_rate": 1.8579955491381546e-06,
      "loss": 0.0556,
      "step": 13926
    },
    {
      "epoch": 0.7191840950167828,
      "grad_norm": 2.64503812789917,
      "learning_rate": 1.8573582191423629e-06,
      "loss": 0.0466,
      "step": 13927
    },
    {
      "epoch": 0.7192357345726826,
      "grad_norm": 5.762581825256348,
      "learning_rate": 1.8567209735393115e-06,
      "loss": 0.0525,
      "step": 13928
    },
    {
      "epoch": 0.7192873741285825,
      "grad_norm": 3.019021987915039,
      "learning_rate": 1.8560838123461138e-06,
      "loss": 0.0173,
      "step": 13929
    },
    {
      "epoch": 0.7193390136844823,
      "grad_norm": 6.021284580230713,
      "learning_rate": 1.8554467355798788e-06,
      "loss": 0.0485,
      "step": 13930
    },
    {
      "epoch": 0.7193906532403821,
      "grad_norm": 4.701417922973633,
      "learning_rate": 1.8548097432577162e-06,
      "loss": 0.0401,
      "step": 13931
    },
    {
      "epoch": 0.7194422927962819,
      "grad_norm": 9.768990516662598,
      "learning_rate": 1.854172835396731e-06,
      "loss": 0.0708,
      "step": 13932
    },
    {
      "epoch": 0.7194939323521817,
      "grad_norm": 2.5837934017181396,
      "learning_rate": 1.8535360120140261e-06,
      "loss": 0.0301,
      "step": 13933
    },
    {
      "epoch": 0.7195455719080815,
      "grad_norm": 5.851423740386963,
      "learning_rate": 1.8528992731267044e-06,
      "loss": 0.0742,
      "step": 13934
    },
    {
      "epoch": 0.7195972114639814,
      "grad_norm": 4.974386692047119,
      "learning_rate": 1.8522626187518656e-06,
      "loss": 0.0585,
      "step": 13935
    },
    {
      "epoch": 0.7196488510198812,
      "grad_norm": 2.599774122238159,
      "learning_rate": 1.851626048906603e-06,
      "loss": 0.0229,
      "step": 13936
    },
    {
      "epoch": 0.719700490575781,
      "grad_norm": 42.85066604614258,
      "learning_rate": 1.8509895636080128e-06,
      "loss": 0.1844,
      "step": 13937
    },
    {
      "epoch": 0.7197521301316808,
      "grad_norm": 12.466097831726074,
      "learning_rate": 1.8503531628731864e-06,
      "loss": 0.2922,
      "step": 13938
    },
    {
      "epoch": 0.7198037696875806,
      "grad_norm": 1.8314738273620605,
      "learning_rate": 1.8497168467192178e-06,
      "loss": 0.0309,
      "step": 13939
    },
    {
      "epoch": 0.7198554092434805,
      "grad_norm": 5.27121639251709,
      "learning_rate": 1.84908061516319e-06,
      "loss": 0.0426,
      "step": 13940
    },
    {
      "epoch": 0.7199070487993803,
      "grad_norm": 4.646339416503906,
      "learning_rate": 1.8484444682221902e-06,
      "loss": 0.0745,
      "step": 13941
    },
    {
      "epoch": 0.7199586883552801,
      "grad_norm": 3.0368881225585938,
      "learning_rate": 1.8478084059133016e-06,
      "loss": 0.0524,
      "step": 13942
    },
    {
      "epoch": 0.7200103279111799,
      "grad_norm": 3.8759145736694336,
      "learning_rate": 1.8471724282536053e-06,
      "loss": 0.0378,
      "step": 13943
    },
    {
      "epoch": 0.7200619674670797,
      "grad_norm": 2.984591245651245,
      "learning_rate": 1.8465365352601788e-06,
      "loss": 0.0514,
      "step": 13944
    },
    {
      "epoch": 0.7201136070229796,
      "grad_norm": 2.4282445907592773,
      "learning_rate": 1.8459007269500996e-06,
      "loss": 0.0324,
      "step": 13945
    },
    {
      "epoch": 0.7201652465788794,
      "grad_norm": 10.863446235656738,
      "learning_rate": 1.8452650033404407e-06,
      "loss": 0.2488,
      "step": 13946
    },
    {
      "epoch": 0.7202168861347792,
      "grad_norm": 2.3629469871520996,
      "learning_rate": 1.844629364448275e-06,
      "loss": 0.0238,
      "step": 13947
    },
    {
      "epoch": 0.720268525690679,
      "grad_norm": 4.640847682952881,
      "learning_rate": 1.8439938102906713e-06,
      "loss": 0.0299,
      "step": 13948
    },
    {
      "epoch": 0.7203201652465788,
      "grad_norm": 6.952390670776367,
      "learning_rate": 1.8433583408846968e-06,
      "loss": 0.0325,
      "step": 13949
    },
    {
      "epoch": 0.7203718048024786,
      "grad_norm": 5.580634117126465,
      "learning_rate": 1.8427229562474185e-06,
      "loss": 0.0937,
      "step": 13950
    },
    {
      "epoch": 0.7204234443583786,
      "grad_norm": 5.150204181671143,
      "learning_rate": 1.842087656395894e-06,
      "loss": 0.059,
      "step": 13951
    },
    {
      "epoch": 0.7204750839142784,
      "grad_norm": 11.32801342010498,
      "learning_rate": 1.8414524413471878e-06,
      "loss": 0.0581,
      "step": 13952
    },
    {
      "epoch": 0.7205267234701782,
      "grad_norm": 2.6283953189849854,
      "learning_rate": 1.840817311118358e-06,
      "loss": 0.014,
      "step": 13953
    },
    {
      "epoch": 0.720578363026078,
      "grad_norm": 3.8139243125915527,
      "learning_rate": 1.8401822657264607e-06,
      "loss": 0.0638,
      "step": 13954
    },
    {
      "epoch": 0.7206300025819778,
      "grad_norm": 8.154830932617188,
      "learning_rate": 1.8395473051885465e-06,
      "loss": 0.1148,
      "step": 13955
    },
    {
      "epoch": 0.7206816421378777,
      "grad_norm": 4.238436698913574,
      "learning_rate": 1.8389124295216687e-06,
      "loss": 0.0546,
      "step": 13956
    },
    {
      "epoch": 0.7207332816937775,
      "grad_norm": 11.445131301879883,
      "learning_rate": 1.8382776387428759e-06,
      "loss": 0.1694,
      "step": 13957
    },
    {
      "epoch": 0.7207849212496773,
      "grad_norm": 4.573535919189453,
      "learning_rate": 1.8376429328692153e-06,
      "loss": 0.0752,
      "step": 13958
    },
    {
      "epoch": 0.7208365608055771,
      "grad_norm": 3.7093842029571533,
      "learning_rate": 1.8370083119177307e-06,
      "loss": 0.0365,
      "step": 13959
    },
    {
      "epoch": 0.7208882003614769,
      "grad_norm": 4.602150917053223,
      "learning_rate": 1.8363737759054644e-06,
      "loss": 0.0582,
      "step": 13960
    },
    {
      "epoch": 0.7209398399173768,
      "grad_norm": 2.6053524017333984,
      "learning_rate": 1.8357393248494565e-06,
      "loss": 0.0342,
      "step": 13961
    },
    {
      "epoch": 0.7209914794732766,
      "grad_norm": 3.371497869491577,
      "learning_rate": 1.8351049587667446e-06,
      "loss": 0.0564,
      "step": 13962
    },
    {
      "epoch": 0.7210431190291764,
      "grad_norm": 14.320452690124512,
      "learning_rate": 1.8344706776743642e-06,
      "loss": 0.1282,
      "step": 13963
    },
    {
      "epoch": 0.7210947585850762,
      "grad_norm": 2.382490396499634,
      "learning_rate": 1.8338364815893477e-06,
      "loss": 0.0334,
      "step": 13964
    },
    {
      "epoch": 0.721146398140976,
      "grad_norm": 2.4801418781280518,
      "learning_rate": 1.8332023705287266e-06,
      "loss": 0.0373,
      "step": 13965
    },
    {
      "epoch": 0.7211980376968758,
      "grad_norm": 2.9835832118988037,
      "learning_rate": 1.8325683445095288e-06,
      "loss": 0.0563,
      "step": 13966
    },
    {
      "epoch": 0.7212496772527757,
      "grad_norm": 3.549744129180908,
      "learning_rate": 1.8319344035487807e-06,
      "loss": 0.0327,
      "step": 13967
    },
    {
      "epoch": 0.7213013168086755,
      "grad_norm": 3.750218629837036,
      "learning_rate": 1.8313005476635081e-06,
      "loss": 0.0582,
      "step": 13968
    },
    {
      "epoch": 0.7213529563645753,
      "grad_norm": 2.539271116256714,
      "learning_rate": 1.8306667768707288e-06,
      "loss": 0.044,
      "step": 13969
    },
    {
      "epoch": 0.7214045959204751,
      "grad_norm": 3.523013114929199,
      "learning_rate": 1.830033091187462e-06,
      "loss": 0.0701,
      "step": 13970
    },
    {
      "epoch": 0.7214562354763749,
      "grad_norm": 9.019218444824219,
      "learning_rate": 1.829399490630729e-06,
      "loss": 0.1524,
      "step": 13971
    },
    {
      "epoch": 0.7215078750322748,
      "grad_norm": 4.215445041656494,
      "learning_rate": 1.828765975217544e-06,
      "loss": 0.0262,
      "step": 13972
    },
    {
      "epoch": 0.7215595145881746,
      "grad_norm": 3.1655077934265137,
      "learning_rate": 1.8281325449649156e-06,
      "loss": 0.0286,
      "step": 13973
    },
    {
      "epoch": 0.7216111541440744,
      "grad_norm": 6.610101699829102,
      "learning_rate": 1.8274991998898566e-06,
      "loss": 0.0814,
      "step": 13974
    },
    {
      "epoch": 0.7216627936999742,
      "grad_norm": 1.2940365076065063,
      "learning_rate": 1.8268659400093747e-06,
      "loss": 0.0211,
      "step": 13975
    },
    {
      "epoch": 0.721714433255874,
      "grad_norm": 2.2039361000061035,
      "learning_rate": 1.826232765340475e-06,
      "loss": 0.0346,
      "step": 13976
    },
    {
      "epoch": 0.7217660728117739,
      "grad_norm": 3.4548544883728027,
      "learning_rate": 1.8255996759001609e-06,
      "loss": 0.0798,
      "step": 13977
    },
    {
      "epoch": 0.7218177123676737,
      "grad_norm": 5.0110015869140625,
      "learning_rate": 1.8249666717054338e-06,
      "loss": 0.0591,
      "step": 13978
    },
    {
      "epoch": 0.7218693519235735,
      "grad_norm": 28.092449188232422,
      "learning_rate": 1.8243337527732925e-06,
      "loss": 0.2156,
      "step": 13979
    },
    {
      "epoch": 0.7219209914794733,
      "grad_norm": 5.906278610229492,
      "learning_rate": 1.8237009191207334e-06,
      "loss": 0.0703,
      "step": 13980
    },
    {
      "epoch": 0.7219726310353731,
      "grad_norm": 8.848555564880371,
      "learning_rate": 1.8230681707647495e-06,
      "loss": 0.0649,
      "step": 13981
    },
    {
      "epoch": 0.7220242705912729,
      "grad_norm": 13.5985746383667,
      "learning_rate": 1.822435507722335e-06,
      "loss": 0.0385,
      "step": 13982
    },
    {
      "epoch": 0.7220759101471728,
      "grad_norm": 2.405172348022461,
      "learning_rate": 1.821802930010479e-06,
      "loss": 0.0358,
      "step": 13983
    },
    {
      "epoch": 0.7221275497030726,
      "grad_norm": 3.230987787246704,
      "learning_rate": 1.8211704376461648e-06,
      "loss": 0.0428,
      "step": 13984
    },
    {
      "epoch": 0.7221791892589724,
      "grad_norm": 7.015964508056641,
      "learning_rate": 1.8205380306463821e-06,
      "loss": 0.0612,
      "step": 13985
    },
    {
      "epoch": 0.7222308288148722,
      "grad_norm": 3.885209798812866,
      "learning_rate": 1.819905709028112e-06,
      "loss": 0.0558,
      "step": 13986
    },
    {
      "epoch": 0.722282468370772,
      "grad_norm": 5.700318813323975,
      "learning_rate": 1.8192734728083373e-06,
      "loss": 0.0802,
      "step": 13987
    },
    {
      "epoch": 0.7223341079266719,
      "grad_norm": 9.102925300598145,
      "learning_rate": 1.818641322004031e-06,
      "loss": 0.0545,
      "step": 13988
    },
    {
      "epoch": 0.7223857474825717,
      "grad_norm": 2.4967477321624756,
      "learning_rate": 1.8180092566321728e-06,
      "loss": 0.0448,
      "step": 13989
    },
    {
      "epoch": 0.7224373870384715,
      "grad_norm": 2.9296646118164062,
      "learning_rate": 1.817377276709734e-06,
      "loss": 0.0511,
      "step": 13990
    },
    {
      "epoch": 0.7224890265943713,
      "grad_norm": 11.563207626342773,
      "learning_rate": 1.8167453822536878e-06,
      "loss": 0.082,
      "step": 13991
    },
    {
      "epoch": 0.7225406661502711,
      "grad_norm": 3.11039662361145,
      "learning_rate": 1.816113573281002e-06,
      "loss": 0.0426,
      "step": 13992
    },
    {
      "epoch": 0.722592305706171,
      "grad_norm": 16.20245361328125,
      "learning_rate": 1.815481849808644e-06,
      "loss": 0.0881,
      "step": 13993
    },
    {
      "epoch": 0.7226439452620708,
      "grad_norm": 4.6614990234375,
      "learning_rate": 1.8148502118535772e-06,
      "loss": 0.049,
      "step": 13994
    },
    {
      "epoch": 0.7226955848179706,
      "grad_norm": 4.748941421508789,
      "learning_rate": 1.8142186594327643e-06,
      "loss": 0.0704,
      "step": 13995
    },
    {
      "epoch": 0.7227472243738704,
      "grad_norm": 8.707247734069824,
      "learning_rate": 1.813587192563165e-06,
      "loss": 0.0854,
      "step": 13996
    },
    {
      "epoch": 0.7227988639297702,
      "grad_norm": 7.4380412101745605,
      "learning_rate": 1.8129558112617367e-06,
      "loss": 0.1088,
      "step": 13997
    },
    {
      "epoch": 0.72285050348567,
      "grad_norm": 6.518710613250732,
      "learning_rate": 1.8123245155454345e-06,
      "loss": 0.1297,
      "step": 13998
    },
    {
      "epoch": 0.7229021430415699,
      "grad_norm": 2.5324342250823975,
      "learning_rate": 1.8116933054312113e-06,
      "loss": 0.0362,
      "step": 13999
    },
    {
      "epoch": 0.7229537825974697,
      "grad_norm": 7.476965427398682,
      "learning_rate": 1.8110621809360173e-06,
      "loss": 0.1195,
      "step": 14000
    },
    {
      "epoch": 0.7230054221533695,
      "grad_norm": 2.0983400344848633,
      "learning_rate": 1.8104311420768034e-06,
      "loss": 0.0317,
      "step": 14001
    },
    {
      "epoch": 0.7230570617092693,
      "grad_norm": 4.784801483154297,
      "learning_rate": 1.8098001888705113e-06,
      "loss": 0.0555,
      "step": 14002
    },
    {
      "epoch": 0.7231087012651691,
      "grad_norm": 2.8162996768951416,
      "learning_rate": 1.8091693213340854e-06,
      "loss": 0.0423,
      "step": 14003
    },
    {
      "epoch": 0.723160340821069,
      "grad_norm": 9.012765884399414,
      "learning_rate": 1.8085385394844695e-06,
      "loss": 0.0638,
      "step": 14004
    },
    {
      "epoch": 0.7232119803769688,
      "grad_norm": 4.855153560638428,
      "learning_rate": 1.8079078433386032e-06,
      "loss": 0.0602,
      "step": 14005
    },
    {
      "epoch": 0.7232636199328686,
      "grad_norm": 7.250383377075195,
      "learning_rate": 1.8072772329134203e-06,
      "loss": 0.0955,
      "step": 14006
    },
    {
      "epoch": 0.7233152594887684,
      "grad_norm": 11.16620922088623,
      "learning_rate": 1.806646708225856e-06,
      "loss": 0.0361,
      "step": 14007
    },
    {
      "epoch": 0.7233668990446682,
      "grad_norm": 12.446487426757812,
      "learning_rate": 1.8060162692928434e-06,
      "loss": 0.1437,
      "step": 14008
    },
    {
      "epoch": 0.723418538600568,
      "grad_norm": 14.403817176818848,
      "learning_rate": 1.805385916131312e-06,
      "loss": 0.0869,
      "step": 14009
    },
    {
      "epoch": 0.7234701781564679,
      "grad_norm": 3.828765630722046,
      "learning_rate": 1.8047556487581886e-06,
      "loss": 0.0605,
      "step": 14010
    },
    {
      "epoch": 0.7235218177123677,
      "grad_norm": 2.665390729904175,
      "learning_rate": 1.8041254671903996e-06,
      "loss": 0.033,
      "step": 14011
    },
    {
      "epoch": 0.7235734572682675,
      "grad_norm": 5.889933109283447,
      "learning_rate": 1.8034953714448676e-06,
      "loss": 0.0496,
      "step": 14012
    },
    {
      "epoch": 0.7236250968241673,
      "grad_norm": 3.212021827697754,
      "learning_rate": 1.8028653615385128e-06,
      "loss": 0.0332,
      "step": 14013
    },
    {
      "epoch": 0.7236767363800671,
      "grad_norm": 2.841923475265503,
      "learning_rate": 1.802235437488254e-06,
      "loss": 0.047,
      "step": 14014
    },
    {
      "epoch": 0.723728375935967,
      "grad_norm": 1.8640680313110352,
      "learning_rate": 1.8016055993110071e-06,
      "loss": 0.0401,
      "step": 14015
    },
    {
      "epoch": 0.7237800154918668,
      "grad_norm": 2.5639238357543945,
      "learning_rate": 1.800975847023687e-06,
      "loss": 0.0261,
      "step": 14016
    },
    {
      "epoch": 0.7238316550477666,
      "grad_norm": 2.8969006538391113,
      "learning_rate": 1.8003461806432016e-06,
      "loss": 0.0589,
      "step": 14017
    },
    {
      "epoch": 0.7238832946036664,
      "grad_norm": 1.125810980796814,
      "learning_rate": 1.799716600186463e-06,
      "loss": 0.0162,
      "step": 14018
    },
    {
      "epoch": 0.7239349341595662,
      "grad_norm": 27.744060516357422,
      "learning_rate": 1.7990871056703779e-06,
      "loss": 0.1651,
      "step": 14019
    },
    {
      "epoch": 0.723986573715466,
      "grad_norm": 25.02839469909668,
      "learning_rate": 1.7984576971118517e-06,
      "loss": 0.0956,
      "step": 14020
    },
    {
      "epoch": 0.7240382132713659,
      "grad_norm": 4.318053245544434,
      "learning_rate": 1.797828374527783e-06,
      "loss": 0.0221,
      "step": 14021
    },
    {
      "epoch": 0.7240898528272657,
      "grad_norm": 12.29378890991211,
      "learning_rate": 1.7971991379350728e-06,
      "loss": 0.1696,
      "step": 14022
    },
    {
      "epoch": 0.7241414923831655,
      "grad_norm": 8.514301300048828,
      "learning_rate": 1.7965699873506227e-06,
      "loss": 0.0567,
      "step": 14023
    },
    {
      "epoch": 0.7241931319390653,
      "grad_norm": 4.174011707305908,
      "learning_rate": 1.7959409227913232e-06,
      "loss": 0.0584,
      "step": 14024
    },
    {
      "epoch": 0.7242447714949651,
      "grad_norm": 6.323703765869141,
      "learning_rate": 1.7953119442740684e-06,
      "loss": 0.0424,
      "step": 14025
    },
    {
      "epoch": 0.724296411050865,
      "grad_norm": 10.479530334472656,
      "learning_rate": 1.7946830518157505e-06,
      "loss": 0.0743,
      "step": 14026
    },
    {
      "epoch": 0.7243480506067648,
      "grad_norm": 4.2207465171813965,
      "learning_rate": 1.794054245433256e-06,
      "loss": 0.0556,
      "step": 14027
    },
    {
      "epoch": 0.7243996901626646,
      "grad_norm": 2.6585803031921387,
      "learning_rate": 1.7934255251434724e-06,
      "loss": 0.0464,
      "step": 14028
    },
    {
      "epoch": 0.7244513297185644,
      "grad_norm": 3.1065711975097656,
      "learning_rate": 1.792796890963282e-06,
      "loss": 0.0313,
      "step": 14029
    },
    {
      "epoch": 0.7245029692744642,
      "grad_norm": 1.7421600818634033,
      "learning_rate": 1.7921683429095676e-06,
      "loss": 0.0257,
      "step": 14030
    },
    {
      "epoch": 0.724554608830364,
      "grad_norm": 2.2207906246185303,
      "learning_rate": 1.7915398809992074e-06,
      "loss": 0.0293,
      "step": 14031
    },
    {
      "epoch": 0.7246062483862639,
      "grad_norm": 5.399433135986328,
      "learning_rate": 1.7909115052490783e-06,
      "loss": 0.0913,
      "step": 14032
    },
    {
      "epoch": 0.7246578879421637,
      "grad_norm": 4.535953521728516,
      "learning_rate": 1.7902832156760547e-06,
      "loss": 0.0327,
      "step": 14033
    },
    {
      "epoch": 0.7247095274980635,
      "grad_norm": 18.473384857177734,
      "learning_rate": 1.7896550122970103e-06,
      "loss": 0.1675,
      "step": 14034
    },
    {
      "epoch": 0.7247611670539633,
      "grad_norm": 2.9229161739349365,
      "learning_rate": 1.789026895128812e-06,
      "loss": 0.0546,
      "step": 14035
    },
    {
      "epoch": 0.7248128066098631,
      "grad_norm": 1.70307195186615,
      "learning_rate": 1.788398864188327e-06,
      "loss": 0.0288,
      "step": 14036
    },
    {
      "epoch": 0.724864446165763,
      "grad_norm": 1.5814588069915771,
      "learning_rate": 1.7877709194924242e-06,
      "loss": 0.0302,
      "step": 14037
    },
    {
      "epoch": 0.7249160857216628,
      "grad_norm": 2.274718999862671,
      "learning_rate": 1.7871430610579659e-06,
      "loss": 0.0274,
      "step": 14038
    },
    {
      "epoch": 0.7249677252775626,
      "grad_norm": 21.39413833618164,
      "learning_rate": 1.78651528890181e-06,
      "loss": 0.0571,
      "step": 14039
    },
    {
      "epoch": 0.7250193648334624,
      "grad_norm": 3.1050336360931396,
      "learning_rate": 1.7858876030408157e-06,
      "loss": 0.0676,
      "step": 14040
    },
    {
      "epoch": 0.7250710043893622,
      "grad_norm": 19.713159561157227,
      "learning_rate": 1.7852600034918377e-06,
      "loss": 0.1797,
      "step": 14041
    },
    {
      "epoch": 0.725122643945262,
      "grad_norm": 2.433621406555176,
      "learning_rate": 1.7846324902717349e-06,
      "loss": 0.031,
      "step": 14042
    },
    {
      "epoch": 0.7251742835011619,
      "grad_norm": 6.798183441162109,
      "learning_rate": 1.7840050633973527e-06,
      "loss": 0.0226,
      "step": 14043
    },
    {
      "epoch": 0.7252259230570617,
      "grad_norm": 4.557009220123291,
      "learning_rate": 1.783377722885542e-06,
      "loss": 0.0429,
      "step": 14044
    },
    {
      "epoch": 0.7252775626129615,
      "grad_norm": 9.988258361816406,
      "learning_rate": 1.7827504687531506e-06,
      "loss": 0.0503,
      "step": 14045
    },
    {
      "epoch": 0.7253292021688613,
      "grad_norm": 1.2819347381591797,
      "learning_rate": 1.782123301017021e-06,
      "loss": 0.0169,
      "step": 14046
    },
    {
      "epoch": 0.7253808417247611,
      "grad_norm": 4.038074016571045,
      "learning_rate": 1.781496219693996e-06,
      "loss": 0.0632,
      "step": 14047
    },
    {
      "epoch": 0.725432481280661,
      "grad_norm": 5.522790431976318,
      "learning_rate": 1.7808692248009164e-06,
      "loss": 0.0538,
      "step": 14048
    },
    {
      "epoch": 0.7254841208365608,
      "grad_norm": 1.8042149543762207,
      "learning_rate": 1.7802423163546196e-06,
      "loss": 0.0162,
      "step": 14049
    },
    {
      "epoch": 0.7255357603924606,
      "grad_norm": 3.75496506690979,
      "learning_rate": 1.7796154943719357e-06,
      "loss": 0.0548,
      "step": 14050
    },
    {
      "epoch": 0.7255873999483604,
      "grad_norm": 9.495460510253906,
      "learning_rate": 1.7789887588697041e-06,
      "loss": 0.0594,
      "step": 14051
    },
    {
      "epoch": 0.7256390395042602,
      "grad_norm": 3.522252082824707,
      "learning_rate": 1.778362109864752e-06,
      "loss": 0.0334,
      "step": 14052
    },
    {
      "epoch": 0.72569067906016,
      "grad_norm": 4.581921577453613,
      "learning_rate": 1.7777355473739094e-06,
      "loss": 0.0902,
      "step": 14053
    },
    {
      "epoch": 0.7257423186160599,
      "grad_norm": 0.7429677248001099,
      "learning_rate": 1.7771090714139989e-06,
      "loss": 0.0039,
      "step": 14054
    },
    {
      "epoch": 0.7257939581719597,
      "grad_norm": 4.059681415557861,
      "learning_rate": 1.776482682001844e-06,
      "loss": 0.0505,
      "step": 14055
    },
    {
      "epoch": 0.7258455977278595,
      "grad_norm": 3.567265272140503,
      "learning_rate": 1.775856379154271e-06,
      "loss": 0.0434,
      "step": 14056
    },
    {
      "epoch": 0.7258972372837593,
      "grad_norm": 2.082026243209839,
      "learning_rate": 1.775230162888093e-06,
      "loss": 0.0299,
      "step": 14057
    },
    {
      "epoch": 0.7259488768396591,
      "grad_norm": 5.111849784851074,
      "learning_rate": 1.7746040332201293e-06,
      "loss": 0.0523,
      "step": 14058
    },
    {
      "epoch": 0.726000516395559,
      "grad_norm": 3.2388663291931152,
      "learning_rate": 1.773977990167192e-06,
      "loss": 0.0312,
      "step": 14059
    },
    {
      "epoch": 0.7260521559514588,
      "grad_norm": 66.56781005859375,
      "learning_rate": 1.7733520337460952e-06,
      "loss": 0.0628,
      "step": 14060
    },
    {
      "epoch": 0.7261037955073586,
      "grad_norm": 1.5906610488891602,
      "learning_rate": 1.7727261639736476e-06,
      "loss": 0.0215,
      "step": 14061
    },
    {
      "epoch": 0.7261554350632584,
      "grad_norm": 3.4523959159851074,
      "learning_rate": 1.7721003808666554e-06,
      "loss": 0.0437,
      "step": 14062
    },
    {
      "epoch": 0.7262070746191582,
      "grad_norm": 10.42981243133545,
      "learning_rate": 1.7714746844419246e-06,
      "loss": 0.1134,
      "step": 14063
    },
    {
      "epoch": 0.726258714175058,
      "grad_norm": 4.494521141052246,
      "learning_rate": 1.7708490747162572e-06,
      "loss": 0.0453,
      "step": 14064
    },
    {
      "epoch": 0.7263103537309579,
      "grad_norm": 3.318431854248047,
      "learning_rate": 1.7702235517064536e-06,
      "loss": 0.0492,
      "step": 14065
    },
    {
      "epoch": 0.7263619932868577,
      "grad_norm": 13.196928977966309,
      "learning_rate": 1.7695981154293112e-06,
      "loss": 0.1459,
      "step": 14066
    },
    {
      "epoch": 0.7264136328427575,
      "grad_norm": 2.920807361602783,
      "learning_rate": 1.7689727659016275e-06,
      "loss": 0.0205,
      "step": 14067
    },
    {
      "epoch": 0.7264652723986573,
      "grad_norm": 2.7209558486938477,
      "learning_rate": 1.7683475031401926e-06,
      "loss": 0.0376,
      "step": 14068
    },
    {
      "epoch": 0.7265169119545571,
      "grad_norm": 2.0581254959106445,
      "learning_rate": 1.7677223271617967e-06,
      "loss": 0.0266,
      "step": 14069
    },
    {
      "epoch": 0.726568551510457,
      "grad_norm": 7.987671375274658,
      "learning_rate": 1.7670972379832318e-06,
      "loss": 0.0783,
      "step": 14070
    },
    {
      "epoch": 0.7266201910663568,
      "grad_norm": 4.828814506530762,
      "learning_rate": 1.7664722356212848e-06,
      "loss": 0.0263,
      "step": 14071
    },
    {
      "epoch": 0.7266718306222566,
      "grad_norm": 2.120913028717041,
      "learning_rate": 1.7658473200927357e-06,
      "loss": 0.0329,
      "step": 14072
    },
    {
      "epoch": 0.7267234701781564,
      "grad_norm": 5.319515705108643,
      "learning_rate": 1.7652224914143673e-06,
      "loss": 0.026,
      "step": 14073
    },
    {
      "epoch": 0.7267751097340562,
      "grad_norm": 1.9382376670837402,
      "learning_rate": 1.764597749602958e-06,
      "loss": 0.0294,
      "step": 14074
    },
    {
      "epoch": 0.726826749289956,
      "grad_norm": 6.619617462158203,
      "learning_rate": 1.7639730946752892e-06,
      "loss": 0.0692,
      "step": 14075
    },
    {
      "epoch": 0.7268783888458559,
      "grad_norm": 3.3474512100219727,
      "learning_rate": 1.7633485266481303e-06,
      "loss": 0.0862,
      "step": 14076
    },
    {
      "epoch": 0.7269300284017557,
      "grad_norm": 3.006852865219116,
      "learning_rate": 1.7627240455382555e-06,
      "loss": 0.0381,
      "step": 14077
    },
    {
      "epoch": 0.7269816679576556,
      "grad_norm": 4.022782802581787,
      "learning_rate": 1.7620996513624345e-06,
      "loss": 0.0505,
      "step": 14078
    },
    {
      "epoch": 0.7270333075135554,
      "grad_norm": 3.144558906555176,
      "learning_rate": 1.7614753441374344e-06,
      "loss": 0.0168,
      "step": 14079
    },
    {
      "epoch": 0.7270849470694553,
      "grad_norm": 7.647284507751465,
      "learning_rate": 1.7608511238800208e-06,
      "loss": 0.0873,
      "step": 14080
    },
    {
      "epoch": 0.7271365866253551,
      "grad_norm": 1.7788074016571045,
      "learning_rate": 1.7602269906069569e-06,
      "loss": 0.0111,
      "step": 14081
    },
    {
      "epoch": 0.7271882261812549,
      "grad_norm": 16.125682830810547,
      "learning_rate": 1.7596029443350049e-06,
      "loss": 0.2388,
      "step": 14082
    },
    {
      "epoch": 0.7272398657371547,
      "grad_norm": 3.1612746715545654,
      "learning_rate": 1.758978985080917e-06,
      "loss": 0.0335,
      "step": 14083
    },
    {
      "epoch": 0.7272915052930545,
      "grad_norm": 3.1137969493865967,
      "learning_rate": 1.7583551128614551e-06,
      "loss": 0.051,
      "step": 14084
    },
    {
      "epoch": 0.7273431448489543,
      "grad_norm": 8.383699417114258,
      "learning_rate": 1.7577313276933705e-06,
      "loss": 0.0794,
      "step": 14085
    },
    {
      "epoch": 0.7273947844048542,
      "grad_norm": 2.412346124649048,
      "learning_rate": 1.757107629593416e-06,
      "loss": 0.0366,
      "step": 14086
    },
    {
      "epoch": 0.727446423960754,
      "grad_norm": 3.697206497192383,
      "learning_rate": 1.7564840185783372e-06,
      "loss": 0.0425,
      "step": 14087
    },
    {
      "epoch": 0.7274980635166538,
      "grad_norm": 2.38578724861145,
      "learning_rate": 1.755860494664881e-06,
      "loss": 0.0467,
      "step": 14088
    },
    {
      "epoch": 0.7275497030725536,
      "grad_norm": 4.689886093139648,
      "learning_rate": 1.755237057869796e-06,
      "loss": 0.0423,
      "step": 14089
    },
    {
      "epoch": 0.7276013426284534,
      "grad_norm": 8.540142059326172,
      "learning_rate": 1.7546137082098197e-06,
      "loss": 0.0724,
      "step": 14090
    },
    {
      "epoch": 0.7276529821843533,
      "grad_norm": 9.332904815673828,
      "learning_rate": 1.753990445701692e-06,
      "loss": 0.0731,
      "step": 14091
    },
    {
      "epoch": 0.7277046217402531,
      "grad_norm": 8.87077522277832,
      "learning_rate": 1.7533672703621514e-06,
      "loss": 0.0357,
      "step": 14092
    },
    {
      "epoch": 0.7277562612961529,
      "grad_norm": 3.5963618755340576,
      "learning_rate": 1.7527441822079321e-06,
      "loss": 0.0457,
      "step": 14093
    },
    {
      "epoch": 0.7278079008520527,
      "grad_norm": 8.371604919433594,
      "learning_rate": 1.7521211812557665e-06,
      "loss": 0.0448,
      "step": 14094
    },
    {
      "epoch": 0.7278595404079525,
      "grad_norm": 12.476273536682129,
      "learning_rate": 1.7514982675223845e-06,
      "loss": 0.0774,
      "step": 14095
    },
    {
      "epoch": 0.7279111799638524,
      "grad_norm": 2.7800447940826416,
      "learning_rate": 1.7508754410245148e-06,
      "loss": 0.0445,
      "step": 14096
    },
    {
      "epoch": 0.7279628195197522,
      "grad_norm": 6.770550727844238,
      "learning_rate": 1.7502527017788818e-06,
      "loss": 0.0441,
      "step": 14097
    },
    {
      "epoch": 0.728014459075652,
      "grad_norm": 2.820073366165161,
      "learning_rate": 1.749630049802209e-06,
      "loss": 0.0332,
      "step": 14098
    },
    {
      "epoch": 0.7280660986315518,
      "grad_norm": 5.1938982009887695,
      "learning_rate": 1.7490074851112176e-06,
      "loss": 0.0791,
      "step": 14099
    },
    {
      "epoch": 0.7281177381874516,
      "grad_norm": 3.2781810760498047,
      "learning_rate": 1.7483850077226267e-06,
      "loss": 0.0418,
      "step": 14100
    },
    {
      "epoch": 0.7281693777433514,
      "grad_norm": 3.7214627265930176,
      "learning_rate": 1.74776261765315e-06,
      "loss": 0.0412,
      "step": 14101
    },
    {
      "epoch": 0.7282210172992513,
      "grad_norm": 2.342602252960205,
      "learning_rate": 1.7471403149194999e-06,
      "loss": 0.043,
      "step": 14102
    },
    {
      "epoch": 0.7282726568551511,
      "grad_norm": 3.245574474334717,
      "learning_rate": 1.7465180995383924e-06,
      "loss": 0.0304,
      "step": 14103
    },
    {
      "epoch": 0.7283242964110509,
      "grad_norm": 2.6893696784973145,
      "learning_rate": 1.7458959715265362e-06,
      "loss": 0.0313,
      "step": 14104
    },
    {
      "epoch": 0.7283759359669507,
      "grad_norm": 5.245311260223389,
      "learning_rate": 1.7452739309006344e-06,
      "loss": 0.0211,
      "step": 14105
    },
    {
      "epoch": 0.7284275755228505,
      "grad_norm": 5.975137233734131,
      "learning_rate": 1.744651977677393e-06,
      "loss": 0.0331,
      "step": 14106
    },
    {
      "epoch": 0.7284792150787504,
      "grad_norm": 4.781938552856445,
      "learning_rate": 1.7440301118735125e-06,
      "loss": 0.0386,
      "step": 14107
    },
    {
      "epoch": 0.7285308546346502,
      "grad_norm": 4.359999179840088,
      "learning_rate": 1.7434083335056973e-06,
      "loss": 0.0305,
      "step": 14108
    },
    {
      "epoch": 0.72858249419055,
      "grad_norm": 2.2426183223724365,
      "learning_rate": 1.7427866425906404e-06,
      "loss": 0.0228,
      "step": 14109
    },
    {
      "epoch": 0.7286341337464498,
      "grad_norm": 6.7300519943237305,
      "learning_rate": 1.7421650391450368e-06,
      "loss": 0.0826,
      "step": 14110
    },
    {
      "epoch": 0.7286857733023496,
      "grad_norm": 3.1002767086029053,
      "learning_rate": 1.7415435231855809e-06,
      "loss": 0.0543,
      "step": 14111
    },
    {
      "epoch": 0.7287374128582494,
      "grad_norm": 11.485355377197266,
      "learning_rate": 1.7409220947289619e-06,
      "loss": 0.1346,
      "step": 14112
    },
    {
      "epoch": 0.7287890524141493,
      "grad_norm": 5.572006702423096,
      "learning_rate": 1.7403007537918676e-06,
      "loss": 0.0237,
      "step": 14113
    },
    {
      "epoch": 0.7288406919700491,
      "grad_norm": 5.286984920501709,
      "learning_rate": 1.7396795003909845e-06,
      "loss": 0.1137,
      "step": 14114
    },
    {
      "epoch": 0.7288923315259489,
      "grad_norm": 5.101585865020752,
      "learning_rate": 1.7390583345429962e-06,
      "loss": 0.0868,
      "step": 14115
    },
    {
      "epoch": 0.7289439710818487,
      "grad_norm": 3.5471370220184326,
      "learning_rate": 1.7384372562645796e-06,
      "loss": 0.0498,
      "step": 14116
    },
    {
      "epoch": 0.7289956106377485,
      "grad_norm": 3.734994411468506,
      "learning_rate": 1.7378162655724174e-06,
      "loss": 0.0483,
      "step": 14117
    },
    {
      "epoch": 0.7290472501936484,
      "grad_norm": 2.9314658641815186,
      "learning_rate": 1.7371953624831844e-06,
      "loss": 0.0328,
      "step": 14118
    },
    {
      "epoch": 0.7290988897495482,
      "grad_norm": 5.368320465087891,
      "learning_rate": 1.7365745470135565e-06,
      "loss": 0.0702,
      "step": 14119
    },
    {
      "epoch": 0.729150529305448,
      "grad_norm": 6.976957321166992,
      "learning_rate": 1.7359538191802006e-06,
      "loss": 0.0898,
      "step": 14120
    },
    {
      "epoch": 0.7292021688613478,
      "grad_norm": 5.143641948699951,
      "learning_rate": 1.7353331789997869e-06,
      "loss": 0.0367,
      "step": 14121
    },
    {
      "epoch": 0.7292538084172476,
      "grad_norm": 10.605318069458008,
      "learning_rate": 1.7347126264889868e-06,
      "loss": 0.1651,
      "step": 14122
    },
    {
      "epoch": 0.7293054479731474,
      "grad_norm": 2.307877779006958,
      "learning_rate": 1.7340921616644595e-06,
      "loss": 0.03,
      "step": 14123
    },
    {
      "epoch": 0.7293570875290473,
      "grad_norm": 5.329637050628662,
      "learning_rate": 1.7334717845428689e-06,
      "loss": 0.0729,
      "step": 14124
    },
    {
      "epoch": 0.7294087270849471,
      "grad_norm": 3.0664422512054443,
      "learning_rate": 1.7328514951408742e-06,
      "loss": 0.0471,
      "step": 14125
    },
    {
      "epoch": 0.7294603666408469,
      "grad_norm": 5.329265594482422,
      "learning_rate": 1.7322312934751334e-06,
      "loss": 0.0699,
      "step": 14126
    },
    {
      "epoch": 0.7295120061967467,
      "grad_norm": 2.7188971042633057,
      "learning_rate": 1.7316111795623008e-06,
      "loss": 0.0197,
      "step": 14127
    },
    {
      "epoch": 0.7295636457526465,
      "grad_norm": 5.2851762771606445,
      "learning_rate": 1.7309911534190293e-06,
      "loss": 0.0479,
      "step": 14128
    },
    {
      "epoch": 0.7296152853085464,
      "grad_norm": 2.2418627738952637,
      "learning_rate": 1.730371215061969e-06,
      "loss": 0.031,
      "step": 14129
    },
    {
      "epoch": 0.7296669248644462,
      "grad_norm": 18.134347915649414,
      "learning_rate": 1.729751364507768e-06,
      "loss": 0.1077,
      "step": 14130
    },
    {
      "epoch": 0.729718564420346,
      "grad_norm": 13.69114875793457,
      "learning_rate": 1.7291316017730714e-06,
      "loss": 0.1503,
      "step": 14131
    },
    {
      "epoch": 0.7297702039762458,
      "grad_norm": 3.3429009914398193,
      "learning_rate": 1.7285119268745227e-06,
      "loss": 0.0586,
      "step": 14132
    },
    {
      "epoch": 0.7298218435321456,
      "grad_norm": 1.379848837852478,
      "learning_rate": 1.7278923398287644e-06,
      "loss": 0.03,
      "step": 14133
    },
    {
      "epoch": 0.7298734830880454,
      "grad_norm": 8.222386360168457,
      "learning_rate": 1.7272728406524315e-06,
      "loss": 0.0816,
      "step": 14134
    },
    {
      "epoch": 0.7299251226439453,
      "grad_norm": 3.36639404296875,
      "learning_rate": 1.72665342936216e-06,
      "loss": 0.0503,
      "step": 14135
    },
    {
      "epoch": 0.7299767621998451,
      "grad_norm": 4.401717662811279,
      "learning_rate": 1.7260341059745872e-06,
      "loss": 0.0743,
      "step": 14136
    },
    {
      "epoch": 0.7300284017557449,
      "grad_norm": 18.49315643310547,
      "learning_rate": 1.725414870506344e-06,
      "loss": 0.1643,
      "step": 14137
    },
    {
      "epoch": 0.7300800413116447,
      "grad_norm": 6.7432122230529785,
      "learning_rate": 1.7247957229740565e-06,
      "loss": 0.046,
      "step": 14138
    },
    {
      "epoch": 0.7301316808675445,
      "grad_norm": 2.961240530014038,
      "learning_rate": 1.724176663394353e-06,
      "loss": 0.0239,
      "step": 14139
    },
    {
      "epoch": 0.7301833204234444,
      "grad_norm": 10.29203987121582,
      "learning_rate": 1.7235576917838554e-06,
      "loss": 0.1012,
      "step": 14140
    },
    {
      "epoch": 0.7302349599793442,
      "grad_norm": 3.038804292678833,
      "learning_rate": 1.7229388081591914e-06,
      "loss": 0.0333,
      "step": 14141
    },
    {
      "epoch": 0.730286599535244,
      "grad_norm": 8.281913757324219,
      "learning_rate": 1.7223200125369754e-06,
      "loss": 0.0952,
      "step": 14142
    },
    {
      "epoch": 0.7303382390911438,
      "grad_norm": 3.453230142593384,
      "learning_rate": 1.7217013049338261e-06,
      "loss": 0.0609,
      "step": 14143
    },
    {
      "epoch": 0.7303898786470436,
      "grad_norm": 19.19853973388672,
      "learning_rate": 1.7210826853663582e-06,
      "loss": 0.3053,
      "step": 14144
    },
    {
      "epoch": 0.7304415182029435,
      "grad_norm": 4.783461570739746,
      "learning_rate": 1.7204641538511846e-06,
      "loss": 0.0968,
      "step": 14145
    },
    {
      "epoch": 0.7304931577588433,
      "grad_norm": 2.5133841037750244,
      "learning_rate": 1.7198457104049148e-06,
      "loss": 0.0238,
      "step": 14146
    },
    {
      "epoch": 0.7305447973147431,
      "grad_norm": 1.4149094820022583,
      "learning_rate": 1.7192273550441573e-06,
      "loss": 0.0107,
      "step": 14147
    },
    {
      "epoch": 0.7305964368706429,
      "grad_norm": 6.029868125915527,
      "learning_rate": 1.7186090877855182e-06,
      "loss": 0.0708,
      "step": 14148
    },
    {
      "epoch": 0.7306480764265427,
      "grad_norm": 2.475008249282837,
      "learning_rate": 1.7179909086455964e-06,
      "loss": 0.0374,
      "step": 14149
    },
    {
      "epoch": 0.7306997159824425,
      "grad_norm": 3.928145408630371,
      "learning_rate": 1.717372817640997e-06,
      "loss": 0.0672,
      "step": 14150
    },
    {
      "epoch": 0.7307513555383424,
      "grad_norm": 2.860905647277832,
      "learning_rate": 1.7167548147883166e-06,
      "loss": 0.0619,
      "step": 14151
    },
    {
      "epoch": 0.7308029950942422,
      "grad_norm": 3.6960339546203613,
      "learning_rate": 1.7161369001041529e-06,
      "loss": 0.0436,
      "step": 14152
    },
    {
      "epoch": 0.730854634650142,
      "grad_norm": 3.2760157585144043,
      "learning_rate": 1.7155190736050957e-06,
      "loss": 0.0471,
      "step": 14153
    },
    {
      "epoch": 0.7309062742060418,
      "grad_norm": 3.369384288787842,
      "learning_rate": 1.714901335307737e-06,
      "loss": 0.0341,
      "step": 14154
    },
    {
      "epoch": 0.7309579137619416,
      "grad_norm": 3.6627516746520996,
      "learning_rate": 1.71428368522867e-06,
      "loss": 0.0412,
      "step": 14155
    },
    {
      "epoch": 0.7310095533178415,
      "grad_norm": 6.511607646942139,
      "learning_rate": 1.713666123384476e-06,
      "loss": 0.0643,
      "step": 14156
    },
    {
      "epoch": 0.7310611928737413,
      "grad_norm": 3.3211214542388916,
      "learning_rate": 1.7130486497917409e-06,
      "loss": 0.0566,
      "step": 14157
    },
    {
      "epoch": 0.7311128324296411,
      "grad_norm": 3.4324214458465576,
      "learning_rate": 1.7124312644670466e-06,
      "loss": 0.0559,
      "step": 14158
    },
    {
      "epoch": 0.7311644719855409,
      "grad_norm": 20.265419006347656,
      "learning_rate": 1.7118139674269724e-06,
      "loss": 0.1589,
      "step": 14159
    },
    {
      "epoch": 0.7312161115414407,
      "grad_norm": 4.2120490074157715,
      "learning_rate": 1.7111967586880945e-06,
      "loss": 0.0356,
      "step": 14160
    },
    {
      "epoch": 0.7312677510973405,
      "grad_norm": 5.736754894256592,
      "learning_rate": 1.7105796382669886e-06,
      "loss": 0.0419,
      "step": 14161
    },
    {
      "epoch": 0.7313193906532404,
      "grad_norm": 3.5699214935302734,
      "learning_rate": 1.7099626061802266e-06,
      "loss": 0.0651,
      "step": 14162
    },
    {
      "epoch": 0.7313710302091402,
      "grad_norm": 3.987856864929199,
      "learning_rate": 1.7093456624443778e-06,
      "loss": 0.0557,
      "step": 14163
    },
    {
      "epoch": 0.73142266976504,
      "grad_norm": 7.454395294189453,
      "learning_rate": 1.7087288070760094e-06,
      "loss": 0.0546,
      "step": 14164
    },
    {
      "epoch": 0.7314743093209398,
      "grad_norm": 10.261038780212402,
      "learning_rate": 1.7081120400916873e-06,
      "loss": 0.0895,
      "step": 14165
    },
    {
      "epoch": 0.7315259488768396,
      "grad_norm": 1.9738476276397705,
      "learning_rate": 1.707495361507976e-06,
      "loss": 0.0201,
      "step": 14166
    },
    {
      "epoch": 0.7315775884327395,
      "grad_norm": 5.422155380249023,
      "learning_rate": 1.7068787713414314e-06,
      "loss": 0.0255,
      "step": 14167
    },
    {
      "epoch": 0.7316292279886393,
      "grad_norm": 1.9983934164047241,
      "learning_rate": 1.7062622696086128e-06,
      "loss": 0.0216,
      "step": 14168
    },
    {
      "epoch": 0.7316808675445391,
      "grad_norm": 3.308859348297119,
      "learning_rate": 1.705645856326078e-06,
      "loss": 0.0645,
      "step": 14169
    },
    {
      "epoch": 0.7317325071004389,
      "grad_norm": 8.770855903625488,
      "learning_rate": 1.705029531510381e-06,
      "loss": 0.1056,
      "step": 14170
    },
    {
      "epoch": 0.7317841466563387,
      "grad_norm": 3.638268232345581,
      "learning_rate": 1.704413295178069e-06,
      "loss": 0.0704,
      "step": 14171
    },
    {
      "epoch": 0.7318357862122385,
      "grad_norm": 4.31083869934082,
      "learning_rate": 1.7037971473456915e-06,
      "loss": 0.0493,
      "step": 14172
    },
    {
      "epoch": 0.7318874257681384,
      "grad_norm": 6.278806209564209,
      "learning_rate": 1.7031810880297944e-06,
      "loss": 0.0671,
      "step": 14173
    },
    {
      "epoch": 0.7319390653240382,
      "grad_norm": 3.855698823928833,
      "learning_rate": 1.702565117246925e-06,
      "loss": 0.0344,
      "step": 14174
    },
    {
      "epoch": 0.731990704879938,
      "grad_norm": 2.803447961807251,
      "learning_rate": 1.7019492350136202e-06,
      "loss": 0.0378,
      "step": 14175
    },
    {
      "epoch": 0.7320423444358378,
      "grad_norm": 2.5168981552124023,
      "learning_rate": 1.7013334413464206e-06,
      "loss": 0.0416,
      "step": 14176
    },
    {
      "epoch": 0.7320939839917376,
      "grad_norm": 3.0102269649505615,
      "learning_rate": 1.7007177362618626e-06,
      "loss": 0.0405,
      "step": 14177
    },
    {
      "epoch": 0.7321456235476375,
      "grad_norm": 3.158912420272827,
      "learning_rate": 1.7001021197764806e-06,
      "loss": 0.0579,
      "step": 14178
    },
    {
      "epoch": 0.7321972631035373,
      "grad_norm": 4.774838447570801,
      "learning_rate": 1.6994865919068059e-06,
      "loss": 0.0728,
      "step": 14179
    },
    {
      "epoch": 0.7322489026594371,
      "grad_norm": 3.996920585632324,
      "learning_rate": 1.6988711526693692e-06,
      "loss": 0.0711,
      "step": 14180
    },
    {
      "epoch": 0.7323005422153369,
      "grad_norm": 1.80556321144104,
      "learning_rate": 1.6982558020806978e-06,
      "loss": 0.0343,
      "step": 14181
    },
    {
      "epoch": 0.7323521817712367,
      "grad_norm": 3.96160626411438,
      "learning_rate": 1.697640540157312e-06,
      "loss": 0.0549,
      "step": 14182
    },
    {
      "epoch": 0.7324038213271366,
      "grad_norm": 5.401858329772949,
      "learning_rate": 1.6970253669157389e-06,
      "loss": 0.0511,
      "step": 14183
    },
    {
      "epoch": 0.7324554608830364,
      "grad_norm": 2.696237564086914,
      "learning_rate": 1.6964102823724986e-06,
      "loss": 0.0363,
      "step": 14184
    },
    {
      "epoch": 0.7325071004389362,
      "grad_norm": 6.30901575088501,
      "learning_rate": 1.6957952865441058e-06,
      "loss": 0.0423,
      "step": 14185
    },
    {
      "epoch": 0.732558739994836,
      "grad_norm": 1.8967262506484985,
      "learning_rate": 1.6951803794470762e-06,
      "loss": 0.0368,
      "step": 14186
    },
    {
      "epoch": 0.7326103795507358,
      "grad_norm": 6.1730241775512695,
      "learning_rate": 1.6945655610979212e-06,
      "loss": 0.084,
      "step": 14187
    },
    {
      "epoch": 0.7326620191066356,
      "grad_norm": 8.839475631713867,
      "learning_rate": 1.693950831513157e-06,
      "loss": 0.0603,
      "step": 14188
    },
    {
      "epoch": 0.7327136586625355,
      "grad_norm": 8.390316009521484,
      "learning_rate": 1.693336190709286e-06,
      "loss": 0.0605,
      "step": 14189
    },
    {
      "epoch": 0.7327652982184353,
      "grad_norm": 1.6416568756103516,
      "learning_rate": 1.692721638702815e-06,
      "loss": 0.0328,
      "step": 14190
    },
    {
      "epoch": 0.7328169377743351,
      "grad_norm": 7.179189682006836,
      "learning_rate": 1.6921071755102481e-06,
      "loss": 0.1087,
      "step": 14191
    },
    {
      "epoch": 0.7328685773302349,
      "grad_norm": 6.908755779266357,
      "learning_rate": 1.6914928011480858e-06,
      "loss": 0.057,
      "step": 14192
    },
    {
      "epoch": 0.7329202168861347,
      "grad_norm": 5.474711894989014,
      "learning_rate": 1.6908785156328262e-06,
      "loss": 0.059,
      "step": 14193
    },
    {
      "epoch": 0.7329718564420346,
      "grad_norm": 3.6895570755004883,
      "learning_rate": 1.6902643189809659e-06,
      "loss": 0.0342,
      "step": 14194
    },
    {
      "epoch": 0.7330234959979344,
      "grad_norm": 5.219858169555664,
      "learning_rate": 1.689650211209e-06,
      "loss": 0.0402,
      "step": 14195
    },
    {
      "epoch": 0.7330751355538342,
      "grad_norm": 3.6522626876831055,
      "learning_rate": 1.6890361923334147e-06,
      "loss": 0.0682,
      "step": 14196
    },
    {
      "epoch": 0.733126775109734,
      "grad_norm": 3.22037935256958,
      "learning_rate": 1.6884222623707047e-06,
      "loss": 0.0491,
      "step": 14197
    },
    {
      "epoch": 0.7331784146656338,
      "grad_norm": 3.3992931842803955,
      "learning_rate": 1.6878084213373536e-06,
      "loss": 0.0294,
      "step": 14198
    },
    {
      "epoch": 0.7332300542215336,
      "grad_norm": 3.2683515548706055,
      "learning_rate": 1.687194669249848e-06,
      "loss": 0.0565,
      "step": 14199
    },
    {
      "epoch": 0.7332816937774335,
      "grad_norm": 1.8941086530685425,
      "learning_rate": 1.6865810061246657e-06,
      "loss": 0.022,
      "step": 14200
    },
    {
      "epoch": 0.7333333333333333,
      "grad_norm": 2.7916018962860107,
      "learning_rate": 1.685967431978287e-06,
      "loss": 0.0439,
      "step": 14201
    },
    {
      "epoch": 0.7333849728892331,
      "grad_norm": 2.882697105407715,
      "learning_rate": 1.6853539468271912e-06,
      "loss": 0.027,
      "step": 14202
    },
    {
      "epoch": 0.7334366124451329,
      "grad_norm": 4.40513801574707,
      "learning_rate": 1.684740550687854e-06,
      "loss": 0.0585,
      "step": 14203
    },
    {
      "epoch": 0.7334882520010327,
      "grad_norm": 6.11610221862793,
      "learning_rate": 1.684127243576743e-06,
      "loss": 0.1029,
      "step": 14204
    },
    {
      "epoch": 0.7335398915569326,
      "grad_norm": 2.559016704559326,
      "learning_rate": 1.6835140255103306e-06,
      "loss": 0.0489,
      "step": 14205
    },
    {
      "epoch": 0.7335915311128325,
      "grad_norm": 3.1299636363983154,
      "learning_rate": 1.6829008965050835e-06,
      "loss": 0.0432,
      "step": 14206
    },
    {
      "epoch": 0.7336431706687323,
      "grad_norm": 3.3008224964141846,
      "learning_rate": 1.6822878565774674e-06,
      "loss": 0.0246,
      "step": 14207
    },
    {
      "epoch": 0.7336948102246321,
      "grad_norm": 4.862478733062744,
      "learning_rate": 1.6816749057439446e-06,
      "loss": 0.042,
      "step": 14208
    },
    {
      "epoch": 0.7337464497805319,
      "grad_norm": 7.384129047393799,
      "learning_rate": 1.6810620440209751e-06,
      "loss": 0.0602,
      "step": 14209
    },
    {
      "epoch": 0.7337980893364318,
      "grad_norm": 3.4681177139282227,
      "learning_rate": 1.680449271425017e-06,
      "loss": 0.0331,
      "step": 14210
    },
    {
      "epoch": 0.7338497288923316,
      "grad_norm": 4.960116386413574,
      "learning_rate": 1.679836587972526e-06,
      "loss": 0.058,
      "step": 14211
    },
    {
      "epoch": 0.7339013684482314,
      "grad_norm": 2.405146598815918,
      "learning_rate": 1.679223993679955e-06,
      "loss": 0.0346,
      "step": 14212
    },
    {
      "epoch": 0.7339530080041312,
      "grad_norm": 2.485304832458496,
      "learning_rate": 1.6786114885637539e-06,
      "loss": 0.0213,
      "step": 14213
    },
    {
      "epoch": 0.734004647560031,
      "grad_norm": 3.1756930351257324,
      "learning_rate": 1.6779990726403733e-06,
      "loss": 0.0581,
      "step": 14214
    },
    {
      "epoch": 0.7340562871159309,
      "grad_norm": 2.1763365268707275,
      "learning_rate": 1.6773867459262544e-06,
      "loss": 0.0165,
      "step": 14215
    },
    {
      "epoch": 0.7341079266718307,
      "grad_norm": 10.09928035736084,
      "learning_rate": 1.6767745084378445e-06,
      "loss": 0.0986,
      "step": 14216
    },
    {
      "epoch": 0.7341595662277305,
      "grad_norm": 2.826106548309326,
      "learning_rate": 1.6761623601915861e-06,
      "loss": 0.0377,
      "step": 14217
    },
    {
      "epoch": 0.7342112057836303,
      "grad_norm": 3.857290029525757,
      "learning_rate": 1.6755503012039138e-06,
      "loss": 0.0371,
      "step": 14218
    },
    {
      "epoch": 0.7342628453395301,
      "grad_norm": 6.128674507141113,
      "learning_rate": 1.6749383314912654e-06,
      "loss": 0.125,
      "step": 14219
    },
    {
      "epoch": 0.7343144848954299,
      "grad_norm": 3.052948236465454,
      "learning_rate": 1.6743264510700735e-06,
      "loss": 0.0327,
      "step": 14220
    },
    {
      "epoch": 0.7343661244513298,
      "grad_norm": 13.580998420715332,
      "learning_rate": 1.6737146599567744e-06,
      "loss": 0.1714,
      "step": 14221
    },
    {
      "epoch": 0.7344177640072296,
      "grad_norm": 42.12531280517578,
      "learning_rate": 1.6731029581677922e-06,
      "loss": 0.1278,
      "step": 14222
    },
    {
      "epoch": 0.7344694035631294,
      "grad_norm": 0.9146215319633484,
      "learning_rate": 1.6724913457195553e-06,
      "loss": 0.0078,
      "step": 14223
    },
    {
      "epoch": 0.7345210431190292,
      "grad_norm": 7.625790596008301,
      "learning_rate": 1.6718798226284882e-06,
      "loss": 0.0652,
      "step": 14224
    },
    {
      "epoch": 0.734572682674929,
      "grad_norm": 1.5819460153579712,
      "learning_rate": 1.671268388911012e-06,
      "loss": 0.0198,
      "step": 14225
    },
    {
      "epoch": 0.7346243222308289,
      "grad_norm": 5.780317783355713,
      "learning_rate": 1.6706570445835468e-06,
      "loss": 0.0601,
      "step": 14226
    },
    {
      "epoch": 0.7346759617867287,
      "grad_norm": 2.987684965133667,
      "learning_rate": 1.6700457896625099e-06,
      "loss": 0.0372,
      "step": 14227
    },
    {
      "epoch": 0.7347276013426285,
      "grad_norm": 2.8670451641082764,
      "learning_rate": 1.6694346241643173e-06,
      "loss": 0.0379,
      "step": 14228
    },
    {
      "epoch": 0.7347792408985283,
      "grad_norm": 6.241431713104248,
      "learning_rate": 1.668823548105376e-06,
      "loss": 0.0438,
      "step": 14229
    },
    {
      "epoch": 0.7348308804544281,
      "grad_norm": 13.935507774353027,
      "learning_rate": 1.6682125615021015e-06,
      "loss": 0.0734,
      "step": 14230
    },
    {
      "epoch": 0.7348825200103279,
      "grad_norm": 2.0736684799194336,
      "learning_rate": 1.6676016643708997e-06,
      "loss": 0.0287,
      "step": 14231
    },
    {
      "epoch": 0.7349341595662278,
      "grad_norm": 4.323572158813477,
      "learning_rate": 1.6669908567281762e-06,
      "loss": 0.0715,
      "step": 14232
    },
    {
      "epoch": 0.7349857991221276,
      "grad_norm": 11.516743659973145,
      "learning_rate": 1.6663801385903316e-06,
      "loss": 0.0972,
      "step": 14233
    },
    {
      "epoch": 0.7350374386780274,
      "grad_norm": 2.264552354812622,
      "learning_rate": 1.6657695099737658e-06,
      "loss": 0.0289,
      "step": 14234
    },
    {
      "epoch": 0.7350890782339272,
      "grad_norm": 3.055544376373291,
      "learning_rate": 1.6651589708948795e-06,
      "loss": 0.0425,
      "step": 14235
    },
    {
      "epoch": 0.735140717789827,
      "grad_norm": 2.23075532913208,
      "learning_rate": 1.664548521370069e-06,
      "loss": 0.0354,
      "step": 14236
    },
    {
      "epoch": 0.7351923573457269,
      "grad_norm": 12.638689041137695,
      "learning_rate": 1.6639381614157235e-06,
      "loss": 0.3188,
      "step": 14237
    },
    {
      "epoch": 0.7352439969016267,
      "grad_norm": 8.244234085083008,
      "learning_rate": 1.6633278910482348e-06,
      "loss": 0.0771,
      "step": 14238
    },
    {
      "epoch": 0.7352956364575265,
      "grad_norm": 3.59560489654541,
      "learning_rate": 1.6627177102839926e-06,
      "loss": 0.0556,
      "step": 14239
    },
    {
      "epoch": 0.7353472760134263,
      "grad_norm": 9.462258338928223,
      "learning_rate": 1.6621076191393815e-06,
      "loss": 0.1604,
      "step": 14240
    },
    {
      "epoch": 0.7353989155693261,
      "grad_norm": 6.2037835121154785,
      "learning_rate": 1.661497617630785e-06,
      "loss": 0.061,
      "step": 14241
    },
    {
      "epoch": 0.735450555125226,
      "grad_norm": 2.2443904876708984,
      "learning_rate": 1.6608877057745848e-06,
      "loss": 0.0258,
      "step": 14242
    },
    {
      "epoch": 0.7355021946811258,
      "grad_norm": 4.590692043304443,
      "learning_rate": 1.6602778835871591e-06,
      "loss": 0.0617,
      "step": 14243
    },
    {
      "epoch": 0.7355538342370256,
      "grad_norm": 3.8026340007781982,
      "learning_rate": 1.6596681510848845e-06,
      "loss": 0.0601,
      "step": 14244
    },
    {
      "epoch": 0.7356054737929254,
      "grad_norm": 1.740297794342041,
      "learning_rate": 1.6590585082841342e-06,
      "loss": 0.024,
      "step": 14245
    },
    {
      "epoch": 0.7356571133488252,
      "grad_norm": 3.108618974685669,
      "learning_rate": 1.6584489552012795e-06,
      "loss": 0.0551,
      "step": 14246
    },
    {
      "epoch": 0.735708752904725,
      "grad_norm": 1.30581533908844,
      "learning_rate": 1.6578394918526918e-06,
      "loss": 0.0329,
      "step": 14247
    },
    {
      "epoch": 0.7357603924606249,
      "grad_norm": 4.2031121253967285,
      "learning_rate": 1.6572301182547324e-06,
      "loss": 0.1323,
      "step": 14248
    },
    {
      "epoch": 0.7358120320165247,
      "grad_norm": 10.184308052062988,
      "learning_rate": 1.6566208344237705e-06,
      "loss": 0.1237,
      "step": 14249
    },
    {
      "epoch": 0.7358636715724245,
      "grad_norm": 2.902963638305664,
      "learning_rate": 1.6560116403761673e-06,
      "loss": 0.0358,
      "step": 14250
    },
    {
      "epoch": 0.7359153111283243,
      "grad_norm": 2.006089210510254,
      "learning_rate": 1.6554025361282795e-06,
      "loss": 0.04,
      "step": 14251
    },
    {
      "epoch": 0.7359669506842241,
      "grad_norm": 3.7682011127471924,
      "learning_rate": 1.6547935216964656e-06,
      "loss": 0.0661,
      "step": 14252
    },
    {
      "epoch": 0.736018590240124,
      "grad_norm": 3.3649380207061768,
      "learning_rate": 1.6541845970970777e-06,
      "loss": 0.0543,
      "step": 14253
    },
    {
      "epoch": 0.7360702297960238,
      "grad_norm": 3.9760525226593018,
      "learning_rate": 1.6535757623464738e-06,
      "loss": 0.0394,
      "step": 14254
    },
    {
      "epoch": 0.7361218693519236,
      "grad_norm": 7.318526268005371,
      "learning_rate": 1.6529670174609985e-06,
      "loss": 0.0351,
      "step": 14255
    },
    {
      "epoch": 0.7361735089078234,
      "grad_norm": 18.556896209716797,
      "learning_rate": 1.6523583624569999e-06,
      "loss": 0.1627,
      "step": 14256
    },
    {
      "epoch": 0.7362251484637232,
      "grad_norm": 9.091301918029785,
      "learning_rate": 1.6517497973508244e-06,
      "loss": 0.1022,
      "step": 14257
    },
    {
      "epoch": 0.736276788019623,
      "grad_norm": 4.081773281097412,
      "learning_rate": 1.6511413221588129e-06,
      "loss": 0.0625,
      "step": 14258
    },
    {
      "epoch": 0.7363284275755229,
      "grad_norm": 2.91920804977417,
      "learning_rate": 1.650532936897306e-06,
      "loss": 0.0432,
      "step": 14259
    },
    {
      "epoch": 0.7363800671314227,
      "grad_norm": 2.8893330097198486,
      "learning_rate": 1.649924641582642e-06,
      "loss": 0.0369,
      "step": 14260
    },
    {
      "epoch": 0.7364317066873225,
      "grad_norm": 3.8131070137023926,
      "learning_rate": 1.649316436231157e-06,
      "loss": 0.0443,
      "step": 14261
    },
    {
      "epoch": 0.7364833462432223,
      "grad_norm": 3.261488437652588,
      "learning_rate": 1.6487083208591787e-06,
      "loss": 0.0412,
      "step": 14262
    },
    {
      "epoch": 0.7365349857991221,
      "grad_norm": 5.551968097686768,
      "learning_rate": 1.6481002954830433e-06,
      "loss": 0.0407,
      "step": 14263
    },
    {
      "epoch": 0.736586625355022,
      "grad_norm": 2.660409450531006,
      "learning_rate": 1.6474923601190762e-06,
      "loss": 0.0446,
      "step": 14264
    },
    {
      "epoch": 0.7366382649109218,
      "grad_norm": 12.873348236083984,
      "learning_rate": 1.6468845147836048e-06,
      "loss": 0.0771,
      "step": 14265
    },
    {
      "epoch": 0.7366899044668216,
      "grad_norm": 5.65117073059082,
      "learning_rate": 1.6462767594929491e-06,
      "loss": 0.07,
      "step": 14266
    },
    {
      "epoch": 0.7367415440227214,
      "grad_norm": 1.3924764394760132,
      "learning_rate": 1.64566909426343e-06,
      "loss": 0.0345,
      "step": 14267
    },
    {
      "epoch": 0.7367931835786212,
      "grad_norm": 4.156986236572266,
      "learning_rate": 1.645061519111369e-06,
      "loss": 0.0639,
      "step": 14268
    },
    {
      "epoch": 0.736844823134521,
      "grad_norm": 2.7211732864379883,
      "learning_rate": 1.6444540340530812e-06,
      "loss": 0.0273,
      "step": 14269
    },
    {
      "epoch": 0.7368964626904209,
      "grad_norm": 1.3750438690185547,
      "learning_rate": 1.643846639104878e-06,
      "loss": 0.0357,
      "step": 14270
    },
    {
      "epoch": 0.7369481022463207,
      "grad_norm": 2.2787630558013916,
      "learning_rate": 1.643239334283071e-06,
      "loss": 0.0436,
      "step": 14271
    },
    {
      "epoch": 0.7369997418022205,
      "grad_norm": 6.400017738342285,
      "learning_rate": 1.6426321196039695e-06,
      "loss": 0.0693,
      "step": 14272
    },
    {
      "epoch": 0.7370513813581203,
      "grad_norm": 1.5482630729675293,
      "learning_rate": 1.6420249950838796e-06,
      "loss": 0.0181,
      "step": 14273
    },
    {
      "epoch": 0.7371030209140201,
      "grad_norm": 2.1268529891967773,
      "learning_rate": 1.641417960739105e-06,
      "loss": 0.0332,
      "step": 14274
    },
    {
      "epoch": 0.73715466046992,
      "grad_norm": 6.948975086212158,
      "learning_rate": 1.6408110165859464e-06,
      "loss": 0.0774,
      "step": 14275
    },
    {
      "epoch": 0.7372063000258198,
      "grad_norm": 1.8915570974349976,
      "learning_rate": 1.640204162640704e-06,
      "loss": 0.0227,
      "step": 14276
    },
    {
      "epoch": 0.7372579395817196,
      "grad_norm": 4.022927761077881,
      "learning_rate": 1.6395973989196735e-06,
      "loss": 0.0417,
      "step": 14277
    },
    {
      "epoch": 0.7373095791376194,
      "grad_norm": 4.996252059936523,
      "learning_rate": 1.6389907254391485e-06,
      "loss": 0.0406,
      "step": 14278
    },
    {
      "epoch": 0.7373612186935192,
      "grad_norm": 3.301896572113037,
      "learning_rate": 1.6383841422154218e-06,
      "loss": 0.0582,
      "step": 14279
    },
    {
      "epoch": 0.737412858249419,
      "grad_norm": 8.292272567749023,
      "learning_rate": 1.6377776492647835e-06,
      "loss": 0.0595,
      "step": 14280
    },
    {
      "epoch": 0.7374644978053189,
      "grad_norm": 4.897268295288086,
      "learning_rate": 1.6371712466035155e-06,
      "loss": 0.0669,
      "step": 14281
    },
    {
      "epoch": 0.7375161373612187,
      "grad_norm": 4.099271774291992,
      "learning_rate": 1.636564934247908e-06,
      "loss": 0.0549,
      "step": 14282
    },
    {
      "epoch": 0.7375677769171185,
      "grad_norm": 14.977356910705566,
      "learning_rate": 1.6359587122142411e-06,
      "loss": 0.2112,
      "step": 14283
    },
    {
      "epoch": 0.7376194164730183,
      "grad_norm": 4.991696834564209,
      "learning_rate": 1.635352580518793e-06,
      "loss": 0.0558,
      "step": 14284
    },
    {
      "epoch": 0.7376710560289181,
      "grad_norm": 2.967669725418091,
      "learning_rate": 1.6347465391778417e-06,
      "loss": 0.0339,
      "step": 14285
    },
    {
      "epoch": 0.737722695584818,
      "grad_norm": 2.1721410751342773,
      "learning_rate": 1.6341405882076599e-06,
      "loss": 0.0213,
      "step": 14286
    },
    {
      "epoch": 0.7377743351407178,
      "grad_norm": 10.789201736450195,
      "learning_rate": 1.6335347276245256e-06,
      "loss": 0.0714,
      "step": 14287
    },
    {
      "epoch": 0.7378259746966176,
      "grad_norm": 3.4207165241241455,
      "learning_rate": 1.6329289574447026e-06,
      "loss": 0.0341,
      "step": 14288
    },
    {
      "epoch": 0.7378776142525174,
      "grad_norm": 3.9229018688201904,
      "learning_rate": 1.6323232776844605e-06,
      "loss": 0.048,
      "step": 14289
    },
    {
      "epoch": 0.7379292538084172,
      "grad_norm": 10.441485404968262,
      "learning_rate": 1.6317176883600644e-06,
      "loss": 0.1092,
      "step": 14290
    },
    {
      "epoch": 0.737980893364317,
      "grad_norm": 3.3276708126068115,
      "learning_rate": 1.6311121894877768e-06,
      "loss": 0.0378,
      "step": 14291
    },
    {
      "epoch": 0.7380325329202169,
      "grad_norm": 3.264265537261963,
      "learning_rate": 1.6305067810838582e-06,
      "loss": 0.0588,
      "step": 14292
    },
    {
      "epoch": 0.7380841724761167,
      "grad_norm": 7.682229042053223,
      "learning_rate": 1.6299014631645653e-06,
      "loss": 0.0441,
      "step": 14293
    },
    {
      "epoch": 0.7381358120320165,
      "grad_norm": 9.840835571289062,
      "learning_rate": 1.6292962357461556e-06,
      "loss": 0.1335,
      "step": 14294
    },
    {
      "epoch": 0.7381874515879163,
      "grad_norm": 8.099678993225098,
      "learning_rate": 1.6286910988448767e-06,
      "loss": 0.0672,
      "step": 14295
    },
    {
      "epoch": 0.7382390911438161,
      "grad_norm": 3.1469919681549072,
      "learning_rate": 1.6280860524769843e-06,
      "loss": 0.0459,
      "step": 14296
    },
    {
      "epoch": 0.738290730699716,
      "grad_norm": 5.907607555389404,
      "learning_rate": 1.6274810966587245e-06,
      "loss": 0.0563,
      "step": 14297
    },
    {
      "epoch": 0.7383423702556158,
      "grad_norm": 1.3300108909606934,
      "learning_rate": 1.6268762314063447e-06,
      "loss": 0.0168,
      "step": 14298
    },
    {
      "epoch": 0.7383940098115156,
      "grad_norm": 6.265170574188232,
      "learning_rate": 1.626271456736085e-06,
      "loss": 0.0533,
      "step": 14299
    },
    {
      "epoch": 0.7384456493674154,
      "grad_norm": 4.379438877105713,
      "learning_rate": 1.6256667726641845e-06,
      "loss": 0.0776,
      "step": 14300
    },
    {
      "epoch": 0.7384972889233152,
      "grad_norm": 5.521923542022705,
      "learning_rate": 1.6250621792068866e-06,
      "loss": 0.0565,
      "step": 14301
    },
    {
      "epoch": 0.738548928479215,
      "grad_norm": 2.2921130657196045,
      "learning_rate": 1.6244576763804259e-06,
      "loss": 0.0235,
      "step": 14302
    },
    {
      "epoch": 0.7386005680351149,
      "grad_norm": 7.319185256958008,
      "learning_rate": 1.623853264201033e-06,
      "loss": 0.1133,
      "step": 14303
    },
    {
      "epoch": 0.7386522075910147,
      "grad_norm": 1.9768251180648804,
      "learning_rate": 1.6232489426849408e-06,
      "loss": 0.0313,
      "step": 14304
    },
    {
      "epoch": 0.7387038471469145,
      "grad_norm": 9.147936820983887,
      "learning_rate": 1.6226447118483768e-06,
      "loss": 0.1275,
      "step": 14305
    },
    {
      "epoch": 0.7387554867028143,
      "grad_norm": 10.920553207397461,
      "learning_rate": 1.6220405717075677e-06,
      "loss": 0.0602,
      "step": 14306
    },
    {
      "epoch": 0.7388071262587141,
      "grad_norm": 3.0802690982818604,
      "learning_rate": 1.6214365222787376e-06,
      "loss": 0.0595,
      "step": 14307
    },
    {
      "epoch": 0.738858765814614,
      "grad_norm": 10.338725090026855,
      "learning_rate": 1.6208325635781064e-06,
      "loss": 0.0418,
      "step": 14308
    },
    {
      "epoch": 0.7389104053705138,
      "grad_norm": 6.992836952209473,
      "learning_rate": 1.6202286956218943e-06,
      "loss": 0.0776,
      "step": 14309
    },
    {
      "epoch": 0.7389620449264136,
      "grad_norm": 1.8129562139511108,
      "learning_rate": 1.6196249184263162e-06,
      "loss": 0.0233,
      "step": 14310
    },
    {
      "epoch": 0.7390136844823134,
      "grad_norm": 17.51304054260254,
      "learning_rate": 1.6190212320075871e-06,
      "loss": 0.0797,
      "step": 14311
    },
    {
      "epoch": 0.7390653240382132,
      "grad_norm": 2.6178438663482666,
      "learning_rate": 1.6184176363819181e-06,
      "loss": 0.0431,
      "step": 14312
    },
    {
      "epoch": 0.739116963594113,
      "grad_norm": 2.258953094482422,
      "learning_rate": 1.6178141315655193e-06,
      "loss": 0.0329,
      "step": 14313
    },
    {
      "epoch": 0.7391686031500129,
      "grad_norm": 4.32627534866333,
      "learning_rate": 1.6172107175745932e-06,
      "loss": 0.0312,
      "step": 14314
    },
    {
      "epoch": 0.7392202427059127,
      "grad_norm": 1.7818189859390259,
      "learning_rate": 1.6166073944253484e-06,
      "loss": 0.0264,
      "step": 14315
    },
    {
      "epoch": 0.7392718822618125,
      "grad_norm": 2.8847389221191406,
      "learning_rate": 1.6160041621339868e-06,
      "loss": 0.0338,
      "step": 14316
    },
    {
      "epoch": 0.7393235218177123,
      "grad_norm": 3.074846029281616,
      "learning_rate": 1.6154010207167041e-06,
      "loss": 0.0313,
      "step": 14317
    },
    {
      "epoch": 0.7393751613736121,
      "grad_norm": 3.647955894470215,
      "learning_rate": 1.6147979701896988e-06,
      "loss": 0.0429,
      "step": 14318
    },
    {
      "epoch": 0.739426800929512,
      "grad_norm": 4.217672348022461,
      "learning_rate": 1.6141950105691634e-06,
      "loss": 0.0401,
      "step": 14319
    },
    {
      "epoch": 0.7394784404854118,
      "grad_norm": 2.9902565479278564,
      "learning_rate": 1.6135921418712959e-06,
      "loss": 0.031,
      "step": 14320
    },
    {
      "epoch": 0.7395300800413116,
      "grad_norm": 4.457846164703369,
      "learning_rate": 1.612989364112279e-06,
      "loss": 0.064,
      "step": 14321
    },
    {
      "epoch": 0.7395817195972114,
      "grad_norm": 3.904324531555176,
      "learning_rate": 1.612386677308303e-06,
      "loss": 0.0347,
      "step": 14322
    },
    {
      "epoch": 0.7396333591531112,
      "grad_norm": 5.457277774810791,
      "learning_rate": 1.6117840814755513e-06,
      "loss": 0.0668,
      "step": 14323
    },
    {
      "epoch": 0.739684998709011,
      "grad_norm": 13.905402183532715,
      "learning_rate": 1.611181576630207e-06,
      "loss": 0.0295,
      "step": 14324
    },
    {
      "epoch": 0.7397366382649109,
      "grad_norm": 2.260861396789551,
      "learning_rate": 1.6105791627884493e-06,
      "loss": 0.0407,
      "step": 14325
    },
    {
      "epoch": 0.7397882778208107,
      "grad_norm": 9.669196128845215,
      "learning_rate": 1.6099768399664557e-06,
      "loss": 0.1232,
      "step": 14326
    },
    {
      "epoch": 0.7398399173767105,
      "grad_norm": 5.374667167663574,
      "learning_rate": 1.6093746081804024e-06,
      "loss": 0.0447,
      "step": 14327
    },
    {
      "epoch": 0.7398915569326103,
      "grad_norm": 3.7351508140563965,
      "learning_rate": 1.6087724674464572e-06,
      "loss": 0.044,
      "step": 14328
    },
    {
      "epoch": 0.7399431964885101,
      "grad_norm": 5.437435626983643,
      "learning_rate": 1.608170417780795e-06,
      "loss": 0.0723,
      "step": 14329
    },
    {
      "epoch": 0.73999483604441,
      "grad_norm": 9.588430404663086,
      "learning_rate": 1.607568459199581e-06,
      "loss": 0.0586,
      "step": 14330
    },
    {
      "epoch": 0.7400464756003098,
      "grad_norm": 4.320042610168457,
      "learning_rate": 1.6069665917189826e-06,
      "loss": 0.0648,
      "step": 14331
    },
    {
      "epoch": 0.7400981151562096,
      "grad_norm": 15.659320831298828,
      "learning_rate": 1.606364815355158e-06,
      "loss": 0.0805,
      "step": 14332
    },
    {
      "epoch": 0.7401497547121095,
      "grad_norm": 5.643511772155762,
      "learning_rate": 1.6057631301242693e-06,
      "loss": 0.0546,
      "step": 14333
    },
    {
      "epoch": 0.7402013942680093,
      "grad_norm": 4.098459243774414,
      "learning_rate": 1.6051615360424754e-06,
      "loss": 0.0491,
      "step": 14334
    },
    {
      "epoch": 0.7402530338239092,
      "grad_norm": 10.074353218078613,
      "learning_rate": 1.604560033125933e-06,
      "loss": 0.1168,
      "step": 14335
    },
    {
      "epoch": 0.740304673379809,
      "grad_norm": 2.7445430755615234,
      "learning_rate": 1.6039586213907909e-06,
      "loss": 0.0401,
      "step": 14336
    },
    {
      "epoch": 0.7403563129357088,
      "grad_norm": 3.563096523284912,
      "learning_rate": 1.6033573008532015e-06,
      "loss": 0.0417,
      "step": 14337
    },
    {
      "epoch": 0.7404079524916086,
      "grad_norm": 12.70132827758789,
      "learning_rate": 1.6027560715293127e-06,
      "loss": 0.1773,
      "step": 14338
    },
    {
      "epoch": 0.7404595920475084,
      "grad_norm": 2.3137991428375244,
      "learning_rate": 1.6021549334352693e-06,
      "loss": 0.023,
      "step": 14339
    },
    {
      "epoch": 0.7405112316034083,
      "grad_norm": 4.714250564575195,
      "learning_rate": 1.6015538865872154e-06,
      "loss": 0.0655,
      "step": 14340
    },
    {
      "epoch": 0.7405628711593081,
      "grad_norm": 3.7383840084075928,
      "learning_rate": 1.6009529310012905e-06,
      "loss": 0.0383,
      "step": 14341
    },
    {
      "epoch": 0.7406145107152079,
      "grad_norm": 1.036422610282898,
      "learning_rate": 1.6003520666936333e-06,
      "loss": 0.0069,
      "step": 14342
    },
    {
      "epoch": 0.7406661502711077,
      "grad_norm": 1.8531721830368042,
      "learning_rate": 1.5997512936803794e-06,
      "loss": 0.0397,
      "step": 14343
    },
    {
      "epoch": 0.7407177898270075,
      "grad_norm": 1.366878628730774,
      "learning_rate": 1.5991506119776617e-06,
      "loss": 0.0174,
      "step": 14344
    },
    {
      "epoch": 0.7407694293829074,
      "grad_norm": 8.542228698730469,
      "learning_rate": 1.598550021601611e-06,
      "loss": 0.0876,
      "step": 14345
    },
    {
      "epoch": 0.7408210689388072,
      "grad_norm": 5.637063026428223,
      "learning_rate": 1.5979495225683577e-06,
      "loss": 0.0817,
      "step": 14346
    },
    {
      "epoch": 0.740872708494707,
      "grad_norm": 8.855091094970703,
      "learning_rate": 1.5973491148940223e-06,
      "loss": 0.0627,
      "step": 14347
    },
    {
      "epoch": 0.7409243480506068,
      "grad_norm": 10.27849006652832,
      "learning_rate": 1.5967487985947333e-06,
      "loss": 0.1777,
      "step": 14348
    },
    {
      "epoch": 0.7409759876065066,
      "grad_norm": 6.142795562744141,
      "learning_rate": 1.5961485736866117e-06,
      "loss": 0.0504,
      "step": 14349
    },
    {
      "epoch": 0.7410276271624064,
      "grad_norm": 10.218998908996582,
      "learning_rate": 1.5955484401857723e-06,
      "loss": 0.1346,
      "step": 14350
    },
    {
      "epoch": 0.7410792667183063,
      "grad_norm": 29.53416633605957,
      "learning_rate": 1.594948398108333e-06,
      "loss": 0.1828,
      "step": 14351
    },
    {
      "epoch": 0.7411309062742061,
      "grad_norm": 8.768967628479004,
      "learning_rate": 1.5943484474704063e-06,
      "loss": 0.0797,
      "step": 14352
    },
    {
      "epoch": 0.7411825458301059,
      "grad_norm": 1.62730872631073,
      "learning_rate": 1.5937485882881076e-06,
      "loss": 0.0274,
      "step": 14353
    },
    {
      "epoch": 0.7412341853860057,
      "grad_norm": 9.665741920471191,
      "learning_rate": 1.5931488205775402e-06,
      "loss": 0.0539,
      "step": 14354
    },
    {
      "epoch": 0.7412858249419055,
      "grad_norm": 2.191194772720337,
      "learning_rate": 1.592549144354813e-06,
      "loss": 0.0166,
      "step": 14355
    },
    {
      "epoch": 0.7413374644978054,
      "grad_norm": 9.013229370117188,
      "learning_rate": 1.5919495596360297e-06,
      "loss": 0.1208,
      "step": 14356
    },
    {
      "epoch": 0.7413891040537052,
      "grad_norm": 0.9756243228912354,
      "learning_rate": 1.5913500664372905e-06,
      "loss": 0.0117,
      "step": 14357
    },
    {
      "epoch": 0.741440743609605,
      "grad_norm": 9.108814239501953,
      "learning_rate": 1.5907506647746951e-06,
      "loss": 0.0999,
      "step": 14358
    },
    {
      "epoch": 0.7414923831655048,
      "grad_norm": 2.011899948120117,
      "learning_rate": 1.5901513546643393e-06,
      "loss": 0.0342,
      "step": 14359
    },
    {
      "epoch": 0.7415440227214046,
      "grad_norm": 1.6187325716018677,
      "learning_rate": 1.5895521361223198e-06,
      "loss": 0.0144,
      "step": 14360
    },
    {
      "epoch": 0.7415956622773044,
      "grad_norm": 3.0587196350097656,
      "learning_rate": 1.5889530091647221e-06,
      "loss": 0.0261,
      "step": 14361
    },
    {
      "epoch": 0.7416473018332043,
      "grad_norm": 2.684488296508789,
      "learning_rate": 1.5883539738076397e-06,
      "loss": 0.0318,
      "step": 14362
    },
    {
      "epoch": 0.7416989413891041,
      "grad_norm": 3.318449020385742,
      "learning_rate": 1.5877550300671591e-06,
      "loss": 0.031,
      "step": 14363
    },
    {
      "epoch": 0.7417505809450039,
      "grad_norm": 4.471019268035889,
      "learning_rate": 1.5871561779593647e-06,
      "loss": 0.0549,
      "step": 14364
    },
    {
      "epoch": 0.7418022205009037,
      "grad_norm": 7.7527689933776855,
      "learning_rate": 1.5865574175003356e-06,
      "loss": 0.0655,
      "step": 14365
    },
    {
      "epoch": 0.7418538600568035,
      "grad_norm": 4.76261568069458,
      "learning_rate": 1.5859587487061506e-06,
      "loss": 0.0447,
      "step": 14366
    },
    {
      "epoch": 0.7419054996127034,
      "grad_norm": 5.207713603973389,
      "learning_rate": 1.5853601715928895e-06,
      "loss": 0.0813,
      "step": 14367
    },
    {
      "epoch": 0.7419571391686032,
      "grad_norm": 2.77255916595459,
      "learning_rate": 1.5847616861766263e-06,
      "loss": 0.0174,
      "step": 14368
    },
    {
      "epoch": 0.742008778724503,
      "grad_norm": 4.1910295486450195,
      "learning_rate": 1.5841632924734307e-06,
      "loss": 0.0422,
      "step": 14369
    },
    {
      "epoch": 0.7420604182804028,
      "grad_norm": 1.98472261428833,
      "learning_rate": 1.5835649904993727e-06,
      "loss": 0.0267,
      "step": 14370
    },
    {
      "epoch": 0.7421120578363026,
      "grad_norm": 2.8782799243927,
      "learning_rate": 1.582966780270519e-06,
      "loss": 0.0419,
      "step": 14371
    },
    {
      "epoch": 0.7421636973922024,
      "grad_norm": 3.30941104888916,
      "learning_rate": 1.5823686618029343e-06,
      "loss": 0.0505,
      "step": 14372
    },
    {
      "epoch": 0.7422153369481023,
      "grad_norm": 1.781007170677185,
      "learning_rate": 1.581770635112681e-06,
      "loss": 0.0224,
      "step": 14373
    },
    {
      "epoch": 0.7422669765040021,
      "grad_norm": 7.283477783203125,
      "learning_rate": 1.5811727002158177e-06,
      "loss": 0.0234,
      "step": 14374
    },
    {
      "epoch": 0.7423186160599019,
      "grad_norm": 14.934311866760254,
      "learning_rate": 1.5805748571284018e-06,
      "loss": 0.1911,
      "step": 14375
    },
    {
      "epoch": 0.7423702556158017,
      "grad_norm": 4.935434818267822,
      "learning_rate": 1.5799771058664882e-06,
      "loss": 0.0424,
      "step": 14376
    },
    {
      "epoch": 0.7424218951717015,
      "grad_norm": 4.373608589172363,
      "learning_rate": 1.5793794464461276e-06,
      "loss": 0.0466,
      "step": 14377
    },
    {
      "epoch": 0.7424735347276014,
      "grad_norm": 1.255223274230957,
      "learning_rate": 1.5787818788833714e-06,
      "loss": 0.009,
      "step": 14378
    },
    {
      "epoch": 0.7425251742835012,
      "grad_norm": 3.2670044898986816,
      "learning_rate": 1.5781844031942673e-06,
      "loss": 0.047,
      "step": 14379
    },
    {
      "epoch": 0.742576813839401,
      "grad_norm": 4.092288970947266,
      "learning_rate": 1.577587019394855e-06,
      "loss": 0.0563,
      "step": 14380
    },
    {
      "epoch": 0.7426284533953008,
      "grad_norm": 3.2750022411346436,
      "learning_rate": 1.576989727501182e-06,
      "loss": 0.0498,
      "step": 14381
    },
    {
      "epoch": 0.7426800929512006,
      "grad_norm": 21.339271545410156,
      "learning_rate": 1.576392527529288e-06,
      "loss": 0.0635,
      "step": 14382
    },
    {
      "epoch": 0.7427317325071005,
      "grad_norm": 3.292558431625366,
      "learning_rate": 1.5757954194952068e-06,
      "loss": 0.0492,
      "step": 14383
    },
    {
      "epoch": 0.7427833720630003,
      "grad_norm": 3.5140974521636963,
      "learning_rate": 1.5751984034149748e-06,
      "loss": 0.0229,
      "step": 14384
    },
    {
      "epoch": 0.7428350116189001,
      "grad_norm": 12.326385498046875,
      "learning_rate": 1.5746014793046228e-06,
      "loss": 0.1401,
      "step": 14385
    },
    {
      "epoch": 0.7428866511747999,
      "grad_norm": 3.922874927520752,
      "learning_rate": 1.5740046471801856e-06,
      "loss": 0.077,
      "step": 14386
    },
    {
      "epoch": 0.7429382907306997,
      "grad_norm": 4.973105430603027,
      "learning_rate": 1.573407907057685e-06,
      "loss": 0.0834,
      "step": 14387
    },
    {
      "epoch": 0.7429899302865995,
      "grad_norm": 17.21048355102539,
      "learning_rate": 1.5728112589531492e-06,
      "loss": 0.1159,
      "step": 14388
    },
    {
      "epoch": 0.7430415698424994,
      "grad_norm": 3.3945748805999756,
      "learning_rate": 1.5722147028825985e-06,
      "loss": 0.0517,
      "step": 14389
    },
    {
      "epoch": 0.7430932093983992,
      "grad_norm": 6.509049415588379,
      "learning_rate": 1.5716182388620542e-06,
      "loss": 0.1101,
      "step": 14390
    },
    {
      "epoch": 0.743144848954299,
      "grad_norm": 4.506285190582275,
      "learning_rate": 1.571021866907534e-06,
      "loss": 0.0765,
      "step": 14391
    },
    {
      "epoch": 0.7431964885101988,
      "grad_norm": 3.9054055213928223,
      "learning_rate": 1.5704255870350521e-06,
      "loss": 0.039,
      "step": 14392
    },
    {
      "epoch": 0.7432481280660986,
      "grad_norm": 5.111131191253662,
      "learning_rate": 1.5698293992606213e-06,
      "loss": 0.0428,
      "step": 14393
    },
    {
      "epoch": 0.7432997676219985,
      "grad_norm": 3.319300413131714,
      "learning_rate": 1.5692333036002515e-06,
      "loss": 0.0246,
      "step": 14394
    },
    {
      "epoch": 0.7433514071778983,
      "grad_norm": 1.7438738346099854,
      "learning_rate": 1.5686373000699506e-06,
      "loss": 0.0333,
      "step": 14395
    },
    {
      "epoch": 0.7434030467337981,
      "grad_norm": 8.55813980102539,
      "learning_rate": 1.5680413886857243e-06,
      "loss": 0.0459,
      "step": 14396
    },
    {
      "epoch": 0.7434546862896979,
      "grad_norm": 3.382472038269043,
      "learning_rate": 1.5674455694635755e-06,
      "loss": 0.0427,
      "step": 14397
    },
    {
      "epoch": 0.7435063258455977,
      "grad_norm": 3.924553155899048,
      "learning_rate": 1.5668498424195023e-06,
      "loss": 0.0583,
      "step": 14398
    },
    {
      "epoch": 0.7435579654014975,
      "grad_norm": 3.482333183288574,
      "learning_rate": 1.5662542075695015e-06,
      "loss": 0.0437,
      "step": 14399
    },
    {
      "epoch": 0.7436096049573974,
      "grad_norm": 4.880770683288574,
      "learning_rate": 1.5656586649295724e-06,
      "loss": 0.0663,
      "step": 14400
    },
    {
      "epoch": 0.7436612445132972,
      "grad_norm": 5.700287818908691,
      "learning_rate": 1.5650632145157074e-06,
      "loss": 0.0616,
      "step": 14401
    },
    {
      "epoch": 0.743712884069197,
      "grad_norm": 2.8597829341888428,
      "learning_rate": 1.5644678563438931e-06,
      "loss": 0.0418,
      "step": 14402
    },
    {
      "epoch": 0.7437645236250968,
      "grad_norm": 10.248821258544922,
      "learning_rate": 1.5638725904301194e-06,
      "loss": 0.1283,
      "step": 14403
    },
    {
      "epoch": 0.7438161631809966,
      "grad_norm": 2.7148425579071045,
      "learning_rate": 1.563277416790372e-06,
      "loss": 0.0421,
      "step": 14404
    },
    {
      "epoch": 0.7438678027368965,
      "grad_norm": 2.553234577178955,
      "learning_rate": 1.562682335440633e-06,
      "loss": 0.0355,
      "step": 14405
    },
    {
      "epoch": 0.7439194422927963,
      "grad_norm": 2.395580291748047,
      "learning_rate": 1.5620873463968827e-06,
      "loss": 0.0405,
      "step": 14406
    },
    {
      "epoch": 0.7439710818486961,
      "grad_norm": 3.633910655975342,
      "learning_rate": 1.5614924496751e-06,
      "loss": 0.0409,
      "step": 14407
    },
    {
      "epoch": 0.7440227214045959,
      "grad_norm": 2.1688663959503174,
      "learning_rate": 1.5608976452912588e-06,
      "loss": 0.0222,
      "step": 14408
    },
    {
      "epoch": 0.7440743609604957,
      "grad_norm": 2.3016409873962402,
      "learning_rate": 1.5603029332613334e-06,
      "loss": 0.014,
      "step": 14409
    },
    {
      "epoch": 0.7441260005163955,
      "grad_norm": 4.1121907234191895,
      "learning_rate": 1.5597083136012935e-06,
      "loss": 0.0362,
      "step": 14410
    },
    {
      "epoch": 0.7441776400722954,
      "grad_norm": 9.272936820983887,
      "learning_rate": 1.5591137863271072e-06,
      "loss": 0.0898,
      "step": 14411
    },
    {
      "epoch": 0.7442292796281952,
      "grad_norm": 4.591395378112793,
      "learning_rate": 1.5585193514547415e-06,
      "loss": 0.0474,
      "step": 14412
    },
    {
      "epoch": 0.744280919184095,
      "grad_norm": 10.980083465576172,
      "learning_rate": 1.5579250090001547e-06,
      "loss": 0.159,
      "step": 14413
    },
    {
      "epoch": 0.7443325587399948,
      "grad_norm": 5.457966327667236,
      "learning_rate": 1.5573307589793125e-06,
      "loss": 0.0759,
      "step": 14414
    },
    {
      "epoch": 0.7443841982958946,
      "grad_norm": 17.23033332824707,
      "learning_rate": 1.5567366014081724e-06,
      "loss": 0.2768,
      "step": 14415
    },
    {
      "epoch": 0.7444358378517945,
      "grad_norm": 7.619363307952881,
      "learning_rate": 1.5561425363026865e-06,
      "loss": 0.0577,
      "step": 14416
    },
    {
      "epoch": 0.7444874774076943,
      "grad_norm": 2.660393476486206,
      "learning_rate": 1.5555485636788104e-06,
      "loss": 0.0435,
      "step": 14417
    },
    {
      "epoch": 0.7445391169635941,
      "grad_norm": 2.7852888107299805,
      "learning_rate": 1.5549546835524927e-06,
      "loss": 0.0428,
      "step": 14418
    },
    {
      "epoch": 0.7445907565194939,
      "grad_norm": 2.6209888458251953,
      "learning_rate": 1.5543608959396855e-06,
      "loss": 0.0335,
      "step": 14419
    },
    {
      "epoch": 0.7446423960753937,
      "grad_norm": 1.271025538444519,
      "learning_rate": 1.553767200856331e-06,
      "loss": 0.0153,
      "step": 14420
    },
    {
      "epoch": 0.7446940356312935,
      "grad_norm": 2.7938387393951416,
      "learning_rate": 1.5531735983183727e-06,
      "loss": 0.0239,
      "step": 14421
    },
    {
      "epoch": 0.7447456751871934,
      "grad_norm": 5.4370574951171875,
      "learning_rate": 1.552580088341752e-06,
      "loss": 0.0468,
      "step": 14422
    },
    {
      "epoch": 0.7447973147430932,
      "grad_norm": 8.799614906311035,
      "learning_rate": 1.551986670942407e-06,
      "loss": 0.0371,
      "step": 14423
    },
    {
      "epoch": 0.744848954298993,
      "grad_norm": 5.117079734802246,
      "learning_rate": 1.5513933461362729e-06,
      "loss": 0.0234,
      "step": 14424
    },
    {
      "epoch": 0.7449005938548928,
      "grad_norm": 13.204034805297852,
      "learning_rate": 1.550800113939283e-06,
      "loss": 0.044,
      "step": 14425
    },
    {
      "epoch": 0.7449522334107926,
      "grad_norm": 3.7598483562469482,
      "learning_rate": 1.5502069743673687e-06,
      "loss": 0.0584,
      "step": 14426
    },
    {
      "epoch": 0.7450038729666925,
      "grad_norm": 7.750575542449951,
      "learning_rate": 1.5496139274364574e-06,
      "loss": 0.1867,
      "step": 14427
    },
    {
      "epoch": 0.7450555125225923,
      "grad_norm": 6.868217468261719,
      "learning_rate": 1.5490209731624756e-06,
      "loss": 0.0273,
      "step": 14428
    },
    {
      "epoch": 0.7451071520784921,
      "grad_norm": 3.2645695209503174,
      "learning_rate": 1.548428111561346e-06,
      "loss": 0.0573,
      "step": 14429
    },
    {
      "epoch": 0.7451587916343919,
      "grad_norm": 2.4956014156341553,
      "learning_rate": 1.5478353426489906e-06,
      "loss": 0.0518,
      "step": 14430
    },
    {
      "epoch": 0.7452104311902917,
      "grad_norm": 5.611934185028076,
      "learning_rate": 1.5472426664413249e-06,
      "loss": 0.0483,
      "step": 14431
    },
    {
      "epoch": 0.7452620707461916,
      "grad_norm": 2.3321008682250977,
      "learning_rate": 1.5466500829542651e-06,
      "loss": 0.0432,
      "step": 14432
    },
    {
      "epoch": 0.7453137103020914,
      "grad_norm": 2.7290096282958984,
      "learning_rate": 1.546057592203727e-06,
      "loss": 0.0301,
      "step": 14433
    },
    {
      "epoch": 0.7453653498579912,
      "grad_norm": 3.102400541305542,
      "learning_rate": 1.5454651942056215e-06,
      "loss": 0.0286,
      "step": 14434
    },
    {
      "epoch": 0.745416989413891,
      "grad_norm": 5.551543235778809,
      "learning_rate": 1.544872888975854e-06,
      "loss": 0.0639,
      "step": 14435
    },
    {
      "epoch": 0.7454686289697908,
      "grad_norm": 2.8570473194122314,
      "learning_rate": 1.544280676530332e-06,
      "loss": 0.0195,
      "step": 14436
    },
    {
      "epoch": 0.7455202685256906,
      "grad_norm": 3.0979671478271484,
      "learning_rate": 1.543688556884958e-06,
      "loss": 0.047,
      "step": 14437
    },
    {
      "epoch": 0.7455719080815905,
      "grad_norm": 5.953902721405029,
      "learning_rate": 1.5430965300556338e-06,
      "loss": 0.0332,
      "step": 14438
    },
    {
      "epoch": 0.7456235476374903,
      "grad_norm": 2.0094428062438965,
      "learning_rate": 1.5425045960582574e-06,
      "loss": 0.0299,
      "step": 14439
    },
    {
      "epoch": 0.7456751871933901,
      "grad_norm": 5.359470367431641,
      "learning_rate": 1.5419127549087244e-06,
      "loss": 0.08,
      "step": 14440
    },
    {
      "epoch": 0.7457268267492899,
      "grad_norm": 11.263044357299805,
      "learning_rate": 1.5413210066229279e-06,
      "loss": 0.1555,
      "step": 14441
    },
    {
      "epoch": 0.7457784663051897,
      "grad_norm": 4.6710028648376465,
      "learning_rate": 1.5407293512167597e-06,
      "loss": 0.0876,
      "step": 14442
    },
    {
      "epoch": 0.7458301058610896,
      "grad_norm": 14.979790687561035,
      "learning_rate": 1.5401377887061076e-06,
      "loss": 0.085,
      "step": 14443
    },
    {
      "epoch": 0.7458817454169894,
      "grad_norm": 0.7957204580307007,
      "learning_rate": 1.539546319106857e-06,
      "loss": 0.007,
      "step": 14444
    },
    {
      "epoch": 0.7459333849728892,
      "grad_norm": 3.165898561477661,
      "learning_rate": 1.5389549424348926e-06,
      "loss": 0.0664,
      "step": 14445
    },
    {
      "epoch": 0.745985024528789,
      "grad_norm": 2.6899280548095703,
      "learning_rate": 1.5383636587060941e-06,
      "loss": 0.0418,
      "step": 14446
    },
    {
      "epoch": 0.7460366640846888,
      "grad_norm": 3.3496785163879395,
      "learning_rate": 1.5377724679363397e-06,
      "loss": 0.0421,
      "step": 14447
    },
    {
      "epoch": 0.7460883036405886,
      "grad_norm": 10.186057090759277,
      "learning_rate": 1.537181370141508e-06,
      "loss": 0.1502,
      "step": 14448
    },
    {
      "epoch": 0.7461399431964885,
      "grad_norm": 17.379358291625977,
      "learning_rate": 1.5365903653374686e-06,
      "loss": 0.1496,
      "step": 14449
    },
    {
      "epoch": 0.7461915827523883,
      "grad_norm": 2.5740256309509277,
      "learning_rate": 1.5359994535400934e-06,
      "loss": 0.0289,
      "step": 14450
    },
    {
      "epoch": 0.7462432223082881,
      "grad_norm": 12.379629135131836,
      "learning_rate": 1.5354086347652504e-06,
      "loss": 0.0754,
      "step": 14451
    },
    {
      "epoch": 0.7462948618641879,
      "grad_norm": 5.081248760223389,
      "learning_rate": 1.534817909028809e-06,
      "loss": 0.0557,
      "step": 14452
    },
    {
      "epoch": 0.7463465014200877,
      "grad_norm": 5.910642147064209,
      "learning_rate": 1.5342272763466287e-06,
      "loss": 0.0467,
      "step": 14453
    },
    {
      "epoch": 0.7463981409759876,
      "grad_norm": 4.001123905181885,
      "learning_rate": 1.5336367367345723e-06,
      "loss": 0.0437,
      "step": 14454
    },
    {
      "epoch": 0.7464497805318874,
      "grad_norm": 5.944972515106201,
      "learning_rate": 1.533046290208497e-06,
      "loss": 0.0668,
      "step": 14455
    },
    {
      "epoch": 0.7465014200877872,
      "grad_norm": 7.562441825866699,
      "learning_rate": 1.5324559367842595e-06,
      "loss": 0.043,
      "step": 14456
    },
    {
      "epoch": 0.746553059643687,
      "grad_norm": 2.152076482772827,
      "learning_rate": 1.531865676477713e-06,
      "loss": 0.0227,
      "step": 14457
    },
    {
      "epoch": 0.7466046991995868,
      "grad_norm": 3.65081787109375,
      "learning_rate": 1.5312755093047083e-06,
      "loss": 0.056,
      "step": 14458
    },
    {
      "epoch": 0.7466563387554866,
      "grad_norm": 4.620945930480957,
      "learning_rate": 1.530685435281094e-06,
      "loss": 0.0372,
      "step": 14459
    },
    {
      "epoch": 0.7467079783113865,
      "grad_norm": 10.762818336486816,
      "learning_rate": 1.5300954544227158e-06,
      "loss": 0.0562,
      "step": 14460
    },
    {
      "epoch": 0.7467596178672864,
      "grad_norm": 11.764421463012695,
      "learning_rate": 1.5295055667454178e-06,
      "loss": 0.0825,
      "step": 14461
    },
    {
      "epoch": 0.7468112574231862,
      "grad_norm": 15.577471733093262,
      "learning_rate": 1.52891577226504e-06,
      "loss": 0.0357,
      "step": 14462
    },
    {
      "epoch": 0.746862896979086,
      "grad_norm": 4.980173110961914,
      "learning_rate": 1.5283260709974223e-06,
      "loss": 0.0445,
      "step": 14463
    },
    {
      "epoch": 0.7469145365349859,
      "grad_norm": 5.072155475616455,
      "learning_rate": 1.5277364629583963e-06,
      "loss": 0.0431,
      "step": 14464
    },
    {
      "epoch": 0.7469661760908857,
      "grad_norm": 2.6254494190216064,
      "learning_rate": 1.5271469481638e-06,
      "loss": 0.0315,
      "step": 14465
    },
    {
      "epoch": 0.7470178156467855,
      "grad_norm": 3.4181978702545166,
      "learning_rate": 1.526557526629463e-06,
      "loss": 0.0259,
      "step": 14466
    },
    {
      "epoch": 0.7470694552026853,
      "grad_norm": 7.98403263092041,
      "learning_rate": 1.5259681983712149e-06,
      "loss": 0.0527,
      "step": 14467
    },
    {
      "epoch": 0.7471210947585851,
      "grad_norm": 1.158977746963501,
      "learning_rate": 1.5253789634048782e-06,
      "loss": 0.0136,
      "step": 14468
    },
    {
      "epoch": 0.7471727343144849,
      "grad_norm": 9.45605754852295,
      "learning_rate": 1.5247898217462782e-06,
      "loss": 0.1066,
      "step": 14469
    },
    {
      "epoch": 0.7472243738703848,
      "grad_norm": 21.117267608642578,
      "learning_rate": 1.5242007734112357e-06,
      "loss": 0.1185,
      "step": 14470
    },
    {
      "epoch": 0.7472760134262846,
      "grad_norm": 4.647895812988281,
      "learning_rate": 1.523611818415569e-06,
      "loss": 0.0633,
      "step": 14471
    },
    {
      "epoch": 0.7473276529821844,
      "grad_norm": 4.899588584899902,
      "learning_rate": 1.5230229567750937e-06,
      "loss": 0.0601,
      "step": 14472
    },
    {
      "epoch": 0.7473792925380842,
      "grad_norm": 7.718667507171631,
      "learning_rate": 1.5224341885056238e-06,
      "loss": 0.0758,
      "step": 14473
    },
    {
      "epoch": 0.747430932093984,
      "grad_norm": 13.235735893249512,
      "learning_rate": 1.5218455136229699e-06,
      "loss": 0.124,
      "step": 14474
    },
    {
      "epoch": 0.7474825716498839,
      "grad_norm": 4.404795169830322,
      "learning_rate": 1.5212569321429393e-06,
      "loss": 0.0365,
      "step": 14475
    },
    {
      "epoch": 0.7475342112057837,
      "grad_norm": 7.661112308502197,
      "learning_rate": 1.5206684440813397e-06,
      "loss": 0.046,
      "step": 14476
    },
    {
      "epoch": 0.7475858507616835,
      "grad_norm": 4.74738073348999,
      "learning_rate": 1.5200800494539724e-06,
      "loss": 0.0299,
      "step": 14477
    },
    {
      "epoch": 0.7476374903175833,
      "grad_norm": 4.432816982269287,
      "learning_rate": 1.5194917482766396e-06,
      "loss": 0.0412,
      "step": 14478
    },
    {
      "epoch": 0.7476891298734831,
      "grad_norm": 3.310445785522461,
      "learning_rate": 1.5189035405651392e-06,
      "loss": 0.0198,
      "step": 14479
    },
    {
      "epoch": 0.747740769429383,
      "grad_norm": 2.5791735649108887,
      "learning_rate": 1.5183154263352666e-06,
      "loss": 0.034,
      "step": 14480
    },
    {
      "epoch": 0.7477924089852828,
      "grad_norm": 2.3163158893585205,
      "learning_rate": 1.517727405602817e-06,
      "loss": 0.0292,
      "step": 14481
    },
    {
      "epoch": 0.7478440485411826,
      "grad_norm": 9.293328285217285,
      "learning_rate": 1.517139478383578e-06,
      "loss": 0.0512,
      "step": 14482
    },
    {
      "epoch": 0.7478956880970824,
      "grad_norm": 9.030403137207031,
      "learning_rate": 1.5165516446933377e-06,
      "loss": 0.093,
      "step": 14483
    },
    {
      "epoch": 0.7479473276529822,
      "grad_norm": 1.4352412223815918,
      "learning_rate": 1.5159639045478852e-06,
      "loss": 0.0216,
      "step": 14484
    },
    {
      "epoch": 0.747998967208882,
      "grad_norm": 3.076262950897217,
      "learning_rate": 1.5153762579630037e-06,
      "loss": 0.0239,
      "step": 14485
    },
    {
      "epoch": 0.7480506067647819,
      "grad_norm": 3.0601143836975098,
      "learning_rate": 1.5147887049544708e-06,
      "loss": 0.0268,
      "step": 14486
    },
    {
      "epoch": 0.7481022463206817,
      "grad_norm": 2.601926326751709,
      "learning_rate": 1.5142012455380656e-06,
      "loss": 0.018,
      "step": 14487
    },
    {
      "epoch": 0.7481538858765815,
      "grad_norm": 4.4953932762146,
      "learning_rate": 1.5136138797295646e-06,
      "loss": 0.0612,
      "step": 14488
    },
    {
      "epoch": 0.7482055254324813,
      "grad_norm": 5.668195724487305,
      "learning_rate": 1.5130266075447402e-06,
      "loss": 0.0321,
      "step": 14489
    },
    {
      "epoch": 0.7482571649883811,
      "grad_norm": 2.3650918006896973,
      "learning_rate": 1.5124394289993638e-06,
      "loss": 0.0431,
      "step": 14490
    },
    {
      "epoch": 0.748308804544281,
      "grad_norm": 4.6339287757873535,
      "learning_rate": 1.5118523441092037e-06,
      "loss": 0.0267,
      "step": 14491
    },
    {
      "epoch": 0.7483604441001808,
      "grad_norm": 4.021702766418457,
      "learning_rate": 1.5112653528900245e-06,
      "loss": 0.048,
      "step": 14492
    },
    {
      "epoch": 0.7484120836560806,
      "grad_norm": 4.328675270080566,
      "learning_rate": 1.5106784553575899e-06,
      "loss": 0.0268,
      "step": 14493
    },
    {
      "epoch": 0.7484637232119804,
      "grad_norm": 2.0029990673065186,
      "learning_rate": 1.5100916515276603e-06,
      "loss": 0.0369,
      "step": 14494
    },
    {
      "epoch": 0.7485153627678802,
      "grad_norm": 2.7821156978607178,
      "learning_rate": 1.5095049414159945e-06,
      "loss": 0.0466,
      "step": 14495
    },
    {
      "epoch": 0.74856700232378,
      "grad_norm": 9.202781677246094,
      "learning_rate": 1.5089183250383487e-06,
      "loss": 0.0732,
      "step": 14496
    },
    {
      "epoch": 0.7486186418796799,
      "grad_norm": 5.552381992340088,
      "learning_rate": 1.5083318024104721e-06,
      "loss": 0.0684,
      "step": 14497
    },
    {
      "epoch": 0.7486702814355797,
      "grad_norm": 3.8645808696746826,
      "learning_rate": 1.5077453735481185e-06,
      "loss": 0.0286,
      "step": 14498
    },
    {
      "epoch": 0.7487219209914795,
      "grad_norm": 3.5456442832946777,
      "learning_rate": 1.5071590384670377e-06,
      "loss": 0.0446,
      "step": 14499
    },
    {
      "epoch": 0.7487735605473793,
      "grad_norm": 7.009406566619873,
      "learning_rate": 1.5065727971829708e-06,
      "loss": 0.0437,
      "step": 14500
    },
    {
      "epoch": 0.7488252001032791,
      "grad_norm": 2.6234383583068848,
      "learning_rate": 1.5059866497116627e-06,
      "loss": 0.0254,
      "step": 14501
    },
    {
      "epoch": 0.748876839659179,
      "grad_norm": 9.812936782836914,
      "learning_rate": 1.5054005960688522e-06,
      "loss": 0.0977,
      "step": 14502
    },
    {
      "epoch": 0.7489284792150788,
      "grad_norm": 5.346309661865234,
      "learning_rate": 1.504814636270282e-06,
      "loss": 0.038,
      "step": 14503
    },
    {
      "epoch": 0.7489801187709786,
      "grad_norm": 19.0749568939209,
      "learning_rate": 1.5042287703316832e-06,
      "loss": 0.0906,
      "step": 14504
    },
    {
      "epoch": 0.7490317583268784,
      "grad_norm": 4.1533894538879395,
      "learning_rate": 1.5036429982687894e-06,
      "loss": 0.077,
      "step": 14505
    },
    {
      "epoch": 0.7490833978827782,
      "grad_norm": 1.7601701021194458,
      "learning_rate": 1.503057320097332e-06,
      "loss": 0.0119,
      "step": 14506
    },
    {
      "epoch": 0.749135037438678,
      "grad_norm": 5.912764549255371,
      "learning_rate": 1.5024717358330381e-06,
      "loss": 0.0963,
      "step": 14507
    },
    {
      "epoch": 0.7491866769945779,
      "grad_norm": 7.000924587249756,
      "learning_rate": 1.5018862454916333e-06,
      "loss": 0.0371,
      "step": 14508
    },
    {
      "epoch": 0.7492383165504777,
      "grad_norm": 14.586352348327637,
      "learning_rate": 1.5013008490888403e-06,
      "loss": 0.0675,
      "step": 14509
    },
    {
      "epoch": 0.7492899561063775,
      "grad_norm": 4.299375057220459,
      "learning_rate": 1.5007155466403806e-06,
      "loss": 0.0447,
      "step": 14510
    },
    {
      "epoch": 0.7493415956622773,
      "grad_norm": 7.335858345031738,
      "learning_rate": 1.5001303381619681e-06,
      "loss": 0.0499,
      "step": 14511
    },
    {
      "epoch": 0.7493932352181771,
      "grad_norm": 5.579354286193848,
      "learning_rate": 1.499545223669322e-06,
      "loss": 0.051,
      "step": 14512
    },
    {
      "epoch": 0.749444874774077,
      "grad_norm": 4.495391368865967,
      "learning_rate": 1.4989602031781536e-06,
      "loss": 0.0445,
      "step": 14513
    },
    {
      "epoch": 0.7494965143299768,
      "grad_norm": 3.555844783782959,
      "learning_rate": 1.4983752767041754e-06,
      "loss": 0.0506,
      "step": 14514
    },
    {
      "epoch": 0.7495481538858766,
      "grad_norm": 15.24841594696045,
      "learning_rate": 1.497790444263091e-06,
      "loss": 0.0225,
      "step": 14515
    },
    {
      "epoch": 0.7495997934417764,
      "grad_norm": 3.916524648666382,
      "learning_rate": 1.4972057058706051e-06,
      "loss": 0.0371,
      "step": 14516
    },
    {
      "epoch": 0.7496514329976762,
      "grad_norm": 3.1961183547973633,
      "learning_rate": 1.496621061542425e-06,
      "loss": 0.0266,
      "step": 14517
    },
    {
      "epoch": 0.749703072553576,
      "grad_norm": 3.454400062561035,
      "learning_rate": 1.4960365112942499e-06,
      "loss": 0.0418,
      "step": 14518
    },
    {
      "epoch": 0.7497547121094759,
      "grad_norm": 19.53978157043457,
      "learning_rate": 1.495452055141774e-06,
      "loss": 0.0872,
      "step": 14519
    },
    {
      "epoch": 0.7498063516653757,
      "grad_norm": 4.47865629196167,
      "learning_rate": 1.4948676931006939e-06,
      "loss": 0.0974,
      "step": 14520
    },
    {
      "epoch": 0.7498579912212755,
      "grad_norm": 4.215468406677246,
      "learning_rate": 1.4942834251867027e-06,
      "loss": 0.0395,
      "step": 14521
    },
    {
      "epoch": 0.7499096307771753,
      "grad_norm": 3.061479091644287,
      "learning_rate": 1.4936992514154901e-06,
      "loss": 0.0305,
      "step": 14522
    },
    {
      "epoch": 0.7499612703330751,
      "grad_norm": 13.497661590576172,
      "learning_rate": 1.4931151718027437e-06,
      "loss": 0.1403,
      "step": 14523
    },
    {
      "epoch": 0.750012909888975,
      "grad_norm": 2.390442132949829,
      "learning_rate": 1.4925311863641479e-06,
      "loss": 0.0309,
      "step": 14524
    },
    {
      "epoch": 0.7500645494448748,
      "grad_norm": 2.1977944374084473,
      "learning_rate": 1.4919472951153852e-06,
      "loss": 0.022,
      "step": 14525
    },
    {
      "epoch": 0.7501161890007746,
      "grad_norm": 1.285332202911377,
      "learning_rate": 1.4913634980721358e-06,
      "loss": 0.0242,
      "step": 14526
    },
    {
      "epoch": 0.7501678285566744,
      "grad_norm": 3.1586732864379883,
      "learning_rate": 1.4907797952500768e-06,
      "loss": 0.0373,
      "step": 14527
    },
    {
      "epoch": 0.7502194681125742,
      "grad_norm": 18.486631393432617,
      "learning_rate": 1.4901961866648835e-06,
      "loss": 0.1956,
      "step": 14528
    },
    {
      "epoch": 0.750271107668474,
      "grad_norm": 2.637002468109131,
      "learning_rate": 1.489612672332229e-06,
      "loss": 0.0365,
      "step": 14529
    },
    {
      "epoch": 0.7503227472243739,
      "grad_norm": 5.521708011627197,
      "learning_rate": 1.489029252267779e-06,
      "loss": 0.0652,
      "step": 14530
    },
    {
      "epoch": 0.7503743867802737,
      "grad_norm": 1.9469698667526245,
      "learning_rate": 1.4884459264872048e-06,
      "loss": 0.0248,
      "step": 14531
    },
    {
      "epoch": 0.7504260263361735,
      "grad_norm": 7.1596245765686035,
      "learning_rate": 1.4878626950061713e-06,
      "loss": 0.0483,
      "step": 14532
    },
    {
      "epoch": 0.7504776658920733,
      "grad_norm": 10.329859733581543,
      "learning_rate": 1.4872795578403376e-06,
      "loss": 0.2442,
      "step": 14533
    },
    {
      "epoch": 0.7505293054479731,
      "grad_norm": 7.017910480499268,
      "learning_rate": 1.4866965150053647e-06,
      "loss": 0.0537,
      "step": 14534
    },
    {
      "epoch": 0.750580945003873,
      "grad_norm": 4.38223934173584,
      "learning_rate": 1.4861135665169086e-06,
      "loss": 0.0387,
      "step": 14535
    },
    {
      "epoch": 0.7506325845597728,
      "grad_norm": 4.276043891906738,
      "learning_rate": 1.4855307123906277e-06,
      "loss": 0.0744,
      "step": 14536
    },
    {
      "epoch": 0.7506842241156726,
      "grad_norm": 9.359420776367188,
      "learning_rate": 1.48494795264217e-06,
      "loss": 0.2223,
      "step": 14537
    },
    {
      "epoch": 0.7507358636715724,
      "grad_norm": 10.602059364318848,
      "learning_rate": 1.4843652872871866e-06,
      "loss": 0.0645,
      "step": 14538
    },
    {
      "epoch": 0.7507875032274722,
      "grad_norm": 2.6280195713043213,
      "learning_rate": 1.4837827163413239e-06,
      "loss": 0.0448,
      "step": 14539
    },
    {
      "epoch": 0.750839142783372,
      "grad_norm": 3.2735836505889893,
      "learning_rate": 1.4832002398202266e-06,
      "loss": 0.0374,
      "step": 14540
    },
    {
      "epoch": 0.7508907823392719,
      "grad_norm": 2.8677165508270264,
      "learning_rate": 1.4826178577395367e-06,
      "loss": 0.0214,
      "step": 14541
    },
    {
      "epoch": 0.7509424218951717,
      "grad_norm": 2.1881837844848633,
      "learning_rate": 1.482035570114893e-06,
      "loss": 0.0233,
      "step": 14542
    },
    {
      "epoch": 0.7509940614510715,
      "grad_norm": 5.071732997894287,
      "learning_rate": 1.4814533769619344e-06,
      "loss": 0.0387,
      "step": 14543
    },
    {
      "epoch": 0.7510457010069713,
      "grad_norm": 1.9483460187911987,
      "learning_rate": 1.480871278296291e-06,
      "loss": 0.0291,
      "step": 14544
    },
    {
      "epoch": 0.7510973405628711,
      "grad_norm": 6.717279434204102,
      "learning_rate": 1.4802892741335973e-06,
      "loss": 0.0303,
      "step": 14545
    },
    {
      "epoch": 0.751148980118771,
      "grad_norm": 3.9653160572052,
      "learning_rate": 1.4797073644894828e-06,
      "loss": 0.042,
      "step": 14546
    },
    {
      "epoch": 0.7512006196746708,
      "grad_norm": 3.945692300796509,
      "learning_rate": 1.4791255493795754e-06,
      "loss": 0.0273,
      "step": 14547
    },
    {
      "epoch": 0.7512522592305706,
      "grad_norm": 7.447824478149414,
      "learning_rate": 1.4785438288194947e-06,
      "loss": 0.0569,
      "step": 14548
    },
    {
      "epoch": 0.7513038987864704,
      "grad_norm": 6.341883182525635,
      "learning_rate": 1.4779622028248642e-06,
      "loss": 0.0663,
      "step": 14549
    },
    {
      "epoch": 0.7513555383423702,
      "grad_norm": 6.598250865936279,
      "learning_rate": 1.477380671411305e-06,
      "loss": 0.0777,
      "step": 14550
    },
    {
      "epoch": 0.75140717789827,
      "grad_norm": 11.954680442810059,
      "learning_rate": 1.4767992345944337e-06,
      "loss": 0.0548,
      "step": 14551
    },
    {
      "epoch": 0.7514588174541699,
      "grad_norm": 2.7104599475860596,
      "learning_rate": 1.4762178923898607e-06,
      "loss": 0.0387,
      "step": 14552
    },
    {
      "epoch": 0.7515104570100697,
      "grad_norm": 7.57103157043457,
      "learning_rate": 1.4756366448131998e-06,
      "loss": 0.076,
      "step": 14553
    },
    {
      "epoch": 0.7515620965659695,
      "grad_norm": 2.0078024864196777,
      "learning_rate": 1.4750554918800597e-06,
      "loss": 0.0273,
      "step": 14554
    },
    {
      "epoch": 0.7516137361218693,
      "grad_norm": 20.721620559692383,
      "learning_rate": 1.4744744336060463e-06,
      "loss": 0.1312,
      "step": 14555
    },
    {
      "epoch": 0.7516653756777691,
      "grad_norm": 7.127549171447754,
      "learning_rate": 1.4738934700067636e-06,
      "loss": 0.048,
      "step": 14556
    },
    {
      "epoch": 0.751717015233669,
      "grad_norm": 5.129033088684082,
      "learning_rate": 1.4733126010978126e-06,
      "loss": 0.0634,
      "step": 14557
    },
    {
      "epoch": 0.7517686547895688,
      "grad_norm": 2.4886999130249023,
      "learning_rate": 1.4727318268947926e-06,
      "loss": 0.0361,
      "step": 14558
    },
    {
      "epoch": 0.7518202943454686,
      "grad_norm": 3.8766028881073,
      "learning_rate": 1.4721511474132993e-06,
      "loss": 0.0483,
      "step": 14559
    },
    {
      "epoch": 0.7518719339013684,
      "grad_norm": 3.7608485221862793,
      "learning_rate": 1.4715705626689264e-06,
      "loss": 0.0584,
      "step": 14560
    },
    {
      "epoch": 0.7519235734572682,
      "grad_norm": 7.058287143707275,
      "learning_rate": 1.4709900726772653e-06,
      "loss": 0.0957,
      "step": 14561
    },
    {
      "epoch": 0.751975213013168,
      "grad_norm": 4.941324234008789,
      "learning_rate": 1.470409677453905e-06,
      "loss": 0.0498,
      "step": 14562
    },
    {
      "epoch": 0.7520268525690679,
      "grad_norm": 3.4391467571258545,
      "learning_rate": 1.4698293770144278e-06,
      "loss": 0.0536,
      "step": 14563
    },
    {
      "epoch": 0.7520784921249677,
      "grad_norm": 6.208892345428467,
      "learning_rate": 1.4692491713744222e-06,
      "loss": 0.0573,
      "step": 14564
    },
    {
      "epoch": 0.7521301316808675,
      "grad_norm": 3.3228085041046143,
      "learning_rate": 1.468669060549468e-06,
      "loss": 0.0323,
      "step": 14565
    },
    {
      "epoch": 0.7521817712367673,
      "grad_norm": 4.375753879547119,
      "learning_rate": 1.468089044555141e-06,
      "loss": 0.0648,
      "step": 14566
    },
    {
      "epoch": 0.7522334107926671,
      "grad_norm": 3.621218204498291,
      "learning_rate": 1.4675091234070183e-06,
      "loss": 0.0552,
      "step": 14567
    },
    {
      "epoch": 0.752285050348567,
      "grad_norm": 3.93056583404541,
      "learning_rate": 1.4669292971206717e-06,
      "loss": 0.0521,
      "step": 14568
    },
    {
      "epoch": 0.7523366899044668,
      "grad_norm": 4.639581680297852,
      "learning_rate": 1.4663495657116766e-06,
      "loss": 0.0335,
      "step": 14569
    },
    {
      "epoch": 0.7523883294603666,
      "grad_norm": 4.469770908355713,
      "learning_rate": 1.4657699291955963e-06,
      "loss": 0.0267,
      "step": 14570
    },
    {
      "epoch": 0.7524399690162664,
      "grad_norm": 11.090950965881348,
      "learning_rate": 1.4651903875879981e-06,
      "loss": 0.097,
      "step": 14571
    },
    {
      "epoch": 0.7524916085721662,
      "grad_norm": 11.294219970703125,
      "learning_rate": 1.4646109409044457e-06,
      "loss": 0.054,
      "step": 14572
    },
    {
      "epoch": 0.7525432481280661,
      "grad_norm": 3.483185052871704,
      "learning_rate": 1.4640315891604984e-06,
      "loss": 0.0317,
      "step": 14573
    },
    {
      "epoch": 0.7525948876839659,
      "grad_norm": 2.9625558853149414,
      "learning_rate": 1.4634523323717148e-06,
      "loss": 0.0327,
      "step": 14574
    },
    {
      "epoch": 0.7526465272398657,
      "grad_norm": 6.644806385040283,
      "learning_rate": 1.4628731705536503e-06,
      "loss": 0.0568,
      "step": 14575
    },
    {
      "epoch": 0.7526981667957655,
      "grad_norm": 2.740650177001953,
      "learning_rate": 1.4622941037218591e-06,
      "loss": 0.0557,
      "step": 14576
    },
    {
      "epoch": 0.7527498063516653,
      "grad_norm": 7.878971576690674,
      "learning_rate": 1.4617151318918876e-06,
      "loss": 0.0528,
      "step": 14577
    },
    {
      "epoch": 0.7528014459075651,
      "grad_norm": 1.8810136318206787,
      "learning_rate": 1.4611362550792867e-06,
      "loss": 0.0372,
      "step": 14578
    },
    {
      "epoch": 0.752853085463465,
      "grad_norm": 7.830737113952637,
      "learning_rate": 1.460557473299602e-06,
      "loss": 0.0686,
      "step": 14579
    },
    {
      "epoch": 0.7529047250193648,
      "grad_norm": 2.7026546001434326,
      "learning_rate": 1.4599787865683762e-06,
      "loss": 0.0506,
      "step": 14580
    },
    {
      "epoch": 0.7529563645752646,
      "grad_norm": 3.8881726264953613,
      "learning_rate": 1.4594001949011465e-06,
      "loss": 0.0461,
      "step": 14581
    },
    {
      "epoch": 0.7530080041311644,
      "grad_norm": 2.485285520553589,
      "learning_rate": 1.4588216983134507e-06,
      "loss": 0.0188,
      "step": 14582
    },
    {
      "epoch": 0.7530596436870642,
      "grad_norm": 11.737533569335938,
      "learning_rate": 1.458243296820827e-06,
      "loss": 0.0461,
      "step": 14583
    },
    {
      "epoch": 0.7531112832429641,
      "grad_norm": 4.597867012023926,
      "learning_rate": 1.4576649904388078e-06,
      "loss": 0.0366,
      "step": 14584
    },
    {
      "epoch": 0.7531629227988639,
      "grad_norm": 4.944797992706299,
      "learning_rate": 1.4570867791829201e-06,
      "loss": 0.0493,
      "step": 14585
    },
    {
      "epoch": 0.7532145623547637,
      "grad_norm": 2.6932950019836426,
      "learning_rate": 1.4565086630686926e-06,
      "loss": 0.0291,
      "step": 14586
    },
    {
      "epoch": 0.7532662019106635,
      "grad_norm": 5.043087005615234,
      "learning_rate": 1.4559306421116492e-06,
      "loss": 0.0499,
      "step": 14587
    },
    {
      "epoch": 0.7533178414665634,
      "grad_norm": 5.237117290496826,
      "learning_rate": 1.4553527163273135e-06,
      "loss": 0.1236,
      "step": 14588
    },
    {
      "epoch": 0.7533694810224633,
      "grad_norm": 5.500089645385742,
      "learning_rate": 1.4547748857312045e-06,
      "loss": 0.0766,
      "step": 14589
    },
    {
      "epoch": 0.7534211205783631,
      "grad_norm": 13.300969123840332,
      "learning_rate": 1.454197150338839e-06,
      "loss": 0.1497,
      "step": 14590
    },
    {
      "epoch": 0.7534727601342629,
      "grad_norm": 2.7384002208709717,
      "learning_rate": 1.453619510165732e-06,
      "loss": 0.036,
      "step": 14591
    },
    {
      "epoch": 0.7535243996901627,
      "grad_norm": 9.258978843688965,
      "learning_rate": 1.453041965227396e-06,
      "loss": 0.0651,
      "step": 14592
    },
    {
      "epoch": 0.7535760392460625,
      "grad_norm": 12.836589813232422,
      "learning_rate": 1.4524645155393391e-06,
      "loss": 0.0627,
      "step": 14593
    },
    {
      "epoch": 0.7536276788019624,
      "grad_norm": 3.7779476642608643,
      "learning_rate": 1.4518871611170688e-06,
      "loss": 0.0484,
      "step": 14594
    },
    {
      "epoch": 0.7536793183578622,
      "grad_norm": 14.629532814025879,
      "learning_rate": 1.4513099019760918e-06,
      "loss": 0.0587,
      "step": 14595
    },
    {
      "epoch": 0.753730957913762,
      "grad_norm": 5.192688465118408,
      "learning_rate": 1.450732738131904e-06,
      "loss": 0.0542,
      "step": 14596
    },
    {
      "epoch": 0.7537825974696618,
      "grad_norm": 19.760936737060547,
      "learning_rate": 1.4501556696000097e-06,
      "loss": 0.2244,
      "step": 14597
    },
    {
      "epoch": 0.7538342370255616,
      "grad_norm": 1.996427059173584,
      "learning_rate": 1.4495786963959058e-06,
      "loss": 0.0269,
      "step": 14598
    },
    {
      "epoch": 0.7538858765814614,
      "grad_norm": 4.588977336883545,
      "learning_rate": 1.4490018185350828e-06,
      "loss": 0.0769,
      "step": 14599
    },
    {
      "epoch": 0.7539375161373613,
      "grad_norm": 5.248540878295898,
      "learning_rate": 1.4484250360330342e-06,
      "loss": 0.1051,
      "step": 14600
    },
    {
      "epoch": 0.7539891556932611,
      "grad_norm": 12.281771659851074,
      "learning_rate": 1.4478483489052476e-06,
      "loss": 0.1143,
      "step": 14601
    },
    {
      "epoch": 0.7540407952491609,
      "grad_norm": 3.1078410148620605,
      "learning_rate": 1.4472717571672134e-06,
      "loss": 0.0315,
      "step": 14602
    },
    {
      "epoch": 0.7540924348050607,
      "grad_norm": 2.4867279529571533,
      "learning_rate": 1.446695260834411e-06,
      "loss": 0.0427,
      "step": 14603
    },
    {
      "epoch": 0.7541440743609605,
      "grad_norm": 9.763087272644043,
      "learning_rate": 1.4461188599223235e-06,
      "loss": 0.0603,
      "step": 14604
    },
    {
      "epoch": 0.7541957139168604,
      "grad_norm": 7.8836588859558105,
      "learning_rate": 1.4455425544464297e-06,
      "loss": 0.1089,
      "step": 14605
    },
    {
      "epoch": 0.7542473534727602,
      "grad_norm": 2.4103176593780518,
      "learning_rate": 1.444966344422205e-06,
      "loss": 0.0478,
      "step": 14606
    },
    {
      "epoch": 0.75429899302866,
      "grad_norm": 6.539015769958496,
      "learning_rate": 1.4443902298651236e-06,
      "loss": 0.0878,
      "step": 14607
    },
    {
      "epoch": 0.7543506325845598,
      "grad_norm": 13.582982063293457,
      "learning_rate": 1.443814210790656e-06,
      "loss": 0.0756,
      "step": 14608
    },
    {
      "epoch": 0.7544022721404596,
      "grad_norm": 3.7859606742858887,
      "learning_rate": 1.4432382872142732e-06,
      "loss": 0.0512,
      "step": 14609
    },
    {
      "epoch": 0.7544539116963594,
      "grad_norm": 21.305574417114258,
      "learning_rate": 1.442662459151436e-06,
      "loss": 0.1197,
      "step": 14610
    },
    {
      "epoch": 0.7545055512522593,
      "grad_norm": 1.7373766899108887,
      "learning_rate": 1.4420867266176114e-06,
      "loss": 0.0278,
      "step": 14611
    },
    {
      "epoch": 0.7545571908081591,
      "grad_norm": 12.905038833618164,
      "learning_rate": 1.4415110896282602e-06,
      "loss": 0.0946,
      "step": 14612
    },
    {
      "epoch": 0.7546088303640589,
      "grad_norm": 4.5355916023254395,
      "learning_rate": 1.4409355481988408e-06,
      "loss": 0.0708,
      "step": 14613
    },
    {
      "epoch": 0.7546604699199587,
      "grad_norm": 12.095227241516113,
      "learning_rate": 1.4403601023448067e-06,
      "loss": 0.0702,
      "step": 14614
    },
    {
      "epoch": 0.7547121094758585,
      "grad_norm": 3.4578628540039062,
      "learning_rate": 1.4397847520816105e-06,
      "loss": 0.0397,
      "step": 14615
    },
    {
      "epoch": 0.7547637490317584,
      "grad_norm": 13.11585807800293,
      "learning_rate": 1.439209497424706e-06,
      "loss": 0.0887,
      "step": 14616
    },
    {
      "epoch": 0.7548153885876582,
      "grad_norm": 3.5138700008392334,
      "learning_rate": 1.4386343383895413e-06,
      "loss": 0.0643,
      "step": 14617
    },
    {
      "epoch": 0.754867028143558,
      "grad_norm": 10.20073127746582,
      "learning_rate": 1.4380592749915579e-06,
      "loss": 0.0401,
      "step": 14618
    },
    {
      "epoch": 0.7549186676994578,
      "grad_norm": 13.991656303405762,
      "learning_rate": 1.4374843072462009e-06,
      "loss": 0.1639,
      "step": 14619
    },
    {
      "epoch": 0.7549703072553576,
      "grad_norm": 5.204644203186035,
      "learning_rate": 1.43690943516891e-06,
      "loss": 0.0947,
      "step": 14620
    },
    {
      "epoch": 0.7550219468112574,
      "grad_norm": 4.080178737640381,
      "learning_rate": 1.436334658775123e-06,
      "loss": 0.055,
      "step": 14621
    },
    {
      "epoch": 0.7550735863671573,
      "grad_norm": 4.223909378051758,
      "learning_rate": 1.4357599780802756e-06,
      "loss": 0.0379,
      "step": 14622
    },
    {
      "epoch": 0.7551252259230571,
      "grad_norm": 1.9589033126831055,
      "learning_rate": 1.4351853930997994e-06,
      "loss": 0.0362,
      "step": 14623
    },
    {
      "epoch": 0.7551768654789569,
      "grad_norm": 2.02340030670166,
      "learning_rate": 1.4346109038491245e-06,
      "loss": 0.0251,
      "step": 14624
    },
    {
      "epoch": 0.7552285050348567,
      "grad_norm": 2.371265172958374,
      "learning_rate": 1.4340365103436792e-06,
      "loss": 0.0364,
      "step": 14625
    },
    {
      "epoch": 0.7552801445907565,
      "grad_norm": 14.754997253417969,
      "learning_rate": 1.433462212598888e-06,
      "loss": 0.0416,
      "step": 14626
    },
    {
      "epoch": 0.7553317841466564,
      "grad_norm": 10.62291145324707,
      "learning_rate": 1.4328880106301718e-06,
      "loss": 0.0676,
      "step": 14627
    },
    {
      "epoch": 0.7553834237025562,
      "grad_norm": 5.011467933654785,
      "learning_rate": 1.4323139044529533e-06,
      "loss": 0.047,
      "step": 14628
    },
    {
      "epoch": 0.755435063258456,
      "grad_norm": 2.3212852478027344,
      "learning_rate": 1.4317398940826444e-06,
      "loss": 0.0451,
      "step": 14629
    },
    {
      "epoch": 0.7554867028143558,
      "grad_norm": 2.101935386657715,
      "learning_rate": 1.4311659795346645e-06,
      "loss": 0.0256,
      "step": 14630
    },
    {
      "epoch": 0.7555383423702556,
      "grad_norm": 2.310966730117798,
      "learning_rate": 1.430592160824425e-06,
      "loss": 0.0237,
      "step": 14631
    },
    {
      "epoch": 0.7555899819261555,
      "grad_norm": 8.626344680786133,
      "learning_rate": 1.4300184379673327e-06,
      "loss": 0.0663,
      "step": 14632
    },
    {
      "epoch": 0.7556416214820553,
      "grad_norm": 16.806488037109375,
      "learning_rate": 1.429444810978796e-06,
      "loss": 0.1363,
      "step": 14633
    },
    {
      "epoch": 0.7556932610379551,
      "grad_norm": 2.0583858489990234,
      "learning_rate": 1.428871279874217e-06,
      "loss": 0.0114,
      "step": 14634
    },
    {
      "epoch": 0.7557449005938549,
      "grad_norm": 4.581770420074463,
      "learning_rate": 1.4282978446690027e-06,
      "loss": 0.0476,
      "step": 14635
    },
    {
      "epoch": 0.7557965401497547,
      "grad_norm": 4.112488746643066,
      "learning_rate": 1.427724505378547e-06,
      "loss": 0.0611,
      "step": 14636
    },
    {
      "epoch": 0.7558481797056545,
      "grad_norm": 4.49662971496582,
      "learning_rate": 1.4271512620182481e-06,
      "loss": 0.0507,
      "step": 14637
    },
    {
      "epoch": 0.7558998192615544,
      "grad_norm": 5.328521251678467,
      "learning_rate": 1.4265781146034996e-06,
      "loss": 0.081,
      "step": 14638
    },
    {
      "epoch": 0.7559514588174542,
      "grad_norm": 2.6268372535705566,
      "learning_rate": 1.4260050631496936e-06,
      "loss": 0.0344,
      "step": 14639
    },
    {
      "epoch": 0.756003098373354,
      "grad_norm": 1.4115182161331177,
      "learning_rate": 1.4254321076722183e-06,
      "loss": 0.0186,
      "step": 14640
    },
    {
      "epoch": 0.7560547379292538,
      "grad_norm": 4.542243480682373,
      "learning_rate": 1.4248592481864604e-06,
      "loss": 0.0346,
      "step": 14641
    },
    {
      "epoch": 0.7561063774851536,
      "grad_norm": 9.480327606201172,
      "learning_rate": 1.4242864847078042e-06,
      "loss": 0.0519,
      "step": 14642
    },
    {
      "epoch": 0.7561580170410535,
      "grad_norm": 10.16129207611084,
      "learning_rate": 1.4237138172516269e-06,
      "loss": 0.0439,
      "step": 14643
    },
    {
      "epoch": 0.7562096565969533,
      "grad_norm": 12.357858657836914,
      "learning_rate": 1.423141245833311e-06,
      "loss": 0.0666,
      "step": 14644
    },
    {
      "epoch": 0.7562612961528531,
      "grad_norm": 3.0767316818237305,
      "learning_rate": 1.4225687704682317e-06,
      "loss": 0.0532,
      "step": 14645
    },
    {
      "epoch": 0.7563129357087529,
      "grad_norm": 8.64576244354248,
      "learning_rate": 1.4219963911717627e-06,
      "loss": 0.0961,
      "step": 14646
    },
    {
      "epoch": 0.7563645752646527,
      "grad_norm": 16.590412139892578,
      "learning_rate": 1.4214241079592722e-06,
      "loss": 0.2134,
      "step": 14647
    },
    {
      "epoch": 0.7564162148205525,
      "grad_norm": 5.969074249267578,
      "learning_rate": 1.4208519208461286e-06,
      "loss": 0.0641,
      "step": 14648
    },
    {
      "epoch": 0.7564678543764524,
      "grad_norm": 1.4205909967422485,
      "learning_rate": 1.4202798298477006e-06,
      "loss": 0.0092,
      "step": 14649
    },
    {
      "epoch": 0.7565194939323522,
      "grad_norm": 20.791301727294922,
      "learning_rate": 1.4197078349793503e-06,
      "loss": 0.1517,
      "step": 14650
    },
    {
      "epoch": 0.756571133488252,
      "grad_norm": 2.108466625213623,
      "learning_rate": 1.4191359362564354e-06,
      "loss": 0.029,
      "step": 14651
    },
    {
      "epoch": 0.7566227730441518,
      "grad_norm": 13.522404670715332,
      "learning_rate": 1.4185641336943157e-06,
      "loss": 0.0659,
      "step": 14652
    },
    {
      "epoch": 0.7566744126000516,
      "grad_norm": 2.560271739959717,
      "learning_rate": 1.4179924273083462e-06,
      "loss": 0.0427,
      "step": 14653
    },
    {
      "epoch": 0.7567260521559515,
      "grad_norm": 3.098191499710083,
      "learning_rate": 1.4174208171138793e-06,
      "loss": 0.0474,
      "step": 14654
    },
    {
      "epoch": 0.7567776917118513,
      "grad_norm": 4.39755392074585,
      "learning_rate": 1.4168493031262652e-06,
      "loss": 0.0515,
      "step": 14655
    },
    {
      "epoch": 0.7568293312677511,
      "grad_norm": 3.810894250869751,
      "learning_rate": 1.4162778853608517e-06,
      "loss": 0.0435,
      "step": 14656
    },
    {
      "epoch": 0.7568809708236509,
      "grad_norm": 20.212928771972656,
      "learning_rate": 1.415706563832983e-06,
      "loss": 0.0237,
      "step": 14657
    },
    {
      "epoch": 0.7569326103795507,
      "grad_norm": 3.659153461456299,
      "learning_rate": 1.4151353385580025e-06,
      "loss": 0.0541,
      "step": 14658
    },
    {
      "epoch": 0.7569842499354505,
      "grad_norm": 1.2490284442901611,
      "learning_rate": 1.414564209551249e-06,
      "loss": 0.02,
      "step": 14659
    },
    {
      "epoch": 0.7570358894913504,
      "grad_norm": 2.68477725982666,
      "learning_rate": 1.4139931768280596e-06,
      "loss": 0.0405,
      "step": 14660
    },
    {
      "epoch": 0.7570875290472502,
      "grad_norm": 4.837786674499512,
      "learning_rate": 1.4134222404037705e-06,
      "loss": 0.0663,
      "step": 14661
    },
    {
      "epoch": 0.75713916860315,
      "grad_norm": 7.970560550689697,
      "learning_rate": 1.4128514002937099e-06,
      "loss": 0.0767,
      "step": 14662
    },
    {
      "epoch": 0.7571908081590498,
      "grad_norm": 3.0470120906829834,
      "learning_rate": 1.4122806565132108e-06,
      "loss": 0.0163,
      "step": 14663
    },
    {
      "epoch": 0.7572424477149496,
      "grad_norm": 2.027639150619507,
      "learning_rate": 1.4117100090776005e-06,
      "loss": 0.0349,
      "step": 14664
    },
    {
      "epoch": 0.7572940872708495,
      "grad_norm": 10.78917121887207,
      "learning_rate": 1.4111394580021997e-06,
      "loss": 0.0554,
      "step": 14665
    },
    {
      "epoch": 0.7573457268267493,
      "grad_norm": 3.3888561725616455,
      "learning_rate": 1.4105690033023323e-06,
      "loss": 0.0598,
      "step": 14666
    },
    {
      "epoch": 0.7573973663826491,
      "grad_norm": 2.88041615486145,
      "learning_rate": 1.4099986449933151e-06,
      "loss": 0.0591,
      "step": 14667
    },
    {
      "epoch": 0.7574490059385489,
      "grad_norm": 1.3548390865325928,
      "learning_rate": 1.4094283830904692e-06,
      "loss": 0.0157,
      "step": 14668
    },
    {
      "epoch": 0.7575006454944487,
      "grad_norm": 2.6218388080596924,
      "learning_rate": 1.4088582176091043e-06,
      "loss": 0.035,
      "step": 14669
    },
    {
      "epoch": 0.7575522850503486,
      "grad_norm": 2.530561923980713,
      "learning_rate": 1.4082881485645328e-06,
      "loss": 0.0288,
      "step": 14670
    },
    {
      "epoch": 0.7576039246062484,
      "grad_norm": 3.8837432861328125,
      "learning_rate": 1.4077181759720638e-06,
      "loss": 0.0613,
      "step": 14671
    },
    {
      "epoch": 0.7576555641621482,
      "grad_norm": 1.1680705547332764,
      "learning_rate": 1.4071482998470027e-06,
      "loss": 0.0122,
      "step": 14672
    },
    {
      "epoch": 0.757707203718048,
      "grad_norm": 3.7963948249816895,
      "learning_rate": 1.4065785202046533e-06,
      "loss": 0.057,
      "step": 14673
    },
    {
      "epoch": 0.7577588432739478,
      "grad_norm": 3.8257148265838623,
      "learning_rate": 1.406008837060317e-06,
      "loss": 0.0569,
      "step": 14674
    },
    {
      "epoch": 0.7578104828298476,
      "grad_norm": 5.67404317855835,
      "learning_rate": 1.4054392504292934e-06,
      "loss": 0.0449,
      "step": 14675
    },
    {
      "epoch": 0.7578621223857475,
      "grad_norm": 3.406780242919922,
      "learning_rate": 1.4048697603268729e-06,
      "loss": 0.0623,
      "step": 14676
    },
    {
      "epoch": 0.7579137619416473,
      "grad_norm": 3.8511767387390137,
      "learning_rate": 1.404300366768354e-06,
      "loss": 0.0922,
      "step": 14677
    },
    {
      "epoch": 0.7579654014975471,
      "grad_norm": 1.2010853290557861,
      "learning_rate": 1.403731069769026e-06,
      "loss": 0.0146,
      "step": 14678
    },
    {
      "epoch": 0.7580170410534469,
      "grad_norm": 1.9404950141906738,
      "learning_rate": 1.4031618693441773e-06,
      "loss": 0.0238,
      "step": 14679
    },
    {
      "epoch": 0.7580686806093467,
      "grad_norm": 9.529067039489746,
      "learning_rate": 1.402592765509091e-06,
      "loss": 0.1025,
      "step": 14680
    },
    {
      "epoch": 0.7581203201652466,
      "grad_norm": 4.087039470672607,
      "learning_rate": 1.4020237582790496e-06,
      "loss": 0.0342,
      "step": 14681
    },
    {
      "epoch": 0.7581719597211464,
      "grad_norm": 4.432721138000488,
      "learning_rate": 1.4014548476693363e-06,
      "loss": 0.0301,
      "step": 14682
    },
    {
      "epoch": 0.7582235992770462,
      "grad_norm": 12.292284965515137,
      "learning_rate": 1.400886033695229e-06,
      "loss": 0.1314,
      "step": 14683
    },
    {
      "epoch": 0.758275238832946,
      "grad_norm": 4.338273525238037,
      "learning_rate": 1.4003173163719992e-06,
      "loss": 0.0414,
      "step": 14684
    },
    {
      "epoch": 0.7583268783888458,
      "grad_norm": 7.280246734619141,
      "learning_rate": 1.3997486957149208e-06,
      "loss": 0.3024,
      "step": 14685
    },
    {
      "epoch": 0.7583785179447456,
      "grad_norm": 3.1270697116851807,
      "learning_rate": 1.399180171739264e-06,
      "loss": 0.0284,
      "step": 14686
    },
    {
      "epoch": 0.7584301575006455,
      "grad_norm": 3.3199703693389893,
      "learning_rate": 1.3986117444602954e-06,
      "loss": 0.0197,
      "step": 14687
    },
    {
      "epoch": 0.7584817970565453,
      "grad_norm": 2.2637767791748047,
      "learning_rate": 1.3980434138932803e-06,
      "loss": 0.0379,
      "step": 14688
    },
    {
      "epoch": 0.7585334366124451,
      "grad_norm": 3.0824828147888184,
      "learning_rate": 1.3974751800534803e-06,
      "loss": 0.0675,
      "step": 14689
    },
    {
      "epoch": 0.7585850761683449,
      "grad_norm": 2.938915252685547,
      "learning_rate": 1.3969070429561543e-06,
      "loss": 0.0354,
      "step": 14690
    },
    {
      "epoch": 0.7586367157242447,
      "grad_norm": 2.6638097763061523,
      "learning_rate": 1.3963390026165596e-06,
      "loss": 0.0739,
      "step": 14691
    },
    {
      "epoch": 0.7586883552801446,
      "grad_norm": 3.2684192657470703,
      "learning_rate": 1.3957710590499508e-06,
      "loss": 0.0661,
      "step": 14692
    },
    {
      "epoch": 0.7587399948360444,
      "grad_norm": 1.1409248113632202,
      "learning_rate": 1.395203212271579e-06,
      "loss": 0.0171,
      "step": 14693
    },
    {
      "epoch": 0.7587916343919442,
      "grad_norm": 5.265031337738037,
      "learning_rate": 1.3946354622966945e-06,
      "loss": 0.051,
      "step": 14694
    },
    {
      "epoch": 0.758843273947844,
      "grad_norm": 9.958283424377441,
      "learning_rate": 1.3940678091405403e-06,
      "loss": 0.1485,
      "step": 14695
    },
    {
      "epoch": 0.7588949135037438,
      "grad_norm": 2.4240784645080566,
      "learning_rate": 1.3935002528183627e-06,
      "loss": 0.0195,
      "step": 14696
    },
    {
      "epoch": 0.7589465530596436,
      "grad_norm": 2.3845269680023193,
      "learning_rate": 1.392932793345405e-06,
      "loss": 0.0347,
      "step": 14697
    },
    {
      "epoch": 0.7589981926155435,
      "grad_norm": 23.939373016357422,
      "learning_rate": 1.3923654307369017e-06,
      "loss": 0.0955,
      "step": 14698
    },
    {
      "epoch": 0.7590498321714433,
      "grad_norm": 3.9150149822235107,
      "learning_rate": 1.3917981650080902e-06,
      "loss": 0.0563,
      "step": 14699
    },
    {
      "epoch": 0.7591014717273431,
      "grad_norm": 1.7365913391113281,
      "learning_rate": 1.3912309961742026e-06,
      "loss": 0.0241,
      "step": 14700
    },
    {
      "epoch": 0.7591531112832429,
      "grad_norm": 13.651687622070312,
      "learning_rate": 1.390663924250475e-06,
      "loss": 0.1754,
      "step": 14701
    },
    {
      "epoch": 0.7592047508391427,
      "grad_norm": 1.8381279706954956,
      "learning_rate": 1.3900969492521298e-06,
      "loss": 0.0229,
      "step": 14702
    },
    {
      "epoch": 0.7592563903950426,
      "grad_norm": 17.366397857666016,
      "learning_rate": 1.3895300711943948e-06,
      "loss": 0.1224,
      "step": 14703
    },
    {
      "epoch": 0.7593080299509424,
      "grad_norm": 10.921330451965332,
      "learning_rate": 1.3889632900924926e-06,
      "loss": 0.1161,
      "step": 14704
    },
    {
      "epoch": 0.7593596695068422,
      "grad_norm": 4.026634216308594,
      "learning_rate": 1.3883966059616443e-06,
      "loss": 0.0306,
      "step": 14705
    },
    {
      "epoch": 0.759411309062742,
      "grad_norm": 11.373847007751465,
      "learning_rate": 1.3878300188170667e-06,
      "loss": 0.0977,
      "step": 14706
    },
    {
      "epoch": 0.7594629486186418,
      "grad_norm": 3.1747806072235107,
      "learning_rate": 1.3872635286739762e-06,
      "loss": 0.0259,
      "step": 14707
    },
    {
      "epoch": 0.7595145881745416,
      "grad_norm": 3.2094948291778564,
      "learning_rate": 1.3866971355475855e-06,
      "loss": 0.0671,
      "step": 14708
    },
    {
      "epoch": 0.7595662277304415,
      "grad_norm": 3.1811201572418213,
      "learning_rate": 1.386130839453101e-06,
      "loss": 0.0363,
      "step": 14709
    },
    {
      "epoch": 0.7596178672863413,
      "grad_norm": 4.494859218597412,
      "learning_rate": 1.3855646404057343e-06,
      "loss": 0.1004,
      "step": 14710
    },
    {
      "epoch": 0.7596695068422411,
      "grad_norm": 3.899555206298828,
      "learning_rate": 1.3849985384206883e-06,
      "loss": 0.0634,
      "step": 14711
    },
    {
      "epoch": 0.7597211463981409,
      "grad_norm": 9.963964462280273,
      "learning_rate": 1.3844325335131675e-06,
      "loss": 0.0966,
      "step": 14712
    },
    {
      "epoch": 0.7597727859540407,
      "grad_norm": 5.413021087646484,
      "learning_rate": 1.3838666256983674e-06,
      "loss": 0.1036,
      "step": 14713
    },
    {
      "epoch": 0.7598244255099406,
      "grad_norm": 5.932299613952637,
      "learning_rate": 1.3833008149914857e-06,
      "loss": 0.0646,
      "step": 14714
    },
    {
      "epoch": 0.7598760650658405,
      "grad_norm": 2.1157312393188477,
      "learning_rate": 1.3827351014077196e-06,
      "loss": 0.0267,
      "step": 14715
    },
    {
      "epoch": 0.7599277046217403,
      "grad_norm": 5.676239490509033,
      "learning_rate": 1.3821694849622603e-06,
      "loss": 0.0353,
      "step": 14716
    },
    {
      "epoch": 0.7599793441776401,
      "grad_norm": 1.3652681112289429,
      "learning_rate": 1.3816039656702946e-06,
      "loss": 0.0196,
      "step": 14717
    },
    {
      "epoch": 0.7600309837335399,
      "grad_norm": 6.0033369064331055,
      "learning_rate": 1.3810385435470102e-06,
      "loss": 0.1164,
      "step": 14718
    },
    {
      "epoch": 0.7600826232894398,
      "grad_norm": 5.276148796081543,
      "learning_rate": 1.380473218607591e-06,
      "loss": 0.0646,
      "step": 14719
    },
    {
      "epoch": 0.7601342628453396,
      "grad_norm": 13.134214401245117,
      "learning_rate": 1.379907990867218e-06,
      "loss": 0.1214,
      "step": 14720
    },
    {
      "epoch": 0.7601859024012394,
      "grad_norm": 8.108122825622559,
      "learning_rate": 1.3793428603410702e-06,
      "loss": 0.1363,
      "step": 14721
    },
    {
      "epoch": 0.7602375419571392,
      "grad_norm": 5.96245813369751,
      "learning_rate": 1.3787778270443235e-06,
      "loss": 0.1119,
      "step": 14722
    },
    {
      "epoch": 0.760289181513039,
      "grad_norm": 4.504332542419434,
      "learning_rate": 1.3782128909921517e-06,
      "loss": 0.0731,
      "step": 14723
    },
    {
      "epoch": 0.7603408210689389,
      "grad_norm": 16.502948760986328,
      "learning_rate": 1.3776480521997255e-06,
      "loss": 0.1093,
      "step": 14724
    },
    {
      "epoch": 0.7603924606248387,
      "grad_norm": 2.541574478149414,
      "learning_rate": 1.3770833106822123e-06,
      "loss": 0.0317,
      "step": 14725
    },
    {
      "epoch": 0.7604441001807385,
      "grad_norm": 11.641517639160156,
      "learning_rate": 1.3765186664547792e-06,
      "loss": 0.0893,
      "step": 14726
    },
    {
      "epoch": 0.7604957397366383,
      "grad_norm": 3.932976245880127,
      "learning_rate": 1.3759541195325893e-06,
      "loss": 0.0408,
      "step": 14727
    },
    {
      "epoch": 0.7605473792925381,
      "grad_norm": 2.7042810916900635,
      "learning_rate": 1.3753896699307995e-06,
      "loss": 0.04,
      "step": 14728
    },
    {
      "epoch": 0.760599018848438,
      "grad_norm": 3.2606959342956543,
      "learning_rate": 1.374825317664571e-06,
      "loss": 0.0556,
      "step": 14729
    },
    {
      "epoch": 0.7606506584043378,
      "grad_norm": 2.6197826862335205,
      "learning_rate": 1.3742610627490604e-06,
      "loss": 0.0217,
      "step": 14730
    },
    {
      "epoch": 0.7607022979602376,
      "grad_norm": 3.122312307357788,
      "learning_rate": 1.3736969051994159e-06,
      "loss": 0.0515,
      "step": 14731
    },
    {
      "epoch": 0.7607539375161374,
      "grad_norm": 4.555870056152344,
      "learning_rate": 1.3731328450307895e-06,
      "loss": 0.0559,
      "step": 14732
    },
    {
      "epoch": 0.7608055770720372,
      "grad_norm": 4.805774688720703,
      "learning_rate": 1.3725688822583277e-06,
      "loss": 0.0668,
      "step": 14733
    },
    {
      "epoch": 0.760857216627937,
      "grad_norm": 6.125369548797607,
      "learning_rate": 1.3720050168971783e-06,
      "loss": 0.0852,
      "step": 14734
    },
    {
      "epoch": 0.7609088561838369,
      "grad_norm": 3.346463441848755,
      "learning_rate": 1.3714412489624802e-06,
      "loss": 0.0364,
      "step": 14735
    },
    {
      "epoch": 0.7609604957397367,
      "grad_norm": 8.86426830291748,
      "learning_rate": 1.3708775784693728e-06,
      "loss": 0.0722,
      "step": 14736
    },
    {
      "epoch": 0.7610121352956365,
      "grad_norm": 2.880110263824463,
      "learning_rate": 1.3703140054329949e-06,
      "loss": 0.0272,
      "step": 14737
    },
    {
      "epoch": 0.7610637748515363,
      "grad_norm": 4.368658065795898,
      "learning_rate": 1.3697505298684794e-06,
      "loss": 0.054,
      "step": 14738
    },
    {
      "epoch": 0.7611154144074361,
      "grad_norm": 20.843666076660156,
      "learning_rate": 1.3691871517909577e-06,
      "loss": 0.0799,
      "step": 14739
    },
    {
      "epoch": 0.761167053963336,
      "grad_norm": 3.492202043533325,
      "learning_rate": 1.3686238712155597e-06,
      "loss": 0.0654,
      "step": 14740
    },
    {
      "epoch": 0.7612186935192358,
      "grad_norm": 2.671074390411377,
      "learning_rate": 1.3680606881574132e-06,
      "loss": 0.0412,
      "step": 14741
    },
    {
      "epoch": 0.7612703330751356,
      "grad_norm": 2.0997486114501953,
      "learning_rate": 1.3674976026316371e-06,
      "loss": 0.0395,
      "step": 14742
    },
    {
      "epoch": 0.7613219726310354,
      "grad_norm": 3.1392366886138916,
      "learning_rate": 1.366934614653357e-06,
      "loss": 0.0428,
      "step": 14743
    },
    {
      "epoch": 0.7613736121869352,
      "grad_norm": 1.8327621221542358,
      "learning_rate": 1.36637172423769e-06,
      "loss": 0.0276,
      "step": 14744
    },
    {
      "epoch": 0.761425251742835,
      "grad_norm": 2.796163320541382,
      "learning_rate": 1.365808931399754e-06,
      "loss": 0.043,
      "step": 14745
    },
    {
      "epoch": 0.7614768912987349,
      "grad_norm": 2.6625311374664307,
      "learning_rate": 1.3652462361546581e-06,
      "loss": 0.036,
      "step": 14746
    },
    {
      "epoch": 0.7615285308546347,
      "grad_norm": 4.432351589202881,
      "learning_rate": 1.3646836385175144e-06,
      "loss": 0.0965,
      "step": 14747
    },
    {
      "epoch": 0.7615801704105345,
      "grad_norm": 2.008634090423584,
      "learning_rate": 1.364121138503433e-06,
      "loss": 0.039,
      "step": 14748
    },
    {
      "epoch": 0.7616318099664343,
      "grad_norm": 3.0954935550689697,
      "learning_rate": 1.3635587361275193e-06,
      "loss": 0.0256,
      "step": 14749
    },
    {
      "epoch": 0.7616834495223341,
      "grad_norm": 3.3267626762390137,
      "learning_rate": 1.362996431404874e-06,
      "loss": 0.0451,
      "step": 14750
    },
    {
      "epoch": 0.761735089078234,
      "grad_norm": 2.434185266494751,
      "learning_rate": 1.3624342243505972e-06,
      "loss": 0.0223,
      "step": 14751
    },
    {
      "epoch": 0.7617867286341338,
      "grad_norm": 2.0452537536621094,
      "learning_rate": 1.361872114979788e-06,
      "loss": 0.0323,
      "step": 14752
    },
    {
      "epoch": 0.7618383681900336,
      "grad_norm": 11.760258674621582,
      "learning_rate": 1.3613101033075404e-06,
      "loss": 0.2236,
      "step": 14753
    },
    {
      "epoch": 0.7618900077459334,
      "grad_norm": 1.8796465396881104,
      "learning_rate": 1.3607481893489472e-06,
      "loss": 0.0379,
      "step": 14754
    },
    {
      "epoch": 0.7619416473018332,
      "grad_norm": 2.0260181427001953,
      "learning_rate": 1.3601863731190978e-06,
      "loss": 0.0411,
      "step": 14755
    },
    {
      "epoch": 0.761993286857733,
      "grad_norm": 4.1007080078125,
      "learning_rate": 1.3596246546330793e-06,
      "loss": 0.0559,
      "step": 14756
    },
    {
      "epoch": 0.7620449264136329,
      "grad_norm": 3.9975132942199707,
      "learning_rate": 1.3590630339059764e-06,
      "loss": 0.054,
      "step": 14757
    },
    {
      "epoch": 0.7620965659695327,
      "grad_norm": 3.4221763610839844,
      "learning_rate": 1.3585015109528703e-06,
      "loss": 0.042,
      "step": 14758
    },
    {
      "epoch": 0.7621482055254325,
      "grad_norm": 14.273699760437012,
      "learning_rate": 1.357940085788841e-06,
      "loss": 0.1307,
      "step": 14759
    },
    {
      "epoch": 0.7621998450813323,
      "grad_norm": 1.6268664598464966,
      "learning_rate": 1.3573787584289656e-06,
      "loss": 0.0118,
      "step": 14760
    },
    {
      "epoch": 0.7622514846372321,
      "grad_norm": 1.4558305740356445,
      "learning_rate": 1.3568175288883145e-06,
      "loss": 0.0183,
      "step": 14761
    },
    {
      "epoch": 0.762303124193132,
      "grad_norm": 5.691262245178223,
      "learning_rate": 1.3562563971819625e-06,
      "loss": 0.0524,
      "step": 14762
    },
    {
      "epoch": 0.7623547637490318,
      "grad_norm": 5.9180779457092285,
      "learning_rate": 1.3556953633249792e-06,
      "loss": 0.04,
      "step": 14763
    },
    {
      "epoch": 0.7624064033049316,
      "grad_norm": 2.4305648803710938,
      "learning_rate": 1.3551344273324274e-06,
      "loss": 0.0368,
      "step": 14764
    },
    {
      "epoch": 0.7624580428608314,
      "grad_norm": 5.059712886810303,
      "learning_rate": 1.3545735892193717e-06,
      "loss": 0.0593,
      "step": 14765
    },
    {
      "epoch": 0.7625096824167312,
      "grad_norm": 3.7323319911956787,
      "learning_rate": 1.3540128490008714e-06,
      "loss": 0.0353,
      "step": 14766
    },
    {
      "epoch": 0.762561321972631,
      "grad_norm": 3.0822818279266357,
      "learning_rate": 1.3534522066919898e-06,
      "loss": 0.0512,
      "step": 14767
    },
    {
      "epoch": 0.7626129615285309,
      "grad_norm": 2.3134400844573975,
      "learning_rate": 1.352891662307777e-06,
      "loss": 0.0237,
      "step": 14768
    },
    {
      "epoch": 0.7626646010844307,
      "grad_norm": 2.7251195907592773,
      "learning_rate": 1.352331215863288e-06,
      "loss": 0.0466,
      "step": 14769
    },
    {
      "epoch": 0.7627162406403305,
      "grad_norm": 5.185105800628662,
      "learning_rate": 1.351770867373573e-06,
      "loss": 0.0533,
      "step": 14770
    },
    {
      "epoch": 0.7627678801962303,
      "grad_norm": 2.720773458480835,
      "learning_rate": 1.3512106168536792e-06,
      "loss": 0.0083,
      "step": 14771
    },
    {
      "epoch": 0.7628195197521301,
      "grad_norm": 1.8710130453109741,
      "learning_rate": 1.3506504643186524e-06,
      "loss": 0.0311,
      "step": 14772
    },
    {
      "epoch": 0.76287115930803,
      "grad_norm": 6.932682991027832,
      "learning_rate": 1.3500904097835343e-06,
      "loss": 0.0643,
      "step": 14773
    },
    {
      "epoch": 0.7629227988639298,
      "grad_norm": 6.3564300537109375,
      "learning_rate": 1.3495304532633669e-06,
      "loss": 0.031,
      "step": 14774
    },
    {
      "epoch": 0.7629744384198296,
      "grad_norm": 14.55994701385498,
      "learning_rate": 1.3489705947731818e-06,
      "loss": 0.1388,
      "step": 14775
    },
    {
      "epoch": 0.7630260779757294,
      "grad_norm": 9.57125186920166,
      "learning_rate": 1.3484108343280188e-06,
      "loss": 0.1107,
      "step": 14776
    },
    {
      "epoch": 0.7630777175316292,
      "grad_norm": 15.243263244628906,
      "learning_rate": 1.3478511719429078e-06,
      "loss": 0.2104,
      "step": 14777
    },
    {
      "epoch": 0.763129357087529,
      "grad_norm": 2.296030044555664,
      "learning_rate": 1.3472916076328797e-06,
      "loss": 0.0267,
      "step": 14778
    },
    {
      "epoch": 0.7631809966434289,
      "grad_norm": 5.924831390380859,
      "learning_rate": 1.346732141412958e-06,
      "loss": 0.0664,
      "step": 14779
    },
    {
      "epoch": 0.7632326361993287,
      "grad_norm": 17.57554817199707,
      "learning_rate": 1.3461727732981665e-06,
      "loss": 0.0797,
      "step": 14780
    },
    {
      "epoch": 0.7632842757552285,
      "grad_norm": 4.658781051635742,
      "learning_rate": 1.34561350330353e-06,
      "loss": 0.0704,
      "step": 14781
    },
    {
      "epoch": 0.7633359153111283,
      "grad_norm": 8.577432632446289,
      "learning_rate": 1.3450543314440673e-06,
      "loss": 0.0909,
      "step": 14782
    },
    {
      "epoch": 0.7633875548670281,
      "grad_norm": 3.050990104675293,
      "learning_rate": 1.3444952577347903e-06,
      "loss": 0.0385,
      "step": 14783
    },
    {
      "epoch": 0.763439194422928,
      "grad_norm": 6.128305912017822,
      "learning_rate": 1.3439362821907148e-06,
      "loss": 0.0611,
      "step": 14784
    },
    {
      "epoch": 0.7634908339788278,
      "grad_norm": 3.241542339324951,
      "learning_rate": 1.3433774048268517e-06,
      "loss": 0.0491,
      "step": 14785
    },
    {
      "epoch": 0.7635424735347276,
      "grad_norm": 6.498305320739746,
      "learning_rate": 1.3428186256582088e-06,
      "loss": 0.054,
      "step": 14786
    },
    {
      "epoch": 0.7635941130906274,
      "grad_norm": 1.494871973991394,
      "learning_rate": 1.3422599446997913e-06,
      "loss": 0.0275,
      "step": 14787
    },
    {
      "epoch": 0.7636457526465272,
      "grad_norm": 1.621440052986145,
      "learning_rate": 1.341701361966603e-06,
      "loss": 0.029,
      "step": 14788
    },
    {
      "epoch": 0.763697392202427,
      "grad_norm": 7.178300857543945,
      "learning_rate": 1.341142877473643e-06,
      "loss": 0.0606,
      "step": 14789
    },
    {
      "epoch": 0.7637490317583269,
      "grad_norm": 5.279527187347412,
      "learning_rate": 1.3405844912359096e-06,
      "loss": 0.0519,
      "step": 14790
    },
    {
      "epoch": 0.7638006713142267,
      "grad_norm": 9.58971118927002,
      "learning_rate": 1.3400262032683976e-06,
      "loss": 0.0811,
      "step": 14791
    },
    {
      "epoch": 0.7638523108701265,
      "grad_norm": 3.9290077686309814,
      "learning_rate": 1.3394680135860993e-06,
      "loss": 0.057,
      "step": 14792
    },
    {
      "epoch": 0.7639039504260263,
      "grad_norm": 4.032032489776611,
      "learning_rate": 1.3389099222040058e-06,
      "loss": 0.0519,
      "step": 14793
    },
    {
      "epoch": 0.7639555899819261,
      "grad_norm": 3.9390957355499268,
      "learning_rate": 1.3383519291370995e-06,
      "loss": 0.0547,
      "step": 14794
    },
    {
      "epoch": 0.764007229537826,
      "grad_norm": 20.185483932495117,
      "learning_rate": 1.33779403440037e-06,
      "loss": 0.1084,
      "step": 14795
    },
    {
      "epoch": 0.7640588690937258,
      "grad_norm": 2.553140878677368,
      "learning_rate": 1.3372362380087977e-06,
      "loss": 0.0374,
      "step": 14796
    },
    {
      "epoch": 0.7641105086496256,
      "grad_norm": 9.090974807739258,
      "learning_rate": 1.3366785399773602e-06,
      "loss": 0.0763,
      "step": 14797
    },
    {
      "epoch": 0.7641621482055254,
      "grad_norm": 7.695197582244873,
      "learning_rate": 1.3361209403210341e-06,
      "loss": 0.1011,
      "step": 14798
    },
    {
      "epoch": 0.7642137877614252,
      "grad_norm": 10.151575088500977,
      "learning_rate": 1.3355634390547929e-06,
      "loss": 0.1064,
      "step": 14799
    },
    {
      "epoch": 0.764265427317325,
      "grad_norm": 6.328211307525635,
      "learning_rate": 1.3350060361936118e-06,
      "loss": 0.0792,
      "step": 14800
    },
    {
      "epoch": 0.7643170668732249,
      "grad_norm": 3.955214738845825,
      "learning_rate": 1.3344487317524547e-06,
      "loss": 0.0614,
      "step": 14801
    },
    {
      "epoch": 0.7643687064291247,
      "grad_norm": 3.5605857372283936,
      "learning_rate": 1.3338915257462887e-06,
      "loss": 0.0602,
      "step": 14802
    },
    {
      "epoch": 0.7644203459850245,
      "grad_norm": 5.1378350257873535,
      "learning_rate": 1.3333344181900781e-06,
      "loss": 0.0504,
      "step": 14803
    },
    {
      "epoch": 0.7644719855409243,
      "grad_norm": 7.386641502380371,
      "learning_rate": 1.3327774090987828e-06,
      "loss": 0.0767,
      "step": 14804
    },
    {
      "epoch": 0.7645236250968241,
      "grad_norm": 2.344541549682617,
      "learning_rate": 1.3322204984873605e-06,
      "loss": 0.0387,
      "step": 14805
    },
    {
      "epoch": 0.764575264652724,
      "grad_norm": 2.4846034049987793,
      "learning_rate": 1.3316636863707672e-06,
      "loss": 0.0183,
      "step": 14806
    },
    {
      "epoch": 0.7646269042086238,
      "grad_norm": 1.9246560335159302,
      "learning_rate": 1.3311069727639563e-06,
      "loss": 0.0241,
      "step": 14807
    },
    {
      "epoch": 0.7646785437645236,
      "grad_norm": 7.258533477783203,
      "learning_rate": 1.330550357681874e-06,
      "loss": 0.1136,
      "step": 14808
    },
    {
      "epoch": 0.7647301833204234,
      "grad_norm": 4.86531400680542,
      "learning_rate": 1.329993841139472e-06,
      "loss": 0.0442,
      "step": 14809
    },
    {
      "epoch": 0.7647818228763232,
      "grad_norm": 8.52638053894043,
      "learning_rate": 1.3294374231516933e-06,
      "loss": 0.0812,
      "step": 14810
    },
    {
      "epoch": 0.764833462432223,
      "grad_norm": 3.623724937438965,
      "learning_rate": 1.328881103733482e-06,
      "loss": 0.0527,
      "step": 14811
    },
    {
      "epoch": 0.7648851019881229,
      "grad_norm": 2.1199376583099365,
      "learning_rate": 1.3283248828997742e-06,
      "loss": 0.0241,
      "step": 14812
    },
    {
      "epoch": 0.7649367415440227,
      "grad_norm": 1.6996737718582153,
      "learning_rate": 1.327768760665506e-06,
      "loss": 0.013,
      "step": 14813
    },
    {
      "epoch": 0.7649883810999225,
      "grad_norm": 3.373061180114746,
      "learning_rate": 1.3272127370456173e-06,
      "loss": 0.0305,
      "step": 14814
    },
    {
      "epoch": 0.7650400206558223,
      "grad_norm": 11.04038143157959,
      "learning_rate": 1.3266568120550349e-06,
      "loss": 0.1059,
      "step": 14815
    },
    {
      "epoch": 0.7650916602117221,
      "grad_norm": 1.5171079635620117,
      "learning_rate": 1.326100985708688e-06,
      "loss": 0.0107,
      "step": 14816
    },
    {
      "epoch": 0.765143299767622,
      "grad_norm": 5.211123466491699,
      "learning_rate": 1.3255452580215044e-06,
      "loss": 0.0661,
      "step": 14817
    },
    {
      "epoch": 0.7651949393235218,
      "grad_norm": 2.629004955291748,
      "learning_rate": 1.3249896290084068e-06,
      "loss": 0.0525,
      "step": 14818
    },
    {
      "epoch": 0.7652465788794216,
      "grad_norm": 2.6143555641174316,
      "learning_rate": 1.3244340986843158e-06,
      "loss": 0.0355,
      "step": 14819
    },
    {
      "epoch": 0.7652982184353214,
      "grad_norm": 6.359645366668701,
      "learning_rate": 1.3238786670641502e-06,
      "loss": 0.0792,
      "step": 14820
    },
    {
      "epoch": 0.7653498579912212,
      "grad_norm": 1.4870229959487915,
      "learning_rate": 1.3233233341628254e-06,
      "loss": 0.0179,
      "step": 14821
    },
    {
      "epoch": 0.7654014975471211,
      "grad_norm": 8.763854026794434,
      "learning_rate": 1.322768099995254e-06,
      "loss": 0.0602,
      "step": 14822
    },
    {
      "epoch": 0.7654531371030209,
      "grad_norm": 6.73459005355835,
      "learning_rate": 1.3222129645763477e-06,
      "loss": 0.0681,
      "step": 14823
    },
    {
      "epoch": 0.7655047766589207,
      "grad_norm": 3.467026948928833,
      "learning_rate": 1.3216579279210123e-06,
      "loss": 0.0246,
      "step": 14824
    },
    {
      "epoch": 0.7655564162148205,
      "grad_norm": 7.951565742492676,
      "learning_rate": 1.3211029900441552e-06,
      "loss": 0.0619,
      "step": 14825
    },
    {
      "epoch": 0.7656080557707203,
      "grad_norm": 2.3923144340515137,
      "learning_rate": 1.320548150960676e-06,
      "loss": 0.0338,
      "step": 14826
    },
    {
      "epoch": 0.7656596953266201,
      "grad_norm": 8.685030937194824,
      "learning_rate": 1.319993410685474e-06,
      "loss": 0.0423,
      "step": 14827
    },
    {
      "epoch": 0.76571133488252,
      "grad_norm": 2.9868109226226807,
      "learning_rate": 1.3194387692334488e-06,
      "loss": 0.033,
      "step": 14828
    },
    {
      "epoch": 0.7657629744384198,
      "grad_norm": 4.771490573883057,
      "learning_rate": 1.3188842266194957e-06,
      "loss": 0.029,
      "step": 14829
    },
    {
      "epoch": 0.7658146139943196,
      "grad_norm": 10.066374778747559,
      "learning_rate": 1.3183297828585028e-06,
      "loss": 0.0732,
      "step": 14830
    },
    {
      "epoch": 0.7658662535502194,
      "grad_norm": 12.967540740966797,
      "learning_rate": 1.3177754379653612e-06,
      "loss": 0.1127,
      "step": 14831
    },
    {
      "epoch": 0.7659178931061192,
      "grad_norm": 4.179154872894287,
      "learning_rate": 1.3172211919549555e-06,
      "loss": 0.0242,
      "step": 14832
    },
    {
      "epoch": 0.7659695326620191,
      "grad_norm": 2.5121588706970215,
      "learning_rate": 1.3166670448421736e-06,
      "loss": 0.049,
      "step": 14833
    },
    {
      "epoch": 0.7660211722179189,
      "grad_norm": 2.3748555183410645,
      "learning_rate": 1.316112996641893e-06,
      "loss": 0.0335,
      "step": 14834
    },
    {
      "epoch": 0.7660728117738187,
      "grad_norm": 6.55875825881958,
      "learning_rate": 1.3155590473689927e-06,
      "loss": 0.1091,
      "step": 14835
    },
    {
      "epoch": 0.7661244513297185,
      "grad_norm": 3.771697998046875,
      "learning_rate": 1.3150051970383488e-06,
      "loss": 0.0406,
      "step": 14836
    },
    {
      "epoch": 0.7661760908856183,
      "grad_norm": 11.488850593566895,
      "learning_rate": 1.314451445664835e-06,
      "loss": 0.1214,
      "step": 14837
    },
    {
      "epoch": 0.7662277304415182,
      "grad_norm": 10.797390937805176,
      "learning_rate": 1.3138977932633213e-06,
      "loss": 0.0516,
      "step": 14838
    },
    {
      "epoch": 0.766279369997418,
      "grad_norm": 2.793205738067627,
      "learning_rate": 1.3133442398486756e-06,
      "loss": 0.0247,
      "step": 14839
    },
    {
      "epoch": 0.7663310095533178,
      "grad_norm": 3.406440496444702,
      "learning_rate": 1.312790785435764e-06,
      "loss": 0.0359,
      "step": 14840
    },
    {
      "epoch": 0.7663826491092176,
      "grad_norm": 6.281130790710449,
      "learning_rate": 1.3122374300394458e-06,
      "loss": 0.0788,
      "step": 14841
    },
    {
      "epoch": 0.7664342886651174,
      "grad_norm": 4.086891174316406,
      "learning_rate": 1.311684173674584e-06,
      "loss": 0.0446,
      "step": 14842
    },
    {
      "epoch": 0.7664859282210174,
      "grad_norm": 14.544541358947754,
      "learning_rate": 1.3111310163560354e-06,
      "loss": 0.1008,
      "step": 14843
    },
    {
      "epoch": 0.7665375677769172,
      "grad_norm": 6.467200756072998,
      "learning_rate": 1.3105779580986549e-06,
      "loss": 0.0499,
      "step": 14844
    },
    {
      "epoch": 0.766589207332817,
      "grad_norm": 3.2390356063842773,
      "learning_rate": 1.3100249989172924e-06,
      "loss": 0.0615,
      "step": 14845
    },
    {
      "epoch": 0.7666408468887168,
      "grad_norm": 3.9460790157318115,
      "learning_rate": 1.3094721388267968e-06,
      "loss": 0.0249,
      "step": 14846
    },
    {
      "epoch": 0.7666924864446166,
      "grad_norm": 11.714254379272461,
      "learning_rate": 1.3089193778420185e-06,
      "loss": 0.1675,
      "step": 14847
    },
    {
      "epoch": 0.7667441260005164,
      "grad_norm": 6.078797340393066,
      "learning_rate": 1.308366715977798e-06,
      "loss": 0.0448,
      "step": 14848
    },
    {
      "epoch": 0.7667957655564163,
      "grad_norm": 3.162978410720825,
      "learning_rate": 1.3078141532489769e-06,
      "loss": 0.0368,
      "step": 14849
    },
    {
      "epoch": 0.7668474051123161,
      "grad_norm": 3.378934860229492,
      "learning_rate": 1.3072616896703948e-06,
      "loss": 0.0438,
      "step": 14850
    },
    {
      "epoch": 0.7668990446682159,
      "grad_norm": 44.22056198120117,
      "learning_rate": 1.3067093252568868e-06,
      "loss": 0.1106,
      "step": 14851
    },
    {
      "epoch": 0.7669506842241157,
      "grad_norm": 3.637251377105713,
      "learning_rate": 1.3061570600232864e-06,
      "loss": 0.0479,
      "step": 14852
    },
    {
      "epoch": 0.7670023237800155,
      "grad_norm": 3.530413866043091,
      "learning_rate": 1.3056048939844246e-06,
      "loss": 0.0488,
      "step": 14853
    },
    {
      "epoch": 0.7670539633359154,
      "grad_norm": 3.3774335384368896,
      "learning_rate": 1.3050528271551282e-06,
      "loss": 0.0528,
      "step": 14854
    },
    {
      "epoch": 0.7671056028918152,
      "grad_norm": 5.709742546081543,
      "learning_rate": 1.304500859550224e-06,
      "loss": 0.0447,
      "step": 14855
    },
    {
      "epoch": 0.767157242447715,
      "grad_norm": 4.518130779266357,
      "learning_rate": 1.3039489911845332e-06,
      "loss": 0.0468,
      "step": 14856
    },
    {
      "epoch": 0.7672088820036148,
      "grad_norm": 1.6616984605789185,
      "learning_rate": 1.3033972220728764e-06,
      "loss": 0.0252,
      "step": 14857
    },
    {
      "epoch": 0.7672605215595146,
      "grad_norm": 21.450908660888672,
      "learning_rate": 1.302845552230072e-06,
      "loss": 0.0745,
      "step": 14858
    },
    {
      "epoch": 0.7673121611154144,
      "grad_norm": 3.1196253299713135,
      "learning_rate": 1.3022939816709323e-06,
      "loss": 0.0362,
      "step": 14859
    },
    {
      "epoch": 0.7673638006713143,
      "grad_norm": 5.074769020080566,
      "learning_rate": 1.3017425104102683e-06,
      "loss": 0.0808,
      "step": 14860
    },
    {
      "epoch": 0.7674154402272141,
      "grad_norm": 3.0170557498931885,
      "learning_rate": 1.3011911384628929e-06,
      "loss": 0.022,
      "step": 14861
    },
    {
      "epoch": 0.7674670797831139,
      "grad_norm": 12.179159164428711,
      "learning_rate": 1.300639865843612e-06,
      "loss": 0.0745,
      "step": 14862
    },
    {
      "epoch": 0.7675187193390137,
      "grad_norm": 7.244026184082031,
      "learning_rate": 1.3000886925672274e-06,
      "loss": 0.0742,
      "step": 14863
    },
    {
      "epoch": 0.7675703588949135,
      "grad_norm": 3.657958745956421,
      "learning_rate": 1.2995376186485415e-06,
      "loss": 0.0717,
      "step": 14864
    },
    {
      "epoch": 0.7676219984508134,
      "grad_norm": 5.202212333679199,
      "learning_rate": 1.2989866441023508e-06,
      "loss": 0.0342,
      "step": 14865
    },
    {
      "epoch": 0.7676736380067132,
      "grad_norm": 3.0268056392669678,
      "learning_rate": 1.2984357689434563e-06,
      "loss": 0.0534,
      "step": 14866
    },
    {
      "epoch": 0.767725277562613,
      "grad_norm": 7.321815490722656,
      "learning_rate": 1.297884993186647e-06,
      "loss": 0.0761,
      "step": 14867
    },
    {
      "epoch": 0.7677769171185128,
      "grad_norm": 6.508870601654053,
      "learning_rate": 1.2973343168467145e-06,
      "loss": 0.0495,
      "step": 14868
    },
    {
      "epoch": 0.7678285566744126,
      "grad_norm": 1.8005146980285645,
      "learning_rate": 1.2967837399384469e-06,
      "loss": 0.026,
      "step": 14869
    },
    {
      "epoch": 0.7678801962303125,
      "grad_norm": 3.6973636150360107,
      "learning_rate": 1.2962332624766288e-06,
      "loss": 0.0494,
      "step": 14870
    },
    {
      "epoch": 0.7679318357862123,
      "grad_norm": 3.1110870838165283,
      "learning_rate": 1.2956828844760438e-06,
      "loss": 0.0297,
      "step": 14871
    },
    {
      "epoch": 0.7679834753421121,
      "grad_norm": 6.0230183601379395,
      "learning_rate": 1.2951326059514714e-06,
      "loss": 0.1165,
      "step": 14872
    },
    {
      "epoch": 0.7680351148980119,
      "grad_norm": 11.80369758605957,
      "learning_rate": 1.29458242691769e-06,
      "loss": 0.1336,
      "step": 14873
    },
    {
      "epoch": 0.7680867544539117,
      "grad_norm": 2.7308835983276367,
      "learning_rate": 1.2940323473894701e-06,
      "loss": 0.0137,
      "step": 14874
    },
    {
      "epoch": 0.7681383940098115,
      "grad_norm": 2.0391576290130615,
      "learning_rate": 1.2934823673815878e-06,
      "loss": 0.0302,
      "step": 14875
    },
    {
      "epoch": 0.7681900335657114,
      "grad_norm": 4.077425479888916,
      "learning_rate": 1.2929324869088105e-06,
      "loss": 0.0187,
      "step": 14876
    },
    {
      "epoch": 0.7682416731216112,
      "grad_norm": 8.004050254821777,
      "learning_rate": 1.2923827059859067e-06,
      "loss": 0.0492,
      "step": 14877
    },
    {
      "epoch": 0.768293312677511,
      "grad_norm": 4.108645439147949,
      "learning_rate": 1.2918330246276379e-06,
      "loss": 0.0425,
      "step": 14878
    },
    {
      "epoch": 0.7683449522334108,
      "grad_norm": 3.7386672496795654,
      "learning_rate": 1.2912834428487635e-06,
      "loss": 0.0327,
      "step": 14879
    },
    {
      "epoch": 0.7683965917893106,
      "grad_norm": 6.374035835266113,
      "learning_rate": 1.2907339606640483e-06,
      "loss": 0.0749,
      "step": 14880
    },
    {
      "epoch": 0.7684482313452105,
      "grad_norm": 3.121424436569214,
      "learning_rate": 1.2901845780882427e-06,
      "loss": 0.0765,
      "step": 14881
    },
    {
      "epoch": 0.7684998709011103,
      "grad_norm": 4.5704779624938965,
      "learning_rate": 1.2896352951361014e-06,
      "loss": 0.034,
      "step": 14882
    },
    {
      "epoch": 0.7685515104570101,
      "grad_norm": 2.4931881427764893,
      "learning_rate": 1.2890861118223752e-06,
      "loss": 0.0185,
      "step": 14883
    },
    {
      "epoch": 0.7686031500129099,
      "grad_norm": 2.256577730178833,
      "learning_rate": 1.2885370281618116e-06,
      "loss": 0.0371,
      "step": 14884
    },
    {
      "epoch": 0.7686547895688097,
      "grad_norm": 25.07283592224121,
      "learning_rate": 1.2879880441691561e-06,
      "loss": 0.0512,
      "step": 14885
    },
    {
      "epoch": 0.7687064291247095,
      "grad_norm": 10.489725112915039,
      "learning_rate": 1.2874391598591506e-06,
      "loss": 0.0693,
      "step": 14886
    },
    {
      "epoch": 0.7687580686806094,
      "grad_norm": 7.481198787689209,
      "learning_rate": 1.2868903752465355e-06,
      "loss": 0.1019,
      "step": 14887
    },
    {
      "epoch": 0.7688097082365092,
      "grad_norm": 3.0248870849609375,
      "learning_rate": 1.2863416903460474e-06,
      "loss": 0.0245,
      "step": 14888
    },
    {
      "epoch": 0.768861347792409,
      "grad_norm": 2.909247875213623,
      "learning_rate": 1.2857931051724209e-06,
      "loss": 0.0702,
      "step": 14889
    },
    {
      "epoch": 0.7689129873483088,
      "grad_norm": 1.8291606903076172,
      "learning_rate": 1.2852446197403879e-06,
      "loss": 0.0171,
      "step": 14890
    },
    {
      "epoch": 0.7689646269042086,
      "grad_norm": 2.6119136810302734,
      "learning_rate": 1.284696234064679e-06,
      "loss": 0.0495,
      "step": 14891
    },
    {
      "epoch": 0.7690162664601085,
      "grad_norm": 2.649137496948242,
      "learning_rate": 1.2841479481600172e-06,
      "loss": 0.0329,
      "step": 14892
    },
    {
      "epoch": 0.7690679060160083,
      "grad_norm": 5.525404453277588,
      "learning_rate": 1.2835997620411267e-06,
      "loss": 0.061,
      "step": 14893
    },
    {
      "epoch": 0.7691195455719081,
      "grad_norm": 17.36046028137207,
      "learning_rate": 1.2830516757227313e-06,
      "loss": 0.1207,
      "step": 14894
    },
    {
      "epoch": 0.7691711851278079,
      "grad_norm": 3.0943174362182617,
      "learning_rate": 1.2825036892195496e-06,
      "loss": 0.0309,
      "step": 14895
    },
    {
      "epoch": 0.7692228246837077,
      "grad_norm": 1.6268750429153442,
      "learning_rate": 1.2819558025462936e-06,
      "loss": 0.04,
      "step": 14896
    },
    {
      "epoch": 0.7692744642396075,
      "grad_norm": 2.7059519290924072,
      "learning_rate": 1.2814080157176784e-06,
      "loss": 0.049,
      "step": 14897
    },
    {
      "epoch": 0.7693261037955074,
      "grad_norm": 4.894850730895996,
      "learning_rate": 1.2808603287484122e-06,
      "loss": 0.0546,
      "step": 14898
    },
    {
      "epoch": 0.7693777433514072,
      "grad_norm": 5.155219554901123,
      "learning_rate": 1.2803127416532079e-06,
      "loss": 0.0555,
      "step": 14899
    },
    {
      "epoch": 0.769429382907307,
      "grad_norm": 8.226153373718262,
      "learning_rate": 1.279765254446766e-06,
      "loss": 0.1047,
      "step": 14900
    },
    {
      "epoch": 0.7694810224632068,
      "grad_norm": 7.942320346832275,
      "learning_rate": 1.2792178671437894e-06,
      "loss": 0.0397,
      "step": 14901
    },
    {
      "epoch": 0.7695326620191066,
      "grad_norm": 5.045856952667236,
      "learning_rate": 1.2786705797589782e-06,
      "loss": 0.0399,
      "step": 14902
    },
    {
      "epoch": 0.7695843015750065,
      "grad_norm": 4.368800640106201,
      "learning_rate": 1.2781233923070291e-06,
      "loss": 0.0687,
      "step": 14903
    },
    {
      "epoch": 0.7696359411309063,
      "grad_norm": 3.5047667026519775,
      "learning_rate": 1.2775763048026362e-06,
      "loss": 0.0395,
      "step": 14904
    },
    {
      "epoch": 0.7696875806868061,
      "grad_norm": 5.296061038970947,
      "learning_rate": 1.277029317260492e-06,
      "loss": 0.104,
      "step": 14905
    },
    {
      "epoch": 0.7697392202427059,
      "grad_norm": 2.201465606689453,
      "learning_rate": 1.2764824296952843e-06,
      "loss": 0.0217,
      "step": 14906
    },
    {
      "epoch": 0.7697908597986057,
      "grad_norm": 5.251378536224365,
      "learning_rate": 1.2759356421216994e-06,
      "loss": 0.0623,
      "step": 14907
    },
    {
      "epoch": 0.7698424993545055,
      "grad_norm": 1.2646806240081787,
      "learning_rate": 1.2753889545544212e-06,
      "loss": 0.0139,
      "step": 14908
    },
    {
      "epoch": 0.7698941389104054,
      "grad_norm": 2.45330810546875,
      "learning_rate": 1.27484236700813e-06,
      "loss": 0.0382,
      "step": 14909
    },
    {
      "epoch": 0.7699457784663052,
      "grad_norm": 9.909590721130371,
      "learning_rate": 1.2742958794975059e-06,
      "loss": 0.0655,
      "step": 14910
    },
    {
      "epoch": 0.769997418022205,
      "grad_norm": 4.534585952758789,
      "learning_rate": 1.2737494920372207e-06,
      "loss": 0.0553,
      "step": 14911
    },
    {
      "epoch": 0.7700490575781048,
      "grad_norm": 2.560089349746704,
      "learning_rate": 1.273203204641948e-06,
      "loss": 0.0238,
      "step": 14912
    },
    {
      "epoch": 0.7701006971340046,
      "grad_norm": 3.399137258529663,
      "learning_rate": 1.2726570173263613e-06,
      "loss": 0.0535,
      "step": 14913
    },
    {
      "epoch": 0.7701523366899045,
      "grad_norm": 10.575098037719727,
      "learning_rate": 1.2721109301051243e-06,
      "loss": 0.1472,
      "step": 14914
    },
    {
      "epoch": 0.7702039762458043,
      "grad_norm": 7.057668209075928,
      "learning_rate": 1.2715649429929027e-06,
      "loss": 0.0444,
      "step": 14915
    },
    {
      "epoch": 0.7702556158017041,
      "grad_norm": 11.242860794067383,
      "learning_rate": 1.271019056004359e-06,
      "loss": 0.0947,
      "step": 14916
    },
    {
      "epoch": 0.7703072553576039,
      "grad_norm": 4.715404033660889,
      "learning_rate": 1.2704732691541522e-06,
      "loss": 0.0371,
      "step": 14917
    },
    {
      "epoch": 0.7703588949135037,
      "grad_norm": 1.8617645502090454,
      "learning_rate": 1.2699275824569385e-06,
      "loss": 0.0198,
      "step": 14918
    },
    {
      "epoch": 0.7704105344694036,
      "grad_norm": 13.29615592956543,
      "learning_rate": 1.2693819959273724e-06,
      "loss": 0.1583,
      "step": 14919
    },
    {
      "epoch": 0.7704621740253034,
      "grad_norm": 2.6630682945251465,
      "learning_rate": 1.268836509580104e-06,
      "loss": 0.0416,
      "step": 14920
    },
    {
      "epoch": 0.7705138135812032,
      "grad_norm": 4.334409713745117,
      "learning_rate": 1.2682911234297835e-06,
      "loss": 0.0344,
      "step": 14921
    },
    {
      "epoch": 0.770565453137103,
      "grad_norm": 23.69034767150879,
      "learning_rate": 1.2677458374910562e-06,
      "loss": 0.258,
      "step": 14922
    },
    {
      "epoch": 0.7706170926930028,
      "grad_norm": 2.450314998626709,
      "learning_rate": 1.2672006517785646e-06,
      "loss": 0.0289,
      "step": 14923
    },
    {
      "epoch": 0.7706687322489026,
      "grad_norm": 6.1543378829956055,
      "learning_rate": 1.2666555663069508e-06,
      "loss": 0.0789,
      "step": 14924
    },
    {
      "epoch": 0.7707203718048025,
      "grad_norm": 3.0334644317626953,
      "learning_rate": 1.2661105810908485e-06,
      "loss": 0.0516,
      "step": 14925
    },
    {
      "epoch": 0.7707720113607023,
      "grad_norm": 2.930666446685791,
      "learning_rate": 1.265565696144897e-06,
      "loss": 0.0319,
      "step": 14926
    },
    {
      "epoch": 0.7708236509166021,
      "grad_norm": 2.918452024459839,
      "learning_rate": 1.2650209114837275e-06,
      "loss": 0.0311,
      "step": 14927
    },
    {
      "epoch": 0.7708752904725019,
      "grad_norm": 4.419105052947998,
      "learning_rate": 1.2644762271219708e-06,
      "loss": 0.0317,
      "step": 14928
    },
    {
      "epoch": 0.7709269300284017,
      "grad_norm": 9.273303031921387,
      "learning_rate": 1.2639316430742515e-06,
      "loss": 0.0679,
      "step": 14929
    },
    {
      "epoch": 0.7709785695843016,
      "grad_norm": 16.590988159179688,
      "learning_rate": 1.2633871593551943e-06,
      "loss": 0.0781,
      "step": 14930
    },
    {
      "epoch": 0.7710302091402014,
      "grad_norm": 2.9356040954589844,
      "learning_rate": 1.2628427759794204e-06,
      "loss": 0.0442,
      "step": 14931
    },
    {
      "epoch": 0.7710818486961012,
      "grad_norm": 3.912095069885254,
      "learning_rate": 1.2622984929615522e-06,
      "loss": 0.0402,
      "step": 14932
    },
    {
      "epoch": 0.771133488252001,
      "grad_norm": 3.281357765197754,
      "learning_rate": 1.2617543103162027e-06,
      "loss": 0.0626,
      "step": 14933
    },
    {
      "epoch": 0.7711851278079008,
      "grad_norm": 6.757861137390137,
      "learning_rate": 1.2612102280579857e-06,
      "loss": 0.1129,
      "step": 14934
    },
    {
      "epoch": 0.7712367673638006,
      "grad_norm": 2.732379913330078,
      "learning_rate": 1.260666246201513e-06,
      "loss": 0.0569,
      "step": 14935
    },
    {
      "epoch": 0.7712884069197005,
      "grad_norm": 9.863664627075195,
      "learning_rate": 1.260122364761392e-06,
      "loss": 0.1278,
      "step": 14936
    },
    {
      "epoch": 0.7713400464756003,
      "grad_norm": 6.304014205932617,
      "learning_rate": 1.2595785837522285e-06,
      "loss": 0.0937,
      "step": 14937
    },
    {
      "epoch": 0.7713916860315001,
      "grad_norm": 2.841660499572754,
      "learning_rate": 1.2590349031886251e-06,
      "loss": 0.0402,
      "step": 14938
    },
    {
      "epoch": 0.7714433255873999,
      "grad_norm": 3.2533600330352783,
      "learning_rate": 1.258491323085182e-06,
      "loss": 0.0428,
      "step": 14939
    },
    {
      "epoch": 0.7714949651432997,
      "grad_norm": 7.166332244873047,
      "learning_rate": 1.2579478434564962e-06,
      "loss": 0.1122,
      "step": 14940
    },
    {
      "epoch": 0.7715466046991996,
      "grad_norm": 3.4872848987579346,
      "learning_rate": 1.2574044643171623e-06,
      "loss": 0.0282,
      "step": 14941
    },
    {
      "epoch": 0.7715982442550994,
      "grad_norm": 11.332138061523438,
      "learning_rate": 1.2568611856817731e-06,
      "loss": 0.1191,
      "step": 14942
    },
    {
      "epoch": 0.7716498838109992,
      "grad_norm": 10.62306022644043,
      "learning_rate": 1.2563180075649184e-06,
      "loss": 0.0769,
      "step": 14943
    },
    {
      "epoch": 0.771701523366899,
      "grad_norm": 5.644564628601074,
      "learning_rate": 1.2557749299811806e-06,
      "loss": 0.0904,
      "step": 14944
    },
    {
      "epoch": 0.7717531629227988,
      "grad_norm": 10.2434663772583,
      "learning_rate": 1.2552319529451485e-06,
      "loss": 0.0685,
      "step": 14945
    },
    {
      "epoch": 0.7718048024786986,
      "grad_norm": 2.0905606746673584,
      "learning_rate": 1.254689076471402e-06,
      "loss": 0.0132,
      "step": 14946
    },
    {
      "epoch": 0.7718564420345985,
      "grad_norm": 1.3480706214904785,
      "learning_rate": 1.2541463005745181e-06,
      "loss": 0.0224,
      "step": 14947
    },
    {
      "epoch": 0.7719080815904983,
      "grad_norm": 2.0465030670166016,
      "learning_rate": 1.253603625269073e-06,
      "loss": 0.0227,
      "step": 14948
    },
    {
      "epoch": 0.7719597211463981,
      "grad_norm": 2.1709628105163574,
      "learning_rate": 1.2530610505696401e-06,
      "loss": 0.0336,
      "step": 14949
    },
    {
      "epoch": 0.7720113607022979,
      "grad_norm": 9.588919639587402,
      "learning_rate": 1.2525185764907894e-06,
      "loss": 0.0565,
      "step": 14950
    },
    {
      "epoch": 0.7720630002581977,
      "grad_norm": 3.164370536804199,
      "learning_rate": 1.251976203047089e-06,
      "loss": 0.055,
      "step": 14951
    },
    {
      "epoch": 0.7721146398140976,
      "grad_norm": 6.880843162536621,
      "learning_rate": 1.2514339302531037e-06,
      "loss": 0.0594,
      "step": 14952
    },
    {
      "epoch": 0.7721662793699974,
      "grad_norm": 7.607063293457031,
      "learning_rate": 1.2508917581233955e-06,
      "loss": 0.1348,
      "step": 14953
    },
    {
      "epoch": 0.7722179189258972,
      "grad_norm": 47.43152618408203,
      "learning_rate": 1.250349686672524e-06,
      "loss": 0.1604,
      "step": 14954
    },
    {
      "epoch": 0.772269558481797,
      "grad_norm": 4.549509525299072,
      "learning_rate": 1.2498077159150467e-06,
      "loss": 0.116,
      "step": 14955
    },
    {
      "epoch": 0.7723211980376968,
      "grad_norm": 4.941812038421631,
      "learning_rate": 1.2492658458655166e-06,
      "loss": 0.0586,
      "step": 14956
    },
    {
      "epoch": 0.7723728375935967,
      "grad_norm": 12.48459529876709,
      "learning_rate": 1.2487240765384873e-06,
      "loss": 0.1788,
      "step": 14957
    },
    {
      "epoch": 0.7724244771494965,
      "grad_norm": 2.3511877059936523,
      "learning_rate": 1.2481824079485033e-06,
      "loss": 0.0403,
      "step": 14958
    },
    {
      "epoch": 0.7724761167053963,
      "grad_norm": 3.969823122024536,
      "learning_rate": 1.2476408401101141e-06,
      "loss": 0.0705,
      "step": 14959
    },
    {
      "epoch": 0.7725277562612961,
      "grad_norm": 1.0452295541763306,
      "learning_rate": 1.2470993730378628e-06,
      "loss": 0.0159,
      "step": 14960
    },
    {
      "epoch": 0.7725793958171959,
      "grad_norm": 4.13523006439209,
      "learning_rate": 1.2465580067462902e-06,
      "loss": 0.0587,
      "step": 14961
    },
    {
      "epoch": 0.7726310353730957,
      "grad_norm": 4.313895225524902,
      "learning_rate": 1.2460167412499324e-06,
      "loss": 0.0561,
      "step": 14962
    },
    {
      "epoch": 0.7726826749289956,
      "grad_norm": 15.69926643371582,
      "learning_rate": 1.2454755765633247e-06,
      "loss": 0.4018,
      "step": 14963
    },
    {
      "epoch": 0.7727343144848954,
      "grad_norm": 1.7741210460662842,
      "learning_rate": 1.2449345127009994e-06,
      "loss": 0.0287,
      "step": 14964
    },
    {
      "epoch": 0.7727859540407952,
      "grad_norm": 2.1126315593719482,
      "learning_rate": 1.24439354967749e-06,
      "loss": 0.0119,
      "step": 14965
    },
    {
      "epoch": 0.772837593596695,
      "grad_norm": 2.269217014312744,
      "learning_rate": 1.2438526875073199e-06,
      "loss": 0.0206,
      "step": 14966
    },
    {
      "epoch": 0.7728892331525948,
      "grad_norm": 2.3968982696533203,
      "learning_rate": 1.2433119262050136e-06,
      "loss": 0.0418,
      "step": 14967
    },
    {
      "epoch": 0.7729408727084947,
      "grad_norm": 3.1819489002227783,
      "learning_rate": 1.2427712657850937e-06,
      "loss": 0.087,
      "step": 14968
    },
    {
      "epoch": 0.7729925122643945,
      "grad_norm": 5.6363444328308105,
      "learning_rate": 1.2422307062620791e-06,
      "loss": 0.0741,
      "step": 14969
    },
    {
      "epoch": 0.7730441518202944,
      "grad_norm": 8.727580070495605,
      "learning_rate": 1.2416902476504861e-06,
      "loss": 0.1401,
      "step": 14970
    },
    {
      "epoch": 0.7730957913761942,
      "grad_norm": 3.931650400161743,
      "learning_rate": 1.2411498899648272e-06,
      "loss": 0.0346,
      "step": 14971
    },
    {
      "epoch": 0.773147430932094,
      "grad_norm": 3.0231258869171143,
      "learning_rate": 1.2406096332196144e-06,
      "loss": 0.0526,
      "step": 14972
    },
    {
      "epoch": 0.7731990704879939,
      "grad_norm": 3.96174955368042,
      "learning_rate": 1.2400694774293554e-06,
      "loss": 0.0454,
      "step": 14973
    },
    {
      "epoch": 0.7732507100438937,
      "grad_norm": 4.651068210601807,
      "learning_rate": 1.2395294226085559e-06,
      "loss": 0.0672,
      "step": 14974
    },
    {
      "epoch": 0.7733023495997935,
      "grad_norm": 9.160962104797363,
      "learning_rate": 1.2389894687717174e-06,
      "loss": 0.1054,
      "step": 14975
    },
    {
      "epoch": 0.7733539891556933,
      "grad_norm": 3.6365513801574707,
      "learning_rate": 1.238449615933343e-06,
      "loss": 0.0631,
      "step": 14976
    },
    {
      "epoch": 0.7734056287115931,
      "grad_norm": 3.1818313598632812,
      "learning_rate": 1.2379098641079245e-06,
      "loss": 0.0334,
      "step": 14977
    },
    {
      "epoch": 0.773457268267493,
      "grad_norm": 1.44497549533844,
      "learning_rate": 1.2373702133099607e-06,
      "loss": 0.0354,
      "step": 14978
    },
    {
      "epoch": 0.7735089078233928,
      "grad_norm": 3.0690972805023193,
      "learning_rate": 1.236830663553944e-06,
      "loss": 0.0301,
      "step": 14979
    },
    {
      "epoch": 0.7735605473792926,
      "grad_norm": 2.649109363555908,
      "learning_rate": 1.2362912148543604e-06,
      "loss": 0.0184,
      "step": 14980
    },
    {
      "epoch": 0.7736121869351924,
      "grad_norm": 7.694766998291016,
      "learning_rate": 1.2357518672256973e-06,
      "loss": 0.0765,
      "step": 14981
    },
    {
      "epoch": 0.7736638264910922,
      "grad_norm": 6.8286309242248535,
      "learning_rate": 1.2352126206824395e-06,
      "loss": 0.0602,
      "step": 14982
    },
    {
      "epoch": 0.773715466046992,
      "grad_norm": 2.764249563217163,
      "learning_rate": 1.2346734752390666e-06,
      "loss": 0.0494,
      "step": 14983
    },
    {
      "epoch": 0.7737671056028919,
      "grad_norm": 4.03234338760376,
      "learning_rate": 1.2341344309100578e-06,
      "loss": 0.0676,
      "step": 14984
    },
    {
      "epoch": 0.7738187451587917,
      "grad_norm": 4.769810199737549,
      "learning_rate": 1.2335954877098883e-06,
      "loss": 0.0454,
      "step": 14985
    },
    {
      "epoch": 0.7738703847146915,
      "grad_norm": 2.155691146850586,
      "learning_rate": 1.2330566456530313e-06,
      "loss": 0.0406,
      "step": 14986
    },
    {
      "epoch": 0.7739220242705913,
      "grad_norm": 2.3489296436309814,
      "learning_rate": 1.2325179047539559e-06,
      "loss": 0.0222,
      "step": 14987
    },
    {
      "epoch": 0.7739736638264911,
      "grad_norm": 7.763875484466553,
      "learning_rate": 1.2319792650271305e-06,
      "loss": 0.0336,
      "step": 14988
    },
    {
      "epoch": 0.774025303382391,
      "grad_norm": 6.666703224182129,
      "learning_rate": 1.23144072648702e-06,
      "loss": 0.0893,
      "step": 14989
    },
    {
      "epoch": 0.7740769429382908,
      "grad_norm": 5.629751205444336,
      "learning_rate": 1.2309022891480864e-06,
      "loss": 0.0697,
      "step": 14990
    },
    {
      "epoch": 0.7741285824941906,
      "grad_norm": 12.512645721435547,
      "learning_rate": 1.2303639530247858e-06,
      "loss": 0.0616,
      "step": 14991
    },
    {
      "epoch": 0.7741802220500904,
      "grad_norm": 2.037075996398926,
      "learning_rate": 1.2298257181315787e-06,
      "loss": 0.0272,
      "step": 14992
    },
    {
      "epoch": 0.7742318616059902,
      "grad_norm": 1.9675195217132568,
      "learning_rate": 1.2292875844829177e-06,
      "loss": 0.0367,
      "step": 14993
    },
    {
      "epoch": 0.77428350116189,
      "grad_norm": 3.439472198486328,
      "learning_rate": 1.2287495520932547e-06,
      "loss": 0.0695,
      "step": 14994
    },
    {
      "epoch": 0.7743351407177899,
      "grad_norm": 3.3449413776397705,
      "learning_rate": 1.2282116209770355e-06,
      "loss": 0.0368,
      "step": 14995
    },
    {
      "epoch": 0.7743867802736897,
      "grad_norm": 3.525587320327759,
      "learning_rate": 1.2276737911487058e-06,
      "loss": 0.0429,
      "step": 14996
    },
    {
      "epoch": 0.7744384198295895,
      "grad_norm": 9.087488174438477,
      "learning_rate": 1.2271360626227113e-06,
      "loss": 0.075,
      "step": 14997
    },
    {
      "epoch": 0.7744900593854893,
      "grad_norm": 2.868992567062378,
      "learning_rate": 1.226598435413493e-06,
      "loss": 0.0375,
      "step": 14998
    },
    {
      "epoch": 0.7745416989413891,
      "grad_norm": 8.404214859008789,
      "learning_rate": 1.2260609095354842e-06,
      "loss": 0.0697,
      "step": 14999
    },
    {
      "epoch": 0.774593338497289,
      "grad_norm": 5.648472309112549,
      "learning_rate": 1.2255234850031217e-06,
      "loss": 0.0594,
      "step": 15000
    },
    {
      "epoch": 0.7746449780531888,
      "grad_norm": 3.4896559715270996,
      "learning_rate": 1.224986161830838e-06,
      "loss": 0.039,
      "step": 15001
    },
    {
      "epoch": 0.7746966176090886,
      "grad_norm": 2.8205065727233887,
      "learning_rate": 1.2244489400330611e-06,
      "loss": 0.0645,
      "step": 15002
    },
    {
      "epoch": 0.7747482571649884,
      "grad_norm": 5.478281497955322,
      "learning_rate": 1.2239118196242195e-06,
      "loss": 0.1045,
      "step": 15003
    },
    {
      "epoch": 0.7747998967208882,
      "grad_norm": 4.937252998352051,
      "learning_rate": 1.2233748006187352e-06,
      "loss": 0.0641,
      "step": 15004
    },
    {
      "epoch": 0.774851536276788,
      "grad_norm": 9.41075611114502,
      "learning_rate": 1.2228378830310305e-06,
      "loss": 0.1091,
      "step": 15005
    },
    {
      "epoch": 0.7749031758326879,
      "grad_norm": 1.3401482105255127,
      "learning_rate": 1.2223010668755237e-06,
      "loss": 0.0203,
      "step": 15006
    },
    {
      "epoch": 0.7749548153885877,
      "grad_norm": 5.914844036102295,
      "learning_rate": 1.22176435216663e-06,
      "loss": 0.0657,
      "step": 15007
    },
    {
      "epoch": 0.7750064549444875,
      "grad_norm": 16.29631233215332,
      "learning_rate": 1.2212277389187628e-06,
      "loss": 0.0584,
      "step": 15008
    },
    {
      "epoch": 0.7750580945003873,
      "grad_norm": 3.081737518310547,
      "learning_rate": 1.2206912271463333e-06,
      "loss": 0.0429,
      "step": 15009
    },
    {
      "epoch": 0.7751097340562871,
      "grad_norm": 8.84710693359375,
      "learning_rate": 1.2201548168637455e-06,
      "loss": 0.0436,
      "step": 15010
    },
    {
      "epoch": 0.775161373612187,
      "grad_norm": 3.5517489910125732,
      "learning_rate": 1.2196185080854084e-06,
      "loss": 0.0369,
      "step": 15011
    },
    {
      "epoch": 0.7752130131680868,
      "grad_norm": 3.1324880123138428,
      "learning_rate": 1.2190823008257236e-06,
      "loss": 0.0486,
      "step": 15012
    },
    {
      "epoch": 0.7752646527239866,
      "grad_norm": 2.481201648712158,
      "learning_rate": 1.2185461950990874e-06,
      "loss": 0.0378,
      "step": 15013
    },
    {
      "epoch": 0.7753162922798864,
      "grad_norm": 6.567210674285889,
      "learning_rate": 1.2180101909198983e-06,
      "loss": 0.0647,
      "step": 15014
    },
    {
      "epoch": 0.7753679318357862,
      "grad_norm": 1.7036312818527222,
      "learning_rate": 1.2174742883025486e-06,
      "loss": 0.0264,
      "step": 15015
    },
    {
      "epoch": 0.775419571391686,
      "grad_norm": 4.655876159667969,
      "learning_rate": 1.2169384872614343e-06,
      "loss": 0.0383,
      "step": 15016
    },
    {
      "epoch": 0.7754712109475859,
      "grad_norm": 1.1663395166397095,
      "learning_rate": 1.2164027878109385e-06,
      "loss": 0.017,
      "step": 15017
    },
    {
      "epoch": 0.7755228505034857,
      "grad_norm": 15.790552139282227,
      "learning_rate": 1.2158671899654496e-06,
      "loss": 0.1463,
      "step": 15018
    },
    {
      "epoch": 0.7755744900593855,
      "grad_norm": 3.9808263778686523,
      "learning_rate": 1.2153316937393493e-06,
      "loss": 0.0588,
      "step": 15019
    },
    {
      "epoch": 0.7756261296152853,
      "grad_norm": 2.477065086364746,
      "learning_rate": 1.2147962991470187e-06,
      "loss": 0.0283,
      "step": 15020
    },
    {
      "epoch": 0.7756777691711851,
      "grad_norm": 3.48423171043396,
      "learning_rate": 1.2142610062028353e-06,
      "loss": 0.0714,
      "step": 15021
    },
    {
      "epoch": 0.775729408727085,
      "grad_norm": 3.916330099105835,
      "learning_rate": 1.2137258149211733e-06,
      "loss": 0.0542,
      "step": 15022
    },
    {
      "epoch": 0.7757810482829848,
      "grad_norm": 3.3548059463500977,
      "learning_rate": 1.213190725316407e-06,
      "loss": 0.0575,
      "step": 15023
    },
    {
      "epoch": 0.7758326878388846,
      "grad_norm": 2.728229284286499,
      "learning_rate": 1.2126557374029009e-06,
      "loss": 0.0372,
      "step": 15024
    },
    {
      "epoch": 0.7758843273947844,
      "grad_norm": 2.79646897315979,
      "learning_rate": 1.212120851195026e-06,
      "loss": 0.0326,
      "step": 15025
    },
    {
      "epoch": 0.7759359669506842,
      "grad_norm": 6.410023212432861,
      "learning_rate": 1.2115860667071449e-06,
      "loss": 0.0181,
      "step": 15026
    },
    {
      "epoch": 0.775987606506584,
      "grad_norm": 1.204035997390747,
      "learning_rate": 1.2110513839536198e-06,
      "loss": 0.0117,
      "step": 15027
    },
    {
      "epoch": 0.7760392460624839,
      "grad_norm": 14.793451309204102,
      "learning_rate": 1.2105168029488068e-06,
      "loss": 0.1073,
      "step": 15028
    },
    {
      "epoch": 0.7760908856183837,
      "grad_norm": 4.049100399017334,
      "learning_rate": 1.2099823237070612e-06,
      "loss": 0.0503,
      "step": 15029
    },
    {
      "epoch": 0.7761425251742835,
      "grad_norm": 3.9579615592956543,
      "learning_rate": 1.2094479462427388e-06,
      "loss": 0.0314,
      "step": 15030
    },
    {
      "epoch": 0.7761941647301833,
      "grad_norm": 22.2725830078125,
      "learning_rate": 1.2089136705701898e-06,
      "loss": 0.1125,
      "step": 15031
    },
    {
      "epoch": 0.7762458042860831,
      "grad_norm": 3.5425596237182617,
      "learning_rate": 1.2083794967037593e-06,
      "loss": 0.0363,
      "step": 15032
    },
    {
      "epoch": 0.776297443841983,
      "grad_norm": 6.8612871170043945,
      "learning_rate": 1.207845424657793e-06,
      "loss": 0.06,
      "step": 15033
    },
    {
      "epoch": 0.7763490833978828,
      "grad_norm": 10.180188179016113,
      "learning_rate": 1.207311454446633e-06,
      "loss": 0.0841,
      "step": 15034
    },
    {
      "epoch": 0.7764007229537826,
      "grad_norm": 3.7228481769561768,
      "learning_rate": 1.2067775860846182e-06,
      "loss": 0.0558,
      "step": 15035
    },
    {
      "epoch": 0.7764523625096824,
      "grad_norm": 3.1009669303894043,
      "learning_rate": 1.206243819586086e-06,
      "loss": 0.0548,
      "step": 15036
    },
    {
      "epoch": 0.7765040020655822,
      "grad_norm": 2.918982982635498,
      "learning_rate": 1.2057101549653693e-06,
      "loss": 0.0338,
      "step": 15037
    },
    {
      "epoch": 0.776555641621482,
      "grad_norm": 2.604461431503296,
      "learning_rate": 1.2051765922368003e-06,
      "loss": 0.0372,
      "step": 15038
    },
    {
      "epoch": 0.7766072811773819,
      "grad_norm": 4.462684154510498,
      "learning_rate": 1.2046431314147062e-06,
      "loss": 0.0709,
      "step": 15039
    },
    {
      "epoch": 0.7766589207332817,
      "grad_norm": 1.854550838470459,
      "learning_rate": 1.2041097725134126e-06,
      "loss": 0.0389,
      "step": 15040
    },
    {
      "epoch": 0.7767105602891815,
      "grad_norm": 1.9206979274749756,
      "learning_rate": 1.2035765155472435e-06,
      "loss": 0.0122,
      "step": 15041
    },
    {
      "epoch": 0.7767621998450813,
      "grad_norm": 3.4304022789001465,
      "learning_rate": 1.2030433605305197e-06,
      "loss": 0.0541,
      "step": 15042
    },
    {
      "epoch": 0.7768138394009811,
      "grad_norm": 2.0246527194976807,
      "learning_rate": 1.202510307477554e-06,
      "loss": 0.035,
      "step": 15043
    },
    {
      "epoch": 0.776865478956881,
      "grad_norm": 3.528651714324951,
      "learning_rate": 1.201977356402666e-06,
      "loss": 0.0528,
      "step": 15044
    },
    {
      "epoch": 0.7769171185127808,
      "grad_norm": 14.83081340789795,
      "learning_rate": 1.2014445073201668e-06,
      "loss": 0.0974,
      "step": 15045
    },
    {
      "epoch": 0.7769687580686806,
      "grad_norm": 3.5776865482330322,
      "learning_rate": 1.2009117602443638e-06,
      "loss": 0.0554,
      "step": 15046
    },
    {
      "epoch": 0.7770203976245804,
      "grad_norm": 3.4018337726593018,
      "learning_rate": 1.2003791151895644e-06,
      "loss": 0.0395,
      "step": 15047
    },
    {
      "epoch": 0.7770720371804802,
      "grad_norm": 2.097996234893799,
      "learning_rate": 1.1998465721700703e-06,
      "loss": 0.0212,
      "step": 15048
    },
    {
      "epoch": 0.77712367673638,
      "grad_norm": 3.7861878871917725,
      "learning_rate": 1.1993141312001871e-06,
      "loss": 0.0557,
      "step": 15049
    },
    {
      "epoch": 0.7771753162922799,
      "grad_norm": 2.656846284866333,
      "learning_rate": 1.1987817922942091e-06,
      "loss": 0.0167,
      "step": 15050
    },
    {
      "epoch": 0.7772269558481797,
      "grad_norm": 14.80810832977295,
      "learning_rate": 1.1982495554664326e-06,
      "loss": 0.0574,
      "step": 15051
    },
    {
      "epoch": 0.7772785954040795,
      "grad_norm": 2.5530576705932617,
      "learning_rate": 1.1977174207311504e-06,
      "loss": 0.059,
      "step": 15052
    },
    {
      "epoch": 0.7773302349599793,
      "grad_norm": 2.1502010822296143,
      "learning_rate": 1.1971853881026529e-06,
      "loss": 0.0369,
      "step": 15053
    },
    {
      "epoch": 0.7773818745158791,
      "grad_norm": 2.83017897605896,
      "learning_rate": 1.1966534575952276e-06,
      "loss": 0.0428,
      "step": 15054
    },
    {
      "epoch": 0.777433514071779,
      "grad_norm": 13.079056739807129,
      "learning_rate": 1.1961216292231581e-06,
      "loss": 0.1034,
      "step": 15055
    },
    {
      "epoch": 0.7774851536276788,
      "grad_norm": 23.36273956298828,
      "learning_rate": 1.1955899030007278e-06,
      "loss": 0.1086,
      "step": 15056
    },
    {
      "epoch": 0.7775367931835786,
      "grad_norm": 3.173760414123535,
      "learning_rate": 1.1950582789422122e-06,
      "loss": 0.048,
      "step": 15057
    },
    {
      "epoch": 0.7775884327394784,
      "grad_norm": 4.970611095428467,
      "learning_rate": 1.1945267570618914e-06,
      "loss": 0.0545,
      "step": 15058
    },
    {
      "epoch": 0.7776400722953782,
      "grad_norm": 8.49023723602295,
      "learning_rate": 1.193995337374037e-06,
      "loss": 0.1155,
      "step": 15059
    },
    {
      "epoch": 0.7776917118512781,
      "grad_norm": 2.055333137512207,
      "learning_rate": 1.1934640198929215e-06,
      "loss": 0.0313,
      "step": 15060
    },
    {
      "epoch": 0.7777433514071779,
      "grad_norm": 3.505922794342041,
      "learning_rate": 1.192932804632811e-06,
      "loss": 0.0656,
      "step": 15061
    },
    {
      "epoch": 0.7777949909630777,
      "grad_norm": 2.190237522125244,
      "learning_rate": 1.1924016916079694e-06,
      "loss": 0.0349,
      "step": 15062
    },
    {
      "epoch": 0.7778466305189775,
      "grad_norm": 3.300868511199951,
      "learning_rate": 1.1918706808326625e-06,
      "loss": 0.0588,
      "step": 15063
    },
    {
      "epoch": 0.7778982700748773,
      "grad_norm": 1.738762617111206,
      "learning_rate": 1.1913397723211506e-06,
      "loss": 0.0186,
      "step": 15064
    },
    {
      "epoch": 0.7779499096307771,
      "grad_norm": 4.778566837310791,
      "learning_rate": 1.190808966087688e-06,
      "loss": 0.0557,
      "step": 15065
    },
    {
      "epoch": 0.778001549186677,
      "grad_norm": 3.898799180984497,
      "learning_rate": 1.1902782621465297e-06,
      "loss": 0.056,
      "step": 15066
    },
    {
      "epoch": 0.7780531887425768,
      "grad_norm": 3.662468910217285,
      "learning_rate": 1.189747660511928e-06,
      "loss": 0.0508,
      "step": 15067
    },
    {
      "epoch": 0.7781048282984766,
      "grad_norm": 6.52589750289917,
      "learning_rate": 1.1892171611981306e-06,
      "loss": 0.0614,
      "step": 15068
    },
    {
      "epoch": 0.7781564678543764,
      "grad_norm": 5.765103816986084,
      "learning_rate": 1.188686764219385e-06,
      "loss": 0.0933,
      "step": 15069
    },
    {
      "epoch": 0.7782081074102762,
      "grad_norm": 6.397564888000488,
      "learning_rate": 1.1881564695899333e-06,
      "loss": 0.0297,
      "step": 15070
    },
    {
      "epoch": 0.7782597469661761,
      "grad_norm": 24.60107421875,
      "learning_rate": 1.1876262773240172e-06,
      "loss": 0.0937,
      "step": 15071
    },
    {
      "epoch": 0.7783113865220759,
      "grad_norm": 4.839584827423096,
      "learning_rate": 1.1870961874358733e-06,
      "loss": 0.0367,
      "step": 15072
    },
    {
      "epoch": 0.7783630260779757,
      "grad_norm": 9.736554145812988,
      "learning_rate": 1.1865661999397376e-06,
      "loss": 0.0563,
      "step": 15073
    },
    {
      "epoch": 0.7784146656338755,
      "grad_norm": 5.442837238311768,
      "learning_rate": 1.186036314849842e-06,
      "loss": 0.0258,
      "step": 15074
    },
    {
      "epoch": 0.7784663051897753,
      "grad_norm": 3.7941343784332275,
      "learning_rate": 1.1855065321804177e-06,
      "loss": 0.0565,
      "step": 15075
    },
    {
      "epoch": 0.7785179447456752,
      "grad_norm": 2.7100138664245605,
      "learning_rate": 1.1849768519456872e-06,
      "loss": 0.0206,
      "step": 15076
    },
    {
      "epoch": 0.778569584301575,
      "grad_norm": 1.4662774801254272,
      "learning_rate": 1.184447274159879e-06,
      "loss": 0.0175,
      "step": 15077
    },
    {
      "epoch": 0.7786212238574748,
      "grad_norm": 2.6372523307800293,
      "learning_rate": 1.1839177988372141e-06,
      "loss": 0.0465,
      "step": 15078
    },
    {
      "epoch": 0.7786728634133746,
      "grad_norm": 5.519954681396484,
      "learning_rate": 1.1833884259919086e-06,
      "loss": 0.0486,
      "step": 15079
    },
    {
      "epoch": 0.7787245029692744,
      "grad_norm": 11.363581657409668,
      "learning_rate": 1.182859155638179e-06,
      "loss": 0.0872,
      "step": 15080
    },
    {
      "epoch": 0.7787761425251742,
      "grad_norm": 9.8609619140625,
      "learning_rate": 1.1823299877902378e-06,
      "loss": 0.0809,
      "step": 15081
    },
    {
      "epoch": 0.7788277820810741,
      "grad_norm": 2.8883919715881348,
      "learning_rate": 1.1818009224622996e-06,
      "loss": 0.0333,
      "step": 15082
    },
    {
      "epoch": 0.7788794216369739,
      "grad_norm": 6.049563884735107,
      "learning_rate": 1.1812719596685668e-06,
      "loss": 0.0751,
      "step": 15083
    },
    {
      "epoch": 0.7789310611928737,
      "grad_norm": 3.7851109504699707,
      "learning_rate": 1.180743099423246e-06,
      "loss": 0.0514,
      "step": 15084
    },
    {
      "epoch": 0.7789827007487735,
      "grad_norm": 2.1941792964935303,
      "learning_rate": 1.18021434174054e-06,
      "loss": 0.0437,
      "step": 15085
    },
    {
      "epoch": 0.7790343403046733,
      "grad_norm": 2.983360767364502,
      "learning_rate": 1.1796856866346467e-06,
      "loss": 0.0291,
      "step": 15086
    },
    {
      "epoch": 0.7790859798605732,
      "grad_norm": 8.940524101257324,
      "learning_rate": 1.1791571341197643e-06,
      "loss": 0.0698,
      "step": 15087
    },
    {
      "epoch": 0.779137619416473,
      "grad_norm": 2.898690700531006,
      "learning_rate": 1.1786286842100852e-06,
      "loss": 0.0294,
      "step": 15088
    },
    {
      "epoch": 0.7791892589723728,
      "grad_norm": 2.9511780738830566,
      "learning_rate": 1.1781003369198023e-06,
      "loss": 0.0347,
      "step": 15089
    },
    {
      "epoch": 0.7792408985282726,
      "grad_norm": 1.813881278038025,
      "learning_rate": 1.1775720922631e-06,
      "loss": 0.0307,
      "step": 15090
    },
    {
      "epoch": 0.7792925380841724,
      "grad_norm": 4.927114009857178,
      "learning_rate": 1.177043950254168e-06,
      "loss": 0.0605,
      "step": 15091
    },
    {
      "epoch": 0.7793441776400722,
      "grad_norm": 5.329047679901123,
      "learning_rate": 1.176515910907187e-06,
      "loss": 0.0206,
      "step": 15092
    },
    {
      "epoch": 0.7793958171959721,
      "grad_norm": 3.4023122787475586,
      "learning_rate": 1.1759879742363394e-06,
      "loss": 0.0587,
      "step": 15093
    },
    {
      "epoch": 0.7794474567518719,
      "grad_norm": 5.003687858581543,
      "learning_rate": 1.1754601402557985e-06,
      "loss": 0.0446,
      "step": 15094
    },
    {
      "epoch": 0.7794990963077717,
      "grad_norm": 9.055323600769043,
      "learning_rate": 1.1749324089797397e-06,
      "loss": 0.0374,
      "step": 15095
    },
    {
      "epoch": 0.7795507358636715,
      "grad_norm": 9.796465873718262,
      "learning_rate": 1.1744047804223373e-06,
      "loss": 0.1062,
      "step": 15096
    },
    {
      "epoch": 0.7796023754195713,
      "grad_norm": 2.861940383911133,
      "learning_rate": 1.17387725459776e-06,
      "loss": 0.0403,
      "step": 15097
    },
    {
      "epoch": 0.7796540149754713,
      "grad_norm": 3.0109260082244873,
      "learning_rate": 1.173349831520172e-06,
      "loss": 0.0205,
      "step": 15098
    },
    {
      "epoch": 0.7797056545313711,
      "grad_norm": 26.314687728881836,
      "learning_rate": 1.1728225112037367e-06,
      "loss": 0.0859,
      "step": 15099
    },
    {
      "epoch": 0.7797572940872709,
      "grad_norm": 12.26321029663086,
      "learning_rate": 1.1722952936626165e-06,
      "loss": 0.0891,
      "step": 15100
    },
    {
      "epoch": 0.7798089336431707,
      "grad_norm": 3.8481743335723877,
      "learning_rate": 1.171768178910968e-06,
      "loss": 0.0768,
      "step": 15101
    },
    {
      "epoch": 0.7798605731990705,
      "grad_norm": 3.39750075340271,
      "learning_rate": 1.171241166962947e-06,
      "loss": 0.0356,
      "step": 15102
    },
    {
      "epoch": 0.7799122127549704,
      "grad_norm": 2.3134765625,
      "learning_rate": 1.1707142578327058e-06,
      "loss": 0.0248,
      "step": 15103
    },
    {
      "epoch": 0.7799638523108702,
      "grad_norm": 7.472953796386719,
      "learning_rate": 1.170187451534394e-06,
      "loss": 0.0626,
      "step": 15104
    },
    {
      "epoch": 0.78001549186677,
      "grad_norm": 5.701663017272949,
      "learning_rate": 1.1696607480821591e-06,
      "loss": 0.0736,
      "step": 15105
    },
    {
      "epoch": 0.7800671314226698,
      "grad_norm": 3.303345203399658,
      "learning_rate": 1.1691341474901441e-06,
      "loss": 0.0418,
      "step": 15106
    },
    {
      "epoch": 0.7801187709785696,
      "grad_norm": 8.05898666381836,
      "learning_rate": 1.1686076497724913e-06,
      "loss": 0.0766,
      "step": 15107
    },
    {
      "epoch": 0.7801704105344694,
      "grad_norm": 3.3506767749786377,
      "learning_rate": 1.1680812549433407e-06,
      "loss": 0.0541,
      "step": 15108
    },
    {
      "epoch": 0.7802220500903693,
      "grad_norm": 5.058639049530029,
      "learning_rate": 1.1675549630168232e-06,
      "loss": 0.0634,
      "step": 15109
    },
    {
      "epoch": 0.7802736896462691,
      "grad_norm": 3.9238293170928955,
      "learning_rate": 1.1670287740070763e-06,
      "loss": 0.0533,
      "step": 15110
    },
    {
      "epoch": 0.7803253292021689,
      "grad_norm": 3.978264808654785,
      "learning_rate": 1.1665026879282309e-06,
      "loss": 0.0354,
      "step": 15111
    },
    {
      "epoch": 0.7803769687580687,
      "grad_norm": 12.520145416259766,
      "learning_rate": 1.1659767047944116e-06,
      "loss": 0.1307,
      "step": 15112
    },
    {
      "epoch": 0.7804286083139685,
      "grad_norm": 3.1262118816375732,
      "learning_rate": 1.1654508246197437e-06,
      "loss": 0.0653,
      "step": 15113
    },
    {
      "epoch": 0.7804802478698684,
      "grad_norm": 13.373217582702637,
      "learning_rate": 1.1649250474183493e-06,
      "loss": 0.0357,
      "step": 15114
    },
    {
      "epoch": 0.7805318874257682,
      "grad_norm": 2.7995047569274902,
      "learning_rate": 1.1643993732043507e-06,
      "loss": 0.0421,
      "step": 15115
    },
    {
      "epoch": 0.780583526981668,
      "grad_norm": 2.6768646240234375,
      "learning_rate": 1.1638738019918605e-06,
      "loss": 0.029,
      "step": 15116
    },
    {
      "epoch": 0.7806351665375678,
      "grad_norm": 11.979048728942871,
      "learning_rate": 1.163348333794994e-06,
      "loss": 0.0572,
      "step": 15117
    },
    {
      "epoch": 0.7806868060934676,
      "grad_norm": 2.495774507522583,
      "learning_rate": 1.162822968627862e-06,
      "loss": 0.037,
      "step": 15118
    },
    {
      "epoch": 0.7807384456493675,
      "grad_norm": 3.679792881011963,
      "learning_rate": 1.1622977065045727e-06,
      "loss": 0.0527,
      "step": 15119
    },
    {
      "epoch": 0.7807900852052673,
      "grad_norm": 8.225113868713379,
      "learning_rate": 1.1617725474392316e-06,
      "loss": 0.0707,
      "step": 15120
    },
    {
      "epoch": 0.7808417247611671,
      "grad_norm": 5.260538101196289,
      "learning_rate": 1.1612474914459415e-06,
      "loss": 0.0553,
      "step": 15121
    },
    {
      "epoch": 0.7808933643170669,
      "grad_norm": 2.4564905166625977,
      "learning_rate": 1.1607225385388038e-06,
      "loss": 0.0275,
      "step": 15122
    },
    {
      "epoch": 0.7809450038729667,
      "grad_norm": 2.556011438369751,
      "learning_rate": 1.1601976887319106e-06,
      "loss": 0.0432,
      "step": 15123
    },
    {
      "epoch": 0.7809966434288665,
      "grad_norm": 2.073437213897705,
      "learning_rate": 1.1596729420393616e-06,
      "loss": 0.0307,
      "step": 15124
    },
    {
      "epoch": 0.7810482829847664,
      "grad_norm": 2.3577723503112793,
      "learning_rate": 1.1591482984752462e-06,
      "loss": 0.0239,
      "step": 15125
    },
    {
      "epoch": 0.7810999225406662,
      "grad_norm": 5.547507286071777,
      "learning_rate": 1.1586237580536547e-06,
      "loss": 0.0611,
      "step": 15126
    },
    {
      "epoch": 0.781151562096566,
      "grad_norm": 2.029691696166992,
      "learning_rate": 1.1580993207886703e-06,
      "loss": 0.0255,
      "step": 15127
    },
    {
      "epoch": 0.7812032016524658,
      "grad_norm": 4.385290622711182,
      "learning_rate": 1.157574986694377e-06,
      "loss": 0.0806,
      "step": 15128
    },
    {
      "epoch": 0.7812548412083656,
      "grad_norm": 2.7348978519439697,
      "learning_rate": 1.157050755784858e-06,
      "loss": 0.0205,
      "step": 15129
    },
    {
      "epoch": 0.7813064807642655,
      "grad_norm": 3.3776144981384277,
      "learning_rate": 1.1565266280741887e-06,
      "loss": 0.0807,
      "step": 15130
    },
    {
      "epoch": 0.7813581203201653,
      "grad_norm": 15.127290725708008,
      "learning_rate": 1.1560026035764443e-06,
      "loss": 0.1292,
      "step": 15131
    },
    {
      "epoch": 0.7814097598760651,
      "grad_norm": 2.538635730743408,
      "learning_rate": 1.1554786823056973e-06,
      "loss": 0.0441,
      "step": 15132
    },
    {
      "epoch": 0.7814613994319649,
      "grad_norm": 2.531005382537842,
      "learning_rate": 1.1549548642760172e-06,
      "loss": 0.0282,
      "step": 15133
    },
    {
      "epoch": 0.7815130389878647,
      "grad_norm": 4.522583961486816,
      "learning_rate": 1.1544311495014703e-06,
      "loss": 0.0517,
      "step": 15134
    },
    {
      "epoch": 0.7815646785437645,
      "grad_norm": 6.0715179443359375,
      "learning_rate": 1.1539075379961207e-06,
      "loss": 0.0815,
      "step": 15135
    },
    {
      "epoch": 0.7816163180996644,
      "grad_norm": 7.595909595489502,
      "learning_rate": 1.1533840297740295e-06,
      "loss": 0.0716,
      "step": 15136
    },
    {
      "epoch": 0.7816679576555642,
      "grad_norm": 1.7701025009155273,
      "learning_rate": 1.1528606248492552e-06,
      "loss": 0.0241,
      "step": 15137
    },
    {
      "epoch": 0.781719597211464,
      "grad_norm": 2.7542638778686523,
      "learning_rate": 1.152337323235853e-06,
      "loss": 0.0496,
      "step": 15138
    },
    {
      "epoch": 0.7817712367673638,
      "grad_norm": 7.32703161239624,
      "learning_rate": 1.1518141249478766e-06,
      "loss": 0.049,
      "step": 15139
    },
    {
      "epoch": 0.7818228763232636,
      "grad_norm": 3.6872692108154297,
      "learning_rate": 1.1512910299993757e-06,
      "loss": 0.0434,
      "step": 15140
    },
    {
      "epoch": 0.7818745158791635,
      "grad_norm": 4.8951005935668945,
      "learning_rate": 1.1507680384043961e-06,
      "loss": 0.0598,
      "step": 15141
    },
    {
      "epoch": 0.7819261554350633,
      "grad_norm": 19.884765625,
      "learning_rate": 1.1502451501769823e-06,
      "loss": 0.2105,
      "step": 15142
    },
    {
      "epoch": 0.7819777949909631,
      "grad_norm": 3.6677639484405518,
      "learning_rate": 1.149722365331178e-06,
      "loss": 0.0458,
      "step": 15143
    },
    {
      "epoch": 0.7820294345468629,
      "grad_norm": 5.607397079467773,
      "learning_rate": 1.149199683881022e-06,
      "loss": 0.0648,
      "step": 15144
    },
    {
      "epoch": 0.7820810741027627,
      "grad_norm": 3.570840358734131,
      "learning_rate": 1.1486771058405489e-06,
      "loss": 0.0517,
      "step": 15145
    },
    {
      "epoch": 0.7821327136586625,
      "grad_norm": 4.478672027587891,
      "learning_rate": 1.1481546312237918e-06,
      "loss": 0.0335,
      "step": 15146
    },
    {
      "epoch": 0.7821843532145624,
      "grad_norm": 5.666256904602051,
      "learning_rate": 1.1476322600447815e-06,
      "loss": 0.0658,
      "step": 15147
    },
    {
      "epoch": 0.7822359927704622,
      "grad_norm": 4.6099772453308105,
      "learning_rate": 1.1471099923175489e-06,
      "loss": 0.0611,
      "step": 15148
    },
    {
      "epoch": 0.782287632326362,
      "grad_norm": 7.406773090362549,
      "learning_rate": 1.146587828056115e-06,
      "loss": 0.0306,
      "step": 15149
    },
    {
      "epoch": 0.7823392718822618,
      "grad_norm": 3.6449100971221924,
      "learning_rate": 1.1460657672745034e-06,
      "loss": 0.0402,
      "step": 15150
    },
    {
      "epoch": 0.7823909114381616,
      "grad_norm": 2.992241621017456,
      "learning_rate": 1.145543809986734e-06,
      "loss": 0.0558,
      "step": 15151
    },
    {
      "epoch": 0.7824425509940615,
      "grad_norm": 2.506964921951294,
      "learning_rate": 1.1450219562068226e-06,
      "loss": 0.0226,
      "step": 15152
    },
    {
      "epoch": 0.7824941905499613,
      "grad_norm": 12.731687545776367,
      "learning_rate": 1.1445002059487843e-06,
      "loss": 0.089,
      "step": 15153
    },
    {
      "epoch": 0.7825458301058611,
      "grad_norm": 1.8049129247665405,
      "learning_rate": 1.1439785592266294e-06,
      "loss": 0.019,
      "step": 15154
    },
    {
      "epoch": 0.7825974696617609,
      "grad_norm": 3.611431121826172,
      "learning_rate": 1.1434570160543684e-06,
      "loss": 0.0339,
      "step": 15155
    },
    {
      "epoch": 0.7826491092176607,
      "grad_norm": 2.639084577560425,
      "learning_rate": 1.1429355764460016e-06,
      "loss": 0.0335,
      "step": 15156
    },
    {
      "epoch": 0.7827007487735606,
      "grad_norm": 2.429248571395874,
      "learning_rate": 1.1424142404155364e-06,
      "loss": 0.034,
      "step": 15157
    },
    {
      "epoch": 0.7827523883294604,
      "grad_norm": 9.625608444213867,
      "learning_rate": 1.1418930079769718e-06,
      "loss": 0.0676,
      "step": 15158
    },
    {
      "epoch": 0.7828040278853602,
      "grad_norm": 2.9155075550079346,
      "learning_rate": 1.1413718791443062e-06,
      "loss": 0.0363,
      "step": 15159
    },
    {
      "epoch": 0.78285566744126,
      "grad_norm": 2.676405191421509,
      "learning_rate": 1.1408508539315305e-06,
      "loss": 0.0279,
      "step": 15160
    },
    {
      "epoch": 0.7829073069971598,
      "grad_norm": 1.7860020399093628,
      "learning_rate": 1.1403299323526379e-06,
      "loss": 0.0194,
      "step": 15161
    },
    {
      "epoch": 0.7829589465530596,
      "grad_norm": 3.6527299880981445,
      "learning_rate": 1.13980911442162e-06,
      "loss": 0.0606,
      "step": 15162
    },
    {
      "epoch": 0.7830105861089595,
      "grad_norm": 2.7267580032348633,
      "learning_rate": 1.139288400152459e-06,
      "loss": 0.0311,
      "step": 15163
    },
    {
      "epoch": 0.7830622256648593,
      "grad_norm": 3.4107367992401123,
      "learning_rate": 1.1387677895591404e-06,
      "loss": 0.0594,
      "step": 15164
    },
    {
      "epoch": 0.7831138652207591,
      "grad_norm": 13.735808372497559,
      "learning_rate": 1.138247282655644e-06,
      "loss": 0.0624,
      "step": 15165
    },
    {
      "epoch": 0.7831655047766589,
      "grad_norm": 4.602413177490234,
      "learning_rate": 1.1377268794559476e-06,
      "loss": 0.048,
      "step": 15166
    },
    {
      "epoch": 0.7832171443325587,
      "grad_norm": 2.602325201034546,
      "learning_rate": 1.1372065799740262e-06,
      "loss": 0.0329,
      "step": 15167
    },
    {
      "epoch": 0.7832687838884586,
      "grad_norm": 8.201436042785645,
      "learning_rate": 1.1366863842238518e-06,
      "loss": 0.063,
      "step": 15168
    },
    {
      "epoch": 0.7833204234443584,
      "grad_norm": 10.104413986206055,
      "learning_rate": 1.1361662922193944e-06,
      "loss": 0.0975,
      "step": 15169
    },
    {
      "epoch": 0.7833720630002582,
      "grad_norm": 11.29484748840332,
      "learning_rate": 1.1356463039746202e-06,
      "loss": 0.1737,
      "step": 15170
    },
    {
      "epoch": 0.783423702556158,
      "grad_norm": 3.212595224380493,
      "learning_rate": 1.1351264195034929e-06,
      "loss": 0.0559,
      "step": 15171
    },
    {
      "epoch": 0.7834753421120578,
      "grad_norm": 1.5931693315505981,
      "learning_rate": 1.1346066388199733e-06,
      "loss": 0.0234,
      "step": 15172
    },
    {
      "epoch": 0.7835269816679576,
      "grad_norm": 3.9170401096343994,
      "learning_rate": 1.1340869619380213e-06,
      "loss": 0.0506,
      "step": 15173
    },
    {
      "epoch": 0.7835786212238575,
      "grad_norm": 2.427793502807617,
      "learning_rate": 1.1335673888715898e-06,
      "loss": 0.0122,
      "step": 15174
    },
    {
      "epoch": 0.7836302607797573,
      "grad_norm": 3.442291498184204,
      "learning_rate": 1.133047919634631e-06,
      "loss": 0.0513,
      "step": 15175
    },
    {
      "epoch": 0.7836819003356571,
      "grad_norm": 1.045413851737976,
      "learning_rate": 1.1325285542410986e-06,
      "loss": 0.0079,
      "step": 15176
    },
    {
      "epoch": 0.7837335398915569,
      "grad_norm": 6.895481586456299,
      "learning_rate": 1.132009292704938e-06,
      "loss": 0.0393,
      "step": 15177
    },
    {
      "epoch": 0.7837851794474567,
      "grad_norm": 11.629209518432617,
      "learning_rate": 1.131490135040092e-06,
      "loss": 0.2248,
      "step": 15178
    },
    {
      "epoch": 0.7838368190033566,
      "grad_norm": 5.118839263916016,
      "learning_rate": 1.1309710812605023e-06,
      "loss": 0.0599,
      "step": 15179
    },
    {
      "epoch": 0.7838884585592564,
      "grad_norm": 2.516068696975708,
      "learning_rate": 1.1304521313801076e-06,
      "loss": 0.0233,
      "step": 15180
    },
    {
      "epoch": 0.7839400981151562,
      "grad_norm": 7.704626083374023,
      "learning_rate": 1.1299332854128476e-06,
      "loss": 0.1105,
      "step": 15181
    },
    {
      "epoch": 0.783991737671056,
      "grad_norm": 4.119483470916748,
      "learning_rate": 1.1294145433726505e-06,
      "loss": 0.0412,
      "step": 15182
    },
    {
      "epoch": 0.7840433772269558,
      "grad_norm": 3.9168426990509033,
      "learning_rate": 1.1288959052734489e-06,
      "loss": 0.0493,
      "step": 15183
    },
    {
      "epoch": 0.7840950167828556,
      "grad_norm": 2.2183094024658203,
      "learning_rate": 1.1283773711291696e-06,
      "loss": 0.0356,
      "step": 15184
    },
    {
      "epoch": 0.7841466563387555,
      "grad_norm": 9.697376251220703,
      "learning_rate": 1.1278589409537377e-06,
      "loss": 0.0863,
      "step": 15185
    },
    {
      "epoch": 0.7841982958946553,
      "grad_norm": 12.807036399841309,
      "learning_rate": 1.1273406147610756e-06,
      "loss": 0.0944,
      "step": 15186
    },
    {
      "epoch": 0.7842499354505551,
      "grad_norm": 2.7862133979797363,
      "learning_rate": 1.1268223925651017e-06,
      "loss": 0.0371,
      "step": 15187
    },
    {
      "epoch": 0.7843015750064549,
      "grad_norm": 2.2868216037750244,
      "learning_rate": 1.1263042743797348e-06,
      "loss": 0.0407,
      "step": 15188
    },
    {
      "epoch": 0.7843532145623547,
      "grad_norm": 2.448176145553589,
      "learning_rate": 1.1257862602188829e-06,
      "loss": 0.0316,
      "step": 15189
    },
    {
      "epoch": 0.7844048541182546,
      "grad_norm": 3.207615852355957,
      "learning_rate": 1.1252683500964622e-06,
      "loss": 0.0339,
      "step": 15190
    },
    {
      "epoch": 0.7844564936741544,
      "grad_norm": 9.259095191955566,
      "learning_rate": 1.124750544026379e-06,
      "loss": 0.1148,
      "step": 15191
    },
    {
      "epoch": 0.7845081332300542,
      "grad_norm": 4.27778959274292,
      "learning_rate": 1.124232842022539e-06,
      "loss": 0.0545,
      "step": 15192
    },
    {
      "epoch": 0.784559772785954,
      "grad_norm": 3.8326494693756104,
      "learning_rate": 1.123715244098843e-06,
      "loss": 0.0647,
      "step": 15193
    },
    {
      "epoch": 0.7846114123418538,
      "grad_norm": 9.264166831970215,
      "learning_rate": 1.1231977502691904e-06,
      "loss": 0.2237,
      "step": 15194
    },
    {
      "epoch": 0.7846630518977536,
      "grad_norm": 2.8365108966827393,
      "learning_rate": 1.1226803605474818e-06,
      "loss": 0.0371,
      "step": 15195
    },
    {
      "epoch": 0.7847146914536535,
      "grad_norm": 8.557668685913086,
      "learning_rate": 1.122163074947607e-06,
      "loss": 0.1012,
      "step": 15196
    },
    {
      "epoch": 0.7847663310095533,
      "grad_norm": 1.4543880224227905,
      "learning_rate": 1.121645893483459e-06,
      "loss": 0.0105,
      "step": 15197
    },
    {
      "epoch": 0.7848179705654531,
      "grad_norm": 2.9978578090667725,
      "learning_rate": 1.121128816168926e-06,
      "loss": 0.0426,
      "step": 15198
    },
    {
      "epoch": 0.7848696101213529,
      "grad_norm": 4.35147762298584,
      "learning_rate": 1.1206118430178936e-06,
      "loss": 0.0759,
      "step": 15199
    },
    {
      "epoch": 0.7849212496772527,
      "grad_norm": 3.7982120513916016,
      "learning_rate": 1.1200949740442446e-06,
      "loss": 0.0566,
      "step": 15200
    },
    {
      "epoch": 0.7849728892331526,
      "grad_norm": 3.956014633178711,
      "learning_rate": 1.1195782092618595e-06,
      "loss": 0.0859,
      "step": 15201
    },
    {
      "epoch": 0.7850245287890524,
      "grad_norm": 3.2850990295410156,
      "learning_rate": 1.119061548684615e-06,
      "loss": 0.025,
      "step": 15202
    },
    {
      "epoch": 0.7850761683449522,
      "grad_norm": 17.01490592956543,
      "learning_rate": 1.1185449923263858e-06,
      "loss": 0.0859,
      "step": 15203
    },
    {
      "epoch": 0.785127807900852,
      "grad_norm": 2.1716196537017822,
      "learning_rate": 1.1180285402010438e-06,
      "loss": 0.0385,
      "step": 15204
    },
    {
      "epoch": 0.7851794474567518,
      "grad_norm": 17.544593811035156,
      "learning_rate": 1.1175121923224575e-06,
      "loss": 0.0837,
      "step": 15205
    },
    {
      "epoch": 0.7852310870126517,
      "grad_norm": 6.347554683685303,
      "learning_rate": 1.116995948704494e-06,
      "loss": 0.0485,
      "step": 15206
    },
    {
      "epoch": 0.7852827265685515,
      "grad_norm": 2.5098986625671387,
      "learning_rate": 1.116479809361014e-06,
      "loss": 0.0838,
      "step": 15207
    },
    {
      "epoch": 0.7853343661244513,
      "grad_norm": 3.6050047874450684,
      "learning_rate": 1.1159637743058783e-06,
      "loss": 0.0456,
      "step": 15208
    },
    {
      "epoch": 0.7853860056803511,
      "grad_norm": 2.9448742866516113,
      "learning_rate": 1.1154478435529475e-06,
      "loss": 0.0476,
      "step": 15209
    },
    {
      "epoch": 0.7854376452362509,
      "grad_norm": 2.1376731395721436,
      "learning_rate": 1.1149320171160754e-06,
      "loss": 0.0248,
      "step": 15210
    },
    {
      "epoch": 0.7854892847921507,
      "grad_norm": 4.240235805511475,
      "learning_rate": 1.1144162950091126e-06,
      "loss": 0.0568,
      "step": 15211
    },
    {
      "epoch": 0.7855409243480506,
      "grad_norm": 6.294912815093994,
      "learning_rate": 1.1139006772459081e-06,
      "loss": 0.1049,
      "step": 15212
    },
    {
      "epoch": 0.7855925639039504,
      "grad_norm": 7.281682014465332,
      "learning_rate": 1.1133851638403086e-06,
      "loss": 0.0843,
      "step": 15213
    },
    {
      "epoch": 0.7856442034598502,
      "grad_norm": 2.2111806869506836,
      "learning_rate": 1.1128697548061612e-06,
      "loss": 0.0353,
      "step": 15214
    },
    {
      "epoch": 0.78569584301575,
      "grad_norm": 7.041902542114258,
      "learning_rate": 1.1123544501573026e-06,
      "loss": 0.0667,
      "step": 15215
    },
    {
      "epoch": 0.7857474825716498,
      "grad_norm": 3.530785322189331,
      "learning_rate": 1.111839249907572e-06,
      "loss": 0.0512,
      "step": 15216
    },
    {
      "epoch": 0.7857991221275497,
      "grad_norm": 2.803149938583374,
      "learning_rate": 1.1113241540708048e-06,
      "loss": 0.0314,
      "step": 15217
    },
    {
      "epoch": 0.7858507616834495,
      "grad_norm": 1.169180154800415,
      "learning_rate": 1.110809162660833e-06,
      "loss": 0.0153,
      "step": 15218
    },
    {
      "epoch": 0.7859024012393493,
      "grad_norm": 1.939335584640503,
      "learning_rate": 1.1102942756914875e-06,
      "loss": 0.0318,
      "step": 15219
    },
    {
      "epoch": 0.7859540407952491,
      "grad_norm": 17.850116729736328,
      "learning_rate": 1.1097794931765937e-06,
      "loss": 0.0928,
      "step": 15220
    },
    {
      "epoch": 0.7860056803511489,
      "grad_norm": 6.321599960327148,
      "learning_rate": 1.1092648151299773e-06,
      "loss": 0.0901,
      "step": 15221
    },
    {
      "epoch": 0.7860573199070487,
      "grad_norm": 2.9577765464782715,
      "learning_rate": 1.108750241565456e-06,
      "loss": 0.0506,
      "step": 15222
    },
    {
      "epoch": 0.7861089594629486,
      "grad_norm": 4.5982279777526855,
      "learning_rate": 1.1082357724968524e-06,
      "loss": 0.0366,
      "step": 15223
    },
    {
      "epoch": 0.7861605990188484,
      "grad_norm": 8.907614707946777,
      "learning_rate": 1.1077214079379793e-06,
      "loss": 0.0704,
      "step": 15224
    },
    {
      "epoch": 0.7862122385747483,
      "grad_norm": 2.022648334503174,
      "learning_rate": 1.1072071479026525e-06,
      "loss": 0.0431,
      "step": 15225
    },
    {
      "epoch": 0.7862638781306481,
      "grad_norm": 2.582672595977783,
      "learning_rate": 1.1066929924046788e-06,
      "loss": 0.0396,
      "step": 15226
    },
    {
      "epoch": 0.786315517686548,
      "grad_norm": 3.719309091567993,
      "learning_rate": 1.1061789414578643e-06,
      "loss": 0.0563,
      "step": 15227
    },
    {
      "epoch": 0.7863671572424478,
      "grad_norm": 1.9296257495880127,
      "learning_rate": 1.105664995076019e-06,
      "loss": 0.0266,
      "step": 15228
    },
    {
      "epoch": 0.7864187967983476,
      "grad_norm": 3.1748180389404297,
      "learning_rate": 1.1051511532729392e-06,
      "loss": 0.0503,
      "step": 15229
    },
    {
      "epoch": 0.7864704363542474,
      "grad_norm": 2.8915677070617676,
      "learning_rate": 1.1046374160624257e-06,
      "loss": 0.0389,
      "step": 15230
    },
    {
      "epoch": 0.7865220759101472,
      "grad_norm": 1.2516701221466064,
      "learning_rate": 1.1041237834582742e-06,
      "loss": 0.0212,
      "step": 15231
    },
    {
      "epoch": 0.786573715466047,
      "grad_norm": 4.687373638153076,
      "learning_rate": 1.1036102554742772e-06,
      "loss": 0.0407,
      "step": 15232
    },
    {
      "epoch": 0.7866253550219469,
      "grad_norm": 2.5586113929748535,
      "learning_rate": 1.1030968321242263e-06,
      "loss": 0.0396,
      "step": 15233
    },
    {
      "epoch": 0.7866769945778467,
      "grad_norm": 2.0956692695617676,
      "learning_rate": 1.1025835134219082e-06,
      "loss": 0.0286,
      "step": 15234
    },
    {
      "epoch": 0.7867286341337465,
      "grad_norm": 1.0024985074996948,
      "learning_rate": 1.1020702993811078e-06,
      "loss": 0.0133,
      "step": 15235
    },
    {
      "epoch": 0.7867802736896463,
      "grad_norm": 3.1602518558502197,
      "learning_rate": 1.101557190015607e-06,
      "loss": 0.0324,
      "step": 15236
    },
    {
      "epoch": 0.7868319132455461,
      "grad_norm": 1.914127230644226,
      "learning_rate": 1.101044185339185e-06,
      "loss": 0.0269,
      "step": 15237
    },
    {
      "epoch": 0.786883552801446,
      "grad_norm": 9.836623191833496,
      "learning_rate": 1.1005312853656175e-06,
      "loss": 0.121,
      "step": 15238
    },
    {
      "epoch": 0.7869351923573458,
      "grad_norm": 1.8795405626296997,
      "learning_rate": 1.1000184901086796e-06,
      "loss": 0.0297,
      "step": 15239
    },
    {
      "epoch": 0.7869868319132456,
      "grad_norm": 5.127349853515625,
      "learning_rate": 1.0995057995821395e-06,
      "loss": 0.0489,
      "step": 15240
    },
    {
      "epoch": 0.7870384714691454,
      "grad_norm": 15.856694221496582,
      "learning_rate": 1.098993213799765e-06,
      "loss": 0.0891,
      "step": 15241
    },
    {
      "epoch": 0.7870901110250452,
      "grad_norm": 2.9542558193206787,
      "learning_rate": 1.098480732775324e-06,
      "loss": 0.0591,
      "step": 15242
    },
    {
      "epoch": 0.787141750580945,
      "grad_norm": 3.758300542831421,
      "learning_rate": 1.097968356522578e-06,
      "loss": 0.062,
      "step": 15243
    },
    {
      "epoch": 0.7871933901368449,
      "grad_norm": 2.751751184463501,
      "learning_rate": 1.097456085055284e-06,
      "loss": 0.0464,
      "step": 15244
    },
    {
      "epoch": 0.7872450296927447,
      "grad_norm": 2.012456178665161,
      "learning_rate": 1.0969439183872e-06,
      "loss": 0.0331,
      "step": 15245
    },
    {
      "epoch": 0.7872966692486445,
      "grad_norm": 1.9116127490997314,
      "learning_rate": 1.0964318565320787e-06,
      "loss": 0.0324,
      "step": 15246
    },
    {
      "epoch": 0.7873483088045443,
      "grad_norm": 5.797489166259766,
      "learning_rate": 1.0959198995036746e-06,
      "loss": 0.0931,
      "step": 15247
    },
    {
      "epoch": 0.7873999483604441,
      "grad_norm": 4.1428070068359375,
      "learning_rate": 1.095408047315732e-06,
      "loss": 0.0775,
      "step": 15248
    },
    {
      "epoch": 0.787451587916344,
      "grad_norm": 14.338287353515625,
      "learning_rate": 1.0948962999819974e-06,
      "loss": 0.0907,
      "step": 15249
    },
    {
      "epoch": 0.7875032274722438,
      "grad_norm": 11.118204116821289,
      "learning_rate": 1.0943846575162136e-06,
      "loss": 0.0832,
      "step": 15250
    },
    {
      "epoch": 0.7875548670281436,
      "grad_norm": 3.1113944053649902,
      "learning_rate": 1.0938731199321202e-06,
      "loss": 0.0433,
      "step": 15251
    },
    {
      "epoch": 0.7876065065840434,
      "grad_norm": 20.801544189453125,
      "learning_rate": 1.0933616872434544e-06,
      "loss": 0.083,
      "step": 15252
    },
    {
      "epoch": 0.7876581461399432,
      "grad_norm": 3.615396738052368,
      "learning_rate": 1.0928503594639496e-06,
      "loss": 0.0436,
      "step": 15253
    },
    {
      "epoch": 0.787709785695843,
      "grad_norm": 3.6714911460876465,
      "learning_rate": 1.0923391366073383e-06,
      "loss": 0.072,
      "step": 15254
    },
    {
      "epoch": 0.7877614252517429,
      "grad_norm": 6.881956577301025,
      "learning_rate": 1.0918280186873453e-06,
      "loss": 0.0478,
      "step": 15255
    },
    {
      "epoch": 0.7878130648076427,
      "grad_norm": 5.079920291900635,
      "learning_rate": 1.0913170057177003e-06,
      "loss": 0.03,
      "step": 15256
    },
    {
      "epoch": 0.7878647043635425,
      "grad_norm": 3.058075189590454,
      "learning_rate": 1.090806097712125e-06,
      "loss": 0.0472,
      "step": 15257
    },
    {
      "epoch": 0.7879163439194423,
      "grad_norm": 1.6236517429351807,
      "learning_rate": 1.0902952946843398e-06,
      "loss": 0.0192,
      "step": 15258
    },
    {
      "epoch": 0.7879679834753421,
      "grad_norm": 14.099394798278809,
      "learning_rate": 1.0897845966480598e-06,
      "loss": 0.1398,
      "step": 15259
    },
    {
      "epoch": 0.788019623031242,
      "grad_norm": 4.808391571044922,
      "learning_rate": 1.089274003616999e-06,
      "loss": 0.0355,
      "step": 15260
    },
    {
      "epoch": 0.7880712625871418,
      "grad_norm": 9.075652122497559,
      "learning_rate": 1.0887635156048736e-06,
      "loss": 0.087,
      "step": 15261
    },
    {
      "epoch": 0.7881229021430416,
      "grad_norm": 7.629123687744141,
      "learning_rate": 1.0882531326253876e-06,
      "loss": 0.0522,
      "step": 15262
    },
    {
      "epoch": 0.7881745416989414,
      "grad_norm": 8.68515682220459,
      "learning_rate": 1.0877428546922487e-06,
      "loss": 0.0888,
      "step": 15263
    },
    {
      "epoch": 0.7882261812548412,
      "grad_norm": 3.414952039718628,
      "learning_rate": 1.0872326818191592e-06,
      "loss": 0.067,
      "step": 15264
    },
    {
      "epoch": 0.788277820810741,
      "grad_norm": 4.9584245681762695,
      "learning_rate": 1.0867226140198195e-06,
      "loss": 0.0449,
      "step": 15265
    },
    {
      "epoch": 0.7883294603666409,
      "grad_norm": 4.103761672973633,
      "learning_rate": 1.086212651307928e-06,
      "loss": 0.0791,
      "step": 15266
    },
    {
      "epoch": 0.7883810999225407,
      "grad_norm": 3.8190531730651855,
      "learning_rate": 1.0857027936971775e-06,
      "loss": 0.0656,
      "step": 15267
    },
    {
      "epoch": 0.7884327394784405,
      "grad_norm": 12.950230598449707,
      "learning_rate": 1.0851930412012619e-06,
      "loss": 0.1161,
      "step": 15268
    },
    {
      "epoch": 0.7884843790343403,
      "grad_norm": 8.027924537658691,
      "learning_rate": 1.0846833938338685e-06,
      "loss": 0.0645,
      "step": 15269
    },
    {
      "epoch": 0.7885360185902401,
      "grad_norm": 2.032034158706665,
      "learning_rate": 1.0841738516086836e-06,
      "loss": 0.0263,
      "step": 15270
    },
    {
      "epoch": 0.78858765814614,
      "grad_norm": 3.031475067138672,
      "learning_rate": 1.083664414539391e-06,
      "loss": 0.0655,
      "step": 15271
    },
    {
      "epoch": 0.7886392977020398,
      "grad_norm": 7.902064800262451,
      "learning_rate": 1.0831550826396725e-06,
      "loss": 0.0801,
      "step": 15272
    },
    {
      "epoch": 0.7886909372579396,
      "grad_norm": 1.3119022846221924,
      "learning_rate": 1.0826458559232028e-06,
      "loss": 0.0111,
      "step": 15273
    },
    {
      "epoch": 0.7887425768138394,
      "grad_norm": 4.954398155212402,
      "learning_rate": 1.082136734403657e-06,
      "loss": 0.0265,
      "step": 15274
    },
    {
      "epoch": 0.7887942163697392,
      "grad_norm": 2.112009286880493,
      "learning_rate": 1.0816277180947094e-06,
      "loss": 0.0258,
      "step": 15275
    },
    {
      "epoch": 0.788845855925639,
      "grad_norm": 5.289237976074219,
      "learning_rate": 1.0811188070100293e-06,
      "loss": 0.0584,
      "step": 15276
    },
    {
      "epoch": 0.7888974954815389,
      "grad_norm": 4.349124431610107,
      "learning_rate": 1.0806100011632803e-06,
      "loss": 0.0291,
      "step": 15277
    },
    {
      "epoch": 0.7889491350374387,
      "grad_norm": 16.603435516357422,
      "learning_rate": 1.0801013005681271e-06,
      "loss": 0.0813,
      "step": 15278
    },
    {
      "epoch": 0.7890007745933385,
      "grad_norm": 2.573814630508423,
      "learning_rate": 1.0795927052382294e-06,
      "loss": 0.0306,
      "step": 15279
    },
    {
      "epoch": 0.7890524141492383,
      "grad_norm": 6.273271560668945,
      "learning_rate": 1.0790842151872493e-06,
      "loss": 0.111,
      "step": 15280
    },
    {
      "epoch": 0.7891040537051381,
      "grad_norm": 10.348172187805176,
      "learning_rate": 1.0785758304288368e-06,
      "loss": 0.1597,
      "step": 15281
    },
    {
      "epoch": 0.789155693261038,
      "grad_norm": 1.9343791007995605,
      "learning_rate": 1.078067550976647e-06,
      "loss": 0.0189,
      "step": 15282
    },
    {
      "epoch": 0.7892073328169378,
      "grad_norm": 2.529592275619507,
      "learning_rate": 1.0775593768443276e-06,
      "loss": 0.0419,
      "step": 15283
    },
    {
      "epoch": 0.7892589723728376,
      "grad_norm": 1.8062320947647095,
      "learning_rate": 1.0770513080455263e-06,
      "loss": 0.0352,
      "step": 15284
    },
    {
      "epoch": 0.7893106119287374,
      "grad_norm": 5.278587818145752,
      "learning_rate": 1.076543344593886e-06,
      "loss": 0.0695,
      "step": 15285
    },
    {
      "epoch": 0.7893622514846372,
      "grad_norm": 3.189910888671875,
      "learning_rate": 1.0760354865030488e-06,
      "loss": 0.0406,
      "step": 15286
    },
    {
      "epoch": 0.789413891040537,
      "grad_norm": 3.392775058746338,
      "learning_rate": 1.0755277337866527e-06,
      "loss": 0.0326,
      "step": 15287
    },
    {
      "epoch": 0.7894655305964369,
      "grad_norm": 1.4592033624649048,
      "learning_rate": 1.0750200864583294e-06,
      "loss": 0.0183,
      "step": 15288
    },
    {
      "epoch": 0.7895171701523367,
      "grad_norm": 3.098013162612915,
      "learning_rate": 1.0745125445317156e-06,
      "loss": 0.0328,
      "step": 15289
    },
    {
      "epoch": 0.7895688097082365,
      "grad_norm": 13.997801780700684,
      "learning_rate": 1.07400510802044e-06,
      "loss": 0.0654,
      "step": 15290
    },
    {
      "epoch": 0.7896204492641363,
      "grad_norm": 3.233154296875,
      "learning_rate": 1.0734977769381294e-06,
      "loss": 0.0566,
      "step": 15291
    },
    {
      "epoch": 0.7896720888200361,
      "grad_norm": 3.525372266769409,
      "learning_rate": 1.0729905512984058e-06,
      "loss": 0.0444,
      "step": 15292
    },
    {
      "epoch": 0.789723728375936,
      "grad_norm": 2.1851742267608643,
      "learning_rate": 1.07248343111489e-06,
      "loss": 0.0203,
      "step": 15293
    },
    {
      "epoch": 0.7897753679318358,
      "grad_norm": 4.207377910614014,
      "learning_rate": 1.0719764164012048e-06,
      "loss": 0.0791,
      "step": 15294
    },
    {
      "epoch": 0.7898270074877356,
      "grad_norm": 13.91635513305664,
      "learning_rate": 1.0714695071709614e-06,
      "loss": 0.1294,
      "step": 15295
    },
    {
      "epoch": 0.7898786470436354,
      "grad_norm": 3.8872179985046387,
      "learning_rate": 1.0709627034377728e-06,
      "loss": 0.0266,
      "step": 15296
    },
    {
      "epoch": 0.7899302865995352,
      "grad_norm": 12.603775978088379,
      "learning_rate": 1.0704560052152503e-06,
      "loss": 0.1992,
      "step": 15297
    },
    {
      "epoch": 0.789981926155435,
      "grad_norm": 4.0267486572265625,
      "learning_rate": 1.069949412516999e-06,
      "loss": 0.0632,
      "step": 15298
    },
    {
      "epoch": 0.7900335657113349,
      "grad_norm": 3.186290979385376,
      "learning_rate": 1.0694429253566246e-06,
      "loss": 0.0611,
      "step": 15299
    },
    {
      "epoch": 0.7900852052672347,
      "grad_norm": 16.223848342895508,
      "learning_rate": 1.0689365437477278e-06,
      "loss": 0.0605,
      "step": 15300
    },
    {
      "epoch": 0.7901368448231345,
      "grad_norm": 3.9295129776000977,
      "learning_rate": 1.0684302677039066e-06,
      "loss": 0.086,
      "step": 15301
    },
    {
      "epoch": 0.7901884843790343,
      "grad_norm": 20.692625045776367,
      "learning_rate": 1.0679240972387572e-06,
      "loss": 0.0447,
      "step": 15302
    },
    {
      "epoch": 0.7902401239349341,
      "grad_norm": 4.10030460357666,
      "learning_rate": 1.0674180323658722e-06,
      "loss": 0.052,
      "step": 15303
    },
    {
      "epoch": 0.790291763490834,
      "grad_norm": 14.37365436553955,
      "learning_rate": 1.0669120730988409e-06,
      "loss": 0.1733,
      "step": 15304
    },
    {
      "epoch": 0.7903434030467338,
      "grad_norm": 4.335737705230713,
      "learning_rate": 1.0664062194512526e-06,
      "loss": 0.051,
      "step": 15305
    },
    {
      "epoch": 0.7903950426026336,
      "grad_norm": 7.918474197387695,
      "learning_rate": 1.0659004714366878e-06,
      "loss": 0.1387,
      "step": 15306
    },
    {
      "epoch": 0.7904466821585334,
      "grad_norm": 4.102680683135986,
      "learning_rate": 1.0653948290687289e-06,
      "loss": 0.037,
      "step": 15307
    },
    {
      "epoch": 0.7904983217144332,
      "grad_norm": 7.65904426574707,
      "learning_rate": 1.0648892923609566e-06,
      "loss": 0.1192,
      "step": 15308
    },
    {
      "epoch": 0.7905499612703331,
      "grad_norm": 7.752985954284668,
      "learning_rate": 1.064383861326947e-06,
      "loss": 0.0852,
      "step": 15309
    },
    {
      "epoch": 0.7906016008262329,
      "grad_norm": 15.91753101348877,
      "learning_rate": 1.06387853598027e-06,
      "loss": 0.1341,
      "step": 15310
    },
    {
      "epoch": 0.7906532403821327,
      "grad_norm": 5.2821173667907715,
      "learning_rate": 1.0633733163344972e-06,
      "loss": 0.0794,
      "step": 15311
    },
    {
      "epoch": 0.7907048799380325,
      "grad_norm": 1.7517694234848022,
      "learning_rate": 1.062868202403194e-06,
      "loss": 0.0353,
      "step": 15312
    },
    {
      "epoch": 0.7907565194939323,
      "grad_norm": 1.7566406726837158,
      "learning_rate": 1.062363194199929e-06,
      "loss": 0.0292,
      "step": 15313
    },
    {
      "epoch": 0.7908081590498321,
      "grad_norm": 7.199812412261963,
      "learning_rate": 1.0618582917382603e-06,
      "loss": 0.0893,
      "step": 15314
    },
    {
      "epoch": 0.790859798605732,
      "grad_norm": 5.176259517669678,
      "learning_rate": 1.0613534950317473e-06,
      "loss": 0.0397,
      "step": 15315
    },
    {
      "epoch": 0.7909114381616318,
      "grad_norm": 1.9046905040740967,
      "learning_rate": 1.0608488040939458e-06,
      "loss": 0.0361,
      "step": 15316
    },
    {
      "epoch": 0.7909630777175316,
      "grad_norm": 1.7416565418243408,
      "learning_rate": 1.0603442189384089e-06,
      "loss": 0.0224,
      "step": 15317
    },
    {
      "epoch": 0.7910147172734314,
      "grad_norm": 4.359591007232666,
      "learning_rate": 1.0598397395786868e-06,
      "loss": 0.0762,
      "step": 15318
    },
    {
      "epoch": 0.7910663568293312,
      "grad_norm": 6.616849899291992,
      "learning_rate": 1.0593353660283273e-06,
      "loss": 0.0325,
      "step": 15319
    },
    {
      "epoch": 0.7911179963852311,
      "grad_norm": 2.677042245864868,
      "learning_rate": 1.058831098300876e-06,
      "loss": 0.0517,
      "step": 15320
    },
    {
      "epoch": 0.7911696359411309,
      "grad_norm": 1.8962076902389526,
      "learning_rate": 1.0583269364098697e-06,
      "loss": 0.0358,
      "step": 15321
    },
    {
      "epoch": 0.7912212754970307,
      "grad_norm": 3.4363696575164795,
      "learning_rate": 1.0578228803688522e-06,
      "loss": 0.0552,
      "step": 15322
    },
    {
      "epoch": 0.7912729150529305,
      "grad_norm": 19.72507667541504,
      "learning_rate": 1.0573189301913577e-06,
      "loss": 0.1708,
      "step": 15323
    },
    {
      "epoch": 0.7913245546088303,
      "grad_norm": 10.395158767700195,
      "learning_rate": 1.0568150858909209e-06,
      "loss": 0.0595,
      "step": 15324
    },
    {
      "epoch": 0.7913761941647302,
      "grad_norm": 2.4393787384033203,
      "learning_rate": 1.0563113474810687e-06,
      "loss": 0.036,
      "step": 15325
    },
    {
      "epoch": 0.79142783372063,
      "grad_norm": 4.970415115356445,
      "learning_rate": 1.0558077149753298e-06,
      "loss": 0.0938,
      "step": 15326
    },
    {
      "epoch": 0.7914794732765298,
      "grad_norm": 2.9838807582855225,
      "learning_rate": 1.0553041883872317e-06,
      "loss": 0.0285,
      "step": 15327
    },
    {
      "epoch": 0.7915311128324296,
      "grad_norm": 3.4339661598205566,
      "learning_rate": 1.054800767730293e-06,
      "loss": 0.0562,
      "step": 15328
    },
    {
      "epoch": 0.7915827523883294,
      "grad_norm": 6.4172821044921875,
      "learning_rate": 1.0542974530180327e-06,
      "loss": 0.044,
      "step": 15329
    },
    {
      "epoch": 0.7916343919442292,
      "grad_norm": 0.7949888110160828,
      "learning_rate": 1.0537942442639675e-06,
      "loss": 0.0086,
      "step": 15330
    },
    {
      "epoch": 0.7916860315001291,
      "grad_norm": 1.3025920391082764,
      "learning_rate": 1.0532911414816116e-06,
      "loss": 0.0211,
      "step": 15331
    },
    {
      "epoch": 0.7917376710560289,
      "grad_norm": 3.157057523727417,
      "learning_rate": 1.052788144684474e-06,
      "loss": 0.0799,
      "step": 15332
    },
    {
      "epoch": 0.7917893106119287,
      "grad_norm": 4.8417181968688965,
      "learning_rate": 1.0522852538860624e-06,
      "loss": 0.0953,
      "step": 15333
    },
    {
      "epoch": 0.7918409501678285,
      "grad_norm": 4.494026184082031,
      "learning_rate": 1.0517824690998818e-06,
      "loss": 0.0598,
      "step": 15334
    },
    {
      "epoch": 0.7918925897237283,
      "grad_norm": 2.9654664993286133,
      "learning_rate": 1.051279790339434e-06,
      "loss": 0.0478,
      "step": 15335
    },
    {
      "epoch": 0.7919442292796282,
      "grad_norm": 2.6400959491729736,
      "learning_rate": 1.050777217618218e-06,
      "loss": 0.0412,
      "step": 15336
    },
    {
      "epoch": 0.791995868835528,
      "grad_norm": 3.001063585281372,
      "learning_rate": 1.0502747509497296e-06,
      "loss": 0.0299,
      "step": 15337
    },
    {
      "epoch": 0.7920475083914278,
      "grad_norm": 4.445459365844727,
      "learning_rate": 1.0497723903474645e-06,
      "loss": 0.0441,
      "step": 15338
    },
    {
      "epoch": 0.7920991479473276,
      "grad_norm": 2.9963204860687256,
      "learning_rate": 1.0492701358249085e-06,
      "loss": 0.0372,
      "step": 15339
    },
    {
      "epoch": 0.7921507875032274,
      "grad_norm": 5.441681861877441,
      "learning_rate": 1.0487679873955509e-06,
      "loss": 0.1028,
      "step": 15340
    },
    {
      "epoch": 0.7922024270591272,
      "grad_norm": 7.901764392852783,
      "learning_rate": 1.0482659450728783e-06,
      "loss": 0.0418,
      "step": 15341
    },
    {
      "epoch": 0.7922540666150271,
      "grad_norm": 0.8833502531051636,
      "learning_rate": 1.0477640088703728e-06,
      "loss": 0.0113,
      "step": 15342
    },
    {
      "epoch": 0.7923057061709269,
      "grad_norm": 3.7363297939300537,
      "learning_rate": 1.0472621788015108e-06,
      "loss": 0.053,
      "step": 15343
    },
    {
      "epoch": 0.7923573457268267,
      "grad_norm": 2.721505641937256,
      "learning_rate": 1.04676045487977e-06,
      "loss": 0.0216,
      "step": 15344
    },
    {
      "epoch": 0.7924089852827265,
      "grad_norm": 4.688263893127441,
      "learning_rate": 1.0462588371186222e-06,
      "loss": 0.0367,
      "step": 15345
    },
    {
      "epoch": 0.7924606248386263,
      "grad_norm": 1.4688981771469116,
      "learning_rate": 1.0457573255315412e-06,
      "loss": 0.0207,
      "step": 15346
    },
    {
      "epoch": 0.7925122643945262,
      "grad_norm": 2.195164918899536,
      "learning_rate": 1.0452559201319918e-06,
      "loss": 0.0427,
      "step": 15347
    },
    {
      "epoch": 0.792563903950426,
      "grad_norm": 9.302826881408691,
      "learning_rate": 1.0447546209334397e-06,
      "loss": 0.0211,
      "step": 15348
    },
    {
      "epoch": 0.7926155435063258,
      "grad_norm": 9.466560363769531,
      "learning_rate": 1.0442534279493465e-06,
      "loss": 0.0999,
      "step": 15349
    },
    {
      "epoch": 0.7926671830622256,
      "grad_norm": 5.6552205085754395,
      "learning_rate": 1.0437523411931717e-06,
      "loss": 0.0858,
      "step": 15350
    },
    {
      "epoch": 0.7927188226181254,
      "grad_norm": 10.745430946350098,
      "learning_rate": 1.0432513606783707e-06,
      "loss": 0.1601,
      "step": 15351
    },
    {
      "epoch": 0.7927704621740254,
      "grad_norm": 6.297542095184326,
      "learning_rate": 1.042750486418398e-06,
      "loss": 0.1674,
      "step": 15352
    },
    {
      "epoch": 0.7928221017299252,
      "grad_norm": 1.9749022722244263,
      "learning_rate": 1.042249718426705e-06,
      "loss": 0.0258,
      "step": 15353
    },
    {
      "epoch": 0.792873741285825,
      "grad_norm": 1.6551308631896973,
      "learning_rate": 1.041749056716735e-06,
      "loss": 0.0113,
      "step": 15354
    },
    {
      "epoch": 0.7929253808417248,
      "grad_norm": 9.154178619384766,
      "learning_rate": 1.041248501301937e-06,
      "loss": 0.0745,
      "step": 15355
    },
    {
      "epoch": 0.7929770203976246,
      "grad_norm": 2.2272818088531494,
      "learning_rate": 1.040748052195752e-06,
      "loss": 0.0276,
      "step": 15356
    },
    {
      "epoch": 0.7930286599535245,
      "grad_norm": 1.846588134765625,
      "learning_rate": 1.0402477094116203e-06,
      "loss": 0.0378,
      "step": 15357
    },
    {
      "epoch": 0.7930802995094243,
      "grad_norm": 17.60834312438965,
      "learning_rate": 1.0397474729629748e-06,
      "loss": 0.126,
      "step": 15358
    },
    {
      "epoch": 0.7931319390653241,
      "grad_norm": 2.7795233726501465,
      "learning_rate": 1.03924734286325e-06,
      "loss": 0.0265,
      "step": 15359
    },
    {
      "epoch": 0.7931835786212239,
      "grad_norm": 5.009191036224365,
      "learning_rate": 1.0387473191258796e-06,
      "loss": 0.0373,
      "step": 15360
    },
    {
      "epoch": 0.7932352181771237,
      "grad_norm": 9.130350112915039,
      "learning_rate": 1.0382474017642873e-06,
      "loss": 0.0711,
      "step": 15361
    },
    {
      "epoch": 0.7932868577330235,
      "grad_norm": 2.9883744716644287,
      "learning_rate": 1.0377475907918999e-06,
      "loss": 0.039,
      "step": 15362
    },
    {
      "epoch": 0.7933384972889234,
      "grad_norm": 6.980311870574951,
      "learning_rate": 1.0372478862221385e-06,
      "loss": 0.073,
      "step": 15363
    },
    {
      "epoch": 0.7933901368448232,
      "grad_norm": 2.006678581237793,
      "learning_rate": 1.0367482880684227e-06,
      "loss": 0.0324,
      "step": 15364
    },
    {
      "epoch": 0.793441776400723,
      "grad_norm": 3.582136631011963,
      "learning_rate": 1.0362487963441693e-06,
      "loss": 0.0531,
      "step": 15365
    },
    {
      "epoch": 0.7934934159566228,
      "grad_norm": 10.698420524597168,
      "learning_rate": 1.0357494110627902e-06,
      "loss": 0.1037,
      "step": 15366
    },
    {
      "epoch": 0.7935450555125226,
      "grad_norm": 1.8678350448608398,
      "learning_rate": 1.0352501322376974e-06,
      "loss": 0.0501,
      "step": 15367
    },
    {
      "epoch": 0.7935966950684225,
      "grad_norm": 7.6296234130859375,
      "learning_rate": 1.0347509598822974e-06,
      "loss": 0.0312,
      "step": 15368
    },
    {
      "epoch": 0.7936483346243223,
      "grad_norm": 2.603091239929199,
      "learning_rate": 1.034251894009996e-06,
      "loss": 0.0287,
      "step": 15369
    },
    {
      "epoch": 0.7936999741802221,
      "grad_norm": 4.297965049743652,
      "learning_rate": 1.0337529346341946e-06,
      "loss": 0.0458,
      "step": 15370
    },
    {
      "epoch": 0.7937516137361219,
      "grad_norm": 3.5734071731567383,
      "learning_rate": 1.033254081768294e-06,
      "loss": 0.0199,
      "step": 15371
    },
    {
      "epoch": 0.7938032532920217,
      "grad_norm": 12.005335807800293,
      "learning_rate": 1.0327553354256876e-06,
      "loss": 0.0566,
      "step": 15372
    },
    {
      "epoch": 0.7938548928479215,
      "grad_norm": 15.618563652038574,
      "learning_rate": 1.0322566956197683e-06,
      "loss": 0.1206,
      "step": 15373
    },
    {
      "epoch": 0.7939065324038214,
      "grad_norm": 2.0122456550598145,
      "learning_rate": 1.03175816236393e-06,
      "loss": 0.0308,
      "step": 15374
    },
    {
      "epoch": 0.7939581719597212,
      "grad_norm": 4.608400344848633,
      "learning_rate": 1.0312597356715598e-06,
      "loss": 0.0624,
      "step": 15375
    },
    {
      "epoch": 0.794009811515621,
      "grad_norm": 14.17820930480957,
      "learning_rate": 1.0307614155560403e-06,
      "loss": 0.0836,
      "step": 15376
    },
    {
      "epoch": 0.7940614510715208,
      "grad_norm": 4.249979496002197,
      "learning_rate": 1.0302632020307545e-06,
      "loss": 0.0556,
      "step": 15377
    },
    {
      "epoch": 0.7941130906274206,
      "grad_norm": 5.477753639221191,
      "learning_rate": 1.0297650951090798e-06,
      "loss": 0.0323,
      "step": 15378
    },
    {
      "epoch": 0.7941647301833205,
      "grad_norm": 2.5465331077575684,
      "learning_rate": 1.029267094804397e-06,
      "loss": 0.0452,
      "step": 15379
    },
    {
      "epoch": 0.7942163697392203,
      "grad_norm": 2.6815149784088135,
      "learning_rate": 1.0287692011300754e-06,
      "loss": 0.0332,
      "step": 15380
    },
    {
      "epoch": 0.7942680092951201,
      "grad_norm": 23.903186798095703,
      "learning_rate": 1.028271414099486e-06,
      "loss": 0.0814,
      "step": 15381
    },
    {
      "epoch": 0.7943196488510199,
      "grad_norm": 8.61947250366211,
      "learning_rate": 1.0277737337259975e-06,
      "loss": 0.0405,
      "step": 15382
    },
    {
      "epoch": 0.7943712884069197,
      "grad_norm": 1.870385766029358,
      "learning_rate": 1.027276160022974e-06,
      "loss": 0.0193,
      "step": 15383
    },
    {
      "epoch": 0.7944229279628195,
      "grad_norm": 2.4802515506744385,
      "learning_rate": 1.0267786930037777e-06,
      "loss": 0.0197,
      "step": 15384
    },
    {
      "epoch": 0.7944745675187194,
      "grad_norm": 7.664467811584473,
      "learning_rate": 1.0262813326817671e-06,
      "loss": 0.0336,
      "step": 15385
    },
    {
      "epoch": 0.7945262070746192,
      "grad_norm": 3.8041985034942627,
      "learning_rate": 1.0257840790702988e-06,
      "loss": 0.0516,
      "step": 15386
    },
    {
      "epoch": 0.794577846630519,
      "grad_norm": 2.282025098800659,
      "learning_rate": 1.0252869321827264e-06,
      "loss": 0.0477,
      "step": 15387
    },
    {
      "epoch": 0.7946294861864188,
      "grad_norm": 4.206849098205566,
      "learning_rate": 1.0247898920323995e-06,
      "loss": 0.0399,
      "step": 15388
    },
    {
      "epoch": 0.7946811257423186,
      "grad_norm": 4.181581974029541,
      "learning_rate": 1.024292958632666e-06,
      "loss": 0.0758,
      "step": 15389
    },
    {
      "epoch": 0.7947327652982185,
      "grad_norm": 6.394500255584717,
      "learning_rate": 1.023796131996872e-06,
      "loss": 0.0928,
      "step": 15390
    },
    {
      "epoch": 0.7947844048541183,
      "grad_norm": 3.0659263134002686,
      "learning_rate": 1.0232994121383567e-06,
      "loss": 0.0601,
      "step": 15391
    },
    {
      "epoch": 0.7948360444100181,
      "grad_norm": 2.930995225906372,
      "learning_rate": 1.0228027990704592e-06,
      "loss": 0.0347,
      "step": 15392
    },
    {
      "epoch": 0.7948876839659179,
      "grad_norm": 1.758232831954956,
      "learning_rate": 1.0223062928065193e-06,
      "loss": 0.0382,
      "step": 15393
    },
    {
      "epoch": 0.7949393235218177,
      "grad_norm": 4.051522254943848,
      "learning_rate": 1.0218098933598659e-06,
      "loss": 0.0898,
      "step": 15394
    },
    {
      "epoch": 0.7949909630777175,
      "grad_norm": 2.8721771240234375,
      "learning_rate": 1.0213136007438312e-06,
      "loss": 0.0241,
      "step": 15395
    },
    {
      "epoch": 0.7950426026336174,
      "grad_norm": 3.455044746398926,
      "learning_rate": 1.0208174149717432e-06,
      "loss": 0.0537,
      "step": 15396
    },
    {
      "epoch": 0.7950942421895172,
      "grad_norm": 16.569290161132812,
      "learning_rate": 1.0203213360569254e-06,
      "loss": 0.1042,
      "step": 15397
    },
    {
      "epoch": 0.795145881745417,
      "grad_norm": 2.2140493392944336,
      "learning_rate": 1.0198253640126998e-06,
      "loss": 0.0308,
      "step": 15398
    },
    {
      "epoch": 0.7951975213013168,
      "grad_norm": 3.9361555576324463,
      "learning_rate": 1.0193294988523855e-06,
      "loss": 0.0838,
      "step": 15399
    },
    {
      "epoch": 0.7952491608572166,
      "grad_norm": 4.1004767417907715,
      "learning_rate": 1.0188337405892985e-06,
      "loss": 0.053,
      "step": 15400
    },
    {
      "epoch": 0.7953008004131165,
      "grad_norm": 3.9575910568237305,
      "learning_rate": 1.0183380892367523e-06,
      "loss": 0.0518,
      "step": 15401
    },
    {
      "epoch": 0.7953524399690163,
      "grad_norm": 2.60136342048645,
      "learning_rate": 1.0178425448080565e-06,
      "loss": 0.0364,
      "step": 15402
    },
    {
      "epoch": 0.7954040795249161,
      "grad_norm": 4.382409572601318,
      "learning_rate": 1.0173471073165187e-06,
      "loss": 0.0381,
      "step": 15403
    },
    {
      "epoch": 0.7954557190808159,
      "grad_norm": 2.830475330352783,
      "learning_rate": 1.0168517767754444e-06,
      "loss": 0.0275,
      "step": 15404
    },
    {
      "epoch": 0.7955073586367157,
      "grad_norm": 3.825233221054077,
      "learning_rate": 1.0163565531981335e-06,
      "loss": 0.0362,
      "step": 15405
    },
    {
      "epoch": 0.7955589981926156,
      "grad_norm": 3.2467844486236572,
      "learning_rate": 1.015861436597884e-06,
      "loss": 0.0693,
      "step": 15406
    },
    {
      "epoch": 0.7956106377485154,
      "grad_norm": 5.065593242645264,
      "learning_rate": 1.0153664269879947e-06,
      "loss": 0.0482,
      "step": 15407
    },
    {
      "epoch": 0.7956622773044152,
      "grad_norm": 7.909816741943359,
      "learning_rate": 1.014871524381759e-06,
      "loss": 0.0818,
      "step": 15408
    },
    {
      "epoch": 0.795713916860315,
      "grad_norm": 17.27851676940918,
      "learning_rate": 1.0143767287924637e-06,
      "loss": 0.0735,
      "step": 15409
    },
    {
      "epoch": 0.7957655564162148,
      "grad_norm": 3.1807453632354736,
      "learning_rate": 1.0138820402333982e-06,
      "loss": 0.057,
      "step": 15410
    },
    {
      "epoch": 0.7958171959721146,
      "grad_norm": 7.414652347564697,
      "learning_rate": 1.0133874587178445e-06,
      "loss": 0.0737,
      "step": 15411
    },
    {
      "epoch": 0.7958688355280145,
      "grad_norm": 4.0166473388671875,
      "learning_rate": 1.012892984259089e-06,
      "loss": 0.051,
      "step": 15412
    },
    {
      "epoch": 0.7959204750839143,
      "grad_norm": 2.3985369205474854,
      "learning_rate": 1.0123986168704059e-06,
      "loss": 0.0194,
      "step": 15413
    },
    {
      "epoch": 0.7959721146398141,
      "grad_norm": 2.5289392471313477,
      "learning_rate": 1.0119043565650728e-06,
      "loss": 0.0328,
      "step": 15414
    },
    {
      "epoch": 0.7960237541957139,
      "grad_norm": 3.012533187866211,
      "learning_rate": 1.0114102033563622e-06,
      "loss": 0.0522,
      "step": 15415
    },
    {
      "epoch": 0.7960753937516137,
      "grad_norm": 2.56374192237854,
      "learning_rate": 1.0109161572575444e-06,
      "loss": 0.0253,
      "step": 15416
    },
    {
      "epoch": 0.7961270333075136,
      "grad_norm": 5.006525039672852,
      "learning_rate": 1.0104222182818856e-06,
      "loss": 0.0678,
      "step": 15417
    },
    {
      "epoch": 0.7961786728634134,
      "grad_norm": 6.776859283447266,
      "learning_rate": 1.0099283864426517e-06,
      "loss": 0.0534,
      "step": 15418
    },
    {
      "epoch": 0.7962303124193132,
      "grad_norm": 3.758890390396118,
      "learning_rate": 1.0094346617531032e-06,
      "loss": 0.0571,
      "step": 15419
    },
    {
      "epoch": 0.796281951975213,
      "grad_norm": 1.924901008605957,
      "learning_rate": 1.0089410442264985e-06,
      "loss": 0.0376,
      "step": 15420
    },
    {
      "epoch": 0.7963335915311128,
      "grad_norm": 7.810619354248047,
      "learning_rate": 1.0084475338760929e-06,
      "loss": 0.1046,
      "step": 15421
    },
    {
      "epoch": 0.7963852310870126,
      "grad_norm": 17.261526107788086,
      "learning_rate": 1.0079541307151404e-06,
      "loss": 0.0623,
      "step": 15422
    },
    {
      "epoch": 0.7964368706429125,
      "grad_norm": 2.9066553115844727,
      "learning_rate": 1.0074608347568914e-06,
      "loss": 0.0423,
      "step": 15423
    },
    {
      "epoch": 0.7964885101988123,
      "grad_norm": 3.361544609069824,
      "learning_rate": 1.00696764601459e-06,
      "loss": 0.0251,
      "step": 15424
    },
    {
      "epoch": 0.7965401497547121,
      "grad_norm": 5.374652862548828,
      "learning_rate": 1.0064745645014813e-06,
      "loss": 0.0524,
      "step": 15425
    },
    {
      "epoch": 0.7965917893106119,
      "grad_norm": 8.371670722961426,
      "learning_rate": 1.005981590230809e-06,
      "loss": 0.0658,
      "step": 15426
    },
    {
      "epoch": 0.7966434288665117,
      "grad_norm": 3.8911778926849365,
      "learning_rate": 1.005488723215809e-06,
      "loss": 0.0445,
      "step": 15427
    },
    {
      "epoch": 0.7966950684224116,
      "grad_norm": 2.7060904502868652,
      "learning_rate": 1.0049959634697177e-06,
      "loss": 0.0369,
      "step": 15428
    },
    {
      "epoch": 0.7967467079783114,
      "grad_norm": 3.633434772491455,
      "learning_rate": 1.0045033110057667e-06,
      "loss": 0.0704,
      "step": 15429
    },
    {
      "epoch": 0.7967983475342112,
      "grad_norm": 2.4314987659454346,
      "learning_rate": 1.0040107658371873e-06,
      "loss": 0.0169,
      "step": 15430
    },
    {
      "epoch": 0.796849987090111,
      "grad_norm": 3.695744514465332,
      "learning_rate": 1.0035183279772053e-06,
      "loss": 0.0454,
      "step": 15431
    },
    {
      "epoch": 0.7969016266460108,
      "grad_norm": 2.922774314880371,
      "learning_rate": 1.0030259974390449e-06,
      "loss": 0.0645,
      "step": 15432
    },
    {
      "epoch": 0.7969532662019106,
      "grad_norm": 2.9933431148529053,
      "learning_rate": 1.0025337742359275e-06,
      "loss": 0.0638,
      "step": 15433
    },
    {
      "epoch": 0.7970049057578105,
      "grad_norm": 2.737128734588623,
      "learning_rate": 1.002041658381071e-06,
      "loss": 0.0317,
      "step": 15434
    },
    {
      "epoch": 0.7970565453137103,
      "grad_norm": 2.272977352142334,
      "learning_rate": 1.0015496498876904e-06,
      "loss": 0.0304,
      "step": 15435
    },
    {
      "epoch": 0.7971081848696101,
      "grad_norm": 2.563845157623291,
      "learning_rate": 1.0010577487689988e-06,
      "loss": 0.0346,
      "step": 15436
    },
    {
      "epoch": 0.7971598244255099,
      "grad_norm": 13.00734806060791,
      "learning_rate": 1.0005659550382068e-06,
      "loss": 0.0558,
      "step": 15437
    },
    {
      "epoch": 0.7972114639814097,
      "grad_norm": 4.069510459899902,
      "learning_rate": 1.0000742687085168e-06,
      "loss": 0.0401,
      "step": 15438
    },
    {
      "epoch": 0.7972631035373096,
      "grad_norm": 3.227160930633545,
      "learning_rate": 9.99582689793137e-07,
      "loss": 0.0291,
      "step": 15439
    },
    {
      "epoch": 0.7973147430932094,
      "grad_norm": 2.403468132019043,
      "learning_rate": 9.990912183052664e-07,
      "loss": 0.0595,
      "step": 15440
    },
    {
      "epoch": 0.7973663826491092,
      "grad_norm": 14.624463081359863,
      "learning_rate": 9.985998542581049e-07,
      "loss": 0.2209,
      "step": 15441
    },
    {
      "epoch": 0.797418022205009,
      "grad_norm": 3.6517856121063232,
      "learning_rate": 9.981085976648452e-07,
      "loss": 0.0614,
      "step": 15442
    },
    {
      "epoch": 0.7974696617609088,
      "grad_norm": 1.5494861602783203,
      "learning_rate": 9.976174485386803e-07,
      "loss": 0.0206,
      "step": 15443
    },
    {
      "epoch": 0.7975213013168087,
      "grad_norm": 3.87186861038208,
      "learning_rate": 9.971264068927994e-07,
      "loss": 0.0299,
      "step": 15444
    },
    {
      "epoch": 0.7975729408727085,
      "grad_norm": 3.32466197013855,
      "learning_rate": 9.966354727403892e-07,
      "loss": 0.0207,
      "step": 15445
    },
    {
      "epoch": 0.7976245804286083,
      "grad_norm": 2.247785806655884,
      "learning_rate": 9.96144646094634e-07,
      "loss": 0.0313,
      "step": 15446
    },
    {
      "epoch": 0.7976762199845081,
      "grad_norm": 13.38101577758789,
      "learning_rate": 9.956539269687133e-07,
      "loss": 0.1118,
      "step": 15447
    },
    {
      "epoch": 0.7977278595404079,
      "grad_norm": 9.571113586425781,
      "learning_rate": 9.951633153758055e-07,
      "loss": 0.1169,
      "step": 15448
    },
    {
      "epoch": 0.7977794990963077,
      "grad_norm": 3.776573419570923,
      "learning_rate": 9.94672811329086e-07,
      "loss": 0.0744,
      "step": 15449
    },
    {
      "epoch": 0.7978311386522076,
      "grad_norm": 2.459467887878418,
      "learning_rate": 9.941824148417255e-07,
      "loss": 0.029,
      "step": 15450
    },
    {
      "epoch": 0.7978827782081074,
      "grad_norm": 3.464038133621216,
      "learning_rate": 9.936921259268944e-07,
      "loss": 0.0298,
      "step": 15451
    },
    {
      "epoch": 0.7979344177640072,
      "grad_norm": 9.006101608276367,
      "learning_rate": 9.932019445977587e-07,
      "loss": 0.0289,
      "step": 15452
    },
    {
      "epoch": 0.797986057319907,
      "grad_norm": 3.028777837753296,
      "learning_rate": 9.927118708674815e-07,
      "loss": 0.0333,
      "step": 15453
    },
    {
      "epoch": 0.7980376968758068,
      "grad_norm": 7.735113143920898,
      "learning_rate": 9.922219047492238e-07,
      "loss": 0.0705,
      "step": 15454
    },
    {
      "epoch": 0.7980893364317067,
      "grad_norm": 8.924846649169922,
      "learning_rate": 9.917320462561431e-07,
      "loss": 0.0738,
      "step": 15455
    },
    {
      "epoch": 0.7981409759876065,
      "grad_norm": 4.023042678833008,
      "learning_rate": 9.912422954013934e-07,
      "loss": 0.048,
      "step": 15456
    },
    {
      "epoch": 0.7981926155435063,
      "grad_norm": 7.029473781585693,
      "learning_rate": 9.907526521981247e-07,
      "loss": 0.0862,
      "step": 15457
    },
    {
      "epoch": 0.7982442550994061,
      "grad_norm": 1.9708126783370972,
      "learning_rate": 9.9026311665949e-07,
      "loss": 0.0442,
      "step": 15458
    },
    {
      "epoch": 0.7982958946553059,
      "grad_norm": 1.9978868961334229,
      "learning_rate": 9.897736887986342e-07,
      "loss": 0.0213,
      "step": 15459
    },
    {
      "epoch": 0.7983475342112057,
      "grad_norm": 3.3803584575653076,
      "learning_rate": 9.89284368628699e-07,
      "loss": 0.0525,
      "step": 15460
    },
    {
      "epoch": 0.7983991737671056,
      "grad_norm": 4.714028835296631,
      "learning_rate": 9.887951561628245e-07,
      "loss": 0.0422,
      "step": 15461
    },
    {
      "epoch": 0.7984508133230054,
      "grad_norm": 2.439209222793579,
      "learning_rate": 9.883060514141495e-07,
      "loss": 0.0245,
      "step": 15462
    },
    {
      "epoch": 0.7985024528789052,
      "grad_norm": 3.605642557144165,
      "learning_rate": 9.878170543958076e-07,
      "loss": 0.0687,
      "step": 15463
    },
    {
      "epoch": 0.798554092434805,
      "grad_norm": 3.1920385360717773,
      "learning_rate": 9.873281651209305e-07,
      "loss": 0.047,
      "step": 15464
    },
    {
      "epoch": 0.7986057319907048,
      "grad_norm": 12.714436531066895,
      "learning_rate": 9.868393836026468e-07,
      "loss": 0.0809,
      "step": 15465
    },
    {
      "epoch": 0.7986573715466047,
      "grad_norm": 7.382978439331055,
      "learning_rate": 9.863507098540826e-07,
      "loss": 0.0895,
      "step": 15466
    },
    {
      "epoch": 0.7987090111025045,
      "grad_norm": 20.652109146118164,
      "learning_rate": 9.858621438883603e-07,
      "loss": 0.1284,
      "step": 15467
    },
    {
      "epoch": 0.7987606506584043,
      "grad_norm": 2.989389657974243,
      "learning_rate": 9.853736857186002e-07,
      "loss": 0.0346,
      "step": 15468
    },
    {
      "epoch": 0.7988122902143041,
      "grad_norm": 14.505245208740234,
      "learning_rate": 9.84885335357919e-07,
      "loss": 0.1152,
      "step": 15469
    },
    {
      "epoch": 0.7988639297702039,
      "grad_norm": 4.1294026374816895,
      "learning_rate": 9.843970928194335e-07,
      "loss": 0.0369,
      "step": 15470
    },
    {
      "epoch": 0.7989155693261037,
      "grad_norm": 2.8172545433044434,
      "learning_rate": 9.839089581162492e-07,
      "loss": 0.0428,
      "step": 15471
    },
    {
      "epoch": 0.7989672088820036,
      "grad_norm": 5.218785762786865,
      "learning_rate": 9.834209312614801e-07,
      "loss": 0.0894,
      "step": 15472
    },
    {
      "epoch": 0.7990188484379034,
      "grad_norm": 1.9562739133834839,
      "learning_rate": 9.82933012268229e-07,
      "loss": 0.0216,
      "step": 15473
    },
    {
      "epoch": 0.7990704879938032,
      "grad_norm": 4.884422779083252,
      "learning_rate": 9.824452011496011e-07,
      "loss": 0.063,
      "step": 15474
    },
    {
      "epoch": 0.799122127549703,
      "grad_norm": 2.0598838329315186,
      "learning_rate": 9.819574979186925e-07,
      "loss": 0.0274,
      "step": 15475
    },
    {
      "epoch": 0.7991737671056028,
      "grad_norm": 2.604296922683716,
      "learning_rate": 9.814699025886004e-07,
      "loss": 0.0235,
      "step": 15476
    },
    {
      "epoch": 0.7992254066615027,
      "grad_norm": 2.388848066329956,
      "learning_rate": 9.80982415172423e-07,
      "loss": 0.0312,
      "step": 15477
    },
    {
      "epoch": 0.7992770462174025,
      "grad_norm": 3.5586113929748535,
      "learning_rate": 9.804950356832465e-07,
      "loss": 0.0263,
      "step": 15478
    },
    {
      "epoch": 0.7993286857733023,
      "grad_norm": 1.5184983015060425,
      "learning_rate": 9.800077641341615e-07,
      "loss": 0.0099,
      "step": 15479
    },
    {
      "epoch": 0.7993803253292022,
      "grad_norm": 3.695981979370117,
      "learning_rate": 9.795206005382525e-07,
      "loss": 0.0745,
      "step": 15480
    },
    {
      "epoch": 0.799431964885102,
      "grad_norm": 9.745147705078125,
      "learning_rate": 9.790335449086014e-07,
      "loss": 0.0796,
      "step": 15481
    },
    {
      "epoch": 0.7994836044410019,
      "grad_norm": 3.4749221801757812,
      "learning_rate": 9.785465972582887e-07,
      "loss": 0.0519,
      "step": 15482
    },
    {
      "epoch": 0.7995352439969017,
      "grad_norm": 6.940175533294678,
      "learning_rate": 9.780597576003903e-07,
      "loss": 0.0221,
      "step": 15483
    },
    {
      "epoch": 0.7995868835528015,
      "grad_norm": 1.9597383737564087,
      "learning_rate": 9.775730259479798e-07,
      "loss": 0.0192,
      "step": 15484
    },
    {
      "epoch": 0.7996385231087013,
      "grad_norm": 3.9128787517547607,
      "learning_rate": 9.77086402314128e-07,
      "loss": 0.0479,
      "step": 15485
    },
    {
      "epoch": 0.7996901626646011,
      "grad_norm": 1.6257779598236084,
      "learning_rate": 9.765998867119031e-07,
      "loss": 0.0344,
      "step": 15486
    },
    {
      "epoch": 0.799741802220501,
      "grad_norm": 2.366159439086914,
      "learning_rate": 9.761134791543698e-07,
      "loss": 0.0168,
      "step": 15487
    },
    {
      "epoch": 0.7997934417764008,
      "grad_norm": 2.452148914337158,
      "learning_rate": 9.756271796545907e-07,
      "loss": 0.0392,
      "step": 15488
    },
    {
      "epoch": 0.7998450813323006,
      "grad_norm": 3.4403796195983887,
      "learning_rate": 9.751409882256236e-07,
      "loss": 0.0277,
      "step": 15489
    },
    {
      "epoch": 0.7998967208882004,
      "grad_norm": 4.477235794067383,
      "learning_rate": 9.746549048805236e-07,
      "loss": 0.046,
      "step": 15490
    },
    {
      "epoch": 0.7999483604441002,
      "grad_norm": 5.334791660308838,
      "learning_rate": 9.741689296323475e-07,
      "loss": 0.0513,
      "step": 15491
    },
    {
      "epoch": 0.8,
      "grad_norm": 14.437620162963867,
      "learning_rate": 9.736830624941451e-07,
      "loss": 0.0598,
      "step": 15492
    },
    {
      "epoch": 0.8000516395558999,
      "grad_norm": 10.439349174499512,
      "learning_rate": 9.731973034789617e-07,
      "loss": 0.0728,
      "step": 15493
    },
    {
      "epoch": 0.8001032791117997,
      "grad_norm": 2.4021663665771484,
      "learning_rate": 9.727116525998426e-07,
      "loss": 0.0461,
      "step": 15494
    },
    {
      "epoch": 0.8001549186676995,
      "grad_norm": 16.452835083007812,
      "learning_rate": 9.722261098698289e-07,
      "loss": 0.1358,
      "step": 15495
    },
    {
      "epoch": 0.8002065582235993,
      "grad_norm": 7.821913719177246,
      "learning_rate": 9.717406753019632e-07,
      "loss": 0.0497,
      "step": 15496
    },
    {
      "epoch": 0.8002065582235993,
      "eval_loss": 0.055103447288274765,
      "eval_runtime": 34.1926,
      "eval_samples_per_second": 45.77,
      "eval_steps_per_second": 5.732,
      "step": 15496
    },
    {
      "epoch": 0.8002581977794991,
      "grad_norm": 5.7163872718811035,
      "learning_rate": 9.712553489092775e-07,
      "loss": 0.0655,
      "step": 15497
    },
    {
      "epoch": 0.800309837335399,
      "grad_norm": 5.161425590515137,
      "learning_rate": 9.707701307048057e-07,
      "loss": 0.0697,
      "step": 15498
    },
    {
      "epoch": 0.8003614768912988,
      "grad_norm": 4.4050374031066895,
      "learning_rate": 9.702850207015778e-07,
      "loss": 0.0304,
      "step": 15499
    },
    {
      "epoch": 0.8004131164471986,
      "grad_norm": 11.557352066040039,
      "learning_rate": 9.698000189126218e-07,
      "loss": 0.1095,
      "step": 15500
    },
    {
      "epoch": 0.8004647560030984,
      "grad_norm": 4.5205912590026855,
      "learning_rate": 9.693151253509614e-07,
      "loss": 0.0445,
      "step": 15501
    },
    {
      "epoch": 0.8005163955589982,
      "grad_norm": 5.282510757446289,
      "learning_rate": 9.688303400296179e-07,
      "loss": 0.0689,
      "step": 15502
    },
    {
      "epoch": 0.800568035114898,
      "grad_norm": 5.063680648803711,
      "learning_rate": 9.68345662961611e-07,
      "loss": 0.0525,
      "step": 15503
    },
    {
      "epoch": 0.8006196746707979,
      "grad_norm": 1.8883973360061646,
      "learning_rate": 9.67861094159953e-07,
      "loss": 0.0302,
      "step": 15504
    },
    {
      "epoch": 0.8006713142266977,
      "grad_norm": 2.529487371444702,
      "learning_rate": 9.6737663363766e-07,
      "loss": 0.0356,
      "step": 15505
    },
    {
      "epoch": 0.8007229537825975,
      "grad_norm": 4.345659255981445,
      "learning_rate": 9.668922814077403e-07,
      "loss": 0.0575,
      "step": 15506
    },
    {
      "epoch": 0.8007745933384973,
      "grad_norm": 5.074063301086426,
      "learning_rate": 9.664080374832018e-07,
      "loss": 0.0472,
      "step": 15507
    },
    {
      "epoch": 0.8008262328943971,
      "grad_norm": 3.491032123565674,
      "learning_rate": 9.659239018770466e-07,
      "loss": 0.0311,
      "step": 15508
    },
    {
      "epoch": 0.800877872450297,
      "grad_norm": 10.982423782348633,
      "learning_rate": 9.654398746022753e-07,
      "loss": 0.1042,
      "step": 15509
    },
    {
      "epoch": 0.8009295120061968,
      "grad_norm": 3.4772441387176514,
      "learning_rate": 9.649559556718901e-07,
      "loss": 0.049,
      "step": 15510
    },
    {
      "epoch": 0.8009811515620966,
      "grad_norm": 7.179123401641846,
      "learning_rate": 9.644721450988819e-07,
      "loss": 0.0403,
      "step": 15511
    },
    {
      "epoch": 0.8010327911179964,
      "grad_norm": 5.791641712188721,
      "learning_rate": 9.639884428962443e-07,
      "loss": 0.0595,
      "step": 15512
    },
    {
      "epoch": 0.8010844306738962,
      "grad_norm": 11.203554153442383,
      "learning_rate": 9.635048490769672e-07,
      "loss": 0.1537,
      "step": 15513
    },
    {
      "epoch": 0.801136070229796,
      "grad_norm": 9.577197074890137,
      "learning_rate": 9.63021363654037e-07,
      "loss": 0.0855,
      "step": 15514
    },
    {
      "epoch": 0.8011877097856959,
      "grad_norm": 4.924509048461914,
      "learning_rate": 9.625379866404366e-07,
      "loss": 0.0803,
      "step": 15515
    },
    {
      "epoch": 0.8012393493415957,
      "grad_norm": 3.0060033798217773,
      "learning_rate": 9.620547180491469e-07,
      "loss": 0.0304,
      "step": 15516
    },
    {
      "epoch": 0.8012909888974955,
      "grad_norm": 2.144885540008545,
      "learning_rate": 9.615715578931461e-07,
      "loss": 0.0238,
      "step": 15517
    },
    {
      "epoch": 0.8013426284533953,
      "grad_norm": 5.871430397033691,
      "learning_rate": 9.610885061854085e-07,
      "loss": 0.0483,
      "step": 15518
    },
    {
      "epoch": 0.8013942680092951,
      "grad_norm": 1.1336696147918701,
      "learning_rate": 9.60605562938906e-07,
      "loss": 0.0113,
      "step": 15519
    },
    {
      "epoch": 0.801445907565195,
      "grad_norm": 3.3752713203430176,
      "learning_rate": 9.601227281666082e-07,
      "loss": 0.0653,
      "step": 15520
    },
    {
      "epoch": 0.8014975471210948,
      "grad_norm": 3.6019837856292725,
      "learning_rate": 9.596400018814822e-07,
      "loss": 0.0601,
      "step": 15521
    },
    {
      "epoch": 0.8015491866769946,
      "grad_norm": 1.5775154829025269,
      "learning_rate": 9.59157384096488e-07,
      "loss": 0.0146,
      "step": 15522
    },
    {
      "epoch": 0.8016008262328944,
      "grad_norm": 2.8606715202331543,
      "learning_rate": 9.586748748245867e-07,
      "loss": 0.0436,
      "step": 15523
    },
    {
      "epoch": 0.8016524657887942,
      "grad_norm": 2.5750491619110107,
      "learning_rate": 9.581924740787374e-07,
      "loss": 0.0222,
      "step": 15524
    },
    {
      "epoch": 0.801704105344694,
      "grad_norm": 2.7407565116882324,
      "learning_rate": 9.577101818718954e-07,
      "loss": 0.0554,
      "step": 15525
    },
    {
      "epoch": 0.8017557449005939,
      "grad_norm": 2.4035580158233643,
      "learning_rate": 9.57227998217009e-07,
      "loss": 0.0275,
      "step": 15526
    },
    {
      "epoch": 0.8018073844564937,
      "grad_norm": 3.36185359954834,
      "learning_rate": 9.567459231270282e-07,
      "loss": 0.0494,
      "step": 15527
    },
    {
      "epoch": 0.8018590240123935,
      "grad_norm": 8.05998706817627,
      "learning_rate": 9.562639566148974e-07,
      "loss": 0.1074,
      "step": 15528
    },
    {
      "epoch": 0.8019106635682933,
      "grad_norm": 4.283556938171387,
      "learning_rate": 9.557820986935634e-07,
      "loss": 0.0561,
      "step": 15529
    },
    {
      "epoch": 0.8019623031241931,
      "grad_norm": 3.7967703342437744,
      "learning_rate": 9.553003493759622e-07,
      "loss": 0.0329,
      "step": 15530
    },
    {
      "epoch": 0.802013942680093,
      "grad_norm": 2.03387451171875,
      "learning_rate": 9.548187086750315e-07,
      "loss": 0.0447,
      "step": 15531
    },
    {
      "epoch": 0.8020655822359928,
      "grad_norm": 15.217244148254395,
      "learning_rate": 9.543371766037063e-07,
      "loss": 0.0961,
      "step": 15532
    },
    {
      "epoch": 0.8021172217918926,
      "grad_norm": 4.419281005859375,
      "learning_rate": 9.538557531749165e-07,
      "loss": 0.0347,
      "step": 15533
    },
    {
      "epoch": 0.8021688613477924,
      "grad_norm": 9.287542343139648,
      "learning_rate": 9.533744384015914e-07,
      "loss": 0.0902,
      "step": 15534
    },
    {
      "epoch": 0.8022205009036922,
      "grad_norm": 5.529110431671143,
      "learning_rate": 9.528932322966555e-07,
      "loss": 0.0701,
      "step": 15535
    },
    {
      "epoch": 0.802272140459592,
      "grad_norm": 4.371700286865234,
      "learning_rate": 9.524121348730331e-07,
      "loss": 0.0462,
      "step": 15536
    },
    {
      "epoch": 0.8023237800154919,
      "grad_norm": 4.442671775817871,
      "learning_rate": 9.519311461436393e-07,
      "loss": 0.0707,
      "step": 15537
    },
    {
      "epoch": 0.8023754195713917,
      "grad_norm": 2.819448232650757,
      "learning_rate": 9.514502661213937e-07,
      "loss": 0.0225,
      "step": 15538
    },
    {
      "epoch": 0.8024270591272915,
      "grad_norm": 3.406395435333252,
      "learning_rate": 9.509694948192105e-07,
      "loss": 0.0341,
      "step": 15539
    },
    {
      "epoch": 0.8024786986831913,
      "grad_norm": 7.043431282043457,
      "learning_rate": 9.5048883225e-07,
      "loss": 0.1112,
      "step": 15540
    },
    {
      "epoch": 0.8025303382390911,
      "grad_norm": 8.19688892364502,
      "learning_rate": 9.500082784266678e-07,
      "loss": 0.0685,
      "step": 15541
    },
    {
      "epoch": 0.802581977794991,
      "grad_norm": 2.9970743656158447,
      "learning_rate": 9.495278333621194e-07,
      "loss": 0.0533,
      "step": 15542
    },
    {
      "epoch": 0.8026336173508908,
      "grad_norm": 2.1498329639434814,
      "learning_rate": 9.490474970692598e-07,
      "loss": 0.022,
      "step": 15543
    },
    {
      "epoch": 0.8026852569067906,
      "grad_norm": 2.965502977371216,
      "learning_rate": 9.485672695609843e-07,
      "loss": 0.0492,
      "step": 15544
    },
    {
      "epoch": 0.8027368964626904,
      "grad_norm": 3.6624324321746826,
      "learning_rate": 9.480871508501904e-07,
      "loss": 0.0345,
      "step": 15545
    },
    {
      "epoch": 0.8027885360185902,
      "grad_norm": 1.5757299661636353,
      "learning_rate": 9.476071409497712e-07,
      "loss": 0.0299,
      "step": 15546
    },
    {
      "epoch": 0.8028401755744901,
      "grad_norm": 9.131103515625,
      "learning_rate": 9.471272398726172e-07,
      "loss": 0.0909,
      "step": 15547
    },
    {
      "epoch": 0.8028918151303899,
      "grad_norm": 7.081198692321777,
      "learning_rate": 9.466474476316151e-07,
      "loss": 0.0767,
      "step": 15548
    },
    {
      "epoch": 0.8029434546862897,
      "grad_norm": 1.834686279296875,
      "learning_rate": 9.461677642396494e-07,
      "loss": 0.026,
      "step": 15549
    },
    {
      "epoch": 0.8029950942421895,
      "grad_norm": 2.6373178958892822,
      "learning_rate": 9.456881897096021e-07,
      "loss": 0.0329,
      "step": 15550
    },
    {
      "epoch": 0.8030467337980893,
      "grad_norm": 1.97957181930542,
      "learning_rate": 9.452087240543512e-07,
      "loss": 0.0315,
      "step": 15551
    },
    {
      "epoch": 0.8030983733539891,
      "grad_norm": 3.468209743499756,
      "learning_rate": 9.447293672867725e-07,
      "loss": 0.0443,
      "step": 15552
    },
    {
      "epoch": 0.803150012909889,
      "grad_norm": 11.694742202758789,
      "learning_rate": 9.442501194197384e-07,
      "loss": 0.1291,
      "step": 15553
    },
    {
      "epoch": 0.8032016524657888,
      "grad_norm": 14.492341995239258,
      "learning_rate": 9.437709804661205e-07,
      "loss": 0.1216,
      "step": 15554
    },
    {
      "epoch": 0.8032532920216886,
      "grad_norm": 6.487968444824219,
      "learning_rate": 9.432919504387827e-07,
      "loss": 0.0595,
      "step": 15555
    },
    {
      "epoch": 0.8033049315775884,
      "grad_norm": 7.223744869232178,
      "learning_rate": 9.428130293505888e-07,
      "loss": 0.0556,
      "step": 15556
    },
    {
      "epoch": 0.8033565711334882,
      "grad_norm": 4.851812362670898,
      "learning_rate": 9.423342172144029e-07,
      "loss": 0.0508,
      "step": 15557
    },
    {
      "epoch": 0.8034082106893881,
      "grad_norm": 2.634883165359497,
      "learning_rate": 9.418555140430824e-07,
      "loss": 0.0177,
      "step": 15558
    },
    {
      "epoch": 0.8034598502452879,
      "grad_norm": 2.567858934402466,
      "learning_rate": 9.413769198494805e-07,
      "loss": 0.032,
      "step": 15559
    },
    {
      "epoch": 0.8035114898011877,
      "grad_norm": 3.4448606967926025,
      "learning_rate": 9.4089843464645e-07,
      "loss": 0.0734,
      "step": 15560
    },
    {
      "epoch": 0.8035631293570875,
      "grad_norm": 3.808039665222168,
      "learning_rate": 9.404200584468393e-07,
      "loss": 0.0615,
      "step": 15561
    },
    {
      "epoch": 0.8036147689129873,
      "grad_norm": 3.2486891746520996,
      "learning_rate": 9.399417912634984e-07,
      "loss": 0.0344,
      "step": 15562
    },
    {
      "epoch": 0.8036664084688872,
      "grad_norm": 5.009175777435303,
      "learning_rate": 9.394636331092671e-07,
      "loss": 0.0433,
      "step": 15563
    },
    {
      "epoch": 0.803718048024787,
      "grad_norm": 2.8207170963287354,
      "learning_rate": 9.389855839969875e-07,
      "loss": 0.0289,
      "step": 15564
    },
    {
      "epoch": 0.8037696875806868,
      "grad_norm": 12.749423027038574,
      "learning_rate": 9.385076439394963e-07,
      "loss": 0.1427,
      "step": 15565
    },
    {
      "epoch": 0.8038213271365866,
      "grad_norm": 2.990882396697998,
      "learning_rate": 9.380298129496284e-07,
      "loss": 0.0368,
      "step": 15566
    },
    {
      "epoch": 0.8038729666924864,
      "grad_norm": 5.24368953704834,
      "learning_rate": 9.375520910402158e-07,
      "loss": 0.0345,
      "step": 15567
    },
    {
      "epoch": 0.8039246062483862,
      "grad_norm": 10.981019973754883,
      "learning_rate": 9.370744782240876e-07,
      "loss": 0.0598,
      "step": 15568
    },
    {
      "epoch": 0.8039762458042861,
      "grad_norm": 1.156894326210022,
      "learning_rate": 9.365969745140702e-07,
      "loss": 0.0071,
      "step": 15569
    },
    {
      "epoch": 0.8040278853601859,
      "grad_norm": 3.543654203414917,
      "learning_rate": 9.361195799229828e-07,
      "loss": 0.0349,
      "step": 15570
    },
    {
      "epoch": 0.8040795249160857,
      "grad_norm": 3.4091250896453857,
      "learning_rate": 9.3564229446365e-07,
      "loss": 0.0313,
      "step": 15571
    },
    {
      "epoch": 0.8041311644719855,
      "grad_norm": 2.0377485752105713,
      "learning_rate": 9.351651181488863e-07,
      "loss": 0.0107,
      "step": 15572
    },
    {
      "epoch": 0.8041828040278853,
      "grad_norm": 2.8673980236053467,
      "learning_rate": 9.346880509915085e-07,
      "loss": 0.0449,
      "step": 15573
    },
    {
      "epoch": 0.8042344435837852,
      "grad_norm": 3.37097430229187,
      "learning_rate": 9.342110930043241e-07,
      "loss": 0.0393,
      "step": 15574
    },
    {
      "epoch": 0.804286083139685,
      "grad_norm": 4.675361156463623,
      "learning_rate": 9.337342442001418e-07,
      "loss": 0.0407,
      "step": 15575
    },
    {
      "epoch": 0.8043377226955848,
      "grad_norm": 2.684621810913086,
      "learning_rate": 9.332575045917708e-07,
      "loss": 0.0222,
      "step": 15576
    },
    {
      "epoch": 0.8043893622514846,
      "grad_norm": 4.220729827880859,
      "learning_rate": 9.327808741920097e-07,
      "loss": 0.0653,
      "step": 15577
    },
    {
      "epoch": 0.8044410018073844,
      "grad_norm": 9.770123481750488,
      "learning_rate": 9.323043530136594e-07,
      "loss": 0.0617,
      "step": 15578
    },
    {
      "epoch": 0.8044926413632842,
      "grad_norm": 4.736082553863525,
      "learning_rate": 9.318279410695158e-07,
      "loss": 0.0308,
      "step": 15579
    },
    {
      "epoch": 0.8045442809191841,
      "grad_norm": 3.4531593322753906,
      "learning_rate": 9.313516383723737e-07,
      "loss": 0.0198,
      "step": 15580
    },
    {
      "epoch": 0.8045959204750839,
      "grad_norm": 2.967525005340576,
      "learning_rate": 9.308754449350221e-07,
      "loss": 0.0245,
      "step": 15581
    },
    {
      "epoch": 0.8046475600309837,
      "grad_norm": 3.552828550338745,
      "learning_rate": 9.3039936077025e-07,
      "loss": 0.0649,
      "step": 15582
    },
    {
      "epoch": 0.8046991995868835,
      "grad_norm": 3.500211715698242,
      "learning_rate": 9.299233858908418e-07,
      "loss": 0.03,
      "step": 15583
    },
    {
      "epoch": 0.8047508391427833,
      "grad_norm": 12.336994171142578,
      "learning_rate": 9.294475203095799e-07,
      "loss": 0.0846,
      "step": 15584
    },
    {
      "epoch": 0.8048024786986832,
      "grad_norm": 3.9085094928741455,
      "learning_rate": 9.289717640392421e-07,
      "loss": 0.0294,
      "step": 15585
    },
    {
      "epoch": 0.804854118254583,
      "grad_norm": 3.753847360610962,
      "learning_rate": 9.284961170926055e-07,
      "loss": 0.0354,
      "step": 15586
    },
    {
      "epoch": 0.8049057578104828,
      "grad_norm": 4.101720809936523,
      "learning_rate": 9.280205794824438e-07,
      "loss": 0.0544,
      "step": 15587
    },
    {
      "epoch": 0.8049573973663826,
      "grad_norm": 3.951568126678467,
      "learning_rate": 9.275451512215244e-07,
      "loss": 0.0468,
      "step": 15588
    },
    {
      "epoch": 0.8050090369222824,
      "grad_norm": 2.9384844303131104,
      "learning_rate": 9.270698323226152e-07,
      "loss": 0.0455,
      "step": 15589
    },
    {
      "epoch": 0.8050606764781822,
      "grad_norm": 2.757255792617798,
      "learning_rate": 9.265946227984823e-07,
      "loss": 0.0509,
      "step": 15590
    },
    {
      "epoch": 0.8051123160340821,
      "grad_norm": 2.893559694290161,
      "learning_rate": 9.261195226618874e-07,
      "loss": 0.0391,
      "step": 15591
    },
    {
      "epoch": 0.8051639555899819,
      "grad_norm": 6.312692642211914,
      "learning_rate": 9.256445319255863e-07,
      "loss": 0.0346,
      "step": 15592
    },
    {
      "epoch": 0.8052155951458817,
      "grad_norm": 7.055330276489258,
      "learning_rate": 9.251696506023361e-07,
      "loss": 0.0502,
      "step": 15593
    },
    {
      "epoch": 0.8052672347017815,
      "grad_norm": 3.0463836193084717,
      "learning_rate": 9.24694878704887e-07,
      "loss": 0.046,
      "step": 15594
    },
    {
      "epoch": 0.8053188742576813,
      "grad_norm": 4.250707149505615,
      "learning_rate": 9.242202162459934e-07,
      "loss": 0.0348,
      "step": 15595
    },
    {
      "epoch": 0.8053705138135812,
      "grad_norm": 2.8174753189086914,
      "learning_rate": 9.237456632383973e-07,
      "loss": 0.0546,
      "step": 15596
    },
    {
      "epoch": 0.805422153369481,
      "grad_norm": 7.7511186599731445,
      "learning_rate": 9.232712196948445e-07,
      "loss": 0.1218,
      "step": 15597
    },
    {
      "epoch": 0.8054737929253808,
      "grad_norm": 2.685405731201172,
      "learning_rate": 9.22796885628075e-07,
      "loss": 0.0577,
      "step": 15598
    },
    {
      "epoch": 0.8055254324812806,
      "grad_norm": 3.075636386871338,
      "learning_rate": 9.223226610508268e-07,
      "loss": 0.0508,
      "step": 15599
    },
    {
      "epoch": 0.8055770720371804,
      "grad_norm": 2.6653671264648438,
      "learning_rate": 9.218485459758347e-07,
      "loss": 0.0421,
      "step": 15600
    },
    {
      "epoch": 0.8056287115930802,
      "grad_norm": 2.5760977268218994,
      "learning_rate": 9.213745404158314e-07,
      "loss": 0.0372,
      "step": 15601
    },
    {
      "epoch": 0.8056803511489801,
      "grad_norm": 11.388620376586914,
      "learning_rate": 9.209006443835461e-07,
      "loss": 0.1662,
      "step": 15602
    },
    {
      "epoch": 0.8057319907048799,
      "grad_norm": 14.722338676452637,
      "learning_rate": 9.204268578917014e-07,
      "loss": 0.0723,
      "step": 15603
    },
    {
      "epoch": 0.8057836302607797,
      "grad_norm": 3.1527931690216064,
      "learning_rate": 9.199531809530244e-07,
      "loss": 0.0533,
      "step": 15604
    },
    {
      "epoch": 0.8058352698166795,
      "grad_norm": 2.6605966091156006,
      "learning_rate": 9.194796135802337e-07,
      "loss": 0.0558,
      "step": 15605
    },
    {
      "epoch": 0.8058869093725793,
      "grad_norm": 11.657906532287598,
      "learning_rate": 9.190061557860481e-07,
      "loss": 0.1526,
      "step": 15606
    },
    {
      "epoch": 0.8059385489284793,
      "grad_norm": 2.097946882247925,
      "learning_rate": 9.18532807583179e-07,
      "loss": 0.0397,
      "step": 15607
    },
    {
      "epoch": 0.8059901884843791,
      "grad_norm": 1.7394570112228394,
      "learning_rate": 9.18059568984338e-07,
      "loss": 0.0335,
      "step": 15608
    },
    {
      "epoch": 0.8060418280402789,
      "grad_norm": 5.792451858520508,
      "learning_rate": 9.175864400022372e-07,
      "loss": 0.0351,
      "step": 15609
    },
    {
      "epoch": 0.8060934675961787,
      "grad_norm": 4.339413166046143,
      "learning_rate": 9.17113420649578e-07,
      "loss": 0.0618,
      "step": 15610
    },
    {
      "epoch": 0.8061451071520785,
      "grad_norm": 2.28509521484375,
      "learning_rate": 9.166405109390647e-07,
      "loss": 0.0188,
      "step": 15611
    },
    {
      "epoch": 0.8061967467079784,
      "grad_norm": 1.415494680404663,
      "learning_rate": 9.161677108833961e-07,
      "loss": 0.021,
      "step": 15612
    },
    {
      "epoch": 0.8062483862638782,
      "grad_norm": 9.076032638549805,
      "learning_rate": 9.156950204952697e-07,
      "loss": 0.0575,
      "step": 15613
    },
    {
      "epoch": 0.806300025819778,
      "grad_norm": 29.58297348022461,
      "learning_rate": 9.152224397873782e-07,
      "loss": 0.1366,
      "step": 15614
    },
    {
      "epoch": 0.8063516653756778,
      "grad_norm": 2.954941987991333,
      "learning_rate": 9.147499687724132e-07,
      "loss": 0.0322,
      "step": 15615
    },
    {
      "epoch": 0.8064033049315776,
      "grad_norm": 3.6377954483032227,
      "learning_rate": 9.142776074630622e-07,
      "loss": 0.0405,
      "step": 15616
    },
    {
      "epoch": 0.8064549444874775,
      "grad_norm": 3.858623504638672,
      "learning_rate": 9.138053558720095e-07,
      "loss": 0.0396,
      "step": 15617
    },
    {
      "epoch": 0.8065065840433773,
      "grad_norm": 11.789875030517578,
      "learning_rate": 9.133332140119378e-07,
      "loss": 0.0537,
      "step": 15618
    },
    {
      "epoch": 0.8065582235992771,
      "grad_norm": 3.7954423427581787,
      "learning_rate": 9.128611818955252e-07,
      "loss": 0.0614,
      "step": 15619
    },
    {
      "epoch": 0.8066098631551769,
      "grad_norm": 5.392528057098389,
      "learning_rate": 9.1238925953545e-07,
      "loss": 0.0309,
      "step": 15620
    },
    {
      "epoch": 0.8066615027110767,
      "grad_norm": 4.816732406616211,
      "learning_rate": 9.119174469443815e-07,
      "loss": 0.0405,
      "step": 15621
    },
    {
      "epoch": 0.8067131422669765,
      "grad_norm": 8.889298439025879,
      "learning_rate": 9.114457441349911e-07,
      "loss": 0.1234,
      "step": 15622
    },
    {
      "epoch": 0.8067647818228764,
      "grad_norm": 3.7264044284820557,
      "learning_rate": 9.109741511199472e-07,
      "loss": 0.0536,
      "step": 15623
    },
    {
      "epoch": 0.8068164213787762,
      "grad_norm": 20.46584701538086,
      "learning_rate": 9.105026679119144e-07,
      "loss": 0.0665,
      "step": 15624
    },
    {
      "epoch": 0.806868060934676,
      "grad_norm": 1.4886287450790405,
      "learning_rate": 9.100312945235518e-07,
      "loss": 0.021,
      "step": 15625
    },
    {
      "epoch": 0.8069197004905758,
      "grad_norm": 10.743993759155273,
      "learning_rate": 9.095600309675189e-07,
      "loss": 0.1132,
      "step": 15626
    },
    {
      "epoch": 0.8069713400464756,
      "grad_norm": 3.8077399730682373,
      "learning_rate": 9.090888772564693e-07,
      "loss": 0.0509,
      "step": 15627
    },
    {
      "epoch": 0.8070229796023755,
      "grad_norm": 5.205528259277344,
      "learning_rate": 9.086178334030599e-07,
      "loss": 0.0575,
      "step": 15628
    },
    {
      "epoch": 0.8070746191582753,
      "grad_norm": 2.492238759994507,
      "learning_rate": 9.081468994199356e-07,
      "loss": 0.0289,
      "step": 15629
    },
    {
      "epoch": 0.8071262587141751,
      "grad_norm": 3.077967405319214,
      "learning_rate": 9.07676075319745e-07,
      "loss": 0.0587,
      "step": 15630
    },
    {
      "epoch": 0.8071778982700749,
      "grad_norm": 4.295677661895752,
      "learning_rate": 9.072053611151305e-07,
      "loss": 0.0578,
      "step": 15631
    },
    {
      "epoch": 0.8072295378259747,
      "grad_norm": 1.6241873502731323,
      "learning_rate": 9.06734756818734e-07,
      "loss": 0.0225,
      "step": 15632
    },
    {
      "epoch": 0.8072811773818745,
      "grad_norm": 4.182626247406006,
      "learning_rate": 9.062642624431922e-07,
      "loss": 0.047,
      "step": 15633
    },
    {
      "epoch": 0.8073328169377744,
      "grad_norm": 11.080793380737305,
      "learning_rate": 9.057938780011399e-07,
      "loss": 0.0712,
      "step": 15634
    },
    {
      "epoch": 0.8073844564936742,
      "grad_norm": 6.3508830070495605,
      "learning_rate": 9.053236035052105e-07,
      "loss": 0.0586,
      "step": 15635
    },
    {
      "epoch": 0.807436096049574,
      "grad_norm": 1.2537455558776855,
      "learning_rate": 9.048534389680292e-07,
      "loss": 0.0214,
      "step": 15636
    },
    {
      "epoch": 0.8074877356054738,
      "grad_norm": 2.345440149307251,
      "learning_rate": 9.043833844022243e-07,
      "loss": 0.0423,
      "step": 15637
    },
    {
      "epoch": 0.8075393751613736,
      "grad_norm": 5.284163951873779,
      "learning_rate": 9.039134398204192e-07,
      "loss": 0.1114,
      "step": 15638
    },
    {
      "epoch": 0.8075910147172735,
      "grad_norm": 3.890864610671997,
      "learning_rate": 9.034436052352335e-07,
      "loss": 0.0371,
      "step": 15639
    },
    {
      "epoch": 0.8076426542731733,
      "grad_norm": 9.108236312866211,
      "learning_rate": 9.02973880659283e-07,
      "loss": 0.0374,
      "step": 15640
    },
    {
      "epoch": 0.8076942938290731,
      "grad_norm": 4.153075695037842,
      "learning_rate": 9.025042661051808e-07,
      "loss": 0.0463,
      "step": 15641
    },
    {
      "epoch": 0.8077459333849729,
      "grad_norm": 12.491914749145508,
      "learning_rate": 9.020347615855423e-07,
      "loss": 0.0766,
      "step": 15642
    },
    {
      "epoch": 0.8077975729408727,
      "grad_norm": 21.488262176513672,
      "learning_rate": 9.015653671129715e-07,
      "loss": 0.0861,
      "step": 15643
    },
    {
      "epoch": 0.8078492124967726,
      "grad_norm": 1.8732908964157104,
      "learning_rate": 9.010960827000748e-07,
      "loss": 0.0284,
      "step": 15644
    },
    {
      "epoch": 0.8079008520526724,
      "grad_norm": 8.7418212890625,
      "learning_rate": 9.006269083594543e-07,
      "loss": 0.0903,
      "step": 15645
    },
    {
      "epoch": 0.8079524916085722,
      "grad_norm": 17.000545501708984,
      "learning_rate": 9.001578441037095e-07,
      "loss": 0.0846,
      "step": 15646
    },
    {
      "epoch": 0.808004131164472,
      "grad_norm": 3.393577814102173,
      "learning_rate": 8.996888899454364e-07,
      "loss": 0.0294,
      "step": 15647
    },
    {
      "epoch": 0.8080557707203718,
      "grad_norm": 11.933768272399902,
      "learning_rate": 8.992200458972295e-07,
      "loss": 0.0683,
      "step": 15648
    },
    {
      "epoch": 0.8081074102762716,
      "grad_norm": 7.662743091583252,
      "learning_rate": 8.987513119716773e-07,
      "loss": 0.0914,
      "step": 15649
    },
    {
      "epoch": 0.8081590498321715,
      "grad_norm": 3.9886326789855957,
      "learning_rate": 8.982826881813683e-07,
      "loss": 0.0751,
      "step": 15650
    },
    {
      "epoch": 0.8082106893880713,
      "grad_norm": 6.480963230133057,
      "learning_rate": 8.978141745388869e-07,
      "loss": 0.0974,
      "step": 15651
    },
    {
      "epoch": 0.8082623289439711,
      "grad_norm": 2.7668559551239014,
      "learning_rate": 8.973457710568145e-07,
      "loss": 0.0468,
      "step": 15652
    },
    {
      "epoch": 0.8083139684998709,
      "grad_norm": 5.5164642333984375,
      "learning_rate": 8.968774777477312e-07,
      "loss": 0.1003,
      "step": 15653
    },
    {
      "epoch": 0.8083656080557707,
      "grad_norm": 12.368444442749023,
      "learning_rate": 8.964092946242092e-07,
      "loss": 0.1179,
      "step": 15654
    },
    {
      "epoch": 0.8084172476116706,
      "grad_norm": 2.7310266494750977,
      "learning_rate": 8.959412216988223e-07,
      "loss": 0.023,
      "step": 15655
    },
    {
      "epoch": 0.8084688871675704,
      "grad_norm": 3.181814193725586,
      "learning_rate": 8.954732589841413e-07,
      "loss": 0.049,
      "step": 15656
    },
    {
      "epoch": 0.8085205267234702,
      "grad_norm": 2.362259864807129,
      "learning_rate": 8.950054064927343e-07,
      "loss": 0.0298,
      "step": 15657
    },
    {
      "epoch": 0.80857216627937,
      "grad_norm": 4.1553635597229,
      "learning_rate": 8.945376642371617e-07,
      "loss": 0.0659,
      "step": 15658
    },
    {
      "epoch": 0.8086238058352698,
      "grad_norm": 16.704830169677734,
      "learning_rate": 8.940700322299856e-07,
      "loss": 0.1466,
      "step": 15659
    },
    {
      "epoch": 0.8086754453911696,
      "grad_norm": 6.527255058288574,
      "learning_rate": 8.936025104837631e-07,
      "loss": 0.0473,
      "step": 15660
    },
    {
      "epoch": 0.8087270849470695,
      "grad_norm": 2.2892045974731445,
      "learning_rate": 8.931350990110521e-07,
      "loss": 0.0405,
      "step": 15661
    },
    {
      "epoch": 0.8087787245029693,
      "grad_norm": 7.759746074676514,
      "learning_rate": 8.926677978244014e-07,
      "loss": 0.0332,
      "step": 15662
    },
    {
      "epoch": 0.8088303640588691,
      "grad_norm": 4.503552436828613,
      "learning_rate": 8.922006069363609e-07,
      "loss": 0.0577,
      "step": 15663
    },
    {
      "epoch": 0.8088820036147689,
      "grad_norm": 2.5613553524017334,
      "learning_rate": 8.917335263594762e-07,
      "loss": 0.0303,
      "step": 15664
    },
    {
      "epoch": 0.8089336431706687,
      "grad_norm": 2.1698367595672607,
      "learning_rate": 8.912665561062911e-07,
      "loss": 0.0351,
      "step": 15665
    },
    {
      "epoch": 0.8089852827265686,
      "grad_norm": 2.3009841442108154,
      "learning_rate": 8.907996961893451e-07,
      "loss": 0.0408,
      "step": 15666
    },
    {
      "epoch": 0.8090369222824684,
      "grad_norm": 4.803154945373535,
      "learning_rate": 8.903329466211752e-07,
      "loss": 0.0629,
      "step": 15667
    },
    {
      "epoch": 0.8090885618383682,
      "grad_norm": 10.256998062133789,
      "learning_rate": 8.898663074143171e-07,
      "loss": 0.1207,
      "step": 15668
    },
    {
      "epoch": 0.809140201394268,
      "grad_norm": 6.055947303771973,
      "learning_rate": 8.893997785812985e-07,
      "loss": 0.039,
      "step": 15669
    },
    {
      "epoch": 0.8091918409501678,
      "grad_norm": 3.9966320991516113,
      "learning_rate": 8.889333601346506e-07,
      "loss": 0.0911,
      "step": 15670
    },
    {
      "epoch": 0.8092434805060676,
      "grad_norm": 6.32362699508667,
      "learning_rate": 8.884670520868976e-07,
      "loss": 0.095,
      "step": 15671
    },
    {
      "epoch": 0.8092951200619675,
      "grad_norm": 2.190372943878174,
      "learning_rate": 8.880008544505637e-07,
      "loss": 0.0438,
      "step": 15672
    },
    {
      "epoch": 0.8093467596178673,
      "grad_norm": 4.940375328063965,
      "learning_rate": 8.875347672381646e-07,
      "loss": 0.0408,
      "step": 15673
    },
    {
      "epoch": 0.8093983991737671,
      "grad_norm": 4.357707500457764,
      "learning_rate": 8.870687904622177e-07,
      "loss": 0.1056,
      "step": 15674
    },
    {
      "epoch": 0.8094500387296669,
      "grad_norm": 2.9459517002105713,
      "learning_rate": 8.866029241352397e-07,
      "loss": 0.0313,
      "step": 15675
    },
    {
      "epoch": 0.8095016782855667,
      "grad_norm": 2.14876127243042,
      "learning_rate": 8.861371682697367e-07,
      "loss": 0.0203,
      "step": 15676
    },
    {
      "epoch": 0.8095533178414666,
      "grad_norm": 1.5668648481369019,
      "learning_rate": 8.856715228782181e-07,
      "loss": 0.0155,
      "step": 15677
    },
    {
      "epoch": 0.8096049573973664,
      "grad_norm": 7.405007839202881,
      "learning_rate": 8.852059879731884e-07,
      "loss": 0.0639,
      "step": 15678
    },
    {
      "epoch": 0.8096565969532662,
      "grad_norm": 3.0682122707366943,
      "learning_rate": 8.847405635671491e-07,
      "loss": 0.0406,
      "step": 15679
    },
    {
      "epoch": 0.809708236509166,
      "grad_norm": 3.364293098449707,
      "learning_rate": 8.842752496725981e-07,
      "loss": 0.0389,
      "step": 15680
    },
    {
      "epoch": 0.8097598760650658,
      "grad_norm": 4.2320146560668945,
      "learning_rate": 8.838100463020315e-07,
      "loss": 0.0264,
      "step": 15681
    },
    {
      "epoch": 0.8098115156209656,
      "grad_norm": 1.183064341545105,
      "learning_rate": 8.833449534679416e-07,
      "loss": 0.0167,
      "step": 15682
    },
    {
      "epoch": 0.8098631551768655,
      "grad_norm": 1.9463831186294556,
      "learning_rate": 8.82879971182819e-07,
      "loss": 0.0237,
      "step": 15683
    },
    {
      "epoch": 0.8099147947327653,
      "grad_norm": 2.4385643005371094,
      "learning_rate": 8.824150994591491e-07,
      "loss": 0.0438,
      "step": 15684
    },
    {
      "epoch": 0.8099664342886651,
      "grad_norm": 5.4254231452941895,
      "learning_rate": 8.819503383094158e-07,
      "loss": 0.0661,
      "step": 15685
    },
    {
      "epoch": 0.8100180738445649,
      "grad_norm": 2.218585252761841,
      "learning_rate": 8.81485687746102e-07,
      "loss": 0.0378,
      "step": 15686
    },
    {
      "epoch": 0.8100697134004647,
      "grad_norm": 7.393226623535156,
      "learning_rate": 8.810211477816821e-07,
      "loss": 0.1098,
      "step": 15687
    },
    {
      "epoch": 0.8101213529563646,
      "grad_norm": 2.1995930671691895,
      "learning_rate": 8.805567184286312e-07,
      "loss": 0.0321,
      "step": 15688
    },
    {
      "epoch": 0.8101729925122644,
      "grad_norm": 3.5457723140716553,
      "learning_rate": 8.800923996994237e-07,
      "loss": 0.0318,
      "step": 15689
    },
    {
      "epoch": 0.8102246320681642,
      "grad_norm": 2.7507193088531494,
      "learning_rate": 8.79628191606528e-07,
      "loss": 0.0415,
      "step": 15690
    },
    {
      "epoch": 0.810276271624064,
      "grad_norm": 3.058339834213257,
      "learning_rate": 8.791640941624079e-07,
      "loss": 0.0551,
      "step": 15691
    },
    {
      "epoch": 0.8103279111799638,
      "grad_norm": 2.921236753463745,
      "learning_rate": 8.787001073795282e-07,
      "loss": 0.0499,
      "step": 15692
    },
    {
      "epoch": 0.8103795507358637,
      "grad_norm": 4.007106781005859,
      "learning_rate": 8.782362312703463e-07,
      "loss": 0.0475,
      "step": 15693
    },
    {
      "epoch": 0.8104311902917635,
      "grad_norm": 7.039465427398682,
      "learning_rate": 8.77772465847323e-07,
      "loss": 0.057,
      "step": 15694
    },
    {
      "epoch": 0.8104828298476633,
      "grad_norm": 2.7151222229003906,
      "learning_rate": 8.773088111229094e-07,
      "loss": 0.0346,
      "step": 15695
    },
    {
      "epoch": 0.8105344694035631,
      "grad_norm": 4.961350440979004,
      "learning_rate": 8.768452671095579e-07,
      "loss": 0.0628,
      "step": 15696
    },
    {
      "epoch": 0.8105861089594629,
      "grad_norm": 6.441860675811768,
      "learning_rate": 8.763818338197155e-07,
      "loss": 0.0581,
      "step": 15697
    },
    {
      "epoch": 0.8106377485153627,
      "grad_norm": 3.8992977142333984,
      "learning_rate": 8.759185112658275e-07,
      "loss": 0.0317,
      "step": 15698
    },
    {
      "epoch": 0.8106893880712626,
      "grad_norm": 2.201475143432617,
      "learning_rate": 8.754552994603365e-07,
      "loss": 0.0285,
      "step": 15699
    },
    {
      "epoch": 0.8107410276271624,
      "grad_norm": 6.26355504989624,
      "learning_rate": 8.749921984156817e-07,
      "loss": 0.0735,
      "step": 15700
    },
    {
      "epoch": 0.8107926671830622,
      "grad_norm": 13.858973503112793,
      "learning_rate": 8.745292081443002e-07,
      "loss": 0.1098,
      "step": 15701
    },
    {
      "epoch": 0.810844306738962,
      "grad_norm": 1.3082220554351807,
      "learning_rate": 8.740663286586215e-07,
      "loss": 0.0178,
      "step": 15702
    },
    {
      "epoch": 0.8108959462948618,
      "grad_norm": 6.197044849395752,
      "learning_rate": 8.736035599710796e-07,
      "loss": 0.0688,
      "step": 15703
    },
    {
      "epoch": 0.8109475858507617,
      "grad_norm": 3.5089728832244873,
      "learning_rate": 8.731409020941e-07,
      "loss": 0.052,
      "step": 15704
    },
    {
      "epoch": 0.8109992254066615,
      "grad_norm": 3.6413214206695557,
      "learning_rate": 8.726783550401086e-07,
      "loss": 0.0496,
      "step": 15705
    },
    {
      "epoch": 0.8110508649625613,
      "grad_norm": 2.7892541885375977,
      "learning_rate": 8.72215918821524e-07,
      "loss": 0.0282,
      "step": 15706
    },
    {
      "epoch": 0.8111025045184611,
      "grad_norm": 12.11876392364502,
      "learning_rate": 8.717535934507654e-07,
      "loss": 0.1362,
      "step": 15707
    },
    {
      "epoch": 0.8111541440743609,
      "grad_norm": 4.562901973724365,
      "learning_rate": 8.712913789402505e-07,
      "loss": 0.0403,
      "step": 15708
    },
    {
      "epoch": 0.8112057836302607,
      "grad_norm": 6.475399494171143,
      "learning_rate": 8.708292753023883e-07,
      "loss": 0.057,
      "step": 15709
    },
    {
      "epoch": 0.8112574231861606,
      "grad_norm": 2.8776891231536865,
      "learning_rate": 8.703672825495901e-07,
      "loss": 0.0289,
      "step": 15710
    },
    {
      "epoch": 0.8113090627420604,
      "grad_norm": 1.7683656215667725,
      "learning_rate": 8.69905400694262e-07,
      "loss": 0.0172,
      "step": 15711
    },
    {
      "epoch": 0.8113607022979602,
      "grad_norm": 1.8041306734085083,
      "learning_rate": 8.694436297488068e-07,
      "loss": 0.0272,
      "step": 15712
    },
    {
      "epoch": 0.81141234185386,
      "grad_norm": 4.057863712310791,
      "learning_rate": 8.689819697256252e-07,
      "loss": 0.0362,
      "step": 15713
    },
    {
      "epoch": 0.8114639814097598,
      "grad_norm": 2.3559486865997314,
      "learning_rate": 8.685204206371151e-07,
      "loss": 0.0462,
      "step": 15714
    },
    {
      "epoch": 0.8115156209656597,
      "grad_norm": 8.260257720947266,
      "learning_rate": 8.680589824956708e-07,
      "loss": 0.0644,
      "step": 15715
    },
    {
      "epoch": 0.8115672605215595,
      "grad_norm": 4.606751918792725,
      "learning_rate": 8.675976553136833e-07,
      "loss": 0.0511,
      "step": 15716
    },
    {
      "epoch": 0.8116189000774593,
      "grad_norm": 1.5680246353149414,
      "learning_rate": 8.671364391035419e-07,
      "loss": 0.0211,
      "step": 15717
    },
    {
      "epoch": 0.8116705396333591,
      "grad_norm": 13.710351943969727,
      "learning_rate": 8.666753338776313e-07,
      "loss": 0.1935,
      "step": 15718
    },
    {
      "epoch": 0.8117221791892589,
      "grad_norm": 8.87840461730957,
      "learning_rate": 8.662143396483364e-07,
      "loss": 0.0456,
      "step": 15719
    },
    {
      "epoch": 0.8117738187451587,
      "grad_norm": 11.045027732849121,
      "learning_rate": 8.657534564280329e-07,
      "loss": 0.1265,
      "step": 15720
    },
    {
      "epoch": 0.8118254583010586,
      "grad_norm": 2.9588935375213623,
      "learning_rate": 8.652926842290987e-07,
      "loss": 0.0491,
      "step": 15721
    },
    {
      "epoch": 0.8118770978569584,
      "grad_norm": 3.3566043376922607,
      "learning_rate": 8.64832023063909e-07,
      "loss": 0.0333,
      "step": 15722
    },
    {
      "epoch": 0.8119287374128582,
      "grad_norm": 3.5816991329193115,
      "learning_rate": 8.643714729448343e-07,
      "loss": 0.0325,
      "step": 15723
    },
    {
      "epoch": 0.811980376968758,
      "grad_norm": 4.727872371673584,
      "learning_rate": 8.63911033884241e-07,
      "loss": 0.0267,
      "step": 15724
    },
    {
      "epoch": 0.8120320165246578,
      "grad_norm": 7.361947536468506,
      "learning_rate": 8.634507058944936e-07,
      "loss": 0.05,
      "step": 15725
    },
    {
      "epoch": 0.8120836560805577,
      "grad_norm": 3.3213727474212646,
      "learning_rate": 8.629904889879537e-07,
      "loss": 0.05,
      "step": 15726
    },
    {
      "epoch": 0.8121352956364575,
      "grad_norm": 3.873868465423584,
      "learning_rate": 8.625303831769827e-07,
      "loss": 0.0238,
      "step": 15727
    },
    {
      "epoch": 0.8121869351923573,
      "grad_norm": 17.35140037536621,
      "learning_rate": 8.620703884739334e-07,
      "loss": 0.0962,
      "step": 15728
    },
    {
      "epoch": 0.8122385747482571,
      "grad_norm": 2.5669009685516357,
      "learning_rate": 8.616105048911594e-07,
      "loss": 0.0373,
      "step": 15729
    },
    {
      "epoch": 0.8122902143041569,
      "grad_norm": 14.53336238861084,
      "learning_rate": 8.611507324410102e-07,
      "loss": 0.0718,
      "step": 15730
    },
    {
      "epoch": 0.8123418538600568,
      "grad_norm": 4.724002361297607,
      "learning_rate": 8.606910711358329e-07,
      "loss": 0.0585,
      "step": 15731
    },
    {
      "epoch": 0.8123934934159566,
      "grad_norm": 3.0063323974609375,
      "learning_rate": 8.602315209879713e-07,
      "loss": 0.0483,
      "step": 15732
    },
    {
      "epoch": 0.8124451329718564,
      "grad_norm": 3.9097554683685303,
      "learning_rate": 8.59772082009766e-07,
      "loss": 0.0376,
      "step": 15733
    },
    {
      "epoch": 0.8124967725277562,
      "grad_norm": 3.443246841430664,
      "learning_rate": 8.593127542135566e-07,
      "loss": 0.0445,
      "step": 15734
    },
    {
      "epoch": 0.8125484120836561,
      "grad_norm": 4.005954742431641,
      "learning_rate": 8.588535376116736e-07,
      "loss": 0.0309,
      "step": 15735
    },
    {
      "epoch": 0.812600051639556,
      "grad_norm": 1.9993809461593628,
      "learning_rate": 8.583944322164528e-07,
      "loss": 0.0249,
      "step": 15736
    },
    {
      "epoch": 0.8126516911954558,
      "grad_norm": 8.074374198913574,
      "learning_rate": 8.579354380402217e-07,
      "loss": 0.083,
      "step": 15737
    },
    {
      "epoch": 0.8127033307513556,
      "grad_norm": 4.342594623565674,
      "learning_rate": 8.574765550953068e-07,
      "loss": 0.0553,
      "step": 15738
    },
    {
      "epoch": 0.8127549703072554,
      "grad_norm": 2.677734136581421,
      "learning_rate": 8.570177833940296e-07,
      "loss": 0.0445,
      "step": 15739
    },
    {
      "epoch": 0.8128066098631552,
      "grad_norm": 1.5520391464233398,
      "learning_rate": 8.565591229487097e-07,
      "loss": 0.0286,
      "step": 15740
    },
    {
      "epoch": 0.812858249419055,
      "grad_norm": 1.9753495454788208,
      "learning_rate": 8.561005737716677e-07,
      "loss": 0.0192,
      "step": 15741
    },
    {
      "epoch": 0.8129098889749549,
      "grad_norm": 20.651519775390625,
      "learning_rate": 8.556421358752137e-07,
      "loss": 0.1188,
      "step": 15742
    },
    {
      "epoch": 0.8129615285308547,
      "grad_norm": 9.128889083862305,
      "learning_rate": 8.551838092716591e-07,
      "loss": 0.0645,
      "step": 15743
    },
    {
      "epoch": 0.8130131680867545,
      "grad_norm": 2.7819104194641113,
      "learning_rate": 8.547255939733134e-07,
      "loss": 0.0473,
      "step": 15744
    },
    {
      "epoch": 0.8130648076426543,
      "grad_norm": 9.6851806640625,
      "learning_rate": 8.542674899924802e-07,
      "loss": 0.0643,
      "step": 15745
    },
    {
      "epoch": 0.8131164471985541,
      "grad_norm": 3.666722297668457,
      "learning_rate": 8.538094973414618e-07,
      "loss": 0.0552,
      "step": 15746
    },
    {
      "epoch": 0.813168086754454,
      "grad_norm": 7.3836588859558105,
      "learning_rate": 8.533516160325578e-07,
      "loss": 0.039,
      "step": 15747
    },
    {
      "epoch": 0.8132197263103538,
      "grad_norm": 3.593038320541382,
      "learning_rate": 8.528938460780633e-07,
      "loss": 0.0397,
      "step": 15748
    },
    {
      "epoch": 0.8132713658662536,
      "grad_norm": 1.987295150756836,
      "learning_rate": 8.524361874902721e-07,
      "loss": 0.025,
      "step": 15749
    },
    {
      "epoch": 0.8133230054221534,
      "grad_norm": 6.380077362060547,
      "learning_rate": 8.519786402814739e-07,
      "loss": 0.0374,
      "step": 15750
    },
    {
      "epoch": 0.8133746449780532,
      "grad_norm": 2.942671060562134,
      "learning_rate": 8.515212044639554e-07,
      "loss": 0.0614,
      "step": 15751
    },
    {
      "epoch": 0.813426284533953,
      "grad_norm": 15.522665023803711,
      "learning_rate": 8.510638800500021e-07,
      "loss": 0.0501,
      "step": 15752
    },
    {
      "epoch": 0.8134779240898529,
      "grad_norm": 7.217593669891357,
      "learning_rate": 8.50606667051892e-07,
      "loss": 0.0409,
      "step": 15753
    },
    {
      "epoch": 0.8135295636457527,
      "grad_norm": 3.1188321113586426,
      "learning_rate": 8.501495654819042e-07,
      "loss": 0.0481,
      "step": 15754
    },
    {
      "epoch": 0.8135812032016525,
      "grad_norm": 3.1646430492401123,
      "learning_rate": 8.496925753523156e-07,
      "loss": 0.039,
      "step": 15755
    },
    {
      "epoch": 0.8136328427575523,
      "grad_norm": 37.88009262084961,
      "learning_rate": 8.492356966753984e-07,
      "loss": 0.0789,
      "step": 15756
    },
    {
      "epoch": 0.8136844823134521,
      "grad_norm": 16.294017791748047,
      "learning_rate": 8.487789294634185e-07,
      "loss": 0.127,
      "step": 15757
    },
    {
      "epoch": 0.813736121869352,
      "grad_norm": 2.6238112449645996,
      "learning_rate": 8.483222737286445e-07,
      "loss": 0.0261,
      "step": 15758
    },
    {
      "epoch": 0.8137877614252518,
      "grad_norm": 5.11217737197876,
      "learning_rate": 8.478657294833381e-07,
      "loss": 0.0588,
      "step": 15759
    },
    {
      "epoch": 0.8138394009811516,
      "grad_norm": 9.841206550598145,
      "learning_rate": 8.474092967397606e-07,
      "loss": 0.1054,
      "step": 15760
    },
    {
      "epoch": 0.8138910405370514,
      "grad_norm": 3.475939989089966,
      "learning_rate": 8.469529755101686e-07,
      "loss": 0.0358,
      "step": 15761
    },
    {
      "epoch": 0.8139426800929512,
      "grad_norm": 1.6219468116760254,
      "learning_rate": 8.464967658068157e-07,
      "loss": 0.0177,
      "step": 15762
    },
    {
      "epoch": 0.813994319648851,
      "grad_norm": 3.371260166168213,
      "learning_rate": 8.46040667641953e-07,
      "loss": 0.04,
      "step": 15763
    },
    {
      "epoch": 0.8140459592047509,
      "grad_norm": 4.027309417724609,
      "learning_rate": 8.455846810278295e-07,
      "loss": 0.0311,
      "step": 15764
    },
    {
      "epoch": 0.8140975987606507,
      "grad_norm": 2.814246416091919,
      "learning_rate": 8.451288059766899e-07,
      "loss": 0.0428,
      "step": 15765
    },
    {
      "epoch": 0.8141492383165505,
      "grad_norm": 9.501434326171875,
      "learning_rate": 8.44673042500776e-07,
      "loss": 0.1195,
      "step": 15766
    },
    {
      "epoch": 0.8142008778724503,
      "grad_norm": 3.2354583740234375,
      "learning_rate": 8.442173906123286e-07,
      "loss": 0.0305,
      "step": 15767
    },
    {
      "epoch": 0.8142525174283501,
      "grad_norm": 4.622481346130371,
      "learning_rate": 8.437618503235789e-07,
      "loss": 0.0242,
      "step": 15768
    },
    {
      "epoch": 0.81430415698425,
      "grad_norm": 2.50315523147583,
      "learning_rate": 8.433064216467657e-07,
      "loss": 0.0257,
      "step": 15769
    },
    {
      "epoch": 0.8143557965401498,
      "grad_norm": 4.080210208892822,
      "learning_rate": 8.428511045941174e-07,
      "loss": 0.0857,
      "step": 15770
    },
    {
      "epoch": 0.8144074360960496,
      "grad_norm": 2.2120330333709717,
      "learning_rate": 8.423958991778597e-07,
      "loss": 0.0435,
      "step": 15771
    },
    {
      "epoch": 0.8144590756519494,
      "grad_norm": 15.299187660217285,
      "learning_rate": 8.419408054102174e-07,
      "loss": 0.2225,
      "step": 15772
    },
    {
      "epoch": 0.8145107152078492,
      "grad_norm": 3.287720203399658,
      "learning_rate": 8.414858233034106e-07,
      "loss": 0.0478,
      "step": 15773
    },
    {
      "epoch": 0.814562354763749,
      "grad_norm": 11.762463569641113,
      "learning_rate": 8.410309528696608e-07,
      "loss": 0.0815,
      "step": 15774
    },
    {
      "epoch": 0.8146139943196489,
      "grad_norm": 2.58632493019104,
      "learning_rate": 8.405761941211799e-07,
      "loss": 0.0384,
      "step": 15775
    },
    {
      "epoch": 0.8146656338755487,
      "grad_norm": 3.027423858642578,
      "learning_rate": 8.401215470701813e-07,
      "loss": 0.0204,
      "step": 15776
    },
    {
      "epoch": 0.8147172734314485,
      "grad_norm": 7.4860920906066895,
      "learning_rate": 8.396670117288735e-07,
      "loss": 0.0908,
      "step": 15777
    },
    {
      "epoch": 0.8147689129873483,
      "grad_norm": 4.3670477867126465,
      "learning_rate": 8.392125881094632e-07,
      "loss": 0.0736,
      "step": 15778
    },
    {
      "epoch": 0.8148205525432481,
      "grad_norm": 1.5913856029510498,
      "learning_rate": 8.387582762241536e-07,
      "loss": 0.0283,
      "step": 15779
    },
    {
      "epoch": 0.814872192099148,
      "grad_norm": 3.171964645385742,
      "learning_rate": 8.383040760851441e-07,
      "loss": 0.0421,
      "step": 15780
    },
    {
      "epoch": 0.8149238316550478,
      "grad_norm": 6.18284273147583,
      "learning_rate": 8.378499877046337e-07,
      "loss": 0.0858,
      "step": 15781
    },
    {
      "epoch": 0.8149754712109476,
      "grad_norm": 2.4865589141845703,
      "learning_rate": 8.373960110948126e-07,
      "loss": 0.023,
      "step": 15782
    },
    {
      "epoch": 0.8150271107668474,
      "grad_norm": 5.529481887817383,
      "learning_rate": 8.369421462678762e-07,
      "loss": 0.0282,
      "step": 15783
    },
    {
      "epoch": 0.8150787503227472,
      "grad_norm": 3.681539535522461,
      "learning_rate": 8.364883932360102e-07,
      "loss": 0.0376,
      "step": 15784
    },
    {
      "epoch": 0.815130389878647,
      "grad_norm": 2.262321949005127,
      "learning_rate": 8.360347520114015e-07,
      "loss": 0.0264,
      "step": 15785
    },
    {
      "epoch": 0.8151820294345469,
      "grad_norm": 1.8500155210494995,
      "learning_rate": 8.355812226062304e-07,
      "loss": 0.0389,
      "step": 15786
    },
    {
      "epoch": 0.8152336689904467,
      "grad_norm": 8.6339693069458,
      "learning_rate": 8.351278050326756e-07,
      "loss": 0.0746,
      "step": 15787
    },
    {
      "epoch": 0.8152853085463465,
      "grad_norm": 2.653012275695801,
      "learning_rate": 8.346744993029149e-07,
      "loss": 0.0387,
      "step": 15788
    },
    {
      "epoch": 0.8153369481022463,
      "grad_norm": 2.256333351135254,
      "learning_rate": 8.34221305429122e-07,
      "loss": 0.0272,
      "step": 15789
    },
    {
      "epoch": 0.8153885876581461,
      "grad_norm": 3.1204841136932373,
      "learning_rate": 8.337682234234651e-07,
      "loss": 0.0329,
      "step": 15790
    },
    {
      "epoch": 0.815440227214046,
      "grad_norm": 5.701000690460205,
      "learning_rate": 8.333152532981121e-07,
      "loss": 0.0841,
      "step": 15791
    },
    {
      "epoch": 0.8154918667699458,
      "grad_norm": 4.087320327758789,
      "learning_rate": 8.328623950652265e-07,
      "loss": 0.067,
      "step": 15792
    },
    {
      "epoch": 0.8155435063258456,
      "grad_norm": 15.812385559082031,
      "learning_rate": 8.324096487369704e-07,
      "loss": 0.0885,
      "step": 15793
    },
    {
      "epoch": 0.8155951458817454,
      "grad_norm": 2.023218870162964,
      "learning_rate": 8.319570143255013e-07,
      "loss": 0.037,
      "step": 15794
    },
    {
      "epoch": 0.8156467854376452,
      "grad_norm": 8.702240943908691,
      "learning_rate": 8.315044918429744e-07,
      "loss": 0.1261,
      "step": 15795
    },
    {
      "epoch": 0.8156984249935451,
      "grad_norm": 15.914907455444336,
      "learning_rate": 8.31052081301541e-07,
      "loss": 0.0657,
      "step": 15796
    },
    {
      "epoch": 0.8157500645494449,
      "grad_norm": 2.831735134124756,
      "learning_rate": 8.305997827133517e-07,
      "loss": 0.0276,
      "step": 15797
    },
    {
      "epoch": 0.8158017041053447,
      "grad_norm": 7.750216484069824,
      "learning_rate": 8.301475960905514e-07,
      "loss": 0.1389,
      "step": 15798
    },
    {
      "epoch": 0.8158533436612445,
      "grad_norm": 2.064635992050171,
      "learning_rate": 8.29695521445284e-07,
      "loss": 0.0294,
      "step": 15799
    },
    {
      "epoch": 0.8159049832171443,
      "grad_norm": 8.530219078063965,
      "learning_rate": 8.292435587896896e-07,
      "loss": 0.072,
      "step": 15800
    },
    {
      "epoch": 0.8159566227730441,
      "grad_norm": 3.930675983428955,
      "learning_rate": 8.287917081359032e-07,
      "loss": 0.033,
      "step": 15801
    },
    {
      "epoch": 0.816008262328944,
      "grad_norm": 3.379579782485962,
      "learning_rate": 8.283399694960608e-07,
      "loss": 0.0314,
      "step": 15802
    },
    {
      "epoch": 0.8160599018848438,
      "grad_norm": 6.967209815979004,
      "learning_rate": 8.278883428822943e-07,
      "loss": 0.0734,
      "step": 15803
    },
    {
      "epoch": 0.8161115414407436,
      "grad_norm": 2.3618195056915283,
      "learning_rate": 8.274368283067291e-07,
      "loss": 0.049,
      "step": 15804
    },
    {
      "epoch": 0.8161631809966434,
      "grad_norm": 11.984146118164062,
      "learning_rate": 8.269854257814919e-07,
      "loss": 0.1493,
      "step": 15805
    },
    {
      "epoch": 0.8162148205525432,
      "grad_norm": 16.425199508666992,
      "learning_rate": 8.265341353187028e-07,
      "loss": 0.0835,
      "step": 15806
    },
    {
      "epoch": 0.8162664601084431,
      "grad_norm": 2.785707950592041,
      "learning_rate": 8.260829569304846e-07,
      "loss": 0.0241,
      "step": 15807
    },
    {
      "epoch": 0.8163180996643429,
      "grad_norm": 1.8708109855651855,
      "learning_rate": 8.256318906289495e-07,
      "loss": 0.0241,
      "step": 15808
    },
    {
      "epoch": 0.8163697392202427,
      "grad_norm": 3.0415828227996826,
      "learning_rate": 8.251809364262125e-07,
      "loss": 0.0461,
      "step": 15809
    },
    {
      "epoch": 0.8164213787761425,
      "grad_norm": 9.385844230651855,
      "learning_rate": 8.247300943343828e-07,
      "loss": 0.0654,
      "step": 15810
    },
    {
      "epoch": 0.8164730183320423,
      "grad_norm": 2.110713243484497,
      "learning_rate": 8.242793643655672e-07,
      "loss": 0.0157,
      "step": 15811
    },
    {
      "epoch": 0.8165246578879422,
      "grad_norm": 2.7882637977600098,
      "learning_rate": 8.238287465318707e-07,
      "loss": 0.034,
      "step": 15812
    },
    {
      "epoch": 0.816576297443842,
      "grad_norm": 2.4220919609069824,
      "learning_rate": 8.233782408453927e-07,
      "loss": 0.0348,
      "step": 15813
    },
    {
      "epoch": 0.8166279369997418,
      "grad_norm": 10.393237113952637,
      "learning_rate": 8.229278473182333e-07,
      "loss": 0.0731,
      "step": 15814
    },
    {
      "epoch": 0.8166795765556416,
      "grad_norm": 4.30955171585083,
      "learning_rate": 8.22477565962484e-07,
      "loss": 0.0524,
      "step": 15815
    },
    {
      "epoch": 0.8167312161115414,
      "grad_norm": 3.8291401863098145,
      "learning_rate": 8.220273967902398e-07,
      "loss": 0.044,
      "step": 15816
    },
    {
      "epoch": 0.8167828556674412,
      "grad_norm": 3.219592809677124,
      "learning_rate": 8.215773398135879e-07,
      "loss": 0.0467,
      "step": 15817
    },
    {
      "epoch": 0.8168344952233411,
      "grad_norm": 2.3278989791870117,
      "learning_rate": 8.211273950446163e-07,
      "loss": 0.0157,
      "step": 15818
    },
    {
      "epoch": 0.8168861347792409,
      "grad_norm": 2.7080721855163574,
      "learning_rate": 8.206775624954055e-07,
      "loss": 0.035,
      "step": 15819
    },
    {
      "epoch": 0.8169377743351407,
      "grad_norm": 2.705754041671753,
      "learning_rate": 8.202278421780341e-07,
      "loss": 0.0347,
      "step": 15820
    },
    {
      "epoch": 0.8169894138910405,
      "grad_norm": 2.1416287422180176,
      "learning_rate": 8.197782341045829e-07,
      "loss": 0.0335,
      "step": 15821
    },
    {
      "epoch": 0.8170410534469403,
      "grad_norm": 9.027409553527832,
      "learning_rate": 8.193287382871246e-07,
      "loss": 0.0459,
      "step": 15822
    },
    {
      "epoch": 0.8170926930028402,
      "grad_norm": 11.364301681518555,
      "learning_rate": 8.188793547377278e-07,
      "loss": 0.0522,
      "step": 15823
    },
    {
      "epoch": 0.81714433255874,
      "grad_norm": 4.946651935577393,
      "learning_rate": 8.184300834684622e-07,
      "loss": 0.0577,
      "step": 15824
    },
    {
      "epoch": 0.8171959721146398,
      "grad_norm": 7.892877101898193,
      "learning_rate": 8.179809244913911e-07,
      "loss": 0.0815,
      "step": 15825
    },
    {
      "epoch": 0.8172476116705396,
      "grad_norm": 5.290426731109619,
      "learning_rate": 8.175318778185782e-07,
      "loss": 0.0538,
      "step": 15826
    },
    {
      "epoch": 0.8172992512264394,
      "grad_norm": 11.739627838134766,
      "learning_rate": 8.170829434620802e-07,
      "loss": 0.0604,
      "step": 15827
    },
    {
      "epoch": 0.8173508907823392,
      "grad_norm": 7.399687767028809,
      "learning_rate": 8.166341214339546e-07,
      "loss": 0.0747,
      "step": 15828
    },
    {
      "epoch": 0.8174025303382391,
      "grad_norm": 3.2501189708709717,
      "learning_rate": 8.161854117462526e-07,
      "loss": 0.0328,
      "step": 15829
    },
    {
      "epoch": 0.8174541698941389,
      "grad_norm": 4.2317705154418945,
      "learning_rate": 8.157368144110251e-07,
      "loss": 0.0614,
      "step": 15830
    },
    {
      "epoch": 0.8175058094500387,
      "grad_norm": 7.888247489929199,
      "learning_rate": 8.15288329440318e-07,
      "loss": 0.1408,
      "step": 15831
    },
    {
      "epoch": 0.8175574490059385,
      "grad_norm": 1.8924267292022705,
      "learning_rate": 8.148399568461751e-07,
      "loss": 0.0394,
      "step": 15832
    },
    {
      "epoch": 0.8176090885618383,
      "grad_norm": 4.744052886962891,
      "learning_rate": 8.143916966406385e-07,
      "loss": 0.0643,
      "step": 15833
    },
    {
      "epoch": 0.8176607281177382,
      "grad_norm": 1.9212538003921509,
      "learning_rate": 8.139435488357423e-07,
      "loss": 0.0429,
      "step": 15834
    },
    {
      "epoch": 0.817712367673638,
      "grad_norm": 2.54266357421875,
      "learning_rate": 8.134955134435241e-07,
      "loss": 0.0518,
      "step": 15835
    },
    {
      "epoch": 0.8177640072295378,
      "grad_norm": 2.469082832336426,
      "learning_rate": 8.130475904760154e-07,
      "loss": 0.0473,
      "step": 15836
    },
    {
      "epoch": 0.8178156467854376,
      "grad_norm": 3.5845463275909424,
      "learning_rate": 8.125997799452435e-07,
      "loss": 0.0513,
      "step": 15837
    },
    {
      "epoch": 0.8178672863413374,
      "grad_norm": 12.88810920715332,
      "learning_rate": 8.121520818632339e-07,
      "loss": 0.1347,
      "step": 15838
    },
    {
      "epoch": 0.8179189258972372,
      "grad_norm": 3.3737566471099854,
      "learning_rate": 8.11704496242009e-07,
      "loss": 0.04,
      "step": 15839
    },
    {
      "epoch": 0.8179705654531371,
      "grad_norm": 4.648149013519287,
      "learning_rate": 8.112570230935907e-07,
      "loss": 0.0277,
      "step": 15840
    },
    {
      "epoch": 0.8180222050090369,
      "grad_norm": 3.2954649925231934,
      "learning_rate": 8.108096624299927e-07,
      "loss": 0.0402,
      "step": 15841
    },
    {
      "epoch": 0.8180738445649367,
      "grad_norm": 2.7658114433288574,
      "learning_rate": 8.103624142632299e-07,
      "loss": 0.0276,
      "step": 15842
    },
    {
      "epoch": 0.8181254841208365,
      "grad_norm": 2.9305613040924072,
      "learning_rate": 8.099152786053116e-07,
      "loss": 0.0321,
      "step": 15843
    },
    {
      "epoch": 0.8181771236767363,
      "grad_norm": 6.965827465057373,
      "learning_rate": 8.094682554682465e-07,
      "loss": 0.0608,
      "step": 15844
    },
    {
      "epoch": 0.8182287632326362,
      "grad_norm": 12.383682250976562,
      "learning_rate": 8.090213448640382e-07,
      "loss": 0.0935,
      "step": 15845
    },
    {
      "epoch": 0.818280402788536,
      "grad_norm": 4.231354236602783,
      "learning_rate": 8.085745468046879e-07,
      "loss": 0.0245,
      "step": 15846
    },
    {
      "epoch": 0.8183320423444358,
      "grad_norm": 11.505820274353027,
      "learning_rate": 8.081278613021959e-07,
      "loss": 0.0456,
      "step": 15847
    },
    {
      "epoch": 0.8183836819003356,
      "grad_norm": 8.44269847869873,
      "learning_rate": 8.076812883685537e-07,
      "loss": 0.0633,
      "step": 15848
    },
    {
      "epoch": 0.8184353214562354,
      "grad_norm": 2.9308927059173584,
      "learning_rate": 8.072348280157566e-07,
      "loss": 0.0582,
      "step": 15849
    },
    {
      "epoch": 0.8184869610121353,
      "grad_norm": 3.27126145362854,
      "learning_rate": 8.067884802557934e-07,
      "loss": 0.0347,
      "step": 15850
    },
    {
      "epoch": 0.8185386005680351,
      "grad_norm": 8.102213859558105,
      "learning_rate": 8.063422451006514e-07,
      "loss": 0.0959,
      "step": 15851
    },
    {
      "epoch": 0.8185902401239349,
      "grad_norm": 6.116328239440918,
      "learning_rate": 8.058961225623113e-07,
      "loss": 0.0636,
      "step": 15852
    },
    {
      "epoch": 0.8186418796798347,
      "grad_norm": 4.112905025482178,
      "learning_rate": 8.054501126527531e-07,
      "loss": 0.0546,
      "step": 15853
    },
    {
      "epoch": 0.8186935192357345,
      "grad_norm": 7.404200077056885,
      "learning_rate": 8.050042153839566e-07,
      "loss": 0.0621,
      "step": 15854
    },
    {
      "epoch": 0.8187451587916343,
      "grad_norm": 4.040617942810059,
      "learning_rate": 8.045584307678961e-07,
      "loss": 0.0211,
      "step": 15855
    },
    {
      "epoch": 0.8187967983475342,
      "grad_norm": 5.943145275115967,
      "learning_rate": 8.041127588165404e-07,
      "loss": 0.0436,
      "step": 15856
    },
    {
      "epoch": 0.818848437903434,
      "grad_norm": 33.33928298950195,
      "learning_rate": 8.036671995418583e-07,
      "loss": 0.1966,
      "step": 15857
    },
    {
      "epoch": 0.8189000774593338,
      "grad_norm": 2.6930837631225586,
      "learning_rate": 8.032217529558156e-07,
      "loss": 0.0444,
      "step": 15858
    },
    {
      "epoch": 0.8189517170152336,
      "grad_norm": 3.8812174797058105,
      "learning_rate": 8.027764190703735e-07,
      "loss": 0.0323,
      "step": 15859
    },
    {
      "epoch": 0.8190033565711334,
      "grad_norm": 7.909648895263672,
      "learning_rate": 8.023311978974918e-07,
      "loss": 0.0726,
      "step": 15860
    },
    {
      "epoch": 0.8190549961270333,
      "grad_norm": 1.6188172101974487,
      "learning_rate": 8.018860894491259e-07,
      "loss": 0.0215,
      "step": 15861
    },
    {
      "epoch": 0.8191066356829332,
      "grad_norm": 11.06074047088623,
      "learning_rate": 8.014410937372292e-07,
      "loss": 0.0769,
      "step": 15862
    },
    {
      "epoch": 0.819158275238833,
      "grad_norm": 2.4428091049194336,
      "learning_rate": 8.009962107737517e-07,
      "loss": 0.0368,
      "step": 15863
    },
    {
      "epoch": 0.8192099147947328,
      "grad_norm": 6.000176906585693,
      "learning_rate": 8.005514405706399e-07,
      "loss": 0.0556,
      "step": 15864
    },
    {
      "epoch": 0.8192615543506326,
      "grad_norm": 2.67758846282959,
      "learning_rate": 8.00106783139838e-07,
      "loss": 0.0342,
      "step": 15865
    },
    {
      "epoch": 0.8193131939065325,
      "grad_norm": 6.2651286125183105,
      "learning_rate": 7.996622384932878e-07,
      "loss": 0.0495,
      "step": 15866
    },
    {
      "epoch": 0.8193648334624323,
      "grad_norm": 2.8174173831939697,
      "learning_rate": 7.992178066429241e-07,
      "loss": 0.0437,
      "step": 15867
    },
    {
      "epoch": 0.8194164730183321,
      "grad_norm": 4.408777713775635,
      "learning_rate": 7.987734876006847e-07,
      "loss": 0.0421,
      "step": 15868
    },
    {
      "epoch": 0.8194681125742319,
      "grad_norm": 7.789888858795166,
      "learning_rate": 7.983292813785015e-07,
      "loss": 0.0339,
      "step": 15869
    },
    {
      "epoch": 0.8195197521301317,
      "grad_norm": 3.495579481124878,
      "learning_rate": 7.978851879883015e-07,
      "loss": 0.0304,
      "step": 15870
    },
    {
      "epoch": 0.8195713916860315,
      "grad_norm": 2.991641044616699,
      "learning_rate": 7.974412074420107e-07,
      "loss": 0.0424,
      "step": 15871
    },
    {
      "epoch": 0.8196230312419314,
      "grad_norm": 3.3588311672210693,
      "learning_rate": 7.969973397515507e-07,
      "loss": 0.0341,
      "step": 15872
    },
    {
      "epoch": 0.8196746707978312,
      "grad_norm": 1.943131685256958,
      "learning_rate": 7.965535849288453e-07,
      "loss": 0.0132,
      "step": 15873
    },
    {
      "epoch": 0.819726310353731,
      "grad_norm": 2.6428778171539307,
      "learning_rate": 7.961099429858072e-07,
      "loss": 0.0378,
      "step": 15874
    },
    {
      "epoch": 0.8197779499096308,
      "grad_norm": 3.4032976627349854,
      "learning_rate": 7.956664139343511e-07,
      "loss": 0.0217,
      "step": 15875
    },
    {
      "epoch": 0.8198295894655306,
      "grad_norm": 4.2589335441589355,
      "learning_rate": 7.952229977863879e-07,
      "loss": 0.0596,
      "step": 15876
    },
    {
      "epoch": 0.8198812290214305,
      "grad_norm": 4.333418369293213,
      "learning_rate": 7.94779694553825e-07,
      "loss": 0.0692,
      "step": 15877
    },
    {
      "epoch": 0.8199328685773303,
      "grad_norm": 6.046959400177002,
      "learning_rate": 7.943365042485668e-07,
      "loss": 0.1092,
      "step": 15878
    },
    {
      "epoch": 0.8199845081332301,
      "grad_norm": 3.432230234146118,
      "learning_rate": 7.93893426882515e-07,
      "loss": 0.034,
      "step": 15879
    },
    {
      "epoch": 0.8200361476891299,
      "grad_norm": 4.342194080352783,
      "learning_rate": 7.93450462467567e-07,
      "loss": 0.0608,
      "step": 15880
    },
    {
      "epoch": 0.8200877872450297,
      "grad_norm": 3.8028972148895264,
      "learning_rate": 7.930076110156199e-07,
      "loss": 0.0568,
      "step": 15881
    },
    {
      "epoch": 0.8201394268009295,
      "grad_norm": 13.275093078613281,
      "learning_rate": 7.925648725385648e-07,
      "loss": 0.1205,
      "step": 15882
    },
    {
      "epoch": 0.8201910663568294,
      "grad_norm": 3.048633337020874,
      "learning_rate": 7.921222470482914e-07,
      "loss": 0.0504,
      "step": 15883
    },
    {
      "epoch": 0.8202427059127292,
      "grad_norm": 3.722191095352173,
      "learning_rate": 7.916797345566874e-07,
      "loss": 0.0265,
      "step": 15884
    },
    {
      "epoch": 0.820294345468629,
      "grad_norm": 15.600488662719727,
      "learning_rate": 7.912373350756336e-07,
      "loss": 0.0598,
      "step": 15885
    },
    {
      "epoch": 0.8203459850245288,
      "grad_norm": 3.772050380706787,
      "learning_rate": 7.907950486170097e-07,
      "loss": 0.0555,
      "step": 15886
    },
    {
      "epoch": 0.8203976245804286,
      "grad_norm": 1.6500623226165771,
      "learning_rate": 7.903528751926959e-07,
      "loss": 0.0097,
      "step": 15887
    },
    {
      "epoch": 0.8204492641363285,
      "grad_norm": 2.465064764022827,
      "learning_rate": 7.899108148145662e-07,
      "loss": 0.033,
      "step": 15888
    },
    {
      "epoch": 0.8205009036922283,
      "grad_norm": 6.396477699279785,
      "learning_rate": 7.894688674944894e-07,
      "loss": 0.0419,
      "step": 15889
    },
    {
      "epoch": 0.8205525432481281,
      "grad_norm": 1.892883539199829,
      "learning_rate": 7.890270332443345e-07,
      "loss": 0.0379,
      "step": 15890
    },
    {
      "epoch": 0.8206041828040279,
      "grad_norm": 2.169041156768799,
      "learning_rate": 7.885853120759668e-07,
      "loss": 0.0404,
      "step": 15891
    },
    {
      "epoch": 0.8206558223599277,
      "grad_norm": 2.030895233154297,
      "learning_rate": 7.881437040012485e-07,
      "loss": 0.0273,
      "step": 15892
    },
    {
      "epoch": 0.8207074619158276,
      "grad_norm": 2.7615716457366943,
      "learning_rate": 7.87702209032038e-07,
      "loss": 0.0535,
      "step": 15893
    },
    {
      "epoch": 0.8207591014717274,
      "grad_norm": 2.554607629776001,
      "learning_rate": 7.872608271801918e-07,
      "loss": 0.0249,
      "step": 15894
    },
    {
      "epoch": 0.8208107410276272,
      "grad_norm": 1.999000072479248,
      "learning_rate": 7.868195584575622e-07,
      "loss": 0.0468,
      "step": 15895
    },
    {
      "epoch": 0.820862380583527,
      "grad_norm": 3.333674669265747,
      "learning_rate": 7.863784028759997e-07,
      "loss": 0.0572,
      "step": 15896
    },
    {
      "epoch": 0.8209140201394268,
      "grad_norm": 3.4091594219207764,
      "learning_rate": 7.859373604473508e-07,
      "loss": 0.0604,
      "step": 15897
    },
    {
      "epoch": 0.8209656596953266,
      "grad_norm": 2.952117681503296,
      "learning_rate": 7.854964311834595e-07,
      "loss": 0.0335,
      "step": 15898
    },
    {
      "epoch": 0.8210172992512265,
      "grad_norm": 5.958659648895264,
      "learning_rate": 7.850556150961663e-07,
      "loss": 0.1042,
      "step": 15899
    },
    {
      "epoch": 0.8210689388071263,
      "grad_norm": 3.409653902053833,
      "learning_rate": 7.84614912197309e-07,
      "loss": 0.041,
      "step": 15900
    },
    {
      "epoch": 0.8211205783630261,
      "grad_norm": 9.239862442016602,
      "learning_rate": 7.841743224987225e-07,
      "loss": 0.1064,
      "step": 15901
    },
    {
      "epoch": 0.8211722179189259,
      "grad_norm": 2.7890779972076416,
      "learning_rate": 7.83733846012239e-07,
      "loss": 0.0214,
      "step": 15902
    },
    {
      "epoch": 0.8212238574748257,
      "grad_norm": 3.108499765396118,
      "learning_rate": 7.832934827496852e-07,
      "loss": 0.031,
      "step": 15903
    },
    {
      "epoch": 0.8212754970307256,
      "grad_norm": 2.839223861694336,
      "learning_rate": 7.828532327228877e-07,
      "loss": 0.0471,
      "step": 15904
    },
    {
      "epoch": 0.8213271365866254,
      "grad_norm": 4.873679161071777,
      "learning_rate": 7.824130959436676e-07,
      "loss": 0.0275,
      "step": 15905
    },
    {
      "epoch": 0.8213787761425252,
      "grad_norm": 3.7804982662200928,
      "learning_rate": 7.819730724238489e-07,
      "loss": 0.0315,
      "step": 15906
    },
    {
      "epoch": 0.821430415698425,
      "grad_norm": 4.890108585357666,
      "learning_rate": 7.815331621752431e-07,
      "loss": 0.035,
      "step": 15907
    },
    {
      "epoch": 0.8214820552543248,
      "grad_norm": 2.8061835765838623,
      "learning_rate": 7.810933652096658e-07,
      "loss": 0.0629,
      "step": 15908
    },
    {
      "epoch": 0.8215336948102246,
      "grad_norm": 8.11593246459961,
      "learning_rate": 7.806536815389265e-07,
      "loss": 0.066,
      "step": 15909
    },
    {
      "epoch": 0.8215853343661245,
      "grad_norm": 7.013167858123779,
      "learning_rate": 7.802141111748334e-07,
      "loss": 0.0402,
      "step": 15910
    },
    {
      "epoch": 0.8216369739220243,
      "grad_norm": 5.254118919372559,
      "learning_rate": 7.797746541291901e-07,
      "loss": 0.0538,
      "step": 15911
    },
    {
      "epoch": 0.8216886134779241,
      "grad_norm": 4.635472774505615,
      "learning_rate": 7.793353104137979e-07,
      "loss": 0.1385,
      "step": 15912
    },
    {
      "epoch": 0.8217402530338239,
      "grad_norm": 9.776246070861816,
      "learning_rate": 7.788960800404555e-07,
      "loss": 0.098,
      "step": 15913
    },
    {
      "epoch": 0.8217918925897237,
      "grad_norm": 4.227450370788574,
      "learning_rate": 7.784569630209571e-07,
      "loss": 0.0574,
      "step": 15914
    },
    {
      "epoch": 0.8218435321456236,
      "grad_norm": 2.5663869380950928,
      "learning_rate": 7.780179593670955e-07,
      "loss": 0.0208,
      "step": 15915
    },
    {
      "epoch": 0.8218951717015234,
      "grad_norm": 3.395419120788574,
      "learning_rate": 7.775790690906598e-07,
      "loss": 0.0468,
      "step": 15916
    },
    {
      "epoch": 0.8219468112574232,
      "grad_norm": 3.536308765411377,
      "learning_rate": 7.771402922034365e-07,
      "loss": 0.0206,
      "step": 15917
    },
    {
      "epoch": 0.821998450813323,
      "grad_norm": 2.5387744903564453,
      "learning_rate": 7.767016287172058e-07,
      "loss": 0.028,
      "step": 15918
    },
    {
      "epoch": 0.8220500903692228,
      "grad_norm": 3.4793970584869385,
      "learning_rate": 7.762630786437508e-07,
      "loss": 0.0417,
      "step": 15919
    },
    {
      "epoch": 0.8221017299251226,
      "grad_norm": 2.7635598182678223,
      "learning_rate": 7.758246419948467e-07,
      "loss": 0.0269,
      "step": 15920
    },
    {
      "epoch": 0.8221533694810225,
      "grad_norm": 3.7670671939849854,
      "learning_rate": 7.753863187822691e-07,
      "loss": 0.0303,
      "step": 15921
    },
    {
      "epoch": 0.8222050090369223,
      "grad_norm": 3.2467970848083496,
      "learning_rate": 7.749481090177869e-07,
      "loss": 0.0378,
      "step": 15922
    },
    {
      "epoch": 0.8222566485928221,
      "grad_norm": 2.265260934829712,
      "learning_rate": 7.745100127131672e-07,
      "loss": 0.0421,
      "step": 15923
    },
    {
      "epoch": 0.8223082881487219,
      "grad_norm": 8.783571243286133,
      "learning_rate": 7.740720298801768e-07,
      "loss": 0.0763,
      "step": 15924
    },
    {
      "epoch": 0.8223599277046217,
      "grad_norm": 2.4531257152557373,
      "learning_rate": 7.736341605305759e-07,
      "loss": 0.0182,
      "step": 15925
    },
    {
      "epoch": 0.8224115672605216,
      "grad_norm": 3.0463075637817383,
      "learning_rate": 7.731964046761231e-07,
      "loss": 0.0266,
      "step": 15926
    },
    {
      "epoch": 0.8224632068164214,
      "grad_norm": 1.5859510898590088,
      "learning_rate": 7.727587623285748e-07,
      "loss": 0.0149,
      "step": 15927
    },
    {
      "epoch": 0.8225148463723212,
      "grad_norm": 5.056273460388184,
      "learning_rate": 7.723212334996827e-07,
      "loss": 0.0614,
      "step": 15928
    },
    {
      "epoch": 0.822566485928221,
      "grad_norm": 17.228580474853516,
      "learning_rate": 7.718838182011973e-07,
      "loss": 0.2011,
      "step": 15929
    },
    {
      "epoch": 0.8226181254841208,
      "grad_norm": 2.900439500808716,
      "learning_rate": 7.714465164448637e-07,
      "loss": 0.0295,
      "step": 15930
    },
    {
      "epoch": 0.8226697650400207,
      "grad_norm": 4.322145938873291,
      "learning_rate": 7.710093282424252e-07,
      "loss": 0.0468,
      "step": 15931
    },
    {
      "epoch": 0.8227214045959205,
      "grad_norm": 16.22789192199707,
      "learning_rate": 7.705722536056237e-07,
      "loss": 0.1168,
      "step": 15932
    },
    {
      "epoch": 0.8227730441518203,
      "grad_norm": 3.0872344970703125,
      "learning_rate": 7.701352925461946e-07,
      "loss": 0.0285,
      "step": 15933
    },
    {
      "epoch": 0.8228246837077201,
      "grad_norm": 13.537806510925293,
      "learning_rate": 7.696984450758732e-07,
      "loss": 0.091,
      "step": 15934
    },
    {
      "epoch": 0.8228763232636199,
      "grad_norm": 1.4689993858337402,
      "learning_rate": 7.692617112063915e-07,
      "loss": 0.0158,
      "step": 15935
    },
    {
      "epoch": 0.8229279628195197,
      "grad_norm": 4.486350059509277,
      "learning_rate": 7.688250909494749e-07,
      "loss": 0.0245,
      "step": 15936
    },
    {
      "epoch": 0.8229796023754196,
      "grad_norm": 2.9538867473602295,
      "learning_rate": 7.683885843168504e-07,
      "loss": 0.0367,
      "step": 15937
    },
    {
      "epoch": 0.8230312419313194,
      "grad_norm": 3.5857455730438232,
      "learning_rate": 7.679521913202382e-07,
      "loss": 0.0547,
      "step": 15938
    },
    {
      "epoch": 0.8230828814872192,
      "grad_norm": 3.256047010421753,
      "learning_rate": 7.675159119713609e-07,
      "loss": 0.049,
      "step": 15939
    },
    {
      "epoch": 0.823134521043119,
      "grad_norm": 3.711271047592163,
      "learning_rate": 7.670797462819307e-07,
      "loss": 0.0479,
      "step": 15940
    },
    {
      "epoch": 0.8231861605990188,
      "grad_norm": 2.2087905406951904,
      "learning_rate": 7.666436942636618e-07,
      "loss": 0.0243,
      "step": 15941
    },
    {
      "epoch": 0.8232378001549187,
      "grad_norm": 3.224104166030884,
      "learning_rate": 7.662077559282638e-07,
      "loss": 0.0379,
      "step": 15942
    },
    {
      "epoch": 0.8232894397108185,
      "grad_norm": 2.907975435256958,
      "learning_rate": 7.657719312874435e-07,
      "loss": 0.0414,
      "step": 15943
    },
    {
      "epoch": 0.8233410792667183,
      "grad_norm": 2.3491525650024414,
      "learning_rate": 7.653362203529046e-07,
      "loss": 0.0243,
      "step": 15944
    },
    {
      "epoch": 0.8233927188226181,
      "grad_norm": 12.647639274597168,
      "learning_rate": 7.649006231363482e-07,
      "loss": 0.1431,
      "step": 15945
    },
    {
      "epoch": 0.8234443583785179,
      "grad_norm": 3.1799044609069824,
      "learning_rate": 7.644651396494707e-07,
      "loss": 0.0433,
      "step": 15946
    },
    {
      "epoch": 0.8234959979344177,
      "grad_norm": 4.969812870025635,
      "learning_rate": 7.640297699039679e-07,
      "loss": 0.0514,
      "step": 15947
    },
    {
      "epoch": 0.8235476374903176,
      "grad_norm": 2.916520595550537,
      "learning_rate": 7.635945139115303e-07,
      "loss": 0.01,
      "step": 15948
    },
    {
      "epoch": 0.8235992770462174,
      "grad_norm": 9.59986400604248,
      "learning_rate": 7.631593716838465e-07,
      "loss": 0.1306,
      "step": 15949
    },
    {
      "epoch": 0.8236509166021172,
      "grad_norm": 2.7523066997528076,
      "learning_rate": 7.627243432326037e-07,
      "loss": 0.0329,
      "step": 15950
    },
    {
      "epoch": 0.823702556158017,
      "grad_norm": 2.205271005630493,
      "learning_rate": 7.622894285694793e-07,
      "loss": 0.0376,
      "step": 15951
    },
    {
      "epoch": 0.8237541957139168,
      "grad_norm": 4.409091949462891,
      "learning_rate": 7.618546277061573e-07,
      "loss": 0.0357,
      "step": 15952
    },
    {
      "epoch": 0.8238058352698167,
      "grad_norm": 29.345840454101562,
      "learning_rate": 7.614199406543127e-07,
      "loss": 0.1491,
      "step": 15953
    },
    {
      "epoch": 0.8238574748257165,
      "grad_norm": 4.452858924865723,
      "learning_rate": 7.60985367425619e-07,
      "loss": 0.0332,
      "step": 15954
    },
    {
      "epoch": 0.8239091143816163,
      "grad_norm": 5.209815979003906,
      "learning_rate": 7.605509080317441e-07,
      "loss": 0.0653,
      "step": 15955
    },
    {
      "epoch": 0.8239607539375161,
      "grad_norm": 2.0751752853393555,
      "learning_rate": 7.601165624843565e-07,
      "loss": 0.0431,
      "step": 15956
    },
    {
      "epoch": 0.8240123934934159,
      "grad_norm": 10.670480728149414,
      "learning_rate": 7.596823307951201e-07,
      "loss": 0.1536,
      "step": 15957
    },
    {
      "epoch": 0.8240640330493157,
      "grad_norm": 21.58632469177246,
      "learning_rate": 7.592482129756956e-07,
      "loss": 0.0929,
      "step": 15958
    },
    {
      "epoch": 0.8241156726052156,
      "grad_norm": 4.928469181060791,
      "learning_rate": 7.58814209037741e-07,
      "loss": 0.0812,
      "step": 15959
    },
    {
      "epoch": 0.8241673121611154,
      "grad_norm": 3.7923789024353027,
      "learning_rate": 7.583803189929106e-07,
      "loss": 0.0516,
      "step": 15960
    },
    {
      "epoch": 0.8242189517170152,
      "grad_norm": 6.584341049194336,
      "learning_rate": 7.579465428528571e-07,
      "loss": 0.0522,
      "step": 15961
    },
    {
      "epoch": 0.824270591272915,
      "grad_norm": 3.9276833534240723,
      "learning_rate": 7.575128806292281e-07,
      "loss": 0.0419,
      "step": 15962
    },
    {
      "epoch": 0.8243222308288148,
      "grad_norm": 3.2679412364959717,
      "learning_rate": 7.5707933233367e-07,
      "loss": 0.0433,
      "step": 15963
    },
    {
      "epoch": 0.8243738703847147,
      "grad_norm": 10.643651008605957,
      "learning_rate": 7.566458979778246e-07,
      "loss": 0.2592,
      "step": 15964
    },
    {
      "epoch": 0.8244255099406145,
      "grad_norm": 4.611304759979248,
      "learning_rate": 7.562125775733325e-07,
      "loss": 0.0415,
      "step": 15965
    },
    {
      "epoch": 0.8244771494965143,
      "grad_norm": 9.885826110839844,
      "learning_rate": 7.55779371131829e-07,
      "loss": 0.0898,
      "step": 15966
    },
    {
      "epoch": 0.8245287890524141,
      "grad_norm": 1.6775981187820435,
      "learning_rate": 7.553462786649479e-07,
      "loss": 0.018,
      "step": 15967
    },
    {
      "epoch": 0.8245804286083139,
      "grad_norm": 8.953819274902344,
      "learning_rate": 7.549133001843207e-07,
      "loss": 0.0684,
      "step": 15968
    },
    {
      "epoch": 0.8246320681642137,
      "grad_norm": 3.3966317176818848,
      "learning_rate": 7.544804357015723e-07,
      "loss": 0.0403,
      "step": 15969
    },
    {
      "epoch": 0.8246837077201136,
      "grad_norm": 3.994260311126709,
      "learning_rate": 7.540476852283263e-07,
      "loss": 0.0423,
      "step": 15970
    },
    {
      "epoch": 0.8247353472760134,
      "grad_norm": 3.3879616260528564,
      "learning_rate": 7.536150487762073e-07,
      "loss": 0.0547,
      "step": 15971
    },
    {
      "epoch": 0.8247869868319132,
      "grad_norm": 2.4207985401153564,
      "learning_rate": 7.531825263568321e-07,
      "loss": 0.0298,
      "step": 15972
    },
    {
      "epoch": 0.824838626387813,
      "grad_norm": 3.8595776557922363,
      "learning_rate": 7.527501179818148e-07,
      "loss": 0.0497,
      "step": 15973
    },
    {
      "epoch": 0.8248902659437128,
      "grad_norm": 3.674280881881714,
      "learning_rate": 7.523178236627671e-07,
      "loss": 0.0845,
      "step": 15974
    },
    {
      "epoch": 0.8249419054996127,
      "grad_norm": 9.571490287780762,
      "learning_rate": 7.518856434112981e-07,
      "loss": 0.1655,
      "step": 15975
    },
    {
      "epoch": 0.8249935450555125,
      "grad_norm": 3.200120449066162,
      "learning_rate": 7.514535772390147e-07,
      "loss": 0.0327,
      "step": 15976
    },
    {
      "epoch": 0.8250451846114123,
      "grad_norm": 2.140427589416504,
      "learning_rate": 7.510216251575187e-07,
      "loss": 0.0207,
      "step": 15977
    },
    {
      "epoch": 0.8250968241673121,
      "grad_norm": 13.054253578186035,
      "learning_rate": 7.505897871784096e-07,
      "loss": 0.0725,
      "step": 15978
    },
    {
      "epoch": 0.8251484637232119,
      "grad_norm": 8.180121421813965,
      "learning_rate": 7.501580633132843e-07,
      "loss": 0.0427,
      "step": 15979
    },
    {
      "epoch": 0.8252001032791118,
      "grad_norm": 4.139436721801758,
      "learning_rate": 7.49726453573737e-07,
      "loss": 0.0283,
      "step": 15980
    },
    {
      "epoch": 0.8252517428350116,
      "grad_norm": 3.89548921585083,
      "learning_rate": 7.49294957971357e-07,
      "loss": 0.0428,
      "step": 15981
    },
    {
      "epoch": 0.8253033823909114,
      "grad_norm": 2.191593885421753,
      "learning_rate": 7.488635765177326e-07,
      "loss": 0.0271,
      "step": 15982
    },
    {
      "epoch": 0.8253550219468112,
      "grad_norm": 11.467832565307617,
      "learning_rate": 7.484323092244489e-07,
      "loss": 0.2113,
      "step": 15983
    },
    {
      "epoch": 0.825406661502711,
      "grad_norm": 4.982699871063232,
      "learning_rate": 7.480011561030836e-07,
      "loss": 0.042,
      "step": 15984
    },
    {
      "epoch": 0.8254583010586108,
      "grad_norm": 6.237845420837402,
      "learning_rate": 7.475701171652189e-07,
      "loss": 0.1279,
      "step": 15985
    },
    {
      "epoch": 0.8255099406145107,
      "grad_norm": 4.161934852600098,
      "learning_rate": 7.471391924224287e-07,
      "loss": 0.0486,
      "step": 15986
    },
    {
      "epoch": 0.8255615801704105,
      "grad_norm": 3.0852863788604736,
      "learning_rate": 7.467083818862853e-07,
      "loss": 0.0311,
      "step": 15987
    },
    {
      "epoch": 0.8256132197263103,
      "grad_norm": 3.2078428268432617,
      "learning_rate": 7.462776855683568e-07,
      "loss": 0.0449,
      "step": 15988
    },
    {
      "epoch": 0.8256648592822102,
      "grad_norm": 3.4847896099090576,
      "learning_rate": 7.458471034802084e-07,
      "loss": 0.0651,
      "step": 15989
    },
    {
      "epoch": 0.82571649883811,
      "grad_norm": 8.234399795532227,
      "learning_rate": 7.454166356334064e-07,
      "loss": 0.0657,
      "step": 15990
    },
    {
      "epoch": 0.8257681383940099,
      "grad_norm": 3.9902026653289795,
      "learning_rate": 7.449862820395071e-07,
      "loss": 0.0369,
      "step": 15991
    },
    {
      "epoch": 0.8258197779499097,
      "grad_norm": 5.003704071044922,
      "learning_rate": 7.44556042710069e-07,
      "loss": 0.0452,
      "step": 15992
    },
    {
      "epoch": 0.8258714175058095,
      "grad_norm": 3.999995231628418,
      "learning_rate": 7.441259176566451e-07,
      "loss": 0.0461,
      "step": 15993
    },
    {
      "epoch": 0.8259230570617093,
      "grad_norm": 2.3268237113952637,
      "learning_rate": 7.436959068907867e-07,
      "loss": 0.0364,
      "step": 15994
    },
    {
      "epoch": 0.8259746966176091,
      "grad_norm": 2.8246068954467773,
      "learning_rate": 7.432660104240408e-07,
      "loss": 0.0267,
      "step": 15995
    },
    {
      "epoch": 0.826026336173509,
      "grad_norm": 4.852828502655029,
      "learning_rate": 7.428362282679513e-07,
      "loss": 0.0395,
      "step": 15996
    },
    {
      "epoch": 0.8260779757294088,
      "grad_norm": 12.435331344604492,
      "learning_rate": 7.424065604340613e-07,
      "loss": 0.1527,
      "step": 15997
    },
    {
      "epoch": 0.8261296152853086,
      "grad_norm": 7.262529373168945,
      "learning_rate": 7.419770069339077e-07,
      "loss": 0.052,
      "step": 15998
    },
    {
      "epoch": 0.8261812548412084,
      "grad_norm": 4.015016555786133,
      "learning_rate": 7.415475677790262e-07,
      "loss": 0.04,
      "step": 15999
    },
    {
      "epoch": 0.8262328943971082,
      "grad_norm": 4.171414852142334,
      "learning_rate": 7.411182429809488e-07,
      "loss": 0.037,
      "step": 16000
    },
    {
      "epoch": 0.826284533953008,
      "grad_norm": 2.91766619682312,
      "learning_rate": 7.406890325512061e-07,
      "loss": 0.0343,
      "step": 16001
    },
    {
      "epoch": 0.8263361735089079,
      "grad_norm": 9.326929092407227,
      "learning_rate": 7.402599365013219e-07,
      "loss": 0.084,
      "step": 16002
    },
    {
      "epoch": 0.8263878130648077,
      "grad_norm": 1.8255606889724731,
      "learning_rate": 7.39830954842819e-07,
      "loss": 0.0252,
      "step": 16003
    },
    {
      "epoch": 0.8264394526207075,
      "grad_norm": 13.945653915405273,
      "learning_rate": 7.394020875872194e-07,
      "loss": 0.0909,
      "step": 16004
    },
    {
      "epoch": 0.8264910921766073,
      "grad_norm": 4.7288336753845215,
      "learning_rate": 7.389733347460403e-07,
      "loss": 0.0628,
      "step": 16005
    },
    {
      "epoch": 0.8265427317325071,
      "grad_norm": 2.155984878540039,
      "learning_rate": 7.38544696330793e-07,
      "loss": 0.0304,
      "step": 16006
    },
    {
      "epoch": 0.826594371288407,
      "grad_norm": 3.166496753692627,
      "learning_rate": 7.381161723529895e-07,
      "loss": 0.0517,
      "step": 16007
    },
    {
      "epoch": 0.8266460108443068,
      "grad_norm": 5.85740852355957,
      "learning_rate": 7.376877628241358e-07,
      "loss": 0.081,
      "step": 16008
    },
    {
      "epoch": 0.8266976504002066,
      "grad_norm": 22.042463302612305,
      "learning_rate": 7.372594677557404e-07,
      "loss": 0.169,
      "step": 16009
    },
    {
      "epoch": 0.8267492899561064,
      "grad_norm": 5.997056007385254,
      "learning_rate": 7.368312871593008e-07,
      "loss": 0.1121,
      "step": 16010
    },
    {
      "epoch": 0.8268009295120062,
      "grad_norm": 2.5055975914001465,
      "learning_rate": 7.364032210463168e-07,
      "loss": 0.0219,
      "step": 16011
    },
    {
      "epoch": 0.826852569067906,
      "grad_norm": 1.094441294670105,
      "learning_rate": 7.35975269428284e-07,
      "loss": 0.0079,
      "step": 16012
    },
    {
      "epoch": 0.8269042086238059,
      "grad_norm": 4.2442240715026855,
      "learning_rate": 7.355474323166945e-07,
      "loss": 0.0309,
      "step": 16013
    },
    {
      "epoch": 0.8269558481797057,
      "grad_norm": 3.3142359256744385,
      "learning_rate": 7.351197097230373e-07,
      "loss": 0.0488,
      "step": 16014
    },
    {
      "epoch": 0.8270074877356055,
      "grad_norm": 3.413276195526123,
      "learning_rate": 7.346921016587988e-07,
      "loss": 0.0418,
      "step": 16015
    },
    {
      "epoch": 0.8270591272915053,
      "grad_norm": 16.73316764831543,
      "learning_rate": 7.342646081354626e-07,
      "loss": 0.1569,
      "step": 16016
    },
    {
      "epoch": 0.8271107668474051,
      "grad_norm": 3.684248208999634,
      "learning_rate": 7.338372291645057e-07,
      "loss": 0.061,
      "step": 16017
    },
    {
      "epoch": 0.827162406403305,
      "grad_norm": 2.1117091178894043,
      "learning_rate": 7.334099647574084e-07,
      "loss": 0.0244,
      "step": 16018
    },
    {
      "epoch": 0.8272140459592048,
      "grad_norm": 3.8388984203338623,
      "learning_rate": 7.329828149256429e-07,
      "loss": 0.0449,
      "step": 16019
    },
    {
      "epoch": 0.8272656855151046,
      "grad_norm": 2.4598090648651123,
      "learning_rate": 7.325557796806815e-07,
      "loss": 0.0401,
      "step": 16020
    },
    {
      "epoch": 0.8273173250710044,
      "grad_norm": 3.853477954864502,
      "learning_rate": 7.321288590339898e-07,
      "loss": 0.0455,
      "step": 16021
    },
    {
      "epoch": 0.8273689646269042,
      "grad_norm": 4.676526069641113,
      "learning_rate": 7.317020529970314e-07,
      "loss": 0.0472,
      "step": 16022
    },
    {
      "epoch": 0.827420604182804,
      "grad_norm": 2.656486749649048,
      "learning_rate": 7.312753615812718e-07,
      "loss": 0.0294,
      "step": 16023
    },
    {
      "epoch": 0.8274722437387039,
      "grad_norm": 2.6597378253936768,
      "learning_rate": 7.308487847981666e-07,
      "loss": 0.03,
      "step": 16024
    },
    {
      "epoch": 0.8275238832946037,
      "grad_norm": 5.083072185516357,
      "learning_rate": 7.304223226591711e-07,
      "loss": 0.057,
      "step": 16025
    },
    {
      "epoch": 0.8275755228505035,
      "grad_norm": 12.408649444580078,
      "learning_rate": 7.29995975175738e-07,
      "loss": 0.0809,
      "step": 16026
    },
    {
      "epoch": 0.8276271624064033,
      "grad_norm": 1.1118468046188354,
      "learning_rate": 7.295697423593168e-07,
      "loss": 0.0167,
      "step": 16027
    },
    {
      "epoch": 0.8276788019623031,
      "grad_norm": 3.441505193710327,
      "learning_rate": 7.291436242213529e-07,
      "loss": 0.0295,
      "step": 16028
    },
    {
      "epoch": 0.827730441518203,
      "grad_norm": 1.9715837240219116,
      "learning_rate": 7.287176207732904e-07,
      "loss": 0.0436,
      "step": 16029
    },
    {
      "epoch": 0.8277820810741028,
      "grad_norm": 8.52976131439209,
      "learning_rate": 7.282917320265681e-07,
      "loss": 0.0405,
      "step": 16030
    },
    {
      "epoch": 0.8278337206300026,
      "grad_norm": 5.2736945152282715,
      "learning_rate": 7.278659579926234e-07,
      "loss": 0.0547,
      "step": 16031
    },
    {
      "epoch": 0.8278853601859024,
      "grad_norm": 11.456533432006836,
      "learning_rate": 7.274402986828899e-07,
      "loss": 0.0766,
      "step": 16032
    },
    {
      "epoch": 0.8279369997418022,
      "grad_norm": 1.47660493850708,
      "learning_rate": 7.270147541087986e-07,
      "loss": 0.0229,
      "step": 16033
    },
    {
      "epoch": 0.8279886392977021,
      "grad_norm": 2.5560646057128906,
      "learning_rate": 7.265893242817779e-07,
      "loss": 0.0287,
      "step": 16034
    },
    {
      "epoch": 0.8280402788536019,
      "grad_norm": 3.156555414199829,
      "learning_rate": 7.261640092132505e-07,
      "loss": 0.0498,
      "step": 16035
    },
    {
      "epoch": 0.8280919184095017,
      "grad_norm": 1.5908751487731934,
      "learning_rate": 7.257388089146366e-07,
      "loss": 0.01,
      "step": 16036
    },
    {
      "epoch": 0.8281435579654015,
      "grad_norm": 10.919655799865723,
      "learning_rate": 7.253137233973584e-07,
      "loss": 0.1755,
      "step": 16037
    },
    {
      "epoch": 0.8281951975213013,
      "grad_norm": 3.984496593475342,
      "learning_rate": 7.248887526728305e-07,
      "loss": 0.0538,
      "step": 16038
    },
    {
      "epoch": 0.8282468370772011,
      "grad_norm": 7.59521484375,
      "learning_rate": 7.244638967524631e-07,
      "loss": 0.0329,
      "step": 16039
    },
    {
      "epoch": 0.828298476633101,
      "grad_norm": 16.849388122558594,
      "learning_rate": 7.240391556476656e-07,
      "loss": 0.1275,
      "step": 16040
    },
    {
      "epoch": 0.8283501161890008,
      "grad_norm": 1.7615940570831299,
      "learning_rate": 7.236145293698437e-07,
      "loss": 0.0218,
      "step": 16041
    },
    {
      "epoch": 0.8284017557449006,
      "grad_norm": 6.616154670715332,
      "learning_rate": 7.231900179304036e-07,
      "loss": 0.041,
      "step": 16042
    },
    {
      "epoch": 0.8284533953008004,
      "grad_norm": 7.13999605178833,
      "learning_rate": 7.227656213407419e-07,
      "loss": 0.0975,
      "step": 16043
    },
    {
      "epoch": 0.8285050348567002,
      "grad_norm": 3.7624638080596924,
      "learning_rate": 7.223413396122558e-07,
      "loss": 0.0519,
      "step": 16044
    },
    {
      "epoch": 0.8285566744126001,
      "grad_norm": 9.46100902557373,
      "learning_rate": 7.219171727563401e-07,
      "loss": 0.1163,
      "step": 16045
    },
    {
      "epoch": 0.8286083139684999,
      "grad_norm": 2.8970589637756348,
      "learning_rate": 7.214931207843845e-07,
      "loss": 0.0235,
      "step": 16046
    },
    {
      "epoch": 0.8286599535243997,
      "grad_norm": 4.867334842681885,
      "learning_rate": 7.210691837077771e-07,
      "loss": 0.046,
      "step": 16047
    },
    {
      "epoch": 0.8287115930802995,
      "grad_norm": 3.75085186958313,
      "learning_rate": 7.206453615379016e-07,
      "loss": 0.0359,
      "step": 16048
    },
    {
      "epoch": 0.8287632326361993,
      "grad_norm": 3.5664918422698975,
      "learning_rate": 7.202216542861412e-07,
      "loss": 0.0679,
      "step": 16049
    },
    {
      "epoch": 0.8288148721920992,
      "grad_norm": 18.721572875976562,
      "learning_rate": 7.197980619638706e-07,
      "loss": 0.0668,
      "step": 16050
    },
    {
      "epoch": 0.828866511747999,
      "grad_norm": 10.54781723022461,
      "learning_rate": 7.193745845824684e-07,
      "loss": 0.1029,
      "step": 16051
    },
    {
      "epoch": 0.8289181513038988,
      "grad_norm": 8.827338218688965,
      "learning_rate": 7.189512221533051e-07,
      "loss": 0.0847,
      "step": 16052
    },
    {
      "epoch": 0.8289697908597986,
      "grad_norm": 4.443270206451416,
      "learning_rate": 7.185279746877516e-07,
      "loss": 0.0762,
      "step": 16053
    },
    {
      "epoch": 0.8290214304156984,
      "grad_norm": 3.3504726886749268,
      "learning_rate": 7.181048421971709e-07,
      "loss": 0.0367,
      "step": 16054
    },
    {
      "epoch": 0.8290730699715982,
      "grad_norm": 2.342125654220581,
      "learning_rate": 7.176818246929263e-07,
      "loss": 0.0418,
      "step": 16055
    },
    {
      "epoch": 0.8291247095274981,
      "grad_norm": 4.005025386810303,
      "learning_rate": 7.172589221863801e-07,
      "loss": 0.0904,
      "step": 16056
    },
    {
      "epoch": 0.8291763490833979,
      "grad_norm": 3.3528876304626465,
      "learning_rate": 7.168361346888869e-07,
      "loss": 0.0539,
      "step": 16057
    },
    {
      "epoch": 0.8292279886392977,
      "grad_norm": 2.766214370727539,
      "learning_rate": 7.164134622118008e-07,
      "loss": 0.0676,
      "step": 16058
    },
    {
      "epoch": 0.8292796281951975,
      "grad_norm": 2.8113789558410645,
      "learning_rate": 7.159909047664715e-07,
      "loss": 0.0318,
      "step": 16059
    },
    {
      "epoch": 0.8293312677510973,
      "grad_norm": 1.827032208442688,
      "learning_rate": 7.155684623642473e-07,
      "loss": 0.0275,
      "step": 16060
    },
    {
      "epoch": 0.8293829073069972,
      "grad_norm": 7.992491722106934,
      "learning_rate": 7.151461350164724e-07,
      "loss": 0.0399,
      "step": 16061
    },
    {
      "epoch": 0.829434546862897,
      "grad_norm": 5.561856746673584,
      "learning_rate": 7.147239227344882e-07,
      "loss": 0.0784,
      "step": 16062
    },
    {
      "epoch": 0.8294861864187968,
      "grad_norm": 21.058408737182617,
      "learning_rate": 7.143018255296318e-07,
      "loss": 0.2119,
      "step": 16063
    },
    {
      "epoch": 0.8295378259746966,
      "grad_norm": 6.484818935394287,
      "learning_rate": 7.138798434132393e-07,
      "loss": 0.0342,
      "step": 16064
    },
    {
      "epoch": 0.8295894655305964,
      "grad_norm": 13.788755416870117,
      "learning_rate": 7.134579763966426e-07,
      "loss": 0.1222,
      "step": 16065
    },
    {
      "epoch": 0.8296411050864962,
      "grad_norm": 3.4166488647460938,
      "learning_rate": 7.130362244911698e-07,
      "loss": 0.0273,
      "step": 16066
    },
    {
      "epoch": 0.8296927446423961,
      "grad_norm": 4.031702041625977,
      "learning_rate": 7.126145877081481e-07,
      "loss": 0.0608,
      "step": 16067
    },
    {
      "epoch": 0.8297443841982959,
      "grad_norm": 4.226919651031494,
      "learning_rate": 7.121930660588983e-07,
      "loss": 0.0705,
      "step": 16068
    },
    {
      "epoch": 0.8297960237541957,
      "grad_norm": 9.37195110321045,
      "learning_rate": 7.117716595547391e-07,
      "loss": 0.0741,
      "step": 16069
    },
    {
      "epoch": 0.8298476633100955,
      "grad_norm": 5.363297462463379,
      "learning_rate": 7.113503682069905e-07,
      "loss": 0.0417,
      "step": 16070
    },
    {
      "epoch": 0.8298993028659953,
      "grad_norm": 10.788708686828613,
      "learning_rate": 7.10929192026965e-07,
      "loss": 0.0697,
      "step": 16071
    },
    {
      "epoch": 0.8299509424218952,
      "grad_norm": 2.066976547241211,
      "learning_rate": 7.10508131025971e-07,
      "loss": 0.0297,
      "step": 16072
    },
    {
      "epoch": 0.830002581977795,
      "grad_norm": 2.6624321937561035,
      "learning_rate": 7.100871852153163e-07,
      "loss": 0.0497,
      "step": 16073
    },
    {
      "epoch": 0.8300542215336948,
      "grad_norm": 6.298370838165283,
      "learning_rate": 7.096663546063054e-07,
      "loss": 0.09,
      "step": 16074
    },
    {
      "epoch": 0.8301058610895946,
      "grad_norm": 8.088493347167969,
      "learning_rate": 7.092456392102398e-07,
      "loss": 0.0602,
      "step": 16075
    },
    {
      "epoch": 0.8301575006454944,
      "grad_norm": 4.232126235961914,
      "learning_rate": 7.088250390384166e-07,
      "loss": 0.0253,
      "step": 16076
    },
    {
      "epoch": 0.8302091402013942,
      "grad_norm": 2.7603750228881836,
      "learning_rate": 7.084045541021312e-07,
      "loss": 0.0531,
      "step": 16077
    },
    {
      "epoch": 0.8302607797572941,
      "grad_norm": 9.44559383392334,
      "learning_rate": 7.079841844126745e-07,
      "loss": 0.1069,
      "step": 16078
    },
    {
      "epoch": 0.8303124193131939,
      "grad_norm": 3.8863344192504883,
      "learning_rate": 7.075639299813364e-07,
      "loss": 0.0437,
      "step": 16079
    },
    {
      "epoch": 0.8303640588690937,
      "grad_norm": 13.248919486999512,
      "learning_rate": 7.071437908194013e-07,
      "loss": 0.1089,
      "step": 16080
    },
    {
      "epoch": 0.8304156984249935,
      "grad_norm": 2.1919960975646973,
      "learning_rate": 7.067237669381527e-07,
      "loss": 0.035,
      "step": 16081
    },
    {
      "epoch": 0.8304673379808933,
      "grad_norm": 1.683916449546814,
      "learning_rate": 7.063038583488701e-07,
      "loss": 0.0153,
      "step": 16082
    },
    {
      "epoch": 0.8305189775367932,
      "grad_norm": 6.0171051025390625,
      "learning_rate": 7.058840650628268e-07,
      "loss": 0.0363,
      "step": 16083
    },
    {
      "epoch": 0.830570617092693,
      "grad_norm": 8.101364135742188,
      "learning_rate": 7.054643870912992e-07,
      "loss": 0.0644,
      "step": 16084
    },
    {
      "epoch": 0.8306222566485928,
      "grad_norm": 2.7190535068511963,
      "learning_rate": 7.050448244455576e-07,
      "loss": 0.0673,
      "step": 16085
    },
    {
      "epoch": 0.8306738962044926,
      "grad_norm": 2.6754515171051025,
      "learning_rate": 7.046253771368666e-07,
      "loss": 0.0487,
      "step": 16086
    },
    {
      "epoch": 0.8307255357603924,
      "grad_norm": 9.21147632598877,
      "learning_rate": 7.042060451764915e-07,
      "loss": 0.2163,
      "step": 16087
    },
    {
      "epoch": 0.8307771753162922,
      "grad_norm": 3.17954421043396,
      "learning_rate": 7.037868285756916e-07,
      "loss": 0.0362,
      "step": 16088
    },
    {
      "epoch": 0.8308288148721921,
      "grad_norm": 3.584394693374634,
      "learning_rate": 7.033677273457279e-07,
      "loss": 0.059,
      "step": 16089
    },
    {
      "epoch": 0.8308804544280919,
      "grad_norm": 15.165169715881348,
      "learning_rate": 7.029487414978514e-07,
      "loss": 0.0691,
      "step": 16090
    },
    {
      "epoch": 0.8309320939839917,
      "grad_norm": 2.3033742904663086,
      "learning_rate": 7.025298710433153e-07,
      "loss": 0.0276,
      "step": 16091
    },
    {
      "epoch": 0.8309837335398915,
      "grad_norm": 10.220747947692871,
      "learning_rate": 7.021111159933674e-07,
      "loss": 0.091,
      "step": 16092
    },
    {
      "epoch": 0.8310353730957913,
      "grad_norm": 2.567990779876709,
      "learning_rate": 7.016924763592537e-07,
      "loss": 0.0266,
      "step": 16093
    },
    {
      "epoch": 0.8310870126516912,
      "grad_norm": 4.249425888061523,
      "learning_rate": 7.012739521522155e-07,
      "loss": 0.034,
      "step": 16094
    },
    {
      "epoch": 0.831138652207591,
      "grad_norm": 2.250133991241455,
      "learning_rate": 7.008555433834924e-07,
      "loss": 0.0349,
      "step": 16095
    },
    {
      "epoch": 0.8311902917634908,
      "grad_norm": 4.446926116943359,
      "learning_rate": 7.004372500643219e-07,
      "loss": 0.0564,
      "step": 16096
    },
    {
      "epoch": 0.8312419313193906,
      "grad_norm": 4.493507385253906,
      "learning_rate": 7.000190722059325e-07,
      "loss": 0.0495,
      "step": 16097
    },
    {
      "epoch": 0.8312935708752904,
      "grad_norm": 3.442084789276123,
      "learning_rate": 6.996010098195583e-07,
      "loss": 0.0212,
      "step": 16098
    },
    {
      "epoch": 0.8313452104311903,
      "grad_norm": 11.422043800354004,
      "learning_rate": 6.991830629164242e-07,
      "loss": 0.0756,
      "step": 16099
    },
    {
      "epoch": 0.8313968499870901,
      "grad_norm": 2.1340363025665283,
      "learning_rate": 6.987652315077553e-07,
      "loss": 0.0263,
      "step": 16100
    },
    {
      "epoch": 0.8314484895429899,
      "grad_norm": 2.260854482650757,
      "learning_rate": 6.983475156047692e-07,
      "loss": 0.0362,
      "step": 16101
    },
    {
      "epoch": 0.8315001290988897,
      "grad_norm": 6.8115668296813965,
      "learning_rate": 6.979299152186841e-07,
      "loss": 0.1089,
      "step": 16102
    },
    {
      "epoch": 0.8315517686547895,
      "grad_norm": 10.316089630126953,
      "learning_rate": 6.975124303607162e-07,
      "loss": 0.0559,
      "step": 16103
    },
    {
      "epoch": 0.8316034082106893,
      "grad_norm": 2.9962997436523438,
      "learning_rate": 6.970950610420768e-07,
      "loss": 0.0426,
      "step": 16104
    },
    {
      "epoch": 0.8316550477665892,
      "grad_norm": 7.388751029968262,
      "learning_rate": 6.96677807273971e-07,
      "loss": 0.04,
      "step": 16105
    },
    {
      "epoch": 0.831706687322489,
      "grad_norm": 4.246959686279297,
      "learning_rate": 6.96260669067606e-07,
      "loss": 0.0672,
      "step": 16106
    },
    {
      "epoch": 0.8317583268783888,
      "grad_norm": 2.2746105194091797,
      "learning_rate": 6.958436464341827e-07,
      "loss": 0.0304,
      "step": 16107
    },
    {
      "epoch": 0.8318099664342886,
      "grad_norm": 1.93830406665802,
      "learning_rate": 6.954267393849007e-07,
      "loss": 0.0358,
      "step": 16108
    },
    {
      "epoch": 0.8318616059901884,
      "grad_norm": 12.909368515014648,
      "learning_rate": 6.950099479309553e-07,
      "loss": 0.1108,
      "step": 16109
    },
    {
      "epoch": 0.8319132455460883,
      "grad_norm": 4.158115386962891,
      "learning_rate": 6.945932720835386e-07,
      "loss": 0.047,
      "step": 16110
    },
    {
      "epoch": 0.8319648851019881,
      "grad_norm": 2.6420462131500244,
      "learning_rate": 6.941767118538401e-07,
      "loss": 0.022,
      "step": 16111
    },
    {
      "epoch": 0.8320165246578879,
      "grad_norm": 3.5240187644958496,
      "learning_rate": 6.937602672530469e-07,
      "loss": 0.0412,
      "step": 16112
    },
    {
      "epoch": 0.8320681642137877,
      "grad_norm": 2.5538244247436523,
      "learning_rate": 6.933439382923418e-07,
      "loss": 0.0335,
      "step": 16113
    },
    {
      "epoch": 0.8321198037696875,
      "grad_norm": 5.314703464508057,
      "learning_rate": 6.929277249829047e-07,
      "loss": 0.0722,
      "step": 16114
    },
    {
      "epoch": 0.8321714433255873,
      "grad_norm": 2.1597065925598145,
      "learning_rate": 6.92511627335914e-07,
      "loss": 0.0273,
      "step": 16115
    },
    {
      "epoch": 0.8322230828814872,
      "grad_norm": 4.188483715057373,
      "learning_rate": 6.920956453625405e-07,
      "loss": 0.0554,
      "step": 16116
    },
    {
      "epoch": 0.8322747224373871,
      "grad_norm": 2.295591354370117,
      "learning_rate": 6.916797790739576e-07,
      "loss": 0.0166,
      "step": 16117
    },
    {
      "epoch": 0.8323263619932869,
      "grad_norm": 5.538080215454102,
      "learning_rate": 6.912640284813338e-07,
      "loss": 0.0956,
      "step": 16118
    },
    {
      "epoch": 0.8323780015491867,
      "grad_norm": 4.70001745223999,
      "learning_rate": 6.908483935958316e-07,
      "loss": 0.036,
      "step": 16119
    },
    {
      "epoch": 0.8324296411050865,
      "grad_norm": 8.813431739807129,
      "learning_rate": 6.904328744286127e-07,
      "loss": 0.0862,
      "step": 16120
    },
    {
      "epoch": 0.8324812806609864,
      "grad_norm": 2.149691581726074,
      "learning_rate": 6.900174709908347e-07,
      "loss": 0.0356,
      "step": 16121
    },
    {
      "epoch": 0.8325329202168862,
      "grad_norm": 2.184624433517456,
      "learning_rate": 6.896021832936562e-07,
      "loss": 0.0338,
      "step": 16122
    },
    {
      "epoch": 0.832584559772786,
      "grad_norm": 2.4634783267974854,
      "learning_rate": 6.891870113482269e-07,
      "loss": 0.0449,
      "step": 16123
    },
    {
      "epoch": 0.8326361993286858,
      "grad_norm": 2.7894322872161865,
      "learning_rate": 6.887719551656957e-07,
      "loss": 0.0439,
      "step": 16124
    },
    {
      "epoch": 0.8326878388845856,
      "grad_norm": 5.457032680511475,
      "learning_rate": 6.883570147572089e-07,
      "loss": 0.0733,
      "step": 16125
    },
    {
      "epoch": 0.8327394784404855,
      "grad_norm": 4.311967372894287,
      "learning_rate": 6.8794219013391e-07,
      "loss": 0.0499,
      "step": 16126
    },
    {
      "epoch": 0.8327911179963853,
      "grad_norm": 5.172555446624756,
      "learning_rate": 6.875274813069382e-07,
      "loss": 0.0533,
      "step": 16127
    },
    {
      "epoch": 0.8328427575522851,
      "grad_norm": 3.0928587913513184,
      "learning_rate": 6.871128882874301e-07,
      "loss": 0.0451,
      "step": 16128
    },
    {
      "epoch": 0.8328943971081849,
      "grad_norm": 6.999874114990234,
      "learning_rate": 6.866984110865205e-07,
      "loss": 0.1588,
      "step": 16129
    },
    {
      "epoch": 0.8329460366640847,
      "grad_norm": 6.354193210601807,
      "learning_rate": 6.862840497153361e-07,
      "loss": 0.0161,
      "step": 16130
    },
    {
      "epoch": 0.8329976762199846,
      "grad_norm": 3.886028289794922,
      "learning_rate": 6.858698041850081e-07,
      "loss": 0.0417,
      "step": 16131
    },
    {
      "epoch": 0.8330493157758844,
      "grad_norm": 10.573071479797363,
      "learning_rate": 6.854556745066593e-07,
      "loss": 0.0368,
      "step": 16132
    },
    {
      "epoch": 0.8331009553317842,
      "grad_norm": 6.036922454833984,
      "learning_rate": 6.850416606914112e-07,
      "loss": 0.0971,
      "step": 16133
    },
    {
      "epoch": 0.833152594887684,
      "grad_norm": 16.084884643554688,
      "learning_rate": 6.846277627503811e-07,
      "loss": 0.0225,
      "step": 16134
    },
    {
      "epoch": 0.8332042344435838,
      "grad_norm": 1.5548707246780396,
      "learning_rate": 6.842139806946824e-07,
      "loss": 0.0152,
      "step": 16135
    },
    {
      "epoch": 0.8332558739994836,
      "grad_norm": 2.8565914630889893,
      "learning_rate": 6.8380031453543e-07,
      "loss": 0.0315,
      "step": 16136
    },
    {
      "epoch": 0.8333075135553835,
      "grad_norm": 2.593815326690674,
      "learning_rate": 6.833867642837316e-07,
      "loss": 0.0274,
      "step": 16137
    },
    {
      "epoch": 0.8333591531112833,
      "grad_norm": 8.987530708312988,
      "learning_rate": 6.829733299506913e-07,
      "loss": 0.0865,
      "step": 16138
    },
    {
      "epoch": 0.8334107926671831,
      "grad_norm": 8.77574348449707,
      "learning_rate": 6.825600115474123e-07,
      "loss": 0.1354,
      "step": 16139
    },
    {
      "epoch": 0.8334624322230829,
      "grad_norm": 5.240549087524414,
      "learning_rate": 6.821468090849942e-07,
      "loss": 0.0918,
      "step": 16140
    },
    {
      "epoch": 0.8335140717789827,
      "grad_norm": 4.97177791595459,
      "learning_rate": 6.817337225745324e-07,
      "loss": 0.0514,
      "step": 16141
    },
    {
      "epoch": 0.8335657113348826,
      "grad_norm": 6.095090389251709,
      "learning_rate": 6.813207520271209e-07,
      "loss": 0.0966,
      "step": 16142
    },
    {
      "epoch": 0.8336173508907824,
      "grad_norm": 9.865385055541992,
      "learning_rate": 6.809078974538486e-07,
      "loss": 0.0983,
      "step": 16143
    },
    {
      "epoch": 0.8336689904466822,
      "grad_norm": 9.055872917175293,
      "learning_rate": 6.804951588658032e-07,
      "loss": 0.1355,
      "step": 16144
    },
    {
      "epoch": 0.833720630002582,
      "grad_norm": 8.217049598693848,
      "learning_rate": 6.800825362740681e-07,
      "loss": 0.0747,
      "step": 16145
    },
    {
      "epoch": 0.8337722695584818,
      "grad_norm": 3.502633571624756,
      "learning_rate": 6.79670029689724e-07,
      "loss": 0.0747,
      "step": 16146
    },
    {
      "epoch": 0.8338239091143816,
      "grad_norm": 6.968038558959961,
      "learning_rate": 6.792576391238481e-07,
      "loss": 0.024,
      "step": 16147
    },
    {
      "epoch": 0.8338755486702815,
      "grad_norm": 3.0600712299346924,
      "learning_rate": 6.788453645875164e-07,
      "loss": 0.0362,
      "step": 16148
    },
    {
      "epoch": 0.8339271882261813,
      "grad_norm": 14.68403434753418,
      "learning_rate": 6.784332060917965e-07,
      "loss": 0.0655,
      "step": 16149
    },
    {
      "epoch": 0.8339788277820811,
      "grad_norm": 8.150108337402344,
      "learning_rate": 6.780211636477596e-07,
      "loss": 0.0601,
      "step": 16150
    },
    {
      "epoch": 0.8340304673379809,
      "grad_norm": 2.900587797164917,
      "learning_rate": 6.776092372664711e-07,
      "loss": 0.0302,
      "step": 16151
    },
    {
      "epoch": 0.8340821068938807,
      "grad_norm": 1.526698112487793,
      "learning_rate": 6.771974269589904e-07,
      "loss": 0.0145,
      "step": 16152
    },
    {
      "epoch": 0.8341337464497806,
      "grad_norm": 3.009937047958374,
      "learning_rate": 6.767857327363775e-07,
      "loss": 0.0506,
      "step": 16153
    },
    {
      "epoch": 0.8341853860056804,
      "grad_norm": 3.453453779220581,
      "learning_rate": 6.763741546096875e-07,
      "loss": 0.0496,
      "step": 16154
    },
    {
      "epoch": 0.8342370255615802,
      "grad_norm": 7.3173604011535645,
      "learning_rate": 6.759626925899753e-07,
      "loss": 0.0519,
      "step": 16155
    },
    {
      "epoch": 0.83428866511748,
      "grad_norm": 2.8092219829559326,
      "learning_rate": 6.755513466882874e-07,
      "loss": 0.0377,
      "step": 16156
    },
    {
      "epoch": 0.8343403046733798,
      "grad_norm": 2.391552209854126,
      "learning_rate": 6.751401169156718e-07,
      "loss": 0.0276,
      "step": 16157
    },
    {
      "epoch": 0.8343919442292796,
      "grad_norm": 1.5719622373580933,
      "learning_rate": 6.74729003283171e-07,
      "loss": 0.0195,
      "step": 16158
    },
    {
      "epoch": 0.8344435837851795,
      "grad_norm": 5.676553249359131,
      "learning_rate": 6.74318005801825e-07,
      "loss": 0.0303,
      "step": 16159
    },
    {
      "epoch": 0.8344952233410793,
      "grad_norm": 22.19106674194336,
      "learning_rate": 6.739071244826712e-07,
      "loss": 0.0268,
      "step": 16160
    },
    {
      "epoch": 0.8345468628969791,
      "grad_norm": 4.460139751434326,
      "learning_rate": 6.734963593367439e-07,
      "loss": 0.0351,
      "step": 16161
    },
    {
      "epoch": 0.8345985024528789,
      "grad_norm": 1.8760725259780884,
      "learning_rate": 6.730857103750737e-07,
      "loss": 0.0192,
      "step": 16162
    },
    {
      "epoch": 0.8346501420087787,
      "grad_norm": 3.3409676551818848,
      "learning_rate": 6.726751776086854e-07,
      "loss": 0.0148,
      "step": 16163
    },
    {
      "epoch": 0.8347017815646786,
      "grad_norm": 14.38018798828125,
      "learning_rate": 6.722647610486072e-07,
      "loss": 0.1858,
      "step": 16164
    },
    {
      "epoch": 0.8347534211205784,
      "grad_norm": 5.512966156005859,
      "learning_rate": 6.718544607058586e-07,
      "loss": 0.0361,
      "step": 16165
    },
    {
      "epoch": 0.8348050606764782,
      "grad_norm": 1.8286247253417969,
      "learning_rate": 6.714442765914597e-07,
      "loss": 0.0281,
      "step": 16166
    },
    {
      "epoch": 0.834856700232378,
      "grad_norm": 1.7921133041381836,
      "learning_rate": 6.710342087164234e-07,
      "loss": 0.0402,
      "step": 16167
    },
    {
      "epoch": 0.8349083397882778,
      "grad_norm": 14.266253471374512,
      "learning_rate": 6.70624257091761e-07,
      "loss": 0.0456,
      "step": 16168
    },
    {
      "epoch": 0.8349599793441776,
      "grad_norm": 2.7622687816619873,
      "learning_rate": 6.70214421728484e-07,
      "loss": 0.0108,
      "step": 16169
    },
    {
      "epoch": 0.8350116189000775,
      "grad_norm": 4.278236389160156,
      "learning_rate": 6.698047026375982e-07,
      "loss": 0.0601,
      "step": 16170
    },
    {
      "epoch": 0.8350632584559773,
      "grad_norm": 2.7955222129821777,
      "learning_rate": 6.693950998301035e-07,
      "loss": 0.0254,
      "step": 16171
    },
    {
      "epoch": 0.8351148980118771,
      "grad_norm": 3.651247978210449,
      "learning_rate": 6.689856133170014e-07,
      "loss": 0.0704,
      "step": 16172
    },
    {
      "epoch": 0.8351665375677769,
      "grad_norm": 2.014040231704712,
      "learning_rate": 6.68576243109288e-07,
      "loss": 0.0309,
      "step": 16173
    },
    {
      "epoch": 0.8352181771236767,
      "grad_norm": 6.154754161834717,
      "learning_rate": 6.68166989217956e-07,
      "loss": 0.0673,
      "step": 16174
    },
    {
      "epoch": 0.8352698166795766,
      "grad_norm": 6.670032978057861,
      "learning_rate": 6.677578516539962e-07,
      "loss": 0.0985,
      "step": 16175
    },
    {
      "epoch": 0.8353214562354764,
      "grad_norm": 5.1128830909729,
      "learning_rate": 6.673488304283948e-07,
      "loss": 0.0358,
      "step": 16176
    },
    {
      "epoch": 0.8353730957913762,
      "grad_norm": 1.4817194938659668,
      "learning_rate": 6.66939925552137e-07,
      "loss": 0.0084,
      "step": 16177
    },
    {
      "epoch": 0.835424735347276,
      "grad_norm": 1.035889983177185,
      "learning_rate": 6.66531137036202e-07,
      "loss": 0.0132,
      "step": 16178
    },
    {
      "epoch": 0.8354763749031758,
      "grad_norm": 2.1491541862487793,
      "learning_rate": 6.661224648915687e-07,
      "loss": 0.0277,
      "step": 16179
    },
    {
      "epoch": 0.8355280144590757,
      "grad_norm": 5.4788713455200195,
      "learning_rate": 6.657139091292108e-07,
      "loss": 0.0654,
      "step": 16180
    },
    {
      "epoch": 0.8355796540149755,
      "grad_norm": 1.7605550289154053,
      "learning_rate": 6.653054697601014e-07,
      "loss": 0.0331,
      "step": 16181
    },
    {
      "epoch": 0.8356312935708753,
      "grad_norm": 6.237532615661621,
      "learning_rate": 6.648971467952048e-07,
      "loss": 0.0401,
      "step": 16182
    },
    {
      "epoch": 0.8356829331267751,
      "grad_norm": 9.761219024658203,
      "learning_rate": 6.644889402454901e-07,
      "loss": 0.0694,
      "step": 16183
    },
    {
      "epoch": 0.8357345726826749,
      "grad_norm": 2.5626678466796875,
      "learning_rate": 6.640808501219193e-07,
      "loss": 0.0558,
      "step": 16184
    },
    {
      "epoch": 0.8357862122385747,
      "grad_norm": 13.968823432922363,
      "learning_rate": 6.636728764354483e-07,
      "loss": 0.0588,
      "step": 16185
    },
    {
      "epoch": 0.8358378517944746,
      "grad_norm": 6.0334882736206055,
      "learning_rate": 6.632650191970347e-07,
      "loss": 0.0625,
      "step": 16186
    },
    {
      "epoch": 0.8358894913503744,
      "grad_norm": 2.2985172271728516,
      "learning_rate": 6.628572784176296e-07,
      "loss": 0.0381,
      "step": 16187
    },
    {
      "epoch": 0.8359411309062742,
      "grad_norm": 3.0485472679138184,
      "learning_rate": 6.624496541081866e-07,
      "loss": 0.0598,
      "step": 16188
    },
    {
      "epoch": 0.835992770462174,
      "grad_norm": 7.019683361053467,
      "learning_rate": 6.620421462796473e-07,
      "loss": 0.095,
      "step": 16189
    },
    {
      "epoch": 0.8360444100180738,
      "grad_norm": 7.231808662414551,
      "learning_rate": 6.616347549429574e-07,
      "loss": 0.0864,
      "step": 16190
    },
    {
      "epoch": 0.8360960495739737,
      "grad_norm": 1.7737253904342651,
      "learning_rate": 6.612274801090562e-07,
      "loss": 0.0223,
      "step": 16191
    },
    {
      "epoch": 0.8361476891298735,
      "grad_norm": 5.354597091674805,
      "learning_rate": 6.608203217888815e-07,
      "loss": 0.033,
      "step": 16192
    },
    {
      "epoch": 0.8361993286857733,
      "grad_norm": 3.8238017559051514,
      "learning_rate": 6.604132799933665e-07,
      "loss": 0.0605,
      "step": 16193
    },
    {
      "epoch": 0.8362509682416731,
      "grad_norm": 4.004965782165527,
      "learning_rate": 6.600063547334418e-07,
      "loss": 0.0881,
      "step": 16194
    },
    {
      "epoch": 0.8363026077975729,
      "grad_norm": 1.589770793914795,
      "learning_rate": 6.595995460200366e-07,
      "loss": 0.02,
      "step": 16195
    },
    {
      "epoch": 0.8363542473534727,
      "grad_norm": 4.5694074630737305,
      "learning_rate": 6.591928538640724e-07,
      "loss": 0.0491,
      "step": 16196
    },
    {
      "epoch": 0.8364058869093726,
      "grad_norm": 5.863138675689697,
      "learning_rate": 6.587862782764725e-07,
      "loss": 0.037,
      "step": 16197
    },
    {
      "epoch": 0.8364575264652724,
      "grad_norm": 3.3716342449188232,
      "learning_rate": 6.58379819268155e-07,
      "loss": 0.0683,
      "step": 16198
    },
    {
      "epoch": 0.8365091660211722,
      "grad_norm": 3.7088990211486816,
      "learning_rate": 6.579734768500362e-07,
      "loss": 0.0359,
      "step": 16199
    },
    {
      "epoch": 0.836560805577072,
      "grad_norm": 2.7119712829589844,
      "learning_rate": 6.575672510330256e-07,
      "loss": 0.0426,
      "step": 16200
    },
    {
      "epoch": 0.8366124451329718,
      "grad_norm": 3.4490551948547363,
      "learning_rate": 6.571611418280322e-07,
      "loss": 0.0566,
      "step": 16201
    },
    {
      "epoch": 0.8366640846888717,
      "grad_norm": 3.5897560119628906,
      "learning_rate": 6.567551492459628e-07,
      "loss": 0.0468,
      "step": 16202
    },
    {
      "epoch": 0.8367157242447715,
      "grad_norm": 6.751486301422119,
      "learning_rate": 6.563492732977217e-07,
      "loss": 0.0629,
      "step": 16203
    },
    {
      "epoch": 0.8367673638006713,
      "grad_norm": 9.967442512512207,
      "learning_rate": 6.559435139942049e-07,
      "loss": 0.04,
      "step": 16204
    },
    {
      "epoch": 0.8368190033565711,
      "grad_norm": 4.408215522766113,
      "learning_rate": 6.555378713463101e-07,
      "loss": 0.0797,
      "step": 16205
    },
    {
      "epoch": 0.8368706429124709,
      "grad_norm": 3.4716384410858154,
      "learning_rate": 6.551323453649305e-07,
      "loss": 0.0386,
      "step": 16206
    },
    {
      "epoch": 0.8369222824683707,
      "grad_norm": 3.262861490249634,
      "learning_rate": 6.547269360609565e-07,
      "loss": 0.0322,
      "step": 16207
    },
    {
      "epoch": 0.8369739220242706,
      "grad_norm": 2.54801082611084,
      "learning_rate": 6.543216434452743e-07,
      "loss": 0.0496,
      "step": 16208
    },
    {
      "epoch": 0.8370255615801704,
      "grad_norm": 3.1276731491088867,
      "learning_rate": 6.53916467528768e-07,
      "loss": 0.0354,
      "step": 16209
    },
    {
      "epoch": 0.8370772011360702,
      "grad_norm": 2.454402208328247,
      "learning_rate": 6.535114083223182e-07,
      "loss": 0.0375,
      "step": 16210
    },
    {
      "epoch": 0.83712884069197,
      "grad_norm": 6.711226940155029,
      "learning_rate": 6.531064658368019e-07,
      "loss": 0.0385,
      "step": 16211
    },
    {
      "epoch": 0.8371804802478698,
      "grad_norm": 11.389935493469238,
      "learning_rate": 6.527016400830949e-07,
      "loss": 0.1003,
      "step": 16212
    },
    {
      "epoch": 0.8372321198037697,
      "grad_norm": 4.4150285720825195,
      "learning_rate": 6.522969310720667e-07,
      "loss": 0.041,
      "step": 16213
    },
    {
      "epoch": 0.8372837593596695,
      "grad_norm": 5.489328861236572,
      "learning_rate": 6.518923388145876e-07,
      "loss": 0.035,
      "step": 16214
    },
    {
      "epoch": 0.8373353989155693,
      "grad_norm": 3.4161956310272217,
      "learning_rate": 6.514878633215188e-07,
      "loss": 0.0585,
      "step": 16215
    },
    {
      "epoch": 0.8373870384714691,
      "grad_norm": 5.8383564949035645,
      "learning_rate": 6.510835046037256e-07,
      "loss": 0.0284,
      "step": 16216
    },
    {
      "epoch": 0.8374386780273689,
      "grad_norm": 3.5546376705169678,
      "learning_rate": 6.506792626720665e-07,
      "loss": 0.0501,
      "step": 16217
    },
    {
      "epoch": 0.8374903175832688,
      "grad_norm": 3.949373483657837,
      "learning_rate": 6.50275137537395e-07,
      "loss": 0.0624,
      "step": 16218
    },
    {
      "epoch": 0.8375419571391686,
      "grad_norm": 4.173783302307129,
      "learning_rate": 6.498711292105648e-07,
      "loss": 0.0275,
      "step": 16219
    },
    {
      "epoch": 0.8375935966950684,
      "grad_norm": 5.868218421936035,
      "learning_rate": 6.494672377024236e-07,
      "loss": 0.0636,
      "step": 16220
    },
    {
      "epoch": 0.8376452362509682,
      "grad_norm": 3.3070068359375,
      "learning_rate": 6.490634630238213e-07,
      "loss": 0.0483,
      "step": 16221
    },
    {
      "epoch": 0.837696875806868,
      "grad_norm": 9.065231323242188,
      "learning_rate": 6.486598051855969e-07,
      "loss": 0.1069,
      "step": 16222
    },
    {
      "epoch": 0.8377485153627678,
      "grad_norm": 1.1010329723358154,
      "learning_rate": 6.482562641985923e-07,
      "loss": 0.0137,
      "step": 16223
    },
    {
      "epoch": 0.8378001549186677,
      "grad_norm": 7.09214973449707,
      "learning_rate": 6.478528400736434e-07,
      "loss": 0.0903,
      "step": 16224
    },
    {
      "epoch": 0.8378517944745675,
      "grad_norm": 1.3405218124389648,
      "learning_rate": 6.474495328215841e-07,
      "loss": 0.0076,
      "step": 16225
    },
    {
      "epoch": 0.8379034340304673,
      "grad_norm": 3.461019515991211,
      "learning_rate": 6.470463424532453e-07,
      "loss": 0.0413,
      "step": 16226
    },
    {
      "epoch": 0.8379550735863671,
      "grad_norm": 5.210290431976318,
      "learning_rate": 6.466432689794533e-07,
      "loss": 0.0321,
      "step": 16227
    },
    {
      "epoch": 0.8380067131422669,
      "grad_norm": 4.241223335266113,
      "learning_rate": 6.462403124110339e-07,
      "loss": 0.0629,
      "step": 16228
    },
    {
      "epoch": 0.8380583526981668,
      "grad_norm": 1.9467670917510986,
      "learning_rate": 6.458374727588052e-07,
      "loss": 0.0395,
      "step": 16229
    },
    {
      "epoch": 0.8381099922540666,
      "grad_norm": 3.172222137451172,
      "learning_rate": 6.454347500335884e-07,
      "loss": 0.0593,
      "step": 16230
    },
    {
      "epoch": 0.8381616318099664,
      "grad_norm": 3.7645177841186523,
      "learning_rate": 6.450321442461959e-07,
      "loss": 0.0528,
      "step": 16231
    },
    {
      "epoch": 0.8382132713658662,
      "grad_norm": 3.0734312534332275,
      "learning_rate": 6.446296554074416e-07,
      "loss": 0.028,
      "step": 16232
    },
    {
      "epoch": 0.838264910921766,
      "grad_norm": 3.908968925476074,
      "learning_rate": 6.442272835281316e-07,
      "loss": 0.0463,
      "step": 16233
    },
    {
      "epoch": 0.8383165504776658,
      "grad_norm": 3.1371283531188965,
      "learning_rate": 6.438250286190711e-07,
      "loss": 0.0546,
      "step": 16234
    },
    {
      "epoch": 0.8383681900335657,
      "grad_norm": 1.724740743637085,
      "learning_rate": 6.434228906910644e-07,
      "loss": 0.0236,
      "step": 16235
    },
    {
      "epoch": 0.8384198295894655,
      "grad_norm": 17.855112075805664,
      "learning_rate": 6.430208697549106e-07,
      "loss": 0.1518,
      "step": 16236
    },
    {
      "epoch": 0.8384714691453653,
      "grad_norm": 2.796205520629883,
      "learning_rate": 6.426189658214038e-07,
      "loss": 0.0404,
      "step": 16237
    },
    {
      "epoch": 0.8385231087012651,
      "grad_norm": 5.4446892738342285,
      "learning_rate": 6.422171789013371e-07,
      "loss": 0.0984,
      "step": 16238
    },
    {
      "epoch": 0.8385747482571649,
      "grad_norm": 2.376617908477783,
      "learning_rate": 6.418155090055006e-07,
      "loss": 0.0329,
      "step": 16239
    },
    {
      "epoch": 0.8386263878130648,
      "grad_norm": 1.814811110496521,
      "learning_rate": 6.414139561446802e-07,
      "loss": 0.0387,
      "step": 16240
    },
    {
      "epoch": 0.8386780273689646,
      "grad_norm": 3.778879404067993,
      "learning_rate": 6.410125203296602e-07,
      "loss": 0.0583,
      "step": 16241
    },
    {
      "epoch": 0.8387296669248644,
      "grad_norm": 2.6465601921081543,
      "learning_rate": 6.406112015712201e-07,
      "loss": 0.046,
      "step": 16242
    },
    {
      "epoch": 0.8387813064807642,
      "grad_norm": 11.768157958984375,
      "learning_rate": 6.402099998801375e-07,
      "loss": 0.1346,
      "step": 16243
    },
    {
      "epoch": 0.8388329460366641,
      "grad_norm": 1.51509690284729,
      "learning_rate": 6.398089152671855e-07,
      "loss": 0.0229,
      "step": 16244
    },
    {
      "epoch": 0.838884585592564,
      "grad_norm": 4.318658351898193,
      "learning_rate": 6.394079477431354e-07,
      "loss": 0.0796,
      "step": 16245
    },
    {
      "epoch": 0.8389362251484638,
      "grad_norm": 1.6837866306304932,
      "learning_rate": 6.390070973187545e-07,
      "loss": 0.0272,
      "step": 16246
    },
    {
      "epoch": 0.8389878647043636,
      "grad_norm": 3.9258158206939697,
      "learning_rate": 6.38606364004809e-07,
      "loss": 0.0324,
      "step": 16247
    },
    {
      "epoch": 0.8390395042602634,
      "grad_norm": 5.966703414916992,
      "learning_rate": 6.382057478120563e-07,
      "loss": 0.0517,
      "step": 16248
    },
    {
      "epoch": 0.8390911438161632,
      "grad_norm": 7.4917449951171875,
      "learning_rate": 6.378052487512576e-07,
      "loss": 0.1177,
      "step": 16249
    },
    {
      "epoch": 0.839142783372063,
      "grad_norm": 19.447040557861328,
      "learning_rate": 6.374048668331689e-07,
      "loss": 0.1346,
      "step": 16250
    },
    {
      "epoch": 0.8391944229279629,
      "grad_norm": 3.458953380584717,
      "learning_rate": 6.370046020685394e-07,
      "loss": 0.057,
      "step": 16251
    },
    {
      "epoch": 0.8392460624838627,
      "grad_norm": 8.75084400177002,
      "learning_rate": 6.366044544681188e-07,
      "loss": 0.0719,
      "step": 16252
    },
    {
      "epoch": 0.8392977020397625,
      "grad_norm": 13.985344886779785,
      "learning_rate": 6.362044240426518e-07,
      "loss": 0.1378,
      "step": 16253
    },
    {
      "epoch": 0.8393493415956623,
      "grad_norm": 5.094115257263184,
      "learning_rate": 6.358045108028838e-07,
      "loss": 0.0485,
      "step": 16254
    },
    {
      "epoch": 0.8394009811515621,
      "grad_norm": 1.5528513193130493,
      "learning_rate": 6.35404714759551e-07,
      "loss": 0.016,
      "step": 16255
    },
    {
      "epoch": 0.839452620707462,
      "grad_norm": 2.1427721977233887,
      "learning_rate": 6.350050359233912e-07,
      "loss": 0.0343,
      "step": 16256
    },
    {
      "epoch": 0.8395042602633618,
      "grad_norm": 6.504336357116699,
      "learning_rate": 6.346054743051366e-07,
      "loss": 0.0489,
      "step": 16257
    },
    {
      "epoch": 0.8395558998192616,
      "grad_norm": 11.35584831237793,
      "learning_rate": 6.342060299155173e-07,
      "loss": 0.0761,
      "step": 16258
    },
    {
      "epoch": 0.8396075393751614,
      "grad_norm": 1.8181653022766113,
      "learning_rate": 6.3380670276526e-07,
      "loss": 0.0388,
      "step": 16259
    },
    {
      "epoch": 0.8396591789310612,
      "grad_norm": 3.2307088375091553,
      "learning_rate": 6.334074928650885e-07,
      "loss": 0.0729,
      "step": 16260
    },
    {
      "epoch": 0.839710818486961,
      "grad_norm": 7.0815653800964355,
      "learning_rate": 6.330084002257242e-07,
      "loss": 0.0833,
      "step": 16261
    },
    {
      "epoch": 0.8397624580428609,
      "grad_norm": 3.102100133895874,
      "learning_rate": 6.326094248578812e-07,
      "loss": 0.0224,
      "step": 16262
    },
    {
      "epoch": 0.8398140975987607,
      "grad_norm": 2.9807791709899902,
      "learning_rate": 6.322105667722766e-07,
      "loss": 0.0355,
      "step": 16263
    },
    {
      "epoch": 0.8398657371546605,
      "grad_norm": 3.150869369506836,
      "learning_rate": 6.318118259796208e-07,
      "loss": 0.0475,
      "step": 16264
    },
    {
      "epoch": 0.8399173767105603,
      "grad_norm": 8.029996871948242,
      "learning_rate": 6.314132024906216e-07,
      "loss": 0.1772,
      "step": 16265
    },
    {
      "epoch": 0.8399690162664601,
      "grad_norm": 2.0952372550964355,
      "learning_rate": 6.310146963159825e-07,
      "loss": 0.0175,
      "step": 16266
    },
    {
      "epoch": 0.84002065582236,
      "grad_norm": 5.83761739730835,
      "learning_rate": 6.306163074664045e-07,
      "loss": 0.0767,
      "step": 16267
    },
    {
      "epoch": 0.8400722953782598,
      "grad_norm": 4.041675567626953,
      "learning_rate": 6.302180359525889e-07,
      "loss": 0.0455,
      "step": 16268
    },
    {
      "epoch": 0.8401239349341596,
      "grad_norm": 14.63342571258545,
      "learning_rate": 6.298198817852297e-07,
      "loss": 0.0737,
      "step": 16269
    },
    {
      "epoch": 0.8401755744900594,
      "grad_norm": 2.5527381896972656,
      "learning_rate": 6.294218449750178e-07,
      "loss": 0.0455,
      "step": 16270
    },
    {
      "epoch": 0.8402272140459592,
      "grad_norm": 5.324949264526367,
      "learning_rate": 6.29023925532643e-07,
      "loss": 0.0523,
      "step": 16271
    },
    {
      "epoch": 0.840278853601859,
      "grad_norm": 7.650274753570557,
      "learning_rate": 6.286261234687907e-07,
      "loss": 0.0417,
      "step": 16272
    },
    {
      "epoch": 0.8403304931577589,
      "grad_norm": 6.559195041656494,
      "learning_rate": 6.28228438794144e-07,
      "loss": 0.0658,
      "step": 16273
    },
    {
      "epoch": 0.8403821327136587,
      "grad_norm": 1.8727531433105469,
      "learning_rate": 6.278308715193821e-07,
      "loss": 0.0258,
      "step": 16274
    },
    {
      "epoch": 0.8404337722695585,
      "grad_norm": 3.28771710395813,
      "learning_rate": 6.274334216551809e-07,
      "loss": 0.0434,
      "step": 16275
    },
    {
      "epoch": 0.8404854118254583,
      "grad_norm": 1.6428624391555786,
      "learning_rate": 6.270360892122141e-07,
      "loss": 0.0212,
      "step": 16276
    },
    {
      "epoch": 0.8405370513813581,
      "grad_norm": 9.059910774230957,
      "learning_rate": 6.266388742011514e-07,
      "loss": 0.0563,
      "step": 16277
    },
    {
      "epoch": 0.840588690937258,
      "grad_norm": 1.2108278274536133,
      "learning_rate": 6.262417766326595e-07,
      "loss": 0.0118,
      "step": 16278
    },
    {
      "epoch": 0.8406403304931578,
      "grad_norm": 5.212985992431641,
      "learning_rate": 6.258447965174031e-07,
      "loss": 0.0907,
      "step": 16279
    },
    {
      "epoch": 0.8406919700490576,
      "grad_norm": 2.7777295112609863,
      "learning_rate": 6.254479338660424e-07,
      "loss": 0.0325,
      "step": 16280
    },
    {
      "epoch": 0.8407436096049574,
      "grad_norm": 2.3389217853546143,
      "learning_rate": 6.250511886892324e-07,
      "loss": 0.0205,
      "step": 16281
    },
    {
      "epoch": 0.8407952491608572,
      "grad_norm": 5.327186107635498,
      "learning_rate": 6.246545609976307e-07,
      "loss": 0.0578,
      "step": 16282
    },
    {
      "epoch": 0.8408468887167571,
      "grad_norm": 2.0156495571136475,
      "learning_rate": 6.242580508018875e-07,
      "loss": 0.0432,
      "step": 16283
    },
    {
      "epoch": 0.8408985282726569,
      "grad_norm": 4.851264953613281,
      "learning_rate": 6.238616581126495e-07,
      "loss": 0.0847,
      "step": 16284
    },
    {
      "epoch": 0.8409501678285567,
      "grad_norm": 5.998858451843262,
      "learning_rate": 6.234653829405618e-07,
      "loss": 0.0463,
      "step": 16285
    },
    {
      "epoch": 0.8410018073844565,
      "grad_norm": 6.3027472496032715,
      "learning_rate": 6.230692252962655e-07,
      "loss": 0.0715,
      "step": 16286
    },
    {
      "epoch": 0.8410534469403563,
      "grad_norm": 11.934480667114258,
      "learning_rate": 6.226731851904022e-07,
      "loss": 0.0828,
      "step": 16287
    },
    {
      "epoch": 0.8411050864962561,
      "grad_norm": 3.745436429977417,
      "learning_rate": 6.222772626336033e-07,
      "loss": 0.0655,
      "step": 16288
    },
    {
      "epoch": 0.841156726052156,
      "grad_norm": 3.4521994590759277,
      "learning_rate": 6.218814576365029e-07,
      "loss": 0.0329,
      "step": 16289
    },
    {
      "epoch": 0.8412083656080558,
      "grad_norm": 2.9836599826812744,
      "learning_rate": 6.214857702097293e-07,
      "loss": 0.0201,
      "step": 16290
    },
    {
      "epoch": 0.8412600051639556,
      "grad_norm": 14.563075065612793,
      "learning_rate": 6.210902003639091e-07,
      "loss": 0.1107,
      "step": 16291
    },
    {
      "epoch": 0.8413116447198554,
      "grad_norm": 1.385168433189392,
      "learning_rate": 6.206947481096642e-07,
      "loss": 0.0224,
      "step": 16292
    },
    {
      "epoch": 0.8413632842757552,
      "grad_norm": 6.013901233673096,
      "learning_rate": 6.20299413457614e-07,
      "loss": 0.0503,
      "step": 16293
    },
    {
      "epoch": 0.8414149238316551,
      "grad_norm": 5.848572731018066,
      "learning_rate": 6.199041964183766e-07,
      "loss": 0.0882,
      "step": 16294
    },
    {
      "epoch": 0.8414665633875549,
      "grad_norm": 13.630045890808105,
      "learning_rate": 6.195090970025619e-07,
      "loss": 0.1214,
      "step": 16295
    },
    {
      "epoch": 0.8415182029434547,
      "grad_norm": 2.2740938663482666,
      "learning_rate": 6.191141152207825e-07,
      "loss": 0.0422,
      "step": 16296
    },
    {
      "epoch": 0.8415698424993545,
      "grad_norm": 3.8097167015075684,
      "learning_rate": 6.187192510836449e-07,
      "loss": 0.0763,
      "step": 16297
    },
    {
      "epoch": 0.8416214820552543,
      "grad_norm": 2.429508924484253,
      "learning_rate": 6.183245046017538e-07,
      "loss": 0.0371,
      "step": 16298
    },
    {
      "epoch": 0.8416731216111542,
      "grad_norm": 4.774104118347168,
      "learning_rate": 6.179298757857072e-07,
      "loss": 0.063,
      "step": 16299
    },
    {
      "epoch": 0.841724761167054,
      "grad_norm": 7.198826313018799,
      "learning_rate": 6.175353646461024e-07,
      "loss": 0.0752,
      "step": 16300
    },
    {
      "epoch": 0.8417764007229538,
      "grad_norm": 12.250945091247559,
      "learning_rate": 6.171409711935361e-07,
      "loss": 0.0544,
      "step": 16301
    },
    {
      "epoch": 0.8418280402788536,
      "grad_norm": 4.330167770385742,
      "learning_rate": 6.167466954385997e-07,
      "loss": 0.0499,
      "step": 16302
    },
    {
      "epoch": 0.8418796798347534,
      "grad_norm": 5.90062141418457,
      "learning_rate": 6.163525373918782e-07,
      "loss": 0.0565,
      "step": 16303
    },
    {
      "epoch": 0.8419313193906532,
      "grad_norm": 4.0332536697387695,
      "learning_rate": 6.159584970639582e-07,
      "loss": 0.0389,
      "step": 16304
    },
    {
      "epoch": 0.8419829589465531,
      "grad_norm": 4.7215752601623535,
      "learning_rate": 6.155645744654204e-07,
      "loss": 0.0451,
      "step": 16305
    },
    {
      "epoch": 0.8420345985024529,
      "grad_norm": 14.661474227905273,
      "learning_rate": 6.151707696068443e-07,
      "loss": 0.1331,
      "step": 16306
    },
    {
      "epoch": 0.8420862380583527,
      "grad_norm": 3.5518760681152344,
      "learning_rate": 6.147770824988048e-07,
      "loss": 0.0416,
      "step": 16307
    },
    {
      "epoch": 0.8421378776142525,
      "grad_norm": 1.7175817489624023,
      "learning_rate": 6.143835131518732e-07,
      "loss": 0.0275,
      "step": 16308
    },
    {
      "epoch": 0.8421895171701523,
      "grad_norm": 7.001785755157471,
      "learning_rate": 6.139900615766192e-07,
      "loss": 0.0383,
      "step": 16309
    },
    {
      "epoch": 0.8422411567260522,
      "grad_norm": 2.929093599319458,
      "learning_rate": 6.135967277836085e-07,
      "loss": 0.0217,
      "step": 16310
    },
    {
      "epoch": 0.842292796281952,
      "grad_norm": 3.7327044010162354,
      "learning_rate": 6.132035117834035e-07,
      "loss": 0.0595,
      "step": 16311
    },
    {
      "epoch": 0.8423444358378518,
      "grad_norm": 6.441672325134277,
      "learning_rate": 6.128104135865642e-07,
      "loss": 0.083,
      "step": 16312
    },
    {
      "epoch": 0.8423960753937516,
      "grad_norm": 4.037454128265381,
      "learning_rate": 6.12417433203647e-07,
      "loss": 0.0451,
      "step": 16313
    },
    {
      "epoch": 0.8424477149496514,
      "grad_norm": 3.7242965698242188,
      "learning_rate": 6.120245706452027e-07,
      "loss": 0.0163,
      "step": 16314
    },
    {
      "epoch": 0.8424993545055512,
      "grad_norm": 3.5139169692993164,
      "learning_rate": 6.116318259217846e-07,
      "loss": 0.0549,
      "step": 16315
    },
    {
      "epoch": 0.8425509940614511,
      "grad_norm": 3.216930627822876,
      "learning_rate": 6.112391990439387e-07,
      "loss": 0.0388,
      "step": 16316
    },
    {
      "epoch": 0.8426026336173509,
      "grad_norm": 2.115696430206299,
      "learning_rate": 6.108466900222066e-07,
      "loss": 0.0326,
      "step": 16317
    },
    {
      "epoch": 0.8426542731732507,
      "grad_norm": 1.5966951847076416,
      "learning_rate": 6.104542988671302e-07,
      "loss": 0.013,
      "step": 16318
    },
    {
      "epoch": 0.8427059127291505,
      "grad_norm": 7.204645156860352,
      "learning_rate": 6.100620255892453e-07,
      "loss": 0.084,
      "step": 16319
    },
    {
      "epoch": 0.8427575522850503,
      "grad_norm": 2.438955068588257,
      "learning_rate": 6.096698701990899e-07,
      "loss": 0.0458,
      "step": 16320
    },
    {
      "epoch": 0.8428091918409502,
      "grad_norm": 2.4835617542266846,
      "learning_rate": 6.092778327071913e-07,
      "loss": 0.0303,
      "step": 16321
    },
    {
      "epoch": 0.84286083139685,
      "grad_norm": 3.8021018505096436,
      "learning_rate": 6.088859131240787e-07,
      "loss": 0.0503,
      "step": 16322
    },
    {
      "epoch": 0.8429124709527498,
      "grad_norm": 1.5628855228424072,
      "learning_rate": 6.084941114602765e-07,
      "loss": 0.0237,
      "step": 16323
    },
    {
      "epoch": 0.8429641105086496,
      "grad_norm": 9.592628479003906,
      "learning_rate": 6.081024277263059e-07,
      "loss": 0.0767,
      "step": 16324
    },
    {
      "epoch": 0.8430157500645494,
      "grad_norm": 3.2902817726135254,
      "learning_rate": 6.077108619326855e-07,
      "loss": 0.0455,
      "step": 16325
    },
    {
      "epoch": 0.8430673896204492,
      "grad_norm": 1.6728899478912354,
      "learning_rate": 6.073194140899308e-07,
      "loss": 0.0251,
      "step": 16326
    },
    {
      "epoch": 0.8431190291763491,
      "grad_norm": 7.344292640686035,
      "learning_rate": 6.069280842085545e-07,
      "loss": 0.0763,
      "step": 16327
    },
    {
      "epoch": 0.8431706687322489,
      "grad_norm": 7.351011276245117,
      "learning_rate": 6.065368722990617e-07,
      "loss": 0.0449,
      "step": 16328
    },
    {
      "epoch": 0.8432223082881487,
      "grad_norm": 6.394491672515869,
      "learning_rate": 6.061457783719621e-07,
      "loss": 0.0775,
      "step": 16329
    },
    {
      "epoch": 0.8432739478440485,
      "grad_norm": 3.170315980911255,
      "learning_rate": 6.057548024377563e-07,
      "loss": 0.0651,
      "step": 16330
    },
    {
      "epoch": 0.8433255873999483,
      "grad_norm": 5.079795837402344,
      "learning_rate": 6.053639445069453e-07,
      "loss": 0.0284,
      "step": 16331
    },
    {
      "epoch": 0.8433772269558482,
      "grad_norm": 5.6213531494140625,
      "learning_rate": 6.049732045900231e-07,
      "loss": 0.0699,
      "step": 16332
    },
    {
      "epoch": 0.843428866511748,
      "grad_norm": 6.030441761016846,
      "learning_rate": 6.045825826974821e-07,
      "loss": 0.0201,
      "step": 16333
    },
    {
      "epoch": 0.8434805060676478,
      "grad_norm": 5.577458381652832,
      "learning_rate": 6.041920788398148e-07,
      "loss": 0.0633,
      "step": 16334
    },
    {
      "epoch": 0.8435321456235476,
      "grad_norm": 4.356272220611572,
      "learning_rate": 6.038016930275075e-07,
      "loss": 0.0673,
      "step": 16335
    },
    {
      "epoch": 0.8435837851794474,
      "grad_norm": 1.8937091827392578,
      "learning_rate": 6.034114252710416e-07,
      "loss": 0.025,
      "step": 16336
    },
    {
      "epoch": 0.8436354247353473,
      "grad_norm": 4.332147598266602,
      "learning_rate": 6.030212755808985e-07,
      "loss": 0.024,
      "step": 16337
    },
    {
      "epoch": 0.8436870642912471,
      "grad_norm": 14.140471458435059,
      "learning_rate": 6.026312439675553e-07,
      "loss": 0.0884,
      "step": 16338
    },
    {
      "epoch": 0.8437387038471469,
      "grad_norm": 5.890823841094971,
      "learning_rate": 6.022413304414854e-07,
      "loss": 0.0865,
      "step": 16339
    },
    {
      "epoch": 0.8437903434030467,
      "grad_norm": 4.838178634643555,
      "learning_rate": 6.018515350131604e-07,
      "loss": 0.0863,
      "step": 16340
    },
    {
      "epoch": 0.8438419829589465,
      "grad_norm": 2.0128965377807617,
      "learning_rate": 6.014618576930475e-07,
      "loss": 0.023,
      "step": 16341
    },
    {
      "epoch": 0.8438936225148463,
      "grad_norm": 5.883184909820557,
      "learning_rate": 6.010722984916112e-07,
      "loss": 0.0858,
      "step": 16342
    },
    {
      "epoch": 0.8439452620707462,
      "grad_norm": 2.0759403705596924,
      "learning_rate": 6.00682857419313e-07,
      "loss": 0.0188,
      "step": 16343
    },
    {
      "epoch": 0.843996901626646,
      "grad_norm": 5.280817031860352,
      "learning_rate": 6.002935344866101e-07,
      "loss": 0.0519,
      "step": 16344
    },
    {
      "epoch": 0.8440485411825458,
      "grad_norm": 2.692293882369995,
      "learning_rate": 5.999043297039586e-07,
      "loss": 0.0343,
      "step": 16345
    },
    {
      "epoch": 0.8441001807384456,
      "grad_norm": 2.931603193283081,
      "learning_rate": 5.995152430818102e-07,
      "loss": 0.0303,
      "step": 16346
    },
    {
      "epoch": 0.8441518202943454,
      "grad_norm": 3.5100622177124023,
      "learning_rate": 5.991262746306109e-07,
      "loss": 0.0491,
      "step": 16347
    },
    {
      "epoch": 0.8442034598502453,
      "grad_norm": 1.720383644104004,
      "learning_rate": 5.987374243608096e-07,
      "loss": 0.0301,
      "step": 16348
    },
    {
      "epoch": 0.8442550994061451,
      "grad_norm": 2.9856958389282227,
      "learning_rate": 5.983486922828475e-07,
      "loss": 0.05,
      "step": 16349
    },
    {
      "epoch": 0.8443067389620449,
      "grad_norm": 17.575927734375,
      "learning_rate": 5.979600784071627e-07,
      "loss": 0.1418,
      "step": 16350
    },
    {
      "epoch": 0.8443583785179447,
      "grad_norm": 4.883927822113037,
      "learning_rate": 5.975715827441908e-07,
      "loss": 0.0723,
      "step": 16351
    },
    {
      "epoch": 0.8444100180738445,
      "grad_norm": 3.5991029739379883,
      "learning_rate": 5.971832053043647e-07,
      "loss": 0.0477,
      "step": 16352
    },
    {
      "epoch": 0.8444616576297443,
      "grad_norm": 2.9868524074554443,
      "learning_rate": 5.967949460981165e-07,
      "loss": 0.0292,
      "step": 16353
    },
    {
      "epoch": 0.8445132971856442,
      "grad_norm": 1.9166040420532227,
      "learning_rate": 5.96406805135869e-07,
      "loss": 0.0166,
      "step": 16354
    },
    {
      "epoch": 0.844564936741544,
      "grad_norm": 2.4454519748687744,
      "learning_rate": 5.960187824280472e-07,
      "loss": 0.0533,
      "step": 16355
    },
    {
      "epoch": 0.8446165762974438,
      "grad_norm": 10.416871070861816,
      "learning_rate": 5.956308779850706e-07,
      "loss": 0.0394,
      "step": 16356
    },
    {
      "epoch": 0.8446682158533436,
      "grad_norm": 2.7354896068573,
      "learning_rate": 5.952430918173563e-07,
      "loss": 0.0315,
      "step": 16357
    },
    {
      "epoch": 0.8447198554092434,
      "grad_norm": 3.0525553226470947,
      "learning_rate": 5.948554239353172e-07,
      "loss": 0.037,
      "step": 16358
    },
    {
      "epoch": 0.8447714949651433,
      "grad_norm": 14.714496612548828,
      "learning_rate": 5.944678743493648e-07,
      "loss": 0.1548,
      "step": 16359
    },
    {
      "epoch": 0.8448231345210431,
      "grad_norm": 4.235710144042969,
      "learning_rate": 5.94080443069906e-07,
      "loss": 0.0542,
      "step": 16360
    },
    {
      "epoch": 0.8448747740769429,
      "grad_norm": 3.615917921066284,
      "learning_rate": 5.936931301073446e-07,
      "loss": 0.0375,
      "step": 16361
    },
    {
      "epoch": 0.8449264136328427,
      "grad_norm": 1.785460114479065,
      "learning_rate": 5.933059354720822e-07,
      "loss": 0.0262,
      "step": 16362
    },
    {
      "epoch": 0.8449780531887425,
      "grad_norm": 4.737611770629883,
      "learning_rate": 5.929188591745156e-07,
      "loss": 0.0628,
      "step": 16363
    },
    {
      "epoch": 0.8450296927446423,
      "grad_norm": 9.264492988586426,
      "learning_rate": 5.925319012250413e-07,
      "loss": 0.0484,
      "step": 16364
    },
    {
      "epoch": 0.8450813323005422,
      "grad_norm": 7.459939956665039,
      "learning_rate": 5.92145061634048e-07,
      "loss": 0.0679,
      "step": 16365
    },
    {
      "epoch": 0.845132971856442,
      "grad_norm": 12.51922607421875,
      "learning_rate": 5.917583404119243e-07,
      "loss": 0.0657,
      "step": 16366
    },
    {
      "epoch": 0.8451846114123418,
      "grad_norm": 4.967652797698975,
      "learning_rate": 5.913717375690575e-07,
      "loss": 0.0755,
      "step": 16367
    },
    {
      "epoch": 0.8452362509682416,
      "grad_norm": 10.250895500183105,
      "learning_rate": 5.909852531158289e-07,
      "loss": 0.0774,
      "step": 16368
    },
    {
      "epoch": 0.8452878905241414,
      "grad_norm": 1.5102418661117554,
      "learning_rate": 5.90598887062615e-07,
      "loss": 0.0117,
      "step": 16369
    },
    {
      "epoch": 0.8453395300800413,
      "grad_norm": 4.4765191078186035,
      "learning_rate": 5.902126394197932e-07,
      "loss": 0.0471,
      "step": 16370
    },
    {
      "epoch": 0.8453911696359411,
      "grad_norm": 6.584122180938721,
      "learning_rate": 5.898265101977352e-07,
      "loss": 0.0906,
      "step": 16371
    },
    {
      "epoch": 0.845442809191841,
      "grad_norm": 2.6046481132507324,
      "learning_rate": 5.894404994068103e-07,
      "loss": 0.0325,
      "step": 16372
    },
    {
      "epoch": 0.8454944487477408,
      "grad_norm": 5.6374897956848145,
      "learning_rate": 5.890546070573843e-07,
      "loss": 0.0429,
      "step": 16373
    },
    {
      "epoch": 0.8455460883036406,
      "grad_norm": 12.234259605407715,
      "learning_rate": 5.886688331598201e-07,
      "loss": 0.1399,
      "step": 16374
    },
    {
      "epoch": 0.8455977278595405,
      "grad_norm": 4.845226287841797,
      "learning_rate": 5.882831777244775e-07,
      "loss": 0.0657,
      "step": 16375
    },
    {
      "epoch": 0.8456493674154403,
      "grad_norm": 9.03059196472168,
      "learning_rate": 5.878976407617126e-07,
      "loss": 0.1263,
      "step": 16376
    },
    {
      "epoch": 0.8457010069713401,
      "grad_norm": 2.544053077697754,
      "learning_rate": 5.875122222818791e-07,
      "loss": 0.0263,
      "step": 16377
    },
    {
      "epoch": 0.8457526465272399,
      "grad_norm": 5.174503803253174,
      "learning_rate": 5.871269222953263e-07,
      "loss": 0.0455,
      "step": 16378
    },
    {
      "epoch": 0.8458042860831397,
      "grad_norm": 13.51972484588623,
      "learning_rate": 5.867417408124027e-07,
      "loss": 0.0445,
      "step": 16379
    },
    {
      "epoch": 0.8458559256390396,
      "grad_norm": 13.362491607666016,
      "learning_rate": 5.863566778434487e-07,
      "loss": 0.092,
      "step": 16380
    },
    {
      "epoch": 0.8459075651949394,
      "grad_norm": 2.660598039627075,
      "learning_rate": 5.859717333988079e-07,
      "loss": 0.0428,
      "step": 16381
    },
    {
      "epoch": 0.8459592047508392,
      "grad_norm": 6.422194957733154,
      "learning_rate": 5.855869074888182e-07,
      "loss": 0.0949,
      "step": 16382
    },
    {
      "epoch": 0.846010844306739,
      "grad_norm": 2.881216287612915,
      "learning_rate": 5.852022001238106e-07,
      "loss": 0.0461,
      "step": 16383
    },
    {
      "epoch": 0.8460624838626388,
      "grad_norm": 57.65004348754883,
      "learning_rate": 5.84817611314118e-07,
      "loss": 0.1462,
      "step": 16384
    },
    {
      "epoch": 0.8461141234185386,
      "grad_norm": 3.9481232166290283,
      "learning_rate": 5.844331410700666e-07,
      "loss": 0.069,
      "step": 16385
    },
    {
      "epoch": 0.8461657629744385,
      "grad_norm": 2.6631784439086914,
      "learning_rate": 5.840487894019841e-07,
      "loss": 0.0349,
      "step": 16386
    },
    {
      "epoch": 0.8462174025303383,
      "grad_norm": 5.544249534606934,
      "learning_rate": 5.836645563201892e-07,
      "loss": 0.0501,
      "step": 16387
    },
    {
      "epoch": 0.8462690420862381,
      "grad_norm": 2.8631441593170166,
      "learning_rate": 5.832804418350007e-07,
      "loss": 0.0423,
      "step": 16388
    },
    {
      "epoch": 0.8463206816421379,
      "grad_norm": 4.31330680847168,
      "learning_rate": 5.828964459567344e-07,
      "loss": 0.065,
      "step": 16389
    },
    {
      "epoch": 0.8463723211980377,
      "grad_norm": 0.844182550907135,
      "learning_rate": 5.825125686957012e-07,
      "loss": 0.0094,
      "step": 16390
    },
    {
      "epoch": 0.8464239607539376,
      "grad_norm": 2.1045114994049072,
      "learning_rate": 5.821288100622102e-07,
      "loss": 0.0321,
      "step": 16391
    },
    {
      "epoch": 0.8464756003098374,
      "grad_norm": 4.668204307556152,
      "learning_rate": 5.817451700665672e-07,
      "loss": 0.0574,
      "step": 16392
    },
    {
      "epoch": 0.8465272398657372,
      "grad_norm": 13.730819702148438,
      "learning_rate": 5.813616487190743e-07,
      "loss": 0.1361,
      "step": 16393
    },
    {
      "epoch": 0.846578879421637,
      "grad_norm": 0.8194041848182678,
      "learning_rate": 5.809782460300301e-07,
      "loss": 0.0078,
      "step": 16394
    },
    {
      "epoch": 0.8466305189775368,
      "grad_norm": 3.460690498352051,
      "learning_rate": 5.805949620097312e-07,
      "loss": 0.0346,
      "step": 16395
    },
    {
      "epoch": 0.8466821585334366,
      "grad_norm": 12.229516983032227,
      "learning_rate": 5.802117966684695e-07,
      "loss": 0.0829,
      "step": 16396
    },
    {
      "epoch": 0.8467337980893365,
      "grad_norm": 1.9047645330429077,
      "learning_rate": 5.798287500165367e-07,
      "loss": 0.0355,
      "step": 16397
    },
    {
      "epoch": 0.8467854376452363,
      "grad_norm": 4.294262886047363,
      "learning_rate": 5.794458220642169e-07,
      "loss": 0.0364,
      "step": 16398
    },
    {
      "epoch": 0.8468370772011361,
      "grad_norm": 2.8837461471557617,
      "learning_rate": 5.790630128217922e-07,
      "loss": 0.031,
      "step": 16399
    },
    {
      "epoch": 0.8468887167570359,
      "grad_norm": 2.6252334117889404,
      "learning_rate": 5.786803222995469e-07,
      "loss": 0.0521,
      "step": 16400
    },
    {
      "epoch": 0.8469403563129357,
      "grad_norm": 2.243586778640747,
      "learning_rate": 5.782977505077536e-07,
      "loss": 0.0252,
      "step": 16401
    },
    {
      "epoch": 0.8469919958688356,
      "grad_norm": 4.2781782150268555,
      "learning_rate": 5.779152974566882e-07,
      "loss": 0.0634,
      "step": 16402
    },
    {
      "epoch": 0.8470436354247354,
      "grad_norm": 2.382272243499756,
      "learning_rate": 5.775329631566201e-07,
      "loss": 0.0278,
      "step": 16403
    },
    {
      "epoch": 0.8470952749806352,
      "grad_norm": 5.175331115722656,
      "learning_rate": 5.771507476178173e-07,
      "loss": 0.06,
      "step": 16404
    },
    {
      "epoch": 0.847146914536535,
      "grad_norm": 1.1235945224761963,
      "learning_rate": 5.767686508505432e-07,
      "loss": 0.0154,
      "step": 16405
    },
    {
      "epoch": 0.8471985540924348,
      "grad_norm": 2.245649576187134,
      "learning_rate": 5.763866728650586e-07,
      "loss": 0.0395,
      "step": 16406
    },
    {
      "epoch": 0.8472501936483346,
      "grad_norm": 1.7404451370239258,
      "learning_rate": 5.760048136716223e-07,
      "loss": 0.0286,
      "step": 16407
    },
    {
      "epoch": 0.8473018332042345,
      "grad_norm": 2.3954172134399414,
      "learning_rate": 5.756230732804874e-07,
      "loss": 0.0283,
      "step": 16408
    },
    {
      "epoch": 0.8473534727601343,
      "grad_norm": 2.4397919178009033,
      "learning_rate": 5.752414517019056e-07,
      "loss": 0.0135,
      "step": 16409
    },
    {
      "epoch": 0.8474051123160341,
      "grad_norm": 3.317279815673828,
      "learning_rate": 5.748599489461254e-07,
      "loss": 0.0277,
      "step": 16410
    },
    {
      "epoch": 0.8474567518719339,
      "grad_norm": 18.361692428588867,
      "learning_rate": 5.744785650233925e-07,
      "loss": 0.1422,
      "step": 16411
    },
    {
      "epoch": 0.8475083914278337,
      "grad_norm": 3.1919710636138916,
      "learning_rate": 5.740972999439454e-07,
      "loss": 0.0321,
      "step": 16412
    },
    {
      "epoch": 0.8475600309837336,
      "grad_norm": 2.9946274757385254,
      "learning_rate": 5.737161537180258e-07,
      "loss": 0.042,
      "step": 16413
    },
    {
      "epoch": 0.8476116705396334,
      "grad_norm": 1.5206693410873413,
      "learning_rate": 5.733351263558684e-07,
      "loss": 0.02,
      "step": 16414
    },
    {
      "epoch": 0.8476633100955332,
      "grad_norm": 2.8186533451080322,
      "learning_rate": 5.729542178677055e-07,
      "loss": 0.05,
      "step": 16415
    },
    {
      "epoch": 0.847714949651433,
      "grad_norm": 4.399238586425781,
      "learning_rate": 5.725734282637646e-07,
      "loss": 0.0486,
      "step": 16416
    },
    {
      "epoch": 0.8477665892073328,
      "grad_norm": 10.887459754943848,
      "learning_rate": 5.721927575542724e-07,
      "loss": 0.1167,
      "step": 16417
    },
    {
      "epoch": 0.8478182287632327,
      "grad_norm": 6.002501487731934,
      "learning_rate": 5.718122057494502e-07,
      "loss": 0.0777,
      "step": 16418
    },
    {
      "epoch": 0.8478698683191325,
      "grad_norm": 3.2440571784973145,
      "learning_rate": 5.714317728595209e-07,
      "loss": 0.0198,
      "step": 16419
    },
    {
      "epoch": 0.8479215078750323,
      "grad_norm": 4.234901428222656,
      "learning_rate": 5.710514588946969e-07,
      "loss": 0.0646,
      "step": 16420
    },
    {
      "epoch": 0.8479731474309321,
      "grad_norm": 2.873328924179077,
      "learning_rate": 5.706712638651928e-07,
      "loss": 0.0387,
      "step": 16421
    },
    {
      "epoch": 0.8480247869868319,
      "grad_norm": 2.3505420684814453,
      "learning_rate": 5.702911877812184e-07,
      "loss": 0.0305,
      "step": 16422
    },
    {
      "epoch": 0.8480764265427317,
      "grad_norm": 4.9976911544799805,
      "learning_rate": 5.699112306529803e-07,
      "loss": 0.042,
      "step": 16423
    },
    {
      "epoch": 0.8481280660986316,
      "grad_norm": 2.0862784385681152,
      "learning_rate": 5.695313924906809e-07,
      "loss": 0.0413,
      "step": 16424
    },
    {
      "epoch": 0.8481797056545314,
      "grad_norm": 5.479255676269531,
      "learning_rate": 5.691516733045221e-07,
      "loss": 0.0709,
      "step": 16425
    },
    {
      "epoch": 0.8482313452104312,
      "grad_norm": 2.713905096054077,
      "learning_rate": 5.687720731046992e-07,
      "loss": 0.0274,
      "step": 16426
    },
    {
      "epoch": 0.848282984766331,
      "grad_norm": 3.594919443130493,
      "learning_rate": 5.683925919014077e-07,
      "loss": 0.0578,
      "step": 16427
    },
    {
      "epoch": 0.8483346243222308,
      "grad_norm": 2.4719583988189697,
      "learning_rate": 5.680132297048369e-07,
      "loss": 0.0529,
      "step": 16428
    },
    {
      "epoch": 0.8483862638781307,
      "grad_norm": 13.291834831237793,
      "learning_rate": 5.676339865251745e-07,
      "loss": 0.122,
      "step": 16429
    },
    {
      "epoch": 0.8484379034340305,
      "grad_norm": 2.3061139583587646,
      "learning_rate": 5.67254862372606e-07,
      "loss": 0.0356,
      "step": 16430
    },
    {
      "epoch": 0.8484895429899303,
      "grad_norm": 1.2738627195358276,
      "learning_rate": 5.668758572573091e-07,
      "loss": 0.0152,
      "step": 16431
    },
    {
      "epoch": 0.8485411825458301,
      "grad_norm": 15.184661865234375,
      "learning_rate": 5.664969711894652e-07,
      "loss": 0.0419,
      "step": 16432
    },
    {
      "epoch": 0.8485928221017299,
      "grad_norm": 14.783492088317871,
      "learning_rate": 5.661182041792484e-07,
      "loss": 0.2514,
      "step": 16433
    },
    {
      "epoch": 0.8486444616576297,
      "grad_norm": 2.09084153175354,
      "learning_rate": 5.657395562368285e-07,
      "loss": 0.0472,
      "step": 16434
    },
    {
      "epoch": 0.8486961012135296,
      "grad_norm": 1.5860203504562378,
      "learning_rate": 5.653610273723742e-07,
      "loss": 0.0242,
      "step": 16435
    },
    {
      "epoch": 0.8487477407694294,
      "grad_norm": 4.193629264831543,
      "learning_rate": 5.649826175960515e-07,
      "loss": 0.0769,
      "step": 16436
    },
    {
      "epoch": 0.8487993803253292,
      "grad_norm": 1.56257963180542,
      "learning_rate": 5.646043269180213e-07,
      "loss": 0.019,
      "step": 16437
    },
    {
      "epoch": 0.848851019881229,
      "grad_norm": 5.457065105438232,
      "learning_rate": 5.642261553484424e-07,
      "loss": 0.0681,
      "step": 16438
    },
    {
      "epoch": 0.8489026594371288,
      "grad_norm": 3.735869884490967,
      "learning_rate": 5.638481028974707e-07,
      "loss": 0.0501,
      "step": 16439
    },
    {
      "epoch": 0.8489542989930287,
      "grad_norm": 8.24803638458252,
      "learning_rate": 5.634701695752582e-07,
      "loss": 0.0643,
      "step": 16440
    },
    {
      "epoch": 0.8490059385489285,
      "grad_norm": 12.931973457336426,
      "learning_rate": 5.630923553919538e-07,
      "loss": 0.0827,
      "step": 16441
    },
    {
      "epoch": 0.8490575781048283,
      "grad_norm": 14.721182823181152,
      "learning_rate": 5.627146603577039e-07,
      "loss": 0.1073,
      "step": 16442
    },
    {
      "epoch": 0.8491092176607281,
      "grad_norm": 1.881372332572937,
      "learning_rate": 5.623370844826503e-07,
      "loss": 0.0209,
      "step": 16443
    },
    {
      "epoch": 0.8491608572166279,
      "grad_norm": 1.9579219818115234,
      "learning_rate": 5.61959627776934e-07,
      "loss": 0.0484,
      "step": 16444
    },
    {
      "epoch": 0.8492124967725277,
      "grad_norm": 2.0952954292297363,
      "learning_rate": 5.615822902506878e-07,
      "loss": 0.0335,
      "step": 16445
    },
    {
      "epoch": 0.8492641363284276,
      "grad_norm": 4.583352088928223,
      "learning_rate": 5.612050719140488e-07,
      "loss": 0.0482,
      "step": 16446
    },
    {
      "epoch": 0.8493157758843274,
      "grad_norm": 2.793437957763672,
      "learning_rate": 5.608279727771443e-07,
      "loss": 0.0416,
      "step": 16447
    },
    {
      "epoch": 0.8493674154402272,
      "grad_norm": 2.3875350952148438,
      "learning_rate": 5.604509928501034e-07,
      "loss": 0.0283,
      "step": 16448
    },
    {
      "epoch": 0.849419054996127,
      "grad_norm": 4.353325843811035,
      "learning_rate": 5.600741321430469e-07,
      "loss": 0.0505,
      "step": 16449
    },
    {
      "epoch": 0.8494706945520268,
      "grad_norm": 5.119873523712158,
      "learning_rate": 5.596973906660951e-07,
      "loss": 0.041,
      "step": 16450
    },
    {
      "epoch": 0.8495223341079267,
      "grad_norm": 4.248696327209473,
      "learning_rate": 5.593207684293667e-07,
      "loss": 0.0648,
      "step": 16451
    },
    {
      "epoch": 0.8495739736638265,
      "grad_norm": 11.604622840881348,
      "learning_rate": 5.589442654429761e-07,
      "loss": 0.1026,
      "step": 16452
    },
    {
      "epoch": 0.8496256132197263,
      "grad_norm": 2.771960496902466,
      "learning_rate": 5.58567881717032e-07,
      "loss": 0.0263,
      "step": 16453
    },
    {
      "epoch": 0.8496772527756261,
      "grad_norm": 2.614163398742676,
      "learning_rate": 5.581916172616425e-07,
      "loss": 0.0134,
      "step": 16454
    },
    {
      "epoch": 0.8497288923315259,
      "grad_norm": 1.8809700012207031,
      "learning_rate": 5.578154720869117e-07,
      "loss": 0.026,
      "step": 16455
    },
    {
      "epoch": 0.8497805318874257,
      "grad_norm": 9.385041236877441,
      "learning_rate": 5.574394462029403e-07,
      "loss": 0.0353,
      "step": 16456
    },
    {
      "epoch": 0.8498321714433256,
      "grad_norm": 2.860640287399292,
      "learning_rate": 5.570635396198276e-07,
      "loss": 0.0278,
      "step": 16457
    },
    {
      "epoch": 0.8498838109992254,
      "grad_norm": 1.9979978799819946,
      "learning_rate": 5.566877523476666e-07,
      "loss": 0.02,
      "step": 16458
    },
    {
      "epoch": 0.8499354505551252,
      "grad_norm": 4.850889205932617,
      "learning_rate": 5.563120843965497e-07,
      "loss": 0.0251,
      "step": 16459
    },
    {
      "epoch": 0.849987090111025,
      "grad_norm": 3.116565227508545,
      "learning_rate": 5.559365357765645e-07,
      "loss": 0.0528,
      "step": 16460
    },
    {
      "epoch": 0.8500387296669248,
      "grad_norm": 10.842744827270508,
      "learning_rate": 5.555611064977967e-07,
      "loss": 0.1166,
      "step": 16461
    },
    {
      "epoch": 0.8500903692228247,
      "grad_norm": 3.274261474609375,
      "learning_rate": 5.551857965703272e-07,
      "loss": 0.0463,
      "step": 16462
    },
    {
      "epoch": 0.8501420087787245,
      "grad_norm": 9.351410865783691,
      "learning_rate": 5.548106060042369e-07,
      "loss": 0.0511,
      "step": 16463
    },
    {
      "epoch": 0.8501936483346243,
      "grad_norm": 3.346837043762207,
      "learning_rate": 5.54435534809597e-07,
      "loss": 0.0419,
      "step": 16464
    },
    {
      "epoch": 0.8502452878905241,
      "grad_norm": 5.3435773849487305,
      "learning_rate": 5.540605829964829e-07,
      "loss": 0.0311,
      "step": 16465
    },
    {
      "epoch": 0.8502969274464239,
      "grad_norm": 1.846593976020813,
      "learning_rate": 5.536857505749638e-07,
      "loss": 0.0335,
      "step": 16466
    },
    {
      "epoch": 0.8503485670023238,
      "grad_norm": 6.542073726654053,
      "learning_rate": 5.533110375551038e-07,
      "loss": 0.0478,
      "step": 16467
    },
    {
      "epoch": 0.8504002065582236,
      "grad_norm": 8.656474113464355,
      "learning_rate": 5.529364439469659e-07,
      "loss": 0.0728,
      "step": 16468
    },
    {
      "epoch": 0.8504518461141234,
      "grad_norm": 6.036065578460693,
      "learning_rate": 5.525619697606088e-07,
      "loss": 0.0535,
      "step": 16469
    },
    {
      "epoch": 0.8505034856700232,
      "grad_norm": 10.405008316040039,
      "learning_rate": 5.521876150060917e-07,
      "loss": 0.0669,
      "step": 16470
    },
    {
      "epoch": 0.850555125225923,
      "grad_norm": 3.3809962272644043,
      "learning_rate": 5.518133796934644e-07,
      "loss": 0.0266,
      "step": 16471
    },
    {
      "epoch": 0.8506067647818228,
      "grad_norm": 16.53380584716797,
      "learning_rate": 5.514392638327775e-07,
      "loss": 0.079,
      "step": 16472
    },
    {
      "epoch": 0.8506584043377227,
      "grad_norm": 2.955961227416992,
      "learning_rate": 5.510652674340777e-07,
      "loss": 0.0537,
      "step": 16473
    },
    {
      "epoch": 0.8507100438936225,
      "grad_norm": 5.263758659362793,
      "learning_rate": 5.50691390507408e-07,
      "loss": 0.0833,
      "step": 16474
    },
    {
      "epoch": 0.8507616834495223,
      "grad_norm": 2.122586727142334,
      "learning_rate": 5.503176330628096e-07,
      "loss": 0.0277,
      "step": 16475
    },
    {
      "epoch": 0.8508133230054221,
      "grad_norm": 2.052182197570801,
      "learning_rate": 5.49943995110318e-07,
      "loss": 0.0211,
      "step": 16476
    },
    {
      "epoch": 0.8508649625613219,
      "grad_norm": 2.4991977214813232,
      "learning_rate": 5.49570476659969e-07,
      "loss": 0.0318,
      "step": 16477
    },
    {
      "epoch": 0.8509166021172218,
      "grad_norm": 6.964813709259033,
      "learning_rate": 5.491970777217892e-07,
      "loss": 0.1464,
      "step": 16478
    },
    {
      "epoch": 0.8509682416731216,
      "grad_norm": 3.808091402053833,
      "learning_rate": 5.488237983058098e-07,
      "loss": 0.0492,
      "step": 16479
    },
    {
      "epoch": 0.8510198812290214,
      "grad_norm": 2.7852022647857666,
      "learning_rate": 5.484506384220533e-07,
      "loss": 0.0392,
      "step": 16480
    },
    {
      "epoch": 0.8510715207849212,
      "grad_norm": 3.874044895172119,
      "learning_rate": 5.480775980805414e-07,
      "loss": 0.0294,
      "step": 16481
    },
    {
      "epoch": 0.851123160340821,
      "grad_norm": 5.397866725921631,
      "learning_rate": 5.477046772912903e-07,
      "loss": 0.0286,
      "step": 16482
    },
    {
      "epoch": 0.8511747998967208,
      "grad_norm": 1.6064131259918213,
      "learning_rate": 5.473318760643137e-07,
      "loss": 0.0256,
      "step": 16483
    },
    {
      "epoch": 0.8512264394526207,
      "grad_norm": 2.445241928100586,
      "learning_rate": 5.469591944096259e-07,
      "loss": 0.0502,
      "step": 16484
    },
    {
      "epoch": 0.8512780790085205,
      "grad_norm": 2.286520004272461,
      "learning_rate": 5.465866323372343e-07,
      "loss": 0.039,
      "step": 16485
    },
    {
      "epoch": 0.8513297185644203,
      "grad_norm": 4.445042133331299,
      "learning_rate": 5.462141898571416e-07,
      "loss": 0.0523,
      "step": 16486
    },
    {
      "epoch": 0.8513813581203201,
      "grad_norm": 5.440056324005127,
      "learning_rate": 5.458418669793503e-07,
      "loss": 0.051,
      "step": 16487
    },
    {
      "epoch": 0.8514329976762199,
      "grad_norm": 4.335728645324707,
      "learning_rate": 5.45469663713859e-07,
      "loss": 0.0617,
      "step": 16488
    },
    {
      "epoch": 0.8514846372321198,
      "grad_norm": 2.442434072494507,
      "learning_rate": 5.450975800706632e-07,
      "loss": 0.0416,
      "step": 16489
    },
    {
      "epoch": 0.8515362767880196,
      "grad_norm": 3.087240695953369,
      "learning_rate": 5.447256160597542e-07,
      "loss": 0.0514,
      "step": 16490
    },
    {
      "epoch": 0.8515879163439194,
      "grad_norm": 3.743431329727173,
      "learning_rate": 5.443537716911212e-07,
      "loss": 0.0175,
      "step": 16491
    },
    {
      "epoch": 0.8516395558998192,
      "grad_norm": 12.125411033630371,
      "learning_rate": 5.439820469747497e-07,
      "loss": 0.0586,
      "step": 16492
    },
    {
      "epoch": 0.851691195455719,
      "grad_norm": 3.7397656440734863,
      "learning_rate": 5.436104419206223e-07,
      "loss": 0.0458,
      "step": 16493
    },
    {
      "epoch": 0.8517428350116188,
      "grad_norm": 3.85807204246521,
      "learning_rate": 5.432389565387175e-07,
      "loss": 0.0459,
      "step": 16494
    },
    {
      "epoch": 0.8517944745675187,
      "grad_norm": 4.6874542236328125,
      "learning_rate": 5.428675908390113e-07,
      "loss": 0.0742,
      "step": 16495
    },
    {
      "epoch": 0.8518461141234185,
      "grad_norm": 2.6159772872924805,
      "learning_rate": 5.42496344831478e-07,
      "loss": 0.0177,
      "step": 16496
    },
    {
      "epoch": 0.8518977536793183,
      "grad_norm": 5.08850622177124,
      "learning_rate": 5.421252185260828e-07,
      "loss": 0.0475,
      "step": 16497
    },
    {
      "epoch": 0.8519493932352181,
      "grad_norm": 4.341768741607666,
      "learning_rate": 5.417542119327968e-07,
      "loss": 0.034,
      "step": 16498
    },
    {
      "epoch": 0.852001032791118,
      "grad_norm": 7.388821125030518,
      "learning_rate": 5.413833250615813e-07,
      "loss": 0.044,
      "step": 16499
    },
    {
      "epoch": 0.8520526723470179,
      "grad_norm": 3.242774248123169,
      "learning_rate": 5.410125579223952e-07,
      "loss": 0.0329,
      "step": 16500
    },
    {
      "epoch": 0.8521043119029177,
      "grad_norm": 2.678351402282715,
      "learning_rate": 5.406419105251953e-07,
      "loss": 0.0474,
      "step": 16501
    },
    {
      "epoch": 0.8521559514588175,
      "grad_norm": 2.592022180557251,
      "learning_rate": 5.40271382879935e-07,
      "loss": 0.028,
      "step": 16502
    },
    {
      "epoch": 0.8522075910147173,
      "grad_norm": 2.928955554962158,
      "learning_rate": 5.399009749965667e-07,
      "loss": 0.0196,
      "step": 16503
    },
    {
      "epoch": 0.8522592305706171,
      "grad_norm": 7.465314865112305,
      "learning_rate": 5.395306868850341e-07,
      "loss": 0.0473,
      "step": 16504
    },
    {
      "epoch": 0.852310870126517,
      "grad_norm": 11.236137390136719,
      "learning_rate": 5.391605185552828e-07,
      "loss": 0.1377,
      "step": 16505
    },
    {
      "epoch": 0.8523625096824168,
      "grad_norm": 7.323098182678223,
      "learning_rate": 5.387904700172531e-07,
      "loss": 0.0725,
      "step": 16506
    },
    {
      "epoch": 0.8524141492383166,
      "grad_norm": 2.4170408248901367,
      "learning_rate": 5.384205412808823e-07,
      "loss": 0.0511,
      "step": 16507
    },
    {
      "epoch": 0.8524657887942164,
      "grad_norm": 1.9136606454849243,
      "learning_rate": 5.380507323561041e-07,
      "loss": 0.025,
      "step": 16508
    },
    {
      "epoch": 0.8525174283501162,
      "grad_norm": 2.093287467956543,
      "learning_rate": 5.376810432528495e-07,
      "loss": 0.0259,
      "step": 16509
    },
    {
      "epoch": 0.852569067906016,
      "grad_norm": 4.236186981201172,
      "learning_rate": 5.373114739810476e-07,
      "loss": 0.0258,
      "step": 16510
    },
    {
      "epoch": 0.8526207074619159,
      "grad_norm": 2.9381794929504395,
      "learning_rate": 5.369420245506196e-07,
      "loss": 0.0596,
      "step": 16511
    },
    {
      "epoch": 0.8526723470178157,
      "grad_norm": 3.8566150665283203,
      "learning_rate": 5.365726949714894e-07,
      "loss": 0.052,
      "step": 16512
    },
    {
      "epoch": 0.8527239865737155,
      "grad_norm": 2.2107150554656982,
      "learning_rate": 5.362034852535741e-07,
      "loss": 0.0318,
      "step": 16513
    },
    {
      "epoch": 0.8527756261296153,
      "grad_norm": 20.691612243652344,
      "learning_rate": 5.358343954067902e-07,
      "loss": 0.1253,
      "step": 16514
    },
    {
      "epoch": 0.8528272656855151,
      "grad_norm": 1.9363698959350586,
      "learning_rate": 5.354654254410463e-07,
      "loss": 0.0424,
      "step": 16515
    },
    {
      "epoch": 0.852878905241415,
      "grad_norm": 3.1243395805358887,
      "learning_rate": 5.350965753662513e-07,
      "loss": 0.0346,
      "step": 16516
    },
    {
      "epoch": 0.8529305447973148,
      "grad_norm": 3.4220340251922607,
      "learning_rate": 5.347278451923121e-07,
      "loss": 0.0399,
      "step": 16517
    },
    {
      "epoch": 0.8529821843532146,
      "grad_norm": 3.8593273162841797,
      "learning_rate": 5.343592349291304e-07,
      "loss": 0.0229,
      "step": 16518
    },
    {
      "epoch": 0.8530338239091144,
      "grad_norm": 2.511110782623291,
      "learning_rate": 5.339907445866039e-07,
      "loss": 0.0205,
      "step": 16519
    },
    {
      "epoch": 0.8530854634650142,
      "grad_norm": 3.5872936248779297,
      "learning_rate": 5.336223741746277e-07,
      "loss": 0.041,
      "step": 16520
    },
    {
      "epoch": 0.8531371030209141,
      "grad_norm": 2.4522643089294434,
      "learning_rate": 5.332541237030952e-07,
      "loss": 0.0434,
      "step": 16521
    },
    {
      "epoch": 0.8531887425768139,
      "grad_norm": 2.5625648498535156,
      "learning_rate": 5.328859931818947e-07,
      "loss": 0.0349,
      "step": 16522
    },
    {
      "epoch": 0.8532403821327137,
      "grad_norm": 13.0870361328125,
      "learning_rate": 5.32517982620912e-07,
      "loss": 0.0566,
      "step": 16523
    },
    {
      "epoch": 0.8532920216886135,
      "grad_norm": 2.4248390197753906,
      "learning_rate": 5.321500920300305e-07,
      "loss": 0.0392,
      "step": 16524
    },
    {
      "epoch": 0.8533436612445133,
      "grad_norm": 3.5525543689727783,
      "learning_rate": 5.317823214191281e-07,
      "loss": 0.0457,
      "step": 16525
    },
    {
      "epoch": 0.8533953008004131,
      "grad_norm": 5.377305030822754,
      "learning_rate": 5.314146707980828e-07,
      "loss": 0.0445,
      "step": 16526
    },
    {
      "epoch": 0.853446940356313,
      "grad_norm": 3.0311994552612305,
      "learning_rate": 5.310471401767658e-07,
      "loss": 0.0399,
      "step": 16527
    },
    {
      "epoch": 0.8534985799122128,
      "grad_norm": 1.7035739421844482,
      "learning_rate": 5.306797295650479e-07,
      "loss": 0.0227,
      "step": 16528
    },
    {
      "epoch": 0.8535502194681126,
      "grad_norm": 7.872087001800537,
      "learning_rate": 5.303124389727959e-07,
      "loss": 0.0784,
      "step": 16529
    },
    {
      "epoch": 0.8536018590240124,
      "grad_norm": 3.4153201580047607,
      "learning_rate": 5.299452684098711e-07,
      "loss": 0.0251,
      "step": 16530
    },
    {
      "epoch": 0.8536534985799122,
      "grad_norm": 4.577719211578369,
      "learning_rate": 5.295782178861347e-07,
      "loss": 0.0807,
      "step": 16531
    },
    {
      "epoch": 0.8537051381358121,
      "grad_norm": 21.349102020263672,
      "learning_rate": 5.292112874114452e-07,
      "loss": 0.0838,
      "step": 16532
    },
    {
      "epoch": 0.8537567776917119,
      "grad_norm": 4.079252243041992,
      "learning_rate": 5.288444769956536e-07,
      "loss": 0.1152,
      "step": 16533
    },
    {
      "epoch": 0.8538084172476117,
      "grad_norm": 3.574414014816284,
      "learning_rate": 5.284777866486112e-07,
      "loss": 0.048,
      "step": 16534
    },
    {
      "epoch": 0.8538600568035115,
      "grad_norm": 2.5506224632263184,
      "learning_rate": 5.281112163801638e-07,
      "loss": 0.0234,
      "step": 16535
    },
    {
      "epoch": 0.8539116963594113,
      "grad_norm": 4.3634419441223145,
      "learning_rate": 5.277447662001589e-07,
      "loss": 0.0649,
      "step": 16536
    },
    {
      "epoch": 0.8539633359153112,
      "grad_norm": 2.7339932918548584,
      "learning_rate": 5.273784361184342e-07,
      "loss": 0.0328,
      "step": 16537
    },
    {
      "epoch": 0.854014975471211,
      "grad_norm": 3.8775012493133545,
      "learning_rate": 5.270122261448274e-07,
      "loss": 0.0641,
      "step": 16538
    },
    {
      "epoch": 0.8540666150271108,
      "grad_norm": 2.4323339462280273,
      "learning_rate": 5.266461362891734e-07,
      "loss": 0.0281,
      "step": 16539
    },
    {
      "epoch": 0.8541182545830106,
      "grad_norm": 3.3914289474487305,
      "learning_rate": 5.262801665613032e-07,
      "loss": 0.0372,
      "step": 16540
    },
    {
      "epoch": 0.8541698941389104,
      "grad_norm": 5.169095039367676,
      "learning_rate": 5.259143169710446e-07,
      "loss": 0.0597,
      "step": 16541
    },
    {
      "epoch": 0.8542215336948102,
      "grad_norm": 4.680836200714111,
      "learning_rate": 5.255485875282212e-07,
      "loss": 0.0537,
      "step": 16542
    },
    {
      "epoch": 0.8542731732507101,
      "grad_norm": 3.4827044010162354,
      "learning_rate": 5.251829782426566e-07,
      "loss": 0.0612,
      "step": 16543
    },
    {
      "epoch": 0.8543248128066099,
      "grad_norm": 6.663698196411133,
      "learning_rate": 5.248174891241653e-07,
      "loss": 0.0637,
      "step": 16544
    },
    {
      "epoch": 0.8543764523625097,
      "grad_norm": 10.397427558898926,
      "learning_rate": 5.244521201825653e-07,
      "loss": 0.1128,
      "step": 16545
    },
    {
      "epoch": 0.8544280919184095,
      "grad_norm": 3.4290761947631836,
      "learning_rate": 5.24086871427667e-07,
      "loss": 0.054,
      "step": 16546
    },
    {
      "epoch": 0.8544797314743093,
      "grad_norm": 2.932399272918701,
      "learning_rate": 5.237217428692803e-07,
      "loss": 0.0324,
      "step": 16547
    },
    {
      "epoch": 0.8545313710302092,
      "grad_norm": 4.025030136108398,
      "learning_rate": 5.233567345172081e-07,
      "loss": 0.0678,
      "step": 16548
    },
    {
      "epoch": 0.854583010586109,
      "grad_norm": 1.5140774250030518,
      "learning_rate": 5.229918463812516e-07,
      "loss": 0.0169,
      "step": 16549
    },
    {
      "epoch": 0.8546346501420088,
      "grad_norm": 3.044654130935669,
      "learning_rate": 5.226270784712134e-07,
      "loss": 0.0311,
      "step": 16550
    },
    {
      "epoch": 0.8546862896979086,
      "grad_norm": 3.737724542617798,
      "learning_rate": 5.222624307968871e-07,
      "loss": 0.0759,
      "step": 16551
    },
    {
      "epoch": 0.8547379292538084,
      "grad_norm": 20.825328826904297,
      "learning_rate": 5.218979033680643e-07,
      "loss": 0.1499,
      "step": 16552
    },
    {
      "epoch": 0.8547895688097082,
      "grad_norm": 4.020538330078125,
      "learning_rate": 5.21533496194534e-07,
      "loss": 0.0519,
      "step": 16553
    },
    {
      "epoch": 0.8548412083656081,
      "grad_norm": 1.2417360544204712,
      "learning_rate": 5.211692092860831e-07,
      "loss": 0.0089,
      "step": 16554
    },
    {
      "epoch": 0.8548928479215079,
      "grad_norm": 20.347904205322266,
      "learning_rate": 5.208050426524935e-07,
      "loss": 0.1532,
      "step": 16555
    },
    {
      "epoch": 0.8549444874774077,
      "grad_norm": 3.805781602859497,
      "learning_rate": 5.204409963035445e-07,
      "loss": 0.0586,
      "step": 16556
    },
    {
      "epoch": 0.8549961270333075,
      "grad_norm": 2.756845235824585,
      "learning_rate": 5.200770702490127e-07,
      "loss": 0.022,
      "step": 16557
    },
    {
      "epoch": 0.8550477665892073,
      "grad_norm": 3.924678087234497,
      "learning_rate": 5.197132644986702e-07,
      "loss": 0.0317,
      "step": 16558
    },
    {
      "epoch": 0.8550994061451072,
      "grad_norm": 3.88134765625,
      "learning_rate": 5.193495790622876e-07,
      "loss": 0.0316,
      "step": 16559
    },
    {
      "epoch": 0.855151045701007,
      "grad_norm": 8.272208213806152,
      "learning_rate": 5.189860139496311e-07,
      "loss": 0.0758,
      "step": 16560
    },
    {
      "epoch": 0.8552026852569068,
      "grad_norm": 6.664799690246582,
      "learning_rate": 5.186225691704633e-07,
      "loss": 0.0865,
      "step": 16561
    },
    {
      "epoch": 0.8552543248128066,
      "grad_norm": 8.891679763793945,
      "learning_rate": 5.182592447345459e-07,
      "loss": 0.0899,
      "step": 16562
    },
    {
      "epoch": 0.8553059643687064,
      "grad_norm": 3.47568416595459,
      "learning_rate": 5.178960406516321e-07,
      "loss": 0.0786,
      "step": 16563
    },
    {
      "epoch": 0.8553576039246062,
      "grad_norm": 10.71408748626709,
      "learning_rate": 5.175329569314786e-07,
      "loss": 0.0983,
      "step": 16564
    },
    {
      "epoch": 0.8554092434805061,
      "grad_norm": 3.9356002807617188,
      "learning_rate": 5.171699935838359e-07,
      "loss": 0.0395,
      "step": 16565
    },
    {
      "epoch": 0.8554608830364059,
      "grad_norm": 5.639175891876221,
      "learning_rate": 5.168071506184491e-07,
      "loss": 0.1159,
      "step": 16566
    },
    {
      "epoch": 0.8555125225923057,
      "grad_norm": 5.462729454040527,
      "learning_rate": 5.164444280450625e-07,
      "loss": 0.0752,
      "step": 16567
    },
    {
      "epoch": 0.8555641621482055,
      "grad_norm": 3.6346962451934814,
      "learning_rate": 5.160818258734163e-07,
      "loss": 0.0427,
      "step": 16568
    },
    {
      "epoch": 0.8556158017041053,
      "grad_norm": 3.0680534839630127,
      "learning_rate": 5.157193441132502e-07,
      "loss": 0.0488,
      "step": 16569
    },
    {
      "epoch": 0.8556674412600052,
      "grad_norm": 12.442106246948242,
      "learning_rate": 5.153569827742955e-07,
      "loss": 0.0352,
      "step": 16570
    },
    {
      "epoch": 0.855719080815905,
      "grad_norm": 4.034297943115234,
      "learning_rate": 5.149947418662843e-07,
      "loss": 0.0309,
      "step": 16571
    },
    {
      "epoch": 0.8557707203718048,
      "grad_norm": 5.1453399658203125,
      "learning_rate": 5.146326213989444e-07,
      "loss": 0.0446,
      "step": 16572
    },
    {
      "epoch": 0.8558223599277046,
      "grad_norm": 3.0736494064331055,
      "learning_rate": 5.142706213819998e-07,
      "loss": 0.0282,
      "step": 16573
    },
    {
      "epoch": 0.8558739994836044,
      "grad_norm": 3.706387758255005,
      "learning_rate": 5.139087418251715e-07,
      "loss": 0.031,
      "step": 16574
    },
    {
      "epoch": 0.8559256390395042,
      "grad_norm": 3.5661232471466064,
      "learning_rate": 5.135469827381784e-07,
      "loss": 0.0585,
      "step": 16575
    },
    {
      "epoch": 0.8559772785954041,
      "grad_norm": 4.522674083709717,
      "learning_rate": 5.13185344130735e-07,
      "loss": 0.0689,
      "step": 16576
    },
    {
      "epoch": 0.8560289181513039,
      "grad_norm": 12.59644889831543,
      "learning_rate": 5.128238260125506e-07,
      "loss": 0.2074,
      "step": 16577
    },
    {
      "epoch": 0.8560805577072037,
      "grad_norm": 5.120115756988525,
      "learning_rate": 5.124624283933366e-07,
      "loss": 0.0469,
      "step": 16578
    },
    {
      "epoch": 0.8561321972631035,
      "grad_norm": 7.957476615905762,
      "learning_rate": 5.121011512827961e-07,
      "loss": 0.0903,
      "step": 16579
    },
    {
      "epoch": 0.8561838368190033,
      "grad_norm": 3.479851722717285,
      "learning_rate": 5.117399946906326e-07,
      "loss": 0.0635,
      "step": 16580
    },
    {
      "epoch": 0.8562354763749032,
      "grad_norm": 3.5189011096954346,
      "learning_rate": 5.113789586265422e-07,
      "loss": 0.0523,
      "step": 16581
    },
    {
      "epoch": 0.856287115930803,
      "grad_norm": 2.9341847896575928,
      "learning_rate": 5.110180431002204e-07,
      "loss": 0.0433,
      "step": 16582
    },
    {
      "epoch": 0.8563387554867028,
      "grad_norm": 9.344758987426758,
      "learning_rate": 5.106572481213612e-07,
      "loss": 0.1457,
      "step": 16583
    },
    {
      "epoch": 0.8563903950426026,
      "grad_norm": 1.2200453281402588,
      "learning_rate": 5.102965736996534e-07,
      "loss": 0.0138,
      "step": 16584
    },
    {
      "epoch": 0.8564420345985024,
      "grad_norm": 3.264932632446289,
      "learning_rate": 5.099360198447806e-07,
      "loss": 0.0546,
      "step": 16585
    },
    {
      "epoch": 0.8564936741544023,
      "grad_norm": 6.923620223999023,
      "learning_rate": 5.095755865664265e-07,
      "loss": 0.1007,
      "step": 16586
    },
    {
      "epoch": 0.8565453137103021,
      "grad_norm": 4.50356912612915,
      "learning_rate": 5.092152738742695e-07,
      "loss": 0.0533,
      "step": 16587
    },
    {
      "epoch": 0.8565969532662019,
      "grad_norm": 8.96914291381836,
      "learning_rate": 5.088550817779858e-07,
      "loss": 0.0737,
      "step": 16588
    },
    {
      "epoch": 0.8566485928221017,
      "grad_norm": 32.21444320678711,
      "learning_rate": 5.084950102872482e-07,
      "loss": 0.0546,
      "step": 16589
    },
    {
      "epoch": 0.8567002323780015,
      "grad_norm": 2.4370908737182617,
      "learning_rate": 5.081350594117263e-07,
      "loss": 0.0335,
      "step": 16590
    },
    {
      "epoch": 0.8567518719339013,
      "grad_norm": 3.1432862281799316,
      "learning_rate": 5.077752291610854e-07,
      "loss": 0.0231,
      "step": 16591
    },
    {
      "epoch": 0.8568035114898012,
      "grad_norm": 11.231073379516602,
      "learning_rate": 5.074155195449887e-07,
      "loss": 0.0342,
      "step": 16592
    },
    {
      "epoch": 0.856855151045701,
      "grad_norm": 4.790775775909424,
      "learning_rate": 5.070559305730966e-07,
      "loss": 0.0345,
      "step": 16593
    },
    {
      "epoch": 0.8569067906016008,
      "grad_norm": 3.4588286876678467,
      "learning_rate": 5.066964622550646e-07,
      "loss": 0.0491,
      "step": 16594
    },
    {
      "epoch": 0.8569584301575006,
      "grad_norm": 14.963730812072754,
      "learning_rate": 5.063371146005475e-07,
      "loss": 0.0671,
      "step": 16595
    },
    {
      "epoch": 0.8570100697134004,
      "grad_norm": 4.5651421546936035,
      "learning_rate": 5.059778876191924e-07,
      "loss": 0.0472,
      "step": 16596
    },
    {
      "epoch": 0.8570617092693003,
      "grad_norm": 6.204768180847168,
      "learning_rate": 5.056187813206482e-07,
      "loss": 0.0388,
      "step": 16597
    },
    {
      "epoch": 0.8571133488252001,
      "grad_norm": 3.7967851161956787,
      "learning_rate": 5.052597957145583e-07,
      "loss": 0.0455,
      "step": 16598
    },
    {
      "epoch": 0.8571649883810999,
      "grad_norm": 4.0359416007995605,
      "learning_rate": 5.049009308105618e-07,
      "loss": 0.0631,
      "step": 16599
    },
    {
      "epoch": 0.8572166279369997,
      "grad_norm": 2.0657448768615723,
      "learning_rate": 5.045421866182965e-07,
      "loss": 0.0241,
      "step": 16600
    },
    {
      "epoch": 0.8572682674928995,
      "grad_norm": 14.35552978515625,
      "learning_rate": 5.041835631473951e-07,
      "loss": 0.1146,
      "step": 16601
    },
    {
      "epoch": 0.8573199070487993,
      "grad_norm": 22.35924530029297,
      "learning_rate": 5.038250604074907e-07,
      "loss": 0.1659,
      "step": 16602
    },
    {
      "epoch": 0.8573715466046992,
      "grad_norm": 2.0093302726745605,
      "learning_rate": 5.034666784082082e-07,
      "loss": 0.0274,
      "step": 16603
    },
    {
      "epoch": 0.857423186160599,
      "grad_norm": 4.395391941070557,
      "learning_rate": 5.031084171591716e-07,
      "loss": 0.0469,
      "step": 16604
    },
    {
      "epoch": 0.8574748257164988,
      "grad_norm": 3.4904446601867676,
      "learning_rate": 5.027502766700027e-07,
      "loss": 0.0446,
      "step": 16605
    },
    {
      "epoch": 0.8575264652723986,
      "grad_norm": 3.3267202377319336,
      "learning_rate": 5.023922569503187e-07,
      "loss": 0.0185,
      "step": 16606
    },
    {
      "epoch": 0.8575781048282984,
      "grad_norm": 26.126399993896484,
      "learning_rate": 5.020343580097336e-07,
      "loss": 0.127,
      "step": 16607
    },
    {
      "epoch": 0.8576297443841983,
      "grad_norm": 1.671340823173523,
      "learning_rate": 5.016765798578588e-07,
      "loss": 0.0287,
      "step": 16608
    },
    {
      "epoch": 0.8576813839400981,
      "grad_norm": 3.803532361984253,
      "learning_rate": 5.01318922504303e-07,
      "loss": 0.053,
      "step": 16609
    },
    {
      "epoch": 0.8577330234959979,
      "grad_norm": 5.221918106079102,
      "learning_rate": 5.009613859586681e-07,
      "loss": 0.0614,
      "step": 16610
    },
    {
      "epoch": 0.8577846630518977,
      "grad_norm": 3.8506112098693848,
      "learning_rate": 5.006039702305576e-07,
      "loss": 0.0313,
      "step": 16611
    },
    {
      "epoch": 0.8578363026077975,
      "grad_norm": 2.1407601833343506,
      "learning_rate": 5.002466753295693e-07,
      "loss": 0.0506,
      "step": 16612
    },
    {
      "epoch": 0.8578879421636973,
      "grad_norm": 5.555994987487793,
      "learning_rate": 4.998895012652982e-07,
      "loss": 0.0744,
      "step": 16613
    },
    {
      "epoch": 0.8579395817195972,
      "grad_norm": 3.1279776096343994,
      "learning_rate": 4.995324480473352e-07,
      "loss": 0.051,
      "step": 16614
    },
    {
      "epoch": 0.857991221275497,
      "grad_norm": 3.4505879878997803,
      "learning_rate": 4.991755156852674e-07,
      "loss": 0.0416,
      "step": 16615
    },
    {
      "epoch": 0.8580428608313968,
      "grad_norm": 29.37928009033203,
      "learning_rate": 4.988187041886822e-07,
      "loss": 0.116,
      "step": 16616
    },
    {
      "epoch": 0.8580945003872966,
      "grad_norm": 8.152345657348633,
      "learning_rate": 4.98462013567162e-07,
      "loss": 0.0222,
      "step": 16617
    },
    {
      "epoch": 0.8581461399431964,
      "grad_norm": 3.773312568664551,
      "learning_rate": 4.981054438302823e-07,
      "loss": 0.0462,
      "step": 16618
    },
    {
      "epoch": 0.8581977794990963,
      "grad_norm": 2.0522093772888184,
      "learning_rate": 4.977489949876208e-07,
      "loss": 0.0371,
      "step": 16619
    },
    {
      "epoch": 0.8582494190549961,
      "grad_norm": 6.622629165649414,
      "learning_rate": 4.973926670487489e-07,
      "loss": 0.0672,
      "step": 16620
    },
    {
      "epoch": 0.8583010586108959,
      "grad_norm": 6.226404190063477,
      "learning_rate": 4.970364600232353e-07,
      "loss": 0.0489,
      "step": 16621
    },
    {
      "epoch": 0.8583526981667957,
      "grad_norm": 1.7187018394470215,
      "learning_rate": 4.96680373920646e-07,
      "loss": 0.0432,
      "step": 16622
    },
    {
      "epoch": 0.8584043377226955,
      "grad_norm": 4.003864288330078,
      "learning_rate": 4.963244087505431e-07,
      "loss": 0.035,
      "step": 16623
    },
    {
      "epoch": 0.8584559772785954,
      "grad_norm": 4.255549430847168,
      "learning_rate": 4.959685645224854e-07,
      "loss": 0.0422,
      "step": 16624
    },
    {
      "epoch": 0.8585076168344952,
      "grad_norm": 8.438992500305176,
      "learning_rate": 4.956128412460298e-07,
      "loss": 0.045,
      "step": 16625
    },
    {
      "epoch": 0.8585592563903951,
      "grad_norm": 3.9533298015594482,
      "learning_rate": 4.952572389307281e-07,
      "loss": 0.0317,
      "step": 16626
    },
    {
      "epoch": 0.8586108959462949,
      "grad_norm": 2.849574089050293,
      "learning_rate": 4.949017575861298e-07,
      "loss": 0.0406,
      "step": 16627
    },
    {
      "epoch": 0.8586625355021947,
      "grad_norm": 2.968151092529297,
      "learning_rate": 4.945463972217818e-07,
      "loss": 0.0179,
      "step": 16628
    },
    {
      "epoch": 0.8587141750580946,
      "grad_norm": 2.5393738746643066,
      "learning_rate": 4.941911578472241e-07,
      "loss": 0.0433,
      "step": 16629
    },
    {
      "epoch": 0.8587658146139944,
      "grad_norm": 1.6754049062728882,
      "learning_rate": 4.938360394719999e-07,
      "loss": 0.0259,
      "step": 16630
    },
    {
      "epoch": 0.8588174541698942,
      "grad_norm": 3.392956256866455,
      "learning_rate": 4.934810421056452e-07,
      "loss": 0.0367,
      "step": 16631
    },
    {
      "epoch": 0.858869093725794,
      "grad_norm": 6.024367809295654,
      "learning_rate": 4.931261657576908e-07,
      "loss": 0.0477,
      "step": 16632
    },
    {
      "epoch": 0.8589207332816938,
      "grad_norm": 3.4769840240478516,
      "learning_rate": 4.927714104376685e-07,
      "loss": 0.0455,
      "step": 16633
    },
    {
      "epoch": 0.8589723728375936,
      "grad_norm": 1.8616935014724731,
      "learning_rate": 4.924167761551024e-07,
      "loss": 0.0256,
      "step": 16634
    },
    {
      "epoch": 0.8590240123934935,
      "grad_norm": 2.1479806900024414,
      "learning_rate": 4.920622629195205e-07,
      "loss": 0.0297,
      "step": 16635
    },
    {
      "epoch": 0.8590756519493933,
      "grad_norm": 3.0846052169799805,
      "learning_rate": 4.917078707404388e-07,
      "loss": 0.0318,
      "step": 16636
    },
    {
      "epoch": 0.8591272915052931,
      "grad_norm": 4.507682800292969,
      "learning_rate": 4.913535996273755e-07,
      "loss": 0.0258,
      "step": 16637
    },
    {
      "epoch": 0.8591789310611929,
      "grad_norm": 1.9754265546798706,
      "learning_rate": 4.90999449589844e-07,
      "loss": 0.0338,
      "step": 16638
    },
    {
      "epoch": 0.8592305706170927,
      "grad_norm": 2.958381175994873,
      "learning_rate": 4.906454206373557e-07,
      "loss": 0.0321,
      "step": 16639
    },
    {
      "epoch": 0.8592822101729926,
      "grad_norm": 2.445643663406372,
      "learning_rate": 4.902915127794161e-07,
      "loss": 0.0382,
      "step": 16640
    },
    {
      "epoch": 0.8593338497288924,
      "grad_norm": 6.090866565704346,
      "learning_rate": 4.899377260255305e-07,
      "loss": 0.0197,
      "step": 16641
    },
    {
      "epoch": 0.8593854892847922,
      "grad_norm": 3.949800968170166,
      "learning_rate": 4.895840603851998e-07,
      "loss": 0.0302,
      "step": 16642
    },
    {
      "epoch": 0.859437128840692,
      "grad_norm": 1.242863416671753,
      "learning_rate": 4.892305158679189e-07,
      "loss": 0.0179,
      "step": 16643
    },
    {
      "epoch": 0.8594887683965918,
      "grad_norm": 10.0715913772583,
      "learning_rate": 4.888770924831842e-07,
      "loss": 0.1045,
      "step": 16644
    },
    {
      "epoch": 0.8595404079524916,
      "grad_norm": 6.485569000244141,
      "learning_rate": 4.885237902404861e-07,
      "loss": 0.0585,
      "step": 16645
    },
    {
      "epoch": 0.8595920475083915,
      "grad_norm": 2.8468854427337646,
      "learning_rate": 4.881706091493122e-07,
      "loss": 0.0373,
      "step": 16646
    },
    {
      "epoch": 0.8596436870642913,
      "grad_norm": 6.573451519012451,
      "learning_rate": 4.878175492191461e-07,
      "loss": 0.0493,
      "step": 16647
    },
    {
      "epoch": 0.8596953266201911,
      "grad_norm": 13.898801803588867,
      "learning_rate": 4.874646104594688e-07,
      "loss": 0.0561,
      "step": 16648
    },
    {
      "epoch": 0.8597469661760909,
      "grad_norm": 72.6748046875,
      "learning_rate": 4.871117928797597e-07,
      "loss": 0.2059,
      "step": 16649
    },
    {
      "epoch": 0.8597986057319907,
      "grad_norm": 2.9855384826660156,
      "learning_rate": 4.86759096489493e-07,
      "loss": 0.0375,
      "step": 16650
    },
    {
      "epoch": 0.8598502452878906,
      "grad_norm": 4.339373588562012,
      "learning_rate": 4.864065212981389e-07,
      "loss": 0.0695,
      "step": 16651
    },
    {
      "epoch": 0.8599018848437904,
      "grad_norm": 2.325944185256958,
      "learning_rate": 4.86054067315166e-07,
      "loss": 0.0318,
      "step": 16652
    },
    {
      "epoch": 0.8599535243996902,
      "grad_norm": 2.9630367755889893,
      "learning_rate": 4.857017345500392e-07,
      "loss": 0.0425,
      "step": 16653
    },
    {
      "epoch": 0.86000516395559,
      "grad_norm": 6.707921504974365,
      "learning_rate": 4.853495230122201e-07,
      "loss": 0.0517,
      "step": 16654
    },
    {
      "epoch": 0.8600568035114898,
      "grad_norm": 22.848833084106445,
      "learning_rate": 4.849974327111673e-07,
      "loss": 0.0804,
      "step": 16655
    },
    {
      "epoch": 0.8601084430673896,
      "grad_norm": 7.874556064605713,
      "learning_rate": 4.846454636563353e-07,
      "loss": 0.0466,
      "step": 16656
    },
    {
      "epoch": 0.8601600826232895,
      "grad_norm": 14.141327857971191,
      "learning_rate": 4.842936158571759e-07,
      "loss": 0.0559,
      "step": 16657
    },
    {
      "epoch": 0.8602117221791893,
      "grad_norm": 2.104173183441162,
      "learning_rate": 4.839418893231385e-07,
      "loss": 0.0266,
      "step": 16658
    },
    {
      "epoch": 0.8602633617350891,
      "grad_norm": 2.873124122619629,
      "learning_rate": 4.835902840636675e-07,
      "loss": 0.0503,
      "step": 16659
    },
    {
      "epoch": 0.8603150012909889,
      "grad_norm": 16.676876068115234,
      "learning_rate": 4.83238800088205e-07,
      "loss": 0.1358,
      "step": 16660
    },
    {
      "epoch": 0.8603666408468887,
      "grad_norm": 7.275219440460205,
      "learning_rate": 4.828874374061915e-07,
      "loss": 0.0663,
      "step": 16661
    },
    {
      "epoch": 0.8604182804027886,
      "grad_norm": 3.6931774616241455,
      "learning_rate": 4.825361960270586e-07,
      "loss": 0.0527,
      "step": 16662
    },
    {
      "epoch": 0.8604699199586884,
      "grad_norm": 6.616250514984131,
      "learning_rate": 4.82185075960242e-07,
      "loss": 0.0943,
      "step": 16663
    },
    {
      "epoch": 0.8605215595145882,
      "grad_norm": 9.98566722869873,
      "learning_rate": 4.818340772151714e-07,
      "loss": 0.082,
      "step": 16664
    },
    {
      "epoch": 0.860573199070488,
      "grad_norm": 13.727246284484863,
      "learning_rate": 4.814831998012692e-07,
      "loss": 0.0638,
      "step": 16665
    },
    {
      "epoch": 0.8606248386263878,
      "grad_norm": 3.818863868713379,
      "learning_rate": 4.811324437279597e-07,
      "loss": 0.0511,
      "step": 16666
    },
    {
      "epoch": 0.8606764781822877,
      "grad_norm": 5.398280143737793,
      "learning_rate": 4.807818090046607e-07,
      "loss": 0.071,
      "step": 16667
    },
    {
      "epoch": 0.8607281177381875,
      "grad_norm": 9.001436233520508,
      "learning_rate": 4.80431295640792e-07,
      "loss": 0.0846,
      "step": 16668
    },
    {
      "epoch": 0.8607797572940873,
      "grad_norm": 2.321868896484375,
      "learning_rate": 4.800809036457621e-07,
      "loss": 0.0368,
      "step": 16669
    },
    {
      "epoch": 0.8608313968499871,
      "grad_norm": 7.260344982147217,
      "learning_rate": 4.797306330289825e-07,
      "loss": 0.0288,
      "step": 16670
    },
    {
      "epoch": 0.8608830364058869,
      "grad_norm": 10.926810264587402,
      "learning_rate": 4.793804837998589e-07,
      "loss": 0.1393,
      "step": 16671
    },
    {
      "epoch": 0.8609346759617867,
      "grad_norm": 2.8736963272094727,
      "learning_rate": 4.790304559677944e-07,
      "loss": 0.0397,
      "step": 16672
    },
    {
      "epoch": 0.8609863155176866,
      "grad_norm": 2.570347547531128,
      "learning_rate": 4.786805495421887e-07,
      "loss": 0.0454,
      "step": 16673
    },
    {
      "epoch": 0.8610379550735864,
      "grad_norm": 4.609570026397705,
      "learning_rate": 4.783307645324381e-07,
      "loss": 0.0769,
      "step": 16674
    },
    {
      "epoch": 0.8610895946294862,
      "grad_norm": 3.371208667755127,
      "learning_rate": 4.779811009479368e-07,
      "loss": 0.0514,
      "step": 16675
    },
    {
      "epoch": 0.861141234185386,
      "grad_norm": 2.5576658248901367,
      "learning_rate": 4.776315587980712e-07,
      "loss": 0.029,
      "step": 16676
    },
    {
      "epoch": 0.8611928737412858,
      "grad_norm": 5.547216892242432,
      "learning_rate": 4.772821380922321e-07,
      "loss": 0.0486,
      "step": 16677
    },
    {
      "epoch": 0.8612445132971857,
      "grad_norm": 1.7045949697494507,
      "learning_rate": 4.769328388398009e-07,
      "loss": 0.0138,
      "step": 16678
    },
    {
      "epoch": 0.8612961528530855,
      "grad_norm": 2.568408727645874,
      "learning_rate": 4.7658366105015895e-07,
      "loss": 0.0324,
      "step": 16679
    },
    {
      "epoch": 0.8613477924089853,
      "grad_norm": 2.451098680496216,
      "learning_rate": 4.76234604732681e-07,
      "loss": 0.0579,
      "step": 16680
    },
    {
      "epoch": 0.8613994319648851,
      "grad_norm": 4.245039939880371,
      "learning_rate": 4.7588566989674067e-07,
      "loss": 0.0627,
      "step": 16681
    },
    {
      "epoch": 0.8614510715207849,
      "grad_norm": 3.655123472213745,
      "learning_rate": 4.75536856551711e-07,
      "loss": 0.0245,
      "step": 16682
    },
    {
      "epoch": 0.8615027110766847,
      "grad_norm": 3.4873976707458496,
      "learning_rate": 4.7518816470695786e-07,
      "loss": 0.0589,
      "step": 16683
    },
    {
      "epoch": 0.8615543506325846,
      "grad_norm": 4.933497905731201,
      "learning_rate": 4.748395943718437e-07,
      "loss": 0.0488,
      "step": 16684
    },
    {
      "epoch": 0.8616059901884844,
      "grad_norm": 2.950587749481201,
      "learning_rate": 4.7449114555572993e-07,
      "loss": 0.03,
      "step": 16685
    },
    {
      "epoch": 0.8616576297443842,
      "grad_norm": 9.60216236114502,
      "learning_rate": 4.741428182679736e-07,
      "loss": 0.0464,
      "step": 16686
    },
    {
      "epoch": 0.861709269300284,
      "grad_norm": 2.5745129585266113,
      "learning_rate": 4.7379461251792937e-07,
      "loss": 0.0282,
      "step": 16687
    },
    {
      "epoch": 0.8617609088561838,
      "grad_norm": 3.615579843521118,
      "learning_rate": 4.7344652831494755e-07,
      "loss": 0.0879,
      "step": 16688
    },
    {
      "epoch": 0.8618125484120837,
      "grad_norm": 13.245875358581543,
      "learning_rate": 4.7309856566837565e-07,
      "loss": 0.047,
      "step": 16689
    },
    {
      "epoch": 0.8618641879679835,
      "grad_norm": 7.977644443511963,
      "learning_rate": 4.727507245875579e-07,
      "loss": 0.0659,
      "step": 16690
    },
    {
      "epoch": 0.8619158275238833,
      "grad_norm": 8.204174995422363,
      "learning_rate": 4.724030050818351e-07,
      "loss": 0.0858,
      "step": 16691
    },
    {
      "epoch": 0.8619674670797831,
      "grad_norm": 13.345893859863281,
      "learning_rate": 4.7205540716054543e-07,
      "loss": 0.039,
      "step": 16692
    },
    {
      "epoch": 0.8620191066356829,
      "grad_norm": 5.022761821746826,
      "learning_rate": 4.717079308330225e-07,
      "loss": 0.0902,
      "step": 16693
    },
    {
      "epoch": 0.8620707461915827,
      "grad_norm": 4.642275810241699,
      "learning_rate": 4.713605761085993e-07,
      "loss": 0.0444,
      "step": 16694
    },
    {
      "epoch": 0.8621223857474826,
      "grad_norm": 2.992044687271118,
      "learning_rate": 4.7101334299660015e-07,
      "loss": 0.0383,
      "step": 16695
    },
    {
      "epoch": 0.8621740253033824,
      "grad_norm": 7.611908435821533,
      "learning_rate": 4.706662315063532e-07,
      "loss": 0.1924,
      "step": 16696
    },
    {
      "epoch": 0.8622256648592822,
      "grad_norm": 2.9976375102996826,
      "learning_rate": 4.703192416471786e-07,
      "loss": 0.0324,
      "step": 16697
    },
    {
      "epoch": 0.862277304415182,
      "grad_norm": 5.712447166442871,
      "learning_rate": 4.6997237342839395e-07,
      "loss": 0.0702,
      "step": 16698
    },
    {
      "epoch": 0.8623289439710818,
      "grad_norm": 3.092747926712036,
      "learning_rate": 4.696256268593147e-07,
      "loss": 0.0292,
      "step": 16699
    },
    {
      "epoch": 0.8623805835269817,
      "grad_norm": 2.9260687828063965,
      "learning_rate": 4.6927900194925104e-07,
      "loss": 0.027,
      "step": 16700
    },
    {
      "epoch": 0.8624322230828815,
      "grad_norm": 14.35085678100586,
      "learning_rate": 4.689324987075139e-07,
      "loss": 0.1432,
      "step": 16701
    },
    {
      "epoch": 0.8624838626387813,
      "grad_norm": 6.663510322570801,
      "learning_rate": 4.6858611714340584e-07,
      "loss": 0.0444,
      "step": 16702
    },
    {
      "epoch": 0.8625355021946811,
      "grad_norm": 9.029451370239258,
      "learning_rate": 4.6823985726623e-07,
      "loss": 0.042,
      "step": 16703
    },
    {
      "epoch": 0.8625871417505809,
      "grad_norm": 4.805056095123291,
      "learning_rate": 4.678937190852839e-07,
      "loss": 0.0432,
      "step": 16704
    },
    {
      "epoch": 0.8626387813064808,
      "grad_norm": 6.487565517425537,
      "learning_rate": 4.6754770260986404e-07,
      "loss": 0.1075,
      "step": 16705
    },
    {
      "epoch": 0.8626904208623806,
      "grad_norm": 4.316120624542236,
      "learning_rate": 4.6720180784926074e-07,
      "loss": 0.0456,
      "step": 16706
    },
    {
      "epoch": 0.8627420604182804,
      "grad_norm": 5.101101398468018,
      "learning_rate": 4.668560348127643e-07,
      "loss": 0.0564,
      "step": 16707
    },
    {
      "epoch": 0.8627936999741802,
      "grad_norm": 3.474675178527832,
      "learning_rate": 4.665103835096602e-07,
      "loss": 0.0226,
      "step": 16708
    },
    {
      "epoch": 0.86284533953008,
      "grad_norm": 9.261344909667969,
      "learning_rate": 4.661648539492275e-07,
      "loss": 0.0928,
      "step": 16709
    },
    {
      "epoch": 0.8628969790859798,
      "grad_norm": 7.306251049041748,
      "learning_rate": 4.6581944614074837e-07,
      "loss": 0.0524,
      "step": 16710
    },
    {
      "epoch": 0.8629486186418797,
      "grad_norm": 3.963615894317627,
      "learning_rate": 4.6547416009349756e-07,
      "loss": 0.0555,
      "step": 16711
    },
    {
      "epoch": 0.8630002581977795,
      "grad_norm": 7.799532890319824,
      "learning_rate": 4.651289958167482e-07,
      "loss": 0.0818,
      "step": 16712
    },
    {
      "epoch": 0.8630518977536793,
      "grad_norm": 1.1853766441345215,
      "learning_rate": 4.6478395331976733e-07,
      "loss": 0.02,
      "step": 16713
    },
    {
      "epoch": 0.8631035373095791,
      "grad_norm": 5.295535087585449,
      "learning_rate": 4.6443903261182146e-07,
      "loss": 0.0549,
      "step": 16714
    },
    {
      "epoch": 0.8631551768654789,
      "grad_norm": 3.2805633544921875,
      "learning_rate": 4.6409423370217477e-07,
      "loss": 0.0256,
      "step": 16715
    },
    {
      "epoch": 0.8632068164213788,
      "grad_norm": 4.8211259841918945,
      "learning_rate": 4.6374955660008434e-07,
      "loss": 0.0551,
      "step": 16716
    },
    {
      "epoch": 0.8632584559772786,
      "grad_norm": 8.494355201721191,
      "learning_rate": 4.634050013148073e-07,
      "loss": 0.074,
      "step": 16717
    },
    {
      "epoch": 0.8633100955331784,
      "grad_norm": 4.198772430419922,
      "learning_rate": 4.6306056785559604e-07,
      "loss": 0.0576,
      "step": 16718
    },
    {
      "epoch": 0.8633617350890782,
      "grad_norm": 1.8770420551300049,
      "learning_rate": 4.627162562317e-07,
      "loss": 0.0352,
      "step": 16719
    },
    {
      "epoch": 0.863413374644978,
      "grad_norm": 3.588831663131714,
      "learning_rate": 4.6237206645236563e-07,
      "loss": 0.0439,
      "step": 16720
    },
    {
      "epoch": 0.8634650142008778,
      "grad_norm": 1.8519635200500488,
      "learning_rate": 4.6202799852683555e-07,
      "loss": 0.0302,
      "step": 16721
    },
    {
      "epoch": 0.8635166537567777,
      "grad_norm": 1.4125944375991821,
      "learning_rate": 4.61684052464349e-07,
      "loss": 0.0158,
      "step": 16722
    },
    {
      "epoch": 0.8635682933126775,
      "grad_norm": 3.0386152267456055,
      "learning_rate": 4.6134022827414307e-07,
      "loss": 0.0173,
      "step": 16723
    },
    {
      "epoch": 0.8636199328685773,
      "grad_norm": 13.985433578491211,
      "learning_rate": 4.6099652596545086e-07,
      "loss": 0.1795,
      "step": 16724
    },
    {
      "epoch": 0.8636715724244771,
      "grad_norm": 3.4040067195892334,
      "learning_rate": 4.6065294554750174e-07,
      "loss": 0.0487,
      "step": 16725
    },
    {
      "epoch": 0.8637232119803769,
      "grad_norm": 8.677512168884277,
      "learning_rate": 4.603094870295233e-07,
      "loss": 0.0418,
      "step": 16726
    },
    {
      "epoch": 0.8637748515362768,
      "grad_norm": 20.810211181640625,
      "learning_rate": 4.59966150420737e-07,
      "loss": 0.0605,
      "step": 16727
    },
    {
      "epoch": 0.8638264910921766,
      "grad_norm": 7.921295166015625,
      "learning_rate": 4.596229357303622e-07,
      "loss": 0.0395,
      "step": 16728
    },
    {
      "epoch": 0.8638781306480764,
      "grad_norm": 4.7157368659973145,
      "learning_rate": 4.5927984296761876e-07,
      "loss": 0.0754,
      "step": 16729
    },
    {
      "epoch": 0.8639297702039762,
      "grad_norm": 1.6832784414291382,
      "learning_rate": 4.5893687214171924e-07,
      "loss": 0.0158,
      "step": 16730
    },
    {
      "epoch": 0.863981409759876,
      "grad_norm": 3.368032455444336,
      "learning_rate": 4.5859402326187184e-07,
      "loss": 0.0409,
      "step": 16731
    },
    {
      "epoch": 0.8640330493157758,
      "grad_norm": 4.717714309692383,
      "learning_rate": 4.582512963372854e-07,
      "loss": 0.0528,
      "step": 16732
    },
    {
      "epoch": 0.8640846888716757,
      "grad_norm": 11.106366157531738,
      "learning_rate": 4.5790869137716134e-07,
      "loss": 0.1268,
      "step": 16733
    },
    {
      "epoch": 0.8641363284275755,
      "grad_norm": 10.545865058898926,
      "learning_rate": 4.575662083907029e-07,
      "loss": 0.1281,
      "step": 16734
    },
    {
      "epoch": 0.8641879679834753,
      "grad_norm": 4.794744491577148,
      "learning_rate": 4.5722384738710543e-07,
      "loss": 0.0558,
      "step": 16735
    },
    {
      "epoch": 0.8642396075393751,
      "grad_norm": 3.737292528152466,
      "learning_rate": 4.568816083755628e-07,
      "loss": 0.0436,
      "step": 16736
    },
    {
      "epoch": 0.8642912470952749,
      "grad_norm": 3.056286573410034,
      "learning_rate": 4.565394913652654e-07,
      "loss": 0.0336,
      "step": 16737
    },
    {
      "epoch": 0.8643428866511748,
      "grad_norm": 8.495723724365234,
      "learning_rate": 4.5619749636540143e-07,
      "loss": 0.0781,
      "step": 16738
    },
    {
      "epoch": 0.8643945262070746,
      "grad_norm": 3.7674310207366943,
      "learning_rate": 4.5585562338515357e-07,
      "loss": 0.0418,
      "step": 16739
    },
    {
      "epoch": 0.8644461657629744,
      "grad_norm": 1.7462012767791748,
      "learning_rate": 4.5551387243370384e-07,
      "loss": 0.0153,
      "step": 16740
    },
    {
      "epoch": 0.8644978053188742,
      "grad_norm": 4.259772777557373,
      "learning_rate": 4.551722435202294e-07,
      "loss": 0.029,
      "step": 16741
    },
    {
      "epoch": 0.864549444874774,
      "grad_norm": 17.869163513183594,
      "learning_rate": 4.548307366539018e-07,
      "loss": 0.1203,
      "step": 16742
    },
    {
      "epoch": 0.8646010844306738,
      "grad_norm": 1.463102102279663,
      "learning_rate": 4.544893518438953e-07,
      "loss": 0.0181,
      "step": 16743
    },
    {
      "epoch": 0.8646527239865737,
      "grad_norm": 3.0303661823272705,
      "learning_rate": 4.54148089099376e-07,
      "loss": 0.0345,
      "step": 16744
    },
    {
      "epoch": 0.8647043635424735,
      "grad_norm": 2.5869429111480713,
      "learning_rate": 4.5380694842950934e-07,
      "loss": 0.0187,
      "step": 16745
    },
    {
      "epoch": 0.8647560030983733,
      "grad_norm": 6.672382831573486,
      "learning_rate": 4.534659298434546e-07,
      "loss": 0.0491,
      "step": 16746
    },
    {
      "epoch": 0.8648076426542731,
      "grad_norm": 1.0742510557174683,
      "learning_rate": 4.531250333503695e-07,
      "loss": 0.0091,
      "step": 16747
    },
    {
      "epoch": 0.8648592822101729,
      "grad_norm": 3.71925687789917,
      "learning_rate": 4.527842589594106e-07,
      "loss": 0.0381,
      "step": 16748
    },
    {
      "epoch": 0.8649109217660728,
      "grad_norm": 3.492401361465454,
      "learning_rate": 4.524436066797272e-07,
      "loss": 0.0338,
      "step": 16749
    },
    {
      "epoch": 0.8649625613219726,
      "grad_norm": 3.1978445053100586,
      "learning_rate": 4.5210307652046757e-07,
      "loss": 0.029,
      "step": 16750
    },
    {
      "epoch": 0.8650142008778724,
      "grad_norm": 22.866437911987305,
      "learning_rate": 4.5176266849077667e-07,
      "loss": 0.082,
      "step": 16751
    },
    {
      "epoch": 0.8650658404337722,
      "grad_norm": 4.533093452453613,
      "learning_rate": 4.51422382599796e-07,
      "loss": 0.0611,
      "step": 16752
    },
    {
      "epoch": 0.865117479989672,
      "grad_norm": 4.29885196685791,
      "learning_rate": 4.510822188566633e-07,
      "loss": 0.0701,
      "step": 16753
    },
    {
      "epoch": 0.865169119545572,
      "grad_norm": 12.401374816894531,
      "learning_rate": 4.507421772705128e-07,
      "loss": 0.0638,
      "step": 16754
    },
    {
      "epoch": 0.8652207591014718,
      "grad_norm": 5.435826301574707,
      "learning_rate": 4.5040225785047687e-07,
      "loss": 0.0442,
      "step": 16755
    },
    {
      "epoch": 0.8652723986573716,
      "grad_norm": 4.15872859954834,
      "learning_rate": 4.500624606056836e-07,
      "loss": 0.0746,
      "step": 16756
    },
    {
      "epoch": 0.8653240382132714,
      "grad_norm": 4.657962322235107,
      "learning_rate": 4.4972278554525795e-07,
      "loss": 0.0858,
      "step": 16757
    },
    {
      "epoch": 0.8653756777691712,
      "grad_norm": 9.219425201416016,
      "learning_rate": 4.4938323267832097e-07,
      "loss": 0.1164,
      "step": 16758
    },
    {
      "epoch": 0.865427317325071,
      "grad_norm": 10.010313987731934,
      "learning_rate": 4.4904380201399254e-07,
      "loss": 0.1497,
      "step": 16759
    },
    {
      "epoch": 0.8654789568809709,
      "grad_norm": 3.8707449436187744,
      "learning_rate": 4.48704493561386e-07,
      "loss": 0.0633,
      "step": 16760
    },
    {
      "epoch": 0.8655305964368707,
      "grad_norm": 2.1605422496795654,
      "learning_rate": 4.483653073296129e-07,
      "loss": 0.0362,
      "step": 16761
    },
    {
      "epoch": 0.8655822359927705,
      "grad_norm": 10.17383098602295,
      "learning_rate": 4.480262433277832e-07,
      "loss": 0.0549,
      "step": 16762
    },
    {
      "epoch": 0.8656338755486703,
      "grad_norm": 7.132932662963867,
      "learning_rate": 4.4768730156500296e-07,
      "loss": 0.0554,
      "step": 16763
    },
    {
      "epoch": 0.8656855151045701,
      "grad_norm": 2.3600013256073,
      "learning_rate": 4.473484820503721e-07,
      "loss": 0.0246,
      "step": 16764
    },
    {
      "epoch": 0.86573715466047,
      "grad_norm": 13.403375625610352,
      "learning_rate": 4.4700978479298996e-07,
      "loss": 0.0621,
      "step": 16765
    },
    {
      "epoch": 0.8657887942163698,
      "grad_norm": 4.5804595947265625,
      "learning_rate": 4.46671209801951e-07,
      "loss": 0.0957,
      "step": 16766
    },
    {
      "epoch": 0.8658404337722696,
      "grad_norm": 3.865513563156128,
      "learning_rate": 4.4633275708635013e-07,
      "loss": 0.0989,
      "step": 16767
    },
    {
      "epoch": 0.8658920733281694,
      "grad_norm": 4.424960613250732,
      "learning_rate": 4.4599442665527406e-07,
      "loss": 0.069,
      "step": 16768
    },
    {
      "epoch": 0.8659437128840692,
      "grad_norm": 20.45699119567871,
      "learning_rate": 4.4565621851780816e-07,
      "loss": 0.1654,
      "step": 16769
    },
    {
      "epoch": 0.8659953524399691,
      "grad_norm": 5.079379081726074,
      "learning_rate": 4.453181326830358e-07,
      "loss": 0.0462,
      "step": 16770
    },
    {
      "epoch": 0.8660469919958689,
      "grad_norm": 4.558706283569336,
      "learning_rate": 4.4498016916003526e-07,
      "loss": 0.0474,
      "step": 16771
    },
    {
      "epoch": 0.8660986315517687,
      "grad_norm": 3.227534770965576,
      "learning_rate": 4.446423279578821e-07,
      "loss": 0.0218,
      "step": 16772
    },
    {
      "epoch": 0.8661502711076685,
      "grad_norm": 3.1508536338806152,
      "learning_rate": 4.4430460908564897e-07,
      "loss": 0.0605,
      "step": 16773
    },
    {
      "epoch": 0.8662019106635683,
      "grad_norm": 2.7699289321899414,
      "learning_rate": 4.439670125524065e-07,
      "loss": 0.033,
      "step": 16774
    },
    {
      "epoch": 0.8662535502194681,
      "grad_norm": 2.2510414123535156,
      "learning_rate": 4.4362953836721734e-07,
      "loss": 0.0303,
      "step": 16775
    },
    {
      "epoch": 0.866305189775368,
      "grad_norm": 3.423795461654663,
      "learning_rate": 4.43292186539146e-07,
      "loss": 0.0582,
      "step": 16776
    },
    {
      "epoch": 0.8663568293312678,
      "grad_norm": 4.338464736938477,
      "learning_rate": 4.4295495707725233e-07,
      "loss": 0.0733,
      "step": 16777
    },
    {
      "epoch": 0.8664084688871676,
      "grad_norm": 20.196317672729492,
      "learning_rate": 4.4261784999059196e-07,
      "loss": 0.1025,
      "step": 16778
    },
    {
      "epoch": 0.8664601084430674,
      "grad_norm": 4.448938369750977,
      "learning_rate": 4.4228086528821654e-07,
      "loss": 0.0521,
      "step": 16779
    },
    {
      "epoch": 0.8665117479989672,
      "grad_norm": 3.512460947036743,
      "learning_rate": 4.4194400297917497e-07,
      "loss": 0.0408,
      "step": 16780
    },
    {
      "epoch": 0.8665633875548671,
      "grad_norm": 8.926407814025879,
      "learning_rate": 4.416072630725166e-07,
      "loss": 0.0969,
      "step": 16781
    },
    {
      "epoch": 0.8666150271107669,
      "grad_norm": 3.763416051864624,
      "learning_rate": 4.4127064557728137e-07,
      "loss": 0.0604,
      "step": 16782
    },
    {
      "epoch": 0.8666666666666667,
      "grad_norm": 6.52785587310791,
      "learning_rate": 4.4093415050250943e-07,
      "loss": 0.1114,
      "step": 16783
    },
    {
      "epoch": 0.8667183062225665,
      "grad_norm": 13.44570541381836,
      "learning_rate": 4.4059777785723736e-07,
      "loss": 0.1037,
      "step": 16784
    },
    {
      "epoch": 0.8667699457784663,
      "grad_norm": 4.849361419677734,
      "learning_rate": 4.402615276504979e-07,
      "loss": 0.0285,
      "step": 16785
    },
    {
      "epoch": 0.8668215853343662,
      "grad_norm": 5.248108386993408,
      "learning_rate": 4.399253998913205e-07,
      "loss": 0.0671,
      "step": 16786
    },
    {
      "epoch": 0.866873224890266,
      "grad_norm": 1.2821296453475952,
      "learning_rate": 4.3958939458873253e-07,
      "loss": 0.0179,
      "step": 16787
    },
    {
      "epoch": 0.8669248644461658,
      "grad_norm": 5.158772945404053,
      "learning_rate": 4.3925351175175656e-07,
      "loss": 0.096,
      "step": 16788
    },
    {
      "epoch": 0.8669765040020656,
      "grad_norm": 9.217573165893555,
      "learning_rate": 4.389177513894122e-07,
      "loss": 0.1197,
      "step": 16789
    },
    {
      "epoch": 0.8670281435579654,
      "grad_norm": 3.8932082653045654,
      "learning_rate": 4.3858211351071655e-07,
      "loss": 0.0533,
      "step": 16790
    },
    {
      "epoch": 0.8670797831138652,
      "grad_norm": 2.6806275844573975,
      "learning_rate": 4.3824659812468197e-07,
      "loss": 0.0578,
      "step": 16791
    },
    {
      "epoch": 0.8671314226697651,
      "grad_norm": 5.43100643157959,
      "learning_rate": 4.3791120524032004e-07,
      "loss": 0.0441,
      "step": 16792
    },
    {
      "epoch": 0.8671830622256649,
      "grad_norm": 1.6578561067581177,
      "learning_rate": 4.375759348666353e-07,
      "loss": 0.0217,
      "step": 16793
    },
    {
      "epoch": 0.8672347017815647,
      "grad_norm": 1.6459848880767822,
      "learning_rate": 4.372407870126311e-07,
      "loss": 0.0254,
      "step": 16794
    },
    {
      "epoch": 0.8672863413374645,
      "grad_norm": 2.3015382289886475,
      "learning_rate": 4.369057616873096e-07,
      "loss": 0.0371,
      "step": 16795
    },
    {
      "epoch": 0.8673379808933643,
      "grad_norm": 10.636249542236328,
      "learning_rate": 4.3657085889966756e-07,
      "loss": 0.1064,
      "step": 16796
    },
    {
      "epoch": 0.8673896204492642,
      "grad_norm": 4.465962886810303,
      "learning_rate": 4.362360786586961e-07,
      "loss": 0.0411,
      "step": 16797
    },
    {
      "epoch": 0.867441260005164,
      "grad_norm": 4.595407962799072,
      "learning_rate": 4.3590142097338696e-07,
      "loss": 0.0396,
      "step": 16798
    },
    {
      "epoch": 0.8674928995610638,
      "grad_norm": 4.99275541305542,
      "learning_rate": 4.3556688585272623e-07,
      "loss": 0.033,
      "step": 16799
    },
    {
      "epoch": 0.8675445391169636,
      "grad_norm": 2.896207571029663,
      "learning_rate": 4.3523247330570005e-07,
      "loss": 0.0502,
      "step": 16800
    },
    {
      "epoch": 0.8675961786728634,
      "grad_norm": 5.821926116943359,
      "learning_rate": 4.3489818334128576e-07,
      "loss": 0.0555,
      "step": 16801
    },
    {
      "epoch": 0.8676478182287632,
      "grad_norm": 2.0826973915100098,
      "learning_rate": 4.345640159684622e-07,
      "loss": 0.0448,
      "step": 16802
    },
    {
      "epoch": 0.8676994577846631,
      "grad_norm": 10.992855072021484,
      "learning_rate": 4.342299711962017e-07,
      "loss": 0.1184,
      "step": 16803
    },
    {
      "epoch": 0.8677510973405629,
      "grad_norm": 4.759527683258057,
      "learning_rate": 4.338960490334759e-07,
      "loss": 0.0405,
      "step": 16804
    },
    {
      "epoch": 0.8678027368964627,
      "grad_norm": 7.994338035583496,
      "learning_rate": 4.335622494892516e-07,
      "loss": 0.0508,
      "step": 16805
    },
    {
      "epoch": 0.8678543764523625,
      "grad_norm": 2.352114677429199,
      "learning_rate": 4.332285725724927e-07,
      "loss": 0.0301,
      "step": 16806
    },
    {
      "epoch": 0.8679060160082623,
      "grad_norm": 3.6931188106536865,
      "learning_rate": 4.328950182921604e-07,
      "loss": 0.0438,
      "step": 16807
    },
    {
      "epoch": 0.8679576555641622,
      "grad_norm": 3.3144030570983887,
      "learning_rate": 4.3256158665720973e-07,
      "loss": 0.0458,
      "step": 16808
    },
    {
      "epoch": 0.868009295120062,
      "grad_norm": 3.1097161769866943,
      "learning_rate": 4.3222827767659737e-07,
      "loss": 0.0219,
      "step": 16809
    },
    {
      "epoch": 0.8680609346759618,
      "grad_norm": 7.137248516082764,
      "learning_rate": 4.3189509135927287e-07,
      "loss": 0.0429,
      "step": 16810
    },
    {
      "epoch": 0.8681125742318616,
      "grad_norm": 15.515637397766113,
      "learning_rate": 4.3156202771418465e-07,
      "loss": 0.0802,
      "step": 16811
    },
    {
      "epoch": 0.8681642137877614,
      "grad_norm": 3.3473708629608154,
      "learning_rate": 4.31229086750275e-07,
      "loss": 0.0557,
      "step": 16812
    },
    {
      "epoch": 0.8682158533436612,
      "grad_norm": 6.0361127853393555,
      "learning_rate": 4.3089626847648445e-07,
      "loss": 0.0636,
      "step": 16813
    },
    {
      "epoch": 0.8682674928995611,
      "grad_norm": 5.562041759490967,
      "learning_rate": 4.3056357290175376e-07,
      "loss": 0.0516,
      "step": 16814
    },
    {
      "epoch": 0.8683191324554609,
      "grad_norm": 4.5463032722473145,
      "learning_rate": 4.3023100003501405e-07,
      "loss": 0.0254,
      "step": 16815
    },
    {
      "epoch": 0.8683707720113607,
      "grad_norm": 4.333404541015625,
      "learning_rate": 4.2989854988519764e-07,
      "loss": 0.0568,
      "step": 16816
    },
    {
      "epoch": 0.8684224115672605,
      "grad_norm": 7.30113410949707,
      "learning_rate": 4.295662224612318e-07,
      "loss": 0.0825,
      "step": 16817
    },
    {
      "epoch": 0.8684740511231603,
      "grad_norm": 2.305307388305664,
      "learning_rate": 4.2923401777204057e-07,
      "loss": 0.0191,
      "step": 16818
    },
    {
      "epoch": 0.8685256906790602,
      "grad_norm": 5.288543701171875,
      "learning_rate": 4.289019358265456e-07,
      "loss": 0.0343,
      "step": 16819
    },
    {
      "epoch": 0.86857733023496,
      "grad_norm": 1.5857163667678833,
      "learning_rate": 4.285699766336648e-07,
      "loss": 0.0155,
      "step": 16820
    },
    {
      "epoch": 0.8686289697908598,
      "grad_norm": 3.402534008026123,
      "learning_rate": 4.282381402023117e-07,
      "loss": 0.0305,
      "step": 16821
    },
    {
      "epoch": 0.8686806093467596,
      "grad_norm": 1.2517088651657104,
      "learning_rate": 4.2790642654139845e-07,
      "loss": 0.012,
      "step": 16822
    },
    {
      "epoch": 0.8687322489026594,
      "grad_norm": 11.20759105682373,
      "learning_rate": 4.275748356598325e-07,
      "loss": 0.1794,
      "step": 16823
    },
    {
      "epoch": 0.8687838884585593,
      "grad_norm": 1.2197439670562744,
      "learning_rate": 4.2724336756651775e-07,
      "loss": 0.0146,
      "step": 16824
    },
    {
      "epoch": 0.8688355280144591,
      "grad_norm": 6.9029107093811035,
      "learning_rate": 4.2691202227035765e-07,
      "loss": 0.0923,
      "step": 16825
    },
    {
      "epoch": 0.8688871675703589,
      "grad_norm": 4.982079029083252,
      "learning_rate": 4.2658079978024735e-07,
      "loss": 0.0805,
      "step": 16826
    },
    {
      "epoch": 0.8689388071262587,
      "grad_norm": 9.926121711730957,
      "learning_rate": 4.262497001050819e-07,
      "loss": 0.0452,
      "step": 16827
    },
    {
      "epoch": 0.8689904466821585,
      "grad_norm": 2.320359468460083,
      "learning_rate": 4.2591872325375484e-07,
      "loss": 0.0291,
      "step": 16828
    },
    {
      "epoch": 0.8690420862380583,
      "grad_norm": 1.189224123954773,
      "learning_rate": 4.255878692351534e-07,
      "loss": 0.0143,
      "step": 16829
    },
    {
      "epoch": 0.8690937257939582,
      "grad_norm": 2.637943983078003,
      "learning_rate": 4.252571380581616e-07,
      "loss": 0.034,
      "step": 16830
    },
    {
      "epoch": 0.869145365349858,
      "grad_norm": 2.9243004322052,
      "learning_rate": 4.249265297316607e-07,
      "loss": 0.0461,
      "step": 16831
    },
    {
      "epoch": 0.8691970049057578,
      "grad_norm": 1.1078336238861084,
      "learning_rate": 4.2459604426452916e-07,
      "loss": 0.0138,
      "step": 16832
    },
    {
      "epoch": 0.8692486444616576,
      "grad_norm": 6.903519630432129,
      "learning_rate": 4.2426568166564373e-07,
      "loss": 0.0329,
      "step": 16833
    },
    {
      "epoch": 0.8693002840175574,
      "grad_norm": 4.873058795928955,
      "learning_rate": 4.239354419438729e-07,
      "loss": 0.0681,
      "step": 16834
    },
    {
      "epoch": 0.8693519235734573,
      "grad_norm": 6.22939920425415,
      "learning_rate": 4.236053251080874e-07,
      "loss": 0.0791,
      "step": 16835
    },
    {
      "epoch": 0.8694035631293571,
      "grad_norm": 15.811564445495605,
      "learning_rate": 4.232753311671506e-07,
      "loss": 0.1549,
      "step": 16836
    },
    {
      "epoch": 0.8694552026852569,
      "grad_norm": 6.377316474914551,
      "learning_rate": 4.2294546012992486e-07,
      "loss": 0.0808,
      "step": 16837
    },
    {
      "epoch": 0.8695068422411567,
      "grad_norm": 4.339272499084473,
      "learning_rate": 4.226157120052687e-07,
      "loss": 0.0315,
      "step": 16838
    },
    {
      "epoch": 0.8695584817970565,
      "grad_norm": 3.1769137382507324,
      "learning_rate": 4.2228608680203733e-07,
      "loss": 0.027,
      "step": 16839
    },
    {
      "epoch": 0.8696101213529563,
      "grad_norm": 1.7761808633804321,
      "learning_rate": 4.2195658452908293e-07,
      "loss": 0.0128,
      "step": 16840
    },
    {
      "epoch": 0.8696617609088562,
      "grad_norm": 20.052593231201172,
      "learning_rate": 4.2162720519525133e-07,
      "loss": 0.1616,
      "step": 16841
    },
    {
      "epoch": 0.869713400464756,
      "grad_norm": 1.5281635522842407,
      "learning_rate": 4.2129794880939103e-07,
      "loss": 0.0177,
      "step": 16842
    },
    {
      "epoch": 0.8697650400206558,
      "grad_norm": 3.9538352489471436,
      "learning_rate": 4.2096881538034207e-07,
      "loss": 0.0201,
      "step": 16843
    },
    {
      "epoch": 0.8698166795765556,
      "grad_norm": 4.39108943939209,
      "learning_rate": 4.206398049169447e-07,
      "loss": 0.0448,
      "step": 16844
    },
    {
      "epoch": 0.8698683191324554,
      "grad_norm": 5.691900730133057,
      "learning_rate": 4.203109174280323e-07,
      "loss": 0.0861,
      "step": 16845
    },
    {
      "epoch": 0.8699199586883553,
      "grad_norm": 8.22472858428955,
      "learning_rate": 4.1998215292243625e-07,
      "loss": 0.1216,
      "step": 16846
    },
    {
      "epoch": 0.8699715982442551,
      "grad_norm": 12.22675895690918,
      "learning_rate": 4.196535114089889e-07,
      "loss": 0.0784,
      "step": 16847
    },
    {
      "epoch": 0.8700232378001549,
      "grad_norm": 3.4608559608459473,
      "learning_rate": 4.193249928965115e-07,
      "loss": 0.0322,
      "step": 16848
    },
    {
      "epoch": 0.8700748773560547,
      "grad_norm": 9.75813102722168,
      "learning_rate": 4.1899659739382815e-07,
      "loss": 0.0463,
      "step": 16849
    },
    {
      "epoch": 0.8701265169119545,
      "grad_norm": 3.416325330734253,
      "learning_rate": 4.1866832490975784e-07,
      "loss": 0.055,
      "step": 16850
    },
    {
      "epoch": 0.8701781564678543,
      "grad_norm": 11.38948917388916,
      "learning_rate": 4.183401754531147e-07,
      "loss": 0.0361,
      "step": 16851
    },
    {
      "epoch": 0.8702297960237542,
      "grad_norm": 7.869091987609863,
      "learning_rate": 4.1801214903271213e-07,
      "loss": 0.0857,
      "step": 16852
    },
    {
      "epoch": 0.870281435579654,
      "grad_norm": 5.347826957702637,
      "learning_rate": 4.176842456573582e-07,
      "loss": 0.0627,
      "step": 16853
    },
    {
      "epoch": 0.8703330751355538,
      "grad_norm": 7.285311698913574,
      "learning_rate": 4.1735646533585917e-07,
      "loss": 0.0517,
      "step": 16854
    },
    {
      "epoch": 0.8703847146914536,
      "grad_norm": 4.722837924957275,
      "learning_rate": 4.1702880807701683e-07,
      "loss": 0.0342,
      "step": 16855
    },
    {
      "epoch": 0.8704363542473534,
      "grad_norm": 1.1072996854782104,
      "learning_rate": 4.167012738896298e-07,
      "loss": 0.0143,
      "step": 16856
    },
    {
      "epoch": 0.8704879938032533,
      "grad_norm": 5.766299724578857,
      "learning_rate": 4.1637386278249427e-07,
      "loss": 0.0819,
      "step": 16857
    },
    {
      "epoch": 0.8705396333591531,
      "grad_norm": 7.889209270477295,
      "learning_rate": 4.160465747644032e-07,
      "loss": 0.0949,
      "step": 16858
    },
    {
      "epoch": 0.8705912729150529,
      "grad_norm": 3.5996248722076416,
      "learning_rate": 4.157194098441436e-07,
      "loss": 0.0389,
      "step": 16859
    },
    {
      "epoch": 0.8706429124709527,
      "grad_norm": 6.20283317565918,
      "learning_rate": 4.1539236803050163e-07,
      "loss": 0.0325,
      "step": 16860
    },
    {
      "epoch": 0.8706945520268525,
      "grad_norm": 3.5761122703552246,
      "learning_rate": 4.1506544933226146e-07,
      "loss": 0.0396,
      "step": 16861
    },
    {
      "epoch": 0.8707461915827523,
      "grad_norm": 2.1954095363616943,
      "learning_rate": 4.147386537582021e-07,
      "loss": 0.0247,
      "step": 16862
    },
    {
      "epoch": 0.8707978311386522,
      "grad_norm": 2.513364315032959,
      "learning_rate": 4.1441198131709716e-07,
      "loss": 0.0287,
      "step": 16863
    },
    {
      "epoch": 0.870849470694552,
      "grad_norm": 18.905792236328125,
      "learning_rate": 4.140854320177207e-07,
      "loss": 0.1052,
      "step": 16864
    },
    {
      "epoch": 0.8709011102504518,
      "grad_norm": 6.214727401733398,
      "learning_rate": 4.137590058688401e-07,
      "loss": 0.1014,
      "step": 16865
    },
    {
      "epoch": 0.8709527498063516,
      "grad_norm": 4.528201103210449,
      "learning_rate": 4.134327028792251e-07,
      "loss": 0.0231,
      "step": 16866
    },
    {
      "epoch": 0.8710043893622514,
      "grad_norm": 5.0406599044799805,
      "learning_rate": 4.131065230576348e-07,
      "loss": 0.0335,
      "step": 16867
    },
    {
      "epoch": 0.8710560289181513,
      "grad_norm": 1.3772082328796387,
      "learning_rate": 4.127804664128293e-07,
      "loss": 0.0097,
      "step": 16868
    },
    {
      "epoch": 0.8711076684740511,
      "grad_norm": 6.9842448234558105,
      "learning_rate": 4.1245453295356454e-07,
      "loss": 0.0575,
      "step": 16869
    },
    {
      "epoch": 0.8711593080299509,
      "grad_norm": 2.5237300395965576,
      "learning_rate": 4.1212872268859395e-07,
      "loss": 0.0249,
      "step": 16870
    },
    {
      "epoch": 0.8712109475858507,
      "grad_norm": 4.492657661437988,
      "learning_rate": 4.118030356266656e-07,
      "loss": 0.072,
      "step": 16871
    },
    {
      "epoch": 0.8712625871417505,
      "grad_norm": 3.5843331813812256,
      "learning_rate": 4.1147747177652684e-07,
      "loss": 0.0347,
      "step": 16872
    },
    {
      "epoch": 0.8713142266976504,
      "grad_norm": 2.9939191341400146,
      "learning_rate": 4.1115203114691917e-07,
      "loss": 0.0554,
      "step": 16873
    },
    {
      "epoch": 0.8713658662535502,
      "grad_norm": 2.8810508251190186,
      "learning_rate": 4.108267137465832e-07,
      "loss": 0.0337,
      "step": 16874
    },
    {
      "epoch": 0.87141750580945,
      "grad_norm": 6.071893215179443,
      "learning_rate": 4.1050151958425377e-07,
      "loss": 0.0645,
      "step": 16875
    },
    {
      "epoch": 0.8714691453653498,
      "grad_norm": 9.011918067932129,
      "learning_rate": 4.101764486686649e-07,
      "loss": 0.0604,
      "step": 16876
    },
    {
      "epoch": 0.8715207849212496,
      "grad_norm": 4.1854987144470215,
      "learning_rate": 4.0985150100854574e-07,
      "loss": 0.0758,
      "step": 16877
    },
    {
      "epoch": 0.8715724244771494,
      "grad_norm": 2.3109374046325684,
      "learning_rate": 4.0952667661262214e-07,
      "loss": 0.0213,
      "step": 16878
    },
    {
      "epoch": 0.8716240640330493,
      "grad_norm": 1.9137935638427734,
      "learning_rate": 4.092019754896154e-07,
      "loss": 0.0211,
      "step": 16879
    },
    {
      "epoch": 0.8716757035889491,
      "grad_norm": 12.535428047180176,
      "learning_rate": 4.0887739764824853e-07,
      "loss": 0.1392,
      "step": 16880
    },
    {
      "epoch": 0.871727343144849,
      "grad_norm": 9.478036880493164,
      "learning_rate": 4.085529430972351e-07,
      "loss": 0.0633,
      "step": 16881
    },
    {
      "epoch": 0.8717789827007488,
      "grad_norm": 6.271520137786865,
      "learning_rate": 4.0822861184528884e-07,
      "loss": 0.0339,
      "step": 16882
    },
    {
      "epoch": 0.8718306222566486,
      "grad_norm": 3.3518483638763428,
      "learning_rate": 4.0790440390111984e-07,
      "loss": 0.0755,
      "step": 16883
    },
    {
      "epoch": 0.8718822618125485,
      "grad_norm": 1.7536025047302246,
      "learning_rate": 4.0758031927343345e-07,
      "loss": 0.0202,
      "step": 16884
    },
    {
      "epoch": 0.8719339013684483,
      "grad_norm": 21.88836669921875,
      "learning_rate": 4.072563579709332e-07,
      "loss": 0.1339,
      "step": 16885
    },
    {
      "epoch": 0.8719855409243481,
      "grad_norm": 2.509276866912842,
      "learning_rate": 4.0693252000231885e-07,
      "loss": 0.0329,
      "step": 16886
    },
    {
      "epoch": 0.8720371804802479,
      "grad_norm": 4.5953593254089355,
      "learning_rate": 4.066088053762868e-07,
      "loss": 0.0608,
      "step": 16887
    },
    {
      "epoch": 0.8720888200361477,
      "grad_norm": 20.598928451538086,
      "learning_rate": 4.062852141015294e-07,
      "loss": 0.0717,
      "step": 16888
    },
    {
      "epoch": 0.8721404595920476,
      "grad_norm": 4.089317798614502,
      "learning_rate": 4.059617461867371e-07,
      "loss": 0.0838,
      "step": 16889
    },
    {
      "epoch": 0.8721920991479474,
      "grad_norm": 2.441354513168335,
      "learning_rate": 4.0563840164059677e-07,
      "loss": 0.0362,
      "step": 16890
    },
    {
      "epoch": 0.8722437387038472,
      "grad_norm": 2.961291551589966,
      "learning_rate": 4.0531518047179144e-07,
      "loss": 0.0463,
      "step": 16891
    },
    {
      "epoch": 0.872295378259747,
      "grad_norm": 2.1315321922302246,
      "learning_rate": 4.049920826889986e-07,
      "loss": 0.0158,
      "step": 16892
    },
    {
      "epoch": 0.8723470178156468,
      "grad_norm": 3.3071460723876953,
      "learning_rate": 4.0466910830089747e-07,
      "loss": 0.1218,
      "step": 16893
    },
    {
      "epoch": 0.8723986573715466,
      "grad_norm": 2.4148130416870117,
      "learning_rate": 4.0434625731615994e-07,
      "loss": 0.033,
      "step": 16894
    },
    {
      "epoch": 0.8724502969274465,
      "grad_norm": 1.8524305820465088,
      "learning_rate": 4.0402352974345747e-07,
      "loss": 0.0259,
      "step": 16895
    },
    {
      "epoch": 0.8725019364833463,
      "grad_norm": 6.502118110656738,
      "learning_rate": 4.037009255914548e-07,
      "loss": 0.0531,
      "step": 16896
    },
    {
      "epoch": 0.8725535760392461,
      "grad_norm": 2.1149239540100098,
      "learning_rate": 4.033784448688155e-07,
      "loss": 0.0401,
      "step": 16897
    },
    {
      "epoch": 0.8726052155951459,
      "grad_norm": 6.634869575500488,
      "learning_rate": 4.030560875841988e-07,
      "loss": 0.0448,
      "step": 16898
    },
    {
      "epoch": 0.8726568551510457,
      "grad_norm": 1.2708629369735718,
      "learning_rate": 4.0273385374626383e-07,
      "loss": 0.017,
      "step": 16899
    },
    {
      "epoch": 0.8727084947069456,
      "grad_norm": 3.2594172954559326,
      "learning_rate": 4.024117433636615e-07,
      "loss": 0.0527,
      "step": 16900
    },
    {
      "epoch": 0.8727601342628454,
      "grad_norm": 8.243070602416992,
      "learning_rate": 4.020897564450432e-07,
      "loss": 0.0751,
      "step": 16901
    },
    {
      "epoch": 0.8728117738187452,
      "grad_norm": 13.4600248336792,
      "learning_rate": 4.017678929990543e-07,
      "loss": 0.0974,
      "step": 16902
    },
    {
      "epoch": 0.872863413374645,
      "grad_norm": 1.9169443845748901,
      "learning_rate": 4.014461530343389e-07,
      "loss": 0.0413,
      "step": 16903
    },
    {
      "epoch": 0.8729150529305448,
      "grad_norm": 8.854406356811523,
      "learning_rate": 4.0112453655953734e-07,
      "loss": 0.1032,
      "step": 16904
    },
    {
      "epoch": 0.8729666924864447,
      "grad_norm": 1.4487569332122803,
      "learning_rate": 4.00803043583286e-07,
      "loss": 0.0186,
      "step": 16905
    },
    {
      "epoch": 0.8730183320423445,
      "grad_norm": 2.9685006141662598,
      "learning_rate": 4.0048167411421756e-07,
      "loss": 0.0534,
      "step": 16906
    },
    {
      "epoch": 0.8730699715982443,
      "grad_norm": 11.755245208740234,
      "learning_rate": 4.0016042816096333e-07,
      "loss": 0.0526,
      "step": 16907
    },
    {
      "epoch": 0.8731216111541441,
      "grad_norm": 3.9881210327148438,
      "learning_rate": 3.9983930573214927e-07,
      "loss": 0.0622,
      "step": 16908
    },
    {
      "epoch": 0.8731732507100439,
      "grad_norm": 1.2044906616210938,
      "learning_rate": 3.9951830683639947e-07,
      "loss": 0.0118,
      "step": 16909
    },
    {
      "epoch": 0.8732248902659437,
      "grad_norm": 17.517335891723633,
      "learning_rate": 3.9919743148233383e-07,
      "loss": 0.0467,
      "step": 16910
    },
    {
      "epoch": 0.8732765298218436,
      "grad_norm": 8.33566951751709,
      "learning_rate": 3.988766796785676e-07,
      "loss": 0.0513,
      "step": 16911
    },
    {
      "epoch": 0.8733281693777434,
      "grad_norm": 4.5575270652771,
      "learning_rate": 3.9855605143371667e-07,
      "loss": 0.0736,
      "step": 16912
    },
    {
      "epoch": 0.8733798089336432,
      "grad_norm": 7.409699440002441,
      "learning_rate": 3.982355467563909e-07,
      "loss": 0.0417,
      "step": 16913
    },
    {
      "epoch": 0.873431448489543,
      "grad_norm": 2.0350544452667236,
      "learning_rate": 3.9791516565519607e-07,
      "loss": 0.0187,
      "step": 16914
    },
    {
      "epoch": 0.8734830880454428,
      "grad_norm": 15.415003776550293,
      "learning_rate": 3.9759490813873533e-07,
      "loss": 0.1042,
      "step": 16915
    },
    {
      "epoch": 0.8735347276013427,
      "grad_norm": 12.119739532470703,
      "learning_rate": 3.9727477421561024e-07,
      "loss": 0.1793,
      "step": 16916
    },
    {
      "epoch": 0.8735863671572425,
      "grad_norm": 3.793271064758301,
      "learning_rate": 3.969547638944171e-07,
      "loss": 0.0451,
      "step": 16917
    },
    {
      "epoch": 0.8736380067131423,
      "grad_norm": 5.259084701538086,
      "learning_rate": 3.9663487718374917e-07,
      "loss": 0.0646,
      "step": 16918
    },
    {
      "epoch": 0.8736896462690421,
      "grad_norm": 1.8967323303222656,
      "learning_rate": 3.9631511409219726e-07,
      "loss": 0.021,
      "step": 16919
    },
    {
      "epoch": 0.8737412858249419,
      "grad_norm": 8.903701782226562,
      "learning_rate": 3.959954746283479e-07,
      "loss": 0.1049,
      "step": 16920
    },
    {
      "epoch": 0.8737929253808417,
      "grad_norm": 7.423748970031738,
      "learning_rate": 3.956759588007852e-07,
      "loss": 0.0754,
      "step": 16921
    },
    {
      "epoch": 0.8738445649367416,
      "grad_norm": 2.0756728649139404,
      "learning_rate": 3.953565666180892e-07,
      "loss": 0.0247,
      "step": 16922
    },
    {
      "epoch": 0.8738962044926414,
      "grad_norm": 2.967010974884033,
      "learning_rate": 3.950372980888367e-07,
      "loss": 0.0453,
      "step": 16923
    },
    {
      "epoch": 0.8739478440485412,
      "grad_norm": 3.6377127170562744,
      "learning_rate": 3.9471815322160267e-07,
      "loss": 0.0397,
      "step": 16924
    },
    {
      "epoch": 0.873999483604441,
      "grad_norm": 3.5406973361968994,
      "learning_rate": 3.943991320249546e-07,
      "loss": 0.0522,
      "step": 16925
    },
    {
      "epoch": 0.8740511231603408,
      "grad_norm": 1.5985251665115356,
      "learning_rate": 3.9408023450746183e-07,
      "loss": 0.0149,
      "step": 16926
    },
    {
      "epoch": 0.8741027627162407,
      "grad_norm": 16.0621337890625,
      "learning_rate": 3.937614606776874e-07,
      "loss": 0.1802,
      "step": 16927
    },
    {
      "epoch": 0.8741544022721405,
      "grad_norm": 4.844631671905518,
      "learning_rate": 3.9344281054419287e-07,
      "loss": 0.0295,
      "step": 16928
    },
    {
      "epoch": 0.8742060418280403,
      "grad_norm": 15.311307907104492,
      "learning_rate": 3.9312428411553306e-07,
      "loss": 0.0616,
      "step": 16929
    },
    {
      "epoch": 0.8742576813839401,
      "grad_norm": 4.417325973510742,
      "learning_rate": 3.9280588140026274e-07,
      "loss": 0.0443,
      "step": 16930
    },
    {
      "epoch": 0.8743093209398399,
      "grad_norm": 4.201869010925293,
      "learning_rate": 3.9248760240693186e-07,
      "loss": 0.0628,
      "step": 16931
    },
    {
      "epoch": 0.8743609604957397,
      "grad_norm": 12.246830940246582,
      "learning_rate": 3.9216944714408957e-07,
      "loss": 0.087,
      "step": 16932
    },
    {
      "epoch": 0.8744126000516396,
      "grad_norm": 3.6142826080322266,
      "learning_rate": 3.918514156202774e-07,
      "loss": 0.0584,
      "step": 16933
    },
    {
      "epoch": 0.8744642396075394,
      "grad_norm": 3.0537197589874268,
      "learning_rate": 3.9153350784403633e-07,
      "loss": 0.0351,
      "step": 16934
    },
    {
      "epoch": 0.8745158791634392,
      "grad_norm": 6.290927886962891,
      "learning_rate": 3.9121572382390336e-07,
      "loss": 0.0601,
      "step": 16935
    },
    {
      "epoch": 0.874567518719339,
      "grad_norm": 6.654223918914795,
      "learning_rate": 3.908980635684134e-07,
      "loss": 0.0318,
      "step": 16936
    },
    {
      "epoch": 0.8746191582752388,
      "grad_norm": 0.47446781396865845,
      "learning_rate": 3.905805270860952e-07,
      "loss": 0.0056,
      "step": 16937
    },
    {
      "epoch": 0.8746707978311387,
      "grad_norm": 6.363857269287109,
      "learning_rate": 3.9026311438547736e-07,
      "loss": 0.0485,
      "step": 16938
    },
    {
      "epoch": 0.8747224373870385,
      "grad_norm": 11.192843437194824,
      "learning_rate": 3.899458254750832e-07,
      "loss": 0.0608,
      "step": 16939
    },
    {
      "epoch": 0.8747740769429383,
      "grad_norm": 2.015545606613159,
      "learning_rate": 3.896286603634325e-07,
      "loss": 0.0296,
      "step": 16940
    },
    {
      "epoch": 0.8748257164988381,
      "grad_norm": 4.711783409118652,
      "learning_rate": 3.8931161905904347e-07,
      "loss": 0.0525,
      "step": 16941
    },
    {
      "epoch": 0.8748773560547379,
      "grad_norm": 5.957987308502197,
      "learning_rate": 3.8899470157042984e-07,
      "loss": 0.0713,
      "step": 16942
    },
    {
      "epoch": 0.8749289956106377,
      "grad_norm": 5.8344526290893555,
      "learning_rate": 3.886779079061026e-07,
      "loss": 0.0686,
      "step": 16943
    },
    {
      "epoch": 0.8749806351665376,
      "grad_norm": 1.4944134950637817,
      "learning_rate": 3.883612380745666e-07,
      "loss": 0.0107,
      "step": 16944
    },
    {
      "epoch": 0.8750322747224374,
      "grad_norm": 17.444843292236328,
      "learning_rate": 3.880446920843284e-07,
      "loss": 0.1131,
      "step": 16945
    },
    {
      "epoch": 0.8750839142783372,
      "grad_norm": 5.610367298126221,
      "learning_rate": 3.877282699438878e-07,
      "loss": 0.0847,
      "step": 16946
    },
    {
      "epoch": 0.875135553834237,
      "grad_norm": 2.6079654693603516,
      "learning_rate": 3.8741197166174137e-07,
      "loss": 0.0442,
      "step": 16947
    },
    {
      "epoch": 0.8751871933901368,
      "grad_norm": 9.770862579345703,
      "learning_rate": 3.870957972463829e-07,
      "loss": 0.1284,
      "step": 16948
    },
    {
      "epoch": 0.8752388329460367,
      "grad_norm": 12.294923782348633,
      "learning_rate": 3.8677974670630393e-07,
      "loss": 0.0842,
      "step": 16949
    },
    {
      "epoch": 0.8752904725019365,
      "grad_norm": 8.374826431274414,
      "learning_rate": 3.8646382004999095e-07,
      "loss": 0.0742,
      "step": 16950
    },
    {
      "epoch": 0.8753421120578363,
      "grad_norm": 4.296138286590576,
      "learning_rate": 3.861480172859278e-07,
      "loss": 0.0968,
      "step": 16951
    },
    {
      "epoch": 0.8753937516137361,
      "grad_norm": 20.15962028503418,
      "learning_rate": 3.858323384225959e-07,
      "loss": 0.0825,
      "step": 16952
    },
    {
      "epoch": 0.8754453911696359,
      "grad_norm": 16.778791427612305,
      "learning_rate": 3.85516783468472e-07,
      "loss": 0.0778,
      "step": 16953
    },
    {
      "epoch": 0.8754970307255358,
      "grad_norm": 11.32385540008545,
      "learning_rate": 3.852013524320297e-07,
      "loss": 0.091,
      "step": 16954
    },
    {
      "epoch": 0.8755486702814356,
      "grad_norm": 6.514889240264893,
      "learning_rate": 3.848860453217401e-07,
      "loss": 0.0898,
      "step": 16955
    },
    {
      "epoch": 0.8756003098373354,
      "grad_norm": 1.8335002660751343,
      "learning_rate": 3.8457086214607084e-07,
      "loss": 0.024,
      "step": 16956
    },
    {
      "epoch": 0.8756519493932352,
      "grad_norm": 2.670060157775879,
      "learning_rate": 3.8425580291348575e-07,
      "loss": 0.0284,
      "step": 16957
    },
    {
      "epoch": 0.875703588949135,
      "grad_norm": 9.945151329040527,
      "learning_rate": 3.8394086763244356e-07,
      "loss": 0.0867,
      "step": 16958
    },
    {
      "epoch": 0.8757552285050348,
      "grad_norm": 6.109004974365234,
      "learning_rate": 3.836260563114036e-07,
      "loss": 0.0894,
      "step": 16959
    },
    {
      "epoch": 0.8758068680609347,
      "grad_norm": 2.8093292713165283,
      "learning_rate": 3.833113689588197e-07,
      "loss": 0.0356,
      "step": 16960
    },
    {
      "epoch": 0.8758585076168345,
      "grad_norm": 1.7894325256347656,
      "learning_rate": 3.829968055831429e-07,
      "loss": 0.0123,
      "step": 16961
    },
    {
      "epoch": 0.8759101471727343,
      "grad_norm": 2.222964286804199,
      "learning_rate": 3.826823661928186e-07,
      "loss": 0.039,
      "step": 16962
    },
    {
      "epoch": 0.8759617867286341,
      "grad_norm": 3.1069765090942383,
      "learning_rate": 3.8236805079629125e-07,
      "loss": 0.0635,
      "step": 16963
    },
    {
      "epoch": 0.8760134262845339,
      "grad_norm": 4.899881839752197,
      "learning_rate": 3.8205385940200346e-07,
      "loss": 0.0758,
      "step": 16964
    },
    {
      "epoch": 0.8760650658404338,
      "grad_norm": 2.560127019882202,
      "learning_rate": 3.8173979201839127e-07,
      "loss": 0.0228,
      "step": 16965
    },
    {
      "epoch": 0.8761167053963336,
      "grad_norm": 3.079333543777466,
      "learning_rate": 3.8142584865388855e-07,
      "loss": 0.0319,
      "step": 16966
    },
    {
      "epoch": 0.8761683449522334,
      "grad_norm": 2.7277112007141113,
      "learning_rate": 3.811120293169257e-07,
      "loss": 0.0326,
      "step": 16967
    },
    {
      "epoch": 0.8762199845081332,
      "grad_norm": 10.758583068847656,
      "learning_rate": 3.807983340159305e-07,
      "loss": 0.0756,
      "step": 16968
    },
    {
      "epoch": 0.876271624064033,
      "grad_norm": 2.247995615005493,
      "learning_rate": 3.8048476275932665e-07,
      "loss": 0.0279,
      "step": 16969
    },
    {
      "epoch": 0.8763232636199328,
      "grad_norm": 3.553952217102051,
      "learning_rate": 3.8017131555553535e-07,
      "loss": 0.0622,
      "step": 16970
    },
    {
      "epoch": 0.8763749031758327,
      "grad_norm": 4.681534290313721,
      "learning_rate": 3.798579924129736e-07,
      "loss": 0.068,
      "step": 16971
    },
    {
      "epoch": 0.8764265427317325,
      "grad_norm": 3.6265106201171875,
      "learning_rate": 3.7954479334005534e-07,
      "loss": 0.0834,
      "step": 16972
    },
    {
      "epoch": 0.8764781822876323,
      "grad_norm": 9.101410865783691,
      "learning_rate": 3.79231718345191e-07,
      "loss": 0.0702,
      "step": 16973
    },
    {
      "epoch": 0.8765298218435321,
      "grad_norm": 3.1002490520477295,
      "learning_rate": 3.7891876743678835e-07,
      "loss": 0.0288,
      "step": 16974
    },
    {
      "epoch": 0.8765814613994319,
      "grad_norm": 11.765478134155273,
      "learning_rate": 3.7860594062325173e-07,
      "loss": 0.2863,
      "step": 16975
    },
    {
      "epoch": 0.8766331009553318,
      "grad_norm": 3.5347790718078613,
      "learning_rate": 3.782932379129817e-07,
      "loss": 0.063,
      "step": 16976
    },
    {
      "epoch": 0.8766847405112316,
      "grad_norm": 5.633111000061035,
      "learning_rate": 3.7798065931437364e-07,
      "loss": 0.0501,
      "step": 16977
    },
    {
      "epoch": 0.8767363800671314,
      "grad_norm": 1.9067760705947876,
      "learning_rate": 3.7766820483582434e-07,
      "loss": 0.0385,
      "step": 16978
    },
    {
      "epoch": 0.8767880196230312,
      "grad_norm": 6.483966827392578,
      "learning_rate": 3.7735587448572363e-07,
      "loss": 0.0691,
      "step": 16979
    },
    {
      "epoch": 0.876839659178931,
      "grad_norm": 5.025321960449219,
      "learning_rate": 3.770436682724582e-07,
      "loss": 0.0332,
      "step": 16980
    },
    {
      "epoch": 0.8768912987348308,
      "grad_norm": 5.95098352432251,
      "learning_rate": 3.767315862044124e-07,
      "loss": 0.0295,
      "step": 16981
    },
    {
      "epoch": 0.8769429382907307,
      "grad_norm": 3.871657371520996,
      "learning_rate": 3.764196282899657e-07,
      "loss": 0.0476,
      "step": 16982
    },
    {
      "epoch": 0.8769945778466305,
      "grad_norm": 4.976700782775879,
      "learning_rate": 3.7610779453749856e-07,
      "loss": 0.0626,
      "step": 16983
    },
    {
      "epoch": 0.8770462174025303,
      "grad_norm": 7.0262250900268555,
      "learning_rate": 3.757960849553821e-07,
      "loss": 0.0908,
      "step": 16984
    },
    {
      "epoch": 0.8770978569584301,
      "grad_norm": 4.760168552398682,
      "learning_rate": 3.7548449955198796e-07,
      "loss": 0.067,
      "step": 16985
    },
    {
      "epoch": 0.8771494965143299,
      "grad_norm": 7.850831508636475,
      "learning_rate": 3.7517303833568387e-07,
      "loss": 0.0428,
      "step": 16986
    },
    {
      "epoch": 0.8772011360702298,
      "grad_norm": 17.919513702392578,
      "learning_rate": 3.7486170131483313e-07,
      "loss": 0.1227,
      "step": 16987
    },
    {
      "epoch": 0.8772527756261296,
      "grad_norm": 1.2156438827514648,
      "learning_rate": 3.7455048849779687e-07,
      "loss": 0.0224,
      "step": 16988
    },
    {
      "epoch": 0.8773044151820294,
      "grad_norm": 13.280365943908691,
      "learning_rate": 3.742393998929322e-07,
      "loss": 0.1064,
      "step": 16989
    },
    {
      "epoch": 0.8773560547379292,
      "grad_norm": 1.828802227973938,
      "learning_rate": 3.7392843550859426e-07,
      "loss": 0.0062,
      "step": 16990
    },
    {
      "epoch": 0.877407694293829,
      "grad_norm": 5.881907939910889,
      "learning_rate": 3.736175953531307e-07,
      "loss": 0.0718,
      "step": 16991
    },
    {
      "epoch": 0.8774593338497289,
      "grad_norm": 7.054234027862549,
      "learning_rate": 3.733068794348915e-07,
      "loss": 0.0607,
      "step": 16992
    },
    {
      "epoch": 0.8775109734056287,
      "grad_norm": 9.79078483581543,
      "learning_rate": 3.7299628776222006e-07,
      "loss": 0.0501,
      "step": 16993
    },
    {
      "epoch": 0.8775626129615285,
      "grad_norm": 3.1667377948760986,
      "learning_rate": 3.72685820343458e-07,
      "loss": 0.0455,
      "step": 16994
    },
    {
      "epoch": 0.8776142525174283,
      "grad_norm": 24.767656326293945,
      "learning_rate": 3.7237547718694034e-07,
      "loss": 0.1165,
      "step": 16995
    },
    {
      "epoch": 0.8776658920733281,
      "grad_norm": 4.357517719268799,
      "learning_rate": 3.72065258301002e-07,
      "loss": 0.0577,
      "step": 16996
    },
    {
      "epoch": 0.8777175316292279,
      "grad_norm": 21.95103645324707,
      "learning_rate": 3.717551636939742e-07,
      "loss": 0.1,
      "step": 16997
    },
    {
      "epoch": 0.8777691711851278,
      "grad_norm": 13.055403709411621,
      "learning_rate": 3.7144519337418473e-07,
      "loss": 0.0972,
      "step": 16998
    },
    {
      "epoch": 0.8778208107410276,
      "grad_norm": 10.48381233215332,
      "learning_rate": 3.7113534734995625e-07,
      "loss": 0.0607,
      "step": 16999
    },
    {
      "epoch": 0.8778724502969274,
      "grad_norm": 5.083340644836426,
      "learning_rate": 3.7082562562961e-07,
      "loss": 0.0701,
      "step": 17000
    },
    {
      "epoch": 0.8779240898528272,
      "grad_norm": 46.17451095581055,
      "learning_rate": 3.7051602822146317e-07,
      "loss": 0.1212,
      "step": 17001
    },
    {
      "epoch": 0.877975729408727,
      "grad_norm": 3.4636037349700928,
      "learning_rate": 3.7020655513382906e-07,
      "loss": 0.0685,
      "step": 17002
    },
    {
      "epoch": 0.8780273689646269,
      "grad_norm": 7.805372714996338,
      "learning_rate": 3.6989720637501947e-07,
      "loss": 0.0473,
      "step": 17003
    },
    {
      "epoch": 0.8780790085205267,
      "grad_norm": 3.2895750999450684,
      "learning_rate": 3.69587981953341e-07,
      "loss": 0.0361,
      "step": 17004
    },
    {
      "epoch": 0.8781306480764265,
      "grad_norm": 8.72175407409668,
      "learning_rate": 3.6927888187709817e-07,
      "loss": 0.0643,
      "step": 17005
    },
    {
      "epoch": 0.8781822876323263,
      "grad_norm": 2.3563506603240967,
      "learning_rate": 3.6896990615459037e-07,
      "loss": 0.0416,
      "step": 17006
    },
    {
      "epoch": 0.8782339271882261,
      "grad_norm": 2.5324177742004395,
      "learning_rate": 3.686610547941161e-07,
      "loss": 0.034,
      "step": 17007
    },
    {
      "epoch": 0.878285566744126,
      "grad_norm": 2.6126317977905273,
      "learning_rate": 3.683523278039697e-07,
      "loss": 0.0377,
      "step": 17008
    },
    {
      "epoch": 0.8783372063000259,
      "grad_norm": 3.0338566303253174,
      "learning_rate": 3.6804372519244013e-07,
      "loss": 0.0572,
      "step": 17009
    },
    {
      "epoch": 0.8783888458559257,
      "grad_norm": 2.7117395401000977,
      "learning_rate": 3.677352469678142e-07,
      "loss": 0.0248,
      "step": 17010
    },
    {
      "epoch": 0.8784404854118255,
      "grad_norm": 4.399319648742676,
      "learning_rate": 3.674268931383779e-07,
      "loss": 0.0428,
      "step": 17011
    },
    {
      "epoch": 0.8784921249677253,
      "grad_norm": 3.856213092803955,
      "learning_rate": 3.6711866371241134e-07,
      "loss": 0.0473,
      "step": 17012
    },
    {
      "epoch": 0.8785437645236251,
      "grad_norm": 13.682027816772461,
      "learning_rate": 3.6681055869819073e-07,
      "loss": 0.1335,
      "step": 17013
    },
    {
      "epoch": 0.878595404079525,
      "grad_norm": 3.242879867553711,
      "learning_rate": 3.665025781039905e-07,
      "loss": 0.0379,
      "step": 17014
    },
    {
      "epoch": 0.8786470436354248,
      "grad_norm": 2.07987380027771,
      "learning_rate": 3.6619472193808015e-07,
      "loss": 0.0389,
      "step": 17015
    },
    {
      "epoch": 0.8786986831913246,
      "grad_norm": 14.017802238464355,
      "learning_rate": 3.6588699020872977e-07,
      "loss": 0.1066,
      "step": 17016
    },
    {
      "epoch": 0.8787503227472244,
      "grad_norm": 3.56223726272583,
      "learning_rate": 3.6557938292419993e-07,
      "loss": 0.0392,
      "step": 17017
    },
    {
      "epoch": 0.8788019623031242,
      "grad_norm": 38.52531433105469,
      "learning_rate": 3.652719000927529e-07,
      "loss": 0.1838,
      "step": 17018
    },
    {
      "epoch": 0.8788536018590241,
      "grad_norm": 17.751710891723633,
      "learning_rate": 3.649645417226455e-07,
      "loss": 0.1706,
      "step": 17019
    },
    {
      "epoch": 0.8789052414149239,
      "grad_norm": 1.6719919443130493,
      "learning_rate": 3.646573078221316e-07,
      "loss": 0.0179,
      "step": 17020
    },
    {
      "epoch": 0.8789568809708237,
      "grad_norm": 12.762763977050781,
      "learning_rate": 3.643501983994613e-07,
      "loss": 0.0282,
      "step": 17021
    },
    {
      "epoch": 0.8790085205267235,
      "grad_norm": 2.433433771133423,
      "learning_rate": 3.640432134628824e-07,
      "loss": 0.0365,
      "step": 17022
    },
    {
      "epoch": 0.8790601600826233,
      "grad_norm": 13.286643028259277,
      "learning_rate": 3.637363530206389e-07,
      "loss": 0.1454,
      "step": 17023
    },
    {
      "epoch": 0.8791117996385232,
      "grad_norm": 4.271561622619629,
      "learning_rate": 3.634296170809687e-07,
      "loss": 0.0679,
      "step": 17024
    },
    {
      "epoch": 0.879163439194423,
      "grad_norm": 3.839550733566284,
      "learning_rate": 3.6312300565211233e-07,
      "loss": 0.0369,
      "step": 17025
    },
    {
      "epoch": 0.8792150787503228,
      "grad_norm": 3.968674421310425,
      "learning_rate": 3.628165187423022e-07,
      "loss": 0.051,
      "step": 17026
    },
    {
      "epoch": 0.8792667183062226,
      "grad_norm": 1.7469555139541626,
      "learning_rate": 3.6251015635976885e-07,
      "loss": 0.0196,
      "step": 17027
    },
    {
      "epoch": 0.8793183578621224,
      "grad_norm": 2.7738449573516846,
      "learning_rate": 3.622039185127385e-07,
      "loss": 0.0459,
      "step": 17028
    },
    {
      "epoch": 0.8793699974180222,
      "grad_norm": 3.225433826446533,
      "learning_rate": 3.618978052094352e-07,
      "loss": 0.0813,
      "step": 17029
    },
    {
      "epoch": 0.8794216369739221,
      "grad_norm": 3.2395472526550293,
      "learning_rate": 3.6159181645808116e-07,
      "loss": 0.0165,
      "step": 17030
    },
    {
      "epoch": 0.8794732765298219,
      "grad_norm": 3.637291193008423,
      "learning_rate": 3.61285952266891e-07,
      "loss": 0.0313,
      "step": 17031
    },
    {
      "epoch": 0.8795249160857217,
      "grad_norm": 6.971301555633545,
      "learning_rate": 3.609802126440792e-07,
      "loss": 0.061,
      "step": 17032
    },
    {
      "epoch": 0.8795765556416215,
      "grad_norm": 3.8432435989379883,
      "learning_rate": 3.606745975978565e-07,
      "loss": 0.0544,
      "step": 17033
    },
    {
      "epoch": 0.8796281951975213,
      "grad_norm": 3.7500078678131104,
      "learning_rate": 3.6036910713642957e-07,
      "loss": 0.0752,
      "step": 17034
    },
    {
      "epoch": 0.8796798347534212,
      "grad_norm": 20.172338485717773,
      "learning_rate": 3.6006374126800247e-07,
      "loss": 0.1157,
      "step": 17035
    },
    {
      "epoch": 0.879731474309321,
      "grad_norm": 10.193780899047852,
      "learning_rate": 3.597585000007747e-07,
      "loss": 0.0397,
      "step": 17036
    },
    {
      "epoch": 0.8797831138652208,
      "grad_norm": 1.990030288696289,
      "learning_rate": 3.594533833429442e-07,
      "loss": 0.0227,
      "step": 17037
    },
    {
      "epoch": 0.8798347534211206,
      "grad_norm": 4.034782409667969,
      "learning_rate": 3.591483913027044e-07,
      "loss": 0.0405,
      "step": 17038
    },
    {
      "epoch": 0.8798863929770204,
      "grad_norm": 1.9552212953567505,
      "learning_rate": 3.5884352388824483e-07,
      "loss": 0.0296,
      "step": 17039
    },
    {
      "epoch": 0.8799380325329202,
      "grad_norm": 2.238645076751709,
      "learning_rate": 3.585387811077534e-07,
      "loss": 0.0387,
      "step": 17040
    },
    {
      "epoch": 0.8799896720888201,
      "grad_norm": 5.5864715576171875,
      "learning_rate": 3.582341629694136e-07,
      "loss": 0.0633,
      "step": 17041
    },
    {
      "epoch": 0.8800413116447199,
      "grad_norm": 2.2492051124572754,
      "learning_rate": 3.57929669481405e-07,
      "loss": 0.032,
      "step": 17042
    },
    {
      "epoch": 0.8800929512006197,
      "grad_norm": 2.9330215454101562,
      "learning_rate": 3.576253006519037e-07,
      "loss": 0.0558,
      "step": 17043
    },
    {
      "epoch": 0.8801445907565195,
      "grad_norm": 2.4298288822174072,
      "learning_rate": 3.5732105648908497e-07,
      "loss": 0.0199,
      "step": 17044
    },
    {
      "epoch": 0.8801962303124193,
      "grad_norm": 2.691067695617676,
      "learning_rate": 3.570169370011195e-07,
      "loss": 0.0264,
      "step": 17045
    },
    {
      "epoch": 0.8802478698683192,
      "grad_norm": 5.13747501373291,
      "learning_rate": 3.5671294219617235e-07,
      "loss": 0.0663,
      "step": 17046
    },
    {
      "epoch": 0.880299509424219,
      "grad_norm": 3.6391303539276123,
      "learning_rate": 3.564090720824076e-07,
      "loss": 0.0564,
      "step": 17047
    },
    {
      "epoch": 0.8803511489801188,
      "grad_norm": 5.691118240356445,
      "learning_rate": 3.561053266679848e-07,
      "loss": 0.0292,
      "step": 17048
    },
    {
      "epoch": 0.8804027885360186,
      "grad_norm": 6.46207857131958,
      "learning_rate": 3.5580170596106247e-07,
      "loss": 0.0793,
      "step": 17049
    },
    {
      "epoch": 0.8804544280919184,
      "grad_norm": 2.023874282836914,
      "learning_rate": 3.554982099697929e-07,
      "loss": 0.0396,
      "step": 17050
    },
    {
      "epoch": 0.8805060676478182,
      "grad_norm": 3.331047296524048,
      "learning_rate": 3.5519483870232575e-07,
      "loss": 0.0809,
      "step": 17051
    },
    {
      "epoch": 0.8805577072037181,
      "grad_norm": 3.2210052013397217,
      "learning_rate": 3.548915921668089e-07,
      "loss": 0.0869,
      "step": 17052
    },
    {
      "epoch": 0.8806093467596179,
      "grad_norm": 7.929963111877441,
      "learning_rate": 3.545884703713848e-07,
      "loss": 0.0703,
      "step": 17053
    },
    {
      "epoch": 0.8806609863155177,
      "grad_norm": 4.898981094360352,
      "learning_rate": 3.5428547332419405e-07,
      "loss": 0.0939,
      "step": 17054
    },
    {
      "epoch": 0.8807126258714175,
      "grad_norm": 2.1293063163757324,
      "learning_rate": 3.539826010333736e-07,
      "loss": 0.0307,
      "step": 17055
    },
    {
      "epoch": 0.8807642654273173,
      "grad_norm": 5.8588995933532715,
      "learning_rate": 3.536798535070568e-07,
      "loss": 0.0655,
      "step": 17056
    },
    {
      "epoch": 0.8808159049832172,
      "grad_norm": 5.2162041664123535,
      "learning_rate": 3.533772307533717e-07,
      "loss": 0.0768,
      "step": 17057
    },
    {
      "epoch": 0.880867544539117,
      "grad_norm": 3.5601084232330322,
      "learning_rate": 3.5307473278044735e-07,
      "loss": 0.0659,
      "step": 17058
    },
    {
      "epoch": 0.8809191840950168,
      "grad_norm": 11.916382789611816,
      "learning_rate": 3.527723595964067e-07,
      "loss": 0.0486,
      "step": 17059
    },
    {
      "epoch": 0.8809708236509166,
      "grad_norm": 5.348159313201904,
      "learning_rate": 3.5247011120936924e-07,
      "loss": 0.0584,
      "step": 17060
    },
    {
      "epoch": 0.8810224632068164,
      "grad_norm": 3.123913288116455,
      "learning_rate": 3.521679876274514e-07,
      "loss": 0.0228,
      "step": 17061
    },
    {
      "epoch": 0.8810741027627162,
      "grad_norm": 1.9886897802352905,
      "learning_rate": 3.5186598885876556e-07,
      "loss": 0.0262,
      "step": 17062
    },
    {
      "epoch": 0.8811257423186161,
      "grad_norm": 1.4557491540908813,
      "learning_rate": 3.515641149114246e-07,
      "loss": 0.0272,
      "step": 17063
    },
    {
      "epoch": 0.8811773818745159,
      "grad_norm": 4.268744945526123,
      "learning_rate": 3.512623657935321e-07,
      "loss": 0.0461,
      "step": 17064
    },
    {
      "epoch": 0.8812290214304157,
      "grad_norm": 3.7389848232269287,
      "learning_rate": 3.5096074151319217e-07,
      "loss": 0.0288,
      "step": 17065
    },
    {
      "epoch": 0.8812806609863155,
      "grad_norm": 11.452338218688965,
      "learning_rate": 3.5065924207850486e-07,
      "loss": 0.0351,
      "step": 17066
    },
    {
      "epoch": 0.8813323005422153,
      "grad_norm": 6.626824855804443,
      "learning_rate": 3.5035786749756607e-07,
      "loss": 0.071,
      "step": 17067
    },
    {
      "epoch": 0.8813839400981152,
      "grad_norm": 6.008721828460693,
      "learning_rate": 3.5005661777847035e-07,
      "loss": 0.0804,
      "step": 17068
    },
    {
      "epoch": 0.881435579654015,
      "grad_norm": 7.693761825561523,
      "learning_rate": 3.4975549292930566e-07,
      "loss": 0.0749,
      "step": 17069
    },
    {
      "epoch": 0.8814872192099148,
      "grad_norm": 2.0663559436798096,
      "learning_rate": 3.4945449295816006e-07,
      "loss": 0.0331,
      "step": 17070
    },
    {
      "epoch": 0.8815388587658146,
      "grad_norm": 1.3552597761154175,
      "learning_rate": 3.4915361787311586e-07,
      "loss": 0.0194,
      "step": 17071
    },
    {
      "epoch": 0.8815904983217144,
      "grad_norm": 2.9881529808044434,
      "learning_rate": 3.4885286768225224e-07,
      "loss": 0.0526,
      "step": 17072
    },
    {
      "epoch": 0.8816421378776143,
      "grad_norm": 5.589273452758789,
      "learning_rate": 3.485522423936466e-07,
      "loss": 0.0872,
      "step": 17073
    },
    {
      "epoch": 0.8816937774335141,
      "grad_norm": 14.332157135009766,
      "learning_rate": 3.4825174201537246e-07,
      "loss": 0.0548,
      "step": 17074
    },
    {
      "epoch": 0.8817454169894139,
      "grad_norm": 4.810156345367432,
      "learning_rate": 3.479513665554973e-07,
      "loss": 0.03,
      "step": 17075
    },
    {
      "epoch": 0.8817970565453137,
      "grad_norm": 6.35607385635376,
      "learning_rate": 3.47651116022088e-07,
      "loss": 0.0643,
      "step": 17076
    },
    {
      "epoch": 0.8818486961012135,
      "grad_norm": 18.768978118896484,
      "learning_rate": 3.473509904232086e-07,
      "loss": 0.0892,
      "step": 17077
    },
    {
      "epoch": 0.8819003356571133,
      "grad_norm": 2.372361660003662,
      "learning_rate": 3.4705098976691995e-07,
      "loss": 0.0337,
      "step": 17078
    },
    {
      "epoch": 0.8819519752130132,
      "grad_norm": 4.479645252227783,
      "learning_rate": 3.46751114061275e-07,
      "loss": 0.0544,
      "step": 17079
    },
    {
      "epoch": 0.882003614768913,
      "grad_norm": 2.2083818912506104,
      "learning_rate": 3.464513633143285e-07,
      "loss": 0.042,
      "step": 17080
    },
    {
      "epoch": 0.8820552543248128,
      "grad_norm": 8.304826736450195,
      "learning_rate": 3.461517375341289e-07,
      "loss": 0.0491,
      "step": 17081
    },
    {
      "epoch": 0.8821068938807126,
      "grad_norm": 3.4369163513183594,
      "learning_rate": 3.458522367287248e-07,
      "loss": 0.0333,
      "step": 17082
    },
    {
      "epoch": 0.8821585334366124,
      "grad_norm": 2.832348346710205,
      "learning_rate": 3.4555286090615645e-07,
      "loss": 0.0288,
      "step": 17083
    },
    {
      "epoch": 0.8822101729925123,
      "grad_norm": 1.2079249620437622,
      "learning_rate": 3.452536100744647e-07,
      "loss": 0.0269,
      "step": 17084
    },
    {
      "epoch": 0.8822618125484121,
      "grad_norm": 14.865492820739746,
      "learning_rate": 3.4495448424168464e-07,
      "loss": 0.0792,
      "step": 17085
    },
    {
      "epoch": 0.8823134521043119,
      "grad_norm": 4.315316200256348,
      "learning_rate": 3.4465548341584997e-07,
      "loss": 0.0546,
      "step": 17086
    },
    {
      "epoch": 0.8823650916602117,
      "grad_norm": 4.804458141326904,
      "learning_rate": 3.443566076049898e-07,
      "loss": 0.0501,
      "step": 17087
    },
    {
      "epoch": 0.8824167312161115,
      "grad_norm": 2.558523416519165,
      "learning_rate": 3.440578568171299e-07,
      "loss": 0.0375,
      "step": 17088
    },
    {
      "epoch": 0.8824683707720113,
      "grad_norm": 14.557670593261719,
      "learning_rate": 3.4375923106029384e-07,
      "loss": 0.0705,
      "step": 17089
    },
    {
      "epoch": 0.8825200103279112,
      "grad_norm": 1.7892330884933472,
      "learning_rate": 3.434607303424986e-07,
      "loss": 0.0234,
      "step": 17090
    },
    {
      "epoch": 0.882571649883811,
      "grad_norm": 3.6766624450683594,
      "learning_rate": 3.4316235467176275e-07,
      "loss": 0.0486,
      "step": 17091
    },
    {
      "epoch": 0.8826232894397108,
      "grad_norm": 5.6851677894592285,
      "learning_rate": 3.428641040560976e-07,
      "loss": 0.0747,
      "step": 17092
    },
    {
      "epoch": 0.8826749289956106,
      "grad_norm": 10.858657836914062,
      "learning_rate": 3.42565978503514e-07,
      "loss": 0.0792,
      "step": 17093
    },
    {
      "epoch": 0.8827265685515104,
      "grad_norm": 6.918763160705566,
      "learning_rate": 3.4226797802201505e-07,
      "loss": 0.0655,
      "step": 17094
    },
    {
      "epoch": 0.8827782081074103,
      "grad_norm": 4.303468704223633,
      "learning_rate": 3.4197010261960427e-07,
      "loss": 0.0405,
      "step": 17095
    },
    {
      "epoch": 0.8828298476633101,
      "grad_norm": 5.04197359085083,
      "learning_rate": 3.416723523042831e-07,
      "loss": 0.0439,
      "step": 17096
    },
    {
      "epoch": 0.8828814872192099,
      "grad_norm": 4.640840530395508,
      "learning_rate": 3.413747270840445e-07,
      "loss": 0.0534,
      "step": 17097
    },
    {
      "epoch": 0.8829331267751097,
      "grad_norm": 14.95908260345459,
      "learning_rate": 3.4107722696688216e-07,
      "loss": 0.0538,
      "step": 17098
    },
    {
      "epoch": 0.8829847663310095,
      "grad_norm": 3.3879306316375732,
      "learning_rate": 3.4077985196078524e-07,
      "loss": 0.0254,
      "step": 17099
    },
    {
      "epoch": 0.8830364058869093,
      "grad_norm": 3.86430287361145,
      "learning_rate": 3.404826020737384e-07,
      "loss": 0.0365,
      "step": 17100
    },
    {
      "epoch": 0.8830880454428092,
      "grad_norm": 3.0245301723480225,
      "learning_rate": 3.4018547731372584e-07,
      "loss": 0.0463,
      "step": 17101
    },
    {
      "epoch": 0.883139684998709,
      "grad_norm": 6.269627571105957,
      "learning_rate": 3.398884776887246e-07,
      "loss": 0.068,
      "step": 17102
    },
    {
      "epoch": 0.8831913245546088,
      "grad_norm": 2.8312196731567383,
      "learning_rate": 3.39591603206712e-07,
      "loss": 0.0538,
      "step": 17103
    },
    {
      "epoch": 0.8832429641105086,
      "grad_norm": 5.663372993469238,
      "learning_rate": 3.3929485387565963e-07,
      "loss": 0.0566,
      "step": 17104
    },
    {
      "epoch": 0.8832946036664084,
      "grad_norm": 2.381901979446411,
      "learning_rate": 3.38998229703536e-07,
      "loss": 0.0323,
      "step": 17105
    },
    {
      "epoch": 0.8833462432223083,
      "grad_norm": 14.33906078338623,
      "learning_rate": 3.3870173069830705e-07,
      "loss": 0.0997,
      "step": 17106
    },
    {
      "epoch": 0.8833978827782081,
      "grad_norm": 4.31951904296875,
      "learning_rate": 3.384053568679363e-07,
      "loss": 0.0517,
      "step": 17107
    },
    {
      "epoch": 0.8834495223341079,
      "grad_norm": 3.9214463233947754,
      "learning_rate": 3.3810910822038023e-07,
      "loss": 0.0363,
      "step": 17108
    },
    {
      "epoch": 0.8835011618900077,
      "grad_norm": 1.506944179534912,
      "learning_rate": 3.378129847635947e-07,
      "loss": 0.0168,
      "step": 17109
    },
    {
      "epoch": 0.8835528014459075,
      "grad_norm": 2.1668548583984375,
      "learning_rate": 3.3751698650553334e-07,
      "loss": 0.0229,
      "step": 17110
    },
    {
      "epoch": 0.8836044410018074,
      "grad_norm": 12.336135864257812,
      "learning_rate": 3.372211134541448e-07,
      "loss": 0.0816,
      "step": 17111
    },
    {
      "epoch": 0.8836560805577072,
      "grad_norm": 6.26698637008667,
      "learning_rate": 3.369253656173732e-07,
      "loss": 0.0674,
      "step": 17112
    },
    {
      "epoch": 0.883707720113607,
      "grad_norm": 3.680612564086914,
      "learning_rate": 3.366297430031612e-07,
      "loss": 0.0509,
      "step": 17113
    },
    {
      "epoch": 0.8837593596695068,
      "grad_norm": 2.110551118850708,
      "learning_rate": 3.363342456194463e-07,
      "loss": 0.0425,
      "step": 17114
    },
    {
      "epoch": 0.8838109992254066,
      "grad_norm": 12.178422927856445,
      "learning_rate": 3.3603887347416706e-07,
      "loss": 0.0941,
      "step": 17115
    },
    {
      "epoch": 0.8838626387813064,
      "grad_norm": 2.5330147743225098,
      "learning_rate": 3.357436265752523e-07,
      "loss": 0.0344,
      "step": 17116
    },
    {
      "epoch": 0.8839142783372063,
      "grad_norm": 13.673796653747559,
      "learning_rate": 3.354485049306311e-07,
      "loss": 0.0931,
      "step": 17117
    },
    {
      "epoch": 0.8839659178931061,
      "grad_norm": 3.1781632900238037,
      "learning_rate": 3.351535085482299e-07,
      "loss": 0.0305,
      "step": 17118
    },
    {
      "epoch": 0.8840175574490059,
      "grad_norm": 3.0990328788757324,
      "learning_rate": 3.348586374359697e-07,
      "loss": 0.0355,
      "step": 17119
    },
    {
      "epoch": 0.8840691970049057,
      "grad_norm": 4.227257251739502,
      "learning_rate": 3.3456389160176903e-07,
      "loss": 0.0626,
      "step": 17120
    },
    {
      "epoch": 0.8841208365608055,
      "grad_norm": 7.686824321746826,
      "learning_rate": 3.342692710535433e-07,
      "loss": 0.0364,
      "step": 17121
    },
    {
      "epoch": 0.8841724761167054,
      "grad_norm": 2.301787853240967,
      "learning_rate": 3.3397477579920457e-07,
      "loss": 0.0369,
      "step": 17122
    },
    {
      "epoch": 0.8842241156726052,
      "grad_norm": 6.186329364776611,
      "learning_rate": 3.336804058466597e-07,
      "loss": 0.0903,
      "step": 17123
    },
    {
      "epoch": 0.884275755228505,
      "grad_norm": 10.902061462402344,
      "learning_rate": 3.333861612038153e-07,
      "loss": 0.0841,
      "step": 17124
    },
    {
      "epoch": 0.8843273947844048,
      "grad_norm": 3.548956871032715,
      "learning_rate": 3.3309204187857213e-07,
      "loss": 0.0439,
      "step": 17125
    },
    {
      "epoch": 0.8843790343403046,
      "grad_norm": 2.5225491523742676,
      "learning_rate": 3.3279804787883006e-07,
      "loss": 0.0231,
      "step": 17126
    },
    {
      "epoch": 0.8844306738962044,
      "grad_norm": 1.4470266103744507,
      "learning_rate": 3.3250417921248225e-07,
      "loss": 0.0107,
      "step": 17127
    },
    {
      "epoch": 0.8844823134521043,
      "grad_norm": 3.3655388355255127,
      "learning_rate": 3.322104358874201e-07,
      "loss": 0.023,
      "step": 17128
    },
    {
      "epoch": 0.8845339530080041,
      "grad_norm": 3.3089544773101807,
      "learning_rate": 3.3191681791153394e-07,
      "loss": 0.0476,
      "step": 17129
    },
    {
      "epoch": 0.8845855925639039,
      "grad_norm": 2.0351479053497314,
      "learning_rate": 3.3162332529270647e-07,
      "loss": 0.0426,
      "step": 17130
    },
    {
      "epoch": 0.8846372321198037,
      "grad_norm": 14.123575210571289,
      "learning_rate": 3.3132995803882015e-07,
      "loss": 0.1271,
      "step": 17131
    },
    {
      "epoch": 0.8846888716757035,
      "grad_norm": 2.2749714851379395,
      "learning_rate": 3.3103671615775325e-07,
      "loss": 0.0328,
      "step": 17132
    },
    {
      "epoch": 0.8847405112316034,
      "grad_norm": 4.284722328186035,
      "learning_rate": 3.3074359965737935e-07,
      "loss": 0.1091,
      "step": 17133
    },
    {
      "epoch": 0.8847921507875032,
      "grad_norm": 26.427663803100586,
      "learning_rate": 3.304506085455711e-07,
      "loss": 0.0622,
      "step": 17134
    },
    {
      "epoch": 0.884843790343403,
      "grad_norm": 3.9378557205200195,
      "learning_rate": 3.301577428301961e-07,
      "loss": 0.0235,
      "step": 17135
    },
    {
      "epoch": 0.8848954298993029,
      "grad_norm": 2.2852039337158203,
      "learning_rate": 3.298650025191186e-07,
      "loss": 0.0235,
      "step": 17136
    },
    {
      "epoch": 0.8849470694552027,
      "grad_norm": 3.7346389293670654,
      "learning_rate": 3.2957238762020063e-07,
      "loss": 0.0525,
      "step": 17137
    },
    {
      "epoch": 0.8849987090111026,
      "grad_norm": 15.66144847869873,
      "learning_rate": 3.292798981412992e-07,
      "loss": 0.1153,
      "step": 17138
    },
    {
      "epoch": 0.8850503485670024,
      "grad_norm": 1.5090267658233643,
      "learning_rate": 3.289875340902698e-07,
      "loss": 0.0208,
      "step": 17139
    },
    {
      "epoch": 0.8851019881229022,
      "grad_norm": 3.0983362197875977,
      "learning_rate": 3.2869529547496335e-07,
      "loss": 0.0436,
      "step": 17140
    },
    {
      "epoch": 0.885153627678802,
      "grad_norm": 2.96633243560791,
      "learning_rate": 3.284031823032269e-07,
      "loss": 0.0293,
      "step": 17141
    },
    {
      "epoch": 0.8852052672347018,
      "grad_norm": 4.696234226226807,
      "learning_rate": 3.2811119458290466e-07,
      "loss": 0.0643,
      "step": 17142
    },
    {
      "epoch": 0.8852569067906016,
      "grad_norm": 2.279855251312256,
      "learning_rate": 3.2781933232183927e-07,
      "loss": 0.038,
      "step": 17143
    },
    {
      "epoch": 0.8853085463465015,
      "grad_norm": 6.628117084503174,
      "learning_rate": 3.2752759552786837e-07,
      "loss": 0.0487,
      "step": 17144
    },
    {
      "epoch": 0.8853601859024013,
      "grad_norm": 2.088892698287964,
      "learning_rate": 3.2723598420882465e-07,
      "loss": 0.0549,
      "step": 17145
    },
    {
      "epoch": 0.8854118254583011,
      "grad_norm": 8.437918663024902,
      "learning_rate": 3.269444983725395e-07,
      "loss": 0.1323,
      "step": 17146
    },
    {
      "epoch": 0.8854634650142009,
      "grad_norm": 1.2590839862823486,
      "learning_rate": 3.266531380268406e-07,
      "loss": 0.0296,
      "step": 17147
    },
    {
      "epoch": 0.8855151045701007,
      "grad_norm": 5.619007587432861,
      "learning_rate": 3.263619031795534e-07,
      "loss": 0.0505,
      "step": 17148
    },
    {
      "epoch": 0.8855667441260006,
      "grad_norm": 2.8555374145507812,
      "learning_rate": 3.260707938384977e-07,
      "loss": 0.034,
      "step": 17149
    },
    {
      "epoch": 0.8856183836819004,
      "grad_norm": 2.7650866508483887,
      "learning_rate": 3.2577981001149063e-07,
      "loss": 0.0288,
      "step": 17150
    },
    {
      "epoch": 0.8856700232378002,
      "grad_norm": 3.136798143386841,
      "learning_rate": 3.2548895170634696e-07,
      "loss": 0.0338,
      "step": 17151
    },
    {
      "epoch": 0.8857216627937,
      "grad_norm": 3.5648088455200195,
      "learning_rate": 3.251982189308772e-07,
      "loss": 0.0536,
      "step": 17152
    },
    {
      "epoch": 0.8857733023495998,
      "grad_norm": 9.279685020446777,
      "learning_rate": 3.2490761169288844e-07,
      "loss": 0.1505,
      "step": 17153
    },
    {
      "epoch": 0.8858249419054997,
      "grad_norm": 3.1166975498199463,
      "learning_rate": 3.246171300001849e-07,
      "loss": 0.0271,
      "step": 17154
    },
    {
      "epoch": 0.8858765814613995,
      "grad_norm": 2.7942047119140625,
      "learning_rate": 3.243267738605682e-07,
      "loss": 0.0296,
      "step": 17155
    },
    {
      "epoch": 0.8859282210172993,
      "grad_norm": 8.226720809936523,
      "learning_rate": 3.240365432818332e-07,
      "loss": 0.0482,
      "step": 17156
    },
    {
      "epoch": 0.8859798605731991,
      "grad_norm": 3.448096513748169,
      "learning_rate": 3.237464382717759e-07,
      "loss": 0.051,
      "step": 17157
    },
    {
      "epoch": 0.8860315001290989,
      "grad_norm": 13.568647384643555,
      "learning_rate": 3.234564588381861e-07,
      "loss": 0.0187,
      "step": 17158
    },
    {
      "epoch": 0.8860831396849987,
      "grad_norm": 7.869222164154053,
      "learning_rate": 3.2316660498885153e-07,
      "loss": 0.1019,
      "step": 17159
    },
    {
      "epoch": 0.8861347792408986,
      "grad_norm": 4.082229137420654,
      "learning_rate": 3.2287687673155434e-07,
      "loss": 0.0468,
      "step": 17160
    },
    {
      "epoch": 0.8861864187967984,
      "grad_norm": 2.3525772094726562,
      "learning_rate": 3.225872740740754e-07,
      "loss": 0.0238,
      "step": 17161
    },
    {
      "epoch": 0.8862380583526982,
      "grad_norm": 3.640167474746704,
      "learning_rate": 3.222977970241942e-07,
      "loss": 0.0552,
      "step": 17162
    },
    {
      "epoch": 0.886289697908598,
      "grad_norm": 2.0075771808624268,
      "learning_rate": 3.22008445589681e-07,
      "loss": 0.0323,
      "step": 17163
    },
    {
      "epoch": 0.8863413374644978,
      "grad_norm": 2.9972312450408936,
      "learning_rate": 3.217192197783076e-07,
      "loss": 0.043,
      "step": 17164
    },
    {
      "epoch": 0.8863929770203977,
      "grad_norm": 1.3266178369522095,
      "learning_rate": 3.214301195978409e-07,
      "loss": 0.0148,
      "step": 17165
    },
    {
      "epoch": 0.8864446165762975,
      "grad_norm": 3.3688809871673584,
      "learning_rate": 3.211411450560442e-07,
      "loss": 0.0385,
      "step": 17166
    },
    {
      "epoch": 0.8864962561321973,
      "grad_norm": 3.893376111984253,
      "learning_rate": 3.2085229616067806e-07,
      "loss": 0.0468,
      "step": 17167
    },
    {
      "epoch": 0.8865478956880971,
      "grad_norm": 1.9258195161819458,
      "learning_rate": 3.2056357291949836e-07,
      "loss": 0.024,
      "step": 17168
    },
    {
      "epoch": 0.8865995352439969,
      "grad_norm": 4.788511753082275,
      "learning_rate": 3.2027497534025953e-07,
      "loss": 0.0628,
      "step": 17169
    },
    {
      "epoch": 0.8866511747998967,
      "grad_norm": 2.36780047416687,
      "learning_rate": 3.199865034307109e-07,
      "loss": 0.0384,
      "step": 17170
    },
    {
      "epoch": 0.8867028143557966,
      "grad_norm": 3.041050434112549,
      "learning_rate": 3.1969815719859907e-07,
      "loss": 0.0435,
      "step": 17171
    },
    {
      "epoch": 0.8867544539116964,
      "grad_norm": 5.920233249664307,
      "learning_rate": 3.194099366516679e-07,
      "loss": 0.0769,
      "step": 17172
    },
    {
      "epoch": 0.8868060934675962,
      "grad_norm": 3.9820659160614014,
      "learning_rate": 3.191218417976577e-07,
      "loss": 0.0375,
      "step": 17173
    },
    {
      "epoch": 0.886857733023496,
      "grad_norm": 2.3265769481658936,
      "learning_rate": 3.188338726443035e-07,
      "loss": 0.0478,
      "step": 17174
    },
    {
      "epoch": 0.8869093725793958,
      "grad_norm": 3.8382601737976074,
      "learning_rate": 3.185460291993386e-07,
      "loss": 0.0452,
      "step": 17175
    },
    {
      "epoch": 0.8869610121352957,
      "grad_norm": 7.081324100494385,
      "learning_rate": 3.1825831147049404e-07,
      "loss": 0.0495,
      "step": 17176
    },
    {
      "epoch": 0.8870126516911955,
      "grad_norm": 2.2397987842559814,
      "learning_rate": 3.1797071946549574e-07,
      "loss": 0.0394,
      "step": 17177
    },
    {
      "epoch": 0.8870642912470953,
      "grad_norm": 2.508453130722046,
      "learning_rate": 3.176832531920665e-07,
      "loss": 0.0245,
      "step": 17178
    },
    {
      "epoch": 0.8871159308029951,
      "grad_norm": 10.549885749816895,
      "learning_rate": 3.173959126579257e-07,
      "loss": 0.0538,
      "step": 17179
    },
    {
      "epoch": 0.8871675703588949,
      "grad_norm": 9.977669715881348,
      "learning_rate": 3.171086978707888e-07,
      "loss": 0.0746,
      "step": 17180
    },
    {
      "epoch": 0.8872192099147947,
      "grad_norm": 2.844176769256592,
      "learning_rate": 3.168216088383719e-07,
      "loss": 0.0514,
      "step": 17181
    },
    {
      "epoch": 0.8872708494706946,
      "grad_norm": 2.5489745140075684,
      "learning_rate": 3.16534645568381e-07,
      "loss": 0.0216,
      "step": 17182
    },
    {
      "epoch": 0.8873224890265944,
      "grad_norm": 9.545365333557129,
      "learning_rate": 3.162478080685244e-07,
      "loss": 0.0269,
      "step": 17183
    },
    {
      "epoch": 0.8873741285824942,
      "grad_norm": 10.419720649719238,
      "learning_rate": 3.1596109634650316e-07,
      "loss": 0.0549,
      "step": 17184
    },
    {
      "epoch": 0.887425768138394,
      "grad_norm": 3.9279286861419678,
      "learning_rate": 3.156745104100184e-07,
      "loss": 0.0643,
      "step": 17185
    },
    {
      "epoch": 0.8874774076942938,
      "grad_norm": 4.594834804534912,
      "learning_rate": 3.15388050266765e-07,
      "loss": 0.063,
      "step": 17186
    },
    {
      "epoch": 0.8875290472501937,
      "grad_norm": 3.0102615356445312,
      "learning_rate": 3.151017159244357e-07,
      "loss": 0.0358,
      "step": 17187
    },
    {
      "epoch": 0.8875806868060935,
      "grad_norm": 11.151906967163086,
      "learning_rate": 3.1481550739072054e-07,
      "loss": 0.0607,
      "step": 17188
    },
    {
      "epoch": 0.8876323263619933,
      "grad_norm": 4.2949724197387695,
      "learning_rate": 3.145294246733033e-07,
      "loss": 0.0572,
      "step": 17189
    },
    {
      "epoch": 0.8876839659178931,
      "grad_norm": 3.1770832538604736,
      "learning_rate": 3.142434677798689e-07,
      "loss": 0.0541,
      "step": 17190
    },
    {
      "epoch": 0.8877356054737929,
      "grad_norm": 2.6184487342834473,
      "learning_rate": 3.139576367180958e-07,
      "loss": 0.0375,
      "step": 17191
    },
    {
      "epoch": 0.8877872450296928,
      "grad_norm": 14.83391284942627,
      "learning_rate": 3.136719314956593e-07,
      "loss": 0.1215,
      "step": 17192
    },
    {
      "epoch": 0.8878388845855926,
      "grad_norm": 5.0054779052734375,
      "learning_rate": 3.133863521202318e-07,
      "loss": 0.0442,
      "step": 17193
    },
    {
      "epoch": 0.8878905241414924,
      "grad_norm": 10.303543090820312,
      "learning_rate": 3.131008985994816e-07,
      "loss": 0.1345,
      "step": 17194
    },
    {
      "epoch": 0.8879421636973922,
      "grad_norm": 10.554515838623047,
      "learning_rate": 3.128155709410763e-07,
      "loss": 0.0951,
      "step": 17195
    },
    {
      "epoch": 0.887993803253292,
      "grad_norm": 7.983242988586426,
      "learning_rate": 3.1253036915267596e-07,
      "loss": 0.0659,
      "step": 17196
    },
    {
      "epoch": 0.8880454428091918,
      "grad_norm": 2.5557539463043213,
      "learning_rate": 3.1224529324194006e-07,
      "loss": 0.0167,
      "step": 17197
    },
    {
      "epoch": 0.8880970823650917,
      "grad_norm": 3.546776533126831,
      "learning_rate": 3.1196034321652467e-07,
      "loss": 0.0266,
      "step": 17198
    },
    {
      "epoch": 0.8881487219209915,
      "grad_norm": 1.872931718826294,
      "learning_rate": 3.1167551908408146e-07,
      "loss": 0.0261,
      "step": 17199
    },
    {
      "epoch": 0.8882003614768913,
      "grad_norm": 10.318763732910156,
      "learning_rate": 3.1139082085225924e-07,
      "loss": 0.1931,
      "step": 17200
    },
    {
      "epoch": 0.8882520010327911,
      "grad_norm": 12.463506698608398,
      "learning_rate": 3.1110624852870254e-07,
      "loss": 0.0837,
      "step": 17201
    },
    {
      "epoch": 0.8883036405886909,
      "grad_norm": 2.999101400375366,
      "learning_rate": 3.108218021210546e-07,
      "loss": 0.0374,
      "step": 17202
    },
    {
      "epoch": 0.8883552801445908,
      "grad_norm": 5.606774806976318,
      "learning_rate": 3.105374816369533e-07,
      "loss": 0.0466,
      "step": 17203
    },
    {
      "epoch": 0.8884069197004906,
      "grad_norm": 9.285202026367188,
      "learning_rate": 3.102532870840336e-07,
      "loss": 0.0847,
      "step": 17204
    },
    {
      "epoch": 0.8884585592563904,
      "grad_norm": 4.138515949249268,
      "learning_rate": 3.099692184699271e-07,
      "loss": 0.0516,
      "step": 17205
    },
    {
      "epoch": 0.8885101988122902,
      "grad_norm": 5.089714050292969,
      "learning_rate": 3.0968527580226395e-07,
      "loss": 0.0541,
      "step": 17206
    },
    {
      "epoch": 0.88856183836819,
      "grad_norm": 3.166614294052124,
      "learning_rate": 3.0940145908866683e-07,
      "loss": 0.044,
      "step": 17207
    },
    {
      "epoch": 0.8886134779240898,
      "grad_norm": 1.7712085247039795,
      "learning_rate": 3.0911776833675744e-07,
      "loss": 0.0199,
      "step": 17208
    },
    {
      "epoch": 0.8886651174799897,
      "grad_norm": 2.414940357208252,
      "learning_rate": 3.0883420355415584e-07,
      "loss": 0.0378,
      "step": 17209
    },
    {
      "epoch": 0.8887167570358895,
      "grad_norm": 4.0146613121032715,
      "learning_rate": 3.08550764748477e-07,
      "loss": 0.0789,
      "step": 17210
    },
    {
      "epoch": 0.8887683965917893,
      "grad_norm": 1.8024719953536987,
      "learning_rate": 3.082674519273304e-07,
      "loss": 0.0217,
      "step": 17211
    },
    {
      "epoch": 0.8888200361476891,
      "grad_norm": 4.175882816314697,
      "learning_rate": 3.07984265098325e-07,
      "loss": 0.0535,
      "step": 17212
    },
    {
      "epoch": 0.8888716757035889,
      "grad_norm": 4.48143196105957,
      "learning_rate": 3.0770120426906527e-07,
      "loss": 0.0582,
      "step": 17213
    },
    {
      "epoch": 0.8889233152594888,
      "grad_norm": 6.259632110595703,
      "learning_rate": 3.0741826944715447e-07,
      "loss": 0.0557,
      "step": 17214
    },
    {
      "epoch": 0.8889749548153886,
      "grad_norm": 4.0360870361328125,
      "learning_rate": 3.071354606401883e-07,
      "loss": 0.0488,
      "step": 17215
    },
    {
      "epoch": 0.8890265943712884,
      "grad_norm": 20.03270721435547,
      "learning_rate": 3.068527778557617e-07,
      "loss": 0.0933,
      "step": 17216
    },
    {
      "epoch": 0.8890782339271882,
      "grad_norm": 5.90153694152832,
      "learning_rate": 3.065702211014665e-07,
      "loss": 0.0602,
      "step": 17217
    },
    {
      "epoch": 0.889129873483088,
      "grad_norm": 5.889223098754883,
      "learning_rate": 3.0628779038489045e-07,
      "loss": 0.0285,
      "step": 17218
    },
    {
      "epoch": 0.8891815130389878,
      "grad_norm": 3.4050211906433105,
      "learning_rate": 3.0600548571361743e-07,
      "loss": 0.0269,
      "step": 17219
    },
    {
      "epoch": 0.8892331525948877,
      "grad_norm": 5.27522087097168,
      "learning_rate": 3.057233070952287e-07,
      "loss": 0.0685,
      "step": 17220
    },
    {
      "epoch": 0.8892847921507875,
      "grad_norm": 5.2187652587890625,
      "learning_rate": 3.0544125453730255e-07,
      "loss": 0.0512,
      "step": 17221
    },
    {
      "epoch": 0.8893364317066873,
      "grad_norm": 13.595230102539062,
      "learning_rate": 3.051593280474119e-07,
      "loss": 0.1161,
      "step": 17222
    },
    {
      "epoch": 0.8893880712625871,
      "grad_norm": 2.813283681869507,
      "learning_rate": 3.0487752763312837e-07,
      "loss": 0.0219,
      "step": 17223
    },
    {
      "epoch": 0.8894397108184869,
      "grad_norm": 1.3962699174880981,
      "learning_rate": 3.0459585330201993e-07,
      "loss": 0.0145,
      "step": 17224
    },
    {
      "epoch": 0.8894913503743868,
      "grad_norm": 16.63381004333496,
      "learning_rate": 3.043143050616509e-07,
      "loss": 0.0791,
      "step": 17225
    },
    {
      "epoch": 0.8895429899302866,
      "grad_norm": 2.2251620292663574,
      "learning_rate": 3.0403288291958045e-07,
      "loss": 0.0386,
      "step": 17226
    },
    {
      "epoch": 0.8895946294861864,
      "grad_norm": 3.0939247608184814,
      "learning_rate": 3.0375158688336625e-07,
      "loss": 0.0422,
      "step": 17227
    },
    {
      "epoch": 0.8896462690420862,
      "grad_norm": 10.396180152893066,
      "learning_rate": 3.0347041696056456e-07,
      "loss": 0.0707,
      "step": 17228
    },
    {
      "epoch": 0.889697908597986,
      "grad_norm": 3.9097046852111816,
      "learning_rate": 3.0318937315872265e-07,
      "loss": 0.0299,
      "step": 17229
    },
    {
      "epoch": 0.8897495481538858,
      "grad_norm": 21.32467269897461,
      "learning_rate": 3.0290845548539007e-07,
      "loss": 0.2679,
      "step": 17230
    },
    {
      "epoch": 0.8898011877097857,
      "grad_norm": 3.8816909790039062,
      "learning_rate": 3.0262766394810907e-07,
      "loss": 0.0528,
      "step": 17231
    },
    {
      "epoch": 0.8898528272656855,
      "grad_norm": 1.8983441591262817,
      "learning_rate": 3.0234699855442083e-07,
      "loss": 0.0194,
      "step": 17232
    },
    {
      "epoch": 0.8899044668215853,
      "grad_norm": 22.92931365966797,
      "learning_rate": 3.0206645931186273e-07,
      "loss": 0.0882,
      "step": 17233
    },
    {
      "epoch": 0.8899561063774851,
      "grad_norm": 2.758122205734253,
      "learning_rate": 3.017860462279676e-07,
      "loss": 0.048,
      "step": 17234
    },
    {
      "epoch": 0.8900077459333849,
      "grad_norm": 3.937147855758667,
      "learning_rate": 3.01505759310266e-07,
      "loss": 0.0468,
      "step": 17235
    },
    {
      "epoch": 0.8900593854892848,
      "grad_norm": 4.500217914581299,
      "learning_rate": 3.012255985662843e-07,
      "loss": 0.0517,
      "step": 17236
    },
    {
      "epoch": 0.8901110250451846,
      "grad_norm": 2.8508715629577637,
      "learning_rate": 3.009455640035469e-07,
      "loss": 0.0477,
      "step": 17237
    },
    {
      "epoch": 0.8901626646010844,
      "grad_norm": 2.7022666931152344,
      "learning_rate": 3.006656556295734e-07,
      "loss": 0.0365,
      "step": 17238
    },
    {
      "epoch": 0.8902143041569842,
      "grad_norm": 12.820054054260254,
      "learning_rate": 3.003858734518816e-07,
      "loss": 0.1449,
      "step": 17239
    },
    {
      "epoch": 0.890265943712884,
      "grad_norm": 10.138269424438477,
      "learning_rate": 3.001062174779823e-07,
      "loss": 0.0707,
      "step": 17240
    },
    {
      "epoch": 0.8903175832687839,
      "grad_norm": 3.7697532176971436,
      "learning_rate": 2.9982668771538603e-07,
      "loss": 0.044,
      "step": 17241
    },
    {
      "epoch": 0.8903692228246837,
      "grad_norm": 3.654587745666504,
      "learning_rate": 2.995472841716013e-07,
      "loss": 0.0371,
      "step": 17242
    },
    {
      "epoch": 0.8904208623805835,
      "grad_norm": 5.9393157958984375,
      "learning_rate": 2.9926800685413037e-07,
      "loss": 0.0269,
      "step": 17243
    },
    {
      "epoch": 0.8904725019364833,
      "grad_norm": 13.815774917602539,
      "learning_rate": 2.9898885577047174e-07,
      "loss": 0.0828,
      "step": 17244
    },
    {
      "epoch": 0.8905241414923831,
      "grad_norm": 15.036989212036133,
      "learning_rate": 2.9870983092812277e-07,
      "loss": 0.1521,
      "step": 17245
    },
    {
      "epoch": 0.8905757810482829,
      "grad_norm": 3.4324729442596436,
      "learning_rate": 2.984309323345752e-07,
      "loss": 0.0528,
      "step": 17246
    },
    {
      "epoch": 0.8906274206041828,
      "grad_norm": 12.631494522094727,
      "learning_rate": 2.9815215999732137e-07,
      "loss": 0.0931,
      "step": 17247
    },
    {
      "epoch": 0.8906790601600826,
      "grad_norm": 2.3911211490631104,
      "learning_rate": 2.9787351392384533e-07,
      "loss": 0.0262,
      "step": 17248
    },
    {
      "epoch": 0.8907306997159824,
      "grad_norm": 2.7362401485443115,
      "learning_rate": 2.9759499412163e-07,
      "loss": 0.0326,
      "step": 17249
    },
    {
      "epoch": 0.8907823392718822,
      "grad_norm": 2.765998363494873,
      "learning_rate": 2.9731660059815483e-07,
      "loss": 0.0305,
      "step": 17250
    },
    {
      "epoch": 0.890833978827782,
      "grad_norm": 2.3333075046539307,
      "learning_rate": 2.970383333608956e-07,
      "loss": 0.0281,
      "step": 17251
    },
    {
      "epoch": 0.8908856183836819,
      "grad_norm": 2.448598623275757,
      "learning_rate": 2.967601924173258e-07,
      "loss": 0.0299,
      "step": 17252
    },
    {
      "epoch": 0.8909372579395817,
      "grad_norm": 4.215813636779785,
      "learning_rate": 2.9648217777491437e-07,
      "loss": 0.0431,
      "step": 17253
    },
    {
      "epoch": 0.8909888974954815,
      "grad_norm": 2.7656404972076416,
      "learning_rate": 2.962042894411277e-07,
      "loss": 0.039,
      "step": 17254
    },
    {
      "epoch": 0.8910405370513813,
      "grad_norm": 9.992660522460938,
      "learning_rate": 2.9592652742342577e-07,
      "loss": 0.0575,
      "step": 17255
    },
    {
      "epoch": 0.8910921766072811,
      "grad_norm": 7.688076972961426,
      "learning_rate": 2.9564889172926993e-07,
      "loss": 0.0357,
      "step": 17256
    },
    {
      "epoch": 0.891143816163181,
      "grad_norm": 5.0441813468933105,
      "learning_rate": 2.953713823661153e-07,
      "loss": 0.0316,
      "step": 17257
    },
    {
      "epoch": 0.8911954557190808,
      "grad_norm": 6.358426094055176,
      "learning_rate": 2.950939993414148e-07,
      "loss": 0.1377,
      "step": 17258
    },
    {
      "epoch": 0.8912470952749806,
      "grad_norm": 2.197564125061035,
      "learning_rate": 2.9481674266261585e-07,
      "loss": 0.0243,
      "step": 17259
    },
    {
      "epoch": 0.8912987348308804,
      "grad_norm": 3.6838691234588623,
      "learning_rate": 2.945396123371641e-07,
      "loss": 0.0215,
      "step": 17260
    },
    {
      "epoch": 0.8913503743867802,
      "grad_norm": 2.701510190963745,
      "learning_rate": 2.942626083725036e-07,
      "loss": 0.0481,
      "step": 17261
    },
    {
      "epoch": 0.89140201394268,
      "grad_norm": 3.17110013961792,
      "learning_rate": 2.939857307760707e-07,
      "loss": 0.0701,
      "step": 17262
    },
    {
      "epoch": 0.89145365349858,
      "grad_norm": 11.384804725646973,
      "learning_rate": 2.937089795553016e-07,
      "loss": 0.0696,
      "step": 17263
    },
    {
      "epoch": 0.8915052930544798,
      "grad_norm": 11.954052925109863,
      "learning_rate": 2.934323547176282e-07,
      "loss": 0.1357,
      "step": 17264
    },
    {
      "epoch": 0.8915569326103796,
      "grad_norm": 3.095891237258911,
      "learning_rate": 2.931558562704789e-07,
      "loss": 0.0212,
      "step": 17265
    },
    {
      "epoch": 0.8916085721662794,
      "grad_norm": 4.577943801879883,
      "learning_rate": 2.92879484221279e-07,
      "loss": 0.057,
      "step": 17266
    },
    {
      "epoch": 0.8916602117221792,
      "grad_norm": 3.2058777809143066,
      "learning_rate": 2.9260323857744963e-07,
      "loss": 0.0332,
      "step": 17267
    },
    {
      "epoch": 0.8917118512780791,
      "grad_norm": 4.270805358886719,
      "learning_rate": 2.9232711934641054e-07,
      "loss": 0.0734,
      "step": 17268
    },
    {
      "epoch": 0.8917634908339789,
      "grad_norm": 2.7197446823120117,
      "learning_rate": 2.920511265355752e-07,
      "loss": 0.0541,
      "step": 17269
    },
    {
      "epoch": 0.8918151303898787,
      "grad_norm": 3.791541337966919,
      "learning_rate": 2.917752601523555e-07,
      "loss": 0.0187,
      "step": 17270
    },
    {
      "epoch": 0.8918667699457785,
      "grad_norm": 3.6942296028137207,
      "learning_rate": 2.914995202041598e-07,
      "loss": 0.0256,
      "step": 17271
    },
    {
      "epoch": 0.8919184095016783,
      "grad_norm": 2.4385836124420166,
      "learning_rate": 2.912239066983935e-07,
      "loss": 0.0248,
      "step": 17272
    },
    {
      "epoch": 0.8919700490575782,
      "grad_norm": 1.7188001871109009,
      "learning_rate": 2.909484196424567e-07,
      "loss": 0.0151,
      "step": 17273
    },
    {
      "epoch": 0.892021688613478,
      "grad_norm": 3.6359896659851074,
      "learning_rate": 2.9067305904374674e-07,
      "loss": 0.0388,
      "step": 17274
    },
    {
      "epoch": 0.8920733281693778,
      "grad_norm": 5.1253342628479,
      "learning_rate": 2.903978249096606e-07,
      "loss": 0.0561,
      "step": 17275
    },
    {
      "epoch": 0.8921249677252776,
      "grad_norm": 6.484332084655762,
      "learning_rate": 2.9012271724758843e-07,
      "loss": 0.0489,
      "step": 17276
    },
    {
      "epoch": 0.8921766072811774,
      "grad_norm": 11.960672378540039,
      "learning_rate": 2.898477360649171e-07,
      "loss": 0.0569,
      "step": 17277
    },
    {
      "epoch": 0.8922282468370772,
      "grad_norm": 2.7089459896087646,
      "learning_rate": 2.895728813690313e-07,
      "loss": 0.033,
      "step": 17278
    },
    {
      "epoch": 0.8922798863929771,
      "grad_norm": 8.736114501953125,
      "learning_rate": 2.8929815316731167e-07,
      "loss": 0.0539,
      "step": 17279
    },
    {
      "epoch": 0.8923315259488769,
      "grad_norm": 2.6578681468963623,
      "learning_rate": 2.8902355146713803e-07,
      "loss": 0.0553,
      "step": 17280
    },
    {
      "epoch": 0.8923831655047767,
      "grad_norm": 7.587307453155518,
      "learning_rate": 2.887490762758821e-07,
      "loss": 0.0929,
      "step": 17281
    },
    {
      "epoch": 0.8924348050606765,
      "grad_norm": 3.6174190044403076,
      "learning_rate": 2.8847472760091586e-07,
      "loss": 0.0499,
      "step": 17282
    },
    {
      "epoch": 0.8924864446165763,
      "grad_norm": 2.72062611579895,
      "learning_rate": 2.8820050544960565e-07,
      "loss": 0.033,
      "step": 17283
    },
    {
      "epoch": 0.8925380841724762,
      "grad_norm": 8.045792579650879,
      "learning_rate": 2.879264098293166e-07,
      "loss": 0.0501,
      "step": 17284
    },
    {
      "epoch": 0.892589723728376,
      "grad_norm": 3.8175182342529297,
      "learning_rate": 2.876524407474085e-07,
      "loss": 0.0312,
      "step": 17285
    },
    {
      "epoch": 0.8926413632842758,
      "grad_norm": 2.848471164703369,
      "learning_rate": 2.873785982112387e-07,
      "loss": 0.024,
      "step": 17286
    },
    {
      "epoch": 0.8926930028401756,
      "grad_norm": 5.1997761726379395,
      "learning_rate": 2.8710488222816246e-07,
      "loss": 0.0717,
      "step": 17287
    },
    {
      "epoch": 0.8927446423960754,
      "grad_norm": 5.99302864074707,
      "learning_rate": 2.868312928055267e-07,
      "loss": 0.0438,
      "step": 17288
    },
    {
      "epoch": 0.8927962819519752,
      "grad_norm": 3.3184144496917725,
      "learning_rate": 2.865578299506816e-07,
      "loss": 0.0262,
      "step": 17289
    },
    {
      "epoch": 0.8928479215078751,
      "grad_norm": 1.7669755220413208,
      "learning_rate": 2.8628449367096967e-07,
      "loss": 0.0316,
      "step": 17290
    },
    {
      "epoch": 0.8928995610637749,
      "grad_norm": 2.810365915298462,
      "learning_rate": 2.8601128397373166e-07,
      "loss": 0.0413,
      "step": 17291
    },
    {
      "epoch": 0.8929512006196747,
      "grad_norm": 4.150808811187744,
      "learning_rate": 2.857382008663034e-07,
      "loss": 0.0464,
      "step": 17292
    },
    {
      "epoch": 0.8930028401755745,
      "grad_norm": 3.3974714279174805,
      "learning_rate": 2.85465244356018e-07,
      "loss": 0.0747,
      "step": 17293
    },
    {
      "epoch": 0.8930544797314743,
      "grad_norm": 2.7240421772003174,
      "learning_rate": 2.8519241445020717e-07,
      "loss": 0.0213,
      "step": 17294
    },
    {
      "epoch": 0.8931061192873742,
      "grad_norm": 3.053260087966919,
      "learning_rate": 2.849197111561963e-07,
      "loss": 0.0329,
      "step": 17295
    },
    {
      "epoch": 0.893157758843274,
      "grad_norm": 2.96083927154541,
      "learning_rate": 2.8464713448130845e-07,
      "loss": 0.0248,
      "step": 17296
    },
    {
      "epoch": 0.8932093983991738,
      "grad_norm": 10.197761535644531,
      "learning_rate": 2.8437468443286374e-07,
      "loss": 0.0834,
      "step": 17297
    },
    {
      "epoch": 0.8932610379550736,
      "grad_norm": 5.6182966232299805,
      "learning_rate": 2.8410236101817865e-07,
      "loss": 0.0505,
      "step": 17298
    },
    {
      "epoch": 0.8933126775109734,
      "grad_norm": 0.8688061237335205,
      "learning_rate": 2.8383016424456624e-07,
      "loss": 0.0139,
      "step": 17299
    },
    {
      "epoch": 0.8933643170668732,
      "grad_norm": 2.9499475955963135,
      "learning_rate": 2.835580941193361e-07,
      "loss": 0.0469,
      "step": 17300
    },
    {
      "epoch": 0.8934159566227731,
      "grad_norm": 2.4856948852539062,
      "learning_rate": 2.8328615064979413e-07,
      "loss": 0.0348,
      "step": 17301
    },
    {
      "epoch": 0.8934675961786729,
      "grad_norm": 9.01941967010498,
      "learning_rate": 2.830143338432434e-07,
      "loss": 0.0624,
      "step": 17302
    },
    {
      "epoch": 0.8935192357345727,
      "grad_norm": 1.8399969339370728,
      "learning_rate": 2.82742643706983e-07,
      "loss": 0.0343,
      "step": 17303
    },
    {
      "epoch": 0.8935708752904725,
      "grad_norm": 3.71982479095459,
      "learning_rate": 2.8247108024830883e-07,
      "loss": 0.0145,
      "step": 17304
    },
    {
      "epoch": 0.8936225148463723,
      "grad_norm": 3.842324733734131,
      "learning_rate": 2.82199643474515e-07,
      "loss": 0.0341,
      "step": 17305
    },
    {
      "epoch": 0.8936741544022722,
      "grad_norm": 5.808692455291748,
      "learning_rate": 2.8192833339288795e-07,
      "loss": 0.0472,
      "step": 17306
    },
    {
      "epoch": 0.893725793958172,
      "grad_norm": 4.954455852508545,
      "learning_rate": 2.8165715001071516e-07,
      "loss": 0.0391,
      "step": 17307
    },
    {
      "epoch": 0.8937774335140718,
      "grad_norm": 3.24957275390625,
      "learning_rate": 2.813860933352791e-07,
      "loss": 0.0481,
      "step": 17308
    },
    {
      "epoch": 0.8938290730699716,
      "grad_norm": 2.0433456897735596,
      "learning_rate": 2.81115163373859e-07,
      "loss": 0.0419,
      "step": 17309
    },
    {
      "epoch": 0.8938807126258714,
      "grad_norm": 3.439263105392456,
      "learning_rate": 2.808443601337291e-07,
      "loss": 0.0425,
      "step": 17310
    },
    {
      "epoch": 0.8939323521817713,
      "grad_norm": 3.2538881301879883,
      "learning_rate": 2.8057368362216285e-07,
      "loss": 0.0167,
      "step": 17311
    },
    {
      "epoch": 0.8939839917376711,
      "grad_norm": 1.9195491075515747,
      "learning_rate": 2.803031338464274e-07,
      "loss": 0.0316,
      "step": 17312
    },
    {
      "epoch": 0.8940356312935709,
      "grad_norm": 15.848185539245605,
      "learning_rate": 2.800327108137907e-07,
      "loss": 0.0621,
      "step": 17313
    },
    {
      "epoch": 0.8940872708494707,
      "grad_norm": 2.091538190841675,
      "learning_rate": 2.797624145315125e-07,
      "loss": 0.0174,
      "step": 17314
    },
    {
      "epoch": 0.8941389104053705,
      "grad_norm": 5.220889091491699,
      "learning_rate": 2.7949224500685214e-07,
      "loss": 0.0622,
      "step": 17315
    },
    {
      "epoch": 0.8941905499612703,
      "grad_norm": 2.4487648010253906,
      "learning_rate": 2.7922220224706477e-07,
      "loss": 0.0251,
      "step": 17316
    },
    {
      "epoch": 0.8942421895171702,
      "grad_norm": 8.160514831542969,
      "learning_rate": 2.7895228625940186e-07,
      "loss": 0.0791,
      "step": 17317
    },
    {
      "epoch": 0.89429382907307,
      "grad_norm": 5.018496990203857,
      "learning_rate": 2.7868249705111204e-07,
      "loss": 0.1248,
      "step": 17318
    },
    {
      "epoch": 0.8943454686289698,
      "grad_norm": 3.549661636352539,
      "learning_rate": 2.784128346294407e-07,
      "loss": 0.0379,
      "step": 17319
    },
    {
      "epoch": 0.8943971081848696,
      "grad_norm": 3.037140130996704,
      "learning_rate": 2.781432990016292e-07,
      "loss": 0.0444,
      "step": 17320
    },
    {
      "epoch": 0.8944487477407694,
      "grad_norm": 1.261414885520935,
      "learning_rate": 2.7787389017491396e-07,
      "loss": 0.0091,
      "step": 17321
    },
    {
      "epoch": 0.8945003872966693,
      "grad_norm": 10.233298301696777,
      "learning_rate": 2.7760460815653146e-07,
      "loss": 0.0759,
      "step": 17322
    },
    {
      "epoch": 0.8945520268525691,
      "grad_norm": 4.914595603942871,
      "learning_rate": 2.7733545295371366e-07,
      "loss": 0.0402,
      "step": 17323
    },
    {
      "epoch": 0.8946036664084689,
      "grad_norm": 12.150134086608887,
      "learning_rate": 2.7706642457368703e-07,
      "loss": 0.0736,
      "step": 17324
    },
    {
      "epoch": 0.8946553059643687,
      "grad_norm": 3.316019296646118,
      "learning_rate": 2.767975230236764e-07,
      "loss": 0.0178,
      "step": 17325
    },
    {
      "epoch": 0.8947069455202685,
      "grad_norm": 12.714073181152344,
      "learning_rate": 2.765287483109019e-07,
      "loss": 0.0691,
      "step": 17326
    },
    {
      "epoch": 0.8947585850761683,
      "grad_norm": 5.705761909484863,
      "learning_rate": 2.762601004425841e-07,
      "loss": 0.0721,
      "step": 17327
    },
    {
      "epoch": 0.8948102246320682,
      "grad_norm": 6.261678695678711,
      "learning_rate": 2.7599157942593436e-07,
      "loss": 0.0702,
      "step": 17328
    },
    {
      "epoch": 0.894861864187968,
      "grad_norm": 8.191514015197754,
      "learning_rate": 2.7572318526816523e-07,
      "loss": 0.1023,
      "step": 17329
    },
    {
      "epoch": 0.8949135037438678,
      "grad_norm": 2.2386257648468018,
      "learning_rate": 2.754549179764837e-07,
      "loss": 0.0317,
      "step": 17330
    },
    {
      "epoch": 0.8949651432997676,
      "grad_norm": 3.8195343017578125,
      "learning_rate": 2.751867775580935e-07,
      "loss": 0.0358,
      "step": 17331
    },
    {
      "epoch": 0.8950167828556674,
      "grad_norm": 7.147584438323975,
      "learning_rate": 2.749187640201956e-07,
      "loss": 0.1178,
      "step": 17332
    },
    {
      "epoch": 0.8950684224115673,
      "grad_norm": 3.0856127738952637,
      "learning_rate": 2.746508773699874e-07,
      "loss": 0.0547,
      "step": 17333
    },
    {
      "epoch": 0.8951200619674671,
      "grad_norm": 2.473684072494507,
      "learning_rate": 2.743831176146622e-07,
      "loss": 0.0306,
      "step": 17334
    },
    {
      "epoch": 0.8951717015233669,
      "grad_norm": 12.666919708251953,
      "learning_rate": 2.7411548476141136e-07,
      "loss": 0.0821,
      "step": 17335
    },
    {
      "epoch": 0.8952233410792667,
      "grad_norm": 3.747267007827759,
      "learning_rate": 2.738479788174209e-07,
      "loss": 0.0404,
      "step": 17336
    },
    {
      "epoch": 0.8952749806351665,
      "grad_norm": 2.2188706398010254,
      "learning_rate": 2.735805997898755e-07,
      "loss": 0.0543,
      "step": 17337
    },
    {
      "epoch": 0.8953266201910663,
      "grad_norm": 7.1790547370910645,
      "learning_rate": 2.733133476859551e-07,
      "loss": 0.0825,
      "step": 17338
    },
    {
      "epoch": 0.8953782597469662,
      "grad_norm": 6.513716220855713,
      "learning_rate": 2.7304622251283555e-07,
      "loss": 0.0445,
      "step": 17339
    },
    {
      "epoch": 0.895429899302866,
      "grad_norm": 2.144138813018799,
      "learning_rate": 2.7277922427769057e-07,
      "loss": 0.0228,
      "step": 17340
    },
    {
      "epoch": 0.8954815388587658,
      "grad_norm": 1.6536754369735718,
      "learning_rate": 2.725123529876905e-07,
      "loss": 0.015,
      "step": 17341
    },
    {
      "epoch": 0.8955331784146656,
      "grad_norm": 8.835492134094238,
      "learning_rate": 2.722456086500036e-07,
      "loss": 0.0714,
      "step": 17342
    },
    {
      "epoch": 0.8955848179705654,
      "grad_norm": 1.610633373260498,
      "learning_rate": 2.719789912717896e-07,
      "loss": 0.0313,
      "step": 17343
    },
    {
      "epoch": 0.8956364575264653,
      "grad_norm": 3.4862475395202637,
      "learning_rate": 2.717125008602106e-07,
      "loss": 0.0534,
      "step": 17344
    },
    {
      "epoch": 0.8956880970823651,
      "grad_norm": 3.3419878482818604,
      "learning_rate": 2.7144613742242256e-07,
      "loss": 0.0556,
      "step": 17345
    },
    {
      "epoch": 0.8957397366382649,
      "grad_norm": 2.740543842315674,
      "learning_rate": 2.71179900965578e-07,
      "loss": 0.0294,
      "step": 17346
    },
    {
      "epoch": 0.8957913761941647,
      "grad_norm": 7.679961681365967,
      "learning_rate": 2.7091379149682683e-07,
      "loss": 0.0723,
      "step": 17347
    },
    {
      "epoch": 0.8958430157500645,
      "grad_norm": 1.9500865936279297,
      "learning_rate": 2.70647809023315e-07,
      "loss": 0.025,
      "step": 17348
    },
    {
      "epoch": 0.8958946553059643,
      "grad_norm": 3.1752309799194336,
      "learning_rate": 2.703819535521851e-07,
      "loss": 0.0354,
      "step": 17349
    },
    {
      "epoch": 0.8959462948618642,
      "grad_norm": 11.328104972839355,
      "learning_rate": 2.70116225090577e-07,
      "loss": 0.0543,
      "step": 17350
    },
    {
      "epoch": 0.895997934417764,
      "grad_norm": 9.739296913146973,
      "learning_rate": 2.6985062364562607e-07,
      "loss": 0.0337,
      "step": 17351
    },
    {
      "epoch": 0.8960495739736638,
      "grad_norm": 17.313074111938477,
      "learning_rate": 2.695851492244644e-07,
      "loss": 0.1023,
      "step": 17352
    },
    {
      "epoch": 0.8961012135295636,
      "grad_norm": 20.20875358581543,
      "learning_rate": 2.69319801834223e-07,
      "loss": 0.1448,
      "step": 17353
    },
    {
      "epoch": 0.8961528530854634,
      "grad_norm": 2.440216064453125,
      "learning_rate": 2.6905458148202447e-07,
      "loss": 0.0232,
      "step": 17354
    },
    {
      "epoch": 0.8962044926413633,
      "grad_norm": 2.005958318710327,
      "learning_rate": 2.6878948817499305e-07,
      "loss": 0.0247,
      "step": 17355
    },
    {
      "epoch": 0.8962561321972631,
      "grad_norm": 6.0630621910095215,
      "learning_rate": 2.6852452192024816e-07,
      "loss": 0.0702,
      "step": 17356
    },
    {
      "epoch": 0.8963077717531629,
      "grad_norm": 3.0442023277282715,
      "learning_rate": 2.6825968272490345e-07,
      "loss": 0.0301,
      "step": 17357
    },
    {
      "epoch": 0.8963594113090627,
      "grad_norm": 2.6774492263793945,
      "learning_rate": 2.6799497059607216e-07,
      "loss": 0.0477,
      "step": 17358
    },
    {
      "epoch": 0.8964110508649625,
      "grad_norm": 1.430830478668213,
      "learning_rate": 2.677303855408614e-07,
      "loss": 0.021,
      "step": 17359
    },
    {
      "epoch": 0.8964626904208624,
      "grad_norm": 2.634246349334717,
      "learning_rate": 2.674659275663793e-07,
      "loss": 0.0348,
      "step": 17360
    },
    {
      "epoch": 0.8965143299767622,
      "grad_norm": 10.451275825500488,
      "learning_rate": 2.672015966797248e-07,
      "loss": 0.1254,
      "step": 17361
    },
    {
      "epoch": 0.896565969532662,
      "grad_norm": 5.74501895904541,
      "learning_rate": 2.6693739288799704e-07,
      "loss": 0.0418,
      "step": 17362
    },
    {
      "epoch": 0.8966176090885618,
      "grad_norm": 3.975740909576416,
      "learning_rate": 2.6667331619829153e-07,
      "loss": 0.0333,
      "step": 17363
    },
    {
      "epoch": 0.8966692486444616,
      "grad_norm": 2.058753728866577,
      "learning_rate": 2.6640936661769924e-07,
      "loss": 0.0168,
      "step": 17364
    },
    {
      "epoch": 0.8967208882003614,
      "grad_norm": 3.8041186332702637,
      "learning_rate": 2.661455441533084e-07,
      "loss": 0.0399,
      "step": 17365
    },
    {
      "epoch": 0.8967725277562613,
      "grad_norm": 2.9819767475128174,
      "learning_rate": 2.6588184881220393e-07,
      "loss": 0.0504,
      "step": 17366
    },
    {
      "epoch": 0.8968241673121611,
      "grad_norm": 4.575070858001709,
      "learning_rate": 2.656182806014673e-07,
      "loss": 0.0551,
      "step": 17367
    },
    {
      "epoch": 0.8968758068680609,
      "grad_norm": 3.3646116256713867,
      "learning_rate": 2.6535483952817577e-07,
      "loss": 0.0257,
      "step": 17368
    },
    {
      "epoch": 0.8969274464239607,
      "grad_norm": 1.7987685203552246,
      "learning_rate": 2.650915255994041e-07,
      "loss": 0.0223,
      "step": 17369
    },
    {
      "epoch": 0.8969790859798605,
      "grad_norm": 4.715688705444336,
      "learning_rate": 2.648283388222234e-07,
      "loss": 0.0335,
      "step": 17370
    },
    {
      "epoch": 0.8970307255357604,
      "grad_norm": 2.222062110900879,
      "learning_rate": 2.6456527920370234e-07,
      "loss": 0.0227,
      "step": 17371
    },
    {
      "epoch": 0.8970823650916602,
      "grad_norm": 10.517180442810059,
      "learning_rate": 2.6430234675090317e-07,
      "loss": 0.0864,
      "step": 17372
    },
    {
      "epoch": 0.89713400464756,
      "grad_norm": 11.508443832397461,
      "learning_rate": 2.6403954147088685e-07,
      "loss": 0.0662,
      "step": 17373
    },
    {
      "epoch": 0.8971856442034598,
      "grad_norm": 3.2647488117218018,
      "learning_rate": 2.637768633707122e-07,
      "loss": 0.0678,
      "step": 17374
    },
    {
      "epoch": 0.8972372837593596,
      "grad_norm": 2.958296298980713,
      "learning_rate": 2.635143124574335e-07,
      "loss": 0.0394,
      "step": 17375
    },
    {
      "epoch": 0.8972889233152594,
      "grad_norm": 13.467794418334961,
      "learning_rate": 2.632518887380997e-07,
      "loss": 0.1114,
      "step": 17376
    },
    {
      "epoch": 0.8973405628711593,
      "grad_norm": 13.50437068939209,
      "learning_rate": 2.629895922197584e-07,
      "loss": 0.0475,
      "step": 17377
    },
    {
      "epoch": 0.8973922024270591,
      "grad_norm": 4.848442077636719,
      "learning_rate": 2.627274229094534e-07,
      "loss": 0.0579,
      "step": 17378
    },
    {
      "epoch": 0.8974438419829589,
      "grad_norm": 4.00551176071167,
      "learning_rate": 2.624653808142258e-07,
      "loss": 0.0465,
      "step": 17379
    },
    {
      "epoch": 0.8974954815388587,
      "grad_norm": 1.503516674041748,
      "learning_rate": 2.62203465941111e-07,
      "loss": 0.018,
      "step": 17380
    },
    {
      "epoch": 0.8975471210947585,
      "grad_norm": 3.60998272895813,
      "learning_rate": 2.61941678297144e-07,
      "loss": 0.0522,
      "step": 17381
    },
    {
      "epoch": 0.8975987606506584,
      "grad_norm": 6.273326873779297,
      "learning_rate": 2.616800178893536e-07,
      "loss": 0.0319,
      "step": 17382
    },
    {
      "epoch": 0.8976504002065582,
      "grad_norm": 10.920385360717773,
      "learning_rate": 2.614184847247675e-07,
      "loss": 0.097,
      "step": 17383
    },
    {
      "epoch": 0.897702039762458,
      "grad_norm": 2.5959248542785645,
      "learning_rate": 2.6115707881040844e-07,
      "loss": 0.0299,
      "step": 17384
    },
    {
      "epoch": 0.8977536793183578,
      "grad_norm": 4.084267616271973,
      "learning_rate": 2.6089580015329573e-07,
      "loss": 0.0566,
      "step": 17385
    },
    {
      "epoch": 0.8978053188742576,
      "grad_norm": 3.7442877292633057,
      "learning_rate": 2.6063464876044717e-07,
      "loss": 0.0269,
      "step": 17386
    },
    {
      "epoch": 0.8978569584301574,
      "grad_norm": 2.5185301303863525,
      "learning_rate": 2.603736246388744e-07,
      "loss": 0.0372,
      "step": 17387
    },
    {
      "epoch": 0.8979085979860573,
      "grad_norm": 5.494979381561279,
      "learning_rate": 2.6011272779558725e-07,
      "loss": 0.0652,
      "step": 17388
    },
    {
      "epoch": 0.8979602375419571,
      "grad_norm": 3.9145493507385254,
      "learning_rate": 2.59851958237593e-07,
      "loss": 0.0526,
      "step": 17389
    },
    {
      "epoch": 0.8980118770978569,
      "grad_norm": 4.425076007843018,
      "learning_rate": 2.595913159718927e-07,
      "loss": 0.0445,
      "step": 17390
    },
    {
      "epoch": 0.8980635166537568,
      "grad_norm": 2.4732420444488525,
      "learning_rate": 2.593308010054868e-07,
      "loss": 0.0356,
      "step": 17391
    },
    {
      "epoch": 0.8981151562096567,
      "grad_norm": 7.6303019523620605,
      "learning_rate": 2.5907041334536976e-07,
      "loss": 0.0374,
      "step": 17392
    },
    {
      "epoch": 0.8981667957655565,
      "grad_norm": 3.1246674060821533,
      "learning_rate": 2.5881015299853654e-07,
      "loss": 0.04,
      "step": 17393
    },
    {
      "epoch": 0.8982184353214563,
      "grad_norm": 13.314600944519043,
      "learning_rate": 2.585500199719737e-07,
      "loss": 0.0785,
      "step": 17394
    },
    {
      "epoch": 0.8982700748773561,
      "grad_norm": 3.6517350673675537,
      "learning_rate": 2.5829001427266797e-07,
      "loss": 0.0381,
      "step": 17395
    },
    {
      "epoch": 0.8983217144332559,
      "grad_norm": 2.606879949569702,
      "learning_rate": 2.5803013590760205e-07,
      "loss": 0.0422,
      "step": 17396
    },
    {
      "epoch": 0.8983733539891557,
      "grad_norm": 22.86966896057129,
      "learning_rate": 2.577703848837537e-07,
      "loss": 0.0564,
      "step": 17397
    },
    {
      "epoch": 0.8984249935450556,
      "grad_norm": 4.589183807373047,
      "learning_rate": 2.5751076120809904e-07,
      "loss": 0.0517,
      "step": 17398
    },
    {
      "epoch": 0.8984766331009554,
      "grad_norm": 2.680262804031372,
      "learning_rate": 2.5725126488760967e-07,
      "loss": 0.03,
      "step": 17399
    },
    {
      "epoch": 0.8985282726568552,
      "grad_norm": 4.601558208465576,
      "learning_rate": 2.5699189592925387e-07,
      "loss": 0.0684,
      "step": 17400
    },
    {
      "epoch": 0.898579912212755,
      "grad_norm": 4.564304828643799,
      "learning_rate": 2.5673265433999727e-07,
      "loss": 0.0787,
      "step": 17401
    },
    {
      "epoch": 0.8986315517686548,
      "grad_norm": 3.400197982788086,
      "learning_rate": 2.5647354012680146e-07,
      "loss": 0.0627,
      "step": 17402
    },
    {
      "epoch": 0.8986831913245547,
      "grad_norm": 2.765951633453369,
      "learning_rate": 2.5621455329662484e-07,
      "loss": 0.0268,
      "step": 17403
    },
    {
      "epoch": 0.8987348308804545,
      "grad_norm": 2.83976674079895,
      "learning_rate": 2.559556938564228e-07,
      "loss": 0.0521,
      "step": 17404
    },
    {
      "epoch": 0.8987864704363543,
      "grad_norm": 8.630792617797852,
      "learning_rate": 2.556969618131444e-07,
      "loss": 0.2635,
      "step": 17405
    },
    {
      "epoch": 0.8988381099922541,
      "grad_norm": 1.9213794469833374,
      "learning_rate": 2.554383571737401e-07,
      "loss": 0.0197,
      "step": 17406
    },
    {
      "epoch": 0.8988897495481539,
      "grad_norm": 3.0987296104431152,
      "learning_rate": 2.5517987994515327e-07,
      "loss": 0.047,
      "step": 17407
    },
    {
      "epoch": 0.8989413891040537,
      "grad_norm": 3.9138541221618652,
      "learning_rate": 2.549215301343266e-07,
      "loss": 0.0342,
      "step": 17408
    },
    {
      "epoch": 0.8989930286599536,
      "grad_norm": 9.20527458190918,
      "learning_rate": 2.546633077481958e-07,
      "loss": 0.0962,
      "step": 17409
    },
    {
      "epoch": 0.8990446682158534,
      "grad_norm": 10.848252296447754,
      "learning_rate": 2.5440521279369635e-07,
      "loss": 0.057,
      "step": 17410
    },
    {
      "epoch": 0.8990963077717532,
      "grad_norm": 10.095298767089844,
      "learning_rate": 2.5414724527775883e-07,
      "loss": 0.1012,
      "step": 17411
    },
    {
      "epoch": 0.899147947327653,
      "grad_norm": 2.3953311443328857,
      "learning_rate": 2.5388940520731096e-07,
      "loss": 0.0317,
      "step": 17412
    },
    {
      "epoch": 0.8991995868835528,
      "grad_norm": 2.112434148788452,
      "learning_rate": 2.5363169258927624e-07,
      "loss": 0.0245,
      "step": 17413
    },
    {
      "epoch": 0.8992512264394527,
      "grad_norm": 2.2945668697357178,
      "learning_rate": 2.533741074305762e-07,
      "loss": 0.048,
      "step": 17414
    },
    {
      "epoch": 0.8993028659953525,
      "grad_norm": 3.3160111904144287,
      "learning_rate": 2.531166497381271e-07,
      "loss": 0.053,
      "step": 17415
    },
    {
      "epoch": 0.8993545055512523,
      "grad_norm": 3.0147242546081543,
      "learning_rate": 2.5285931951884333e-07,
      "loss": 0.0491,
      "step": 17416
    },
    {
      "epoch": 0.8994061451071521,
      "grad_norm": 4.53860330581665,
      "learning_rate": 2.52602116779635e-07,
      "loss": 0.0525,
      "step": 17417
    },
    {
      "epoch": 0.8994577846630519,
      "grad_norm": 5.128458499908447,
      "learning_rate": 2.5234504152740926e-07,
      "loss": 0.0545,
      "step": 17418
    },
    {
      "epoch": 0.8995094242189517,
      "grad_norm": 7.727163791656494,
      "learning_rate": 2.520880937690695e-07,
      "loss": 0.1144,
      "step": 17419
    },
    {
      "epoch": 0.8995610637748516,
      "grad_norm": 2.2968802452087402,
      "learning_rate": 2.518312735115158e-07,
      "loss": 0.0293,
      "step": 17420
    },
    {
      "epoch": 0.8996127033307514,
      "grad_norm": 1.9517637491226196,
      "learning_rate": 2.515745807616454e-07,
      "loss": 0.0186,
      "step": 17421
    },
    {
      "epoch": 0.8996643428866512,
      "grad_norm": 2.9435925483703613,
      "learning_rate": 2.5131801552635115e-07,
      "loss": 0.0413,
      "step": 17422
    },
    {
      "epoch": 0.899715982442551,
      "grad_norm": 1.7760424613952637,
      "learning_rate": 2.5106157781252194e-07,
      "loss": 0.0108,
      "step": 17423
    },
    {
      "epoch": 0.8997676219984508,
      "grad_norm": 5.23618745803833,
      "learning_rate": 2.508052676270445e-07,
      "loss": 0.0966,
      "step": 17424
    },
    {
      "epoch": 0.8998192615543507,
      "grad_norm": 3.807543992996216,
      "learning_rate": 2.505490849768033e-07,
      "loss": 0.034,
      "step": 17425
    },
    {
      "epoch": 0.8998709011102505,
      "grad_norm": 3.901048183441162,
      "learning_rate": 2.502930298686773e-07,
      "loss": 0.0468,
      "step": 17426
    },
    {
      "epoch": 0.8999225406661503,
      "grad_norm": 4.760001182556152,
      "learning_rate": 2.5003710230954105e-07,
      "loss": 0.0654,
      "step": 17427
    },
    {
      "epoch": 0.8999741802220501,
      "grad_norm": 1.6050928831100464,
      "learning_rate": 2.4978130230626897e-07,
      "loss": 0.0259,
      "step": 17428
    },
    {
      "epoch": 0.9000258197779499,
      "grad_norm": 6.579190254211426,
      "learning_rate": 2.4952562986572946e-07,
      "loss": 0.0391,
      "step": 17429
    },
    {
      "epoch": 0.9000774593338497,
      "grad_norm": 5.194089889526367,
      "learning_rate": 2.492700849947888e-07,
      "loss": 0.0513,
      "step": 17430
    },
    {
      "epoch": 0.9001290988897496,
      "grad_norm": 13.83438777923584,
      "learning_rate": 2.490146677003091e-07,
      "loss": 0.0578,
      "step": 17431
    },
    {
      "epoch": 0.9001807384456494,
      "grad_norm": 3.390921115875244,
      "learning_rate": 2.487593779891495e-07,
      "loss": 0.0572,
      "step": 17432
    },
    {
      "epoch": 0.9002323780015492,
      "grad_norm": 6.672588348388672,
      "learning_rate": 2.4850421586816554e-07,
      "loss": 0.1002,
      "step": 17433
    },
    {
      "epoch": 0.9002323780015492,
      "eval_loss": 0.053167372941970825,
      "eval_runtime": 34.1863,
      "eval_samples_per_second": 45.779,
      "eval_steps_per_second": 5.733,
      "step": 17433
    },
    {
      "epoch": 0.900284017557449,
      "grad_norm": 20.34788703918457,
      "learning_rate": 2.482491813442095e-07,
      "loss": 0.2069,
      "step": 17434
    },
    {
      "epoch": 0.9003356571133488,
      "grad_norm": 23.334327697753906,
      "learning_rate": 2.479942744241304e-07,
      "loss": 0.0406,
      "step": 17435
    },
    {
      "epoch": 0.9003872966692487,
      "grad_norm": 6.882468223571777,
      "learning_rate": 2.477394951147727e-07,
      "loss": 0.0581,
      "step": 17436
    },
    {
      "epoch": 0.9004389362251485,
      "grad_norm": 2.1471498012542725,
      "learning_rate": 2.474848434229793e-07,
      "loss": 0.0401,
      "step": 17437
    },
    {
      "epoch": 0.9004905757810483,
      "grad_norm": 3.392108201980591,
      "learning_rate": 2.4723031935558694e-07,
      "loss": 0.0497,
      "step": 17438
    },
    {
      "epoch": 0.9005422153369481,
      "grad_norm": 2.7749571800231934,
      "learning_rate": 2.4697592291943185e-07,
      "loss": 0.0359,
      "step": 17439
    },
    {
      "epoch": 0.9005938548928479,
      "grad_norm": 2.3931524753570557,
      "learning_rate": 2.467216541213463e-07,
      "loss": 0.0278,
      "step": 17440
    },
    {
      "epoch": 0.9006454944487478,
      "grad_norm": 3.710867404937744,
      "learning_rate": 2.4646751296815764e-07,
      "loss": 0.0578,
      "step": 17441
    },
    {
      "epoch": 0.9006971340046476,
      "grad_norm": 24.986604690551758,
      "learning_rate": 2.462134994666904e-07,
      "loss": 0.0517,
      "step": 17442
    },
    {
      "epoch": 0.9007487735605474,
      "grad_norm": 1.9406594038009644,
      "learning_rate": 2.459596136237652e-07,
      "loss": 0.0237,
      "step": 17443
    },
    {
      "epoch": 0.9008004131164472,
      "grad_norm": 4.606014251708984,
      "learning_rate": 2.457058554462022e-07,
      "loss": 0.0533,
      "step": 17444
    },
    {
      "epoch": 0.900852052672347,
      "grad_norm": 2.544631004333496,
      "learning_rate": 2.4545222494081367e-07,
      "loss": 0.0351,
      "step": 17445
    },
    {
      "epoch": 0.9009036922282468,
      "grad_norm": 10.476608276367188,
      "learning_rate": 2.451987221144109e-07,
      "loss": 0.1006,
      "step": 17446
    },
    {
      "epoch": 0.9009553317841467,
      "grad_norm": 5.222413063049316,
      "learning_rate": 2.449453469738028e-07,
      "loss": 0.051,
      "step": 17447
    },
    {
      "epoch": 0.9010069713400465,
      "grad_norm": 2.8391573429107666,
      "learning_rate": 2.446920995257918e-07,
      "loss": 0.0214,
      "step": 17448
    },
    {
      "epoch": 0.9010586108959463,
      "grad_norm": 2.149693012237549,
      "learning_rate": 2.4443897977718014e-07,
      "loss": 0.0268,
      "step": 17449
    },
    {
      "epoch": 0.9011102504518461,
      "grad_norm": 1.7591617107391357,
      "learning_rate": 2.441859877347641e-07,
      "loss": 0.0181,
      "step": 17450
    },
    {
      "epoch": 0.9011618900077459,
      "grad_norm": 4.742827415466309,
      "learning_rate": 2.4393312340533823e-07,
      "loss": 0.0621,
      "step": 17451
    },
    {
      "epoch": 0.9012135295636458,
      "grad_norm": 3.9353978633880615,
      "learning_rate": 2.4368038679569274e-07,
      "loss": 0.0523,
      "step": 17452
    },
    {
      "epoch": 0.9012651691195456,
      "grad_norm": 3.1586101055145264,
      "learning_rate": 2.434277779126143e-07,
      "loss": 0.0421,
      "step": 17453
    },
    {
      "epoch": 0.9013168086754454,
      "grad_norm": 3.947988748550415,
      "learning_rate": 2.4317529676288644e-07,
      "loss": 0.0376,
      "step": 17454
    },
    {
      "epoch": 0.9013684482313452,
      "grad_norm": 3.0774033069610596,
      "learning_rate": 2.4292294335329034e-07,
      "loss": 0.0627,
      "step": 17455
    },
    {
      "epoch": 0.901420087787245,
      "grad_norm": 1.6401110887527466,
      "learning_rate": 2.426707176906018e-07,
      "loss": 0.0283,
      "step": 17456
    },
    {
      "epoch": 0.9014717273431448,
      "grad_norm": 2.893130302429199,
      "learning_rate": 2.4241861978159366e-07,
      "loss": 0.0444,
      "step": 17457
    },
    {
      "epoch": 0.9015233668990447,
      "grad_norm": 2.7271840572357178,
      "learning_rate": 2.421666496330366e-07,
      "loss": 0.0296,
      "step": 17458
    },
    {
      "epoch": 0.9015750064549445,
      "grad_norm": 13.736299514770508,
      "learning_rate": 2.419148072516975e-07,
      "loss": 0.0621,
      "step": 17459
    },
    {
      "epoch": 0.9016266460108443,
      "grad_norm": 5.523118019104004,
      "learning_rate": 2.4166309264433865e-07,
      "loss": 0.0767,
      "step": 17460
    },
    {
      "epoch": 0.9016782855667441,
      "grad_norm": 4.033493518829346,
      "learning_rate": 2.414115058177191e-07,
      "loss": 0.059,
      "step": 17461
    },
    {
      "epoch": 0.9017299251226439,
      "grad_norm": 5.555064678192139,
      "learning_rate": 2.4116004677859573e-07,
      "loss": 0.054,
      "step": 17462
    },
    {
      "epoch": 0.9017815646785438,
      "grad_norm": 6.595864772796631,
      "learning_rate": 2.4090871553372086e-07,
      "loss": 0.0314,
      "step": 17463
    },
    {
      "epoch": 0.9018332042344436,
      "grad_norm": 5.616371154785156,
      "learning_rate": 2.406575120898441e-07,
      "loss": 0.0552,
      "step": 17464
    },
    {
      "epoch": 0.9018848437903434,
      "grad_norm": 3.400820732116699,
      "learning_rate": 2.404064364537112e-07,
      "loss": 0.0458,
      "step": 17465
    },
    {
      "epoch": 0.9019364833462432,
      "grad_norm": 4.135263919830322,
      "learning_rate": 2.401554886320645e-07,
      "loss": 0.081,
      "step": 17466
    },
    {
      "epoch": 0.901988122902143,
      "grad_norm": 7.7817206382751465,
      "learning_rate": 2.3990466863164307e-07,
      "loss": 0.1052,
      "step": 17467
    },
    {
      "epoch": 0.9020397624580428,
      "grad_norm": 5.359416961669922,
      "learning_rate": 2.396539764591826e-07,
      "loss": 0.0582,
      "step": 17468
    },
    {
      "epoch": 0.9020914020139427,
      "grad_norm": 15.402661323547363,
      "learning_rate": 2.3940341212141447e-07,
      "loss": 0.1893,
      "step": 17469
    },
    {
      "epoch": 0.9021430415698425,
      "grad_norm": 3.863898992538452,
      "learning_rate": 2.3915297562506877e-07,
      "loss": 0.0489,
      "step": 17470
    },
    {
      "epoch": 0.9021946811257423,
      "grad_norm": 4.00190544128418,
      "learning_rate": 2.3890266697686847e-07,
      "loss": 0.0468,
      "step": 17471
    },
    {
      "epoch": 0.9022463206816421,
      "grad_norm": 1.52483069896698,
      "learning_rate": 2.386524861835371e-07,
      "loss": 0.0275,
      "step": 17472
    },
    {
      "epoch": 0.9022979602375419,
      "grad_norm": 10.619468688964844,
      "learning_rate": 2.3840243325179312e-07,
      "loss": 0.0759,
      "step": 17473
    },
    {
      "epoch": 0.9023495997934418,
      "grad_norm": 4.507717609405518,
      "learning_rate": 2.3815250818835124e-07,
      "loss": 0.0251,
      "step": 17474
    },
    {
      "epoch": 0.9024012393493416,
      "grad_norm": 9.412747383117676,
      "learning_rate": 2.3790271099992212e-07,
      "loss": 0.0412,
      "step": 17475
    },
    {
      "epoch": 0.9024528789052414,
      "grad_norm": 1.5737158060073853,
      "learning_rate": 2.3765304169321378e-07,
      "loss": 0.0141,
      "step": 17476
    },
    {
      "epoch": 0.9025045184611412,
      "grad_norm": 1.9468697309494019,
      "learning_rate": 2.374035002749331e-07,
      "loss": 0.0196,
      "step": 17477
    },
    {
      "epoch": 0.902556158017041,
      "grad_norm": 4.3724565505981445,
      "learning_rate": 2.371540867517791e-07,
      "loss": 0.0261,
      "step": 17478
    },
    {
      "epoch": 0.9026077975729409,
      "grad_norm": 2.477160930633545,
      "learning_rate": 2.3690480113044978e-07,
      "loss": 0.035,
      "step": 17479
    },
    {
      "epoch": 0.9026594371288407,
      "grad_norm": 4.102935314178467,
      "learning_rate": 2.3665564341764037e-07,
      "loss": 0.0316,
      "step": 17480
    },
    {
      "epoch": 0.9027110766847405,
      "grad_norm": 16.723159790039062,
      "learning_rate": 2.3640661362004046e-07,
      "loss": 0.1206,
      "step": 17481
    },
    {
      "epoch": 0.9027627162406403,
      "grad_norm": 11.201249122619629,
      "learning_rate": 2.361577117443392e-07,
      "loss": 0.2891,
      "step": 17482
    },
    {
      "epoch": 0.9028143557965401,
      "grad_norm": 8.822382926940918,
      "learning_rate": 2.3590893779721903e-07,
      "loss": 0.0706,
      "step": 17483
    },
    {
      "epoch": 0.9028659953524399,
      "grad_norm": 3.88630747795105,
      "learning_rate": 2.3566029178536175e-07,
      "loss": 0.0305,
      "step": 17484
    },
    {
      "epoch": 0.9029176349083398,
      "grad_norm": 3.49965238571167,
      "learning_rate": 2.3541177371544377e-07,
      "loss": 0.0212,
      "step": 17485
    },
    {
      "epoch": 0.9029692744642396,
      "grad_norm": 0.9385899901390076,
      "learning_rate": 2.3516338359413915e-07,
      "loss": 0.0117,
      "step": 17486
    },
    {
      "epoch": 0.9030209140201394,
      "grad_norm": 4.2820916175842285,
      "learning_rate": 2.3491512142811755e-07,
      "loss": 0.0441,
      "step": 17487
    },
    {
      "epoch": 0.9030725535760392,
      "grad_norm": 2.604322671890259,
      "learning_rate": 2.3466698722404757e-07,
      "loss": 0.0394,
      "step": 17488
    },
    {
      "epoch": 0.903124193131939,
      "grad_norm": 5.7762370109558105,
      "learning_rate": 2.344189809885905e-07,
      "loss": 0.0798,
      "step": 17489
    },
    {
      "epoch": 0.9031758326878389,
      "grad_norm": 5.8537187576293945,
      "learning_rate": 2.3417110272840716e-07,
      "loss": 0.0376,
      "step": 17490
    },
    {
      "epoch": 0.9032274722437387,
      "grad_norm": 1.7423757314682007,
      "learning_rate": 2.3392335245015386e-07,
      "loss": 0.0148,
      "step": 17491
    },
    {
      "epoch": 0.9032791117996385,
      "grad_norm": 3.4113054275512695,
      "learning_rate": 2.3367573016048528e-07,
      "loss": 0.0374,
      "step": 17492
    },
    {
      "epoch": 0.9033307513555383,
      "grad_norm": 0.9746551513671875,
      "learning_rate": 2.334282358660489e-07,
      "loss": 0.0086,
      "step": 17493
    },
    {
      "epoch": 0.9033823909114381,
      "grad_norm": 3.324160575866699,
      "learning_rate": 2.3318086957349218e-07,
      "loss": 0.0459,
      "step": 17494
    },
    {
      "epoch": 0.903434030467338,
      "grad_norm": 6.941092491149902,
      "learning_rate": 2.3293363128945646e-07,
      "loss": 0.0406,
      "step": 17495
    },
    {
      "epoch": 0.9034856700232378,
      "grad_norm": 2.1393449306488037,
      "learning_rate": 2.326865210205842e-07,
      "loss": 0.037,
      "step": 17496
    },
    {
      "epoch": 0.9035373095791376,
      "grad_norm": 3.879002332687378,
      "learning_rate": 2.324395387735079e-07,
      "loss": 0.0388,
      "step": 17497
    },
    {
      "epoch": 0.9035889491350374,
      "grad_norm": 1.4913809299468994,
      "learning_rate": 2.321926845548622e-07,
      "loss": 0.0171,
      "step": 17498
    },
    {
      "epoch": 0.9036405886909372,
      "grad_norm": 2.3998219966888428,
      "learning_rate": 2.319459583712752e-07,
      "loss": 0.027,
      "step": 17499
    },
    {
      "epoch": 0.903692228246837,
      "grad_norm": 5.793314456939697,
      "learning_rate": 2.3169936022937268e-07,
      "loss": 0.038,
      "step": 17500
    },
    {
      "epoch": 0.9037438678027369,
      "grad_norm": 7.387577533721924,
      "learning_rate": 2.3145289013577655e-07,
      "loss": 0.0983,
      "step": 17501
    },
    {
      "epoch": 0.9037955073586367,
      "grad_norm": 5.035586357116699,
      "learning_rate": 2.3120654809710596e-07,
      "loss": 0.0683,
      "step": 17502
    },
    {
      "epoch": 0.9038471469145365,
      "grad_norm": 3.593688726425171,
      "learning_rate": 2.309603341199773e-07,
      "loss": 0.0478,
      "step": 17503
    },
    {
      "epoch": 0.9038987864704363,
      "grad_norm": 3.3656275272369385,
      "learning_rate": 2.307142482109992e-07,
      "loss": 0.0503,
      "step": 17504
    },
    {
      "epoch": 0.9039504260263361,
      "grad_norm": 4.465698719024658,
      "learning_rate": 2.3046829037678298e-07,
      "loss": 0.0601,
      "step": 17505
    },
    {
      "epoch": 0.904002065582236,
      "grad_norm": 6.435846328735352,
      "learning_rate": 2.3022246062393284e-07,
      "loss": 0.0425,
      "step": 17506
    },
    {
      "epoch": 0.9040537051381358,
      "grad_norm": 1.892943263053894,
      "learning_rate": 2.299767589590507e-07,
      "loss": 0.0266,
      "step": 17507
    },
    {
      "epoch": 0.9041053446940356,
      "grad_norm": 4.181267738342285,
      "learning_rate": 2.297311853887335e-07,
      "loss": 0.054,
      "step": 17508
    },
    {
      "epoch": 0.9041569842499354,
      "grad_norm": 10.454975128173828,
      "learning_rate": 2.2948573991957544e-07,
      "loss": 0.0486,
      "step": 17509
    },
    {
      "epoch": 0.9042086238058352,
      "grad_norm": 2.7442188262939453,
      "learning_rate": 2.292404225581707e-07,
      "loss": 0.0095,
      "step": 17510
    },
    {
      "epoch": 0.904260263361735,
      "grad_norm": 11.016021728515625,
      "learning_rate": 2.2899523331110453e-07,
      "loss": 0.0673,
      "step": 17511
    },
    {
      "epoch": 0.9043119029176349,
      "grad_norm": 3.1713032722473145,
      "learning_rate": 2.2875017218496166e-07,
      "loss": 0.053,
      "step": 17512
    },
    {
      "epoch": 0.9043635424735347,
      "grad_norm": 6.740580081939697,
      "learning_rate": 2.2850523918632294e-07,
      "loss": 0.0702,
      "step": 17513
    },
    {
      "epoch": 0.9044151820294345,
      "grad_norm": 2.289311170578003,
      "learning_rate": 2.2826043432176648e-07,
      "loss": 0.0191,
      "step": 17514
    },
    {
      "epoch": 0.9044668215853343,
      "grad_norm": 4.701004981994629,
      "learning_rate": 2.2801575759786588e-07,
      "loss": 0.058,
      "step": 17515
    },
    {
      "epoch": 0.9045184611412341,
      "grad_norm": 9.359827995300293,
      "learning_rate": 2.2777120902119144e-07,
      "loss": 0.1086,
      "step": 17516
    },
    {
      "epoch": 0.904570100697134,
      "grad_norm": 2.6086173057556152,
      "learning_rate": 2.2752678859831066e-07,
      "loss": 0.0364,
      "step": 17517
    },
    {
      "epoch": 0.9046217402530339,
      "grad_norm": 6.713955402374268,
      "learning_rate": 2.2728249633578724e-07,
      "loss": 0.0587,
      "step": 17518
    },
    {
      "epoch": 0.9046733798089337,
      "grad_norm": 4.1425676345825195,
      "learning_rate": 2.2703833224018147e-07,
      "loss": 0.0799,
      "step": 17519
    },
    {
      "epoch": 0.9047250193648335,
      "grad_norm": 4.341177463531494,
      "learning_rate": 2.2679429631804973e-07,
      "loss": 0.0455,
      "step": 17520
    },
    {
      "epoch": 0.9047766589207333,
      "grad_norm": 23.681461334228516,
      "learning_rate": 2.2655038857594625e-07,
      "loss": 0.1347,
      "step": 17521
    },
    {
      "epoch": 0.9048282984766332,
      "grad_norm": 3.1652936935424805,
      "learning_rate": 2.263066090204197e-07,
      "loss": 0.0336,
      "step": 17522
    },
    {
      "epoch": 0.904879938032533,
      "grad_norm": 3.1124165058135986,
      "learning_rate": 2.2606295765801644e-07,
      "loss": 0.0496,
      "step": 17523
    },
    {
      "epoch": 0.9049315775884328,
      "grad_norm": 25.763357162475586,
      "learning_rate": 2.2581943449528132e-07,
      "loss": 0.1521,
      "step": 17524
    },
    {
      "epoch": 0.9049832171443326,
      "grad_norm": 21.91136932373047,
      "learning_rate": 2.2557603953875352e-07,
      "loss": 0.0905,
      "step": 17525
    },
    {
      "epoch": 0.9050348567002324,
      "grad_norm": 8.52686595916748,
      "learning_rate": 2.2533277279496723e-07,
      "loss": 0.0561,
      "step": 17526
    },
    {
      "epoch": 0.9050864962561322,
      "grad_norm": 1.6538037061691284,
      "learning_rate": 2.2508963427045726e-07,
      "loss": 0.0105,
      "step": 17527
    },
    {
      "epoch": 0.9051381358120321,
      "grad_norm": 1.9289827346801758,
      "learning_rate": 2.2484662397175117e-07,
      "loss": 0.024,
      "step": 17528
    },
    {
      "epoch": 0.9051897753679319,
      "grad_norm": 3.524123191833496,
      "learning_rate": 2.2460374190537647e-07,
      "loss": 0.0549,
      "step": 17529
    },
    {
      "epoch": 0.9052414149238317,
      "grad_norm": 16.455570220947266,
      "learning_rate": 2.2436098807785468e-07,
      "loss": 0.1673,
      "step": 17530
    },
    {
      "epoch": 0.9052930544797315,
      "grad_norm": 13.248175621032715,
      "learning_rate": 2.241183624957044e-07,
      "loss": 0.2394,
      "step": 17531
    },
    {
      "epoch": 0.9053446940356313,
      "grad_norm": 5.828254222869873,
      "learning_rate": 2.2387586516544157e-07,
      "loss": 0.0307,
      "step": 17532
    },
    {
      "epoch": 0.9053963335915312,
      "grad_norm": 2.0084896087646484,
      "learning_rate": 2.2363349609357877e-07,
      "loss": 0.0354,
      "step": 17533
    },
    {
      "epoch": 0.905447973147431,
      "grad_norm": 4.037087440490723,
      "learning_rate": 2.2339125528662297e-07,
      "loss": 0.0653,
      "step": 17534
    },
    {
      "epoch": 0.9054996127033308,
      "grad_norm": 5.501960754394531,
      "learning_rate": 2.2314914275108123e-07,
      "loss": 0.0518,
      "step": 17535
    },
    {
      "epoch": 0.9055512522592306,
      "grad_norm": 10.763802528381348,
      "learning_rate": 2.2290715849345446e-07,
      "loss": 0.0926,
      "step": 17536
    },
    {
      "epoch": 0.9056028918151304,
      "grad_norm": 1.6660161018371582,
      "learning_rate": 2.2266530252023967e-07,
      "loss": 0.03,
      "step": 17537
    },
    {
      "epoch": 0.9056545313710302,
      "grad_norm": 3.0523455142974854,
      "learning_rate": 2.2242357483793385e-07,
      "loss": 0.0333,
      "step": 17538
    },
    {
      "epoch": 0.9057061709269301,
      "grad_norm": 2.1304640769958496,
      "learning_rate": 2.2218197545302688e-07,
      "loss": 0.0377,
      "step": 17539
    },
    {
      "epoch": 0.9057578104828299,
      "grad_norm": 13.347274780273438,
      "learning_rate": 2.21940504372008e-07,
      "loss": 0.1368,
      "step": 17540
    },
    {
      "epoch": 0.9058094500387297,
      "grad_norm": 2.753631591796875,
      "learning_rate": 2.2169916160136029e-07,
      "loss": 0.0249,
      "step": 17541
    },
    {
      "epoch": 0.9058610895946295,
      "grad_norm": 8.887395858764648,
      "learning_rate": 2.2145794714756475e-07,
      "loss": 0.05,
      "step": 17542
    },
    {
      "epoch": 0.9059127291505293,
      "grad_norm": 3.6736974716186523,
      "learning_rate": 2.2121686101710116e-07,
      "loss": 0.0421,
      "step": 17543
    },
    {
      "epoch": 0.9059643687064292,
      "grad_norm": 10.640625,
      "learning_rate": 2.2097590321644103e-07,
      "loss": 0.0589,
      "step": 17544
    },
    {
      "epoch": 0.906016008262329,
      "grad_norm": 3.0128543376922607,
      "learning_rate": 2.2073507375205583e-07,
      "loss": 0.0283,
      "step": 17545
    },
    {
      "epoch": 0.9060676478182288,
      "grad_norm": 4.087791919708252,
      "learning_rate": 2.2049437263041374e-07,
      "loss": 0.0228,
      "step": 17546
    },
    {
      "epoch": 0.9061192873741286,
      "grad_norm": 12.101259231567383,
      "learning_rate": 2.202537998579779e-07,
      "loss": 0.062,
      "step": 17547
    },
    {
      "epoch": 0.9061709269300284,
      "grad_norm": 1.2034907341003418,
      "learning_rate": 2.2001335544120818e-07,
      "loss": 0.0151,
      "step": 17548
    },
    {
      "epoch": 0.9062225664859282,
      "grad_norm": 2.5290873050689697,
      "learning_rate": 2.1977303938656215e-07,
      "loss": 0.0528,
      "step": 17549
    },
    {
      "epoch": 0.9062742060418281,
      "grad_norm": 3.4478416442871094,
      "learning_rate": 2.1953285170049355e-07,
      "loss": 0.0359,
      "step": 17550
    },
    {
      "epoch": 0.9063258455977279,
      "grad_norm": 4.956068515777588,
      "learning_rate": 2.1929279238945166e-07,
      "loss": 0.1092,
      "step": 17551
    },
    {
      "epoch": 0.9063774851536277,
      "grad_norm": 9.474433898925781,
      "learning_rate": 2.190528614598836e-07,
      "loss": 0.0811,
      "step": 17552
    },
    {
      "epoch": 0.9064291247095275,
      "grad_norm": 1.748453140258789,
      "learning_rate": 2.188130589182319e-07,
      "loss": 0.0238,
      "step": 17553
    },
    {
      "epoch": 0.9064807642654273,
      "grad_norm": 8.504152297973633,
      "learning_rate": 2.185733847709376e-07,
      "loss": 0.092,
      "step": 17554
    },
    {
      "epoch": 0.9065324038213272,
      "grad_norm": 9.42210865020752,
      "learning_rate": 2.183338390244355e-07,
      "loss": 0.052,
      "step": 17555
    },
    {
      "epoch": 0.906584043377227,
      "grad_norm": 6.101425647735596,
      "learning_rate": 2.180944216851577e-07,
      "loss": 0.0567,
      "step": 17556
    },
    {
      "epoch": 0.9066356829331268,
      "grad_norm": 14.931321144104004,
      "learning_rate": 2.1785513275953518e-07,
      "loss": 0.2544,
      "step": 17557
    },
    {
      "epoch": 0.9066873224890266,
      "grad_norm": 13.422306060791016,
      "learning_rate": 2.1761597225399444e-07,
      "loss": 0.1619,
      "step": 17558
    },
    {
      "epoch": 0.9067389620449264,
      "grad_norm": 17.15857696533203,
      "learning_rate": 2.1737694017495592e-07,
      "loss": 0.0624,
      "step": 17559
    },
    {
      "epoch": 0.9067906016008263,
      "grad_norm": 2.732569456100464,
      "learning_rate": 2.1713803652883948e-07,
      "loss": 0.0242,
      "step": 17560
    },
    {
      "epoch": 0.9068422411567261,
      "grad_norm": 2.717162609100342,
      "learning_rate": 2.168992613220594e-07,
      "loss": 0.0305,
      "step": 17561
    },
    {
      "epoch": 0.9068938807126259,
      "grad_norm": 3.2742161750793457,
      "learning_rate": 2.1666061456103115e-07,
      "loss": 0.0587,
      "step": 17562
    },
    {
      "epoch": 0.9069455202685257,
      "grad_norm": 7.968153953552246,
      "learning_rate": 2.1642209625216015e-07,
      "loss": 0.0506,
      "step": 17563
    },
    {
      "epoch": 0.9069971598244255,
      "grad_norm": 2.3831207752227783,
      "learning_rate": 2.1618370640185238e-07,
      "loss": 0.0247,
      "step": 17564
    },
    {
      "epoch": 0.9070487993803253,
      "grad_norm": 2.259882688522339,
      "learning_rate": 2.159454450165105e-07,
      "loss": 0.0374,
      "step": 17565
    },
    {
      "epoch": 0.9071004389362252,
      "grad_norm": 4.596446990966797,
      "learning_rate": 2.157073121025316e-07,
      "loss": 0.0593,
      "step": 17566
    },
    {
      "epoch": 0.907152078492125,
      "grad_norm": 4.176618576049805,
      "learning_rate": 2.154693076663117e-07,
      "loss": 0.0519,
      "step": 17567
    },
    {
      "epoch": 0.9072037180480248,
      "grad_norm": 3.599382162094116,
      "learning_rate": 2.1523143171424066e-07,
      "loss": 0.0308,
      "step": 17568
    },
    {
      "epoch": 0.9072553576039246,
      "grad_norm": 10.928150177001953,
      "learning_rate": 2.1499368425270838e-07,
      "loss": 0.0727,
      "step": 17569
    },
    {
      "epoch": 0.9073069971598244,
      "grad_norm": 5.390889644622803,
      "learning_rate": 2.1475606528809755e-07,
      "loss": 0.0928,
      "step": 17570
    },
    {
      "epoch": 0.9073586367157243,
      "grad_norm": 2.746218681335449,
      "learning_rate": 2.1451857482678974e-07,
      "loss": 0.0316,
      "step": 17571
    },
    {
      "epoch": 0.9074102762716241,
      "grad_norm": 2.3461787700653076,
      "learning_rate": 2.1428121287516314e-07,
      "loss": 0.0331,
      "step": 17572
    },
    {
      "epoch": 0.9074619158275239,
      "grad_norm": 2.2024006843566895,
      "learning_rate": 2.140439794395921e-07,
      "loss": 0.022,
      "step": 17573
    },
    {
      "epoch": 0.9075135553834237,
      "grad_norm": 1.3225504159927368,
      "learning_rate": 2.13806874526446e-07,
      "loss": 0.0168,
      "step": 17574
    },
    {
      "epoch": 0.9075651949393235,
      "grad_norm": 12.538565635681152,
      "learning_rate": 2.135698981420925e-07,
      "loss": 0.1667,
      "step": 17575
    },
    {
      "epoch": 0.9076168344952233,
      "grad_norm": 4.628666877746582,
      "learning_rate": 2.1333305029289652e-07,
      "loss": 0.0343,
      "step": 17576
    },
    {
      "epoch": 0.9076684740511232,
      "grad_norm": 8.475297927856445,
      "learning_rate": 2.1309633098521687e-07,
      "loss": 0.0654,
      "step": 17577
    },
    {
      "epoch": 0.907720113607023,
      "grad_norm": 8.549561500549316,
      "learning_rate": 2.1285974022541121e-07,
      "loss": 0.0875,
      "step": 17578
    },
    {
      "epoch": 0.9077717531629228,
      "grad_norm": 5.977415084838867,
      "learning_rate": 2.1262327801983284e-07,
      "loss": 0.0675,
      "step": 17579
    },
    {
      "epoch": 0.9078233927188226,
      "grad_norm": 5.592618942260742,
      "learning_rate": 2.1238694437483166e-07,
      "loss": 0.0494,
      "step": 17580
    },
    {
      "epoch": 0.9078750322747224,
      "grad_norm": 3.499526262283325,
      "learning_rate": 2.1215073929675366e-07,
      "loss": 0.0546,
      "step": 17581
    },
    {
      "epoch": 0.9079266718306223,
      "grad_norm": 2.47104811668396,
      "learning_rate": 2.119146627919433e-07,
      "loss": 0.0275,
      "step": 17582
    },
    {
      "epoch": 0.9079783113865221,
      "grad_norm": 6.02549934387207,
      "learning_rate": 2.116787148667393e-07,
      "loss": 0.0635,
      "step": 17583
    },
    {
      "epoch": 0.9080299509424219,
      "grad_norm": 3.4916346073150635,
      "learning_rate": 2.1144289552747776e-07,
      "loss": 0.0329,
      "step": 17584
    },
    {
      "epoch": 0.9080815904983217,
      "grad_norm": 2.9168999195098877,
      "learning_rate": 2.1120720478049138e-07,
      "loss": 0.0335,
      "step": 17585
    },
    {
      "epoch": 0.9081332300542215,
      "grad_norm": 5.175474166870117,
      "learning_rate": 2.109716426321101e-07,
      "loss": 0.0322,
      "step": 17586
    },
    {
      "epoch": 0.9081848696101213,
      "grad_norm": 9.549698829650879,
      "learning_rate": 2.1073620908865945e-07,
      "loss": 0.0906,
      "step": 17587
    },
    {
      "epoch": 0.9082365091660212,
      "grad_norm": 3.0905704498291016,
      "learning_rate": 2.1050090415646097e-07,
      "loss": 0.0378,
      "step": 17588
    },
    {
      "epoch": 0.908288148721921,
      "grad_norm": 6.556323051452637,
      "learning_rate": 2.1026572784183352e-07,
      "loss": 0.0859,
      "step": 17589
    },
    {
      "epoch": 0.9083397882778208,
      "grad_norm": 2.1259782314300537,
      "learning_rate": 2.1003068015109372e-07,
      "loss": 0.0224,
      "step": 17590
    },
    {
      "epoch": 0.9083914278337206,
      "grad_norm": 8.914006233215332,
      "learning_rate": 2.0979576109055377e-07,
      "loss": 0.0901,
      "step": 17591
    },
    {
      "epoch": 0.9084430673896204,
      "grad_norm": 8.369852066040039,
      "learning_rate": 2.095609706665208e-07,
      "loss": 0.0768,
      "step": 17592
    },
    {
      "epoch": 0.9084947069455203,
      "grad_norm": 2.5190179347991943,
      "learning_rate": 2.0932630888530037e-07,
      "loss": 0.0287,
      "step": 17593
    },
    {
      "epoch": 0.9085463465014201,
      "grad_norm": 1.403131365776062,
      "learning_rate": 2.0909177575319407e-07,
      "loss": 0.0093,
      "step": 17594
    },
    {
      "epoch": 0.9085979860573199,
      "grad_norm": 2.9008045196533203,
      "learning_rate": 2.0885737127650075e-07,
      "loss": 0.0217,
      "step": 17595
    },
    {
      "epoch": 0.9086496256132197,
      "grad_norm": 5.195286750793457,
      "learning_rate": 2.0862309546151482e-07,
      "loss": 0.0647,
      "step": 17596
    },
    {
      "epoch": 0.9087012651691195,
      "grad_norm": 6.361313819885254,
      "learning_rate": 2.0838894831452684e-07,
      "loss": 0.0394,
      "step": 17597
    },
    {
      "epoch": 0.9087529047250194,
      "grad_norm": 3.954216957092285,
      "learning_rate": 2.0815492984182505e-07,
      "loss": 0.0403,
      "step": 17598
    },
    {
      "epoch": 0.9088045442809192,
      "grad_norm": 11.402932167053223,
      "learning_rate": 2.0792104004969393e-07,
      "loss": 0.0775,
      "step": 17599
    },
    {
      "epoch": 0.908856183836819,
      "grad_norm": 8.749982833862305,
      "learning_rate": 2.0768727894441453e-07,
      "loss": 0.1005,
      "step": 17600
    },
    {
      "epoch": 0.9089078233927188,
      "grad_norm": 4.429752349853516,
      "learning_rate": 2.0745364653226408e-07,
      "loss": 0.042,
      "step": 17601
    },
    {
      "epoch": 0.9089594629486186,
      "grad_norm": 2.9112095832824707,
      "learning_rate": 2.0722014281951696e-07,
      "loss": 0.046,
      "step": 17602
    },
    {
      "epoch": 0.9090111025045184,
      "grad_norm": 6.615159511566162,
      "learning_rate": 2.0698676781244208e-07,
      "loss": 0.0692,
      "step": 17603
    },
    {
      "epoch": 0.9090627420604183,
      "grad_norm": 8.437524795532227,
      "learning_rate": 2.067535215173083e-07,
      "loss": 0.1212,
      "step": 17604
    },
    {
      "epoch": 0.9091143816163181,
      "grad_norm": 2.947580099105835,
      "learning_rate": 2.0652040394037897e-07,
      "loss": 0.0347,
      "step": 17605
    },
    {
      "epoch": 0.9091660211722179,
      "grad_norm": 6.578843116760254,
      "learning_rate": 2.062874150879146e-07,
      "loss": 0.1016,
      "step": 17606
    },
    {
      "epoch": 0.9092176607281177,
      "grad_norm": 4.484181880950928,
      "learning_rate": 2.0605455496617022e-07,
      "loss": 0.0508,
      "step": 17607
    },
    {
      "epoch": 0.9092693002840175,
      "grad_norm": 6.340688705444336,
      "learning_rate": 2.0582182358139968e-07,
      "loss": 0.0412,
      "step": 17608
    },
    {
      "epoch": 0.9093209398399174,
      "grad_norm": 4.043738842010498,
      "learning_rate": 2.055892209398541e-07,
      "loss": 0.0499,
      "step": 17609
    },
    {
      "epoch": 0.9093725793958172,
      "grad_norm": 16.187641143798828,
      "learning_rate": 2.0535674704777852e-07,
      "loss": 0.1317,
      "step": 17610
    },
    {
      "epoch": 0.909424218951717,
      "grad_norm": 1.0646297931671143,
      "learning_rate": 2.0512440191141624e-07,
      "loss": 0.0093,
      "step": 17611
    },
    {
      "epoch": 0.9094758585076168,
      "grad_norm": 2.3603434562683105,
      "learning_rate": 2.0489218553700673e-07,
      "loss": 0.0446,
      "step": 17612
    },
    {
      "epoch": 0.9095274980635166,
      "grad_norm": 4.112326145172119,
      "learning_rate": 2.0466009793078557e-07,
      "loss": 0.0762,
      "step": 17613
    },
    {
      "epoch": 0.9095791376194164,
      "grad_norm": 5.667076110839844,
      "learning_rate": 2.0442813909898552e-07,
      "loss": 0.074,
      "step": 17614
    },
    {
      "epoch": 0.9096307771753163,
      "grad_norm": 3.7325010299682617,
      "learning_rate": 2.041963090478355e-07,
      "loss": 0.0195,
      "step": 17615
    },
    {
      "epoch": 0.9096824167312161,
      "grad_norm": 9.201976776123047,
      "learning_rate": 2.0396460778356108e-07,
      "loss": 0.0976,
      "step": 17616
    },
    {
      "epoch": 0.9097340562871159,
      "grad_norm": 7.143703460693359,
      "learning_rate": 2.0373303531238507e-07,
      "loss": 0.0512,
      "step": 17617
    },
    {
      "epoch": 0.9097856958430157,
      "grad_norm": 4.546424865722656,
      "learning_rate": 2.035015916405253e-07,
      "loss": 0.0471,
      "step": 17618
    },
    {
      "epoch": 0.9098373353989155,
      "grad_norm": 1.9815834760665894,
      "learning_rate": 2.0327027677419676e-07,
      "loss": 0.0371,
      "step": 17619
    },
    {
      "epoch": 0.9098889749548154,
      "grad_norm": 2.0857653617858887,
      "learning_rate": 2.030390907196128e-07,
      "loss": 0.0285,
      "step": 17620
    },
    {
      "epoch": 0.9099406145107152,
      "grad_norm": 3.983046054840088,
      "learning_rate": 2.0280803348298018e-07,
      "loss": 0.0613,
      "step": 17621
    },
    {
      "epoch": 0.909992254066615,
      "grad_norm": 5.956101894378662,
      "learning_rate": 2.0257710507050277e-07,
      "loss": 0.0666,
      "step": 17622
    },
    {
      "epoch": 0.9100438936225148,
      "grad_norm": 2.0756707191467285,
      "learning_rate": 2.0234630548838453e-07,
      "loss": 0.0227,
      "step": 17623
    },
    {
      "epoch": 0.9100955331784146,
      "grad_norm": 3.867842674255371,
      "learning_rate": 2.021156347428227e-07,
      "loss": 0.0316,
      "step": 17624
    },
    {
      "epoch": 0.9101471727343144,
      "grad_norm": 8.826570510864258,
      "learning_rate": 2.0188509284001067e-07,
      "loss": 0.1124,
      "step": 17625
    },
    {
      "epoch": 0.9101988122902143,
      "grad_norm": 2.378701686859131,
      "learning_rate": 2.016546797861396e-07,
      "loss": 0.0321,
      "step": 17626
    },
    {
      "epoch": 0.9102504518461141,
      "grad_norm": 5.8281049728393555,
      "learning_rate": 2.014243955873968e-07,
      "loss": 0.0758,
      "step": 17627
    },
    {
      "epoch": 0.9103020914020139,
      "grad_norm": 4.4957356452941895,
      "learning_rate": 2.011942402499678e-07,
      "loss": 0.0568,
      "step": 17628
    },
    {
      "epoch": 0.9103537309579137,
      "grad_norm": 3.2098536491394043,
      "learning_rate": 2.0096421378003217e-07,
      "loss": 0.0384,
      "step": 17629
    },
    {
      "epoch": 0.9104053705138135,
      "grad_norm": 2.5532078742980957,
      "learning_rate": 2.0073431618376714e-07,
      "loss": 0.0293,
      "step": 17630
    },
    {
      "epoch": 0.9104570100697134,
      "grad_norm": 3.96267032623291,
      "learning_rate": 2.0050454746734615e-07,
      "loss": 0.0533,
      "step": 17631
    },
    {
      "epoch": 0.9105086496256132,
      "grad_norm": 1.5848591327667236,
      "learning_rate": 2.0027490763694034e-07,
      "loss": 0.0091,
      "step": 17632
    },
    {
      "epoch": 0.910560289181513,
      "grad_norm": 2.597332239151001,
      "learning_rate": 2.0004539669871536e-07,
      "loss": 0.0324,
      "step": 17633
    },
    {
      "epoch": 0.9106119287374128,
      "grad_norm": 11.732217788696289,
      "learning_rate": 1.9981601465883516e-07,
      "loss": 0.0765,
      "step": 17634
    },
    {
      "epoch": 0.9106635682933126,
      "grad_norm": 5.200380802154541,
      "learning_rate": 1.995867615234598e-07,
      "loss": 0.0529,
      "step": 17635
    },
    {
      "epoch": 0.9107152078492124,
      "grad_norm": 6.836597442626953,
      "learning_rate": 1.9935763729874435e-07,
      "loss": 0.0523,
      "step": 17636
    },
    {
      "epoch": 0.9107668474051123,
      "grad_norm": 1.9412975311279297,
      "learning_rate": 1.9912864199084337e-07,
      "loss": 0.0376,
      "step": 17637
    },
    {
      "epoch": 0.9108184869610121,
      "grad_norm": 3.234762668609619,
      "learning_rate": 1.988997756059058e-07,
      "loss": 0.0516,
      "step": 17638
    },
    {
      "epoch": 0.9108701265169119,
      "grad_norm": 3.5650739669799805,
      "learning_rate": 1.9867103815007727e-07,
      "loss": 0.0331,
      "step": 17639
    },
    {
      "epoch": 0.9109217660728117,
      "grad_norm": 2.2584903240203857,
      "learning_rate": 1.984424296295001e-07,
      "loss": 0.0255,
      "step": 17640
    },
    {
      "epoch": 0.9109734056287115,
      "grad_norm": 8.458324432373047,
      "learning_rate": 1.982139500503133e-07,
      "loss": 0.0493,
      "step": 17641
    },
    {
      "epoch": 0.9110250451846114,
      "grad_norm": 7.794546604156494,
      "learning_rate": 1.979855994186547e-07,
      "loss": 0.0608,
      "step": 17642
    },
    {
      "epoch": 0.9110766847405112,
      "grad_norm": 2.9401590824127197,
      "learning_rate": 1.977573777406533e-07,
      "loss": 0.04,
      "step": 17643
    },
    {
      "epoch": 0.911128324296411,
      "grad_norm": 3.403621196746826,
      "learning_rate": 1.9752928502243973e-07,
      "loss": 0.0379,
      "step": 17644
    },
    {
      "epoch": 0.9111799638523108,
      "grad_norm": 1.5977991819381714,
      "learning_rate": 1.9730132127013802e-07,
      "loss": 0.0147,
      "step": 17645
    },
    {
      "epoch": 0.9112316034082107,
      "grad_norm": 1.4355863332748413,
      "learning_rate": 1.9707348648987102e-07,
      "loss": 0.0211,
      "step": 17646
    },
    {
      "epoch": 0.9112832429641106,
      "grad_norm": 10.748343467712402,
      "learning_rate": 1.9684578068775662e-07,
      "loss": 0.0701,
      "step": 17647
    },
    {
      "epoch": 0.9113348825200104,
      "grad_norm": 11.502498626708984,
      "learning_rate": 1.9661820386990936e-07,
      "loss": 0.0608,
      "step": 17648
    },
    {
      "epoch": 0.9113865220759102,
      "grad_norm": 8.410953521728516,
      "learning_rate": 1.9639075604244163e-07,
      "loss": 0.2169,
      "step": 17649
    },
    {
      "epoch": 0.91143816163181,
      "grad_norm": 3.6351795196533203,
      "learning_rate": 1.9616343721145903e-07,
      "loss": 0.0375,
      "step": 17650
    },
    {
      "epoch": 0.9114898011877098,
      "grad_norm": 4.9469685554504395,
      "learning_rate": 1.9593624738306838e-07,
      "loss": 0.0736,
      "step": 17651
    },
    {
      "epoch": 0.9115414407436097,
      "grad_norm": 14.161441802978516,
      "learning_rate": 1.957091865633698e-07,
      "loss": 0.1337,
      "step": 17652
    },
    {
      "epoch": 0.9115930802995095,
      "grad_norm": 7.633440971374512,
      "learning_rate": 1.9548225475846117e-07,
      "loss": 0.1053,
      "step": 17653
    },
    {
      "epoch": 0.9116447198554093,
      "grad_norm": 4.424452304840088,
      "learning_rate": 1.9525545197443542e-07,
      "loss": 0.0573,
      "step": 17654
    },
    {
      "epoch": 0.9116963594113091,
      "grad_norm": 5.471057891845703,
      "learning_rate": 1.950287782173832e-07,
      "loss": 0.0569,
      "step": 17655
    },
    {
      "epoch": 0.9117479989672089,
      "grad_norm": 6.570034503936768,
      "learning_rate": 1.94802233493393e-07,
      "loss": 0.0805,
      "step": 17656
    },
    {
      "epoch": 0.9117996385231087,
      "grad_norm": 4.6006059646606445,
      "learning_rate": 1.9457581780854883e-07,
      "loss": 0.04,
      "step": 17657
    },
    {
      "epoch": 0.9118512780790086,
      "grad_norm": 9.728150367736816,
      "learning_rate": 1.9434953116892863e-07,
      "loss": 0.0525,
      "step": 17658
    },
    {
      "epoch": 0.9119029176349084,
      "grad_norm": 3.7793350219726562,
      "learning_rate": 1.9412337358061027e-07,
      "loss": 0.0255,
      "step": 17659
    },
    {
      "epoch": 0.9119545571908082,
      "grad_norm": 5.770754814147949,
      "learning_rate": 1.9389734504966727e-07,
      "loss": 0.0601,
      "step": 17660
    },
    {
      "epoch": 0.912006196746708,
      "grad_norm": 7.846729278564453,
      "learning_rate": 1.9367144558216867e-07,
      "loss": 0.0517,
      "step": 17661
    },
    {
      "epoch": 0.9120578363026078,
      "grad_norm": 4.5397138595581055,
      "learning_rate": 1.934456751841818e-07,
      "loss": 0.0557,
      "step": 17662
    },
    {
      "epoch": 0.9121094758585077,
      "grad_norm": 3.3401505947113037,
      "learning_rate": 1.9322003386176845e-07,
      "loss": 0.0658,
      "step": 17663
    },
    {
      "epoch": 0.9121611154144075,
      "grad_norm": 3.825387954711914,
      "learning_rate": 1.9299452162098886e-07,
      "loss": 0.046,
      "step": 17664
    },
    {
      "epoch": 0.9122127549703073,
      "grad_norm": 16.77959442138672,
      "learning_rate": 1.927691384678987e-07,
      "loss": 0.0663,
      "step": 17665
    },
    {
      "epoch": 0.9122643945262071,
      "grad_norm": 2.3857874870300293,
      "learning_rate": 1.925438844085503e-07,
      "loss": 0.0484,
      "step": 17666
    },
    {
      "epoch": 0.9123160340821069,
      "grad_norm": 2.790130138397217,
      "learning_rate": 1.923187594489928e-07,
      "loss": 0.0345,
      "step": 17667
    },
    {
      "epoch": 0.9123676736380067,
      "grad_norm": 28.952123641967773,
      "learning_rate": 1.9209376359527187e-07,
      "loss": 0.1452,
      "step": 17668
    },
    {
      "epoch": 0.9124193131939066,
      "grad_norm": 3.48983097076416,
      "learning_rate": 1.918688968534288e-07,
      "loss": 0.0369,
      "step": 17669
    },
    {
      "epoch": 0.9124709527498064,
      "grad_norm": 2.8294081687927246,
      "learning_rate": 1.9164415922950264e-07,
      "loss": 0.0253,
      "step": 17670
    },
    {
      "epoch": 0.9125225923057062,
      "grad_norm": 12.269003868103027,
      "learning_rate": 1.9141955072952966e-07,
      "loss": 0.1137,
      "step": 17671
    },
    {
      "epoch": 0.912574231861606,
      "grad_norm": 2.298081159591675,
      "learning_rate": 1.9119507135953952e-07,
      "loss": 0.0198,
      "step": 17672
    },
    {
      "epoch": 0.9126258714175058,
      "grad_norm": 20.42461395263672,
      "learning_rate": 1.9097072112556124e-07,
      "loss": 0.1522,
      "step": 17673
    },
    {
      "epoch": 0.9126775109734057,
      "grad_norm": 5.259271621704102,
      "learning_rate": 1.907465000336195e-07,
      "loss": 0.092,
      "step": 17674
    },
    {
      "epoch": 0.9127291505293055,
      "grad_norm": 5.606030464172363,
      "learning_rate": 1.905224080897361e-07,
      "loss": 0.0866,
      "step": 17675
    },
    {
      "epoch": 0.9127807900852053,
      "grad_norm": 3.26149845123291,
      "learning_rate": 1.9029844529992846e-07,
      "loss": 0.0359,
      "step": 17676
    },
    {
      "epoch": 0.9128324296411051,
      "grad_norm": 1.26911199092865,
      "learning_rate": 1.900746116702107e-07,
      "loss": 0.0257,
      "step": 17677
    },
    {
      "epoch": 0.9128840691970049,
      "grad_norm": 3.002241373062134,
      "learning_rate": 1.8985090720659348e-07,
      "loss": 0.0338,
      "step": 17678
    },
    {
      "epoch": 0.9129357087529048,
      "grad_norm": 0.9909816980361938,
      "learning_rate": 1.8962733191508486e-07,
      "loss": 0.0078,
      "step": 17679
    },
    {
      "epoch": 0.9129873483088046,
      "grad_norm": 3.775545597076416,
      "learning_rate": 1.8940388580168834e-07,
      "loss": 0.0367,
      "step": 17680
    },
    {
      "epoch": 0.9130389878647044,
      "grad_norm": 17.13195037841797,
      "learning_rate": 1.891805688724041e-07,
      "loss": 0.0609,
      "step": 17681
    },
    {
      "epoch": 0.9130906274206042,
      "grad_norm": 6.12872314453125,
      "learning_rate": 1.8895738113323015e-07,
      "loss": 0.0719,
      "step": 17682
    },
    {
      "epoch": 0.913142266976504,
      "grad_norm": 2.1133224964141846,
      "learning_rate": 1.8873432259015777e-07,
      "loss": 0.0417,
      "step": 17683
    },
    {
      "epoch": 0.9131939065324038,
      "grad_norm": 6.343148708343506,
      "learning_rate": 1.8851139324917945e-07,
      "loss": 0.0531,
      "step": 17684
    },
    {
      "epoch": 0.9132455460883037,
      "grad_norm": 1.05375075340271,
      "learning_rate": 1.882885931162809e-07,
      "loss": 0.0066,
      "step": 17685
    },
    {
      "epoch": 0.9132971856442035,
      "grad_norm": 3.046417474746704,
      "learning_rate": 1.8806592219744512e-07,
      "loss": 0.0366,
      "step": 17686
    },
    {
      "epoch": 0.9133488252001033,
      "grad_norm": 3.8658783435821533,
      "learning_rate": 1.8784338049865182e-07,
      "loss": 0.0407,
      "step": 17687
    },
    {
      "epoch": 0.9134004647560031,
      "grad_norm": 3.9240479469299316,
      "learning_rate": 1.8762096802587559e-07,
      "loss": 0.0482,
      "step": 17688
    },
    {
      "epoch": 0.9134521043119029,
      "grad_norm": 1.9084011316299438,
      "learning_rate": 1.8739868478509172e-07,
      "loss": 0.0407,
      "step": 17689
    },
    {
      "epoch": 0.9135037438678028,
      "grad_norm": 4.697027683258057,
      "learning_rate": 1.8717653078226873e-07,
      "loss": 0.0929,
      "step": 17690
    },
    {
      "epoch": 0.9135553834237026,
      "grad_norm": 4.55647611618042,
      "learning_rate": 1.8695450602337128e-07,
      "loss": 0.0413,
      "step": 17691
    },
    {
      "epoch": 0.9136070229796024,
      "grad_norm": 3.437631607055664,
      "learning_rate": 1.8673261051436187e-07,
      "loss": 0.0269,
      "step": 17692
    },
    {
      "epoch": 0.9136586625355022,
      "grad_norm": 3.1797714233398438,
      "learning_rate": 1.8651084426120014e-07,
      "loss": 0.0679,
      "step": 17693
    },
    {
      "epoch": 0.913710302091402,
      "grad_norm": 2.2149558067321777,
      "learning_rate": 1.8628920726984079e-07,
      "loss": 0.0609,
      "step": 17694
    },
    {
      "epoch": 0.9137619416473018,
      "grad_norm": 3.630636215209961,
      "learning_rate": 1.860676995462357e-07,
      "loss": 0.043,
      "step": 17695
    },
    {
      "epoch": 0.9138135812032017,
      "grad_norm": 10.561867713928223,
      "learning_rate": 1.858463210963335e-07,
      "loss": 0.0686,
      "step": 17696
    },
    {
      "epoch": 0.9138652207591015,
      "grad_norm": 3.531808376312256,
      "learning_rate": 1.8562507192607882e-07,
      "loss": 0.0449,
      "step": 17697
    },
    {
      "epoch": 0.9139168603150013,
      "grad_norm": 2.2877845764160156,
      "learning_rate": 1.8540395204141358e-07,
      "loss": 0.0223,
      "step": 17698
    },
    {
      "epoch": 0.9139684998709011,
      "grad_norm": 4.348926067352295,
      "learning_rate": 1.851829614482753e-07,
      "loss": 0.0555,
      "step": 17699
    },
    {
      "epoch": 0.9140201394268009,
      "grad_norm": 2.8353071212768555,
      "learning_rate": 1.8496210015259807e-07,
      "loss": 0.0415,
      "step": 17700
    },
    {
      "epoch": 0.9140717789827008,
      "grad_norm": 2.5666158199310303,
      "learning_rate": 1.8474136816031497e-07,
      "loss": 0.0249,
      "step": 17701
    },
    {
      "epoch": 0.9141234185386006,
      "grad_norm": 1.6914616823196411,
      "learning_rate": 1.8452076547735009e-07,
      "loss": 0.0221,
      "step": 17702
    },
    {
      "epoch": 0.9141750580945004,
      "grad_norm": 5.02382755279541,
      "learning_rate": 1.8430029210963042e-07,
      "loss": 0.0563,
      "step": 17703
    },
    {
      "epoch": 0.9142266976504002,
      "grad_norm": 6.183992385864258,
      "learning_rate": 1.8407994806307618e-07,
      "loss": 0.0595,
      "step": 17704
    },
    {
      "epoch": 0.9142783372063,
      "grad_norm": 2.1538729667663574,
      "learning_rate": 1.8385973334360375e-07,
      "loss": 0.0303,
      "step": 17705
    },
    {
      "epoch": 0.9143299767621998,
      "grad_norm": 10.67163372039795,
      "learning_rate": 1.836396479571262e-07,
      "loss": 0.0839,
      "step": 17706
    },
    {
      "epoch": 0.9143816163180997,
      "grad_norm": 1.429053544998169,
      "learning_rate": 1.8341969190955433e-07,
      "loss": 0.0233,
      "step": 17707
    },
    {
      "epoch": 0.9144332558739995,
      "grad_norm": 3.596141815185547,
      "learning_rate": 1.8319986520679622e-07,
      "loss": 0.0196,
      "step": 17708
    },
    {
      "epoch": 0.9144848954298993,
      "grad_norm": 6.269039154052734,
      "learning_rate": 1.8298016785475327e-07,
      "loss": 0.0896,
      "step": 17709
    },
    {
      "epoch": 0.9145365349857991,
      "grad_norm": 6.516014099121094,
      "learning_rate": 1.827605998593257e-07,
      "loss": 0.1142,
      "step": 17710
    },
    {
      "epoch": 0.9145881745416989,
      "grad_norm": 13.690577507019043,
      "learning_rate": 1.8254116122641053e-07,
      "loss": 0.143,
      "step": 17711
    },
    {
      "epoch": 0.9146398140975988,
      "grad_norm": 12.115764617919922,
      "learning_rate": 1.8232185196189967e-07,
      "loss": 0.1202,
      "step": 17712
    },
    {
      "epoch": 0.9146914536534986,
      "grad_norm": 5.690013408660889,
      "learning_rate": 1.8210267207168286e-07,
      "loss": 0.0537,
      "step": 17713
    },
    {
      "epoch": 0.9147430932093984,
      "grad_norm": 9.412628173828125,
      "learning_rate": 1.8188362156164597e-07,
      "loss": 0.1016,
      "step": 17714
    },
    {
      "epoch": 0.9147947327652982,
      "grad_norm": 15.063261985778809,
      "learning_rate": 1.8166470043767204e-07,
      "loss": 0.0849,
      "step": 17715
    },
    {
      "epoch": 0.914846372321198,
      "grad_norm": 7.827045440673828,
      "learning_rate": 1.8144590870563806e-07,
      "loss": 0.0812,
      "step": 17716
    },
    {
      "epoch": 0.9148980118770978,
      "grad_norm": 3.2942490577697754,
      "learning_rate": 1.8122724637142152e-07,
      "loss": 0.0422,
      "step": 17717
    },
    {
      "epoch": 0.9149496514329977,
      "grad_norm": 5.026919364929199,
      "learning_rate": 1.8100871344089388e-07,
      "loss": 0.0795,
      "step": 17718
    },
    {
      "epoch": 0.9150012909888975,
      "grad_norm": 1.9972840547561646,
      "learning_rate": 1.8079030991992373e-07,
      "loss": 0.0379,
      "step": 17719
    },
    {
      "epoch": 0.9150529305447973,
      "grad_norm": 7.6906890869140625,
      "learning_rate": 1.8057203581437533e-07,
      "loss": 0.0827,
      "step": 17720
    },
    {
      "epoch": 0.9151045701006971,
      "grad_norm": 7.517714977264404,
      "learning_rate": 1.8035389113010948e-07,
      "loss": 0.1087,
      "step": 17721
    },
    {
      "epoch": 0.9151562096565969,
      "grad_norm": 6.529158592224121,
      "learning_rate": 1.8013587587298653e-07,
      "loss": 0.0666,
      "step": 17722
    },
    {
      "epoch": 0.9152078492124968,
      "grad_norm": 8.977866172790527,
      "learning_rate": 1.7991799004886067e-07,
      "loss": 0.1261,
      "step": 17723
    },
    {
      "epoch": 0.9152594887683966,
      "grad_norm": 2.1174349784851074,
      "learning_rate": 1.797002336635817e-07,
      "loss": 0.0331,
      "step": 17724
    },
    {
      "epoch": 0.9153111283242964,
      "grad_norm": 10.433281898498535,
      "learning_rate": 1.7948260672299767e-07,
      "loss": 0.0459,
      "step": 17725
    },
    {
      "epoch": 0.9153627678801962,
      "grad_norm": 5.1244215965271,
      "learning_rate": 1.7926510923295337e-07,
      "loss": 0.0472,
      "step": 17726
    },
    {
      "epoch": 0.915414407436096,
      "grad_norm": 19.05767059326172,
      "learning_rate": 1.7904774119928858e-07,
      "loss": 0.2525,
      "step": 17727
    },
    {
      "epoch": 0.9154660469919959,
      "grad_norm": 11.507705688476562,
      "learning_rate": 1.788305026278414e-07,
      "loss": 0.0834,
      "step": 17728
    },
    {
      "epoch": 0.9155176865478957,
      "grad_norm": 4.4177680015563965,
      "learning_rate": 1.7861339352444551e-07,
      "loss": 0.0603,
      "step": 17729
    },
    {
      "epoch": 0.9155693261037955,
      "grad_norm": 1.9468730688095093,
      "learning_rate": 1.7839641389493011e-07,
      "loss": 0.0434,
      "step": 17730
    },
    {
      "epoch": 0.9156209656596953,
      "grad_norm": 1.7899843454360962,
      "learning_rate": 1.7817956374512334e-07,
      "loss": 0.0244,
      "step": 17731
    },
    {
      "epoch": 0.9156726052155951,
      "grad_norm": 2.5083062648773193,
      "learning_rate": 1.7796284308084776e-07,
      "loss": 0.0214,
      "step": 17732
    },
    {
      "epoch": 0.9157242447714949,
      "grad_norm": 5.633573532104492,
      "learning_rate": 1.7774625190792373e-07,
      "loss": 0.0373,
      "step": 17733
    },
    {
      "epoch": 0.9157758843273948,
      "grad_norm": 1.5960124731063843,
      "learning_rate": 1.7752979023216766e-07,
      "loss": 0.0272,
      "step": 17734
    },
    {
      "epoch": 0.9158275238832946,
      "grad_norm": 2.033029079437256,
      "learning_rate": 1.773134580593905e-07,
      "loss": 0.0266,
      "step": 17735
    },
    {
      "epoch": 0.9158791634391944,
      "grad_norm": 3.526606559753418,
      "learning_rate": 1.7709725539540423e-07,
      "loss": 0.0432,
      "step": 17736
    },
    {
      "epoch": 0.9159308029950942,
      "grad_norm": 2.051797389984131,
      "learning_rate": 1.7688118224601368e-07,
      "loss": 0.0393,
      "step": 17737
    },
    {
      "epoch": 0.915982442550994,
      "grad_norm": 10.419937133789062,
      "learning_rate": 1.7666523861702146e-07,
      "loss": 0.1285,
      "step": 17738
    },
    {
      "epoch": 0.9160340821068939,
      "grad_norm": 6.300732612609863,
      "learning_rate": 1.764494245142262e-07,
      "loss": 0.0483,
      "step": 17739
    },
    {
      "epoch": 0.9160857216627937,
      "grad_norm": 3.5813698768615723,
      "learning_rate": 1.7623373994342275e-07,
      "loss": 0.0496,
      "step": 17740
    },
    {
      "epoch": 0.9161373612186935,
      "grad_norm": 4.7960076332092285,
      "learning_rate": 1.7601818491040535e-07,
      "loss": 0.0403,
      "step": 17741
    },
    {
      "epoch": 0.9161890007745933,
      "grad_norm": 3.168712615966797,
      "learning_rate": 1.7580275942096047e-07,
      "loss": 0.0409,
      "step": 17742
    },
    {
      "epoch": 0.9162406403304931,
      "grad_norm": 3.945025682449341,
      "learning_rate": 1.7558746348087353e-07,
      "loss": 0.0538,
      "step": 17743
    },
    {
      "epoch": 0.916292279886393,
      "grad_norm": 2.4769763946533203,
      "learning_rate": 1.7537229709592707e-07,
      "loss": 0.0119,
      "step": 17744
    },
    {
      "epoch": 0.9163439194422928,
      "grad_norm": 2.8189504146575928,
      "learning_rate": 1.7515726027189816e-07,
      "loss": 0.033,
      "step": 17745
    },
    {
      "epoch": 0.9163955589981926,
      "grad_norm": 4.326710224151611,
      "learning_rate": 1.7494235301456165e-07,
      "loss": 0.0459,
      "step": 17746
    },
    {
      "epoch": 0.9164471985540924,
      "grad_norm": 2.4502828121185303,
      "learning_rate": 1.74727575329689e-07,
      "loss": 0.0376,
      "step": 17747
    },
    {
      "epoch": 0.9164988381099922,
      "grad_norm": 6.130019664764404,
      "learning_rate": 1.7451292722304836e-07,
      "loss": 0.0577,
      "step": 17748
    },
    {
      "epoch": 0.916550477665892,
      "grad_norm": 3.694627046585083,
      "learning_rate": 1.7429840870040182e-07,
      "loss": 0.0463,
      "step": 17749
    },
    {
      "epoch": 0.9166021172217919,
      "grad_norm": 3.011033058166504,
      "learning_rate": 1.74084019767512e-07,
      "loss": 0.0371,
      "step": 17750
    },
    {
      "epoch": 0.9166537567776917,
      "grad_norm": 4.746526718139648,
      "learning_rate": 1.7386976043013591e-07,
      "loss": 0.0668,
      "step": 17751
    },
    {
      "epoch": 0.9167053963335915,
      "grad_norm": 5.16208553314209,
      "learning_rate": 1.7365563069402736e-07,
      "loss": 0.0664,
      "step": 17752
    },
    {
      "epoch": 0.9167570358894913,
      "grad_norm": 3.9138293266296387,
      "learning_rate": 1.7344163056493612e-07,
      "loss": 0.0319,
      "step": 17753
    },
    {
      "epoch": 0.9168086754453911,
      "grad_norm": 3.0317225456237793,
      "learning_rate": 1.7322776004860764e-07,
      "loss": 0.034,
      "step": 17754
    },
    {
      "epoch": 0.916860315001291,
      "grad_norm": 1.6772267818450928,
      "learning_rate": 1.7301401915078786e-07,
      "loss": 0.0294,
      "step": 17755
    },
    {
      "epoch": 0.9169119545571908,
      "grad_norm": 3.519205093383789,
      "learning_rate": 1.7280040787721552e-07,
      "loss": 0.0508,
      "step": 17756
    },
    {
      "epoch": 0.9169635941130906,
      "grad_norm": 5.843351364135742,
      "learning_rate": 1.725869262336266e-07,
      "loss": 0.0954,
      "step": 17757
    },
    {
      "epoch": 0.9170152336689904,
      "grad_norm": 3.3268918991088867,
      "learning_rate": 1.7237357422575373e-07,
      "loss": 0.0399,
      "step": 17758
    },
    {
      "epoch": 0.9170668732248902,
      "grad_norm": 13.84549331665039,
      "learning_rate": 1.7216035185932734e-07,
      "loss": 0.114,
      "step": 17759
    },
    {
      "epoch": 0.91711851278079,
      "grad_norm": 2.152972936630249,
      "learning_rate": 1.7194725914007226e-07,
      "loss": 0.0299,
      "step": 17760
    },
    {
      "epoch": 0.9171701523366899,
      "grad_norm": 2.7508459091186523,
      "learning_rate": 1.7173429607371117e-07,
      "loss": 0.0408,
      "step": 17761
    },
    {
      "epoch": 0.9172217918925897,
      "grad_norm": 2.173459529876709,
      "learning_rate": 1.7152146266596338e-07,
      "loss": 0.0347,
      "step": 17762
    },
    {
      "epoch": 0.9172734314484895,
      "grad_norm": 2.8781328201293945,
      "learning_rate": 1.7130875892254428e-07,
      "loss": 0.0431,
      "step": 17763
    },
    {
      "epoch": 0.9173250710043893,
      "grad_norm": 2.0608367919921875,
      "learning_rate": 1.7109618484916545e-07,
      "loss": 0.0158,
      "step": 17764
    },
    {
      "epoch": 0.9173767105602891,
      "grad_norm": 3.2741544246673584,
      "learning_rate": 1.7088374045153566e-07,
      "loss": 0.0174,
      "step": 17765
    },
    {
      "epoch": 0.917428350116189,
      "grad_norm": 8.340556144714355,
      "learning_rate": 1.7067142573535977e-07,
      "loss": 0.0773,
      "step": 17766
    },
    {
      "epoch": 0.9174799896720888,
      "grad_norm": 3.020218849182129,
      "learning_rate": 1.704592407063399e-07,
      "loss": 0.0321,
      "step": 17767
    },
    {
      "epoch": 0.9175316292279886,
      "grad_norm": 2.5500195026397705,
      "learning_rate": 1.702471853701726e-07,
      "loss": 0.0219,
      "step": 17768
    },
    {
      "epoch": 0.9175832687838884,
      "grad_norm": 2.266798257827759,
      "learning_rate": 1.700352597325533e-07,
      "loss": 0.0391,
      "step": 17769
    },
    {
      "epoch": 0.9176349083397882,
      "grad_norm": 1.923902988433838,
      "learning_rate": 1.6982346379917414e-07,
      "loss": 0.019,
      "step": 17770
    },
    {
      "epoch": 0.917686547895688,
      "grad_norm": 1.514967441558838,
      "learning_rate": 1.696117975757211e-07,
      "loss": 0.0317,
      "step": 17771
    },
    {
      "epoch": 0.9177381874515879,
      "grad_norm": 1.4284995794296265,
      "learning_rate": 1.694002610678791e-07,
      "loss": 0.0191,
      "step": 17772
    },
    {
      "epoch": 0.9177898270074878,
      "grad_norm": 2.6849730014801025,
      "learning_rate": 1.6918885428132748e-07,
      "loss": 0.0153,
      "step": 17773
    },
    {
      "epoch": 0.9178414665633876,
      "grad_norm": 6.689449310302734,
      "learning_rate": 1.6897757722174613e-07,
      "loss": 0.0265,
      "step": 17774
    },
    {
      "epoch": 0.9178931061192874,
      "grad_norm": 4.922466278076172,
      "learning_rate": 1.6876642989480607e-07,
      "loss": 0.0827,
      "step": 17775
    },
    {
      "epoch": 0.9179447456751872,
      "grad_norm": 8.263707160949707,
      "learning_rate": 1.6855541230617832e-07,
      "loss": 0.1084,
      "step": 17776
    },
    {
      "epoch": 0.9179963852310871,
      "grad_norm": 4.737066268920898,
      "learning_rate": 1.6834452446153005e-07,
      "loss": 0.0228,
      "step": 17777
    },
    {
      "epoch": 0.9180480247869869,
      "grad_norm": 9.527359008789062,
      "learning_rate": 1.6813376636652335e-07,
      "loss": 0.0802,
      "step": 17778
    },
    {
      "epoch": 0.9180996643428867,
      "grad_norm": 2.765294313430786,
      "learning_rate": 1.6792313802681925e-07,
      "loss": 0.0179,
      "step": 17779
    },
    {
      "epoch": 0.9181513038987865,
      "grad_norm": 1.7525614500045776,
      "learning_rate": 1.677126394480727e-07,
      "loss": 0.0367,
      "step": 17780
    },
    {
      "epoch": 0.9182029434546863,
      "grad_norm": 3.358797550201416,
      "learning_rate": 1.6750227063593806e-07,
      "loss": 0.0295,
      "step": 17781
    },
    {
      "epoch": 0.9182545830105862,
      "grad_norm": 3.292994499206543,
      "learning_rate": 1.6729203159606245e-07,
      "loss": 0.0226,
      "step": 17782
    },
    {
      "epoch": 0.918306222566486,
      "grad_norm": 6.7205610275268555,
      "learning_rate": 1.6708192233409359e-07,
      "loss": 0.0358,
      "step": 17783
    },
    {
      "epoch": 0.9183578621223858,
      "grad_norm": 5.613820552825928,
      "learning_rate": 1.668719428556731e-07,
      "loss": 0.037,
      "step": 17784
    },
    {
      "epoch": 0.9184095016782856,
      "grad_norm": 8.500504493713379,
      "learning_rate": 1.6666209316643978e-07,
      "loss": 0.0615,
      "step": 17785
    },
    {
      "epoch": 0.9184611412341854,
      "grad_norm": 3.6257970333099365,
      "learning_rate": 1.6645237327202857e-07,
      "loss": 0.0497,
      "step": 17786
    },
    {
      "epoch": 0.9185127807900852,
      "grad_norm": 4.9650397300720215,
      "learning_rate": 1.6624278317807107e-07,
      "loss": 0.0689,
      "step": 17787
    },
    {
      "epoch": 0.9185644203459851,
      "grad_norm": 2.686657428741455,
      "learning_rate": 1.6603332289019612e-07,
      "loss": 0.0487,
      "step": 17788
    },
    {
      "epoch": 0.9186160599018849,
      "grad_norm": 3.430065631866455,
      "learning_rate": 1.658239924140298e-07,
      "loss": 0.028,
      "step": 17789
    },
    {
      "epoch": 0.9186676994577847,
      "grad_norm": 15.049951553344727,
      "learning_rate": 1.6561479175519147e-07,
      "loss": 0.106,
      "step": 17790
    },
    {
      "epoch": 0.9187193390136845,
      "grad_norm": 13.457694053649902,
      "learning_rate": 1.6540572091929995e-07,
      "loss": 0.1932,
      "step": 17791
    },
    {
      "epoch": 0.9187709785695843,
      "grad_norm": 2.549335241317749,
      "learning_rate": 1.651967799119697e-07,
      "loss": 0.0359,
      "step": 17792
    },
    {
      "epoch": 0.9188226181254842,
      "grad_norm": 3.8533480167388916,
      "learning_rate": 1.6498796873881119e-07,
      "loss": 0.0316,
      "step": 17793
    },
    {
      "epoch": 0.918874257681384,
      "grad_norm": 9.863030433654785,
      "learning_rate": 1.6477928740543214e-07,
      "loss": 0.1163,
      "step": 17794
    },
    {
      "epoch": 0.9189258972372838,
      "grad_norm": 6.073263645172119,
      "learning_rate": 1.64570735917437e-07,
      "loss": 0.084,
      "step": 17795
    },
    {
      "epoch": 0.9189775367931836,
      "grad_norm": 2.922111988067627,
      "learning_rate": 1.6436231428042516e-07,
      "loss": 0.0427,
      "step": 17796
    },
    {
      "epoch": 0.9190291763490834,
      "grad_norm": 3.167217254638672,
      "learning_rate": 1.6415402249999434e-07,
      "loss": 0.0474,
      "step": 17797
    },
    {
      "epoch": 0.9190808159049833,
      "grad_norm": 2.81896710395813,
      "learning_rate": 1.6394586058173789e-07,
      "loss": 0.0382,
      "step": 17798
    },
    {
      "epoch": 0.9191324554608831,
      "grad_norm": 7.994415760040283,
      "learning_rate": 1.6373782853124577e-07,
      "loss": 0.0539,
      "step": 17799
    },
    {
      "epoch": 0.9191840950167829,
      "grad_norm": 4.044332504272461,
      "learning_rate": 1.635299263541046e-07,
      "loss": 0.062,
      "step": 17800
    },
    {
      "epoch": 0.9192357345726827,
      "grad_norm": 2.271151065826416,
      "learning_rate": 1.6332215405589657e-07,
      "loss": 0.0446,
      "step": 17801
    },
    {
      "epoch": 0.9192873741285825,
      "grad_norm": 3.117391347885132,
      "learning_rate": 1.6311451164220226e-07,
      "loss": 0.0271,
      "step": 17802
    },
    {
      "epoch": 0.9193390136844823,
      "grad_norm": 2.413249969482422,
      "learning_rate": 1.629069991185983e-07,
      "loss": 0.0268,
      "step": 17803
    },
    {
      "epoch": 0.9193906532403822,
      "grad_norm": 3.328887462615967,
      "learning_rate": 1.6269961649065525e-07,
      "loss": 0.0469,
      "step": 17804
    },
    {
      "epoch": 0.919442292796282,
      "grad_norm": 5.377439022064209,
      "learning_rate": 1.624923637639436e-07,
      "loss": 0.0576,
      "step": 17805
    },
    {
      "epoch": 0.9194939323521818,
      "grad_norm": 3.4792566299438477,
      "learning_rate": 1.6228524094402788e-07,
      "loss": 0.0379,
      "step": 17806
    },
    {
      "epoch": 0.9195455719080816,
      "grad_norm": 3.9051403999328613,
      "learning_rate": 1.6207824803647187e-07,
      "loss": 0.0573,
      "step": 17807
    },
    {
      "epoch": 0.9195972114639814,
      "grad_norm": 1.8848817348480225,
      "learning_rate": 1.6187138504683285e-07,
      "loss": 0.0276,
      "step": 17808
    },
    {
      "epoch": 0.9196488510198813,
      "grad_norm": 2.5127604007720947,
      "learning_rate": 1.616646519806664e-07,
      "loss": 0.0494,
      "step": 17809
    },
    {
      "epoch": 0.9197004905757811,
      "grad_norm": 12.901549339294434,
      "learning_rate": 1.6145804884352356e-07,
      "loss": 0.0489,
      "step": 17810
    },
    {
      "epoch": 0.9197521301316809,
      "grad_norm": 15.16250991821289,
      "learning_rate": 1.6125157564095329e-07,
      "loss": 0.1285,
      "step": 17811
    },
    {
      "epoch": 0.9198037696875807,
      "grad_norm": 20.67245101928711,
      "learning_rate": 1.6104523237849944e-07,
      "loss": 0.0928,
      "step": 17812
    },
    {
      "epoch": 0.9198554092434805,
      "grad_norm": 2.88846492767334,
      "learning_rate": 1.608390190617043e-07,
      "loss": 0.0306,
      "step": 17813
    },
    {
      "epoch": 0.9199070487993803,
      "grad_norm": 2.4039690494537354,
      "learning_rate": 1.606329356961045e-07,
      "loss": 0.0297,
      "step": 17814
    },
    {
      "epoch": 0.9199586883552802,
      "grad_norm": 5.929817199707031,
      "learning_rate": 1.6042698228723398e-07,
      "loss": 0.0173,
      "step": 17815
    },
    {
      "epoch": 0.92001032791118,
      "grad_norm": 14.92834758758545,
      "learning_rate": 1.6022115884062438e-07,
      "loss": 0.1337,
      "step": 17816
    },
    {
      "epoch": 0.9200619674670798,
      "grad_norm": 2.1207995414733887,
      "learning_rate": 1.6001546536180246e-07,
      "loss": 0.0246,
      "step": 17817
    },
    {
      "epoch": 0.9201136070229796,
      "grad_norm": 2.409075975418091,
      "learning_rate": 1.598099018562932e-07,
      "loss": 0.0331,
      "step": 17818
    },
    {
      "epoch": 0.9201652465788794,
      "grad_norm": 19.63414764404297,
      "learning_rate": 1.596044683296144e-07,
      "loss": 0.152,
      "step": 17819
    },
    {
      "epoch": 0.9202168861347793,
      "grad_norm": 2.5733578205108643,
      "learning_rate": 1.5939916478728335e-07,
      "loss": 0.0388,
      "step": 17820
    },
    {
      "epoch": 0.9202685256906791,
      "grad_norm": 11.132497787475586,
      "learning_rate": 1.5919399123481448e-07,
      "loss": 0.1809,
      "step": 17821
    },
    {
      "epoch": 0.9203201652465789,
      "grad_norm": 9.282556533813477,
      "learning_rate": 1.5898894767771788e-07,
      "loss": 0.071,
      "step": 17822
    },
    {
      "epoch": 0.9203718048024787,
      "grad_norm": 4.390185356140137,
      "learning_rate": 1.58784034121498e-07,
      "loss": 0.0366,
      "step": 17823
    },
    {
      "epoch": 0.9204234443583785,
      "grad_norm": 2.430985689163208,
      "learning_rate": 1.585792505716588e-07,
      "loss": 0.0167,
      "step": 17824
    },
    {
      "epoch": 0.9204750839142783,
      "grad_norm": 2.011925458908081,
      "learning_rate": 1.5837459703369916e-07,
      "loss": 0.0394,
      "step": 17825
    },
    {
      "epoch": 0.9205267234701782,
      "grad_norm": 3.493436336517334,
      "learning_rate": 1.5817007351311476e-07,
      "loss": 0.0246,
      "step": 17826
    },
    {
      "epoch": 0.920578363026078,
      "grad_norm": 2.244807243347168,
      "learning_rate": 1.5796568001539835e-07,
      "loss": 0.0224,
      "step": 17827
    },
    {
      "epoch": 0.9206300025819778,
      "grad_norm": 3.0558149814605713,
      "learning_rate": 1.577614165460384e-07,
      "loss": 0.0468,
      "step": 17828
    },
    {
      "epoch": 0.9206816421378776,
      "grad_norm": 2.7795615196228027,
      "learning_rate": 1.5755728311052044e-07,
      "loss": 0.0549,
      "step": 17829
    },
    {
      "epoch": 0.9207332816937774,
      "grad_norm": 13.620710372924805,
      "learning_rate": 1.5735327971432623e-07,
      "loss": 0.1142,
      "step": 17830
    },
    {
      "epoch": 0.9207849212496773,
      "grad_norm": 37.690921783447266,
      "learning_rate": 1.5714940636293362e-07,
      "loss": 0.1626,
      "step": 17831
    },
    {
      "epoch": 0.9208365608055771,
      "grad_norm": 3.6386351585388184,
      "learning_rate": 1.5694566306181769e-07,
      "loss": 0.0567,
      "step": 17832
    },
    {
      "epoch": 0.9208882003614769,
      "grad_norm": 3.1763179302215576,
      "learning_rate": 1.567420498164507e-07,
      "loss": 0.0286,
      "step": 17833
    },
    {
      "epoch": 0.9209398399173767,
      "grad_norm": 3.2150886058807373,
      "learning_rate": 1.5653856663229883e-07,
      "loss": 0.0712,
      "step": 17834
    },
    {
      "epoch": 0.9209914794732765,
      "grad_norm": 8.534891128540039,
      "learning_rate": 1.5633521351482772e-07,
      "loss": 0.0977,
      "step": 17835
    },
    {
      "epoch": 0.9210431190291763,
      "grad_norm": 2.644369602203369,
      "learning_rate": 1.56131990469498e-07,
      "loss": 0.0309,
      "step": 17836
    },
    {
      "epoch": 0.9210947585850762,
      "grad_norm": 4.315389633178711,
      "learning_rate": 1.5592889750176698e-07,
      "loss": 0.0802,
      "step": 17837
    },
    {
      "epoch": 0.921146398140976,
      "grad_norm": 3.528303861618042,
      "learning_rate": 1.5572593461708807e-07,
      "loss": 0.0529,
      "step": 17838
    },
    {
      "epoch": 0.9211980376968758,
      "grad_norm": 3.4201009273529053,
      "learning_rate": 1.5552310182091136e-07,
      "loss": 0.0834,
      "step": 17839
    },
    {
      "epoch": 0.9212496772527756,
      "grad_norm": 2.456158399581909,
      "learning_rate": 1.5532039911868523e-07,
      "loss": 0.0404,
      "step": 17840
    },
    {
      "epoch": 0.9213013168086754,
      "grad_norm": 1.24663507938385,
      "learning_rate": 1.5511782651585204e-07,
      "loss": 0.0096,
      "step": 17841
    },
    {
      "epoch": 0.9213529563645753,
      "grad_norm": 2.3967809677124023,
      "learning_rate": 1.5491538401785132e-07,
      "loss": 0.0322,
      "step": 17842
    },
    {
      "epoch": 0.9214045959204751,
      "grad_norm": 18.62125587463379,
      "learning_rate": 1.547130716301204e-07,
      "loss": 0.1445,
      "step": 17843
    },
    {
      "epoch": 0.9214562354763749,
      "grad_norm": 6.717881202697754,
      "learning_rate": 1.5451088935809212e-07,
      "loss": 0.0398,
      "step": 17844
    },
    {
      "epoch": 0.9215078750322747,
      "grad_norm": 10.93815803527832,
      "learning_rate": 1.5430883720719493e-07,
      "loss": 0.0484,
      "step": 17845
    },
    {
      "epoch": 0.9215595145881745,
      "grad_norm": 3.1920838356018066,
      "learning_rate": 1.5410691518285615e-07,
      "loss": 0.0345,
      "step": 17846
    },
    {
      "epoch": 0.9216111541440744,
      "grad_norm": 6.260924816131592,
      "learning_rate": 1.5390512329049701e-07,
      "loss": 0.0659,
      "step": 17847
    },
    {
      "epoch": 0.9216627936999742,
      "grad_norm": 5.509549140930176,
      "learning_rate": 1.5370346153553705e-07,
      "loss": 0.0692,
      "step": 17848
    },
    {
      "epoch": 0.921714433255874,
      "grad_norm": 3.6976940631866455,
      "learning_rate": 1.535019299233914e-07,
      "loss": 0.0803,
      "step": 17849
    },
    {
      "epoch": 0.9217660728117738,
      "grad_norm": 2.2476918697357178,
      "learning_rate": 1.5330052845947184e-07,
      "loss": 0.0464,
      "step": 17850
    },
    {
      "epoch": 0.9218177123676736,
      "grad_norm": 7.630711555480957,
      "learning_rate": 1.530992571491885e-07,
      "loss": 0.0887,
      "step": 17851
    },
    {
      "epoch": 0.9218693519235734,
      "grad_norm": 11.719878196716309,
      "learning_rate": 1.528981159979437e-07,
      "loss": 0.0633,
      "step": 17852
    },
    {
      "epoch": 0.9219209914794733,
      "grad_norm": 3.453334093093872,
      "learning_rate": 1.5269710501113977e-07,
      "loss": 0.0319,
      "step": 17853
    },
    {
      "epoch": 0.9219726310353731,
      "grad_norm": 3.579457998275757,
      "learning_rate": 1.524962241941763e-07,
      "loss": 0.0433,
      "step": 17854
    },
    {
      "epoch": 0.9220242705912729,
      "grad_norm": 1.5867555141448975,
      "learning_rate": 1.5229547355244621e-07,
      "loss": 0.0189,
      "step": 17855
    },
    {
      "epoch": 0.9220759101471727,
      "grad_norm": 5.081505298614502,
      "learning_rate": 1.5209485309134076e-07,
      "loss": 0.1054,
      "step": 17856
    },
    {
      "epoch": 0.9221275497030725,
      "grad_norm": 5.967776775360107,
      "learning_rate": 1.5189436281624726e-07,
      "loss": 0.0581,
      "step": 17857
    },
    {
      "epoch": 0.9221791892589724,
      "grad_norm": 2.8716635704040527,
      "learning_rate": 1.516940027325503e-07,
      "loss": 0.039,
      "step": 17858
    },
    {
      "epoch": 0.9222308288148722,
      "grad_norm": 3.78412127494812,
      "learning_rate": 1.514937728456295e-07,
      "loss": 0.0467,
      "step": 17859
    },
    {
      "epoch": 0.922282468370772,
      "grad_norm": 32.50461196899414,
      "learning_rate": 1.5129367316086218e-07,
      "loss": 0.081,
      "step": 17860
    },
    {
      "epoch": 0.9223341079266718,
      "grad_norm": 5.214203357696533,
      "learning_rate": 1.5109370368362242e-07,
      "loss": 0.074,
      "step": 17861
    },
    {
      "epoch": 0.9223857474825716,
      "grad_norm": 16.60062599182129,
      "learning_rate": 1.5089386441927978e-07,
      "loss": 0.0939,
      "step": 17862
    },
    {
      "epoch": 0.9224373870384714,
      "grad_norm": 3.0362002849578857,
      "learning_rate": 1.5069415537320054e-07,
      "loss": 0.0323,
      "step": 17863
    },
    {
      "epoch": 0.9224890265943713,
      "grad_norm": 2.3142566680908203,
      "learning_rate": 1.5049457655074818e-07,
      "loss": 0.017,
      "step": 17864
    },
    {
      "epoch": 0.9225406661502711,
      "grad_norm": 16.941171646118164,
      "learning_rate": 1.502951279572812e-07,
      "loss": 0.1611,
      "step": 17865
    },
    {
      "epoch": 0.9225923057061709,
      "grad_norm": 4.053772926330566,
      "learning_rate": 1.5009580959815695e-07,
      "loss": 0.0241,
      "step": 17866
    },
    {
      "epoch": 0.9226439452620707,
      "grad_norm": 2.95369553565979,
      "learning_rate": 1.498966214787273e-07,
      "loss": 0.0592,
      "step": 17867
    },
    {
      "epoch": 0.9226955848179705,
      "grad_norm": 8.717344284057617,
      "learning_rate": 1.4969756360434128e-07,
      "loss": 0.1104,
      "step": 17868
    },
    {
      "epoch": 0.9227472243738704,
      "grad_norm": 2.6080596446990967,
      "learning_rate": 1.4949863598034464e-07,
      "loss": 0.0348,
      "step": 17869
    },
    {
      "epoch": 0.9227988639297702,
      "grad_norm": 4.216770172119141,
      "learning_rate": 1.492998386120792e-07,
      "loss": 0.058,
      "step": 17870
    },
    {
      "epoch": 0.92285050348567,
      "grad_norm": 3.0170328617095947,
      "learning_rate": 1.491011715048829e-07,
      "loss": 0.0273,
      "step": 17871
    },
    {
      "epoch": 0.9229021430415698,
      "grad_norm": 19.727556228637695,
      "learning_rate": 1.4890263466409095e-07,
      "loss": 0.0948,
      "step": 17872
    },
    {
      "epoch": 0.9229537825974696,
      "grad_norm": 5.391790866851807,
      "learning_rate": 1.4870422809503626e-07,
      "loss": 0.0435,
      "step": 17873
    },
    {
      "epoch": 0.9230054221533694,
      "grad_norm": 5.744159698486328,
      "learning_rate": 1.4850595180304573e-07,
      "loss": 0.0736,
      "step": 17874
    },
    {
      "epoch": 0.9230570617092693,
      "grad_norm": 2.581146240234375,
      "learning_rate": 1.4830780579344338e-07,
      "loss": 0.0212,
      "step": 17875
    },
    {
      "epoch": 0.9231087012651691,
      "grad_norm": 17.150300979614258,
      "learning_rate": 1.4810979007155112e-07,
      "loss": 0.1156,
      "step": 17876
    },
    {
      "epoch": 0.9231603408210689,
      "grad_norm": 2.1997973918914795,
      "learning_rate": 1.4791190464268633e-07,
      "loss": 0.0207,
      "step": 17877
    },
    {
      "epoch": 0.9232119803769687,
      "grad_norm": 7.171018123626709,
      "learning_rate": 1.4771414951216255e-07,
      "loss": 0.0613,
      "step": 17878
    },
    {
      "epoch": 0.9232636199328685,
      "grad_norm": 4.968437671661377,
      "learning_rate": 1.4751652468529108e-07,
      "loss": 0.0699,
      "step": 17879
    },
    {
      "epoch": 0.9233152594887684,
      "grad_norm": 19.443838119506836,
      "learning_rate": 1.4731903016737826e-07,
      "loss": 0.147,
      "step": 17880
    },
    {
      "epoch": 0.9233668990446682,
      "grad_norm": 6.639900207519531,
      "learning_rate": 1.4712166596372813e-07,
      "loss": 0.1028,
      "step": 17881
    },
    {
      "epoch": 0.923418538600568,
      "grad_norm": 2.5602943897247314,
      "learning_rate": 1.4692443207964035e-07,
      "loss": 0.0406,
      "step": 17882
    },
    {
      "epoch": 0.9234701781564678,
      "grad_norm": 4.079664707183838,
      "learning_rate": 1.4672732852041181e-07,
      "loss": 0.0525,
      "step": 17883
    },
    {
      "epoch": 0.9235218177123676,
      "grad_norm": 3.6014184951782227,
      "learning_rate": 1.4653035529133608e-07,
      "loss": 0.049,
      "step": 17884
    },
    {
      "epoch": 0.9235734572682675,
      "grad_norm": 2.1073663234710693,
      "learning_rate": 1.4633351239770055e-07,
      "loss": 0.0247,
      "step": 17885
    },
    {
      "epoch": 0.9236250968241673,
      "grad_norm": 5.8342695236206055,
      "learning_rate": 1.4613679984479323e-07,
      "loss": 0.0402,
      "step": 17886
    },
    {
      "epoch": 0.9236767363800671,
      "grad_norm": 1.7189818620681763,
      "learning_rate": 1.45940217637896e-07,
      "loss": 0.0146,
      "step": 17887
    },
    {
      "epoch": 0.9237283759359669,
      "grad_norm": 1.673832654953003,
      "learning_rate": 1.4574376578228856e-07,
      "loss": 0.016,
      "step": 17888
    },
    {
      "epoch": 0.9237800154918667,
      "grad_norm": 8.2725830078125,
      "learning_rate": 1.455474442832455e-07,
      "loss": 0.0646,
      "step": 17889
    },
    {
      "epoch": 0.9238316550477665,
      "grad_norm": 5.719880104064941,
      "learning_rate": 1.453512531460388e-07,
      "loss": 0.0757,
      "step": 17890
    },
    {
      "epoch": 0.9238832946036664,
      "grad_norm": 2.040292501449585,
      "learning_rate": 1.4515519237593756e-07,
      "loss": 0.0125,
      "step": 17891
    },
    {
      "epoch": 0.9239349341595662,
      "grad_norm": 14.794493675231934,
      "learning_rate": 1.4495926197820697e-07,
      "loss": 0.1389,
      "step": 17892
    },
    {
      "epoch": 0.923986573715466,
      "grad_norm": 1.5165021419525146,
      "learning_rate": 1.4476346195810786e-07,
      "loss": 0.0266,
      "step": 17893
    },
    {
      "epoch": 0.9240382132713658,
      "grad_norm": 4.825996398925781,
      "learning_rate": 1.445677923208988e-07,
      "loss": 0.035,
      "step": 17894
    },
    {
      "epoch": 0.9240898528272656,
      "grad_norm": 10.663732528686523,
      "learning_rate": 1.4437225307183444e-07,
      "loss": 0.1035,
      "step": 17895
    },
    {
      "epoch": 0.9241414923831655,
      "grad_norm": 2.2540171146392822,
      "learning_rate": 1.4417684421616506e-07,
      "loss": 0.0364,
      "step": 17896
    },
    {
      "epoch": 0.9241931319390653,
      "grad_norm": 10.50268268585205,
      "learning_rate": 1.4398156575913923e-07,
      "loss": 0.0926,
      "step": 17897
    },
    {
      "epoch": 0.9242447714949651,
      "grad_norm": 3.5354437828063965,
      "learning_rate": 1.43786417706e-07,
      "loss": 0.0294,
      "step": 17898
    },
    {
      "epoch": 0.9242964110508649,
      "grad_norm": 6.562088489532471,
      "learning_rate": 1.4359140006198812e-07,
      "loss": 0.0706,
      "step": 17899
    },
    {
      "epoch": 0.9243480506067648,
      "grad_norm": 23.137737274169922,
      "learning_rate": 1.4339651283234113e-07,
      "loss": 0.0905,
      "step": 17900
    },
    {
      "epoch": 0.9243996901626647,
      "grad_norm": 1.2144548892974854,
      "learning_rate": 1.4320175602229203e-07,
      "loss": 0.0217,
      "step": 17901
    },
    {
      "epoch": 0.9244513297185645,
      "grad_norm": 3.3620941638946533,
      "learning_rate": 1.4300712963707163e-07,
      "loss": 0.0553,
      "step": 17902
    },
    {
      "epoch": 0.9245029692744643,
      "grad_norm": 2.6236228942871094,
      "learning_rate": 1.4281263368190524e-07,
      "loss": 0.0198,
      "step": 17903
    },
    {
      "epoch": 0.9245546088303641,
      "grad_norm": 12.539839744567871,
      "learning_rate": 1.4261826816201584e-07,
      "loss": 0.1642,
      "step": 17904
    },
    {
      "epoch": 0.9246062483862639,
      "grad_norm": 5.91408634185791,
      "learning_rate": 1.4242403308262377e-07,
      "loss": 0.0318,
      "step": 17905
    },
    {
      "epoch": 0.9246578879421637,
      "grad_norm": 2.7498867511749268,
      "learning_rate": 1.4222992844894589e-07,
      "loss": 0.0623,
      "step": 17906
    },
    {
      "epoch": 0.9247095274980636,
      "grad_norm": 2.660398483276367,
      "learning_rate": 1.4203595426619255e-07,
      "loss": 0.0623,
      "step": 17907
    },
    {
      "epoch": 0.9247611670539634,
      "grad_norm": 7.863595485687256,
      "learning_rate": 1.4184211053957397e-07,
      "loss": 0.0526,
      "step": 17908
    },
    {
      "epoch": 0.9248128066098632,
      "grad_norm": 9.380439758300781,
      "learning_rate": 1.4164839727429546e-07,
      "loss": 0.0862,
      "step": 17909
    },
    {
      "epoch": 0.924864446165763,
      "grad_norm": 5.85766077041626,
      "learning_rate": 1.4145481447555953e-07,
      "loss": 0.072,
      "step": 17910
    },
    {
      "epoch": 0.9249160857216628,
      "grad_norm": 7.0650315284729,
      "learning_rate": 1.4126136214856368e-07,
      "loss": 0.0368,
      "step": 17911
    },
    {
      "epoch": 0.9249677252775627,
      "grad_norm": 2.4545369148254395,
      "learning_rate": 1.4106804029850374e-07,
      "loss": 0.0423,
      "step": 17912
    },
    {
      "epoch": 0.9250193648334625,
      "grad_norm": 3.787891149520874,
      "learning_rate": 1.4087484893057057e-07,
      "loss": 0.0377,
      "step": 17913
    },
    {
      "epoch": 0.9250710043893623,
      "grad_norm": 3.029254674911499,
      "learning_rate": 1.4068178804995225e-07,
      "loss": 0.0577,
      "step": 17914
    },
    {
      "epoch": 0.9251226439452621,
      "grad_norm": 4.815049171447754,
      "learning_rate": 1.404888576618335e-07,
      "loss": 0.0549,
      "step": 17915
    },
    {
      "epoch": 0.9251742835011619,
      "grad_norm": 5.137897491455078,
      "learning_rate": 1.402960577713952e-07,
      "loss": 0.0579,
      "step": 17916
    },
    {
      "epoch": 0.9252259230570617,
      "grad_norm": 11.3931245803833,
      "learning_rate": 1.4010338838381543e-07,
      "loss": 0.0983,
      "step": 17917
    },
    {
      "epoch": 0.9252775626129616,
      "grad_norm": 5.121138572692871,
      "learning_rate": 1.3991084950426613e-07,
      "loss": 0.045,
      "step": 17918
    },
    {
      "epoch": 0.9253292021688614,
      "grad_norm": 1.985507845878601,
      "learning_rate": 1.3971844113792043e-07,
      "loss": 0.0195,
      "step": 17919
    },
    {
      "epoch": 0.9253808417247612,
      "grad_norm": 3.562286138534546,
      "learning_rate": 1.3952616328994362e-07,
      "loss": 0.0478,
      "step": 17920
    },
    {
      "epoch": 0.925432481280661,
      "grad_norm": 2.4226014614105225,
      "learning_rate": 1.393340159654999e-07,
      "loss": 0.0552,
      "step": 17921
    },
    {
      "epoch": 0.9254841208365608,
      "grad_norm": 4.023039817810059,
      "learning_rate": 1.391419991697479e-07,
      "loss": 0.0556,
      "step": 17922
    },
    {
      "epoch": 0.9255357603924607,
      "grad_norm": 12.620320320129395,
      "learning_rate": 1.3895011290784578e-07,
      "loss": 0.1135,
      "step": 17923
    },
    {
      "epoch": 0.9255873999483605,
      "grad_norm": 16.448938369750977,
      "learning_rate": 1.3875835718494546e-07,
      "loss": 0.1376,
      "step": 17924
    },
    {
      "epoch": 0.9256390395042603,
      "grad_norm": 13.805315971374512,
      "learning_rate": 1.3856673200619618e-07,
      "loss": 0.2028,
      "step": 17925
    },
    {
      "epoch": 0.9256906790601601,
      "grad_norm": 2.6917428970336914,
      "learning_rate": 1.383752373767444e-07,
      "loss": 0.0462,
      "step": 17926
    },
    {
      "epoch": 0.9257423186160599,
      "grad_norm": 4.551211357116699,
      "learning_rate": 1.3818387330173267e-07,
      "loss": 0.0626,
      "step": 17927
    },
    {
      "epoch": 0.9257939581719598,
      "grad_norm": 2.6701011657714844,
      "learning_rate": 1.3799263978629962e-07,
      "loss": 0.0251,
      "step": 17928
    },
    {
      "epoch": 0.9258455977278596,
      "grad_norm": 1.4518556594848633,
      "learning_rate": 1.3780153683558063e-07,
      "loss": 0.0131,
      "step": 17929
    },
    {
      "epoch": 0.9258972372837594,
      "grad_norm": 5.885551452636719,
      "learning_rate": 1.3761056445470712e-07,
      "loss": 0.0591,
      "step": 17930
    },
    {
      "epoch": 0.9259488768396592,
      "grad_norm": 2.4256207942962646,
      "learning_rate": 1.374197226488083e-07,
      "loss": 0.0211,
      "step": 17931
    },
    {
      "epoch": 0.926000516395559,
      "grad_norm": 2.5177254676818848,
      "learning_rate": 1.3722901142300905e-07,
      "loss": 0.036,
      "step": 17932
    },
    {
      "epoch": 0.9260521559514588,
      "grad_norm": 2.7536332607269287,
      "learning_rate": 1.3703843078242962e-07,
      "loss": 0.0227,
      "step": 17933
    },
    {
      "epoch": 0.9261037955073587,
      "grad_norm": 10.923823356628418,
      "learning_rate": 1.368479807321893e-07,
      "loss": 0.0749,
      "step": 17934
    },
    {
      "epoch": 0.9261554350632585,
      "grad_norm": 2.403118371963501,
      "learning_rate": 1.3665766127740177e-07,
      "loss": 0.0287,
      "step": 17935
    },
    {
      "epoch": 0.9262070746191583,
      "grad_norm": 2.5804643630981445,
      "learning_rate": 1.3646747242317792e-07,
      "loss": 0.0315,
      "step": 17936
    },
    {
      "epoch": 0.9262587141750581,
      "grad_norm": 4.519477367401123,
      "learning_rate": 1.3627741417462425e-07,
      "loss": 0.0656,
      "step": 17937
    },
    {
      "epoch": 0.9263103537309579,
      "grad_norm": 4.215553283691406,
      "learning_rate": 1.360874865368461e-07,
      "loss": 0.0569,
      "step": 17938
    },
    {
      "epoch": 0.9263619932868578,
      "grad_norm": 11.869786262512207,
      "learning_rate": 1.3589768951494386e-07,
      "loss": 0.0859,
      "step": 17939
    },
    {
      "epoch": 0.9264136328427576,
      "grad_norm": 4.153776168823242,
      "learning_rate": 1.3570802311401288e-07,
      "loss": 0.0669,
      "step": 17940
    },
    {
      "epoch": 0.9264652723986574,
      "grad_norm": 3.7379086017608643,
      "learning_rate": 1.3551848733914742e-07,
      "loss": 0.0292,
      "step": 17941
    },
    {
      "epoch": 0.9265169119545572,
      "grad_norm": 6.6488871574401855,
      "learning_rate": 1.3532908219543672e-07,
      "loss": 0.0845,
      "step": 17942
    },
    {
      "epoch": 0.926568551510457,
      "grad_norm": 6.019897937774658,
      "learning_rate": 1.3513980768796785e-07,
      "loss": 0.0492,
      "step": 17943
    },
    {
      "epoch": 0.9266201910663568,
      "grad_norm": 33.529090881347656,
      "learning_rate": 1.3495066382182342e-07,
      "loss": 0.1529,
      "step": 17944
    },
    {
      "epoch": 0.9266718306222567,
      "grad_norm": 3.0086545944213867,
      "learning_rate": 1.3476165060208214e-07,
      "loss": 0.0155,
      "step": 17945
    },
    {
      "epoch": 0.9267234701781565,
      "grad_norm": 7.737370491027832,
      "learning_rate": 1.345727680338199e-07,
      "loss": 0.0233,
      "step": 17946
    },
    {
      "epoch": 0.9267751097340563,
      "grad_norm": 4.521970748901367,
      "learning_rate": 1.3438401612210938e-07,
      "loss": 0.0263,
      "step": 17947
    },
    {
      "epoch": 0.9268267492899561,
      "grad_norm": 11.217955589294434,
      "learning_rate": 1.3419539487201927e-07,
      "loss": 0.0546,
      "step": 17948
    },
    {
      "epoch": 0.9268783888458559,
      "grad_norm": 1.6347566843032837,
      "learning_rate": 1.3400690428861495e-07,
      "loss": 0.0348,
      "step": 17949
    },
    {
      "epoch": 0.9269300284017558,
      "grad_norm": 5.5612921714782715,
      "learning_rate": 1.3381854437695852e-07,
      "loss": 0.0549,
      "step": 17950
    },
    {
      "epoch": 0.9269816679576556,
      "grad_norm": 1.9876080751419067,
      "learning_rate": 1.3363031514210644e-07,
      "loss": 0.022,
      "step": 17951
    },
    {
      "epoch": 0.9270333075135554,
      "grad_norm": 2.0800535678863525,
      "learning_rate": 1.3344221658911472e-07,
      "loss": 0.0284,
      "step": 17952
    },
    {
      "epoch": 0.9270849470694552,
      "grad_norm": 3.197984457015991,
      "learning_rate": 1.3325424872303538e-07,
      "loss": 0.0355,
      "step": 17953
    },
    {
      "epoch": 0.927136586625355,
      "grad_norm": 1.6183199882507324,
      "learning_rate": 1.3306641154891498e-07,
      "loss": 0.0274,
      "step": 17954
    },
    {
      "epoch": 0.9271882261812548,
      "grad_norm": 4.90952205657959,
      "learning_rate": 1.3287870507179724e-07,
      "loss": 0.0916,
      "step": 17955
    },
    {
      "epoch": 0.9272398657371547,
      "grad_norm": 2.316289186477661,
      "learning_rate": 1.3269112929672369e-07,
      "loss": 0.0296,
      "step": 17956
    },
    {
      "epoch": 0.9272915052930545,
      "grad_norm": 2.859694719314575,
      "learning_rate": 1.3250368422873195e-07,
      "loss": 0.0398,
      "step": 17957
    },
    {
      "epoch": 0.9273431448489543,
      "grad_norm": 4.278389930725098,
      "learning_rate": 1.323163698728547e-07,
      "loss": 0.0519,
      "step": 17958
    },
    {
      "epoch": 0.9273947844048541,
      "grad_norm": 3.5144965648651123,
      "learning_rate": 1.3212918623412286e-07,
      "loss": 0.0397,
      "step": 17959
    },
    {
      "epoch": 0.9274464239607539,
      "grad_norm": 10.30978012084961,
      "learning_rate": 1.3194213331756246e-07,
      "loss": 0.1344,
      "step": 17960
    },
    {
      "epoch": 0.9274980635166538,
      "grad_norm": 19.060813903808594,
      "learning_rate": 1.3175521112819667e-07,
      "loss": 0.0383,
      "step": 17961
    },
    {
      "epoch": 0.9275497030725536,
      "grad_norm": 2.213716745376587,
      "learning_rate": 1.3156841967104538e-07,
      "loss": 0.0303,
      "step": 17962
    },
    {
      "epoch": 0.9276013426284534,
      "grad_norm": 1.7292957305908203,
      "learning_rate": 1.3138175895112514e-07,
      "loss": 0.0203,
      "step": 17963
    },
    {
      "epoch": 0.9276529821843532,
      "grad_norm": 17.147470474243164,
      "learning_rate": 1.3119522897344805e-07,
      "loss": 0.0662,
      "step": 17964
    },
    {
      "epoch": 0.927704621740253,
      "grad_norm": 16.59575653076172,
      "learning_rate": 1.3100882974302232e-07,
      "loss": 0.073,
      "step": 17965
    },
    {
      "epoch": 0.9277562612961529,
      "grad_norm": 2.1243977546691895,
      "learning_rate": 1.3082256126485503e-07,
      "loss": 0.0314,
      "step": 17966
    },
    {
      "epoch": 0.9278079008520527,
      "grad_norm": 5.400331974029541,
      "learning_rate": 1.3063642354394778e-07,
      "loss": 0.0711,
      "step": 17967
    },
    {
      "epoch": 0.9278595404079525,
      "grad_norm": 1.8487789630889893,
      "learning_rate": 1.3045041658529932e-07,
      "loss": 0.025,
      "step": 17968
    },
    {
      "epoch": 0.9279111799638523,
      "grad_norm": 3.481858968734741,
      "learning_rate": 1.3026454039390347e-07,
      "loss": 0.04,
      "step": 17969
    },
    {
      "epoch": 0.9279628195197521,
      "grad_norm": 13.518381118774414,
      "learning_rate": 1.3007879497475284e-07,
      "loss": 0.1075,
      "step": 17970
    },
    {
      "epoch": 0.9280144590756519,
      "grad_norm": 7.920684337615967,
      "learning_rate": 1.2989318033283517e-07,
      "loss": 0.0598,
      "step": 17971
    },
    {
      "epoch": 0.9280660986315518,
      "grad_norm": 12.365325927734375,
      "learning_rate": 1.2970769647313587e-07,
      "loss": 0.0796,
      "step": 17972
    },
    {
      "epoch": 0.9281177381874516,
      "grad_norm": 1.5766596794128418,
      "learning_rate": 1.2952234340063484e-07,
      "loss": 0.0299,
      "step": 17973
    },
    {
      "epoch": 0.9281693777433514,
      "grad_norm": 9.939820289611816,
      "learning_rate": 1.2933712112030984e-07,
      "loss": 0.1428,
      "step": 17974
    },
    {
      "epoch": 0.9282210172992512,
      "grad_norm": 3.810154914855957,
      "learning_rate": 1.2915202963713457e-07,
      "loss": 0.0398,
      "step": 17975
    },
    {
      "epoch": 0.928272656855151,
      "grad_norm": 3.8199071884155273,
      "learning_rate": 1.2896706895607957e-07,
      "loss": 0.0544,
      "step": 17976
    },
    {
      "epoch": 0.9283242964110509,
      "grad_norm": 2.238377094268799,
      "learning_rate": 1.2878223908211253e-07,
      "loss": 0.0391,
      "step": 17977
    },
    {
      "epoch": 0.9283759359669507,
      "grad_norm": 4.6992292404174805,
      "learning_rate": 1.285975400201961e-07,
      "loss": 0.0481,
      "step": 17978
    },
    {
      "epoch": 0.9284275755228505,
      "grad_norm": 4.617436408996582,
      "learning_rate": 1.284129717752902e-07,
      "loss": 0.0281,
      "step": 17979
    },
    {
      "epoch": 0.9284792150787503,
      "grad_norm": 15.144369125366211,
      "learning_rate": 1.2822853435235205e-07,
      "loss": 0.0655,
      "step": 17980
    },
    {
      "epoch": 0.9285308546346501,
      "grad_norm": 16.381980895996094,
      "learning_rate": 1.2804422775633373e-07,
      "loss": 0.0453,
      "step": 17981
    },
    {
      "epoch": 0.92858249419055,
      "grad_norm": 31.971403121948242,
      "learning_rate": 1.2786005199218465e-07,
      "loss": 0.1164,
      "step": 17982
    },
    {
      "epoch": 0.9286341337464498,
      "grad_norm": 5.529568195343018,
      "learning_rate": 1.2767600706485195e-07,
      "loss": 0.0221,
      "step": 17983
    },
    {
      "epoch": 0.9286857733023496,
      "grad_norm": 9.941832542419434,
      "learning_rate": 1.2749209297927557e-07,
      "loss": 0.1393,
      "step": 17984
    },
    {
      "epoch": 0.9287374128582494,
      "grad_norm": 17.60953712463379,
      "learning_rate": 1.2730830974039597e-07,
      "loss": 0.1331,
      "step": 17985
    },
    {
      "epoch": 0.9287890524141492,
      "grad_norm": 4.347851753234863,
      "learning_rate": 1.2712465735314926e-07,
      "loss": 0.0273,
      "step": 17986
    },
    {
      "epoch": 0.928840691970049,
      "grad_norm": 1.8367058038711548,
      "learning_rate": 1.2694113582246538e-07,
      "loss": 0.0372,
      "step": 17987
    },
    {
      "epoch": 0.9288923315259489,
      "grad_norm": 1.3388175964355469,
      "learning_rate": 1.2675774515327366e-07,
      "loss": 0.0166,
      "step": 17988
    },
    {
      "epoch": 0.9289439710818487,
      "grad_norm": 9.906472206115723,
      "learning_rate": 1.2657448535049855e-07,
      "loss": 0.1174,
      "step": 17989
    },
    {
      "epoch": 0.9289956106377485,
      "grad_norm": 3.4489145278930664,
      "learning_rate": 1.263913564190622e-07,
      "loss": 0.0446,
      "step": 17990
    },
    {
      "epoch": 0.9290472501936483,
      "grad_norm": 2.5023155212402344,
      "learning_rate": 1.2620835836388124e-07,
      "loss": 0.0302,
      "step": 17991
    },
    {
      "epoch": 0.9290988897495481,
      "grad_norm": 18.689783096313477,
      "learning_rate": 1.2602549118987006e-07,
      "loss": 0.0947,
      "step": 17992
    },
    {
      "epoch": 0.929150529305448,
      "grad_norm": 17.683767318725586,
      "learning_rate": 1.2584275490193975e-07,
      "loss": 0.1295,
      "step": 17993
    },
    {
      "epoch": 0.9292021688613478,
      "grad_norm": 2.8543314933776855,
      "learning_rate": 1.256601495049975e-07,
      "loss": 0.0245,
      "step": 17994
    },
    {
      "epoch": 0.9292538084172476,
      "grad_norm": 3.004509925842285,
      "learning_rate": 1.2547767500394714e-07,
      "loss": 0.0285,
      "step": 17995
    },
    {
      "epoch": 0.9293054479731474,
      "grad_norm": 11.118755340576172,
      "learning_rate": 1.2529533140368867e-07,
      "loss": 0.1205,
      "step": 17996
    },
    {
      "epoch": 0.9293570875290472,
      "grad_norm": 2.7012524604797363,
      "learning_rate": 1.251131187091187e-07,
      "loss": 0.0318,
      "step": 17997
    },
    {
      "epoch": 0.929408727084947,
      "grad_norm": 3.373518228530884,
      "learning_rate": 1.2493103692513e-07,
      "loss": 0.031,
      "step": 17998
    },
    {
      "epoch": 0.9294603666408469,
      "grad_norm": 4.384315490722656,
      "learning_rate": 1.2474908605661306e-07,
      "loss": 0.076,
      "step": 17999
    },
    {
      "epoch": 0.9295120061967467,
      "grad_norm": 3.7837750911712646,
      "learning_rate": 1.2456726610845348e-07,
      "loss": 0.0563,
      "step": 18000
    },
    {
      "epoch": 0.9295636457526465,
      "grad_norm": 4.621321678161621,
      "learning_rate": 1.2438557708553457e-07,
      "loss": 0.0434,
      "step": 18001
    },
    {
      "epoch": 0.9296152853085463,
      "grad_norm": 4.024881362915039,
      "learning_rate": 1.2420401899273405e-07,
      "loss": 0.0484,
      "step": 18002
    },
    {
      "epoch": 0.9296669248644461,
      "grad_norm": 3.2408783435821533,
      "learning_rate": 1.2402259183492805e-07,
      "loss": 0.0334,
      "step": 18003
    },
    {
      "epoch": 0.929718564420346,
      "grad_norm": 3.367175817489624,
      "learning_rate": 1.2384129561698877e-07,
      "loss": 0.0486,
      "step": 18004
    },
    {
      "epoch": 0.9297702039762458,
      "grad_norm": 19.54967498779297,
      "learning_rate": 1.2366013034378622e-07,
      "loss": 0.1129,
      "step": 18005
    },
    {
      "epoch": 0.9298218435321456,
      "grad_norm": 12.911552429199219,
      "learning_rate": 1.234790960201826e-07,
      "loss": 0.0642,
      "step": 18006
    },
    {
      "epoch": 0.9298734830880454,
      "grad_norm": 6.596431732177734,
      "learning_rate": 1.232981926510418e-07,
      "loss": 0.0848,
      "step": 18007
    },
    {
      "epoch": 0.9299251226439452,
      "grad_norm": 1.917428731918335,
      "learning_rate": 1.2311742024121997e-07,
      "loss": 0.0272,
      "step": 18008
    },
    {
      "epoch": 0.929976762199845,
      "grad_norm": 1.758300542831421,
      "learning_rate": 1.229367787955732e-07,
      "loss": 0.0364,
      "step": 18009
    },
    {
      "epoch": 0.9300284017557449,
      "grad_norm": 3.5900774002075195,
      "learning_rate": 1.2275626831895148e-07,
      "loss": 0.0451,
      "step": 18010
    },
    {
      "epoch": 0.9300800413116447,
      "grad_norm": 3.5779218673706055,
      "learning_rate": 1.2257588881620208e-07,
      "loss": 0.0418,
      "step": 18011
    },
    {
      "epoch": 0.9301316808675445,
      "grad_norm": 9.899713516235352,
      "learning_rate": 1.2239564029216998e-07,
      "loss": 0.0587,
      "step": 18012
    },
    {
      "epoch": 0.9301833204234443,
      "grad_norm": 15.192575454711914,
      "learning_rate": 1.2221552275169468e-07,
      "loss": 0.3089,
      "step": 18013
    },
    {
      "epoch": 0.9302349599793441,
      "grad_norm": 6.7862114906311035,
      "learning_rate": 1.2203553619961338e-07,
      "loss": 0.0591,
      "step": 18014
    },
    {
      "epoch": 0.930286599535244,
      "grad_norm": 34.388668060302734,
      "learning_rate": 1.2185568064075947e-07,
      "loss": 0.1514,
      "step": 18015
    },
    {
      "epoch": 0.9303382390911438,
      "grad_norm": 4.139667510986328,
      "learning_rate": 1.2167595607996296e-07,
      "loss": 0.0291,
      "step": 18016
    },
    {
      "epoch": 0.9303898786470436,
      "grad_norm": 2.8202874660491943,
      "learning_rate": 1.2149636252204944e-07,
      "loss": 0.0367,
      "step": 18017
    },
    {
      "epoch": 0.9304415182029434,
      "grad_norm": 6.237051010131836,
      "learning_rate": 1.2131689997184225e-07,
      "loss": 0.0548,
      "step": 18018
    },
    {
      "epoch": 0.9304931577588432,
      "grad_norm": 2.939612627029419,
      "learning_rate": 1.2113756843416146e-07,
      "loss": 0.0709,
      "step": 18019
    },
    {
      "epoch": 0.930544797314743,
      "grad_norm": 3.8085732460021973,
      "learning_rate": 1.2095836791382155e-07,
      "loss": 0.097,
      "step": 18020
    },
    {
      "epoch": 0.9305964368706429,
      "grad_norm": 6.036711692810059,
      "learning_rate": 1.2077929841563586e-07,
      "loss": 0.0981,
      "step": 18021
    },
    {
      "epoch": 0.9306480764265427,
      "grad_norm": 2.1080071926116943,
      "learning_rate": 1.2060035994441165e-07,
      "loss": 0.0302,
      "step": 18022
    },
    {
      "epoch": 0.9306997159824425,
      "grad_norm": 3.5622756481170654,
      "learning_rate": 1.2042155250495568e-07,
      "loss": 0.0464,
      "step": 18023
    },
    {
      "epoch": 0.9307513555383423,
      "grad_norm": 4.282547473907471,
      "learning_rate": 1.2024287610206909e-07,
      "loss": 0.054,
      "step": 18024
    },
    {
      "epoch": 0.9308029950942421,
      "grad_norm": 8.754303932189941,
      "learning_rate": 1.2006433074055023e-07,
      "loss": 0.0906,
      "step": 18025
    },
    {
      "epoch": 0.930854634650142,
      "grad_norm": 3.7717173099517822,
      "learning_rate": 1.198859164251931e-07,
      "loss": 0.0658,
      "step": 18026
    },
    {
      "epoch": 0.9309062742060418,
      "grad_norm": 2.9449384212493896,
      "learning_rate": 1.1970763316078993e-07,
      "loss": 0.0483,
      "step": 18027
    },
    {
      "epoch": 0.9309579137619417,
      "grad_norm": 2.8954834938049316,
      "learning_rate": 1.1952948095212745e-07,
      "loss": 0.0437,
      "step": 18028
    },
    {
      "epoch": 0.9310095533178415,
      "grad_norm": 1.097401738166809,
      "learning_rate": 1.1935145980399022e-07,
      "loss": 0.0169,
      "step": 18029
    },
    {
      "epoch": 0.9310611928737413,
      "grad_norm": 12.779051780700684,
      "learning_rate": 1.1917356972115879e-07,
      "loss": 0.0566,
      "step": 18030
    },
    {
      "epoch": 0.9311128324296412,
      "grad_norm": 5.9369049072265625,
      "learning_rate": 1.1899581070840993e-07,
      "loss": 0.0545,
      "step": 18031
    },
    {
      "epoch": 0.931164471985541,
      "grad_norm": 6.462536334991455,
      "learning_rate": 1.1881818277051704e-07,
      "loss": 0.049,
      "step": 18032
    },
    {
      "epoch": 0.9312161115414408,
      "grad_norm": 6.803020000457764,
      "learning_rate": 1.1864068591225131e-07,
      "loss": 0.0859,
      "step": 18033
    },
    {
      "epoch": 0.9312677510973406,
      "grad_norm": 2.9456629753112793,
      "learning_rate": 1.1846332013837836e-07,
      "loss": 0.0646,
      "step": 18034
    },
    {
      "epoch": 0.9313193906532404,
      "grad_norm": 3.4751360416412354,
      "learning_rate": 1.182860854536616e-07,
      "loss": 0.0497,
      "step": 18035
    },
    {
      "epoch": 0.9313710302091402,
      "grad_norm": 1.557693600654602,
      "learning_rate": 1.1810898186285891e-07,
      "loss": 0.0165,
      "step": 18036
    },
    {
      "epoch": 0.9314226697650401,
      "grad_norm": 4.636733531951904,
      "learning_rate": 1.1793200937072869e-07,
      "loss": 0.0259,
      "step": 18037
    },
    {
      "epoch": 0.9314743093209399,
      "grad_norm": 10.002543449401855,
      "learning_rate": 1.1775516798202213e-07,
      "loss": 0.1216,
      "step": 18038
    },
    {
      "epoch": 0.9315259488768397,
      "grad_norm": 4.494680404663086,
      "learning_rate": 1.1757845770148824e-07,
      "loss": 0.0468,
      "step": 18039
    },
    {
      "epoch": 0.9315775884327395,
      "grad_norm": 1.6366162300109863,
      "learning_rate": 1.1740187853387208e-07,
      "loss": 0.0202,
      "step": 18040
    },
    {
      "epoch": 0.9316292279886393,
      "grad_norm": 4.894298553466797,
      "learning_rate": 1.1722543048391599e-07,
      "loss": 0.0469,
      "step": 18041
    },
    {
      "epoch": 0.9316808675445392,
      "grad_norm": 4.296062469482422,
      "learning_rate": 1.1704911355635785e-07,
      "loss": 0.0481,
      "step": 18042
    },
    {
      "epoch": 0.931732507100439,
      "grad_norm": 3.302814245223999,
      "learning_rate": 1.1687292775593329e-07,
      "loss": 0.0381,
      "step": 18043
    },
    {
      "epoch": 0.9317841466563388,
      "grad_norm": 1.8867937326431274,
      "learning_rate": 1.1669687308737299e-07,
      "loss": 0.0222,
      "step": 18044
    },
    {
      "epoch": 0.9318357862122386,
      "grad_norm": 3.1340348720550537,
      "learning_rate": 1.1652094955540483e-07,
      "loss": 0.0461,
      "step": 18045
    },
    {
      "epoch": 0.9318874257681384,
      "grad_norm": 6.375030040740967,
      "learning_rate": 1.1634515716475336e-07,
      "loss": 0.0671,
      "step": 18046
    },
    {
      "epoch": 0.9319390653240383,
      "grad_norm": 5.521002292633057,
      "learning_rate": 1.1616949592013871e-07,
      "loss": 0.0897,
      "step": 18047
    },
    {
      "epoch": 0.9319907048799381,
      "grad_norm": 1.4054169654846191,
      "learning_rate": 1.1599396582627875e-07,
      "loss": 0.0227,
      "step": 18048
    },
    {
      "epoch": 0.9320423444358379,
      "grad_norm": 20.405141830444336,
      "learning_rate": 1.1581856688788695e-07,
      "loss": 0.117,
      "step": 18049
    },
    {
      "epoch": 0.9320939839917377,
      "grad_norm": 19.09389305114746,
      "learning_rate": 1.1564329910967287e-07,
      "loss": 0.2564,
      "step": 18050
    },
    {
      "epoch": 0.9321456235476375,
      "grad_norm": 5.039989948272705,
      "learning_rate": 1.1546816249634441e-07,
      "loss": 0.0744,
      "step": 18051
    },
    {
      "epoch": 0.9321972631035373,
      "grad_norm": 10.095355987548828,
      "learning_rate": 1.1529315705260446e-07,
      "loss": 0.0651,
      "step": 18052
    },
    {
      "epoch": 0.9322489026594372,
      "grad_norm": 2.456916570663452,
      "learning_rate": 1.1511828278315151e-07,
      "loss": 0.0358,
      "step": 18053
    },
    {
      "epoch": 0.932300542215337,
      "grad_norm": 2.6987898349761963,
      "learning_rate": 1.1494353969268235e-07,
      "loss": 0.0239,
      "step": 18054
    },
    {
      "epoch": 0.9323521817712368,
      "grad_norm": 5.871132850646973,
      "learning_rate": 1.1476892778588933e-07,
      "loss": 0.1162,
      "step": 18055
    },
    {
      "epoch": 0.9324038213271366,
      "grad_norm": 2.7500064373016357,
      "learning_rate": 1.145944470674626e-07,
      "loss": 0.022,
      "step": 18056
    },
    {
      "epoch": 0.9324554608830364,
      "grad_norm": 3.2934370040893555,
      "learning_rate": 1.1442009754208617e-07,
      "loss": 0.0322,
      "step": 18057
    },
    {
      "epoch": 0.9325071004389363,
      "grad_norm": 5.781211853027344,
      "learning_rate": 1.1424587921444242e-07,
      "loss": 0.0797,
      "step": 18058
    },
    {
      "epoch": 0.9325587399948361,
      "grad_norm": 2.526416063308716,
      "learning_rate": 1.1407179208921037e-07,
      "loss": 0.0454,
      "step": 18059
    },
    {
      "epoch": 0.9326103795507359,
      "grad_norm": 15.286535263061523,
      "learning_rate": 1.1389783617106465e-07,
      "loss": 0.0362,
      "step": 18060
    },
    {
      "epoch": 0.9326620191066357,
      "grad_norm": 1.7364740371704102,
      "learning_rate": 1.1372401146467648e-07,
      "loss": 0.0339,
      "step": 18061
    },
    {
      "epoch": 0.9327136586625355,
      "grad_norm": 4.622199058532715,
      "learning_rate": 1.1355031797471383e-07,
      "loss": 0.0719,
      "step": 18062
    },
    {
      "epoch": 0.9327652982184353,
      "grad_norm": 4.824285507202148,
      "learning_rate": 1.1337675570584184e-07,
      "loss": 0.057,
      "step": 18063
    },
    {
      "epoch": 0.9328169377743352,
      "grad_norm": 3.1355512142181396,
      "learning_rate": 1.1320332466272011e-07,
      "loss": 0.0534,
      "step": 18064
    },
    {
      "epoch": 0.932868577330235,
      "grad_norm": 1.1705223321914673,
      "learning_rate": 1.1303002485000658e-07,
      "loss": 0.0078,
      "step": 18065
    },
    {
      "epoch": 0.9329202168861348,
      "grad_norm": 4.7993364334106445,
      "learning_rate": 1.1285685627235531e-07,
      "loss": 0.0393,
      "step": 18066
    },
    {
      "epoch": 0.9329718564420346,
      "grad_norm": 13.0940580368042,
      "learning_rate": 1.1268381893441704e-07,
      "loss": 0.0811,
      "step": 18067
    },
    {
      "epoch": 0.9330234959979344,
      "grad_norm": 10.589311599731445,
      "learning_rate": 1.1251091284083693e-07,
      "loss": 0.0955,
      "step": 18068
    },
    {
      "epoch": 0.9330751355538343,
      "grad_norm": 3.24387264251709,
      "learning_rate": 1.1233813799625903e-07,
      "loss": 0.0467,
      "step": 18069
    },
    {
      "epoch": 0.9331267751097341,
      "grad_norm": 2.1116342544555664,
      "learning_rate": 1.1216549440532298e-07,
      "loss": 0.0267,
      "step": 18070
    },
    {
      "epoch": 0.9331784146656339,
      "grad_norm": 4.829451084136963,
      "learning_rate": 1.1199298207266618e-07,
      "loss": 0.0591,
      "step": 18071
    },
    {
      "epoch": 0.9332300542215337,
      "grad_norm": 3.777696132659912,
      "learning_rate": 1.1182060100291937e-07,
      "loss": 0.0448,
      "step": 18072
    },
    {
      "epoch": 0.9332816937774335,
      "grad_norm": 13.553984642028809,
      "learning_rate": 1.1164835120071271e-07,
      "loss": 0.134,
      "step": 18073
    },
    {
      "epoch": 0.9333333333333333,
      "grad_norm": 2.4484007358551025,
      "learning_rate": 1.1147623267067199e-07,
      "loss": 0.056,
      "step": 18074
    },
    {
      "epoch": 0.9333849728892332,
      "grad_norm": 7.714473247528076,
      "learning_rate": 1.1130424541741846e-07,
      "loss": 0.1392,
      "step": 18075
    },
    {
      "epoch": 0.933436612445133,
      "grad_norm": 9.273513793945312,
      "learning_rate": 1.1113238944557125e-07,
      "loss": 0.0895,
      "step": 18076
    },
    {
      "epoch": 0.9334882520010328,
      "grad_norm": 2.5816457271575928,
      "learning_rate": 1.1096066475974554e-07,
      "loss": 0.0546,
      "step": 18077
    },
    {
      "epoch": 0.9335398915569326,
      "grad_norm": 2.6101887226104736,
      "learning_rate": 1.1078907136455264e-07,
      "loss": 0.0253,
      "step": 18078
    },
    {
      "epoch": 0.9335915311128324,
      "grad_norm": 7.603453636169434,
      "learning_rate": 1.1061760926459997e-07,
      "loss": 0.0952,
      "step": 18079
    },
    {
      "epoch": 0.9336431706687323,
      "grad_norm": 5.571260929107666,
      "learning_rate": 1.1044627846449329e-07,
      "loss": 0.079,
      "step": 18080
    },
    {
      "epoch": 0.9336948102246321,
      "grad_norm": 25.661054611206055,
      "learning_rate": 1.1027507896883227e-07,
      "loss": 0.0925,
      "step": 18081
    },
    {
      "epoch": 0.9337464497805319,
      "grad_norm": 16.330015182495117,
      "learning_rate": 1.1010401078221489e-07,
      "loss": 0.1325,
      "step": 18082
    },
    {
      "epoch": 0.9337980893364317,
      "grad_norm": 10.296379089355469,
      "learning_rate": 1.0993307390923469e-07,
      "loss": 0.0583,
      "step": 18083
    },
    {
      "epoch": 0.9338497288923315,
      "grad_norm": 16.216920852661133,
      "learning_rate": 1.0976226835448245e-07,
      "loss": 0.2251,
      "step": 18084
    },
    {
      "epoch": 0.9339013684482314,
      "grad_norm": 4.3524580001831055,
      "learning_rate": 1.0959159412254505e-07,
      "loss": 0.0645,
      "step": 18085
    },
    {
      "epoch": 0.9339530080041312,
      "grad_norm": 15.443341255187988,
      "learning_rate": 1.094210512180055e-07,
      "loss": 0.1769,
      "step": 18086
    },
    {
      "epoch": 0.934004647560031,
      "grad_norm": 3.5095183849334717,
      "learning_rate": 1.0925063964544346e-07,
      "loss": 0.0817,
      "step": 18087
    },
    {
      "epoch": 0.9340562871159308,
      "grad_norm": 8.979382514953613,
      "learning_rate": 1.0908035940943474e-07,
      "loss": 0.0324,
      "step": 18088
    },
    {
      "epoch": 0.9341079266718306,
      "grad_norm": 5.113129615783691,
      "learning_rate": 1.0891021051455397e-07,
      "loss": 0.0584,
      "step": 18089
    },
    {
      "epoch": 0.9341595662277304,
      "grad_norm": 4.340865135192871,
      "learning_rate": 1.0874019296536808e-07,
      "loss": 0.049,
      "step": 18090
    },
    {
      "epoch": 0.9342112057836303,
      "grad_norm": 2.568477153778076,
      "learning_rate": 1.0857030676644398e-07,
      "loss": 0.0402,
      "step": 18091
    },
    {
      "epoch": 0.9342628453395301,
      "grad_norm": 5.366862773895264,
      "learning_rate": 1.0840055192234355e-07,
      "loss": 0.0501,
      "step": 18092
    },
    {
      "epoch": 0.9343144848954299,
      "grad_norm": 1.123766541481018,
      "learning_rate": 1.0823092843762484e-07,
      "loss": 0.011,
      "step": 18093
    },
    {
      "epoch": 0.9343661244513297,
      "grad_norm": 2.2118701934814453,
      "learning_rate": 1.0806143631684419e-07,
      "loss": 0.0155,
      "step": 18094
    },
    {
      "epoch": 0.9344177640072295,
      "grad_norm": 3.167583465576172,
      "learning_rate": 1.0789207556455183e-07,
      "loss": 0.0459,
      "step": 18095
    },
    {
      "epoch": 0.9344694035631294,
      "grad_norm": 2.551485538482666,
      "learning_rate": 1.0772284618529694e-07,
      "loss": 0.0472,
      "step": 18096
    },
    {
      "epoch": 0.9345210431190292,
      "grad_norm": 3.2805516719818115,
      "learning_rate": 1.0755374818362307e-07,
      "loss": 0.026,
      "step": 18097
    },
    {
      "epoch": 0.934572682674929,
      "grad_norm": 1.9845727682113647,
      "learning_rate": 1.0738478156407162e-07,
      "loss": 0.0142,
      "step": 18098
    },
    {
      "epoch": 0.9346243222308288,
      "grad_norm": 13.191499710083008,
      "learning_rate": 1.072159463311795e-07,
      "loss": 0.0821,
      "step": 18099
    },
    {
      "epoch": 0.9346759617867286,
      "grad_norm": 2.749264717102051,
      "learning_rate": 1.0704724248948251e-07,
      "loss": 0.0324,
      "step": 18100
    },
    {
      "epoch": 0.9347276013426284,
      "grad_norm": 15.080977439880371,
      "learning_rate": 1.0687867004350872e-07,
      "loss": 0.0557,
      "step": 18101
    },
    {
      "epoch": 0.9347792408985283,
      "grad_norm": 2.5061819553375244,
      "learning_rate": 1.0671022899778505e-07,
      "loss": 0.0164,
      "step": 18102
    },
    {
      "epoch": 0.9348308804544281,
      "grad_norm": 4.729485988616943,
      "learning_rate": 1.0654191935683678e-07,
      "loss": 0.0759,
      "step": 18103
    },
    {
      "epoch": 0.9348825200103279,
      "grad_norm": 3.2370035648345947,
      "learning_rate": 1.063737411251825e-07,
      "loss": 0.0293,
      "step": 18104
    },
    {
      "epoch": 0.9349341595662277,
      "grad_norm": 12.857150077819824,
      "learning_rate": 1.0620569430733862e-07,
      "loss": 0.0534,
      "step": 18105
    },
    {
      "epoch": 0.9349857991221275,
      "grad_norm": 3.485177993774414,
      "learning_rate": 1.0603777890781763e-07,
      "loss": 0.0343,
      "step": 18106
    },
    {
      "epoch": 0.9350374386780274,
      "grad_norm": 3.759899854660034,
      "learning_rate": 1.0586999493112871e-07,
      "loss": 0.0428,
      "step": 18107
    },
    {
      "epoch": 0.9350890782339272,
      "grad_norm": 6.058359146118164,
      "learning_rate": 1.0570234238177824e-07,
      "loss": 0.0623,
      "step": 18108
    },
    {
      "epoch": 0.935140717789827,
      "grad_norm": 8.578051567077637,
      "learning_rate": 1.0553482126426762e-07,
      "loss": 0.0722,
      "step": 18109
    },
    {
      "epoch": 0.9351923573457268,
      "grad_norm": 3.390772819519043,
      "learning_rate": 1.0536743158309603e-07,
      "loss": 0.0367,
      "step": 18110
    },
    {
      "epoch": 0.9352439969016266,
      "grad_norm": 6.157176494598389,
      "learning_rate": 1.0520017334275823e-07,
      "loss": 0.0778,
      "step": 18111
    },
    {
      "epoch": 0.9352956364575264,
      "grad_norm": 1.8316410779953003,
      "learning_rate": 1.050330465477456e-07,
      "loss": 0.015,
      "step": 18112
    },
    {
      "epoch": 0.9353472760134263,
      "grad_norm": 9.919333457946777,
      "learning_rate": 1.0486605120254679e-07,
      "loss": 0.1343,
      "step": 18113
    },
    {
      "epoch": 0.9353989155693261,
      "grad_norm": 8.204739570617676,
      "learning_rate": 1.0469918731164597e-07,
      "loss": 0.0679,
      "step": 18114
    },
    {
      "epoch": 0.9354505551252259,
      "grad_norm": 3.564642906188965,
      "learning_rate": 1.0453245487952457e-07,
      "loss": 0.0523,
      "step": 18115
    },
    {
      "epoch": 0.9355021946811257,
      "grad_norm": 11.247703552246094,
      "learning_rate": 1.0436585391065845e-07,
      "loss": 0.1442,
      "step": 18116
    },
    {
      "epoch": 0.9355538342370255,
      "grad_norm": 14.224648475646973,
      "learning_rate": 1.0419938440952348e-07,
      "loss": 0.1492,
      "step": 18117
    },
    {
      "epoch": 0.9356054737929254,
      "grad_norm": 4.357693195343018,
      "learning_rate": 1.0403304638058942e-07,
      "loss": 0.0588,
      "step": 18118
    },
    {
      "epoch": 0.9356571133488252,
      "grad_norm": 5.08120059967041,
      "learning_rate": 1.0386683982832213e-07,
      "loss": 0.0574,
      "step": 18119
    },
    {
      "epoch": 0.935708752904725,
      "grad_norm": 3.3550453186035156,
      "learning_rate": 1.0370076475718638e-07,
      "loss": 0.0513,
      "step": 18120
    },
    {
      "epoch": 0.9357603924606248,
      "grad_norm": 4.085882663726807,
      "learning_rate": 1.0353482117164026e-07,
      "loss": 0.0587,
      "step": 18121
    },
    {
      "epoch": 0.9358120320165246,
      "grad_norm": 2.9744980335235596,
      "learning_rate": 1.0336900907614245e-07,
      "loss": 0.0504,
      "step": 18122
    },
    {
      "epoch": 0.9358636715724244,
      "grad_norm": 2.1561050415039062,
      "learning_rate": 1.032033284751438e-07,
      "loss": 0.0279,
      "step": 18123
    },
    {
      "epoch": 0.9359153111283243,
      "grad_norm": 8.133273124694824,
      "learning_rate": 1.0303777937309356e-07,
      "loss": 0.0641,
      "step": 18124
    },
    {
      "epoch": 0.9359669506842241,
      "grad_norm": 2.7499797344207764,
      "learning_rate": 1.0287236177443871e-07,
      "loss": 0.0403,
      "step": 18125
    },
    {
      "epoch": 0.9360185902401239,
      "grad_norm": 2.280024766921997,
      "learning_rate": 1.027070756836196e-07,
      "loss": 0.0443,
      "step": 18126
    },
    {
      "epoch": 0.9360702297960237,
      "grad_norm": 12.367984771728516,
      "learning_rate": 1.0254192110507655e-07,
      "loss": 0.1384,
      "step": 18127
    },
    {
      "epoch": 0.9361218693519235,
      "grad_norm": 6.9474873542785645,
      "learning_rate": 1.0237689804324325e-07,
      "loss": 0.036,
      "step": 18128
    },
    {
      "epoch": 0.9361735089078234,
      "grad_norm": 4.267338752746582,
      "learning_rate": 1.0221200650255281e-07,
      "loss": 0.0463,
      "step": 18129
    },
    {
      "epoch": 0.9362251484637232,
      "grad_norm": 13.842117309570312,
      "learning_rate": 1.0204724648743114e-07,
      "loss": 0.2579,
      "step": 18130
    },
    {
      "epoch": 0.936276788019623,
      "grad_norm": 3.341963529586792,
      "learning_rate": 1.0188261800230415e-07,
      "loss": 0.0427,
      "step": 18131
    },
    {
      "epoch": 0.9363284275755228,
      "grad_norm": 5.7085161209106445,
      "learning_rate": 1.0171812105159274e-07,
      "loss": 0.096,
      "step": 18132
    },
    {
      "epoch": 0.9363800671314226,
      "grad_norm": 2.046488046646118,
      "learning_rate": 1.0155375563971393e-07,
      "loss": 0.0314,
      "step": 18133
    },
    {
      "epoch": 0.9364317066873225,
      "grad_norm": 1.7454832792282104,
      "learning_rate": 1.0138952177108197e-07,
      "loss": 0.0199,
      "step": 18134
    },
    {
      "epoch": 0.9364833462432223,
      "grad_norm": 3.9714620113372803,
      "learning_rate": 1.0122541945010611e-07,
      "loss": 0.0428,
      "step": 18135
    },
    {
      "epoch": 0.9365349857991221,
      "grad_norm": 3.758582830429077,
      "learning_rate": 1.0106144868119506e-07,
      "loss": 0.0345,
      "step": 18136
    },
    {
      "epoch": 0.9365866253550219,
      "grad_norm": 2.092135190963745,
      "learning_rate": 1.0089760946875082e-07,
      "loss": 0.0234,
      "step": 18137
    },
    {
      "epoch": 0.9366382649109217,
      "grad_norm": 3.1832475662231445,
      "learning_rate": 1.0073390181717324e-07,
      "loss": 0.0474,
      "step": 18138
    },
    {
      "epoch": 0.9366899044668215,
      "grad_norm": 4.420958518981934,
      "learning_rate": 1.0057032573085879e-07,
      "loss": 0.104,
      "step": 18139
    },
    {
      "epoch": 0.9367415440227214,
      "grad_norm": 6.326009750366211,
      "learning_rate": 1.0040688121420006e-07,
      "loss": 0.0621,
      "step": 18140
    },
    {
      "epoch": 0.9367931835786212,
      "grad_norm": 5.608681678771973,
      "learning_rate": 1.0024356827158632e-07,
      "loss": 0.046,
      "step": 18141
    },
    {
      "epoch": 0.936844823134521,
      "grad_norm": 5.414803504943848,
      "learning_rate": 1.0008038690740296e-07,
      "loss": 0.0244,
      "step": 18142
    },
    {
      "epoch": 0.9368964626904208,
      "grad_norm": 3.1910104751586914,
      "learning_rate": 9.991733712603202e-08,
      "loss": 0.0458,
      "step": 18143
    },
    {
      "epoch": 0.9369481022463206,
      "grad_norm": 2.8657352924346924,
      "learning_rate": 9.975441893185222e-08,
      "loss": 0.0191,
      "step": 18144
    },
    {
      "epoch": 0.9369997418022205,
      "grad_norm": 6.611483573913574,
      "learning_rate": 9.959163232923896e-08,
      "loss": 0.0802,
      "step": 18145
    },
    {
      "epoch": 0.9370513813581203,
      "grad_norm": 3.948119878768921,
      "learning_rate": 9.942897732256318e-08,
      "loss": 0.0382,
      "step": 18146
    },
    {
      "epoch": 0.9371030209140201,
      "grad_norm": 7.776339054107666,
      "learning_rate": 9.92664539161925e-08,
      "loss": 0.0762,
      "step": 18147
    },
    {
      "epoch": 0.9371546604699199,
      "grad_norm": 14.451101303100586,
      "learning_rate": 9.910406211449286e-08,
      "loss": 0.0597,
      "step": 18148
    },
    {
      "epoch": 0.9372063000258197,
      "grad_norm": 3.0321826934814453,
      "learning_rate": 9.894180192182301e-08,
      "loss": 0.048,
      "step": 18149
    },
    {
      "epoch": 0.9372579395817195,
      "grad_norm": 4.205024242401123,
      "learning_rate": 9.877967334254168e-08,
      "loss": 0.0443,
      "step": 18150
    },
    {
      "epoch": 0.9373095791376194,
      "grad_norm": 3.305952310562134,
      "learning_rate": 9.861767638100317e-08,
      "loss": 0.0447,
      "step": 18151
    },
    {
      "epoch": 0.9373612186935192,
      "grad_norm": 7.679361820220947,
      "learning_rate": 9.845581104155621e-08,
      "loss": 0.0885,
      "step": 18152
    },
    {
      "epoch": 0.937412858249419,
      "grad_norm": 7.866631984710693,
      "learning_rate": 9.829407732854846e-08,
      "loss": 0.0979,
      "step": 18153
    },
    {
      "epoch": 0.9374644978053188,
      "grad_norm": 9.028581619262695,
      "learning_rate": 9.813247524632253e-08,
      "loss": 0.0467,
      "step": 18154
    },
    {
      "epoch": 0.9375161373612187,
      "grad_norm": 5.454813480377197,
      "learning_rate": 9.797100479921939e-08,
      "loss": 0.0348,
      "step": 18155
    },
    {
      "epoch": 0.9375677769171186,
      "grad_norm": 2.5595669746398926,
      "learning_rate": 9.780966599157393e-08,
      "loss": 0.0404,
      "step": 18156
    },
    {
      "epoch": 0.9376194164730184,
      "grad_norm": 1.9669557809829712,
      "learning_rate": 9.764845882771878e-08,
      "loss": 0.0233,
      "step": 18157
    },
    {
      "epoch": 0.9376710560289182,
      "grad_norm": 3.99358868598938,
      "learning_rate": 9.748738331198382e-08,
      "loss": 0.052,
      "step": 18158
    },
    {
      "epoch": 0.937722695584818,
      "grad_norm": 3.764427900314331,
      "learning_rate": 9.73264394486939e-08,
      "loss": 0.019,
      "step": 18159
    },
    {
      "epoch": 0.9377743351407178,
      "grad_norm": 2.8909425735473633,
      "learning_rate": 9.716562724217172e-08,
      "loss": 0.0424,
      "step": 18160
    },
    {
      "epoch": 0.9378259746966177,
      "grad_norm": 2.7657957077026367,
      "learning_rate": 9.700494669673488e-08,
      "loss": 0.0555,
      "step": 18161
    },
    {
      "epoch": 0.9378776142525175,
      "grad_norm": 2.7116506099700928,
      "learning_rate": 9.684439781669941e-08,
      "loss": 0.0619,
      "step": 18162
    },
    {
      "epoch": 0.9379292538084173,
      "grad_norm": 5.199524879455566,
      "learning_rate": 9.668398060637518e-08,
      "loss": 0.0859,
      "step": 18163
    },
    {
      "epoch": 0.9379808933643171,
      "grad_norm": 14.520220756530762,
      "learning_rate": 9.652369507007098e-08,
      "loss": 0.1004,
      "step": 18164
    },
    {
      "epoch": 0.9380325329202169,
      "grad_norm": 4.402139186859131,
      "learning_rate": 9.636354121209113e-08,
      "loss": 0.0291,
      "step": 18165
    },
    {
      "epoch": 0.9380841724761168,
      "grad_norm": 3.1875460147857666,
      "learning_rate": 9.620351903673663e-08,
      "loss": 0.0136,
      "step": 18166
    },
    {
      "epoch": 0.9381358120320166,
      "grad_norm": 3.6878576278686523,
      "learning_rate": 9.604362854830462e-08,
      "loss": 0.0243,
      "step": 18167
    },
    {
      "epoch": 0.9381874515879164,
      "grad_norm": 3.7230918407440186,
      "learning_rate": 9.588386975108777e-08,
      "loss": 0.0457,
      "step": 18168
    },
    {
      "epoch": 0.9382390911438162,
      "grad_norm": 2.9931998252868652,
      "learning_rate": 9.572424264937708e-08,
      "loss": 0.0146,
      "step": 18169
    },
    {
      "epoch": 0.938290730699716,
      "grad_norm": 3.757460117340088,
      "learning_rate": 9.556474724746023e-08,
      "loss": 0.0535,
      "step": 18170
    },
    {
      "epoch": 0.9383423702556158,
      "grad_norm": 3.1481642723083496,
      "learning_rate": 9.540538354961881e-08,
      "loss": 0.0589,
      "step": 18171
    },
    {
      "epoch": 0.9383940098115157,
      "grad_norm": 4.131826400756836,
      "learning_rate": 9.524615156013273e-08,
      "loss": 0.0667,
      "step": 18172
    },
    {
      "epoch": 0.9384456493674155,
      "grad_norm": 2.3791615962982178,
      "learning_rate": 9.5087051283278e-08,
      "loss": 0.0462,
      "step": 18173
    },
    {
      "epoch": 0.9384972889233153,
      "grad_norm": 3.416940450668335,
      "learning_rate": 9.492808272332731e-08,
      "loss": 0.0471,
      "step": 18174
    },
    {
      "epoch": 0.9385489284792151,
      "grad_norm": 2.47528338432312,
      "learning_rate": 9.47692458845495e-08,
      "loss": 0.0337,
      "step": 18175
    },
    {
      "epoch": 0.9386005680351149,
      "grad_norm": 2.058502674102783,
      "learning_rate": 9.461054077121057e-08,
      "loss": 0.0287,
      "step": 18176
    },
    {
      "epoch": 0.9386522075910148,
      "grad_norm": 2.721240997314453,
      "learning_rate": 9.445196738757156e-08,
      "loss": 0.0508,
      "step": 18177
    },
    {
      "epoch": 0.9387038471469146,
      "grad_norm": 2.9753975868225098,
      "learning_rate": 9.429352573789075e-08,
      "loss": 0.0479,
      "step": 18178
    },
    {
      "epoch": 0.9387554867028144,
      "grad_norm": 2.489778995513916,
      "learning_rate": 9.41352158264236e-08,
      "loss": 0.0422,
      "step": 18179
    },
    {
      "epoch": 0.9388071262587142,
      "grad_norm": 3.836254596710205,
      "learning_rate": 9.397703765742116e-08,
      "loss": 0.0433,
      "step": 18180
    },
    {
      "epoch": 0.938858765814614,
      "grad_norm": 3.223825216293335,
      "learning_rate": 9.381899123513172e-08,
      "loss": 0.0224,
      "step": 18181
    },
    {
      "epoch": 0.9389104053705138,
      "grad_norm": 9.842671394348145,
      "learning_rate": 9.366107656379797e-08,
      "loss": 0.0933,
      "step": 18182
    },
    {
      "epoch": 0.9389620449264137,
      "grad_norm": 6.708298683166504,
      "learning_rate": 9.350329364766153e-08,
      "loss": 0.039,
      "step": 18183
    },
    {
      "epoch": 0.9390136844823135,
      "grad_norm": 4.978942394256592,
      "learning_rate": 9.334564249096011e-08,
      "loss": 0.0371,
      "step": 18184
    },
    {
      "epoch": 0.9390653240382133,
      "grad_norm": 4.086799621582031,
      "learning_rate": 9.318812309792591e-08,
      "loss": 0.0616,
      "step": 18185
    },
    {
      "epoch": 0.9391169635941131,
      "grad_norm": 7.634232521057129,
      "learning_rate": 9.303073547278995e-08,
      "loss": 0.0636,
      "step": 18186
    },
    {
      "epoch": 0.9391686031500129,
      "grad_norm": 5.2768402099609375,
      "learning_rate": 9.287347961977833e-08,
      "loss": 0.0752,
      "step": 18187
    },
    {
      "epoch": 0.9392202427059128,
      "grad_norm": 4.924057483673096,
      "learning_rate": 9.271635554311431e-08,
      "loss": 0.0684,
      "step": 18188
    },
    {
      "epoch": 0.9392718822618126,
      "grad_norm": 1.578589677810669,
      "learning_rate": 9.255936324701732e-08,
      "loss": 0.024,
      "step": 18189
    },
    {
      "epoch": 0.9393235218177124,
      "grad_norm": 2.702108860015869,
      "learning_rate": 9.240250273570228e-08,
      "loss": 0.0411,
      "step": 18190
    },
    {
      "epoch": 0.9393751613736122,
      "grad_norm": 1.9560078382492065,
      "learning_rate": 9.224577401338309e-08,
      "loss": 0.0252,
      "step": 18191
    },
    {
      "epoch": 0.939426800929512,
      "grad_norm": 3.8343827724456787,
      "learning_rate": 9.208917708426746e-08,
      "loss": 0.0452,
      "step": 18192
    },
    {
      "epoch": 0.9394784404854118,
      "grad_norm": 2.108350992202759,
      "learning_rate": 9.193271195256037e-08,
      "loss": 0.0288,
      "step": 18193
    },
    {
      "epoch": 0.9395300800413117,
      "grad_norm": 5.219727516174316,
      "learning_rate": 9.177637862246513e-08,
      "loss": 0.0538,
      "step": 18194
    },
    {
      "epoch": 0.9395817195972115,
      "grad_norm": 6.582928657531738,
      "learning_rate": 9.162017709817894e-08,
      "loss": 0.1484,
      "step": 18195
    },
    {
      "epoch": 0.9396333591531113,
      "grad_norm": 0.7225218415260315,
      "learning_rate": 9.14641073838951e-08,
      "loss": 0.0077,
      "step": 18196
    },
    {
      "epoch": 0.9396849987090111,
      "grad_norm": 2.3871021270751953,
      "learning_rate": 9.130816948380695e-08,
      "loss": 0.0225,
      "step": 18197
    },
    {
      "epoch": 0.9397366382649109,
      "grad_norm": 5.827293872833252,
      "learning_rate": 9.115236340210166e-08,
      "loss": 0.0377,
      "step": 18198
    },
    {
      "epoch": 0.9397882778208108,
      "grad_norm": 12.133318901062012,
      "learning_rate": 9.099668914296255e-08,
      "loss": 0.1473,
      "step": 18199
    },
    {
      "epoch": 0.9398399173767106,
      "grad_norm": 2.4843504428863525,
      "learning_rate": 9.084114671057021e-08,
      "loss": 0.0299,
      "step": 18200
    },
    {
      "epoch": 0.9398915569326104,
      "grad_norm": 7.087238788604736,
      "learning_rate": 9.068573610910069e-08,
      "loss": 0.0519,
      "step": 18201
    },
    {
      "epoch": 0.9399431964885102,
      "grad_norm": 31.045629501342773,
      "learning_rate": 9.053045734272959e-08,
      "loss": 0.1362,
      "step": 18202
    },
    {
      "epoch": 0.93999483604441,
      "grad_norm": 5.754273414611816,
      "learning_rate": 9.037531041562575e-08,
      "loss": 0.039,
      "step": 18203
    },
    {
      "epoch": 0.9400464756003098,
      "grad_norm": 2.622774124145508,
      "learning_rate": 9.022029533195476e-08,
      "loss": 0.0434,
      "step": 18204
    },
    {
      "epoch": 0.9400981151562097,
      "grad_norm": 7.320806980133057,
      "learning_rate": 9.006541209587994e-08,
      "loss": 0.0489,
      "step": 18205
    },
    {
      "epoch": 0.9401497547121095,
      "grad_norm": 2.802246570587158,
      "learning_rate": 8.991066071156074e-08,
      "loss": 0.0483,
      "step": 18206
    },
    {
      "epoch": 0.9402013942680093,
      "grad_norm": 9.985138893127441,
      "learning_rate": 8.975604118315273e-08,
      "loss": 0.0965,
      "step": 18207
    },
    {
      "epoch": 0.9402530338239091,
      "grad_norm": 3.3458216190338135,
      "learning_rate": 8.960155351480815e-08,
      "loss": 0.0226,
      "step": 18208
    },
    {
      "epoch": 0.9403046733798089,
      "grad_norm": 2.3074071407318115,
      "learning_rate": 8.944719771067534e-08,
      "loss": 0.041,
      "step": 18209
    },
    {
      "epoch": 0.9403563129357088,
      "grad_norm": 5.171643257141113,
      "learning_rate": 8.929297377489987e-08,
      "loss": 0.0408,
      "step": 18210
    },
    {
      "epoch": 0.9404079524916086,
      "grad_norm": 2.5066168308258057,
      "learning_rate": 8.913888171162288e-08,
      "loss": 0.0472,
      "step": 18211
    },
    {
      "epoch": 0.9404595920475084,
      "grad_norm": 6.336357593536377,
      "learning_rate": 8.898492152498217e-08,
      "loss": 0.0543,
      "step": 18212
    },
    {
      "epoch": 0.9405112316034082,
      "grad_norm": 3.633155107498169,
      "learning_rate": 8.883109321911276e-08,
      "loss": 0.0236,
      "step": 18213
    },
    {
      "epoch": 0.940562871159308,
      "grad_norm": 7.281436443328857,
      "learning_rate": 8.867739679814579e-08,
      "loss": 0.0823,
      "step": 18214
    },
    {
      "epoch": 0.9406145107152079,
      "grad_norm": 1.8114426136016846,
      "learning_rate": 8.852383226620743e-08,
      "loss": 0.017,
      "step": 18215
    },
    {
      "epoch": 0.9406661502711077,
      "grad_norm": 1.2010949850082397,
      "learning_rate": 8.837039962742267e-08,
      "loss": 0.014,
      "step": 18216
    },
    {
      "epoch": 0.9407177898270075,
      "grad_norm": 9.03003978729248,
      "learning_rate": 8.821709888591156e-08,
      "loss": 0.0838,
      "step": 18217
    },
    {
      "epoch": 0.9407694293829073,
      "grad_norm": 2.819490432739258,
      "learning_rate": 8.806393004579028e-08,
      "loss": 0.0541,
      "step": 18218
    },
    {
      "epoch": 0.9408210689388071,
      "grad_norm": 8.846904754638672,
      "learning_rate": 8.791089311117218e-08,
      "loss": 0.0764,
      "step": 18219
    },
    {
      "epoch": 0.9408727084947069,
      "grad_norm": 5.309942245483398,
      "learning_rate": 8.77579880861673e-08,
      "loss": 0.0564,
      "step": 18220
    },
    {
      "epoch": 0.9409243480506068,
      "grad_norm": 2.6333813667297363,
      "learning_rate": 8.760521497488184e-08,
      "loss": 0.0461,
      "step": 18221
    },
    {
      "epoch": 0.9409759876065066,
      "grad_norm": 14.919047355651855,
      "learning_rate": 8.745257378141802e-08,
      "loss": 0.0792,
      "step": 18222
    },
    {
      "epoch": 0.9410276271624064,
      "grad_norm": 6.122187614440918,
      "learning_rate": 8.730006450987539e-08,
      "loss": 0.0603,
      "step": 18223
    },
    {
      "epoch": 0.9410792667183062,
      "grad_norm": 2.825528383255005,
      "learning_rate": 8.714768716434897e-08,
      "loss": 0.026,
      "step": 18224
    },
    {
      "epoch": 0.941130906274206,
      "grad_norm": 4.799598693847656,
      "learning_rate": 8.699544174893048e-08,
      "loss": 0.0809,
      "step": 18225
    },
    {
      "epoch": 0.9411825458301059,
      "grad_norm": 4.011090278625488,
      "learning_rate": 8.684332826770891e-08,
      "loss": 0.0461,
      "step": 18226
    },
    {
      "epoch": 0.9412341853860057,
      "grad_norm": 4.391225814819336,
      "learning_rate": 8.669134672476876e-08,
      "loss": 0.0414,
      "step": 18227
    },
    {
      "epoch": 0.9412858249419055,
      "grad_norm": 2.8946638107299805,
      "learning_rate": 8.653949712419174e-08,
      "loss": 0.0373,
      "step": 18228
    },
    {
      "epoch": 0.9413374644978053,
      "grad_norm": 2.409773111343384,
      "learning_rate": 8.638777947005516e-08,
      "loss": 0.0409,
      "step": 18229
    },
    {
      "epoch": 0.9413891040537051,
      "grad_norm": 6.24342155456543,
      "learning_rate": 8.623619376643356e-08,
      "loss": 0.0357,
      "step": 18230
    },
    {
      "epoch": 0.941440743609605,
      "grad_norm": 2.3951938152313232,
      "learning_rate": 8.608474001739753e-08,
      "loss": 0.0437,
      "step": 18231
    },
    {
      "epoch": 0.9414923831655048,
      "grad_norm": 3.3017983436584473,
      "learning_rate": 8.59334182270144e-08,
      "loss": 0.0549,
      "step": 18232
    },
    {
      "epoch": 0.9415440227214046,
      "grad_norm": 15.771820068359375,
      "learning_rate": 8.578222839934758e-08,
      "loss": 0.1484,
      "step": 18233
    },
    {
      "epoch": 0.9415956622773044,
      "grad_norm": 19.461336135864258,
      "learning_rate": 8.56311705384566e-08,
      "loss": 0.0645,
      "step": 18234
    },
    {
      "epoch": 0.9416473018332042,
      "grad_norm": 2.1143605709075928,
      "learning_rate": 8.548024464839876e-08,
      "loss": 0.024,
      "step": 18235
    },
    {
      "epoch": 0.941698941389104,
      "grad_norm": 3.4218385219573975,
      "learning_rate": 8.532945073322752e-08,
      "loss": 0.0471,
      "step": 18236
    },
    {
      "epoch": 0.9417505809450039,
      "grad_norm": 3.0627033710479736,
      "learning_rate": 8.517878879699127e-08,
      "loss": 0.051,
      "step": 18237
    },
    {
      "epoch": 0.9418022205009037,
      "grad_norm": 2.602259397506714,
      "learning_rate": 8.502825884373622e-08,
      "loss": 0.0363,
      "step": 18238
    },
    {
      "epoch": 0.9418538600568035,
      "grad_norm": 4.073053359985352,
      "learning_rate": 8.487786087750416e-08,
      "loss": 0.0763,
      "step": 18239
    },
    {
      "epoch": 0.9419054996127033,
      "grad_norm": 3.330080986022949,
      "learning_rate": 8.472759490233517e-08,
      "loss": 0.0557,
      "step": 18240
    },
    {
      "epoch": 0.9419571391686031,
      "grad_norm": 5.512397289276123,
      "learning_rate": 8.457746092226382e-08,
      "loss": 0.0275,
      "step": 18241
    },
    {
      "epoch": 0.942008778724503,
      "grad_norm": 2.631336212158203,
      "learning_rate": 8.442745894132132e-08,
      "loss": 0.0415,
      "step": 18242
    },
    {
      "epoch": 0.9420604182804028,
      "grad_norm": 8.369905471801758,
      "learning_rate": 8.427758896353666e-08,
      "loss": 0.1239,
      "step": 18243
    },
    {
      "epoch": 0.9421120578363026,
      "grad_norm": 2.7321319580078125,
      "learning_rate": 8.412785099293385e-08,
      "loss": 0.0366,
      "step": 18244
    },
    {
      "epoch": 0.9421636973922024,
      "grad_norm": 2.915330410003662,
      "learning_rate": 8.397824503353413e-08,
      "loss": 0.0382,
      "step": 18245
    },
    {
      "epoch": 0.9422153369481022,
      "grad_norm": 2.4564104080200195,
      "learning_rate": 8.382877108935539e-08,
      "loss": 0.0499,
      "step": 18246
    },
    {
      "epoch": 0.942266976504002,
      "grad_norm": 2.9697349071502686,
      "learning_rate": 8.367942916441219e-08,
      "loss": 0.032,
      "step": 18247
    },
    {
      "epoch": 0.9423186160599019,
      "grad_norm": 9.9804048538208,
      "learning_rate": 8.353021926271242e-08,
      "loss": 0.0486,
      "step": 18248
    },
    {
      "epoch": 0.9423702556158017,
      "grad_norm": 4.524704933166504,
      "learning_rate": 8.33811413882657e-08,
      "loss": 0.0475,
      "step": 18249
    },
    {
      "epoch": 0.9424218951717015,
      "grad_norm": 2.4163269996643066,
      "learning_rate": 8.323219554507489e-08,
      "loss": 0.0287,
      "step": 18250
    },
    {
      "epoch": 0.9424735347276013,
      "grad_norm": 4.556933403015137,
      "learning_rate": 8.308338173713848e-08,
      "loss": 0.0717,
      "step": 18251
    },
    {
      "epoch": 0.9425251742835011,
      "grad_norm": 18.959056854248047,
      "learning_rate": 8.293469996845327e-08,
      "loss": 0.1467,
      "step": 18252
    },
    {
      "epoch": 0.942576813839401,
      "grad_norm": 3.982618570327759,
      "learning_rate": 8.278615024301218e-08,
      "loss": 0.0404,
      "step": 18253
    },
    {
      "epoch": 0.9426284533953008,
      "grad_norm": 2.2382471561431885,
      "learning_rate": 8.263773256480534e-08,
      "loss": 0.0231,
      "step": 18254
    },
    {
      "epoch": 0.9426800929512006,
      "grad_norm": 2.130481004714966,
      "learning_rate": 8.24894469378168e-08,
      "loss": 0.0203,
      "step": 18255
    },
    {
      "epoch": 0.9427317325071004,
      "grad_norm": 11.829090118408203,
      "learning_rate": 8.234129336602947e-08,
      "loss": 0.1058,
      "step": 18256
    },
    {
      "epoch": 0.9427833720630002,
      "grad_norm": 16.070531845092773,
      "learning_rate": 8.21932718534213e-08,
      "loss": 0.1281,
      "step": 18257
    },
    {
      "epoch": 0.9428350116189,
      "grad_norm": 2.147430419921875,
      "learning_rate": 8.204538240396797e-08,
      "loss": 0.0554,
      "step": 18258
    },
    {
      "epoch": 0.9428866511747999,
      "grad_norm": 2.2909796237945557,
      "learning_rate": 8.189762502164078e-08,
      "loss": 0.0366,
      "step": 18259
    },
    {
      "epoch": 0.9429382907306997,
      "grad_norm": 15.49622631072998,
      "learning_rate": 8.174999971040709e-08,
      "loss": 0.131,
      "step": 18260
    },
    {
      "epoch": 0.9429899302865995,
      "grad_norm": 12.41664981842041,
      "learning_rate": 8.160250647423207e-08,
      "loss": 0.0813,
      "step": 18261
    },
    {
      "epoch": 0.9430415698424993,
      "grad_norm": 6.893320560455322,
      "learning_rate": 8.145514531707533e-08,
      "loss": 0.0606,
      "step": 18262
    },
    {
      "epoch": 0.9430932093983991,
      "grad_norm": 2.0224924087524414,
      "learning_rate": 8.130791624289536e-08,
      "loss": 0.0365,
      "step": 18263
    },
    {
      "epoch": 0.943144848954299,
      "grad_norm": 4.535565376281738,
      "learning_rate": 8.116081925564456e-08,
      "loss": 0.1152,
      "step": 18264
    },
    {
      "epoch": 0.9431964885101988,
      "grad_norm": 2.711691379547119,
      "learning_rate": 8.101385435927478e-08,
      "loss": 0.0542,
      "step": 18265
    },
    {
      "epoch": 0.9432481280660986,
      "grad_norm": 3.103959083557129,
      "learning_rate": 8.086702155773174e-08,
      "loss": 0.0442,
      "step": 18266
    },
    {
      "epoch": 0.9432997676219984,
      "grad_norm": 13.616941452026367,
      "learning_rate": 8.072032085495729e-08,
      "loss": 0.1597,
      "step": 18267
    },
    {
      "epoch": 0.9433514071778982,
      "grad_norm": 5.5990753173828125,
      "learning_rate": 8.057375225489327e-08,
      "loss": 0.1289,
      "step": 18268
    },
    {
      "epoch": 0.943403046733798,
      "grad_norm": 2.580015182495117,
      "learning_rate": 8.042731576147434e-08,
      "loss": 0.0435,
      "step": 18269
    },
    {
      "epoch": 0.9434546862896979,
      "grad_norm": 2.9542415142059326,
      "learning_rate": 8.02810113786323e-08,
      "loss": 0.0306,
      "step": 18270
    },
    {
      "epoch": 0.9435063258455977,
      "grad_norm": 1.3372937440872192,
      "learning_rate": 8.013483911029741e-08,
      "loss": 0.0101,
      "step": 18271
    },
    {
      "epoch": 0.9435579654014975,
      "grad_norm": 4.745906829833984,
      "learning_rate": 7.998879896039424e-08,
      "loss": 0.0368,
      "step": 18272
    },
    {
      "epoch": 0.9436096049573973,
      "grad_norm": 9.913569450378418,
      "learning_rate": 7.984289093284414e-08,
      "loss": 0.0472,
      "step": 18273
    },
    {
      "epoch": 0.9436612445132971,
      "grad_norm": 2.2222774028778076,
      "learning_rate": 7.969711503156674e-08,
      "loss": 0.0205,
      "step": 18274
    },
    {
      "epoch": 0.943712884069197,
      "grad_norm": 1.5493844747543335,
      "learning_rate": 7.955147126047502e-08,
      "loss": 0.0207,
      "step": 18275
    },
    {
      "epoch": 0.9437645236250968,
      "grad_norm": 3.96061635017395,
      "learning_rate": 7.94059596234814e-08,
      "loss": 0.0309,
      "step": 18276
    },
    {
      "epoch": 0.9438161631809966,
      "grad_norm": 2.605435371398926,
      "learning_rate": 7.926058012449278e-08,
      "loss": 0.0356,
      "step": 18277
    },
    {
      "epoch": 0.9438678027368964,
      "grad_norm": 3.472771406173706,
      "learning_rate": 7.911533276741379e-08,
      "loss": 0.0504,
      "step": 18278
    },
    {
      "epoch": 0.9439194422927962,
      "grad_norm": 1.6918092966079712,
      "learning_rate": 7.897021755614465e-08,
      "loss": 0.0256,
      "step": 18279
    },
    {
      "epoch": 0.943971081848696,
      "grad_norm": 18.529972076416016,
      "learning_rate": 7.882523449458169e-08,
      "loss": 0.1275,
      "step": 18280
    },
    {
      "epoch": 0.9440227214045959,
      "grad_norm": 4.22249698638916,
      "learning_rate": 7.868038358661845e-08,
      "loss": 0.0503,
      "step": 18281
    },
    {
      "epoch": 0.9440743609604957,
      "grad_norm": 7.612735748291016,
      "learning_rate": 7.853566483614572e-08,
      "loss": 0.0969,
      "step": 18282
    },
    {
      "epoch": 0.9441260005163956,
      "grad_norm": 3.437147855758667,
      "learning_rate": 7.839107824704928e-08,
      "loss": 0.0419,
      "step": 18283
    },
    {
      "epoch": 0.9441776400722954,
      "grad_norm": 2.353652238845825,
      "learning_rate": 7.824662382321157e-08,
      "loss": 0.0406,
      "step": 18284
    },
    {
      "epoch": 0.9442292796281953,
      "grad_norm": 3.3343453407287598,
      "learning_rate": 7.810230156851173e-08,
      "loss": 0.055,
      "step": 18285
    },
    {
      "epoch": 0.9442809191840951,
      "grad_norm": 20.011213302612305,
      "learning_rate": 7.795811148682552e-08,
      "loss": 0.198,
      "step": 18286
    },
    {
      "epoch": 0.9443325587399949,
      "grad_norm": 4.953403472900391,
      "learning_rate": 7.781405358202599e-08,
      "loss": 0.0844,
      "step": 18287
    },
    {
      "epoch": 0.9443841982958947,
      "grad_norm": 3.5187745094299316,
      "learning_rate": 7.767012785798e-08,
      "loss": 0.0329,
      "step": 18288
    },
    {
      "epoch": 0.9444358378517945,
      "grad_norm": 4.929616928100586,
      "learning_rate": 7.752633431855395e-08,
      "loss": 0.0458,
      "step": 18289
    },
    {
      "epoch": 0.9444874774076943,
      "grad_norm": 1.8027609586715698,
      "learning_rate": 7.738267296760805e-08,
      "loss": 0.0117,
      "step": 18290
    },
    {
      "epoch": 0.9445391169635942,
      "grad_norm": 2.92971134185791,
      "learning_rate": 7.723914380900144e-08,
      "loss": 0.0686,
      "step": 18291
    },
    {
      "epoch": 0.944590756519494,
      "grad_norm": 12.633201599121094,
      "learning_rate": 7.709574684658772e-08,
      "loss": 0.1037,
      "step": 18292
    },
    {
      "epoch": 0.9446423960753938,
      "grad_norm": 9.396183013916016,
      "learning_rate": 7.695248208421768e-08,
      "loss": 0.1214,
      "step": 18293
    },
    {
      "epoch": 0.9446940356312936,
      "grad_norm": 3.6528289318084717,
      "learning_rate": 7.680934952573938e-08,
      "loss": 0.0398,
      "step": 18294
    },
    {
      "epoch": 0.9447456751871934,
      "grad_norm": 6.247204780578613,
      "learning_rate": 7.666634917499472e-08,
      "loss": 0.0744,
      "step": 18295
    },
    {
      "epoch": 0.9447973147430933,
      "grad_norm": 6.220203399658203,
      "learning_rate": 7.652348103582563e-08,
      "loss": 0.1154,
      "step": 18296
    },
    {
      "epoch": 0.9448489542989931,
      "grad_norm": 2.0136587619781494,
      "learning_rate": 7.638074511206794e-08,
      "loss": 0.0303,
      "step": 18297
    },
    {
      "epoch": 0.9449005938548929,
      "grad_norm": 1.7430331707000732,
      "learning_rate": 7.623814140755526e-08,
      "loss": 0.0205,
      "step": 18298
    },
    {
      "epoch": 0.9449522334107927,
      "grad_norm": 2.938009262084961,
      "learning_rate": 7.609566992611616e-08,
      "loss": 0.0534,
      "step": 18299
    },
    {
      "epoch": 0.9450038729666925,
      "grad_norm": 7.03842830657959,
      "learning_rate": 7.595333067157706e-08,
      "loss": 0.0822,
      "step": 18300
    },
    {
      "epoch": 0.9450555125225923,
      "grad_norm": 8.336853981018066,
      "learning_rate": 7.581112364776044e-08,
      "loss": 0.0689,
      "step": 18301
    },
    {
      "epoch": 0.9451071520784922,
      "grad_norm": 4.321858882904053,
      "learning_rate": 7.566904885848491e-08,
      "loss": 0.0859,
      "step": 18302
    },
    {
      "epoch": 0.945158791634392,
      "grad_norm": 12.286932945251465,
      "learning_rate": 7.552710630756632e-08,
      "loss": 0.1205,
      "step": 18303
    },
    {
      "epoch": 0.9452104311902918,
      "grad_norm": 2.303405523300171,
      "learning_rate": 7.53852959988155e-08,
      "loss": 0.0391,
      "step": 18304
    },
    {
      "epoch": 0.9452620707461916,
      "grad_norm": 3.369788408279419,
      "learning_rate": 7.524361793604107e-08,
      "loss": 0.0529,
      "step": 18305
    },
    {
      "epoch": 0.9453137103020914,
      "grad_norm": 7.105695724487305,
      "learning_rate": 7.510207212304776e-08,
      "loss": 0.0739,
      "step": 18306
    },
    {
      "epoch": 0.9453653498579913,
      "grad_norm": 1.080240249633789,
      "learning_rate": 7.496065856363699e-08,
      "loss": 0.0131,
      "step": 18307
    },
    {
      "epoch": 0.9454169894138911,
      "grad_norm": 3.829648017883301,
      "learning_rate": 7.481937726160571e-08,
      "loss": 0.0654,
      "step": 18308
    },
    {
      "epoch": 0.9454686289697909,
      "grad_norm": 11.337993621826172,
      "learning_rate": 7.467822822074755e-08,
      "loss": 0.0548,
      "step": 18309
    },
    {
      "epoch": 0.9455202685256907,
      "grad_norm": 2.337125062942505,
      "learning_rate": 7.453721144485393e-08,
      "loss": 0.0223,
      "step": 18310
    },
    {
      "epoch": 0.9455719080815905,
      "grad_norm": 2.4106764793395996,
      "learning_rate": 7.439632693771126e-08,
      "loss": 0.0262,
      "step": 18311
    },
    {
      "epoch": 0.9456235476374903,
      "grad_norm": 3.504324197769165,
      "learning_rate": 7.425557470310318e-08,
      "loss": 0.0371,
      "step": 18312
    },
    {
      "epoch": 0.9456751871933902,
      "grad_norm": 4.10626745223999,
      "learning_rate": 7.411495474480834e-08,
      "loss": 0.0596,
      "step": 18313
    },
    {
      "epoch": 0.94572682674929,
      "grad_norm": 7.401426315307617,
      "learning_rate": 7.397446706660428e-08,
      "loss": 0.0679,
      "step": 18314
    },
    {
      "epoch": 0.9457784663051898,
      "grad_norm": 5.19426965713501,
      "learning_rate": 7.383411167226295e-08,
      "loss": 0.1005,
      "step": 18315
    },
    {
      "epoch": 0.9458301058610896,
      "grad_norm": 4.844019889831543,
      "learning_rate": 7.369388856555415e-08,
      "loss": 0.0562,
      "step": 18316
    },
    {
      "epoch": 0.9458817454169894,
      "grad_norm": 5.21873664855957,
      "learning_rate": 7.355379775024207e-08,
      "loss": 0.0723,
      "step": 18317
    },
    {
      "epoch": 0.9459333849728893,
      "grad_norm": 12.30720043182373,
      "learning_rate": 7.341383923009038e-08,
      "loss": 0.19,
      "step": 18318
    },
    {
      "epoch": 0.9459850245287891,
      "grad_norm": 3.222283363342285,
      "learning_rate": 7.327401300885607e-08,
      "loss": 0.0275,
      "step": 18319
    },
    {
      "epoch": 0.9460366640846889,
      "grad_norm": 1.7743743658065796,
      "learning_rate": 7.313431909029556e-08,
      "loss": 0.0355,
      "step": 18320
    },
    {
      "epoch": 0.9460883036405887,
      "grad_norm": 2.44742488861084,
      "learning_rate": 7.299475747815865e-08,
      "loss": 0.0518,
      "step": 18321
    },
    {
      "epoch": 0.9461399431964885,
      "grad_norm": 7.248490333557129,
      "learning_rate": 7.285532817619456e-08,
      "loss": 0.0544,
      "step": 18322
    },
    {
      "epoch": 0.9461915827523883,
      "grad_norm": 3.5527913570404053,
      "learning_rate": 7.271603118814641e-08,
      "loss": 0.0292,
      "step": 18323
    },
    {
      "epoch": 0.9462432223082882,
      "grad_norm": 7.344142913818359,
      "learning_rate": 7.257686651775509e-08,
      "loss": 0.0631,
      "step": 18324
    },
    {
      "epoch": 0.946294861864188,
      "grad_norm": 6.345138072967529,
      "learning_rate": 7.243783416875816e-08,
      "loss": 0.0529,
      "step": 18325
    },
    {
      "epoch": 0.9463465014200878,
      "grad_norm": 3.1771490573883057,
      "learning_rate": 7.229893414488932e-08,
      "loss": 0.043,
      "step": 18326
    },
    {
      "epoch": 0.9463981409759876,
      "grad_norm": 1.834535002708435,
      "learning_rate": 7.216016644987778e-08,
      "loss": 0.0391,
      "step": 18327
    },
    {
      "epoch": 0.9464497805318874,
      "grad_norm": 2.9761574268341064,
      "learning_rate": 7.202153108745114e-08,
      "loss": 0.0363,
      "step": 18328
    },
    {
      "epoch": 0.9465014200877873,
      "grad_norm": 2.729055643081665,
      "learning_rate": 7.188302806133141e-08,
      "loss": 0.0318,
      "step": 18329
    },
    {
      "epoch": 0.9465530596436871,
      "grad_norm": 8.28593635559082,
      "learning_rate": 7.174465737523783e-08,
      "loss": 0.063,
      "step": 18330
    },
    {
      "epoch": 0.9466046991995869,
      "grad_norm": 3.5362446308135986,
      "learning_rate": 7.160641903288745e-08,
      "loss": 0.0492,
      "step": 18331
    },
    {
      "epoch": 0.9466563387554867,
      "grad_norm": 8.333283424377441,
      "learning_rate": 7.146831303799173e-08,
      "loss": 0.1149,
      "step": 18332
    },
    {
      "epoch": 0.9467079783113865,
      "grad_norm": 2.778050661087036,
      "learning_rate": 7.133033939425826e-08,
      "loss": 0.039,
      "step": 18333
    },
    {
      "epoch": 0.9467596178672864,
      "grad_norm": 18.042388916015625,
      "learning_rate": 7.119249810539464e-08,
      "loss": 0.1722,
      "step": 18334
    },
    {
      "epoch": 0.9468112574231862,
      "grad_norm": 11.5087308883667,
      "learning_rate": 7.105478917510067e-08,
      "loss": 0.0657,
      "step": 18335
    },
    {
      "epoch": 0.946862896979086,
      "grad_norm": 3.4057605266571045,
      "learning_rate": 7.091721260707451e-08,
      "loss": 0.0661,
      "step": 18336
    },
    {
      "epoch": 0.9469145365349858,
      "grad_norm": 4.513022422790527,
      "learning_rate": 7.077976840501155e-08,
      "loss": 0.0376,
      "step": 18337
    },
    {
      "epoch": 0.9469661760908856,
      "grad_norm": 2.365980625152588,
      "learning_rate": 7.064245657260215e-08,
      "loss": 0.0208,
      "step": 18338
    },
    {
      "epoch": 0.9470178156467854,
      "grad_norm": 2.850050449371338,
      "learning_rate": 7.050527711353339e-08,
      "loss": 0.0233,
      "step": 18339
    },
    {
      "epoch": 0.9470694552026853,
      "grad_norm": 2.3564345836639404,
      "learning_rate": 7.036823003148952e-08,
      "loss": 0.0213,
      "step": 18340
    },
    {
      "epoch": 0.9471210947585851,
      "grad_norm": 4.166586399078369,
      "learning_rate": 7.023131533015092e-08,
      "loss": 0.0235,
      "step": 18341
    },
    {
      "epoch": 0.9471727343144849,
      "grad_norm": 2.5708208084106445,
      "learning_rate": 7.009453301319413e-08,
      "loss": 0.0445,
      "step": 18342
    },
    {
      "epoch": 0.9472243738703847,
      "grad_norm": 14.082508087158203,
      "learning_rate": 6.99578830842923e-08,
      "loss": 0.0991,
      "step": 18343
    },
    {
      "epoch": 0.9472760134262845,
      "grad_norm": 3.98872709274292,
      "learning_rate": 6.982136554711471e-08,
      "loss": 0.0215,
      "step": 18344
    },
    {
      "epoch": 0.9473276529821844,
      "grad_norm": 8.853179931640625,
      "learning_rate": 6.968498040532845e-08,
      "loss": 0.0563,
      "step": 18345
    },
    {
      "epoch": 0.9473792925380842,
      "grad_norm": 2.0011045932769775,
      "learning_rate": 6.954872766259502e-08,
      "loss": 0.0219,
      "step": 18346
    },
    {
      "epoch": 0.947430932093984,
      "grad_norm": 1.590083122253418,
      "learning_rate": 6.941260732257315e-08,
      "loss": 0.0203,
      "step": 18347
    },
    {
      "epoch": 0.9474825716498838,
      "grad_norm": 3.3710312843322754,
      "learning_rate": 6.927661938891883e-08,
      "loss": 0.0456,
      "step": 18348
    },
    {
      "epoch": 0.9475342112057836,
      "grad_norm": 2.2705299854278564,
      "learning_rate": 6.914076386528468e-08,
      "loss": 0.0459,
      "step": 18349
    },
    {
      "epoch": 0.9475858507616834,
      "grad_norm": 1.651000738143921,
      "learning_rate": 6.900504075531723e-08,
      "loss": 0.0324,
      "step": 18350
    },
    {
      "epoch": 0.9476374903175833,
      "grad_norm": 6.384292125701904,
      "learning_rate": 6.886945006266244e-08,
      "loss": 0.0736,
      "step": 18351
    },
    {
      "epoch": 0.9476891298734831,
      "grad_norm": 3.3478457927703857,
      "learning_rate": 6.873399179096073e-08,
      "loss": 0.0382,
      "step": 18352
    },
    {
      "epoch": 0.9477407694293829,
      "grad_norm": 8.621712684631348,
      "learning_rate": 6.859866594385034e-08,
      "loss": 0.0475,
      "step": 18353
    },
    {
      "epoch": 0.9477924089852827,
      "grad_norm": 5.483999252319336,
      "learning_rate": 6.846347252496499e-08,
      "loss": 0.035,
      "step": 18354
    },
    {
      "epoch": 0.9478440485411825,
      "grad_norm": 2.742504119873047,
      "learning_rate": 6.832841153793568e-08,
      "loss": 0.0491,
      "step": 18355
    },
    {
      "epoch": 0.9478956880970824,
      "grad_norm": 2.2790074348449707,
      "learning_rate": 6.819348298638839e-08,
      "loss": 0.0196,
      "step": 18356
    },
    {
      "epoch": 0.9479473276529822,
      "grad_norm": 26.406442642211914,
      "learning_rate": 6.80586868739469e-08,
      "loss": 0.093,
      "step": 18357
    },
    {
      "epoch": 0.947998967208882,
      "grad_norm": 1.910752296447754,
      "learning_rate": 6.792402320423163e-08,
      "loss": 0.0182,
      "step": 18358
    },
    {
      "epoch": 0.9480506067647818,
      "grad_norm": 8.189854621887207,
      "learning_rate": 6.778949198085805e-08,
      "loss": 0.0649,
      "step": 18359
    },
    {
      "epoch": 0.9481022463206816,
      "grad_norm": 2.73127818107605,
      "learning_rate": 6.765509320743879e-08,
      "loss": 0.046,
      "step": 18360
    },
    {
      "epoch": 0.9481538858765814,
      "grad_norm": 3.909329891204834,
      "learning_rate": 6.752082688758432e-08,
      "loss": 0.0427,
      "step": 18361
    },
    {
      "epoch": 0.9482055254324813,
      "grad_norm": 5.559345722198486,
      "learning_rate": 6.738669302489842e-08,
      "loss": 0.0571,
      "step": 18362
    },
    {
      "epoch": 0.9482571649883811,
      "grad_norm": 2.7117273807525635,
      "learning_rate": 6.725269162298486e-08,
      "loss": 0.0639,
      "step": 18363
    },
    {
      "epoch": 0.9483088045442809,
      "grad_norm": 1.8093165159225464,
      "learning_rate": 6.711882268544079e-08,
      "loss": 0.0155,
      "step": 18364
    },
    {
      "epoch": 0.9483604441001807,
      "grad_norm": 3.0903029441833496,
      "learning_rate": 6.69850862158622e-08,
      "loss": 0.0436,
      "step": 18365
    },
    {
      "epoch": 0.9484120836560805,
      "grad_norm": 3.598454713821411,
      "learning_rate": 6.685148221783899e-08,
      "loss": 0.0303,
      "step": 18366
    },
    {
      "epoch": 0.9484637232119804,
      "grad_norm": 4.9358720779418945,
      "learning_rate": 6.671801069496053e-08,
      "loss": 0.0736,
      "step": 18367
    },
    {
      "epoch": 0.9485153627678802,
      "grad_norm": 2.7637195587158203,
      "learning_rate": 6.658467165081006e-08,
      "loss": 0.0583,
      "step": 18368
    },
    {
      "epoch": 0.94856700232378,
      "grad_norm": 1.8163954019546509,
      "learning_rate": 6.645146508896916e-08,
      "loss": 0.0183,
      "step": 18369
    },
    {
      "epoch": 0.9486186418796798,
      "grad_norm": 3.9800329208374023,
      "learning_rate": 6.631839101301385e-08,
      "loss": 0.0453,
      "step": 18370
    },
    {
      "epoch": 0.9486702814355796,
      "grad_norm": 7.118109226226807,
      "learning_rate": 6.61854494265185e-08,
      "loss": 0.0664,
      "step": 18371
    },
    {
      "epoch": 0.9487219209914795,
      "grad_norm": 15.915522575378418,
      "learning_rate": 6.605264033305303e-08,
      "loss": 0.062,
      "step": 18372
    },
    {
      "epoch": 0.9487735605473793,
      "grad_norm": 3.9850854873657227,
      "learning_rate": 6.591996373618403e-08,
      "loss": 0.0391,
      "step": 18373
    },
    {
      "epoch": 0.9488252001032791,
      "grad_norm": 0.9923912882804871,
      "learning_rate": 6.578741963947365e-08,
      "loss": 0.0085,
      "step": 18374
    },
    {
      "epoch": 0.9488768396591789,
      "grad_norm": 7.109520435333252,
      "learning_rate": 6.565500804648239e-08,
      "loss": 0.0757,
      "step": 18375
    },
    {
      "epoch": 0.9489284792150787,
      "grad_norm": 3.953666925430298,
      "learning_rate": 6.552272896076518e-08,
      "loss": 0.0446,
      "step": 18376
    },
    {
      "epoch": 0.9489801187709785,
      "grad_norm": 14.718595504760742,
      "learning_rate": 6.53905823858747e-08,
      "loss": 0.0679,
      "step": 18377
    },
    {
      "epoch": 0.9490317583268784,
      "grad_norm": 4.209982395172119,
      "learning_rate": 6.525856832535926e-08,
      "loss": 0.0425,
      "step": 18378
    },
    {
      "epoch": 0.9490833978827782,
      "grad_norm": 2.857417106628418,
      "learning_rate": 6.512668678276379e-08,
      "loss": 0.0161,
      "step": 18379
    },
    {
      "epoch": 0.949135037438678,
      "grad_norm": 7.4703264236450195,
      "learning_rate": 6.499493776163101e-08,
      "loss": 0.035,
      "step": 18380
    },
    {
      "epoch": 0.9491866769945778,
      "grad_norm": 3.688330888748169,
      "learning_rate": 6.486332126549754e-08,
      "loss": 0.0311,
      "step": 18381
    },
    {
      "epoch": 0.9492383165504776,
      "grad_norm": 4.645468711853027,
      "learning_rate": 6.47318372978989e-08,
      "loss": 0.045,
      "step": 18382
    },
    {
      "epoch": 0.9492899561063775,
      "grad_norm": 4.556355953216553,
      "learning_rate": 6.4600485862365e-08,
      "loss": 0.0429,
      "step": 18383
    },
    {
      "epoch": 0.9493415956622773,
      "grad_norm": 2.159410238265991,
      "learning_rate": 6.446926696242362e-08,
      "loss": 0.0242,
      "step": 18384
    },
    {
      "epoch": 0.9493932352181771,
      "grad_norm": 9.382959365844727,
      "learning_rate": 6.433818060159858e-08,
      "loss": 0.1143,
      "step": 18385
    },
    {
      "epoch": 0.9494448747740769,
      "grad_norm": 17.37484359741211,
      "learning_rate": 6.420722678341041e-08,
      "loss": 0.1832,
      "step": 18386
    },
    {
      "epoch": 0.9494965143299767,
      "grad_norm": 2.6866390705108643,
      "learning_rate": 6.407640551137462e-08,
      "loss": 0.0526,
      "step": 18387
    },
    {
      "epoch": 0.9495481538858765,
      "grad_norm": 4.442991256713867,
      "learning_rate": 6.394571678900563e-08,
      "loss": 0.0531,
      "step": 18388
    },
    {
      "epoch": 0.9495997934417764,
      "grad_norm": 5.1302595138549805,
      "learning_rate": 6.38151606198123e-08,
      "loss": 0.0887,
      "step": 18389
    },
    {
      "epoch": 0.9496514329976762,
      "grad_norm": 2.8716371059417725,
      "learning_rate": 6.368473700730071e-08,
      "loss": 0.0299,
      "step": 18390
    },
    {
      "epoch": 0.949703072553576,
      "grad_norm": 4.531135559082031,
      "learning_rate": 6.355444595497307e-08,
      "loss": 0.0495,
      "step": 18391
    },
    {
      "epoch": 0.9497547121094758,
      "grad_norm": 9.70846939086914,
      "learning_rate": 6.342428746632878e-08,
      "loss": 0.0792,
      "step": 18392
    },
    {
      "epoch": 0.9498063516653756,
      "grad_norm": 2.212613105773926,
      "learning_rate": 6.329426154486229e-08,
      "loss": 0.0125,
      "step": 18393
    },
    {
      "epoch": 0.9498579912212755,
      "grad_norm": 2.831609010696411,
      "learning_rate": 6.316436819406579e-08,
      "loss": 0.0298,
      "step": 18394
    },
    {
      "epoch": 0.9499096307771753,
      "grad_norm": 2.51127290725708,
      "learning_rate": 6.303460741742762e-08,
      "loss": 0.0296,
      "step": 18395
    },
    {
      "epoch": 0.9499612703330751,
      "grad_norm": 2.724092721939087,
      "learning_rate": 6.290497921843219e-08,
      "loss": 0.0334,
      "step": 18396
    },
    {
      "epoch": 0.9500129098889749,
      "grad_norm": 3.5458824634552,
      "learning_rate": 6.277548360056118e-08,
      "loss": 0.0607,
      "step": 18397
    },
    {
      "epoch": 0.9500645494448747,
      "grad_norm": 9.138142585754395,
      "learning_rate": 6.264612056729014e-08,
      "loss": 0.0569,
      "step": 18398
    },
    {
      "epoch": 0.9501161890007745,
      "grad_norm": 7.726706504821777,
      "learning_rate": 6.251689012209517e-08,
      "loss": 0.1747,
      "step": 18399
    },
    {
      "epoch": 0.9501678285566744,
      "grad_norm": 17.480485916137695,
      "learning_rate": 6.238779226844571e-08,
      "loss": 0.092,
      "step": 18400
    },
    {
      "epoch": 0.9502194681125742,
      "grad_norm": 3.993452787399292,
      "learning_rate": 6.225882700980845e-08,
      "loss": 0.0708,
      "step": 18401
    },
    {
      "epoch": 0.950271107668474,
      "grad_norm": 5.131587505340576,
      "learning_rate": 6.212999434964672e-08,
      "loss": 0.058,
      "step": 18402
    },
    {
      "epoch": 0.9503227472243738,
      "grad_norm": 1.9086582660675049,
      "learning_rate": 6.200129429142054e-08,
      "loss": 0.0304,
      "step": 18403
    },
    {
      "epoch": 0.9503743867802736,
      "grad_norm": 2.531420946121216,
      "learning_rate": 6.187272683858547e-08,
      "loss": 0.0277,
      "step": 18404
    },
    {
      "epoch": 0.9504260263361735,
      "grad_norm": 10.054430961608887,
      "learning_rate": 6.174429199459486e-08,
      "loss": 0.0798,
      "step": 18405
    },
    {
      "epoch": 0.9504776658920733,
      "grad_norm": 4.694578170776367,
      "learning_rate": 6.161598976289707e-08,
      "loss": 0.0588,
      "step": 18406
    },
    {
      "epoch": 0.9505293054479731,
      "grad_norm": 3.5766966342926025,
      "learning_rate": 6.14878201469371e-08,
      "loss": 0.0628,
      "step": 18407
    },
    {
      "epoch": 0.9505809450038729,
      "grad_norm": 3.313063859939575,
      "learning_rate": 6.135978315015779e-08,
      "loss": 0.0618,
      "step": 18408
    },
    {
      "epoch": 0.9506325845597727,
      "grad_norm": 9.314468383789062,
      "learning_rate": 6.123187877599745e-08,
      "loss": 0.0784,
      "step": 18409
    },
    {
      "epoch": 0.9506842241156727,
      "grad_norm": 2.8429527282714844,
      "learning_rate": 6.110410702789005e-08,
      "loss": 0.0359,
      "step": 18410
    },
    {
      "epoch": 0.9507358636715725,
      "grad_norm": 1.4247180223464966,
      "learning_rate": 6.097646790926781e-08,
      "loss": 0.0237,
      "step": 18411
    },
    {
      "epoch": 0.9507875032274723,
      "grad_norm": 2.1509628295898438,
      "learning_rate": 6.08489614235569e-08,
      "loss": 0.0135,
      "step": 18412
    },
    {
      "epoch": 0.9508391427833721,
      "grad_norm": 4.629695892333984,
      "learning_rate": 6.072158757418234e-08,
      "loss": 0.0511,
      "step": 18413
    },
    {
      "epoch": 0.9508907823392719,
      "grad_norm": 2.0337026119232178,
      "learning_rate": 6.059434636456473e-08,
      "loss": 0.0081,
      "step": 18414
    },
    {
      "epoch": 0.9509424218951718,
      "grad_norm": 8.457499504089355,
      "learning_rate": 6.046723779812135e-08,
      "loss": 0.0463,
      "step": 18415
    },
    {
      "epoch": 0.9509940614510716,
      "grad_norm": 3.8889102935791016,
      "learning_rate": 6.034026187826503e-08,
      "loss": 0.0469,
      "step": 18416
    },
    {
      "epoch": 0.9510457010069714,
      "grad_norm": 1.4818620681762695,
      "learning_rate": 6.021341860840469e-08,
      "loss": 0.0091,
      "step": 18417
    },
    {
      "epoch": 0.9510973405628712,
      "grad_norm": 1.9982415437698364,
      "learning_rate": 6.008670799194816e-08,
      "loss": 0.0331,
      "step": 18418
    },
    {
      "epoch": 0.951148980118771,
      "grad_norm": 1.8013617992401123,
      "learning_rate": 5.996013003229717e-08,
      "loss": 0.0245,
      "step": 18419
    },
    {
      "epoch": 0.9512006196746708,
      "grad_norm": 1.8453885316848755,
      "learning_rate": 5.983368473285178e-08,
      "loss": 0.0281,
      "step": 18420
    },
    {
      "epoch": 0.9512522592305707,
      "grad_norm": 2.1989450454711914,
      "learning_rate": 5.97073720970065e-08,
      "loss": 0.0214,
      "step": 18421
    },
    {
      "epoch": 0.9513038987864705,
      "grad_norm": 26.312496185302734,
      "learning_rate": 5.958119212815361e-08,
      "loss": 0.0633,
      "step": 18422
    },
    {
      "epoch": 0.9513555383423703,
      "grad_norm": 5.686659336090088,
      "learning_rate": 5.9455144829682065e-08,
      "loss": 0.0692,
      "step": 18423
    },
    {
      "epoch": 0.9514071778982701,
      "grad_norm": 10.613699913024902,
      "learning_rate": 5.9329230204976383e-08,
      "loss": 0.0624,
      "step": 18424
    },
    {
      "epoch": 0.9514588174541699,
      "grad_norm": 4.50939416885376,
      "learning_rate": 5.920344825741775e-08,
      "loss": 0.0247,
      "step": 18425
    },
    {
      "epoch": 0.9515104570100698,
      "grad_norm": 3.686312437057495,
      "learning_rate": 5.9077798990384575e-08,
      "loss": 0.0268,
      "step": 18426
    },
    {
      "epoch": 0.9515620965659696,
      "grad_norm": 4.848339080810547,
      "learning_rate": 5.8952282407249703e-08,
      "loss": 0.0682,
      "step": 18427
    },
    {
      "epoch": 0.9516137361218694,
      "grad_norm": 2.0697433948516846,
      "learning_rate": 5.8826898511385457e-08,
      "loss": 0.0146,
      "step": 18428
    },
    {
      "epoch": 0.9516653756777692,
      "grad_norm": 7.872291564941406,
      "learning_rate": 5.8701647306158016e-08,
      "loss": 0.0908,
      "step": 18429
    },
    {
      "epoch": 0.951717015233669,
      "grad_norm": 4.09749174118042,
      "learning_rate": 5.857652879493081e-08,
      "loss": 0.0541,
      "step": 18430
    },
    {
      "epoch": 0.9517686547895688,
      "grad_norm": 16.754636764526367,
      "learning_rate": 5.845154298106337e-08,
      "loss": 0.1495,
      "step": 18431
    },
    {
      "epoch": 0.9518202943454687,
      "grad_norm": 12.933677673339844,
      "learning_rate": 5.832668986791357e-08,
      "loss": 0.1797,
      "step": 18432
    },
    {
      "epoch": 0.9518719339013685,
      "grad_norm": 4.6319193840026855,
      "learning_rate": 5.820196945883316e-08,
      "loss": 0.0753,
      "step": 18433
    },
    {
      "epoch": 0.9519235734572683,
      "grad_norm": 3.4668750762939453,
      "learning_rate": 5.8077381757171146e-08,
      "loss": 0.0401,
      "step": 18434
    },
    {
      "epoch": 0.9519752130131681,
      "grad_norm": 3.460035562515259,
      "learning_rate": 5.7952926766273735e-08,
      "loss": 0.0393,
      "step": 18435
    },
    {
      "epoch": 0.9520268525690679,
      "grad_norm": 7.85791015625,
      "learning_rate": 5.7828604489482686e-08,
      "loss": 0.0524,
      "step": 18436
    },
    {
      "epoch": 0.9520784921249678,
      "grad_norm": 7.0037126541137695,
      "learning_rate": 5.770441493013701e-08,
      "loss": 0.097,
      "step": 18437
    },
    {
      "epoch": 0.9521301316808676,
      "grad_norm": 11.158761978149414,
      "learning_rate": 5.7580358091571255e-08,
      "loss": 0.1134,
      "step": 18438
    },
    {
      "epoch": 0.9521817712367674,
      "grad_norm": 2.331317901611328,
      "learning_rate": 5.745643397711776e-08,
      "loss": 0.0218,
      "step": 18439
    },
    {
      "epoch": 0.9522334107926672,
      "grad_norm": 5.401488304138184,
      "learning_rate": 5.733264259010274e-08,
      "loss": 0.0295,
      "step": 18440
    },
    {
      "epoch": 0.952285050348567,
      "grad_norm": 2.6956024169921875,
      "learning_rate": 5.720898393385188e-08,
      "loss": 0.045,
      "step": 18441
    },
    {
      "epoch": 0.9523366899044668,
      "grad_norm": 3.1729931831359863,
      "learning_rate": 5.708545801168586e-08,
      "loss": 0.0402,
      "step": 18442
    },
    {
      "epoch": 0.9523883294603667,
      "grad_norm": 2.1957502365112305,
      "learning_rate": 5.69620648269209e-08,
      "loss": 0.0354,
      "step": 18443
    },
    {
      "epoch": 0.9524399690162665,
      "grad_norm": 3.357856273651123,
      "learning_rate": 5.683880438287215e-08,
      "loss": 0.0474,
      "step": 18444
    },
    {
      "epoch": 0.9524916085721663,
      "grad_norm": 2.83516001701355,
      "learning_rate": 5.67156766828475e-08,
      "loss": 0.0462,
      "step": 18445
    },
    {
      "epoch": 0.9525432481280661,
      "grad_norm": 2.538644313812256,
      "learning_rate": 5.659268173015542e-08,
      "loss": 0.0635,
      "step": 18446
    },
    {
      "epoch": 0.9525948876839659,
      "grad_norm": 7.257087230682373,
      "learning_rate": 5.646981952809771e-08,
      "loss": 0.0645,
      "step": 18447
    },
    {
      "epoch": 0.9526465272398658,
      "grad_norm": 3.1522693634033203,
      "learning_rate": 5.634709007997452e-08,
      "loss": 0.0294,
      "step": 18448
    },
    {
      "epoch": 0.9526981667957656,
      "grad_norm": 2.618065595626831,
      "learning_rate": 5.622449338908098e-08,
      "loss": 0.0238,
      "step": 18449
    },
    {
      "epoch": 0.9527498063516654,
      "grad_norm": 4.243284225463867,
      "learning_rate": 5.610202945870891e-08,
      "loss": 0.0469,
      "step": 18450
    },
    {
      "epoch": 0.9528014459075652,
      "grad_norm": 4.555278301239014,
      "learning_rate": 5.597969829214789e-08,
      "loss": 0.0414,
      "step": 18451
    },
    {
      "epoch": 0.952853085463465,
      "grad_norm": 2.234105110168457,
      "learning_rate": 5.585749989268308e-08,
      "loss": 0.0229,
      "step": 18452
    },
    {
      "epoch": 0.9529047250193649,
      "grad_norm": 4.554159641265869,
      "learning_rate": 5.573543426359518e-08,
      "loss": 0.0503,
      "step": 18453
    },
    {
      "epoch": 0.9529563645752647,
      "grad_norm": 2.386599540710449,
      "learning_rate": 5.5613501408162685e-08,
      "loss": 0.0461,
      "step": 18454
    },
    {
      "epoch": 0.9530080041311645,
      "grad_norm": 3.6513547897338867,
      "learning_rate": 5.54917013296602e-08,
      "loss": 0.06,
      "step": 18455
    },
    {
      "epoch": 0.9530596436870643,
      "grad_norm": 10.603874206542969,
      "learning_rate": 5.537003403135788e-08,
      "loss": 0.0939,
      "step": 18456
    },
    {
      "epoch": 0.9531112832429641,
      "grad_norm": 1.8882861137390137,
      "learning_rate": 5.5248499516523116e-08,
      "loss": 0.0286,
      "step": 18457
    },
    {
      "epoch": 0.9531629227988639,
      "grad_norm": 2.361921548843384,
      "learning_rate": 5.512709778842051e-08,
      "loss": 0.0329,
      "step": 18458
    },
    {
      "epoch": 0.9532145623547638,
      "grad_norm": 3.4058988094329834,
      "learning_rate": 5.500582885030914e-08,
      "loss": 0.0315,
      "step": 18459
    },
    {
      "epoch": 0.9532662019106636,
      "grad_norm": 7.523262023925781,
      "learning_rate": 5.488469270544583e-08,
      "loss": 0.0857,
      "step": 18460
    },
    {
      "epoch": 0.9533178414665634,
      "grad_norm": 2.713304281234741,
      "learning_rate": 5.476368935708354e-08,
      "loss": 0.0291,
      "step": 18461
    },
    {
      "epoch": 0.9533694810224632,
      "grad_norm": 4.786985874176025,
      "learning_rate": 5.4642818808472444e-08,
      "loss": 0.0308,
      "step": 18462
    },
    {
      "epoch": 0.953421120578363,
      "grad_norm": 2.871706247329712,
      "learning_rate": 5.4522081062857726e-08,
      "loss": 0.0372,
      "step": 18463
    },
    {
      "epoch": 0.9534727601342629,
      "grad_norm": 2.1901803016662598,
      "learning_rate": 5.4401476123481235e-08,
      "loss": 0.0196,
      "step": 18464
    },
    {
      "epoch": 0.9535243996901627,
      "grad_norm": 2.3038508892059326,
      "learning_rate": 5.42810039935826e-08,
      "loss": 0.0196,
      "step": 18465
    },
    {
      "epoch": 0.9535760392460625,
      "grad_norm": 2.44442081451416,
      "learning_rate": 5.4160664676397025e-08,
      "loss": 0.0321,
      "step": 18466
    },
    {
      "epoch": 0.9536276788019623,
      "grad_norm": 11.604424476623535,
      "learning_rate": 5.404045817515524e-08,
      "loss": 0.0483,
      "step": 18467
    },
    {
      "epoch": 0.9536793183578621,
      "grad_norm": 4.251537322998047,
      "learning_rate": 5.3920384493085783e-08,
      "loss": 0.021,
      "step": 18468
    },
    {
      "epoch": 0.953730957913762,
      "grad_norm": 5.848080158233643,
      "learning_rate": 5.3800443633412744e-08,
      "loss": 0.0475,
      "step": 18469
    },
    {
      "epoch": 0.9537825974696618,
      "grad_norm": 2.140801429748535,
      "learning_rate": 5.3680635599358544e-08,
      "loss": 0.0226,
      "step": 18470
    },
    {
      "epoch": 0.9538342370255616,
      "grad_norm": 1.2975044250488281,
      "learning_rate": 5.3560960394138386e-08,
      "loss": 0.0247,
      "step": 18471
    },
    {
      "epoch": 0.9538858765814614,
      "grad_norm": 3.33866286277771,
      "learning_rate": 5.344141802096747e-08,
      "loss": 0.0342,
      "step": 18472
    },
    {
      "epoch": 0.9539375161373612,
      "grad_norm": 4.470611095428467,
      "learning_rate": 5.332200848305547e-08,
      "loss": 0.0551,
      "step": 18473
    },
    {
      "epoch": 0.953989155693261,
      "grad_norm": 3.932288646697998,
      "learning_rate": 5.320273178360868e-08,
      "loss": 0.0311,
      "step": 18474
    },
    {
      "epoch": 0.9540407952491609,
      "grad_norm": 3.402381420135498,
      "learning_rate": 5.308358792583068e-08,
      "loss": 0.0565,
      "step": 18475
    },
    {
      "epoch": 0.9540924348050607,
      "grad_norm": 1.997925877571106,
      "learning_rate": 5.2964576912921114e-08,
      "loss": 0.0411,
      "step": 18476
    },
    {
      "epoch": 0.9541440743609605,
      "grad_norm": 6.760553359985352,
      "learning_rate": 5.284569874807577e-08,
      "loss": 0.1191,
      "step": 18477
    },
    {
      "epoch": 0.9541957139168603,
      "grad_norm": 2.9330809116363525,
      "learning_rate": 5.2726953434486524e-08,
      "loss": 0.0431,
      "step": 18478
    },
    {
      "epoch": 0.9542473534727601,
      "grad_norm": 26.01620864868164,
      "learning_rate": 5.2608340975343066e-08,
      "loss": 0.0612,
      "step": 18479
    },
    {
      "epoch": 0.95429899302866,
      "grad_norm": 1.8513530492782593,
      "learning_rate": 5.2489861373829495e-08,
      "loss": 0.015,
      "step": 18480
    },
    {
      "epoch": 0.9543506325845598,
      "grad_norm": 5.304808139801025,
      "learning_rate": 5.2371514633128836e-08,
      "loss": 0.0551,
      "step": 18481
    },
    {
      "epoch": 0.9544022721404596,
      "grad_norm": 0.9552797675132751,
      "learning_rate": 5.225330075641799e-08,
      "loss": 0.018,
      "step": 18482
    },
    {
      "epoch": 0.9544539116963594,
      "grad_norm": 4.503178119659424,
      "learning_rate": 5.213521974687108e-08,
      "loss": 0.0237,
      "step": 18483
    },
    {
      "epoch": 0.9545055512522592,
      "grad_norm": 12.2725191116333,
      "learning_rate": 5.2017271607661124e-08,
      "loss": 0.0415,
      "step": 18484
    },
    {
      "epoch": 0.954557190808159,
      "grad_norm": 5.590943813323975,
      "learning_rate": 5.1899456341953924e-08,
      "loss": 0.051,
      "step": 18485
    },
    {
      "epoch": 0.9546088303640589,
      "grad_norm": 2.852356195449829,
      "learning_rate": 5.178177395291362e-08,
      "loss": 0.0498,
      "step": 18486
    },
    {
      "epoch": 0.9546604699199587,
      "grad_norm": 3.342984437942505,
      "learning_rate": 5.166422444370045e-08,
      "loss": 0.0661,
      "step": 18487
    },
    {
      "epoch": 0.9547121094758585,
      "grad_norm": 10.016255378723145,
      "learning_rate": 5.1546807817471345e-08,
      "loss": 0.0571,
      "step": 18488
    },
    {
      "epoch": 0.9547637490317583,
      "grad_norm": 2.427915573120117,
      "learning_rate": 5.142952407737878e-08,
      "loss": 0.0414,
      "step": 18489
    },
    {
      "epoch": 0.9548153885876581,
      "grad_norm": 2.7263474464416504,
      "learning_rate": 5.131237322657301e-08,
      "loss": 0.0512,
      "step": 18490
    },
    {
      "epoch": 0.954867028143558,
      "grad_norm": 1.2603896856307983,
      "learning_rate": 5.1195355268199854e-08,
      "loss": 0.0117,
      "step": 18491
    },
    {
      "epoch": 0.9549186676994578,
      "grad_norm": 2.533355951309204,
      "learning_rate": 5.10784702054018e-08,
      "loss": 0.0162,
      "step": 18492
    },
    {
      "epoch": 0.9549703072553576,
      "grad_norm": 1.6599522829055786,
      "learning_rate": 5.096171804131689e-08,
      "loss": 0.0235,
      "step": 18493
    },
    {
      "epoch": 0.9550219468112574,
      "grad_norm": 4.554549217224121,
      "learning_rate": 5.0845098779081504e-08,
      "loss": 0.0445,
      "step": 18494
    },
    {
      "epoch": 0.9550735863671572,
      "grad_norm": 17.059473037719727,
      "learning_rate": 5.0728612421827026e-08,
      "loss": 0.1067,
      "step": 18495
    },
    {
      "epoch": 0.955125225923057,
      "grad_norm": 3.2602596282958984,
      "learning_rate": 5.0612258972681517e-08,
      "loss": 0.0481,
      "step": 18496
    },
    {
      "epoch": 0.9551768654789569,
      "grad_norm": 2.731472969055176,
      "learning_rate": 5.049603843476858e-08,
      "loss": 0.0188,
      "step": 18497
    },
    {
      "epoch": 0.9552285050348567,
      "grad_norm": 2.4748449325561523,
      "learning_rate": 5.037995081121072e-08,
      "loss": 0.0269,
      "step": 18498
    },
    {
      "epoch": 0.9552801445907565,
      "grad_norm": 12.516131401062012,
      "learning_rate": 5.026399610512489e-08,
      "loss": 0.0628,
      "step": 18499
    },
    {
      "epoch": 0.9553317841466563,
      "grad_norm": 2.9705810546875,
      "learning_rate": 5.014817431962415e-08,
      "loss": 0.0383,
      "step": 18500
    },
    {
      "epoch": 0.9553834237025561,
      "grad_norm": 4.190913677215576,
      "learning_rate": 5.003248545781991e-08,
      "loss": 0.0303,
      "step": 18501
    },
    {
      "epoch": 0.955435063258456,
      "grad_norm": 3.6587047576904297,
      "learning_rate": 4.991692952281746e-08,
      "loss": 0.0564,
      "step": 18502
    },
    {
      "epoch": 0.9554867028143558,
      "grad_norm": 4.37802267074585,
      "learning_rate": 4.980150651772209e-08,
      "loss": 0.0519,
      "step": 18503
    },
    {
      "epoch": 0.9555383423702556,
      "grad_norm": 8.98040771484375,
      "learning_rate": 4.968621644563188e-08,
      "loss": 0.097,
      "step": 18504
    },
    {
      "epoch": 0.9555899819261554,
      "grad_norm": 4.652971267700195,
      "learning_rate": 4.957105930964268e-08,
      "loss": 0.0449,
      "step": 18505
    },
    {
      "epoch": 0.9556416214820552,
      "grad_norm": 2.5004351139068604,
      "learning_rate": 4.9456035112847577e-08,
      "loss": 0.0425,
      "step": 18506
    },
    {
      "epoch": 0.955693261037955,
      "grad_norm": 11.640000343322754,
      "learning_rate": 4.934114385833522e-08,
      "loss": 0.0988,
      "step": 18507
    },
    {
      "epoch": 0.9557449005938549,
      "grad_norm": 2.26975154876709,
      "learning_rate": 4.9226385549190905e-08,
      "loss": 0.0373,
      "step": 18508
    },
    {
      "epoch": 0.9557965401497547,
      "grad_norm": 1.359783411026001,
      "learning_rate": 4.9111760188496614e-08,
      "loss": 0.0145,
      "step": 18509
    },
    {
      "epoch": 0.9558481797056545,
      "grad_norm": 3.7494351863861084,
      "learning_rate": 4.8997267779329896e-08,
      "loss": 0.0557,
      "step": 18510
    },
    {
      "epoch": 0.9558998192615543,
      "grad_norm": 3.8317513465881348,
      "learning_rate": 4.8882908324766056e-08,
      "loss": 0.0481,
      "step": 18511
    },
    {
      "epoch": 0.9559514588174541,
      "grad_norm": 12.355366706848145,
      "learning_rate": 4.876868182787542e-08,
      "loss": 0.0928,
      "step": 18512
    },
    {
      "epoch": 0.956003098373354,
      "grad_norm": 3.3335583209991455,
      "learning_rate": 4.8654588291725534e-08,
      "loss": 0.061,
      "step": 18513
    },
    {
      "epoch": 0.9560547379292538,
      "grad_norm": 5.153214931488037,
      "learning_rate": 4.854062771938173e-08,
      "loss": 0.0563,
      "step": 18514
    },
    {
      "epoch": 0.9561063774851536,
      "grad_norm": 31.726377487182617,
      "learning_rate": 4.8426800113902105e-08,
      "loss": 0.0474,
      "step": 18515
    },
    {
      "epoch": 0.9561580170410534,
      "grad_norm": 5.263953685760498,
      "learning_rate": 4.831310547834422e-08,
      "loss": 0.0405,
      "step": 18516
    },
    {
      "epoch": 0.9562096565969532,
      "grad_norm": 2.323157548904419,
      "learning_rate": 4.819954381576175e-08,
      "loss": 0.028,
      "step": 18517
    },
    {
      "epoch": 0.956261296152853,
      "grad_norm": 3.561818838119507,
      "learning_rate": 4.808611512920447e-08,
      "loss": 0.018,
      "step": 18518
    },
    {
      "epoch": 0.9563129357087529,
      "grad_norm": 10.993321418762207,
      "learning_rate": 4.7972819421717165e-08,
      "loss": 0.1502,
      "step": 18519
    },
    {
      "epoch": 0.9563645752646527,
      "grad_norm": 2.5590877532958984,
      "learning_rate": 4.785965669634351e-08,
      "loss": 0.0483,
      "step": 18520
    },
    {
      "epoch": 0.9564162148205525,
      "grad_norm": 8.360563278198242,
      "learning_rate": 4.774662695612164e-08,
      "loss": 0.0713,
      "step": 18521
    },
    {
      "epoch": 0.9564678543764523,
      "grad_norm": 3.405702829360962,
      "learning_rate": 4.7633730204086906e-08,
      "loss": 0.0391,
      "step": 18522
    },
    {
      "epoch": 0.9565194939323521,
      "grad_norm": 1.9874789714813232,
      "learning_rate": 4.752096644327131e-08,
      "loss": 0.0299,
      "step": 18523
    },
    {
      "epoch": 0.956571133488252,
      "grad_norm": 2.7388999462127686,
      "learning_rate": 4.7408335676703e-08,
      "loss": 0.0378,
      "step": 18524
    },
    {
      "epoch": 0.9566227730441518,
      "grad_norm": 2.0667715072631836,
      "learning_rate": 4.729583790740677e-08,
      "loss": 0.0363,
      "step": 18525
    },
    {
      "epoch": 0.9566744126000516,
      "grad_norm": 5.182273864746094,
      "learning_rate": 4.718347313840299e-08,
      "loss": 0.044,
      "step": 18526
    },
    {
      "epoch": 0.9567260521559514,
      "grad_norm": 3.690507173538208,
      "learning_rate": 4.7071241372709795e-08,
      "loss": 0.0258,
      "step": 18527
    },
    {
      "epoch": 0.9567776917118512,
      "grad_norm": 6.505270957946777,
      "learning_rate": 4.6959142613340894e-08,
      "loss": 0.094,
      "step": 18528
    },
    {
      "epoch": 0.956829331267751,
      "grad_norm": 3.1532394886016846,
      "learning_rate": 4.684717686330664e-08,
      "loss": 0.0568,
      "step": 18529
    },
    {
      "epoch": 0.9568809708236509,
      "grad_norm": 4.331991195678711,
      "learning_rate": 4.673534412561298e-08,
      "loss": 0.0905,
      "step": 18530
    },
    {
      "epoch": 0.9569326103795507,
      "grad_norm": 10.120368003845215,
      "learning_rate": 4.662364440326361e-08,
      "loss": 0.1594,
      "step": 18531
    },
    {
      "epoch": 0.9569842499354505,
      "grad_norm": 3.871986150741577,
      "learning_rate": 4.651207769925892e-08,
      "loss": 0.0685,
      "step": 18532
    },
    {
      "epoch": 0.9570358894913503,
      "grad_norm": 3.614384889602661,
      "learning_rate": 4.6400644016594274e-08,
      "loss": 0.0282,
      "step": 18533
    },
    {
      "epoch": 0.9570875290472501,
      "grad_norm": 3.6877779960632324,
      "learning_rate": 4.628934335826174e-08,
      "loss": 0.0448,
      "step": 18534
    },
    {
      "epoch": 0.95713916860315,
      "grad_norm": 2.361522674560547,
      "learning_rate": 4.617817572725003e-08,
      "loss": 0.0444,
      "step": 18535
    },
    {
      "epoch": 0.9571908081590498,
      "grad_norm": 4.55108118057251,
      "learning_rate": 4.60671411265462e-08,
      "loss": 0.0545,
      "step": 18536
    },
    {
      "epoch": 0.9572424477149497,
      "grad_norm": 13.136666297912598,
      "learning_rate": 4.5956239559130086e-08,
      "loss": 0.0756,
      "step": 18537
    },
    {
      "epoch": 0.9572940872708495,
      "grad_norm": 25.160497665405273,
      "learning_rate": 4.584547102798043e-08,
      "loss": 0.0713,
      "step": 18538
    },
    {
      "epoch": 0.9573457268267493,
      "grad_norm": 16.024503707885742,
      "learning_rate": 4.5734835536072054e-08,
      "loss": 0.0907,
      "step": 18539
    },
    {
      "epoch": 0.9573973663826492,
      "grad_norm": 3.784095287322998,
      "learning_rate": 4.562433308637537e-08,
      "loss": 0.0713,
      "step": 18540
    },
    {
      "epoch": 0.957449005938549,
      "grad_norm": 5.37794828414917,
      "learning_rate": 4.551396368185856e-08,
      "loss": 0.0677,
      "step": 18541
    },
    {
      "epoch": 0.9575006454944488,
      "grad_norm": 6.636687278747559,
      "learning_rate": 4.540372732548537e-08,
      "loss": 0.0454,
      "step": 18542
    },
    {
      "epoch": 0.9575522850503486,
      "grad_norm": 6.700018405914307,
      "learning_rate": 4.5293624020216196e-08,
      "loss": 0.0615,
      "step": 18543
    },
    {
      "epoch": 0.9576039246062484,
      "grad_norm": 2.4294943809509277,
      "learning_rate": 4.518365376900702e-08,
      "loss": 0.0222,
      "step": 18544
    },
    {
      "epoch": 0.9576555641621483,
      "grad_norm": 2.287576913833618,
      "learning_rate": 4.507381657481102e-08,
      "loss": 0.0286,
      "step": 18545
    },
    {
      "epoch": 0.9577072037180481,
      "grad_norm": 6.842397212982178,
      "learning_rate": 4.4964112440579186e-08,
      "loss": 0.046,
      "step": 18546
    },
    {
      "epoch": 0.9577588432739479,
      "grad_norm": 2.2834386825561523,
      "learning_rate": 4.485454136925638e-08,
      "loss": 0.031,
      "step": 18547
    },
    {
      "epoch": 0.9578104828298477,
      "grad_norm": 3.8916969299316406,
      "learning_rate": 4.474510336378468e-08,
      "loss": 0.0544,
      "step": 18548
    },
    {
      "epoch": 0.9578621223857475,
      "grad_norm": 3.928767442703247,
      "learning_rate": 4.463579842710342e-08,
      "loss": 0.0777,
      "step": 18549
    },
    {
      "epoch": 0.9579137619416473,
      "grad_norm": 9.106792449951172,
      "learning_rate": 4.452662656214801e-08,
      "loss": 0.0567,
      "step": 18550
    },
    {
      "epoch": 0.9579654014975472,
      "grad_norm": 9.228455543518066,
      "learning_rate": 4.4417587771850566e-08,
      "loss": 0.1239,
      "step": 18551
    },
    {
      "epoch": 0.958017041053447,
      "grad_norm": 4.00870943069458,
      "learning_rate": 4.430868205913819e-08,
      "loss": 0.0589,
      "step": 18552
    },
    {
      "epoch": 0.9580686806093468,
      "grad_norm": 5.557629108428955,
      "learning_rate": 4.41999094269363e-08,
      "loss": 0.0507,
      "step": 18553
    },
    {
      "epoch": 0.9581203201652466,
      "grad_norm": 7.351378917694092,
      "learning_rate": 4.409126987816536e-08,
      "loss": 0.0259,
      "step": 18554
    },
    {
      "epoch": 0.9581719597211464,
      "grad_norm": 3.7472403049468994,
      "learning_rate": 4.398276341574303e-08,
      "loss": 0.0354,
      "step": 18555
    },
    {
      "epoch": 0.9582235992770463,
      "grad_norm": 4.788790702819824,
      "learning_rate": 4.387439004258309e-08,
      "loss": 0.1119,
      "step": 18556
    },
    {
      "epoch": 0.9582752388329461,
      "grad_norm": 3.173598051071167,
      "learning_rate": 4.376614976159599e-08,
      "loss": 0.0421,
      "step": 18557
    },
    {
      "epoch": 0.9583268783888459,
      "grad_norm": 4.575285911560059,
      "learning_rate": 4.3658042575688285e-08,
      "loss": 0.0404,
      "step": 18558
    },
    {
      "epoch": 0.9583785179447457,
      "grad_norm": 4.405795097351074,
      "learning_rate": 4.3550068487762663e-08,
      "loss": 0.0268,
      "step": 18559
    },
    {
      "epoch": 0.9584301575006455,
      "grad_norm": 1.6834721565246582,
      "learning_rate": 4.344222750071958e-08,
      "loss": 0.0153,
      "step": 18560
    },
    {
      "epoch": 0.9584817970565453,
      "grad_norm": 6.712220191955566,
      "learning_rate": 4.333451961745394e-08,
      "loss": 0.0247,
      "step": 18561
    },
    {
      "epoch": 0.9585334366124452,
      "grad_norm": 4.110872745513916,
      "learning_rate": 4.322694484085954e-08,
      "loss": 0.0439,
      "step": 18562
    },
    {
      "epoch": 0.958585076168345,
      "grad_norm": 35.926944732666016,
      "learning_rate": 4.311950317382352e-08,
      "loss": 0.1529,
      "step": 18563
    },
    {
      "epoch": 0.9586367157242448,
      "grad_norm": 3.264641523361206,
      "learning_rate": 4.301219461923245e-08,
      "loss": 0.0364,
      "step": 18564
    },
    {
      "epoch": 0.9586883552801446,
      "grad_norm": 4.49968147277832,
      "learning_rate": 4.2905019179967365e-08,
      "loss": 0.0438,
      "step": 18565
    },
    {
      "epoch": 0.9587399948360444,
      "grad_norm": 1.995869755744934,
      "learning_rate": 4.279797685890652e-08,
      "loss": 0.0323,
      "step": 18566
    },
    {
      "epoch": 0.9587916343919443,
      "grad_norm": 2.898303508758545,
      "learning_rate": 4.269106765892428e-08,
      "loss": 0.036,
      "step": 18567
    },
    {
      "epoch": 0.9588432739478441,
      "grad_norm": 3.264511823654175,
      "learning_rate": 4.258429158289168e-08,
      "loss": 0.0695,
      "step": 18568
    },
    {
      "epoch": 0.9588949135037439,
      "grad_norm": 12.272640228271484,
      "learning_rate": 4.247764863367698e-08,
      "loss": 0.0736,
      "step": 18569
    },
    {
      "epoch": 0.9589465530596437,
      "grad_norm": 6.4856414794921875,
      "learning_rate": 4.237113881414234e-08,
      "loss": 0.0806,
      "step": 18570
    },
    {
      "epoch": 0.9589981926155435,
      "grad_norm": 6.412520885467529,
      "learning_rate": 4.226476212714881e-08,
      "loss": 0.0576,
      "step": 18571
    },
    {
      "epoch": 0.9590498321714434,
      "grad_norm": 9.550643920898438,
      "learning_rate": 4.215851857555353e-08,
      "loss": 0.1164,
      "step": 18572
    },
    {
      "epoch": 0.9591014717273432,
      "grad_norm": 1.4000989198684692,
      "learning_rate": 4.205240816220868e-08,
      "loss": 0.0179,
      "step": 18573
    },
    {
      "epoch": 0.959153111283243,
      "grad_norm": 2.435499668121338,
      "learning_rate": 4.19464308899642e-08,
      "loss": 0.0316,
      "step": 18574
    },
    {
      "epoch": 0.9592047508391428,
      "grad_norm": 2.8084921836853027,
      "learning_rate": 4.184058676166613e-08,
      "loss": 0.0414,
      "step": 18575
    },
    {
      "epoch": 0.9592563903950426,
      "grad_norm": 3.9180867671966553,
      "learning_rate": 4.173487578015667e-08,
      "loss": 0.0378,
      "step": 18576
    },
    {
      "epoch": 0.9593080299509424,
      "grad_norm": 2.3330395221710205,
      "learning_rate": 4.1629297948274084e-08,
      "loss": 0.036,
      "step": 18577
    },
    {
      "epoch": 0.9593596695068423,
      "grad_norm": 4.770862102508545,
      "learning_rate": 4.152385326885444e-08,
      "loss": 0.0714,
      "step": 18578
    },
    {
      "epoch": 0.9594113090627421,
      "grad_norm": 32.768699645996094,
      "learning_rate": 4.1418541744728815e-08,
      "loss": 0.0728,
      "step": 18579
    },
    {
      "epoch": 0.9594629486186419,
      "grad_norm": 3.730536460876465,
      "learning_rate": 4.131336337872549e-08,
      "loss": 0.044,
      "step": 18580
    },
    {
      "epoch": 0.9595145881745417,
      "grad_norm": 7.691086292266846,
      "learning_rate": 4.120831817366888e-08,
      "loss": 0.0436,
      "step": 18581
    },
    {
      "epoch": 0.9595662277304415,
      "grad_norm": 12.307195663452148,
      "learning_rate": 4.110340613237951e-08,
      "loss": 0.0656,
      "step": 18582
    },
    {
      "epoch": 0.9596178672863414,
      "grad_norm": 2.179394006729126,
      "learning_rate": 4.0998627257675673e-08,
      "loss": 0.0318,
      "step": 18583
    },
    {
      "epoch": 0.9596695068422412,
      "grad_norm": 13.647083282470703,
      "learning_rate": 4.089398155237012e-08,
      "loss": 0.3105,
      "step": 18584
    },
    {
      "epoch": 0.959721146398141,
      "grad_norm": 5.436222553253174,
      "learning_rate": 4.0789469019273384e-08,
      "loss": 0.0415,
      "step": 18585
    },
    {
      "epoch": 0.9597727859540408,
      "grad_norm": 2.4438469409942627,
      "learning_rate": 4.0685089661192114e-08,
      "loss": 0.0207,
      "step": 18586
    },
    {
      "epoch": 0.9598244255099406,
      "grad_norm": 1.4332605600357056,
      "learning_rate": 4.058084348092905e-08,
      "loss": 0.0112,
      "step": 18587
    },
    {
      "epoch": 0.9598760650658404,
      "grad_norm": 6.810298919677734,
      "learning_rate": 4.047673048128364e-08,
      "loss": 0.0476,
      "step": 18588
    },
    {
      "epoch": 0.9599277046217403,
      "grad_norm": 3.5081305503845215,
      "learning_rate": 4.037275066505253e-08,
      "loss": 0.0365,
      "step": 18589
    },
    {
      "epoch": 0.9599793441776401,
      "grad_norm": 4.5122389793396,
      "learning_rate": 4.0268904035026815e-08,
      "loss": 0.047,
      "step": 18590
    },
    {
      "epoch": 0.9600309837335399,
      "grad_norm": 2.5867502689361572,
      "learning_rate": 4.0165190593995927e-08,
      "loss": 0.0269,
      "step": 18591
    },
    {
      "epoch": 0.9600826232894397,
      "grad_norm": 12.454012870788574,
      "learning_rate": 4.006161034474487e-08,
      "loss": 0.0802,
      "step": 18592
    },
    {
      "epoch": 0.9601342628453395,
      "grad_norm": 6.228479862213135,
      "learning_rate": 3.9958163290055304e-08,
      "loss": 0.0485,
      "step": 18593
    },
    {
      "epoch": 0.9601859024012394,
      "grad_norm": 1.7419971227645874,
      "learning_rate": 3.985484943270501e-08,
      "loss": 0.0314,
      "step": 18594
    },
    {
      "epoch": 0.9602375419571392,
      "grad_norm": 5.055090427398682,
      "learning_rate": 3.9751668775468435e-08,
      "loss": 0.0605,
      "step": 18595
    },
    {
      "epoch": 0.960289181513039,
      "grad_norm": 1.8104645013809204,
      "learning_rate": 3.9648621321116134e-08,
      "loss": 0.0341,
      "step": 18596
    },
    {
      "epoch": 0.9603408210689388,
      "grad_norm": 2.69077730178833,
      "learning_rate": 3.9545707072415895e-08,
      "loss": 0.0477,
      "step": 18597
    },
    {
      "epoch": 0.9603924606248386,
      "grad_norm": 1.1431065797805786,
      "learning_rate": 3.944292603213107e-08,
      "loss": 0.0174,
      "step": 18598
    },
    {
      "epoch": 0.9604441001807384,
      "grad_norm": 5.060548782348633,
      "learning_rate": 3.9340278203021664e-08,
      "loss": 0.0616,
      "step": 18599
    },
    {
      "epoch": 0.9604957397366383,
      "grad_norm": 4.495251178741455,
      "learning_rate": 3.9237763587844926e-08,
      "loss": 0.041,
      "step": 18600
    },
    {
      "epoch": 0.9605473792925381,
      "grad_norm": 1.8953912258148193,
      "learning_rate": 3.913538218935198e-08,
      "loss": 0.0248,
      "step": 18601
    },
    {
      "epoch": 0.9605990188484379,
      "grad_norm": 4.628069877624512,
      "learning_rate": 3.9033134010294513e-08,
      "loss": 0.0429,
      "step": 18602
    },
    {
      "epoch": 0.9606506584043377,
      "grad_norm": 8.665072441101074,
      "learning_rate": 3.893101905341701e-08,
      "loss": 0.0849,
      "step": 18603
    },
    {
      "epoch": 0.9607022979602375,
      "grad_norm": 3.5284626483917236,
      "learning_rate": 3.882903732146115e-08,
      "loss": 0.0277,
      "step": 18604
    },
    {
      "epoch": 0.9607539375161374,
      "grad_norm": 6.439159870147705,
      "learning_rate": 3.872718881716697e-08,
      "loss": 0.0569,
      "step": 18605
    },
    {
      "epoch": 0.9608055770720372,
      "grad_norm": 4.2030792236328125,
      "learning_rate": 3.86254735432684e-08,
      "loss": 0.0305,
      "step": 18606
    },
    {
      "epoch": 0.960857216627937,
      "grad_norm": 2.7023227214813232,
      "learning_rate": 3.852389150249769e-08,
      "loss": 0.0452,
      "step": 18607
    },
    {
      "epoch": 0.9609088561838368,
      "grad_norm": 19.313480377197266,
      "learning_rate": 3.842244269758211e-08,
      "loss": 0.0453,
      "step": 18608
    },
    {
      "epoch": 0.9609604957397366,
      "grad_norm": 0.6996645927429199,
      "learning_rate": 3.832112713124614e-08,
      "loss": 0.0039,
      "step": 18609
    },
    {
      "epoch": 0.9610121352956364,
      "grad_norm": 4.611501693725586,
      "learning_rate": 3.821994480621038e-08,
      "loss": 0.036,
      "step": 18610
    },
    {
      "epoch": 0.9610637748515363,
      "grad_norm": 11.025161743164062,
      "learning_rate": 3.811889572519267e-08,
      "loss": 0.0587,
      "step": 18611
    },
    {
      "epoch": 0.9611154144074361,
      "grad_norm": 2.7953004837036133,
      "learning_rate": 3.801797989090583e-08,
      "loss": 0.049,
      "step": 18612
    },
    {
      "epoch": 0.9611670539633359,
      "grad_norm": 2.178942918777466,
      "learning_rate": 3.7917197306060474e-08,
      "loss": 0.0352,
      "step": 18613
    },
    {
      "epoch": 0.9612186935192357,
      "grad_norm": 1.4022568464279175,
      "learning_rate": 3.7816547973362205e-08,
      "loss": 0.0129,
      "step": 18614
    },
    {
      "epoch": 0.9612703330751355,
      "grad_norm": 7.699594497680664,
      "learning_rate": 3.7716031895514425e-08,
      "loss": 0.1002,
      "step": 18615
    },
    {
      "epoch": 0.9613219726310354,
      "grad_norm": 1.0346142053604126,
      "learning_rate": 3.7615649075217195e-08,
      "loss": 0.0125,
      "step": 18616
    },
    {
      "epoch": 0.9613736121869352,
      "grad_norm": 4.936123847961426,
      "learning_rate": 3.751539951516447e-08,
      "loss": 0.0733,
      "step": 18617
    },
    {
      "epoch": 0.961425251742835,
      "grad_norm": 3.7655270099639893,
      "learning_rate": 3.7415283218049655e-08,
      "loss": 0.0586,
      "step": 18618
    },
    {
      "epoch": 0.9614768912987348,
      "grad_norm": 4.595409870147705,
      "learning_rate": 3.73153001865606e-08,
      "loss": 0.0705,
      "step": 18619
    },
    {
      "epoch": 0.9615285308546346,
      "grad_norm": 3.897453784942627,
      "learning_rate": 3.721545042338237e-08,
      "loss": 0.0567,
      "step": 18620
    },
    {
      "epoch": 0.9615801704105345,
      "grad_norm": 6.25705099105835,
      "learning_rate": 3.711573393119672e-08,
      "loss": 0.1139,
      "step": 18621
    },
    {
      "epoch": 0.9616318099664343,
      "grad_norm": 2.5319952964782715,
      "learning_rate": 3.70161507126815e-08,
      "loss": 0.0354,
      "step": 18622
    },
    {
      "epoch": 0.9616834495223341,
      "grad_norm": 22.500736236572266,
      "learning_rate": 3.691670077051013e-08,
      "loss": 0.0861,
      "step": 18623
    },
    {
      "epoch": 0.9617350890782339,
      "grad_norm": 1.5334975719451904,
      "learning_rate": 3.6817384107353804e-08,
      "loss": 0.0153,
      "step": 18624
    },
    {
      "epoch": 0.9617867286341337,
      "grad_norm": 7.183643341064453,
      "learning_rate": 3.6718200725879835e-08,
      "loss": 0.0267,
      "step": 18625
    },
    {
      "epoch": 0.9618383681900335,
      "grad_norm": 5.514160633087158,
      "learning_rate": 3.66191506287511e-08,
      "loss": 0.0328,
      "step": 18626
    },
    {
      "epoch": 0.9618900077459334,
      "grad_norm": 3.568145990371704,
      "learning_rate": 3.652023381862824e-08,
      "loss": 0.073,
      "step": 18627
    },
    {
      "epoch": 0.9619416473018332,
      "grad_norm": 2.530886650085449,
      "learning_rate": 3.6421450298166904e-08,
      "loss": 0.0147,
      "step": 18628
    },
    {
      "epoch": 0.961993286857733,
      "grad_norm": 3.261043071746826,
      "learning_rate": 3.6322800070019426e-08,
      "loss": 0.0343,
      "step": 18629
    },
    {
      "epoch": 0.9620449264136328,
      "grad_norm": 3.6960368156433105,
      "learning_rate": 3.622428313683646e-08,
      "loss": 0.0567,
      "step": 18630
    },
    {
      "epoch": 0.9620965659695326,
      "grad_norm": 13.441186904907227,
      "learning_rate": 3.612589950126255e-08,
      "loss": 0.1058,
      "step": 18631
    },
    {
      "epoch": 0.9621482055254325,
      "grad_norm": 6.961384296417236,
      "learning_rate": 3.602764916593948e-08,
      "loss": 0.1544,
      "step": 18632
    },
    {
      "epoch": 0.9621998450813323,
      "grad_norm": 10.764646530151367,
      "learning_rate": 3.592953213350625e-08,
      "loss": 0.0794,
      "step": 18633
    },
    {
      "epoch": 0.9622514846372321,
      "grad_norm": 7.981380939483643,
      "learning_rate": 3.5831548406597415e-08,
      "loss": 0.0596,
      "step": 18634
    },
    {
      "epoch": 0.9623031241931319,
      "grad_norm": 4.217655181884766,
      "learning_rate": 3.5733697987844763e-08,
      "loss": 0.044,
      "step": 18635
    },
    {
      "epoch": 0.9623547637490317,
      "grad_norm": 2.638108968734741,
      "learning_rate": 3.563598087987508e-08,
      "loss": 0.0522,
      "step": 18636
    },
    {
      "epoch": 0.9624064033049315,
      "grad_norm": 3.2872822284698486,
      "learning_rate": 3.5538397085313484e-08,
      "loss": 0.063,
      "step": 18637
    },
    {
      "epoch": 0.9624580428608314,
      "grad_norm": 4.564493656158447,
      "learning_rate": 3.5440946606779545e-08,
      "loss": 0.032,
      "step": 18638
    },
    {
      "epoch": 0.9625096824167312,
      "grad_norm": 11.77763557434082,
      "learning_rate": 3.5343629446890625e-08,
      "loss": 0.1032,
      "step": 18639
    },
    {
      "epoch": 0.962561321972631,
      "grad_norm": 3.483938694000244,
      "learning_rate": 3.5246445608260186e-08,
      "loss": 0.0427,
      "step": 18640
    },
    {
      "epoch": 0.9626129615285308,
      "grad_norm": 6.723477840423584,
      "learning_rate": 3.5149395093497796e-08,
      "loss": 0.0538,
      "step": 18641
    },
    {
      "epoch": 0.9626646010844306,
      "grad_norm": 2.674821615219116,
      "learning_rate": 3.5052477905209716e-08,
      "loss": 0.0254,
      "step": 18642
    },
    {
      "epoch": 0.9627162406403305,
      "grad_norm": 2.243553876876831,
      "learning_rate": 3.495569404599885e-08,
      "loss": 0.0313,
      "step": 18643
    },
    {
      "epoch": 0.9627678801962303,
      "grad_norm": 11.057101249694824,
      "learning_rate": 3.48590435184637e-08,
      "loss": 0.061,
      "step": 18644
    },
    {
      "epoch": 0.9628195197521301,
      "grad_norm": 2.7247626781463623,
      "learning_rate": 3.476252632519994e-08,
      "loss": 0.0518,
      "step": 18645
    },
    {
      "epoch": 0.9628711593080299,
      "grad_norm": 2.155266523361206,
      "learning_rate": 3.466614246879996e-08,
      "loss": 0.0248,
      "step": 18646
    },
    {
      "epoch": 0.9629227988639297,
      "grad_norm": 3.961005449295044,
      "learning_rate": 3.456989195185167e-08,
      "loss": 0.0399,
      "step": 18647
    },
    {
      "epoch": 0.9629744384198295,
      "grad_norm": 2.9578423500061035,
      "learning_rate": 3.447377477693914e-08,
      "loss": 0.0537,
      "step": 18648
    },
    {
      "epoch": 0.9630260779757294,
      "grad_norm": 11.56752872467041,
      "learning_rate": 3.437779094664473e-08,
      "loss": 0.0686,
      "step": 18649
    },
    {
      "epoch": 0.9630777175316292,
      "grad_norm": 5.333116054534912,
      "learning_rate": 3.428194046354527e-08,
      "loss": 0.0464,
      "step": 18650
    },
    {
      "epoch": 0.963129357087529,
      "grad_norm": 2.8839638233184814,
      "learning_rate": 3.4186223330214816e-08,
      "loss": 0.0459,
      "step": 18651
    },
    {
      "epoch": 0.9631809966434288,
      "grad_norm": 36.54017639160156,
      "learning_rate": 3.409063954922354e-08,
      "loss": 0.0641,
      "step": 18652
    },
    {
      "epoch": 0.9632326361993286,
      "grad_norm": 5.5677056312561035,
      "learning_rate": 3.3995189123138816e-08,
      "loss": 0.0778,
      "step": 18653
    },
    {
      "epoch": 0.9632842757552285,
      "grad_norm": 8.293766021728516,
      "learning_rate": 3.389987205452305e-08,
      "loss": 0.1247,
      "step": 18654
    },
    {
      "epoch": 0.9633359153111283,
      "grad_norm": 12.527503967285156,
      "learning_rate": 3.3804688345936975e-08,
      "loss": 0.0756,
      "step": 18655
    },
    {
      "epoch": 0.9633875548670281,
      "grad_norm": 1.0211966037750244,
      "learning_rate": 3.370963799993632e-08,
      "loss": 0.0066,
      "step": 18656
    },
    {
      "epoch": 0.9634391944229279,
      "grad_norm": 2.212805986404419,
      "learning_rate": 3.3614721019072926e-08,
      "loss": 0.0458,
      "step": 18657
    },
    {
      "epoch": 0.9634908339788277,
      "grad_norm": 5.951212406158447,
      "learning_rate": 3.3519937405895874e-08,
      "loss": 0.0453,
      "step": 18658
    },
    {
      "epoch": 0.9635424735347276,
      "grad_norm": 4.428443908691406,
      "learning_rate": 3.3425287162951456e-08,
      "loss": 0.0339,
      "step": 18659
    },
    {
      "epoch": 0.9635941130906274,
      "grad_norm": 1.7684190273284912,
      "learning_rate": 3.333077029278042e-08,
      "loss": 0.0212,
      "step": 18660
    },
    {
      "epoch": 0.9636457526465272,
      "grad_norm": 3.2634010314941406,
      "learning_rate": 3.3236386797921293e-08,
      "loss": 0.0372,
      "step": 18661
    },
    {
      "epoch": 0.963697392202427,
      "grad_norm": 4.132810592651367,
      "learning_rate": 3.314213668090816e-08,
      "loss": 0.065,
      "step": 18662
    },
    {
      "epoch": 0.9637490317583268,
      "grad_norm": 2.011345148086548,
      "learning_rate": 3.304801994427287e-08,
      "loss": 0.0443,
      "step": 18663
    },
    {
      "epoch": 0.9638006713142266,
      "grad_norm": 4.801492214202881,
      "learning_rate": 3.29540365905423e-08,
      "loss": 0.0451,
      "step": 18664
    },
    {
      "epoch": 0.9638523108701266,
      "grad_norm": 2.7469801902770996,
      "learning_rate": 3.286018662224055e-08,
      "loss": 0.0538,
      "step": 18665
    },
    {
      "epoch": 0.9639039504260264,
      "grad_norm": 4.87105655670166,
      "learning_rate": 3.276647004188782e-08,
      "loss": 0.0525,
      "step": 18666
    },
    {
      "epoch": 0.9639555899819262,
      "grad_norm": 13.822921752929688,
      "learning_rate": 3.2672886851999877e-08,
      "loss": 0.04,
      "step": 18667
    },
    {
      "epoch": 0.964007229537826,
      "grad_norm": 8.892708778381348,
      "learning_rate": 3.257943705509192e-08,
      "loss": 0.058,
      "step": 18668
    },
    {
      "epoch": 0.9640588690937258,
      "grad_norm": 6.530367851257324,
      "learning_rate": 3.24861206536714e-08,
      "loss": 0.0616,
      "step": 18669
    },
    {
      "epoch": 0.9641105086496257,
      "grad_norm": 2.8720743656158447,
      "learning_rate": 3.239293765024465e-08,
      "loss": 0.0256,
      "step": 18670
    },
    {
      "epoch": 0.9641621482055255,
      "grad_norm": 2.5778439044952393,
      "learning_rate": 3.229988804731521e-08,
      "loss": 0.0336,
      "step": 18671
    },
    {
      "epoch": 0.9642137877614253,
      "grad_norm": 8.060490608215332,
      "learning_rate": 3.220697184738053e-08,
      "loss": 0.0628,
      "step": 18672
    },
    {
      "epoch": 0.9642654273173251,
      "grad_norm": 5.0750532150268555,
      "learning_rate": 3.211418905293639e-08,
      "loss": 0.067,
      "step": 18673
    },
    {
      "epoch": 0.9643170668732249,
      "grad_norm": 4.800419807434082,
      "learning_rate": 3.2021539666474146e-08,
      "loss": 0.0319,
      "step": 18674
    },
    {
      "epoch": 0.9643687064291248,
      "grad_norm": 11.463306427001953,
      "learning_rate": 3.1929023690482344e-08,
      "loss": 0.1873,
      "step": 18675
    },
    {
      "epoch": 0.9644203459850246,
      "grad_norm": 7.043959617614746,
      "learning_rate": 3.1836641127444e-08,
      "loss": 0.066,
      "step": 18676
    },
    {
      "epoch": 0.9644719855409244,
      "grad_norm": 0.6467103362083435,
      "learning_rate": 3.1744391979841584e-08,
      "loss": 0.0042,
      "step": 18677
    },
    {
      "epoch": 0.9645236250968242,
      "grad_norm": 2.310002565383911,
      "learning_rate": 3.165227625015199e-08,
      "loss": 0.0262,
      "step": 18678
    },
    {
      "epoch": 0.964575264652724,
      "grad_norm": 9.974333763122559,
      "learning_rate": 3.1560293940848255e-08,
      "loss": 0.0767,
      "step": 18679
    },
    {
      "epoch": 0.9646269042086238,
      "grad_norm": 3.2681331634521484,
      "learning_rate": 3.146844505440116e-08,
      "loss": 0.0277,
      "step": 18680
    },
    {
      "epoch": 0.9646785437645237,
      "grad_norm": 4.5597662925720215,
      "learning_rate": 3.1376729593276534e-08,
      "loss": 0.0441,
      "step": 18681
    },
    {
      "epoch": 0.9647301833204235,
      "grad_norm": 3.4990994930267334,
      "learning_rate": 3.128514755993795e-08,
      "loss": 0.0323,
      "step": 18682
    },
    {
      "epoch": 0.9647818228763233,
      "grad_norm": 5.840734481811523,
      "learning_rate": 3.119369895684454e-08,
      "loss": 0.0217,
      "step": 18683
    },
    {
      "epoch": 0.9648334624322231,
      "grad_norm": 5.720447540283203,
      "learning_rate": 3.110238378645214e-08,
      "loss": 0.0546,
      "step": 18684
    },
    {
      "epoch": 0.9648851019881229,
      "grad_norm": 3.700517177581787,
      "learning_rate": 3.101120205121266e-08,
      "loss": 0.0896,
      "step": 18685
    },
    {
      "epoch": 0.9649367415440228,
      "grad_norm": 4.622447490692139,
      "learning_rate": 3.0920153753574714e-08,
      "loss": 0.0495,
      "step": 18686
    },
    {
      "epoch": 0.9649883810999226,
      "grad_norm": 2.1876564025878906,
      "learning_rate": 3.082923889598411e-08,
      "loss": 0.0293,
      "step": 18687
    },
    {
      "epoch": 0.9650400206558224,
      "grad_norm": 1.8225435018539429,
      "learning_rate": 3.073845748088111e-08,
      "loss": 0.0367,
      "step": 18688
    },
    {
      "epoch": 0.9650916602117222,
      "grad_norm": 3.8455615043640137,
      "learning_rate": 3.064780951070434e-08,
      "loss": 0.1057,
      "step": 18689
    },
    {
      "epoch": 0.965143299767622,
      "grad_norm": 3.159125566482544,
      "learning_rate": 3.05572949878874e-08,
      "loss": 0.0487,
      "step": 18690
    },
    {
      "epoch": 0.9651949393235218,
      "grad_norm": 10.518220901489258,
      "learning_rate": 3.046691391486167e-08,
      "loss": 0.0804,
      "step": 18691
    },
    {
      "epoch": 0.9652465788794217,
      "grad_norm": 16.523338317871094,
      "learning_rate": 3.037666629405411e-08,
      "loss": 0.0573,
      "step": 18692
    },
    {
      "epoch": 0.9652982184353215,
      "grad_norm": 3.608440637588501,
      "learning_rate": 3.028655212788834e-08,
      "loss": 0.0537,
      "step": 18693
    },
    {
      "epoch": 0.9653498579912213,
      "grad_norm": 2.171748638153076,
      "learning_rate": 3.019657141878407e-08,
      "loss": 0.027,
      "step": 18694
    },
    {
      "epoch": 0.9654014975471211,
      "grad_norm": 5.057189464569092,
      "learning_rate": 3.010672416915716e-08,
      "loss": 0.0707,
      "step": 18695
    },
    {
      "epoch": 0.9654531371030209,
      "grad_norm": 3.269365072250366,
      "learning_rate": 3.001701038142124e-08,
      "loss": 0.037,
      "step": 18696
    },
    {
      "epoch": 0.9655047766589208,
      "grad_norm": 2.486631155014038,
      "learning_rate": 2.9927430057985485e-08,
      "loss": 0.0178,
      "step": 18697
    },
    {
      "epoch": 0.9655564162148206,
      "grad_norm": 18.36591339111328,
      "learning_rate": 2.983798320125464e-08,
      "loss": 0.1123,
      "step": 18698
    },
    {
      "epoch": 0.9656080557707204,
      "grad_norm": 17.424528121948242,
      "learning_rate": 2.9748669813631226e-08,
      "loss": 0.0853,
      "step": 18699
    },
    {
      "epoch": 0.9656596953266202,
      "grad_norm": 2.943662166595459,
      "learning_rate": 2.9659489897513326e-08,
      "loss": 0.0336,
      "step": 18700
    },
    {
      "epoch": 0.96571133488252,
      "grad_norm": 3.7838423252105713,
      "learning_rate": 2.9570443455296804e-08,
      "loss": 0.0405,
      "step": 18701
    },
    {
      "epoch": 0.9657629744384199,
      "grad_norm": 3.6475841999053955,
      "learning_rate": 2.9481530489371414e-08,
      "loss": 0.0544,
      "step": 18702
    },
    {
      "epoch": 0.9658146139943197,
      "grad_norm": 4.662166595458984,
      "learning_rate": 2.939275100212635e-08,
      "loss": 0.0282,
      "step": 18703
    },
    {
      "epoch": 0.9658662535502195,
      "grad_norm": 12.640030860900879,
      "learning_rate": 2.9304104995944717e-08,
      "loss": 0.1583,
      "step": 18704
    },
    {
      "epoch": 0.9659178931061193,
      "grad_norm": 1.9677075147628784,
      "learning_rate": 2.9215592473207378e-08,
      "loss": 0.0178,
      "step": 18705
    },
    {
      "epoch": 0.9659695326620191,
      "grad_norm": 3.3945584297180176,
      "learning_rate": 2.9127213436290768e-08,
      "loss": 0.0666,
      "step": 18706
    },
    {
      "epoch": 0.9660211722179189,
      "grad_norm": 4.714353084564209,
      "learning_rate": 2.9038967887569104e-08,
      "loss": 0.0317,
      "step": 18707
    },
    {
      "epoch": 0.9660728117738188,
      "grad_norm": 9.861553192138672,
      "learning_rate": 2.895085582941215e-08,
      "loss": 0.0762,
      "step": 18708
    },
    {
      "epoch": 0.9661244513297186,
      "grad_norm": 10.669057846069336,
      "learning_rate": 2.8862877264184685e-08,
      "loss": 0.0838,
      "step": 18709
    },
    {
      "epoch": 0.9661760908856184,
      "grad_norm": 16.889564514160156,
      "learning_rate": 2.8775032194250375e-08,
      "loss": 0.0958,
      "step": 18710
    },
    {
      "epoch": 0.9662277304415182,
      "grad_norm": 4.8049726486206055,
      "learning_rate": 2.8687320621967886e-08,
      "loss": 0.051,
      "step": 18711
    },
    {
      "epoch": 0.966279369997418,
      "grad_norm": 7.5887908935546875,
      "learning_rate": 2.8599742549693666e-08,
      "loss": 0.0888,
      "step": 18712
    },
    {
      "epoch": 0.9663310095533179,
      "grad_norm": 1.8592818975448608,
      "learning_rate": 2.8512297979778058e-08,
      "loss": 0.0292,
      "step": 18713
    },
    {
      "epoch": 0.9663826491092177,
      "grad_norm": 8.311213493347168,
      "learning_rate": 2.842498691456974e-08,
      "loss": 0.0612,
      "step": 18714
    },
    {
      "epoch": 0.9664342886651175,
      "grad_norm": 7.626840591430664,
      "learning_rate": 2.8337809356414058e-08,
      "loss": 0.0457,
      "step": 18715
    },
    {
      "epoch": 0.9664859282210173,
      "grad_norm": 24.949726104736328,
      "learning_rate": 2.8250765307650808e-08,
      "loss": 0.2283,
      "step": 18716
    },
    {
      "epoch": 0.9665375677769171,
      "grad_norm": 4.797102451324463,
      "learning_rate": 2.816385477061867e-08,
      "loss": 0.0367,
      "step": 18717
    },
    {
      "epoch": 0.966589207332817,
      "grad_norm": 1.5824379920959473,
      "learning_rate": 2.8077077747651338e-08,
      "loss": 0.0237,
      "step": 18718
    },
    {
      "epoch": 0.9666408468887168,
      "grad_norm": 12.585012435913086,
      "learning_rate": 2.7990434241078614e-08,
      "loss": 0.0767,
      "step": 18719
    },
    {
      "epoch": 0.9666924864446166,
      "grad_norm": 1.38398277759552,
      "learning_rate": 2.7903924253227523e-08,
      "loss": 0.0141,
      "step": 18720
    },
    {
      "epoch": 0.9667441260005164,
      "grad_norm": 6.356092929840088,
      "learning_rate": 2.7817547786421208e-08,
      "loss": 0.0417,
      "step": 18721
    },
    {
      "epoch": 0.9667957655564162,
      "grad_norm": 1.4322822093963623,
      "learning_rate": 2.7731304842979477e-08,
      "loss": 0.0063,
      "step": 18722
    },
    {
      "epoch": 0.966847405112316,
      "grad_norm": 3.1345863342285156,
      "learning_rate": 2.764519542521771e-08,
      "loss": 0.042,
      "step": 18723
    },
    {
      "epoch": 0.9668990446682159,
      "grad_norm": 1.8405604362487793,
      "learning_rate": 2.7559219535449044e-08,
      "loss": 0.0309,
      "step": 18724
    },
    {
      "epoch": 0.9669506842241157,
      "grad_norm": 9.453624725341797,
      "learning_rate": 2.747337717598164e-08,
      "loss": 0.1136,
      "step": 18725
    },
    {
      "epoch": 0.9670023237800155,
      "grad_norm": 2.4098215103149414,
      "learning_rate": 2.738766834912143e-08,
      "loss": 0.0481,
      "step": 18726
    },
    {
      "epoch": 0.9670539633359153,
      "grad_norm": 12.973008155822754,
      "learning_rate": 2.73020930571688e-08,
      "loss": 0.0818,
      "step": 18727
    },
    {
      "epoch": 0.9671056028918151,
      "grad_norm": 3.3653109073638916,
      "learning_rate": 2.721665130242246e-08,
      "loss": 0.0374,
      "step": 18728
    },
    {
      "epoch": 0.967157242447715,
      "grad_norm": 8.703644752502441,
      "learning_rate": 2.7131343087177796e-08,
      "loss": 0.0418,
      "step": 18729
    },
    {
      "epoch": 0.9672088820036148,
      "grad_norm": 3.054290533065796,
      "learning_rate": 2.704616841372465e-08,
      "loss": 0.0703,
      "step": 18730
    },
    {
      "epoch": 0.9672605215595146,
      "grad_norm": 8.844278335571289,
      "learning_rate": 2.6961127284350075e-08,
      "loss": 0.0504,
      "step": 18731
    },
    {
      "epoch": 0.9673121611154144,
      "grad_norm": 7.282419681549072,
      "learning_rate": 2.6876219701338356e-08,
      "loss": 0.0502,
      "step": 18732
    },
    {
      "epoch": 0.9673638006713142,
      "grad_norm": 10.175894737243652,
      "learning_rate": 2.679144566696934e-08,
      "loss": 0.0406,
      "step": 18733
    },
    {
      "epoch": 0.967415440227214,
      "grad_norm": 3.700671911239624,
      "learning_rate": 2.670680518352009e-08,
      "loss": 0.0383,
      "step": 18734
    },
    {
      "epoch": 0.9674670797831139,
      "grad_norm": 4.862968444824219,
      "learning_rate": 2.6622298253263236e-08,
      "loss": 0.0453,
      "step": 18735
    },
    {
      "epoch": 0.9675187193390137,
      "grad_norm": 2.4207663536071777,
      "learning_rate": 2.6537924878467513e-08,
      "loss": 0.0539,
      "step": 18736
    },
    {
      "epoch": 0.9675703588949135,
      "grad_norm": 4.818243980407715,
      "learning_rate": 2.6453685061400004e-08,
      "loss": 0.0625,
      "step": 18737
    },
    {
      "epoch": 0.9676219984508133,
      "grad_norm": 16.060882568359375,
      "learning_rate": 2.6369578804321117e-08,
      "loss": 0.0937,
      "step": 18738
    },
    {
      "epoch": 0.9676736380067131,
      "grad_norm": 11.052643775939941,
      "learning_rate": 2.628560610949127e-08,
      "loss": 0.0668,
      "step": 18739
    },
    {
      "epoch": 0.967725277562613,
      "grad_norm": 13.124037742614746,
      "learning_rate": 2.620176697916421e-08,
      "loss": 0.0567,
      "step": 18740
    },
    {
      "epoch": 0.9677769171185128,
      "grad_norm": 2.793912410736084,
      "learning_rate": 2.6118061415592033e-08,
      "loss": 0.0358,
      "step": 18741
    },
    {
      "epoch": 0.9678285566744126,
      "grad_norm": 3.8800203800201416,
      "learning_rate": 2.6034489421021823e-08,
      "loss": 0.0749,
      "step": 18742
    },
    {
      "epoch": 0.9678801962303124,
      "grad_norm": 7.780601978302002,
      "learning_rate": 2.5951050997698458e-08,
      "loss": 0.1142,
      "step": 18743
    },
    {
      "epoch": 0.9679318357862122,
      "grad_norm": 3.9709465503692627,
      "learning_rate": 2.5867746147862914e-08,
      "loss": 0.0651,
      "step": 18744
    },
    {
      "epoch": 0.967983475342112,
      "grad_norm": 3.760685443878174,
      "learning_rate": 2.5784574873751743e-08,
      "loss": 0.0391,
      "step": 18745
    },
    {
      "epoch": 0.9680351148980119,
      "grad_norm": 2.9379189014434814,
      "learning_rate": 2.5701537177598158e-08,
      "loss": 0.0426,
      "step": 18746
    },
    {
      "epoch": 0.9680867544539117,
      "grad_norm": 1.5495408773422241,
      "learning_rate": 2.561863306163259e-08,
      "loss": 0.023,
      "step": 18747
    },
    {
      "epoch": 0.9681383940098115,
      "grad_norm": 2.703216791152954,
      "learning_rate": 2.55358625280816e-08,
      "loss": 0.0191,
      "step": 18748
    },
    {
      "epoch": 0.9681900335657113,
      "grad_norm": 10.432302474975586,
      "learning_rate": 2.545322557916674e-08,
      "loss": 0.0766,
      "step": 18749
    },
    {
      "epoch": 0.9682416731216111,
      "grad_norm": 1.862697958946228,
      "learning_rate": 2.5370722217108458e-08,
      "loss": 0.0211,
      "step": 18750
    },
    {
      "epoch": 0.968293312677511,
      "grad_norm": 3.4075562953948975,
      "learning_rate": 2.5288352444121643e-08,
      "loss": 0.0308,
      "step": 18751
    },
    {
      "epoch": 0.9683449522334108,
      "grad_norm": 2.750648260116577,
      "learning_rate": 2.5206116262418422e-08,
      "loss": 0.0433,
      "step": 18752
    },
    {
      "epoch": 0.9683965917893106,
      "grad_norm": 3.186016321182251,
      "learning_rate": 2.5124013674207026e-08,
      "loss": 0.0376,
      "step": 18753
    },
    {
      "epoch": 0.9684482313452104,
      "grad_norm": 16.670557022094727,
      "learning_rate": 2.5042044681691803e-08,
      "loss": 0.0799,
      "step": 18754
    },
    {
      "epoch": 0.9684998709011102,
      "grad_norm": 4.265800952911377,
      "learning_rate": 2.496020928707543e-08,
      "loss": 0.0429,
      "step": 18755
    },
    {
      "epoch": 0.96855151045701,
      "grad_norm": 4.102229118347168,
      "learning_rate": 2.487850749255394e-08,
      "loss": 0.0684,
      "step": 18756
    },
    {
      "epoch": 0.9686031500129099,
      "grad_norm": 3.201113700866699,
      "learning_rate": 2.4796939300322232e-08,
      "loss": 0.0568,
      "step": 18757
    },
    {
      "epoch": 0.9686547895688097,
      "grad_norm": 2.428126335144043,
      "learning_rate": 2.4715504712570227e-08,
      "loss": 0.0563,
      "step": 18758
    },
    {
      "epoch": 0.9687064291247095,
      "grad_norm": 2.710453748703003,
      "learning_rate": 2.463420373148562e-08,
      "loss": 0.0321,
      "step": 18759
    },
    {
      "epoch": 0.9687580686806093,
      "grad_norm": 3.056461811065674,
      "learning_rate": 2.455303635925055e-08,
      "loss": 0.0562,
      "step": 18760
    },
    {
      "epoch": 0.9688097082365091,
      "grad_norm": 6.885476112365723,
      "learning_rate": 2.4472002598045498e-08,
      "loss": 0.0373,
      "step": 18761
    },
    {
      "epoch": 0.968861347792409,
      "grad_norm": 3.946976900100708,
      "learning_rate": 2.43911024500465e-08,
      "loss": 0.0662,
      "step": 18762
    },
    {
      "epoch": 0.9689129873483088,
      "grad_norm": 2.2880725860595703,
      "learning_rate": 2.4310335917425708e-08,
      "loss": 0.0196,
      "step": 18763
    },
    {
      "epoch": 0.9689646269042086,
      "grad_norm": 2.5380630493164062,
      "learning_rate": 2.4229703002351946e-08,
      "loss": 0.0295,
      "step": 18764
    },
    {
      "epoch": 0.9690162664601084,
      "grad_norm": 3.92486572265625,
      "learning_rate": 2.4149203706991255e-08,
      "loss": 0.0761,
      "step": 18765
    },
    {
      "epoch": 0.9690679060160082,
      "grad_norm": 2.6534945964813232,
      "learning_rate": 2.4068838033504127e-08,
      "loss": 0.0577,
      "step": 18766
    },
    {
      "epoch": 0.969119545571908,
      "grad_norm": 2.6620686054229736,
      "learning_rate": 2.3988605984050506e-08,
      "loss": 0.0439,
      "step": 18767
    },
    {
      "epoch": 0.9691711851278079,
      "grad_norm": 2.9425108432769775,
      "learning_rate": 2.3908507560783113e-08,
      "loss": 0.054,
      "step": 18768
    },
    {
      "epoch": 0.9692228246837077,
      "grad_norm": 9.195430755615234,
      "learning_rate": 2.382854276585411e-08,
      "loss": 0.0489,
      "step": 18769
    },
    {
      "epoch": 0.9692744642396075,
      "grad_norm": 10.803889274597168,
      "learning_rate": 2.3748711601410125e-08,
      "loss": 0.189,
      "step": 18770
    },
    {
      "epoch": 0.9693261037955073,
      "grad_norm": 1.4087140560150146,
      "learning_rate": 2.3669014069595542e-08,
      "loss": 0.0204,
      "step": 18771
    },
    {
      "epoch": 0.9693777433514071,
      "grad_norm": 2.5337319374084473,
      "learning_rate": 2.358945017255032e-08,
      "loss": 0.0313,
      "step": 18772
    },
    {
      "epoch": 0.969429382907307,
      "grad_norm": 15.387697219848633,
      "learning_rate": 2.3510019912411086e-08,
      "loss": 0.2175,
      "step": 18773
    },
    {
      "epoch": 0.9694810224632068,
      "grad_norm": 3.148834228515625,
      "learning_rate": 2.3430723291311132e-08,
      "loss": 0.0517,
      "step": 18774
    },
    {
      "epoch": 0.9695326620191066,
      "grad_norm": 1.9523341655731201,
      "learning_rate": 2.3351560311379305e-08,
      "loss": 0.0212,
      "step": 18775
    },
    {
      "epoch": 0.9695843015750064,
      "grad_norm": 2.7656636238098145,
      "learning_rate": 2.327253097474169e-08,
      "loss": 0.0564,
      "step": 18776
    },
    {
      "epoch": 0.9696359411309062,
      "grad_norm": 2.01375412940979,
      "learning_rate": 2.3193635283520478e-08,
      "loss": 0.0245,
      "step": 18777
    },
    {
      "epoch": 0.969687580686806,
      "grad_norm": 2.687960147857666,
      "learning_rate": 2.3114873239835078e-08,
      "loss": 0.0478,
      "step": 18778
    },
    {
      "epoch": 0.9697392202427059,
      "grad_norm": 4.961716651916504,
      "learning_rate": 2.3036244845799916e-08,
      "loss": 0.0776,
      "step": 18779
    },
    {
      "epoch": 0.9697908597986057,
      "grad_norm": 6.680037021636963,
      "learning_rate": 2.2957750103526077e-08,
      "loss": 0.0703,
      "step": 18780
    },
    {
      "epoch": 0.9698424993545055,
      "grad_norm": 4.260068416595459,
      "learning_rate": 2.2879389015122434e-08,
      "loss": 0.0497,
      "step": 18781
    },
    {
      "epoch": 0.9698941389104053,
      "grad_norm": 2.341942071914673,
      "learning_rate": 2.2801161582692856e-08,
      "loss": 0.0102,
      "step": 18782
    },
    {
      "epoch": 0.9699457784663051,
      "grad_norm": 1.9730950593948364,
      "learning_rate": 2.2723067808337328e-08,
      "loss": 0.0296,
      "step": 18783
    },
    {
      "epoch": 0.969997418022205,
      "grad_norm": 2.7735846042633057,
      "learning_rate": 2.2645107694154177e-08,
      "loss": 0.0409,
      "step": 18784
    },
    {
      "epoch": 0.9700490575781048,
      "grad_norm": 11.414812088012695,
      "learning_rate": 2.256728124223673e-08,
      "loss": 0.1589,
      "step": 18785
    },
    {
      "epoch": 0.9701006971340046,
      "grad_norm": 0.48237088322639465,
      "learning_rate": 2.2489588454674414e-08,
      "loss": 0.0042,
      "step": 18786
    },
    {
      "epoch": 0.9701523366899044,
      "grad_norm": 10.552881240844727,
      "learning_rate": 2.241202933355391e-08,
      "loss": 0.1171,
      "step": 18787
    },
    {
      "epoch": 0.9702039762458042,
      "grad_norm": 2.4611072540283203,
      "learning_rate": 2.2334603880957428e-08,
      "loss": 0.0366,
      "step": 18788
    },
    {
      "epoch": 0.970255615801704,
      "grad_norm": 4.180549621582031,
      "learning_rate": 2.2257312098965532e-08,
      "loss": 0.0618,
      "step": 18789
    },
    {
      "epoch": 0.9703072553576039,
      "grad_norm": 2.067111015319824,
      "learning_rate": 2.218015398965212e-08,
      "loss": 0.0411,
      "step": 18790
    },
    {
      "epoch": 0.9703588949135037,
      "grad_norm": 1.9002559185028076,
      "learning_rate": 2.210312955509053e-08,
      "loss": 0.0277,
      "step": 18791
    },
    {
      "epoch": 0.9704105344694036,
      "grad_norm": 2.8045365810394287,
      "learning_rate": 2.202623879734911e-08,
      "loss": 0.0371,
      "step": 18792
    },
    {
      "epoch": 0.9704621740253034,
      "grad_norm": 1.8936340808868408,
      "learning_rate": 2.194948171849176e-08,
      "loss": 0.0263,
      "step": 18793
    },
    {
      "epoch": 0.9705138135812033,
      "grad_norm": 8.433895111083984,
      "learning_rate": 2.187285832058017e-08,
      "loss": 0.055,
      "step": 18794
    },
    {
      "epoch": 0.9705654531371031,
      "grad_norm": 2.1352524757385254,
      "learning_rate": 2.1796368605672137e-08,
      "loss": 0.0347,
      "step": 18795
    },
    {
      "epoch": 0.9706170926930029,
      "grad_norm": 2.719552993774414,
      "learning_rate": 2.172001257582157e-08,
      "loss": 0.0383,
      "step": 18796
    },
    {
      "epoch": 0.9706687322489027,
      "grad_norm": 7.623614311218262,
      "learning_rate": 2.1643790233079053e-08,
      "loss": 0.2015,
      "step": 18797
    },
    {
      "epoch": 0.9707203718048025,
      "grad_norm": 2.809490442276001,
      "learning_rate": 2.156770157949184e-08,
      "loss": 0.0398,
      "step": 18798
    },
    {
      "epoch": 0.9707720113607023,
      "grad_norm": 1.9733037948608398,
      "learning_rate": 2.1491746617102184e-08,
      "loss": 0.0188,
      "step": 18799
    },
    {
      "epoch": 0.9708236509166022,
      "grad_norm": 2.7836742401123047,
      "learning_rate": 2.1415925347950673e-08,
      "loss": 0.044,
      "step": 18800
    },
    {
      "epoch": 0.970875290472502,
      "grad_norm": 7.715173244476318,
      "learning_rate": 2.1340237774073457e-08,
      "loss": 0.0415,
      "step": 18801
    },
    {
      "epoch": 0.9709269300284018,
      "grad_norm": 5.19154691696167,
      "learning_rate": 2.1264683897502248e-08,
      "loss": 0.0837,
      "step": 18802
    },
    {
      "epoch": 0.9709785695843016,
      "grad_norm": 3.970710039138794,
      "learning_rate": 2.118926372026653e-08,
      "loss": 0.0405,
      "step": 18803
    },
    {
      "epoch": 0.9710302091402014,
      "grad_norm": 2.2608134746551514,
      "learning_rate": 2.1113977244391904e-08,
      "loss": 0.0214,
      "step": 18804
    },
    {
      "epoch": 0.9710818486961013,
      "grad_norm": 2.582414388656616,
      "learning_rate": 2.1038824471899532e-08,
      "loss": 0.0414,
      "step": 18805
    },
    {
      "epoch": 0.9711334882520011,
      "grad_norm": 5.825781345367432,
      "learning_rate": 2.09638054048078e-08,
      "loss": 0.0418,
      "step": 18806
    },
    {
      "epoch": 0.9711851278079009,
      "grad_norm": 2.3855276107788086,
      "learning_rate": 2.0888920045131765e-08,
      "loss": 0.037,
      "step": 18807
    },
    {
      "epoch": 0.9712367673638007,
      "grad_norm": 2.6454455852508545,
      "learning_rate": 2.0814168394881484e-08,
      "loss": 0.0365,
      "step": 18808
    },
    {
      "epoch": 0.9712884069197005,
      "grad_norm": 5.876735210418701,
      "learning_rate": 2.0739550456064795e-08,
      "loss": 0.0882,
      "step": 18809
    },
    {
      "epoch": 0.9713400464756003,
      "grad_norm": 2.2955336570739746,
      "learning_rate": 2.0665066230686205e-08,
      "loss": 0.0202,
      "step": 18810
    },
    {
      "epoch": 0.9713916860315002,
      "grad_norm": 4.474154949188232,
      "learning_rate": 2.059071572074467e-08,
      "loss": 0.063,
      "step": 18811
    },
    {
      "epoch": 0.9714433255874,
      "grad_norm": 4.817800045013428,
      "learning_rate": 2.0516498928237483e-08,
      "loss": 0.0784,
      "step": 18812
    },
    {
      "epoch": 0.9714949651432998,
      "grad_norm": 1.4672482013702393,
      "learning_rate": 2.0442415855157494e-08,
      "loss": 0.016,
      "step": 18813
    },
    {
      "epoch": 0.9715466046991996,
      "grad_norm": 3.0916800498962402,
      "learning_rate": 2.036846650349422e-08,
      "loss": 0.0255,
      "step": 18814
    },
    {
      "epoch": 0.9715982442550994,
      "grad_norm": 8.374160766601562,
      "learning_rate": 2.0294650875233303e-08,
      "loss": 0.079,
      "step": 18815
    },
    {
      "epoch": 0.9716498838109993,
      "grad_norm": 1.4499589204788208,
      "learning_rate": 2.0220968972357592e-08,
      "loss": 0.0188,
      "step": 18816
    },
    {
      "epoch": 0.9717015233668991,
      "grad_norm": 2.498790740966797,
      "learning_rate": 2.014742079684495e-08,
      "loss": 0.0211,
      "step": 18817
    },
    {
      "epoch": 0.9717531629227989,
      "grad_norm": 3.4604196548461914,
      "learning_rate": 2.0074006350671028e-08,
      "loss": 0.0493,
      "step": 18818
    },
    {
      "epoch": 0.9718048024786987,
      "grad_norm": 2.6200056076049805,
      "learning_rate": 2.0000725635807018e-08,
      "loss": 0.0608,
      "step": 18819
    },
    {
      "epoch": 0.9718564420345985,
      "grad_norm": 2.294055700302124,
      "learning_rate": 1.9927578654220792e-08,
      "loss": 0.026,
      "step": 18820
    },
    {
      "epoch": 0.9719080815904984,
      "grad_norm": 4.182422161102295,
      "learning_rate": 1.9854565407876892e-08,
      "loss": 0.0217,
      "step": 18821
    },
    {
      "epoch": 0.9719597211463982,
      "grad_norm": 3.615565061569214,
      "learning_rate": 1.978168589873597e-08,
      "loss": 0.1034,
      "step": 18822
    },
    {
      "epoch": 0.972011360702298,
      "grad_norm": 1.7825777530670166,
      "learning_rate": 1.9708940128754794e-08,
      "loss": 0.0202,
      "step": 18823
    },
    {
      "epoch": 0.9720630002581978,
      "grad_norm": 2.611273765563965,
      "learning_rate": 1.9636328099886804e-08,
      "loss": 0.0389,
      "step": 18824
    },
    {
      "epoch": 0.9721146398140976,
      "grad_norm": 6.164133548736572,
      "learning_rate": 1.9563849814082658e-08,
      "loss": 0.0437,
      "step": 18825
    },
    {
      "epoch": 0.9721662793699974,
      "grad_norm": 2.9270317554473877,
      "learning_rate": 1.949150527328858e-08,
      "loss": 0.0324,
      "step": 18826
    },
    {
      "epoch": 0.9722179189258973,
      "grad_norm": 3.164088249206543,
      "learning_rate": 1.9419294479446348e-08,
      "loss": 0.0301,
      "step": 18827
    },
    {
      "epoch": 0.9722695584817971,
      "grad_norm": 18.359947204589844,
      "learning_rate": 1.934721743449608e-08,
      "loss": 0.1217,
      "step": 18828
    },
    {
      "epoch": 0.9723211980376969,
      "grad_norm": 6.7209906578063965,
      "learning_rate": 1.9275274140373447e-08,
      "loss": 0.0747,
      "step": 18829
    },
    {
      "epoch": 0.9723728375935967,
      "grad_norm": 2.418506145477295,
      "learning_rate": 1.9203464599009126e-08,
      "loss": 0.0325,
      "step": 18830
    },
    {
      "epoch": 0.9724244771494965,
      "grad_norm": 8.327317237854004,
      "learning_rate": 1.913178881233324e-08,
      "loss": 0.0998,
      "step": 18831
    },
    {
      "epoch": 0.9724761167053964,
      "grad_norm": 2.0214686393737793,
      "learning_rate": 1.9060246782269254e-08,
      "loss": 0.0237,
      "step": 18832
    },
    {
      "epoch": 0.9725277562612962,
      "grad_norm": 2.413578510284424,
      "learning_rate": 1.898883851073896e-08,
      "loss": 0.0601,
      "step": 18833
    },
    {
      "epoch": 0.972579395817196,
      "grad_norm": 4.889910697937012,
      "learning_rate": 1.8917563999659715e-08,
      "loss": 0.0839,
      "step": 18834
    },
    {
      "epoch": 0.9726310353730958,
      "grad_norm": 16.6942081451416,
      "learning_rate": 1.8846423250945546e-08,
      "loss": 0.0898,
      "step": 18835
    },
    {
      "epoch": 0.9726826749289956,
      "grad_norm": 1.9515151977539062,
      "learning_rate": 1.877541626650714e-08,
      "loss": 0.0302,
      "step": 18836
    },
    {
      "epoch": 0.9727343144848954,
      "grad_norm": 1.9367650747299194,
      "learning_rate": 1.8704543048251312e-08,
      "loss": 0.0197,
      "step": 18837
    },
    {
      "epoch": 0.9727859540407953,
      "grad_norm": 2.163072347640991,
      "learning_rate": 1.863380359808098e-08,
      "loss": 0.0343,
      "step": 18838
    },
    {
      "epoch": 0.9728375935966951,
      "grad_norm": 11.674485206604004,
      "learning_rate": 1.8563197917895736e-08,
      "loss": 0.0516,
      "step": 18839
    },
    {
      "epoch": 0.9728892331525949,
      "grad_norm": 13.181214332580566,
      "learning_rate": 1.8492726009592398e-08,
      "loss": 0.0964,
      "step": 18840
    },
    {
      "epoch": 0.9729408727084947,
      "grad_norm": 2.8523216247558594,
      "learning_rate": 1.8422387875062232e-08,
      "loss": 0.0276,
      "step": 18841
    },
    {
      "epoch": 0.9729925122643945,
      "grad_norm": 4.992142200469971,
      "learning_rate": 1.8352183516195387e-08,
      "loss": 0.0654,
      "step": 18842
    },
    {
      "epoch": 0.9730441518202944,
      "grad_norm": 4.562366485595703,
      "learning_rate": 1.828211293487592e-08,
      "loss": 0.0221,
      "step": 18843
    },
    {
      "epoch": 0.9730957913761942,
      "grad_norm": 0.8929346203804016,
      "learning_rate": 1.8212176132986758e-08,
      "loss": 0.0117,
      "step": 18844
    },
    {
      "epoch": 0.973147430932094,
      "grad_norm": 8.803160667419434,
      "learning_rate": 1.8142373112404744e-08,
      "loss": 0.0653,
      "step": 18845
    },
    {
      "epoch": 0.9731990704879938,
      "grad_norm": 42.58123016357422,
      "learning_rate": 1.8072703875005036e-08,
      "loss": 0.1692,
      "step": 18846
    },
    {
      "epoch": 0.9732507100438936,
      "grad_norm": 19.51511573791504,
      "learning_rate": 1.800316842265837e-08,
      "loss": 0.1479,
      "step": 18847
    },
    {
      "epoch": 0.9733023495997934,
      "grad_norm": 4.469274520874023,
      "learning_rate": 1.793376675723213e-08,
      "loss": 0.0315,
      "step": 18848
    },
    {
      "epoch": 0.9733539891556933,
      "grad_norm": 3.6638882160186768,
      "learning_rate": 1.7864498880590388e-08,
      "loss": 0.0659,
      "step": 18849
    },
    {
      "epoch": 0.9734056287115931,
      "grad_norm": 7.965056896209717,
      "learning_rate": 1.7795364794592206e-08,
      "loss": 0.0435,
      "step": 18850
    },
    {
      "epoch": 0.9734572682674929,
      "grad_norm": 1.750670075416565,
      "learning_rate": 1.772636450109555e-08,
      "loss": 0.0171,
      "step": 18851
    },
    {
      "epoch": 0.9735089078233927,
      "grad_norm": 10.796610832214355,
      "learning_rate": 1.7657498001952266e-08,
      "loss": 0.0942,
      "step": 18852
    },
    {
      "epoch": 0.9735605473792925,
      "grad_norm": 2.8958141803741455,
      "learning_rate": 1.7588765299011988e-08,
      "loss": 0.0447,
      "step": 18853
    },
    {
      "epoch": 0.9736121869351924,
      "grad_norm": 2.8118348121643066,
      "learning_rate": 1.752016639412102e-08,
      "loss": 0.0341,
      "step": 18854
    },
    {
      "epoch": 0.9736638264910922,
      "grad_norm": 1.970179796218872,
      "learning_rate": 1.7451701289120658e-08,
      "loss": 0.0523,
      "step": 18855
    },
    {
      "epoch": 0.973715466046992,
      "grad_norm": 2.4482340812683105,
      "learning_rate": 1.7383369985849997e-08,
      "loss": 0.0393,
      "step": 18856
    },
    {
      "epoch": 0.9737671056028918,
      "grad_norm": 2.2621355056762695,
      "learning_rate": 1.7315172486143673e-08,
      "loss": 0.0271,
      "step": 18857
    },
    {
      "epoch": 0.9738187451587916,
      "grad_norm": 6.670032024383545,
      "learning_rate": 1.7247108791833554e-08,
      "loss": 0.0345,
      "step": 18858
    },
    {
      "epoch": 0.9738703847146915,
      "grad_norm": 3.145066976547241,
      "learning_rate": 1.7179178904746518e-08,
      "loss": 0.0318,
      "step": 18859
    },
    {
      "epoch": 0.9739220242705913,
      "grad_norm": 7.971834182739258,
      "learning_rate": 1.711138282670777e-08,
      "loss": 0.0777,
      "step": 18860
    },
    {
      "epoch": 0.9739736638264911,
      "grad_norm": 2.3822364807128906,
      "learning_rate": 1.7043720559538067e-08,
      "loss": 0.0338,
      "step": 18861
    },
    {
      "epoch": 0.9740253033823909,
      "grad_norm": 4.477641582489014,
      "learning_rate": 1.697619210505319e-08,
      "loss": 0.0496,
      "step": 18862
    },
    {
      "epoch": 0.9740769429382907,
      "grad_norm": 2.510615825653076,
      "learning_rate": 1.6908797465067795e-08,
      "loss": 0.0259,
      "step": 18863
    },
    {
      "epoch": 0.9741285824941905,
      "grad_norm": 3.9368298053741455,
      "learning_rate": 1.684153664139043e-08,
      "loss": 0.0882,
      "step": 18864
    },
    {
      "epoch": 0.9741802220500904,
      "grad_norm": 2.2352702617645264,
      "learning_rate": 1.6774409635828547e-08,
      "loss": 0.0458,
      "step": 18865
    },
    {
      "epoch": 0.9742318616059902,
      "grad_norm": 4.90532922744751,
      "learning_rate": 1.6707416450184034e-08,
      "loss": 0.0445,
      "step": 18866
    },
    {
      "epoch": 0.97428350116189,
      "grad_norm": 4.6408491134643555,
      "learning_rate": 1.664055708625656e-08,
      "loss": 0.0405,
      "step": 18867
    },
    {
      "epoch": 0.9743351407177898,
      "grad_norm": 9.036515235900879,
      "learning_rate": 1.6573831545840802e-08,
      "loss": 0.1044,
      "step": 18868
    },
    {
      "epoch": 0.9743867802736896,
      "grad_norm": 3.250706672668457,
      "learning_rate": 1.6507239830729214e-08,
      "loss": 0.0727,
      "step": 18869
    },
    {
      "epoch": 0.9744384198295895,
      "grad_norm": 6.99859094619751,
      "learning_rate": 1.644078194270926e-08,
      "loss": 0.0944,
      "step": 18870
    },
    {
      "epoch": 0.9744900593854893,
      "grad_norm": 5.497206211090088,
      "learning_rate": 1.637445788356673e-08,
      "loss": 0.0403,
      "step": 18871
    },
    {
      "epoch": 0.9745416989413891,
      "grad_norm": 6.389498710632324,
      "learning_rate": 1.6308267655081866e-08,
      "loss": 0.0815,
      "step": 18872
    },
    {
      "epoch": 0.9745933384972889,
      "grad_norm": 1.9238876104354858,
      "learning_rate": 1.6242211259032693e-08,
      "loss": 0.0232,
      "step": 18873
    },
    {
      "epoch": 0.9746449780531887,
      "grad_norm": 6.717078685760498,
      "learning_rate": 1.6176288697192787e-08,
      "loss": 0.0361,
      "step": 18874
    },
    {
      "epoch": 0.9746966176090885,
      "grad_norm": 4.667872905731201,
      "learning_rate": 1.6110499971332405e-08,
      "loss": 0.0452,
      "step": 18875
    },
    {
      "epoch": 0.9747482571649884,
      "grad_norm": 3.025846004486084,
      "learning_rate": 1.604484508321791e-08,
      "loss": 0.0309,
      "step": 18876
    },
    {
      "epoch": 0.9747998967208882,
      "grad_norm": 1.9917320013046265,
      "learning_rate": 1.597932403461344e-08,
      "loss": 0.0238,
      "step": 18877
    },
    {
      "epoch": 0.974851536276788,
      "grad_norm": 3.0739126205444336,
      "learning_rate": 1.59139368272776e-08,
      "loss": 0.0295,
      "step": 18878
    },
    {
      "epoch": 0.9749031758326878,
      "grad_norm": 9.056900024414062,
      "learning_rate": 1.5848683462966196e-08,
      "loss": 0.1074,
      "step": 18879
    },
    {
      "epoch": 0.9749548153885876,
      "grad_norm": 5.616401195526123,
      "learning_rate": 1.5783563943432835e-08,
      "loss": 0.0409,
      "step": 18880
    },
    {
      "epoch": 0.9750064549444875,
      "grad_norm": 3.6278789043426514,
      "learning_rate": 1.5718578270424445e-08,
      "loss": 0.0475,
      "step": 18881
    },
    {
      "epoch": 0.9750580945003873,
      "grad_norm": 3.3536911010742188,
      "learning_rate": 1.5653726445687413e-08,
      "loss": 0.051,
      "step": 18882
    },
    {
      "epoch": 0.9751097340562871,
      "grad_norm": 2.541194438934326,
      "learning_rate": 1.5589008470963118e-08,
      "loss": 0.0471,
      "step": 18883
    },
    {
      "epoch": 0.9751613736121869,
      "grad_norm": 2.623450517654419,
      "learning_rate": 1.5524424347989064e-08,
      "loss": 0.0344,
      "step": 18884
    },
    {
      "epoch": 0.9752130131680867,
      "grad_norm": 3.274135112762451,
      "learning_rate": 1.545997407849942e-08,
      "loss": 0.0528,
      "step": 18885
    },
    {
      "epoch": 0.9752646527239865,
      "grad_norm": 4.210451126098633,
      "learning_rate": 1.5395657664225573e-08,
      "loss": 0.0391,
      "step": 18886
    },
    {
      "epoch": 0.9753162922798864,
      "grad_norm": 6.56658411026001,
      "learning_rate": 1.533147510689448e-08,
      "loss": 0.0328,
      "step": 18887
    },
    {
      "epoch": 0.9753679318357862,
      "grad_norm": 20.02571678161621,
      "learning_rate": 1.5267426408229756e-08,
      "loss": 0.1166,
      "step": 18888
    },
    {
      "epoch": 0.975419571391686,
      "grad_norm": 5.351398944854736,
      "learning_rate": 1.52035115699517e-08,
      "loss": 0.0584,
      "step": 18889
    },
    {
      "epoch": 0.9754712109475858,
      "grad_norm": 2.489919424057007,
      "learning_rate": 1.5139730593775604e-08,
      "loss": 0.0446,
      "step": 18890
    },
    {
      "epoch": 0.9755228505034856,
      "grad_norm": 2.0241291522979736,
      "learning_rate": 1.5076083481415095e-08,
      "loss": 0.0177,
      "step": 18891
    },
    {
      "epoch": 0.9755744900593855,
      "grad_norm": 8.22432804107666,
      "learning_rate": 1.501257023457936e-08,
      "loss": 0.0368,
      "step": 18892
    },
    {
      "epoch": 0.9756261296152853,
      "grad_norm": 3.170261859893799,
      "learning_rate": 1.494919085497315e-08,
      "loss": 0.0261,
      "step": 18893
    },
    {
      "epoch": 0.9756777691711851,
      "grad_norm": 2.612821102142334,
      "learning_rate": 1.4885945344299546e-08,
      "loss": 0.0246,
      "step": 18894
    },
    {
      "epoch": 0.9757294087270849,
      "grad_norm": 2.902409553527832,
      "learning_rate": 1.4822833704257189e-08,
      "loss": 0.0443,
      "step": 18895
    },
    {
      "epoch": 0.9757810482829847,
      "grad_norm": 10.369284629821777,
      "learning_rate": 1.475985593653917e-08,
      "loss": 0.1025,
      "step": 18896
    },
    {
      "epoch": 0.9758326878388845,
      "grad_norm": 8.486371994018555,
      "learning_rate": 1.4697012042838021e-08,
      "loss": 0.0369,
      "step": 18897
    },
    {
      "epoch": 0.9758843273947844,
      "grad_norm": 5.662904739379883,
      "learning_rate": 1.4634302024841285e-08,
      "loss": 0.0873,
      "step": 18898
    },
    {
      "epoch": 0.9759359669506842,
      "grad_norm": 8.943097114562988,
      "learning_rate": 1.4571725884232612e-08,
      "loss": 0.0907,
      "step": 18899
    },
    {
      "epoch": 0.975987606506584,
      "grad_norm": 2.159403085708618,
      "learning_rate": 1.4509283622692882e-08,
      "loss": 0.0254,
      "step": 18900
    },
    {
      "epoch": 0.9760392460624838,
      "grad_norm": 10.720433235168457,
      "learning_rate": 1.4446975241898532e-08,
      "loss": 0.0456,
      "step": 18901
    },
    {
      "epoch": 0.9760908856183836,
      "grad_norm": 1.7548762559890747,
      "learning_rate": 1.4384800743522664e-08,
      "loss": 0.026,
      "step": 18902
    },
    {
      "epoch": 0.9761425251742835,
      "grad_norm": 3.027127265930176,
      "learning_rate": 1.4322760129235058e-08,
      "loss": 0.0438,
      "step": 18903
    },
    {
      "epoch": 0.9761941647301833,
      "grad_norm": 3.1299445629119873,
      "learning_rate": 1.4260853400702156e-08,
      "loss": 0.0553,
      "step": 18904
    },
    {
      "epoch": 0.9762458042860831,
      "grad_norm": 2.05273175239563,
      "learning_rate": 1.4199080559585409e-08,
      "loss": 0.0169,
      "step": 18905
    },
    {
      "epoch": 0.9762974438419829,
      "grad_norm": 7.201337814331055,
      "learning_rate": 1.4137441607545154e-08,
      "loss": 0.0702,
      "step": 18906
    },
    {
      "epoch": 0.9763490833978827,
      "grad_norm": 2.8381433486938477,
      "learning_rate": 1.4075936546235625e-08,
      "loss": 0.0481,
      "step": 18907
    },
    {
      "epoch": 0.9764007229537826,
      "grad_norm": 6.787026882171631,
      "learning_rate": 1.4014565377308276e-08,
      "loss": 0.0809,
      "step": 18908
    },
    {
      "epoch": 0.9764523625096824,
      "grad_norm": 2.251542806625366,
      "learning_rate": 1.3953328102412345e-08,
      "loss": 0.0382,
      "step": 18909
    },
    {
      "epoch": 0.9765040020655822,
      "grad_norm": 4.079498291015625,
      "learning_rate": 1.3892224723190961e-08,
      "loss": 0.0338,
      "step": 18910
    },
    {
      "epoch": 0.976555641621482,
      "grad_norm": 2.006887912750244,
      "learning_rate": 1.383125524128559e-08,
      "loss": 0.031,
      "step": 18911
    },
    {
      "epoch": 0.9766072811773818,
      "grad_norm": 14.573841094970703,
      "learning_rate": 1.377041965833381e-08,
      "loss": 0.1049,
      "step": 18912
    },
    {
      "epoch": 0.9766589207332816,
      "grad_norm": 4.091441631317139,
      "learning_rate": 1.3709717975969317e-08,
      "loss": 0.0293,
      "step": 18913
    },
    {
      "epoch": 0.9767105602891815,
      "grad_norm": 5.935326099395752,
      "learning_rate": 1.3649150195821358e-08,
      "loss": 0.0871,
      "step": 18914
    },
    {
      "epoch": 0.9767621998450813,
      "grad_norm": 4.322079181671143,
      "learning_rate": 1.358871631951697e-08,
      "loss": 0.0353,
      "step": 18915
    },
    {
      "epoch": 0.9768138394009811,
      "grad_norm": 7.859781742095947,
      "learning_rate": 1.3528416348679297e-08,
      "loss": 0.1087,
      "step": 18916
    },
    {
      "epoch": 0.9768654789568809,
      "grad_norm": 1.905853509902954,
      "learning_rate": 1.3468250284927043e-08,
      "loss": 0.0203,
      "step": 18917
    },
    {
      "epoch": 0.9769171185127807,
      "grad_norm": 7.846630096435547,
      "learning_rate": 1.3408218129876693e-08,
      "loss": 0.1108,
      "step": 18918
    },
    {
      "epoch": 0.9769687580686806,
      "grad_norm": 4.785872936248779,
      "learning_rate": 1.3348319885139183e-08,
      "loss": 0.0502,
      "step": 18919
    },
    {
      "epoch": 0.9770203976245805,
      "grad_norm": 8.359865188598633,
      "learning_rate": 1.3288555552324333e-08,
      "loss": 0.0397,
      "step": 18920
    },
    {
      "epoch": 0.9770720371804803,
      "grad_norm": 4.02146577835083,
      "learning_rate": 1.3228925133036419e-08,
      "loss": 0.0434,
      "step": 18921
    },
    {
      "epoch": 0.9771236767363801,
      "grad_norm": 6.994601726531982,
      "learning_rate": 1.3169428628876934e-08,
      "loss": 0.046,
      "step": 18922
    },
    {
      "epoch": 0.9771753162922799,
      "grad_norm": 4.291754245758057,
      "learning_rate": 1.3110066041442937e-08,
      "loss": 0.0522,
      "step": 18923
    },
    {
      "epoch": 0.9772269558481798,
      "grad_norm": 11.058145523071289,
      "learning_rate": 1.3050837372329816e-08,
      "loss": 0.1611,
      "step": 18924
    },
    {
      "epoch": 0.9772785954040796,
      "grad_norm": 2.0942959785461426,
      "learning_rate": 1.2991742623126857e-08,
      "loss": 0.0286,
      "step": 18925
    },
    {
      "epoch": 0.9773302349599794,
      "grad_norm": 10.023359298706055,
      "learning_rate": 1.2932781795421678e-08,
      "loss": 0.0717,
      "step": 18926
    },
    {
      "epoch": 0.9773818745158792,
      "grad_norm": 11.180147171020508,
      "learning_rate": 1.2873954890797457e-08,
      "loss": 0.0479,
      "step": 18927
    },
    {
      "epoch": 0.977433514071779,
      "grad_norm": 14.99605655670166,
      "learning_rate": 1.2815261910834597e-08,
      "loss": 0.1702,
      "step": 18928
    },
    {
      "epoch": 0.9774851536276788,
      "grad_norm": 4.167405128479004,
      "learning_rate": 1.275670285710795e-08,
      "loss": 0.0604,
      "step": 18929
    },
    {
      "epoch": 0.9775367931835787,
      "grad_norm": 1.7959012985229492,
      "learning_rate": 1.2698277731190701e-08,
      "loss": 0.0198,
      "step": 18930
    },
    {
      "epoch": 0.9775884327394785,
      "grad_norm": 4.540087699890137,
      "learning_rate": 1.2639986534651593e-08,
      "loss": 0.0283,
      "step": 18931
    },
    {
      "epoch": 0.9776400722953783,
      "grad_norm": 3.68385648727417,
      "learning_rate": 1.2581829269056601e-08,
      "loss": 0.0376,
      "step": 18932
    },
    {
      "epoch": 0.9776917118512781,
      "grad_norm": 3.6378679275512695,
      "learning_rate": 1.2523805935967249e-08,
      "loss": 0.0253,
      "step": 18933
    },
    {
      "epoch": 0.9777433514071779,
      "grad_norm": 2.8007702827453613,
      "learning_rate": 1.246591653694118e-08,
      "loss": 0.018,
      "step": 18934
    },
    {
      "epoch": 0.9777949909630778,
      "grad_norm": 2.397367238998413,
      "learning_rate": 1.2408161073533265e-08,
      "loss": 0.0408,
      "step": 18935
    },
    {
      "epoch": 0.9778466305189776,
      "grad_norm": 3.6296281814575195,
      "learning_rate": 1.2350539547294483e-08,
      "loss": 0.0613,
      "step": 18936
    },
    {
      "epoch": 0.9778982700748774,
      "grad_norm": 18.1807861328125,
      "learning_rate": 1.2293051959772484e-08,
      "loss": 0.0839,
      "step": 18937
    },
    {
      "epoch": 0.9779499096307772,
      "grad_norm": 3.1248199939727783,
      "learning_rate": 1.2235698312511035e-08,
      "loss": 0.0321,
      "step": 18938
    },
    {
      "epoch": 0.978001549186677,
      "grad_norm": 4.483521938323975,
      "learning_rate": 1.217847860704946e-08,
      "loss": 0.0392,
      "step": 18939
    },
    {
      "epoch": 0.9780531887425769,
      "grad_norm": 3.3233370780944824,
      "learning_rate": 1.2121392844925417e-08,
      "loss": 0.0394,
      "step": 18940
    },
    {
      "epoch": 0.9781048282984767,
      "grad_norm": 4.358649253845215,
      "learning_rate": 1.206444102767157e-08,
      "loss": 0.0542,
      "step": 18941
    },
    {
      "epoch": 0.9781564678543765,
      "grad_norm": 3.6161909103393555,
      "learning_rate": 1.200762315681725e-08,
      "loss": 0.0492,
      "step": 18942
    },
    {
      "epoch": 0.9782081074102763,
      "grad_norm": 9.624553680419922,
      "learning_rate": 1.1950939233887903e-08,
      "loss": 0.1287,
      "step": 18943
    },
    {
      "epoch": 0.9782597469661761,
      "grad_norm": 2.296391248703003,
      "learning_rate": 1.1894389260405647e-08,
      "loss": 0.0463,
      "step": 18944
    },
    {
      "epoch": 0.9783113865220759,
      "grad_norm": 4.185965538024902,
      "learning_rate": 1.1837973237889821e-08,
      "loss": 0.0477,
      "step": 18945
    },
    {
      "epoch": 0.9783630260779758,
      "grad_norm": 5.232689380645752,
      "learning_rate": 1.1781691167854769e-08,
      "loss": 0.06,
      "step": 18946
    },
    {
      "epoch": 0.9784146656338756,
      "grad_norm": 6.020728588104248,
      "learning_rate": 1.1725543051812615e-08,
      "loss": 0.1111,
      "step": 18947
    },
    {
      "epoch": 0.9784663051897754,
      "grad_norm": 4.232213973999023,
      "learning_rate": 1.1669528891270487e-08,
      "loss": 0.0469,
      "step": 18948
    },
    {
      "epoch": 0.9785179447456752,
      "grad_norm": 2.057352304458618,
      "learning_rate": 1.1613648687732737e-08,
      "loss": 0.0282,
      "step": 18949
    },
    {
      "epoch": 0.978569584301575,
      "grad_norm": 1.43333899974823,
      "learning_rate": 1.1557902442700385e-08,
      "loss": 0.0126,
      "step": 18950
    },
    {
      "epoch": 0.9786212238574749,
      "grad_norm": 20.362918853759766,
      "learning_rate": 1.150229015766946e-08,
      "loss": 0.1262,
      "step": 18951
    },
    {
      "epoch": 0.9786728634133747,
      "grad_norm": 10.330398559570312,
      "learning_rate": 1.1446811834134875e-08,
      "loss": 0.1233,
      "step": 18952
    },
    {
      "epoch": 0.9787245029692745,
      "grad_norm": 2.1441195011138916,
      "learning_rate": 1.1391467473584883e-08,
      "loss": 0.0374,
      "step": 18953
    },
    {
      "epoch": 0.9787761425251743,
      "grad_norm": 3.6426353454589844,
      "learning_rate": 1.1336257077506629e-08,
      "loss": 0.0262,
      "step": 18954
    },
    {
      "epoch": 0.9788277820810741,
      "grad_norm": 3.7648472785949707,
      "learning_rate": 1.1281180647382816e-08,
      "loss": 0.0416,
      "step": 18955
    },
    {
      "epoch": 0.978879421636974,
      "grad_norm": 2.2534358501434326,
      "learning_rate": 1.1226238184691706e-08,
      "loss": 0.0466,
      "step": 18956
    },
    {
      "epoch": 0.9789310611928738,
      "grad_norm": 2.8686420917510986,
      "learning_rate": 1.1171429690909896e-08,
      "loss": 0.0442,
      "step": 18957
    },
    {
      "epoch": 0.9789827007487736,
      "grad_norm": 4.614761829376221,
      "learning_rate": 1.1116755167507875e-08,
      "loss": 0.0766,
      "step": 18958
    },
    {
      "epoch": 0.9790343403046734,
      "grad_norm": 5.264932155609131,
      "learning_rate": 1.1062214615955024e-08,
      "loss": 0.0698,
      "step": 18959
    },
    {
      "epoch": 0.9790859798605732,
      "grad_norm": 2.645508050918579,
      "learning_rate": 1.1007808037715173e-08,
      "loss": 0.0183,
      "step": 18960
    },
    {
      "epoch": 0.979137619416473,
      "grad_norm": 12.383028030395508,
      "learning_rate": 1.0953535434249929e-08,
      "loss": 0.0785,
      "step": 18961
    },
    {
      "epoch": 0.9791892589723729,
      "grad_norm": 3.995652437210083,
      "learning_rate": 1.089939680701646e-08,
      "loss": 0.0631,
      "step": 18962
    },
    {
      "epoch": 0.9792408985282727,
      "grad_norm": 2.2520041465759277,
      "learning_rate": 1.0845392157468604e-08,
      "loss": 0.0441,
      "step": 18963
    },
    {
      "epoch": 0.9792925380841725,
      "grad_norm": 2.170255184173584,
      "learning_rate": 1.079152148705631e-08,
      "loss": 0.0293,
      "step": 18964
    },
    {
      "epoch": 0.9793441776400723,
      "grad_norm": 18.675405502319336,
      "learning_rate": 1.0737784797226759e-08,
      "loss": 0.0965,
      "step": 18965
    },
    {
      "epoch": 0.9793958171959721,
      "grad_norm": 3.0401415824890137,
      "learning_rate": 1.0684182089423234e-08,
      "loss": 0.0309,
      "step": 18966
    },
    {
      "epoch": 0.979447456751872,
      "grad_norm": 5.473809242248535,
      "learning_rate": 1.0630713365084034e-08,
      "loss": 0.0505,
      "step": 18967
    },
    {
      "epoch": 0.9794990963077718,
      "grad_norm": 2.0484695434570312,
      "learning_rate": 1.0577378625646339e-08,
      "loss": 0.0251,
      "step": 18968
    },
    {
      "epoch": 0.9795507358636716,
      "grad_norm": 4.010880470275879,
      "learning_rate": 1.0524177872541231e-08,
      "loss": 0.074,
      "step": 18969
    },
    {
      "epoch": 0.9796023754195714,
      "grad_norm": 15.093324661254883,
      "learning_rate": 1.0471111107198672e-08,
      "loss": 0.0839,
      "step": 18970
    },
    {
      "epoch": 0.9796540149754712,
      "grad_norm": 5.258695602416992,
      "learning_rate": 1.0418178331042528e-08,
      "loss": 0.059,
      "step": 18971
    },
    {
      "epoch": 0.979705654531371,
      "grad_norm": 5.168452262878418,
      "learning_rate": 1.0365379545494436e-08,
      "loss": 0.0443,
      "step": 18972
    },
    {
      "epoch": 0.9797572940872709,
      "grad_norm": 1.919724464416504,
      "learning_rate": 1.0312714751972707e-08,
      "loss": 0.0273,
      "step": 18973
    },
    {
      "epoch": 0.9798089336431707,
      "grad_norm": 1.1318449974060059,
      "learning_rate": 1.0260183951891767e-08,
      "loss": 0.0135,
      "step": 18974
    },
    {
      "epoch": 0.9798605731990705,
      "grad_norm": 5.053642749786377,
      "learning_rate": 1.0207787146661597e-08,
      "loss": 0.0798,
      "step": 18975
    },
    {
      "epoch": 0.9799122127549703,
      "grad_norm": 3.6600987911224365,
      "learning_rate": 1.0155524337689405e-08,
      "loss": 0.0334,
      "step": 18976
    },
    {
      "epoch": 0.9799638523108701,
      "grad_norm": 2.452035427093506,
      "learning_rate": 1.0103395526379068e-08,
      "loss": 0.0337,
      "step": 18977
    },
    {
      "epoch": 0.98001549186677,
      "grad_norm": 8.97156047821045,
      "learning_rate": 1.0051400714130576e-08,
      "loss": 0.1028,
      "step": 18978
    },
    {
      "epoch": 0.9800671314226698,
      "grad_norm": 2.1177453994750977,
      "learning_rate": 9.999539902339484e-09,
      "loss": 0.0414,
      "step": 18979
    },
    {
      "epoch": 0.9801187709785696,
      "grad_norm": 5.671103477478027,
      "learning_rate": 9.947813092399118e-09,
      "loss": 0.0479,
      "step": 18980
    },
    {
      "epoch": 0.9801704105344694,
      "grad_norm": 1.7681454420089722,
      "learning_rate": 9.896220285698365e-09,
      "loss": 0.0328,
      "step": 18981
    },
    {
      "epoch": 0.9802220500903692,
      "grad_norm": 3.519059419631958,
      "learning_rate": 9.844761483622234e-09,
      "loss": 0.0424,
      "step": 18982
    },
    {
      "epoch": 0.980273689646269,
      "grad_norm": 8.256855010986328,
      "learning_rate": 9.793436687553503e-09,
      "loss": 0.1592,
      "step": 18983
    },
    {
      "epoch": 0.9803253292021689,
      "grad_norm": 4.113079071044922,
      "learning_rate": 9.742245898869407e-09,
      "loss": 0.0544,
      "step": 18984
    },
    {
      "epoch": 0.9803769687580687,
      "grad_norm": 3.7551000118255615,
      "learning_rate": 9.69118911894551e-09,
      "loss": 0.0539,
      "step": 18985
    },
    {
      "epoch": 0.9804286083139685,
      "grad_norm": 1.3181747198104858,
      "learning_rate": 9.640266349152383e-09,
      "loss": 0.0161,
      "step": 18986
    },
    {
      "epoch": 0.9804802478698683,
      "grad_norm": 11.917967796325684,
      "learning_rate": 9.58947759085782e-09,
      "loss": 0.0421,
      "step": 18987
    },
    {
      "epoch": 0.9805318874257681,
      "grad_norm": 3.315739631652832,
      "learning_rate": 9.538822845425732e-09,
      "loss": 0.0509,
      "step": 18988
    },
    {
      "epoch": 0.980583526981668,
      "grad_norm": 5.7465057373046875,
      "learning_rate": 9.488302114216142e-09,
      "loss": 0.0688,
      "step": 18989
    },
    {
      "epoch": 0.9806351665375678,
      "grad_norm": 10.702590942382812,
      "learning_rate": 9.437915398585739e-09,
      "loss": 0.0979,
      "step": 18990
    },
    {
      "epoch": 0.9806868060934676,
      "grad_norm": 11.380437850952148,
      "learning_rate": 9.38766269988789e-09,
      "loss": 0.041,
      "step": 18991
    },
    {
      "epoch": 0.9807384456493674,
      "grad_norm": 1.3541538715362549,
      "learning_rate": 9.337544019471512e-09,
      "loss": 0.0333,
      "step": 18992
    },
    {
      "epoch": 0.9807900852052672,
      "grad_norm": 13.422927856445312,
      "learning_rate": 9.287559358683306e-09,
      "loss": 0.0834,
      "step": 18993
    },
    {
      "epoch": 0.980841724761167,
      "grad_norm": 17.945545196533203,
      "learning_rate": 9.23770871886498e-09,
      "loss": 0.0764,
      "step": 18994
    },
    {
      "epoch": 0.9808933643170669,
      "grad_norm": 1.9580756425857544,
      "learning_rate": 9.187992101355458e-09,
      "loss": 0.0419,
      "step": 18995
    },
    {
      "epoch": 0.9809450038729667,
      "grad_norm": 5.3470306396484375,
      "learning_rate": 9.138409507489787e-09,
      "loss": 0.0302,
      "step": 18996
    },
    {
      "epoch": 0.9809966434288665,
      "grad_norm": 8.059379577636719,
      "learning_rate": 9.088960938599123e-09,
      "loss": 0.08,
      "step": 18997
    },
    {
      "epoch": 0.9810482829847663,
      "grad_norm": 2.969160318374634,
      "learning_rate": 9.039646396011848e-09,
      "loss": 0.0424,
      "step": 18998
    },
    {
      "epoch": 0.9810999225406661,
      "grad_norm": 5.964473247528076,
      "learning_rate": 8.990465881052456e-09,
      "loss": 0.0414,
      "step": 18999
    },
    {
      "epoch": 0.981151562096566,
      "grad_norm": 3.3504552841186523,
      "learning_rate": 8.941419395041006e-09,
      "loss": 0.0376,
      "step": 19000
    },
    {
      "epoch": 0.9812032016524658,
      "grad_norm": 5.107679843902588,
      "learning_rate": 8.892506939294776e-09,
      "loss": 0.0295,
      "step": 19001
    },
    {
      "epoch": 0.9812548412083656,
      "grad_norm": 7.629509925842285,
      "learning_rate": 8.843728515127715e-09,
      "loss": 0.0892,
      "step": 19002
    },
    {
      "epoch": 0.9813064807642654,
      "grad_norm": 4.941861152648926,
      "learning_rate": 8.795084123848774e-09,
      "loss": 0.0766,
      "step": 19003
    },
    {
      "epoch": 0.9813581203201652,
      "grad_norm": 1.785809874534607,
      "learning_rate": 8.746573766765243e-09,
      "loss": 0.0308,
      "step": 19004
    },
    {
      "epoch": 0.981409759876065,
      "grad_norm": 11.68289852142334,
      "learning_rate": 8.698197445179412e-09,
      "loss": 0.1431,
      "step": 19005
    },
    {
      "epoch": 0.9814613994319649,
      "grad_norm": 12.438888549804688,
      "learning_rate": 8.649955160390245e-09,
      "loss": 0.0971,
      "step": 19006
    },
    {
      "epoch": 0.9815130389878647,
      "grad_norm": 3.1325833797454834,
      "learning_rate": 8.601846913693923e-09,
      "loss": 0.03,
      "step": 19007
    },
    {
      "epoch": 0.9815646785437645,
      "grad_norm": 4.394988059997559,
      "learning_rate": 8.553872706381638e-09,
      "loss": 0.0631,
      "step": 19008
    },
    {
      "epoch": 0.9816163180996643,
      "grad_norm": 2.296316385269165,
      "learning_rate": 8.50603253974125e-09,
      "loss": 0.0233,
      "step": 19009
    },
    {
      "epoch": 0.9816679576555641,
      "grad_norm": 3.2513034343719482,
      "learning_rate": 8.458326415058393e-09,
      "loss": 0.0371,
      "step": 19010
    },
    {
      "epoch": 0.981719597211464,
      "grad_norm": 2.8559865951538086,
      "learning_rate": 8.410754333614269e-09,
      "loss": 0.0271,
      "step": 19011
    },
    {
      "epoch": 0.9817712367673638,
      "grad_norm": 6.3361968994140625,
      "learning_rate": 8.363316296685631e-09,
      "loss": 0.0294,
      "step": 19012
    },
    {
      "epoch": 0.9818228763232636,
      "grad_norm": 4.478699207305908,
      "learning_rate": 8.316012305547017e-09,
      "loss": 0.0566,
      "step": 19013
    },
    {
      "epoch": 0.9818745158791634,
      "grad_norm": 2.995326280593872,
      "learning_rate": 8.268842361467966e-09,
      "loss": 0.0217,
      "step": 19014
    },
    {
      "epoch": 0.9819261554350632,
      "grad_norm": 3.6814839839935303,
      "learning_rate": 8.221806465715797e-09,
      "loss": 0.0381,
      "step": 19015
    },
    {
      "epoch": 0.981977794990963,
      "grad_norm": 8.631786346435547,
      "learning_rate": 8.174904619553947e-09,
      "loss": 0.129,
      "step": 19016
    },
    {
      "epoch": 0.9820294345468629,
      "grad_norm": 3.960899829864502,
      "learning_rate": 8.128136824241405e-09,
      "loss": 0.0539,
      "step": 19017
    },
    {
      "epoch": 0.9820810741027627,
      "grad_norm": 4.139713764190674,
      "learning_rate": 8.081503081033837e-09,
      "loss": 0.0661,
      "step": 19018
    },
    {
      "epoch": 0.9821327136586625,
      "grad_norm": 13.584704399108887,
      "learning_rate": 8.035003391183571e-09,
      "loss": 0.0669,
      "step": 19019
    },
    {
      "epoch": 0.9821843532145623,
      "grad_norm": 11.57344913482666,
      "learning_rate": 7.988637755940165e-09,
      "loss": 0.0745,
      "step": 19020
    },
    {
      "epoch": 0.9822359927704621,
      "grad_norm": 3.4802207946777344,
      "learning_rate": 7.942406176547623e-09,
      "loss": 0.0483,
      "step": 19021
    },
    {
      "epoch": 0.982287632326362,
      "grad_norm": 2.577211856842041,
      "learning_rate": 7.896308654248285e-09,
      "loss": 0.0334,
      "step": 19022
    },
    {
      "epoch": 0.9823392718822618,
      "grad_norm": 6.9157915115356445,
      "learning_rate": 7.85034519028005e-09,
      "loss": 0.0665,
      "step": 19023
    },
    {
      "epoch": 0.9823909114381616,
      "grad_norm": 1.6536566019058228,
      "learning_rate": 7.804515785876376e-09,
      "loss": 0.0272,
      "step": 19024
    },
    {
      "epoch": 0.9824425509940614,
      "grad_norm": 5.680605411529541,
      "learning_rate": 7.7588204422685e-09,
      "loss": 0.0253,
      "step": 19025
    },
    {
      "epoch": 0.9824941905499612,
      "grad_norm": 6.366537570953369,
      "learning_rate": 7.713259160683772e-09,
      "loss": 0.037,
      "step": 19026
    },
    {
      "epoch": 0.982545830105861,
      "grad_norm": 12.589628219604492,
      "learning_rate": 7.66783194234566e-09,
      "loss": 0.0937,
      "step": 19027
    },
    {
      "epoch": 0.9825974696617609,
      "grad_norm": 9.112232208251953,
      "learning_rate": 7.622538788473743e-09,
      "loss": 0.0661,
      "step": 19028
    },
    {
      "epoch": 0.9826491092176607,
      "grad_norm": 4.711857795715332,
      "learning_rate": 7.577379700284271e-09,
      "loss": 0.0963,
      "step": 19029
    },
    {
      "epoch": 0.9827007487735605,
      "grad_norm": 13.785157203674316,
      "learning_rate": 7.532354678990161e-09,
      "loss": 0.0797,
      "step": 19030
    },
    {
      "epoch": 0.9827523883294603,
      "grad_norm": 2.2989532947540283,
      "learning_rate": 7.487463725801002e-09,
      "loss": 0.031,
      "step": 19031
    },
    {
      "epoch": 0.9828040278853601,
      "grad_norm": 5.840073585510254,
      "learning_rate": 7.442706841921388e-09,
      "loss": 0.0333,
      "step": 19032
    },
    {
      "epoch": 0.98285566744126,
      "grad_norm": 5.616692066192627,
      "learning_rate": 7.398084028553687e-09,
      "loss": 0.0867,
      "step": 19033
    },
    {
      "epoch": 0.9829073069971598,
      "grad_norm": 4.477748870849609,
      "learning_rate": 7.353595286895831e-09,
      "loss": 0.0686,
      "step": 19034
    },
    {
      "epoch": 0.9829589465530596,
      "grad_norm": 2.4079411029815674,
      "learning_rate": 7.309240618142977e-09,
      "loss": 0.0572,
      "step": 19035
    },
    {
      "epoch": 0.9830105861089594,
      "grad_norm": 1.4980883598327637,
      "learning_rate": 7.265020023486391e-09,
      "loss": 0.0234,
      "step": 19036
    },
    {
      "epoch": 0.9830622256648592,
      "grad_norm": 6.808041095733643,
      "learning_rate": 7.2209335041134584e-09,
      "loss": 0.0394,
      "step": 19037
    },
    {
      "epoch": 0.983113865220759,
      "grad_norm": 7.403524875640869,
      "learning_rate": 7.176981061207122e-09,
      "loss": 0.1127,
      "step": 19038
    },
    {
      "epoch": 0.9831655047766589,
      "grad_norm": 11.191444396972656,
      "learning_rate": 7.133162695949214e-09,
      "loss": 0.1071,
      "step": 19039
    },
    {
      "epoch": 0.9832171443325587,
      "grad_norm": 2.961386203765869,
      "learning_rate": 7.08947840951546e-09,
      "loss": 0.0275,
      "step": 19040
    },
    {
      "epoch": 0.9832687838884585,
      "grad_norm": 9.759089469909668,
      "learning_rate": 7.045928203079366e-09,
      "loss": 0.0529,
      "step": 19041
    },
    {
      "epoch": 0.9833204234443583,
      "grad_norm": 6.349772930145264,
      "learning_rate": 7.002512077809997e-09,
      "loss": 0.0757,
      "step": 19042
    },
    {
      "epoch": 0.9833720630002581,
      "grad_norm": 4.2314653396606445,
      "learning_rate": 6.959230034873088e-09,
      "loss": 0.0615,
      "step": 19043
    },
    {
      "epoch": 0.983423702556158,
      "grad_norm": 18.19862174987793,
      "learning_rate": 6.916082075431596e-09,
      "loss": 0.0643,
      "step": 19044
    },
    {
      "epoch": 0.9834753421120578,
      "grad_norm": 4.373860836029053,
      "learning_rate": 6.8730682006445945e-09,
      "loss": 0.0548,
      "step": 19045
    },
    {
      "epoch": 0.9835269816679576,
      "grad_norm": 2.4795265197753906,
      "learning_rate": 6.830188411666161e-09,
      "loss": 0.0298,
      "step": 19046
    },
    {
      "epoch": 0.9835786212238575,
      "grad_norm": 2.5182061195373535,
      "learning_rate": 6.787442709648151e-09,
      "loss": 0.0395,
      "step": 19047
    },
    {
      "epoch": 0.9836302607797573,
      "grad_norm": 2.8804237842559814,
      "learning_rate": 6.744831095738535e-09,
      "loss": 0.0311,
      "step": 19048
    },
    {
      "epoch": 0.9836819003356572,
      "grad_norm": 7.651074409484863,
      "learning_rate": 6.702353571081399e-09,
      "loss": 0.0611,
      "step": 19049
    },
    {
      "epoch": 0.983733539891557,
      "grad_norm": 3.3762810230255127,
      "learning_rate": 6.660010136818051e-09,
      "loss": 0.0356,
      "step": 19050
    },
    {
      "epoch": 0.9837851794474568,
      "grad_norm": 12.187315940856934,
      "learning_rate": 6.617800794084805e-09,
      "loss": 0.1079,
      "step": 19051
    },
    {
      "epoch": 0.9838368190033566,
      "grad_norm": 18.349700927734375,
      "learning_rate": 6.575725544015754e-09,
      "loss": 0.0724,
      "step": 19052
    },
    {
      "epoch": 0.9838884585592564,
      "grad_norm": 1.8873205184936523,
      "learning_rate": 6.53378438774055e-09,
      "loss": 0.0343,
      "step": 19053
    },
    {
      "epoch": 0.9839400981151563,
      "grad_norm": 3.228708028793335,
      "learning_rate": 6.491977326385513e-09,
      "loss": 0.0579,
      "step": 19054
    },
    {
      "epoch": 0.9839917376710561,
      "grad_norm": 13.14412784576416,
      "learning_rate": 6.450304361073079e-09,
      "loss": 0.0401,
      "step": 19055
    },
    {
      "epoch": 0.9840433772269559,
      "grad_norm": 3.7133522033691406,
      "learning_rate": 6.408765492922353e-09,
      "loss": 0.044,
      "step": 19056
    },
    {
      "epoch": 0.9840950167828557,
      "grad_norm": 2.759256601333618,
      "learning_rate": 6.367360723049665e-09,
      "loss": 0.0336,
      "step": 19057
    },
    {
      "epoch": 0.9841466563387555,
      "grad_norm": 11.756532669067383,
      "learning_rate": 6.326090052565792e-09,
      "loss": 0.1072,
      "step": 19058
    },
    {
      "epoch": 0.9841982958946554,
      "grad_norm": 2.8471620082855225,
      "learning_rate": 6.284953482579292e-09,
      "loss": 0.0542,
      "step": 19059
    },
    {
      "epoch": 0.9842499354505552,
      "grad_norm": 16.666990280151367,
      "learning_rate": 6.243951014195393e-09,
      "loss": 0.1195,
      "step": 19060
    },
    {
      "epoch": 0.984301575006455,
      "grad_norm": 7.150744438171387,
      "learning_rate": 6.20308264851488e-09,
      "loss": 0.1156,
      "step": 19061
    },
    {
      "epoch": 0.9843532145623548,
      "grad_norm": 5.579035758972168,
      "learning_rate": 6.162348386635208e-09,
      "loss": 0.0441,
      "step": 19062
    },
    {
      "epoch": 0.9844048541182546,
      "grad_norm": 8.171073913574219,
      "learning_rate": 6.121748229649948e-09,
      "loss": 0.0979,
      "step": 19063
    },
    {
      "epoch": 0.9844564936741544,
      "grad_norm": 2.548339605331421,
      "learning_rate": 6.081282178649894e-09,
      "loss": 0.0485,
      "step": 19064
    },
    {
      "epoch": 0.9845081332300543,
      "grad_norm": 2.073312282562256,
      "learning_rate": 6.040950234721399e-09,
      "loss": 0.0433,
      "step": 19065
    },
    {
      "epoch": 0.9845597727859541,
      "grad_norm": 1.9998395442962646,
      "learning_rate": 6.00075239894804e-09,
      "loss": 0.0404,
      "step": 19066
    },
    {
      "epoch": 0.9846114123418539,
      "grad_norm": 2.499690532684326,
      "learning_rate": 5.960688672408399e-09,
      "loss": 0.0235,
      "step": 19067
    },
    {
      "epoch": 0.9846630518977537,
      "grad_norm": 7.711825847625732,
      "learning_rate": 5.920759056178837e-09,
      "loss": 0.0672,
      "step": 19068
    },
    {
      "epoch": 0.9847146914536535,
      "grad_norm": 2.2412054538726807,
      "learning_rate": 5.880963551331831e-09,
      "loss": 0.0457,
      "step": 19069
    },
    {
      "epoch": 0.9847663310095534,
      "grad_norm": 3.48757004737854,
      "learning_rate": 5.841302158935969e-09,
      "loss": 0.0346,
      "step": 19070
    },
    {
      "epoch": 0.9848179705654532,
      "grad_norm": 2.786466598510742,
      "learning_rate": 5.801774880055955e-09,
      "loss": 0.0532,
      "step": 19071
    },
    {
      "epoch": 0.984869610121353,
      "grad_norm": 1.9803276062011719,
      "learning_rate": 5.7623817157537175e-09,
      "loss": 0.0495,
      "step": 19072
    },
    {
      "epoch": 0.9849212496772528,
      "grad_norm": 2.0506350994110107,
      "learning_rate": 5.7231226670867444e-09,
      "loss": 0.0214,
      "step": 19073
    },
    {
      "epoch": 0.9849728892331526,
      "grad_norm": 4.4790496826171875,
      "learning_rate": 5.683997735109747e-09,
      "loss": 0.036,
      "step": 19074
    },
    {
      "epoch": 0.9850245287890524,
      "grad_norm": 1.909365177154541,
      "learning_rate": 5.645006920872998e-09,
      "loss": 0.0355,
      "step": 19075
    },
    {
      "epoch": 0.9850761683449523,
      "grad_norm": 2.932042121887207,
      "learning_rate": 5.606150225423435e-09,
      "loss": 0.0213,
      "step": 19076
    },
    {
      "epoch": 0.9851278079008521,
      "grad_norm": 4.082035064697266,
      "learning_rate": 5.567427649804669e-09,
      "loss": 0.044,
      "step": 19077
    },
    {
      "epoch": 0.9851794474567519,
      "grad_norm": 5.23157262802124,
      "learning_rate": 5.528839195056979e-09,
      "loss": 0.0471,
      "step": 19078
    },
    {
      "epoch": 0.9852310870126517,
      "grad_norm": 3.042186737060547,
      "learning_rate": 5.490384862216203e-09,
      "loss": 0.0467,
      "step": 19079
    },
    {
      "epoch": 0.9852827265685515,
      "grad_norm": 5.339335918426514,
      "learning_rate": 5.452064652315403e-09,
      "loss": 0.0394,
      "step": 19080
    },
    {
      "epoch": 0.9853343661244514,
      "grad_norm": 1.8101351261138916,
      "learning_rate": 5.413878566382647e-09,
      "loss": 0.0258,
      "step": 19081
    },
    {
      "epoch": 0.9853860056803512,
      "grad_norm": 2.2800025939941406,
      "learning_rate": 5.37582660544489e-09,
      "loss": 0.0332,
      "step": 19082
    },
    {
      "epoch": 0.985437645236251,
      "grad_norm": 2.4433605670928955,
      "learning_rate": 5.3379087705224265e-09,
      "loss": 0.0405,
      "step": 19083
    },
    {
      "epoch": 0.9854892847921508,
      "grad_norm": 5.586819648742676,
      "learning_rate": 5.300125062634442e-09,
      "loss": 0.0739,
      "step": 19084
    },
    {
      "epoch": 0.9855409243480506,
      "grad_norm": 3.2599785327911377,
      "learning_rate": 5.262475482795682e-09,
      "loss": 0.0559,
      "step": 19085
    },
    {
      "epoch": 0.9855925639039504,
      "grad_norm": 5.047484397888184,
      "learning_rate": 5.224960032016446e-09,
      "loss": 0.0456,
      "step": 19086
    },
    {
      "epoch": 0.9856442034598503,
      "grad_norm": 4.241508483886719,
      "learning_rate": 5.187578711304819e-09,
      "loss": 0.0294,
      "step": 19087
    },
    {
      "epoch": 0.9856958430157501,
      "grad_norm": 20.577062606811523,
      "learning_rate": 5.150331521664442e-09,
      "loss": 0.0491,
      "step": 19088
    },
    {
      "epoch": 0.9857474825716499,
      "grad_norm": 12.978013038635254,
      "learning_rate": 5.113218464095626e-09,
      "loss": 0.1659,
      "step": 19089
    },
    {
      "epoch": 0.9857991221275497,
      "grad_norm": 5.618610382080078,
      "learning_rate": 5.076239539594796e-09,
      "loss": 0.0499,
      "step": 19090
    },
    {
      "epoch": 0.9858507616834495,
      "grad_norm": 3.011904239654541,
      "learning_rate": 5.039394749155046e-09,
      "loss": 0.0419,
      "step": 19091
    },
    {
      "epoch": 0.9859024012393494,
      "grad_norm": 9.378484725952148,
      "learning_rate": 5.0026840937661415e-09,
      "loss": 0.0838,
      "step": 19092
    },
    {
      "epoch": 0.9859540407952492,
      "grad_norm": 9.536698341369629,
      "learning_rate": 4.966107574413404e-09,
      "loss": 0.1389,
      "step": 19093
    },
    {
      "epoch": 0.986005680351149,
      "grad_norm": 5.5167927742004395,
      "learning_rate": 4.92966519207938e-09,
      "loss": 0.0646,
      "step": 19094
    },
    {
      "epoch": 0.9860573199070488,
      "grad_norm": 6.143557548522949,
      "learning_rate": 4.893356947742178e-09,
      "loss": 0.1226,
      "step": 19095
    },
    {
      "epoch": 0.9861089594629486,
      "grad_norm": 7.324565410614014,
      "learning_rate": 4.857182842378238e-09,
      "loss": 0.0817,
      "step": 19096
    },
    {
      "epoch": 0.9861605990188484,
      "grad_norm": 1.884588360786438,
      "learning_rate": 4.821142876957341e-09,
      "loss": 0.0133,
      "step": 19097
    },
    {
      "epoch": 0.9862122385747483,
      "grad_norm": 2.361037492752075,
      "learning_rate": 4.785237052448155e-09,
      "loss": 0.0435,
      "step": 19098
    },
    {
      "epoch": 0.9862638781306481,
      "grad_norm": 29.807796478271484,
      "learning_rate": 4.749465369814909e-09,
      "loss": 0.0291,
      "step": 19099
    },
    {
      "epoch": 0.9863155176865479,
      "grad_norm": 10.082571029663086,
      "learning_rate": 4.713827830017392e-09,
      "loss": 0.1655,
      "step": 19100
    },
    {
      "epoch": 0.9863671572424477,
      "grad_norm": 3.0340232849121094,
      "learning_rate": 4.678324434013726e-09,
      "loss": 0.0392,
      "step": 19101
    },
    {
      "epoch": 0.9864187967983475,
      "grad_norm": 3.6594290733337402,
      "learning_rate": 4.642955182757036e-09,
      "loss": 0.031,
      "step": 19102
    },
    {
      "epoch": 0.9864704363542474,
      "grad_norm": 2.2284045219421387,
      "learning_rate": 4.607720077196564e-09,
      "loss": 0.0398,
      "step": 19103
    },
    {
      "epoch": 0.9865220759101472,
      "grad_norm": 2.9869203567504883,
      "learning_rate": 4.572619118279331e-09,
      "loss": 0.0562,
      "step": 19104
    },
    {
      "epoch": 0.986573715466047,
      "grad_norm": 2.9554648399353027,
      "learning_rate": 4.537652306946805e-09,
      "loss": 0.0479,
      "step": 19105
    },
    {
      "epoch": 0.9866253550219468,
      "grad_norm": 4.807577610015869,
      "learning_rate": 4.5028196441393446e-09,
      "loss": 0.0615,
      "step": 19106
    },
    {
      "epoch": 0.9866769945778466,
      "grad_norm": 10.31383228302002,
      "learning_rate": 4.468121130791758e-09,
      "loss": 0.1027,
      "step": 19107
    },
    {
      "epoch": 0.9867286341337465,
      "grad_norm": 4.044548034667969,
      "learning_rate": 4.433556767835523e-09,
      "loss": 0.0578,
      "step": 19108
    },
    {
      "epoch": 0.9867802736896463,
      "grad_norm": 5.233489990234375,
      "learning_rate": 4.399126556198785e-09,
      "loss": 0.0708,
      "step": 19109
    },
    {
      "epoch": 0.9868319132455461,
      "grad_norm": 2.6301522254943848,
      "learning_rate": 4.364830496806915e-09,
      "loss": 0.0283,
      "step": 19110
    },
    {
      "epoch": 0.9868835528014459,
      "grad_norm": 2.1663877964019775,
      "learning_rate": 4.330668590580289e-09,
      "loss": 0.0231,
      "step": 19111
    },
    {
      "epoch": 0.9869351923573457,
      "grad_norm": 15.412763595581055,
      "learning_rate": 4.2966408384359505e-09,
      "loss": 0.0878,
      "step": 19112
    },
    {
      "epoch": 0.9869868319132455,
      "grad_norm": 3.7152669429779053,
      "learning_rate": 4.262747241288723e-09,
      "loss": 0.0395,
      "step": 19113
    },
    {
      "epoch": 0.9870384714691454,
      "grad_norm": 5.1793317794799805,
      "learning_rate": 4.228987800047879e-09,
      "loss": 0.0319,
      "step": 19114
    },
    {
      "epoch": 0.9870901110250452,
      "grad_norm": 3.901414155960083,
      "learning_rate": 4.1953625156210265e-09,
      "loss": 0.0438,
      "step": 19115
    },
    {
      "epoch": 0.987141750580945,
      "grad_norm": 1.351683497428894,
      "learning_rate": 4.161871388909666e-09,
      "loss": 0.0221,
      "step": 19116
    },
    {
      "epoch": 0.9871933901368448,
      "grad_norm": 2.142099142074585,
      "learning_rate": 4.1285144208147445e-09,
      "loss": 0.0315,
      "step": 19117
    },
    {
      "epoch": 0.9872450296927446,
      "grad_norm": 1.534387469291687,
      "learning_rate": 4.095291612230545e-09,
      "loss": 0.0223,
      "step": 19118
    },
    {
      "epoch": 0.9872966692486445,
      "grad_norm": 3.225041389465332,
      "learning_rate": 4.062202964050244e-09,
      "loss": 0.0351,
      "step": 19119
    },
    {
      "epoch": 0.9873483088045443,
      "grad_norm": 11.601298332214355,
      "learning_rate": 4.029248477162018e-09,
      "loss": 0.054,
      "step": 19120
    },
    {
      "epoch": 0.9873999483604441,
      "grad_norm": 4.809789180755615,
      "learning_rate": 3.99642815245127e-09,
      "loss": 0.0152,
      "step": 19121
    },
    {
      "epoch": 0.9874515879163439,
      "grad_norm": 12.347987174987793,
      "learning_rate": 3.9637419907984086e-09,
      "loss": 0.0857,
      "step": 19122
    },
    {
      "epoch": 0.9875032274722437,
      "grad_norm": 2.336008310317993,
      "learning_rate": 3.931189993082174e-09,
      "loss": 0.0168,
      "step": 19123
    },
    {
      "epoch": 0.9875548670281435,
      "grad_norm": 2.6859469413757324,
      "learning_rate": 3.898772160176312e-09,
      "loss": 0.0234,
      "step": 19124
    },
    {
      "epoch": 0.9876065065840434,
      "grad_norm": 2.2259914875030518,
      "learning_rate": 3.866488492951237e-09,
      "loss": 0.0276,
      "step": 19125
    },
    {
      "epoch": 0.9876581461399432,
      "grad_norm": 15.54182243347168,
      "learning_rate": 3.834338992274589e-09,
      "loss": 0.0845,
      "step": 19126
    },
    {
      "epoch": 0.987709785695843,
      "grad_norm": 1.8386173248291016,
      "learning_rate": 3.802323659008456e-09,
      "loss": 0.0285,
      "step": 19127
    },
    {
      "epoch": 0.9877614252517428,
      "grad_norm": 2.8373637199401855,
      "learning_rate": 3.770442494013815e-09,
      "loss": 0.0279,
      "step": 19128
    },
    {
      "epoch": 0.9878130648076426,
      "grad_norm": 10.742891311645508,
      "learning_rate": 3.738695498146094e-09,
      "loss": 0.0816,
      "step": 19129
    },
    {
      "epoch": 0.9878647043635425,
      "grad_norm": 4.716490745544434,
      "learning_rate": 3.7070826722579447e-09,
      "loss": 0.0871,
      "step": 19130
    },
    {
      "epoch": 0.9879163439194423,
      "grad_norm": 11.202505111694336,
      "learning_rate": 3.6756040171986864e-09,
      "loss": 0.0664,
      "step": 19131
    },
    {
      "epoch": 0.9879679834753421,
      "grad_norm": 6.599956035614014,
      "learning_rate": 3.6442595338131994e-09,
      "loss": 0.0319,
      "step": 19132
    },
    {
      "epoch": 0.9880196230312419,
      "grad_norm": 2.558197021484375,
      "learning_rate": 3.6130492229435874e-09,
      "loss": 0.0337,
      "step": 19133
    },
    {
      "epoch": 0.9880712625871417,
      "grad_norm": 3.0188241004943848,
      "learning_rate": 3.58197308542696e-09,
      "loss": 0.0206,
      "step": 19134
    },
    {
      "epoch": 0.9881229021430415,
      "grad_norm": 4.992822647094727,
      "learning_rate": 3.551031122099313e-09,
      "loss": 0.0851,
      "step": 19135
    },
    {
      "epoch": 0.9881745416989414,
      "grad_norm": 3.141023635864258,
      "learning_rate": 3.520223333791095e-09,
      "loss": 0.0712,
      "step": 19136
    },
    {
      "epoch": 0.9882261812548412,
      "grad_norm": 9.665513038635254,
      "learning_rate": 3.4895497213288667e-09,
      "loss": 0.0927,
      "step": 19137
    },
    {
      "epoch": 0.988277820810741,
      "grad_norm": 3.933595657348633,
      "learning_rate": 3.459010285536968e-09,
      "loss": 0.0761,
      "step": 19138
    },
    {
      "epoch": 0.9883294603666408,
      "grad_norm": 2.9572582244873047,
      "learning_rate": 3.4286050272352987e-09,
      "loss": 0.0685,
      "step": 19139
    },
    {
      "epoch": 0.9883810999225406,
      "grad_norm": 4.291078090667725,
      "learning_rate": 3.398333947240984e-09,
      "loss": 0.0446,
      "step": 19140
    },
    {
      "epoch": 0.9884327394784405,
      "grad_norm": 20.254167556762695,
      "learning_rate": 3.368197046366151e-09,
      "loss": 0.1568,
      "step": 19141
    },
    {
      "epoch": 0.9884843790343403,
      "grad_norm": 9.306461334228516,
      "learning_rate": 3.338194325420152e-09,
      "loss": 0.0475,
      "step": 19142
    },
    {
      "epoch": 0.9885360185902401,
      "grad_norm": 4.1766037940979,
      "learning_rate": 3.30832578520901e-09,
      "loss": 0.0568,
      "step": 19143
    },
    {
      "epoch": 0.9885876581461399,
      "grad_norm": 1.7746460437774658,
      "learning_rate": 3.2785914265348605e-09,
      "loss": 0.0196,
      "step": 19144
    },
    {
      "epoch": 0.9886392977020397,
      "grad_norm": 2.715770959854126,
      "learning_rate": 3.248991250195954e-09,
      "loss": 0.0308,
      "step": 19145
    },
    {
      "epoch": 0.9886909372579396,
      "grad_norm": 2.2956817150115967,
      "learning_rate": 3.2195252569877655e-09,
      "loss": 0.0356,
      "step": 19146
    },
    {
      "epoch": 0.9887425768138394,
      "grad_norm": 6.070857048034668,
      "learning_rate": 3.1901934477002183e-09,
      "loss": 0.0532,
      "step": 19147
    },
    {
      "epoch": 0.9887942163697392,
      "grad_norm": 4.2004923820495605,
      "learning_rate": 3.160995823122126e-09,
      "loss": 0.0846,
      "step": 19148
    },
    {
      "epoch": 0.988845855925639,
      "grad_norm": 3.6824638843536377,
      "learning_rate": 3.1319323840378613e-09,
      "loss": 0.0581,
      "step": 19149
    },
    {
      "epoch": 0.9888974954815388,
      "grad_norm": 2.681847095489502,
      "learning_rate": 3.1030031312268005e-09,
      "loss": 0.0467,
      "step": 19150
    },
    {
      "epoch": 0.9889491350374386,
      "grad_norm": 2.8633971214294434,
      "learning_rate": 3.0742080654661e-09,
      "loss": 0.0435,
      "step": 19151
    },
    {
      "epoch": 0.9890007745933385,
      "grad_norm": 6.7784953117370605,
      "learning_rate": 3.045547187529585e-09,
      "loss": 0.0627,
      "step": 19152
    },
    {
      "epoch": 0.9890524141492383,
      "grad_norm": 10.115556716918945,
      "learning_rate": 3.0170204981866403e-09,
      "loss": 0.1067,
      "step": 19153
    },
    {
      "epoch": 0.9891040537051381,
      "grad_norm": 3.8235056400299072,
      "learning_rate": 2.9886279982033195e-09,
      "loss": 0.0392,
      "step": 19154
    },
    {
      "epoch": 0.9891556932610379,
      "grad_norm": 14.89112377166748,
      "learning_rate": 2.960369688341791e-09,
      "loss": 0.0791,
      "step": 19155
    },
    {
      "epoch": 0.9892073328169377,
      "grad_norm": 3.7460808753967285,
      "learning_rate": 2.932245569360892e-09,
      "loss": 0.034,
      "step": 19156
    },
    {
      "epoch": 0.9892589723728376,
      "grad_norm": 1.3196306228637695,
      "learning_rate": 2.9042556420166846e-09,
      "loss": 0.0171,
      "step": 19157
    },
    {
      "epoch": 0.9893106119287374,
      "grad_norm": 4.0024614334106445,
      "learning_rate": 2.8763999070596793e-09,
      "loss": 0.0485,
      "step": 19158
    },
    {
      "epoch": 0.9893622514846372,
      "grad_norm": 3.2195963859558105,
      "learning_rate": 2.848678365239277e-09,
      "loss": 0.0304,
      "step": 19159
    },
    {
      "epoch": 0.989413891040537,
      "grad_norm": 3.3545563220977783,
      "learning_rate": 2.8210910172987716e-09,
      "loss": 0.0543,
      "step": 19160
    },
    {
      "epoch": 0.9894655305964368,
      "grad_norm": 4.264334201812744,
      "learning_rate": 2.793637863978682e-09,
      "loss": 0.0339,
      "step": 19161
    },
    {
      "epoch": 0.9895171701523366,
      "grad_norm": 5.714608669281006,
      "learning_rate": 2.766318906017862e-09,
      "loss": 0.079,
      "step": 19162
    },
    {
      "epoch": 0.9895688097082365,
      "grad_norm": 2.53128719329834,
      "learning_rate": 2.739134144148503e-09,
      "loss": 0.0422,
      "step": 19163
    },
    {
      "epoch": 0.9896204492641363,
      "grad_norm": 3.2400364875793457,
      "learning_rate": 2.7120835791011324e-09,
      "loss": 0.023,
      "step": 19164
    },
    {
      "epoch": 0.9896720888200361,
      "grad_norm": 2.8372576236724854,
      "learning_rate": 2.6851672116023906e-09,
      "loss": 0.04,
      "step": 19165
    },
    {
      "epoch": 0.9897237283759359,
      "grad_norm": 15.65835952758789,
      "learning_rate": 2.6583850423744784e-09,
      "loss": 0.0997,
      "step": 19166
    },
    {
      "epoch": 0.9897753679318357,
      "grad_norm": 2.1407618522644043,
      "learning_rate": 2.6317370721368195e-09,
      "loss": 0.0287,
      "step": 19167
    },
    {
      "epoch": 0.9898270074877356,
      "grad_norm": 6.819632053375244,
      "learning_rate": 2.6052233016055084e-09,
      "loss": 0.0645,
      "step": 19168
    },
    {
      "epoch": 0.9898786470436354,
      "grad_norm": 3.620722532272339,
      "learning_rate": 2.578843731492198e-09,
      "loss": 0.0403,
      "step": 19169
    },
    {
      "epoch": 0.9899302865995352,
      "grad_norm": 3.780601739883423,
      "learning_rate": 2.5525983625052097e-09,
      "loss": 0.0507,
      "step": 19170
    },
    {
      "epoch": 0.989981926155435,
      "grad_norm": 1.2328269481658936,
      "learning_rate": 2.5264871953495364e-09,
      "loss": 0.0159,
      "step": 19171
    },
    {
      "epoch": 0.9900335657113348,
      "grad_norm": 2.3193359375,
      "learning_rate": 2.500510230726283e-09,
      "loss": 0.0376,
      "step": 19172
    },
    {
      "epoch": 0.9900852052672346,
      "grad_norm": 2.2547097206115723,
      "learning_rate": 2.4746674693332252e-09,
      "loss": 0.0292,
      "step": 19173
    },
    {
      "epoch": 0.9901368448231346,
      "grad_norm": 3.7358744144439697,
      "learning_rate": 2.4489589118636967e-09,
      "loss": 0.0187,
      "step": 19174
    },
    {
      "epoch": 0.9901884843790344,
      "grad_norm": 3.713393211364746,
      "learning_rate": 2.423384559008257e-09,
      "loss": 0.0462,
      "step": 19175
    },
    {
      "epoch": 0.9902401239349342,
      "grad_norm": 3.3626692295074463,
      "learning_rate": 2.397944411454689e-09,
      "loss": 0.0696,
      "step": 19176
    },
    {
      "epoch": 0.990291763490834,
      "grad_norm": 2.297826051712036,
      "learning_rate": 2.3726384698852243e-09,
      "loss": 0.0354,
      "step": 19177
    },
    {
      "epoch": 0.9903434030467339,
      "grad_norm": 4.90449857711792,
      "learning_rate": 2.34746673497932e-09,
      "loss": 0.0907,
      "step": 19178
    },
    {
      "epoch": 0.9903950426026337,
      "grad_norm": 7.111511707305908,
      "learning_rate": 2.3224292074131016e-09,
      "loss": 0.0995,
      "step": 19179
    },
    {
      "epoch": 0.9904466821585335,
      "grad_norm": 3.780674695968628,
      "learning_rate": 2.2975258878593642e-09,
      "loss": 0.0452,
      "step": 19180
    },
    {
      "epoch": 0.9904983217144333,
      "grad_norm": 2.508378744125366,
      "learning_rate": 2.2727567769864624e-09,
      "loss": 0.0329,
      "step": 19181
    },
    {
      "epoch": 0.9905499612703331,
      "grad_norm": 14.728068351745605,
      "learning_rate": 2.248121875459974e-09,
      "loss": 0.0973,
      "step": 19182
    },
    {
      "epoch": 0.9906016008262329,
      "grad_norm": 2.556492328643799,
      "learning_rate": 2.223621183940483e-09,
      "loss": 0.0685,
      "step": 19183
    },
    {
      "epoch": 0.9906532403821328,
      "grad_norm": 4.6050310134887695,
      "learning_rate": 2.199254703086906e-09,
      "loss": 0.0536,
      "step": 19184
    },
    {
      "epoch": 0.9907048799380326,
      "grad_norm": 3.411087989807129,
      "learning_rate": 2.1750224335537194e-09,
      "loss": 0.0236,
      "step": 19185
    },
    {
      "epoch": 0.9907565194939324,
      "grad_norm": 4.460360527038574,
      "learning_rate": 2.150924375990404e-09,
      "loss": 0.0424,
      "step": 19186
    },
    {
      "epoch": 0.9908081590498322,
      "grad_norm": 7.72755241394043,
      "learning_rate": 2.12696053104533e-09,
      "loss": 0.1405,
      "step": 19187
    },
    {
      "epoch": 0.990859798605732,
      "grad_norm": 3.4551055431365967,
      "learning_rate": 2.1031308993618713e-09,
      "loss": 0.0199,
      "step": 19188
    },
    {
      "epoch": 0.9909114381616319,
      "grad_norm": 4.239871025085449,
      "learning_rate": 2.0794354815789618e-09,
      "loss": 0.0437,
      "step": 19189
    },
    {
      "epoch": 0.9909630777175317,
      "grad_norm": 4.984833240509033,
      "learning_rate": 2.055874278333869e-09,
      "loss": 0.0559,
      "step": 19190
    },
    {
      "epoch": 0.9910147172734315,
      "grad_norm": 3.1398355960845947,
      "learning_rate": 2.0324472902588654e-09,
      "loss": 0.016,
      "step": 19191
    },
    {
      "epoch": 0.9910663568293313,
      "grad_norm": 3.061049461364746,
      "learning_rate": 2.009154517983447e-09,
      "loss": 0.0529,
      "step": 19192
    },
    {
      "epoch": 0.9911179963852311,
      "grad_norm": 7.570728778839111,
      "learning_rate": 1.985995962133225e-09,
      "loss": 0.106,
      "step": 19193
    },
    {
      "epoch": 0.9911696359411309,
      "grad_norm": 3.4028775691986084,
      "learning_rate": 1.962971623329368e-09,
      "loss": 0.0322,
      "step": 19194
    },
    {
      "epoch": 0.9912212754970308,
      "grad_norm": 4.234867572784424,
      "learning_rate": 1.940081502190827e-09,
      "loss": 0.0326,
      "step": 19195
    },
    {
      "epoch": 0.9912729150529306,
      "grad_norm": 1.9919581413269043,
      "learning_rate": 1.9173255993315544e-09,
      "loss": 0.034,
      "step": 19196
    },
    {
      "epoch": 0.9913245546088304,
      "grad_norm": 1.9255754947662354,
      "learning_rate": 1.894703915363838e-09,
      "loss": 0.0293,
      "step": 19197
    },
    {
      "epoch": 0.9913761941647302,
      "grad_norm": 33.58498764038086,
      "learning_rate": 1.87221645089386e-09,
      "loss": 0.1067,
      "step": 19198
    },
    {
      "epoch": 0.99142783372063,
      "grad_norm": 20.4117374420166,
      "learning_rate": 1.8498632065266919e-09,
      "loss": 0.1104,
      "step": 19199
    },
    {
      "epoch": 0.9914794732765299,
      "grad_norm": 6.822972297668457,
      "learning_rate": 1.8276441828618542e-09,
      "loss": 0.0685,
      "step": 19200
    },
    {
      "epoch": 0.9915311128324297,
      "grad_norm": 1.3662689924240112,
      "learning_rate": 1.8055593804960913e-09,
      "loss": 0.012,
      "step": 19201
    },
    {
      "epoch": 0.9915827523883295,
      "grad_norm": 2.21087646484375,
      "learning_rate": 1.783608800022263e-09,
      "loss": 0.0165,
      "step": 19202
    },
    {
      "epoch": 0.9916343919442293,
      "grad_norm": 5.000943183898926,
      "learning_rate": 1.7617924420304522e-09,
      "loss": 0.0459,
      "step": 19203
    },
    {
      "epoch": 0.9916860315001291,
      "grad_norm": 3.7869529724121094,
      "learning_rate": 1.7401103071063018e-09,
      "loss": 0.0533,
      "step": 19204
    },
    {
      "epoch": 0.991737671056029,
      "grad_norm": 3.145023822784424,
      "learning_rate": 1.718562395832124e-09,
      "loss": 0.0168,
      "step": 19205
    },
    {
      "epoch": 0.9917893106119288,
      "grad_norm": 10.518610000610352,
      "learning_rate": 1.6971487087857896e-09,
      "loss": 0.1579,
      "step": 19206
    },
    {
      "epoch": 0.9918409501678286,
      "grad_norm": 1.762193202972412,
      "learning_rate": 1.6758692465435045e-09,
      "loss": 0.0305,
      "step": 19207
    },
    {
      "epoch": 0.9918925897237284,
      "grad_norm": 6.597837448120117,
      "learning_rate": 1.6547240096759232e-09,
      "loss": 0.0634,
      "step": 19208
    },
    {
      "epoch": 0.9919442292796282,
      "grad_norm": 5.30997896194458,
      "learning_rate": 1.63371299875148e-09,
      "loss": 0.0768,
      "step": 19209
    },
    {
      "epoch": 0.991995868835528,
      "grad_norm": 2.3795371055603027,
      "learning_rate": 1.612836214333613e-09,
      "loss": 0.0178,
      "step": 19210
    },
    {
      "epoch": 0.9920475083914279,
      "grad_norm": 4.8357672691345215,
      "learning_rate": 1.59209365698354e-09,
      "loss": 0.059,
      "step": 19211
    },
    {
      "epoch": 0.9920991479473277,
      "grad_norm": 2.8766486644744873,
      "learning_rate": 1.5714853272580377e-09,
      "loss": 0.0378,
      "step": 19212
    },
    {
      "epoch": 0.9921507875032275,
      "grad_norm": 5.927275657653809,
      "learning_rate": 1.5510112257111077e-09,
      "loss": 0.0222,
      "step": 19213
    },
    {
      "epoch": 0.9922024270591273,
      "grad_norm": 1.925194263458252,
      "learning_rate": 1.530671352891755e-09,
      "loss": 0.0223,
      "step": 19214
    },
    {
      "epoch": 0.9922540666150271,
      "grad_norm": 4.593131065368652,
      "learning_rate": 1.5104657093462094e-09,
      "loss": 0.0688,
      "step": 19215
    },
    {
      "epoch": 0.992305706170927,
      "grad_norm": 7.2107768058776855,
      "learning_rate": 1.490394295617925e-09,
      "loss": 0.0767,
      "step": 19216
    },
    {
      "epoch": 0.9923573457268268,
      "grad_norm": 2.3456804752349854,
      "learning_rate": 1.4704571122448053e-09,
      "loss": 0.0219,
      "step": 19217
    },
    {
      "epoch": 0.9924089852827266,
      "grad_norm": 2.9721405506134033,
      "learning_rate": 1.4506541597630875e-09,
      "loss": 0.0414,
      "step": 19218
    },
    {
      "epoch": 0.9924606248386264,
      "grad_norm": 3.352585792541504,
      "learning_rate": 1.4309854387045686e-09,
      "loss": 0.0429,
      "step": 19219
    },
    {
      "epoch": 0.9925122643945262,
      "grad_norm": 3.879473924636841,
      "learning_rate": 1.4114509495966045e-09,
      "loss": 0.0597,
      "step": 19220
    },
    {
      "epoch": 0.992563903950426,
      "grad_norm": 3.2509424686431885,
      "learning_rate": 1.3920506929643307e-09,
      "loss": 0.0236,
      "step": 19221
    },
    {
      "epoch": 0.9926155435063259,
      "grad_norm": 3.7868382930755615,
      "learning_rate": 1.372784669328997e-09,
      "loss": 0.0731,
      "step": 19222
    },
    {
      "epoch": 0.9926671830622257,
      "grad_norm": 3.7875258922576904,
      "learning_rate": 1.3536528792074123e-09,
      "loss": 0.0461,
      "step": 19223
    },
    {
      "epoch": 0.9927188226181255,
      "grad_norm": 3.7960774898529053,
      "learning_rate": 1.3346553231136094e-09,
      "loss": 0.0405,
      "step": 19224
    },
    {
      "epoch": 0.9927704621740253,
      "grad_norm": 3.0971226692199707,
      "learning_rate": 1.3157920015582915e-09,
      "loss": 0.0438,
      "step": 19225
    },
    {
      "epoch": 0.9928221017299251,
      "grad_norm": 7.47879695892334,
      "learning_rate": 1.2970629150466097e-09,
      "loss": 0.0514,
      "step": 19226
    },
    {
      "epoch": 0.992873741285825,
      "grad_norm": 5.404688835144043,
      "learning_rate": 1.2784680640831604e-09,
      "loss": 0.0701,
      "step": 19227
    },
    {
      "epoch": 0.9929253808417248,
      "grad_norm": 3.812453269958496,
      "learning_rate": 1.2600074491658786e-09,
      "loss": 0.0736,
      "step": 19228
    },
    {
      "epoch": 0.9929770203976246,
      "grad_norm": 2.46016526222229,
      "learning_rate": 1.2416810707915893e-09,
      "loss": 0.0399,
      "step": 19229
    },
    {
      "epoch": 0.9930286599535244,
      "grad_norm": 2.864100933074951,
      "learning_rate": 1.223488929451566e-09,
      "loss": 0.0267,
      "step": 19230
    },
    {
      "epoch": 0.9930802995094242,
      "grad_norm": 3.6158316135406494,
      "learning_rate": 1.2054310256348623e-09,
      "loss": 0.071,
      "step": 19231
    },
    {
      "epoch": 0.993131939065324,
      "grad_norm": 6.830013275146484,
      "learning_rate": 1.187507359826645e-09,
      "loss": 0.0928,
      "step": 19232
    },
    {
      "epoch": 0.9931835786212239,
      "grad_norm": 2.6245174407958984,
      "learning_rate": 1.1697179325076413e-09,
      "loss": 0.0372,
      "step": 19233
    },
    {
      "epoch": 0.9932352181771237,
      "grad_norm": 3.6325111389160156,
      "learning_rate": 1.152062744155802e-09,
      "loss": 0.0528,
      "step": 19234
    },
    {
      "epoch": 0.9932868577330235,
      "grad_norm": 4.937611103057861,
      "learning_rate": 1.134541795245192e-09,
      "loss": 0.0664,
      "step": 19235
    },
    {
      "epoch": 0.9933384972889233,
      "grad_norm": 6.204514503479004,
      "learning_rate": 1.1171550862465463e-09,
      "loss": 0.0563,
      "step": 19236
    },
    {
      "epoch": 0.9933901368448231,
      "grad_norm": 2.4247305393218994,
      "learning_rate": 1.0999026176267135e-09,
      "loss": 0.0325,
      "step": 19237
    },
    {
      "epoch": 0.993441776400723,
      "grad_norm": 3.5018460750579834,
      "learning_rate": 1.0827843898486567e-09,
      "loss": 0.0407,
      "step": 19238
    },
    {
      "epoch": 0.9934934159566228,
      "grad_norm": 6.370998859405518,
      "learning_rate": 1.0658004033720083e-09,
      "loss": 0.0889,
      "step": 19239
    },
    {
      "epoch": 0.9935450555125226,
      "grad_norm": 3.0601425170898438,
      "learning_rate": 1.0489506586536246e-09,
      "loss": 0.03,
      "step": 19240
    },
    {
      "epoch": 0.9935966950684224,
      "grad_norm": 1.921518325805664,
      "learning_rate": 1.032235156145367e-09,
      "loss": 0.0269,
      "step": 19241
    },
    {
      "epoch": 0.9936483346243222,
      "grad_norm": 10.258472442626953,
      "learning_rate": 1.0156538962963202e-09,
      "loss": 0.0672,
      "step": 19242
    },
    {
      "epoch": 0.993699974180222,
      "grad_norm": 2.4473114013671875,
      "learning_rate": 9.99206879551684e-10,
      "loss": 0.032,
      "step": 19243
    },
    {
      "epoch": 0.9937516137361219,
      "grad_norm": 2.9944334030151367,
      "learning_rate": 9.828941063527719e-10,
      "loss": 0.0204,
      "step": 19244
    },
    {
      "epoch": 0.9938032532920217,
      "grad_norm": 3.531092643737793,
      "learning_rate": 9.667155771386771e-10,
      "loss": 0.0486,
      "step": 19245
    },
    {
      "epoch": 0.9938548928479215,
      "grad_norm": 3.674107074737549,
      "learning_rate": 9.506712923429417e-10,
      "loss": 0.0312,
      "step": 19246
    },
    {
      "epoch": 0.9939065324038213,
      "grad_norm": 4.058895587921143,
      "learning_rate": 9.347612523963323e-10,
      "loss": 0.0401,
      "step": 19247
    },
    {
      "epoch": 0.9939581719597211,
      "grad_norm": 12.184467315673828,
      "learning_rate": 9.189854577268397e-10,
      "loss": 0.1166,
      "step": 19248
    },
    {
      "epoch": 0.994009811515621,
      "grad_norm": 3.167649745941162,
      "learning_rate": 9.033439087574592e-10,
      "loss": 0.0502,
      "step": 19249
    },
    {
      "epoch": 0.9940614510715208,
      "grad_norm": 2.878601551055908,
      "learning_rate": 8.878366059089649e-10,
      "loss": 0.0377,
      "step": 19250
    },
    {
      "epoch": 0.9941130906274206,
      "grad_norm": 5.383821964263916,
      "learning_rate": 8.724635495965805e-10,
      "loss": 0.0647,
      "step": 19251
    },
    {
      "epoch": 0.9941647301833204,
      "grad_norm": 2.1680448055267334,
      "learning_rate": 8.572247402344191e-10,
      "loss": 0.0441,
      "step": 19252
    },
    {
      "epoch": 0.9942163697392202,
      "grad_norm": 1.7835291624069214,
      "learning_rate": 8.421201782310429e-10,
      "loss": 0.0303,
      "step": 19253
    },
    {
      "epoch": 0.99426800929512,
      "grad_norm": 2.8426520824432373,
      "learning_rate": 8.271498639922382e-10,
      "loss": 0.0271,
      "step": 19254
    },
    {
      "epoch": 0.9943196488510199,
      "grad_norm": 3.176637887954712,
      "learning_rate": 8.123137979199058e-10,
      "loss": 0.0447,
      "step": 19255
    },
    {
      "epoch": 0.9943712884069197,
      "grad_norm": 3.4887659549713135,
      "learning_rate": 7.976119804120608e-10,
      "loss": 0.0432,
      "step": 19256
    },
    {
      "epoch": 0.9944229279628195,
      "grad_norm": 2.1863274574279785,
      "learning_rate": 7.830444118644976e-10,
      "loss": 0.0236,
      "step": 19257
    },
    {
      "epoch": 0.9944745675187193,
      "grad_norm": 4.980393886566162,
      "learning_rate": 7.686110926680146e-10,
      "loss": 0.0733,
      "step": 19258
    },
    {
      "epoch": 0.9945262070746191,
      "grad_norm": 1.538256287574768,
      "learning_rate": 7.543120232095246e-10,
      "loss": 0.0294,
      "step": 19259
    },
    {
      "epoch": 0.994577846630519,
      "grad_norm": 1.7425483465194702,
      "learning_rate": 7.401472038737201e-10,
      "loss": 0.0193,
      "step": 19260
    },
    {
      "epoch": 0.9946294861864188,
      "grad_norm": 3.6551012992858887,
      "learning_rate": 7.261166350408521e-10,
      "loss": 0.0582,
      "step": 19261
    },
    {
      "epoch": 0.9946811257423186,
      "grad_norm": 8.973848342895508,
      "learning_rate": 7.122203170878417e-10,
      "loss": 0.0475,
      "step": 19262
    },
    {
      "epoch": 0.9947327652982184,
      "grad_norm": 7.321830749511719,
      "learning_rate": 6.984582503877235e-10,
      "loss": 0.0561,
      "step": 19263
    },
    {
      "epoch": 0.9947844048541182,
      "grad_norm": 1.5950736999511719,
      "learning_rate": 6.848304353096469e-10,
      "loss": 0.0459,
      "step": 19264
    },
    {
      "epoch": 0.994836044410018,
      "grad_norm": 4.003497123718262,
      "learning_rate": 6.713368722199854e-10,
      "loss": 0.0499,
      "step": 19265
    },
    {
      "epoch": 0.9948876839659179,
      "grad_norm": 2.405212640762329,
      "learning_rate": 6.579775614812267e-10,
      "loss": 0.0376,
      "step": 19266
    },
    {
      "epoch": 0.9949393235218177,
      "grad_norm": 1.9426664113998413,
      "learning_rate": 6.44752503451973e-10,
      "loss": 0.0298,
      "step": 19267
    },
    {
      "epoch": 0.9949909630777175,
      "grad_norm": 3.781860113143921,
      "learning_rate": 6.316616984874957e-10,
      "loss": 0.0389,
      "step": 19268
    },
    {
      "epoch": 0.9950426026336173,
      "grad_norm": 3.3828506469726562,
      "learning_rate": 6.187051469391803e-10,
      "loss": 0.0298,
      "step": 19269
    },
    {
      "epoch": 0.9950942421895171,
      "grad_norm": 2.645230531692505,
      "learning_rate": 6.058828491550817e-10,
      "loss": 0.05,
      "step": 19270
    },
    {
      "epoch": 0.995145881745417,
      "grad_norm": 4.6816253662109375,
      "learning_rate": 5.931948054788139e-10,
      "loss": 0.0404,
      "step": 19271
    },
    {
      "epoch": 0.9951975213013168,
      "grad_norm": 1.9336364269256592,
      "learning_rate": 5.806410162523257e-10,
      "loss": 0.0307,
      "step": 19272
    },
    {
      "epoch": 0.9952491608572166,
      "grad_norm": 2.256253480911255,
      "learning_rate": 5.682214818120146e-10,
      "loss": 0.0398,
      "step": 19273
    },
    {
      "epoch": 0.9953008004131164,
      "grad_norm": 7.515839576721191,
      "learning_rate": 5.559362024915027e-10,
      "loss": 0.0536,
      "step": 19274
    },
    {
      "epoch": 0.9953524399690162,
      "grad_norm": 4.20047664642334,
      "learning_rate": 5.437851786205262e-10,
      "loss": 0.0396,
      "step": 19275
    },
    {
      "epoch": 0.995404079524916,
      "grad_norm": 8.906717300415039,
      "learning_rate": 5.317684105254905e-10,
      "loss": 0.0895,
      "step": 19276
    },
    {
      "epoch": 0.9954557190808159,
      "grad_norm": 11.980323791503906,
      "learning_rate": 5.198858985294708e-10,
      "loss": 0.0743,
      "step": 19277
    },
    {
      "epoch": 0.9955073586367157,
      "grad_norm": 11.891302108764648,
      "learning_rate": 5.081376429511009e-10,
      "loss": 0.0621,
      "step": 19278
    },
    {
      "epoch": 0.9955589981926155,
      "grad_norm": 3.148383617401123,
      "learning_rate": 4.965236441062394e-10,
      "loss": 0.042,
      "step": 19279
    },
    {
      "epoch": 0.9956106377485153,
      "grad_norm": 7.963416576385498,
      "learning_rate": 4.850439023063036e-10,
      "loss": 0.1737,
      "step": 19280
    },
    {
      "epoch": 0.9956622773044151,
      "grad_norm": 19.154722213745117,
      "learning_rate": 4.736984178599357e-10,
      "loss": 0.0587,
      "step": 19281
    },
    {
      "epoch": 0.995713916860315,
      "grad_norm": 3.0829384326934814,
      "learning_rate": 4.62487191071892e-10,
      "loss": 0.0221,
      "step": 19282
    },
    {
      "epoch": 0.9957655564162148,
      "grad_norm": 7.894060134887695,
      "learning_rate": 4.5141022224304276e-10,
      "loss": 0.0699,
      "step": 19283
    },
    {
      "epoch": 0.9958171959721146,
      "grad_norm": 2.320816993713379,
      "learning_rate": 4.404675116703727e-10,
      "loss": 0.0167,
      "step": 19284
    },
    {
      "epoch": 0.9958688355280144,
      "grad_norm": 4.581374645233154,
      "learning_rate": 4.296590596486461e-10,
      "loss": 0.0347,
      "step": 19285
    },
    {
      "epoch": 0.9959204750839142,
      "grad_norm": 4.984432220458984,
      "learning_rate": 4.1898486646763104e-10,
      "loss": 0.1008,
      "step": 19286
    },
    {
      "epoch": 0.995972114639814,
      "grad_norm": 7.025736331939697,
      "learning_rate": 4.0844493241432026e-10,
      "loss": 0.0996,
      "step": 19287
    },
    {
      "epoch": 0.9960237541957139,
      "grad_norm": 2.188156843185425,
      "learning_rate": 3.980392577707104e-10,
      "loss": 0.0425,
      "step": 19288
    },
    {
      "epoch": 0.9960753937516137,
      "grad_norm": 12.713910102844238,
      "learning_rate": 3.8776784281768787e-10,
      "loss": 0.1327,
      "step": 19289
    },
    {
      "epoch": 0.9961270333075135,
      "grad_norm": 2.899770736694336,
      "learning_rate": 3.776306878300329e-10,
      "loss": 0.0258,
      "step": 19290
    },
    {
      "epoch": 0.9961786728634133,
      "grad_norm": 2.9949162006378174,
      "learning_rate": 3.676277930808603e-10,
      "loss": 0.0454,
      "step": 19291
    },
    {
      "epoch": 0.9962303124193131,
      "grad_norm": 26.13300323486328,
      "learning_rate": 3.5775915883773384e-10,
      "loss": 0.1255,
      "step": 19292
    },
    {
      "epoch": 0.996281951975213,
      "grad_norm": 12.137913703918457,
      "learning_rate": 3.4802478536599683e-10,
      "loss": 0.047,
      "step": 19293
    },
    {
      "epoch": 0.9963335915311128,
      "grad_norm": 3.3717854022979736,
      "learning_rate": 3.3842467292766193e-10,
      "loss": 0.042,
      "step": 19294
    },
    {
      "epoch": 0.9963852310870126,
      "grad_norm": 4.355104923248291,
      "learning_rate": 3.289588217803008e-10,
      "loss": 0.0759,
      "step": 19295
    },
    {
      "epoch": 0.9964368706429124,
      "grad_norm": 1.027786374092102,
      "learning_rate": 3.196272321775995e-10,
      "loss": 0.0145,
      "step": 19296
    },
    {
      "epoch": 0.9964885101988122,
      "grad_norm": 4.666506290435791,
      "learning_rate": 3.104299043704684e-10,
      "loss": 0.0688,
      "step": 19297
    },
    {
      "epoch": 0.9965401497547121,
      "grad_norm": 2.798063278198242,
      "learning_rate": 3.013668386059321e-10,
      "loss": 0.0221,
      "step": 19298
    },
    {
      "epoch": 0.9965917893106119,
      "grad_norm": 14.922015190124512,
      "learning_rate": 2.9243803512712945e-10,
      "loss": 0.0582,
      "step": 19299
    },
    {
      "epoch": 0.9966434288665117,
      "grad_norm": 4.6494903564453125,
      "learning_rate": 2.8364349417442373e-10,
      "loss": 0.0279,
      "step": 19300
    },
    {
      "epoch": 0.9966950684224115,
      "grad_norm": 9.438918113708496,
      "learning_rate": 2.7498321598373733e-10,
      "loss": 0.0677,
      "step": 19301
    },
    {
      "epoch": 0.9967467079783114,
      "grad_norm": 3.546125650405884,
      "learning_rate": 2.664572007871069e-10,
      "loss": 0.0421,
      "step": 19302
    },
    {
      "epoch": 0.9967983475342113,
      "grad_norm": 2.82464861869812,
      "learning_rate": 2.580654488137935e-10,
      "loss": 0.043,
      "step": 19303
    },
    {
      "epoch": 0.9968499870901111,
      "grad_norm": 4.213798999786377,
      "learning_rate": 2.498079602897274e-10,
      "loss": 0.1008,
      "step": 19304
    },
    {
      "epoch": 0.9969016266460109,
      "grad_norm": 5.265811443328857,
      "learning_rate": 2.4168473543584316e-10,
      "loss": 0.0619,
      "step": 19305
    },
    {
      "epoch": 0.9969532662019107,
      "grad_norm": 2.610457181930542,
      "learning_rate": 2.3369577447085455e-10,
      "loss": 0.0293,
      "step": 19306
    },
    {
      "epoch": 0.9970049057578105,
      "grad_norm": 3.293855667114258,
      "learning_rate": 2.2584107760847963e-10,
      "loss": 0.0421,
      "step": 19307
    },
    {
      "epoch": 0.9970565453137104,
      "grad_norm": 1.6802762746810913,
      "learning_rate": 2.1812064506077091e-10,
      "loss": 0.0143,
      "step": 19308
    },
    {
      "epoch": 0.9971081848696102,
      "grad_norm": 9.049948692321777,
      "learning_rate": 2.105344770342299e-10,
      "loss": 0.0187,
      "step": 19309
    },
    {
      "epoch": 0.99715982442551,
      "grad_norm": 2.6485488414764404,
      "learning_rate": 2.0308257373313766e-10,
      "loss": 0.0263,
      "step": 19310
    },
    {
      "epoch": 0.9972114639814098,
      "grad_norm": 2.118243455886841,
      "learning_rate": 1.957649353573343e-10,
      "loss": 0.0273,
      "step": 19311
    },
    {
      "epoch": 0.9972631035373096,
      "grad_norm": 4.892057418823242,
      "learning_rate": 1.8858156210277424e-10,
      "loss": 0.0729,
      "step": 19312
    },
    {
      "epoch": 0.9973147430932094,
      "grad_norm": 3.0214593410491943,
      "learning_rate": 1.815324541631913e-10,
      "loss": 0.0239,
      "step": 19313
    },
    {
      "epoch": 0.9973663826491093,
      "grad_norm": 3.7971746921539307,
      "learning_rate": 1.746176117278786e-10,
      "loss": 0.0334,
      "step": 19314
    },
    {
      "epoch": 0.9974180222050091,
      "grad_norm": 4.073285102844238,
      "learning_rate": 1.6783703498224334e-10,
      "loss": 0.0569,
      "step": 19315
    },
    {
      "epoch": 0.9974696617609089,
      "grad_norm": 1.747355580329895,
      "learning_rate": 1.61190724107807e-10,
      "loss": 0.0144,
      "step": 19316
    },
    {
      "epoch": 0.9975213013168087,
      "grad_norm": 1.9340214729309082,
      "learning_rate": 1.5467867928442571e-10,
      "loss": 0.0283,
      "step": 19317
    },
    {
      "epoch": 0.9975729408727085,
      "grad_norm": 2.731785535812378,
      "learning_rate": 1.4830090068529424e-10,
      "loss": 0.0348,
      "step": 19318
    },
    {
      "epoch": 0.9976245804286084,
      "grad_norm": 17.245269775390625,
      "learning_rate": 1.4205738848305227e-10,
      "loss": 0.0649,
      "step": 19319
    },
    {
      "epoch": 0.9976762199845082,
      "grad_norm": 8.773381233215332,
      "learning_rate": 1.359481428447884e-10,
      "loss": 0.0964,
      "step": 19320
    },
    {
      "epoch": 0.997727859540408,
      "grad_norm": 5.511264324188232,
      "learning_rate": 1.2997316393481562e-10,
      "loss": 0.0897,
      "step": 19321
    },
    {
      "epoch": 0.9977794990963078,
      "grad_norm": 12.015739440917969,
      "learning_rate": 1.2413245191300604e-10,
      "loss": 0.0492,
      "step": 19322
    },
    {
      "epoch": 0.9978311386522076,
      "grad_norm": 1.3373364210128784,
      "learning_rate": 1.1842600693701133e-10,
      "loss": 0.009,
      "step": 19323
    },
    {
      "epoch": 0.9978827782081074,
      "grad_norm": 5.317378520965576,
      "learning_rate": 1.1285382915948717e-10,
      "loss": 0.0849,
      "step": 19324
    },
    {
      "epoch": 0.9979344177640073,
      "grad_norm": 1.366349697113037,
      "learning_rate": 1.0741591873031365e-10,
      "loss": 0.0115,
      "step": 19325
    },
    {
      "epoch": 0.9979860573199071,
      "grad_norm": 9.187067031860352,
      "learning_rate": 1.021122757954851e-10,
      "loss": 0.0998,
      "step": 19326
    },
    {
      "epoch": 0.9980376968758069,
      "grad_norm": 4.740352153778076,
      "learning_rate": 9.694290049711008e-11,
      "loss": 0.0504,
      "step": 19327
    },
    {
      "epoch": 0.9980893364317067,
      "grad_norm": 6.785051345825195,
      "learning_rate": 9.190779297452157e-11,
      "loss": 0.0393,
      "step": 19328
    },
    {
      "epoch": 0.9981409759876065,
      "grad_norm": 17.750444412231445,
      "learning_rate": 8.700695336261167e-11,
      "loss": 0.1081,
      "step": 19329
    },
    {
      "epoch": 0.9981926155435064,
      "grad_norm": 1.8866472244262695,
      "learning_rate": 8.224038179294181e-11,
      "loss": 0.0204,
      "step": 19330
    },
    {
      "epoch": 0.9982442550994062,
      "grad_norm": 5.440040588378906,
      "learning_rate": 7.760807839374273e-11,
      "loss": 0.0331,
      "step": 19331
    },
    {
      "epoch": 0.998295894655306,
      "grad_norm": 2.2563860416412354,
      "learning_rate": 7.311004328991455e-11,
      "loss": 0.013,
      "step": 19332
    },
    {
      "epoch": 0.9983475342112058,
      "grad_norm": 2.8770663738250732,
      "learning_rate": 6.874627660080624e-11,
      "loss": 0.0551,
      "step": 19333
    },
    {
      "epoch": 0.9983991737671056,
      "grad_norm": 4.1083502769470215,
      "learning_rate": 6.451677844521165e-11,
      "loss": 0.0601,
      "step": 19334
    },
    {
      "epoch": 0.9984508133230054,
      "grad_norm": 2.2962422370910645,
      "learning_rate": 6.042154893526331e-11,
      "loss": 0.0325,
      "step": 19335
    },
    {
      "epoch": 0.9985024528789053,
      "grad_norm": 3.675499677658081,
      "learning_rate": 5.646058818198352e-11,
      "loss": 0.0569,
      "step": 19336
    },
    {
      "epoch": 0.9985540924348051,
      "grad_norm": 1.397897720336914,
      "learning_rate": 5.263389629139859e-11,
      "loss": 0.0242,
      "step": 19337
    },
    {
      "epoch": 0.9986057319907049,
      "grad_norm": 1.997186303138733,
      "learning_rate": 4.894147336620414e-11,
      "loss": 0.0432,
      "step": 19338
    },
    {
      "epoch": 0.9986573715466047,
      "grad_norm": 6.653584957122803,
      "learning_rate": 4.538331950521002e-11,
      "loss": 0.1343,
      "step": 19339
    },
    {
      "epoch": 0.9987090111025045,
      "grad_norm": 3.828169822692871,
      "learning_rate": 4.195943480500564e-11,
      "loss": 0.0807,
      "step": 19340
    },
    {
      "epoch": 0.9987606506584044,
      "grad_norm": 5.441251754760742,
      "learning_rate": 3.866981935662928e-11,
      "loss": 0.0565,
      "step": 19341
    },
    {
      "epoch": 0.9988122902143042,
      "grad_norm": 3.8793141841888428,
      "learning_rate": 3.551447324834367e-11,
      "loss": 0.0984,
      "step": 19342
    },
    {
      "epoch": 0.998863929770204,
      "grad_norm": 2.546114206314087,
      "learning_rate": 3.2493396565635995e-11,
      "loss": 0.0282,
      "step": 19343
    },
    {
      "epoch": 0.9989155693261038,
      "grad_norm": 2.650031566619873,
      "learning_rate": 2.960658938899741e-11,
      "loss": 0.0235,
      "step": 19344
    },
    {
      "epoch": 0.9989672088820036,
      "grad_norm": 2.270355224609375,
      "learning_rate": 2.6854051796698644e-11,
      "loss": 0.0303,
      "step": 19345
    },
    {
      "epoch": 0.9990188484379035,
      "grad_norm": 1.7122106552124023,
      "learning_rate": 2.4235783861459304e-11,
      "loss": 0.0304,
      "step": 19346
    },
    {
      "epoch": 0.9990704879938033,
      "grad_norm": 2.030507802963257,
      "learning_rate": 2.175178565488878e-11,
      "loss": 0.0182,
      "step": 19347
    },
    {
      "epoch": 0.9991221275497031,
      "grad_norm": 2.490400791168213,
      "learning_rate": 1.9402057242490223e-11,
      "loss": 0.0513,
      "step": 19348
    },
    {
      "epoch": 0.9991737671056029,
      "grad_norm": 1.5484899282455444,
      "learning_rate": 1.7186598688101464e-11,
      "loss": 0.0243,
      "step": 19349
    },
    {
      "epoch": 0.9992254066615027,
      "grad_norm": 4.199741840362549,
      "learning_rate": 1.510541005111943e-11,
      "loss": 0.04,
      "step": 19350
    },
    {
      "epoch": 0.9992770462174025,
      "grad_norm": 3.717794418334961,
      "learning_rate": 1.3158491387610383e-11,
      "loss": 0.0364,
      "step": 19351
    },
    {
      "epoch": 0.9993286857733024,
      "grad_norm": 4.437006950378418,
      "learning_rate": 1.1345842749199699e-11,
      "loss": 0.0652,
      "step": 19352
    },
    {
      "epoch": 0.9993803253292022,
      "grad_norm": 1.9170482158660889,
      "learning_rate": 9.667464185292296e-12,
      "loss": 0.0274,
      "step": 19353
    },
    {
      "epoch": 0.999431964885102,
      "grad_norm": 5.959677696228027,
      "learning_rate": 8.1233557402971e-12,
      "loss": 0.0418,
      "step": 19354
    },
    {
      "epoch": 0.9994836044410018,
      "grad_norm": 3.2795159816741943,
      "learning_rate": 6.713517456402585e-12,
      "loss": 0.0455,
      "step": 19355
    },
    {
      "epoch": 0.9995352439969016,
      "grad_norm": 3.632662057876587,
      "learning_rate": 5.437949370801221e-12,
      "loss": 0.0313,
      "step": 19356
    },
    {
      "epoch": 0.9995868835528015,
      "grad_norm": 3.7904953956604004,
      "learning_rate": 4.296651517909922e-12,
      "loss": 0.0632,
      "step": 19357
    },
    {
      "epoch": 0.9996385231087013,
      "grad_norm": 4.811854839324951,
      "learning_rate": 3.2896239288149333e-12,
      "loss": 0.056,
      "step": 19358
    },
    {
      "epoch": 0.9996901626646011,
      "grad_norm": 12.49565315246582,
      "learning_rate": 2.416866630161607e-12,
      "loss": 0.1051,
      "step": 19359
    },
    {
      "epoch": 0.9997418022205009,
      "grad_norm": 17.051841735839844,
      "learning_rate": 1.6783796458197387e-12,
      "loss": 0.1195,
      "step": 19360
    },
    {
      "epoch": 0.9997934417764007,
      "grad_norm": 0.8962852358818054,
      "learning_rate": 1.0741629946631194e-12,
      "loss": 0.0195,
      "step": 19361
    },
    {
      "epoch": 0.9998450813323005,
      "grad_norm": 4.103510856628418,
      "learning_rate": 6.042166939002059e-13,
      "loss": 0.0421,
      "step": 19362
    },
    {
      "epoch": 0.9998967208882004,
      "grad_norm": 6.032886505126953,
      "learning_rate": 2.685407557434516e-13,
      "loss": 0.0913,
      "step": 19363
    },
    {
      "epoch": 0.9999483604441002,
      "grad_norm": 2.5319411754608154,
      "learning_rate": 6.71351896297523e-14,
      "loss": 0.0313,
      "step": 19364
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5839954614639282,
      "learning_rate": 0.0,
      "loss": 0.0175,
      "step": 19365
    },
    {
      "epoch": 1.0,
      "step": 19365,
      "total_flos": 2.725260927518638e+18,
      "train_loss": 0.07305596203520642,
      "train_runtime": 19013.0111,
      "train_samples_per_second": 8.148,
      "train_steps_per_second": 1.019
    }
  ],
  "logging_steps": 1,
  "max_steps": 19365,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 19365,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 2.725260927518638e+18,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}