{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.33104200338831224,
  "eval_steps": 500,
  "global_step": 1700,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00019473059022841898,
      "grad_norm": 0.97265625,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 3.0625,
      "step": 1
    },
    {
      "epoch": 0.00038946118045683797,
      "grad_norm": 0.95703125,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 3.0592,
      "step": 2
    },
    {
      "epoch": 0.0005841917706852569,
      "grad_norm": 0.91796875,
      "learning_rate": 3e-06,
      "loss": 3.0624,
      "step": 3
    },
    {
      "epoch": 0.0007789223609136759,
      "grad_norm": 0.8515625,
      "learning_rate": 4.000000000000001e-06,
      "loss": 3.0779,
      "step": 4
    },
    {
      "epoch": 0.0009736529511420949,
      "grad_norm": 0.82421875,
      "learning_rate": 5e-06,
      "loss": 3.0704,
      "step": 5
    },
    {
      "epoch": 0.0011683835413705138,
      "grad_norm": 0.84375,
      "learning_rate": 6e-06,
      "loss": 3.0795,
      "step": 6
    },
    {
      "epoch": 0.0013631141315989328,
      "grad_norm": 0.94140625,
      "learning_rate": 7e-06,
      "loss": 3.0531,
      "step": 7
    },
    {
      "epoch": 0.0015578447218273519,
      "grad_norm": 0.91015625,
      "learning_rate": 8.000000000000001e-06,
      "loss": 3.0557,
      "step": 8
    },
    {
      "epoch": 0.001752575312055771,
      "grad_norm": 0.80859375,
      "learning_rate": 9e-06,
      "loss": 3.0463,
      "step": 9
    },
    {
      "epoch": 0.0019473059022841897,
      "grad_norm": 0.77734375,
      "learning_rate": 1e-05,
      "loss": 3.0676,
      "step": 10
    },
    {
      "epoch": 0.0021420364925126088,
      "grad_norm": 0.83203125,
      "learning_rate": 1.1000000000000001e-05,
      "loss": 3.0434,
      "step": 11
    },
    {
      "epoch": 0.0023367670827410276,
      "grad_norm": 1.0390625,
      "learning_rate": 1.2e-05,
      "loss": 3.0498,
      "step": 12
    },
    {
      "epoch": 0.002531497672969447,
      "grad_norm": 1.71875,
      "learning_rate": 1.3000000000000001e-05,
      "loss": 3.0523,
      "step": 13
    },
    {
      "epoch": 0.0027262282631978657,
      "grad_norm": 1.3671875,
      "learning_rate": 1.4e-05,
      "loss": 3.0558,
      "step": 14
    },
    {
      "epoch": 0.002920958853426285,
      "grad_norm": 0.90234375,
      "learning_rate": 1.5000000000000002e-05,
      "loss": 3.0365,
      "step": 15
    },
    {
      "epoch": 0.0031156894436547037,
      "grad_norm": 0.85546875,
      "learning_rate": 1.6000000000000003e-05,
      "loss": 3.0144,
      "step": 16
    },
    {
      "epoch": 0.0033104200338831226,
      "grad_norm": 0.95703125,
      "learning_rate": 1.7e-05,
      "loss": 3.0222,
      "step": 17
    },
    {
      "epoch": 0.003505150624111542,
      "grad_norm": 0.95703125,
      "learning_rate": 1.8e-05,
      "loss": 3.018,
      "step": 18
    },
    {
      "epoch": 0.0036998812143399606,
      "grad_norm": 1.2421875,
      "learning_rate": 1.9e-05,
      "loss": 3.0269,
      "step": 19
    },
    {
      "epoch": 0.0038946118045683794,
      "grad_norm": 2.015625,
      "learning_rate": 2e-05,
      "loss": 3.0331,
      "step": 20
    },
    {
      "epoch": 0.004089342394796799,
      "grad_norm": 1.1640625,
      "learning_rate": 2.1000000000000002e-05,
      "loss": 3.0341,
      "step": 21
    },
    {
      "epoch": 0.0042840729850252175,
      "grad_norm": 0.99609375,
      "learning_rate": 2.2000000000000003e-05,
      "loss": 2.9996,
      "step": 22
    },
    {
      "epoch": 0.004478803575253636,
      "grad_norm": 1.2734375,
      "learning_rate": 2.3e-05,
      "loss": 3.022,
      "step": 23
    },
    {
      "epoch": 0.004673534165482055,
      "grad_norm": 1.796875,
      "learning_rate": 2.4e-05,
      "loss": 2.9947,
      "step": 24
    },
    {
      "epoch": 0.004868264755710475,
      "grad_norm": 1.8203125,
      "learning_rate": 2.5e-05,
      "loss": 3.0017,
      "step": 25
    },
    {
      "epoch": 0.005062995345938894,
      "grad_norm": 1.1171875,
      "learning_rate": 2.6000000000000002e-05,
      "loss": 3.0014,
      "step": 26
    },
    {
      "epoch": 0.0052577259361673125,
      "grad_norm": 1.4921875,
      "learning_rate": 2.7000000000000002e-05,
      "loss": 3.0155,
      "step": 27
    },
    {
      "epoch": 0.005452456526395731,
      "grad_norm": 1.609375,
      "learning_rate": 2.8e-05,
      "loss": 3.0046,
      "step": 28
    },
    {
      "epoch": 0.00564718711662415,
      "grad_norm": 1.4765625,
      "learning_rate": 2.9e-05,
      "loss": 3.0053,
      "step": 29
    },
    {
      "epoch": 0.00584191770685257,
      "grad_norm": 1.7578125,
      "learning_rate": 3.0000000000000004e-05,
      "loss": 2.9963,
      "step": 30
    },
    {
      "epoch": 0.006036648297080989,
      "grad_norm": 2.09375,
      "learning_rate": 3.1e-05,
      "loss": 2.9719,
      "step": 31
    },
    {
      "epoch": 0.0062313788873094075,
      "grad_norm": 1.5,
      "learning_rate": 3.2000000000000005e-05,
      "loss": 2.9868,
      "step": 32
    },
    {
      "epoch": 0.006426109477537826,
      "grad_norm": 1.8984375,
      "learning_rate": 3.3e-05,
      "loss": 2.9856,
      "step": 33
    },
    {
      "epoch": 0.006620840067766245,
      "grad_norm": 1.5390625,
      "learning_rate": 3.4e-05,
      "loss": 3.0007,
      "step": 34
    },
    {
      "epoch": 0.006815570657994665,
      "grad_norm": 1.359375,
      "learning_rate": 3.5000000000000004e-05,
      "loss": 2.9758,
      "step": 35
    },
    {
      "epoch": 0.007010301248223084,
      "grad_norm": 2.359375,
      "learning_rate": 3.6e-05,
      "loss": 2.9728,
      "step": 36
    },
    {
      "epoch": 0.007205031838451502,
      "grad_norm": 2.0625,
      "learning_rate": 3.7000000000000005e-05,
      "loss": 2.9665,
      "step": 37
    },
    {
      "epoch": 0.007399762428679921,
      "grad_norm": 1.8984375,
      "learning_rate": 3.8e-05,
      "loss": 2.9688,
      "step": 38
    },
    {
      "epoch": 0.00759449301890834,
      "grad_norm": 1.21875,
      "learning_rate": 3.9e-05,
      "loss": 2.988,
      "step": 39
    },
    {
      "epoch": 0.007789223609136759,
      "grad_norm": 2.0,
      "learning_rate": 4e-05,
      "loss": 2.9744,
      "step": 40
    },
    {
      "epoch": 0.007983954199365179,
      "grad_norm": 1.8359375,
      "learning_rate": 3.9999996198006636e-05,
      "loss": 2.9476,
      "step": 41
    },
    {
      "epoch": 0.008178684789593597,
      "grad_norm": 2.75,
      "learning_rate": 3.9999984792027976e-05,
      "loss": 2.942,
      "step": 42
    },
    {
      "epoch": 0.008373415379822016,
      "grad_norm": 1.578125,
      "learning_rate": 3.999996578206837e-05,
      "loss": 2.9624,
      "step": 43
    },
    {
      "epoch": 0.008568145970050435,
      "grad_norm": 3.15625,
      "learning_rate": 3.999993916813503e-05,
      "loss": 2.966,
      "step": 44
    },
    {
      "epoch": 0.008762876560278854,
      "grad_norm": 1.515625,
      "learning_rate": 3.9999904950238085e-05,
      "loss": 2.9464,
      "step": 45
    },
    {
      "epoch": 0.008957607150507273,
      "grad_norm": 4.25,
      "learning_rate": 3.9999863128390545e-05,
      "loss": 2.9666,
      "step": 46
    },
    {
      "epoch": 0.009152337740735692,
      "grad_norm": 3.15625,
      "learning_rate": 3.9999813702608305e-05,
      "loss": 2.9594,
      "step": 47
    },
    {
      "epoch": 0.00934706833096411,
      "grad_norm": 3.703125,
      "learning_rate": 3.9999756672910155e-05,
      "loss": 2.9664,
      "step": 48
    },
    {
      "epoch": 0.009541798921192531,
      "grad_norm": 3.296875,
      "learning_rate": 3.999969203931779e-05,
      "loss": 2.9415,
      "step": 49
    },
    {
      "epoch": 0.00973652951142095,
      "grad_norm": 2.890625,
      "learning_rate": 3.9999619801855775e-05,
      "loss": 2.9606,
      "step": 50
    },
    {
      "epoch": 0.009931260101649369,
      "grad_norm": 2.8125,
      "learning_rate": 3.999953996055157e-05,
      "loss": 2.9471,
      "step": 51
    },
    {
      "epoch": 0.010125990691877787,
      "grad_norm": 2.328125,
      "learning_rate": 3.9999452515435545e-05,
      "loss": 2.9346,
      "step": 52
    },
    {
      "epoch": 0.010320721282106206,
      "grad_norm": 3.296875,
      "learning_rate": 3.9999357466540934e-05,
      "loss": 2.9223,
      "step": 53
    },
    {
      "epoch": 0.010515451872334625,
      "grad_norm": 2.265625,
      "learning_rate": 3.999925481390388e-05,
      "loss": 2.9388,
      "step": 54
    },
    {
      "epoch": 0.010710182462563044,
      "grad_norm": 3.265625,
      "learning_rate": 3.99991445575634e-05,
      "loss": 2.9453,
      "step": 55
    },
    {
      "epoch": 0.010904913052791463,
      "grad_norm": 2.265625,
      "learning_rate": 3.9999026697561426e-05,
      "loss": 2.9142,
      "step": 56
    },
    {
      "epoch": 0.011099643643019881,
      "grad_norm": 2.9375,
      "learning_rate": 3.999890123394277e-05,
      "loss": 2.9351,
      "step": 57
    },
    {
      "epoch": 0.0112943742332483,
      "grad_norm": 2.40625,
      "learning_rate": 3.999876816675513e-05,
      "loss": 2.9431,
      "step": 58
    },
    {
      "epoch": 0.01148910482347672,
      "grad_norm": 1.8046875,
      "learning_rate": 3.99986274960491e-05,
      "loss": 2.9416,
      "step": 59
    },
    {
      "epoch": 0.01168383541370514,
      "grad_norm": 4.28125,
      "learning_rate": 3.9998479221878155e-05,
      "loss": 2.9497,
      "step": 60
    },
    {
      "epoch": 0.011878566003933558,
      "grad_norm": 3.3125,
      "learning_rate": 3.999832334429867e-05,
      "loss": 2.9299,
      "step": 61
    },
    {
      "epoch": 0.012073296594161977,
      "grad_norm": 5.0625,
      "learning_rate": 3.9998159863369916e-05,
      "loss": 2.951,
      "step": 62
    },
    {
      "epoch": 0.012268027184390396,
      "grad_norm": 4.25,
      "learning_rate": 3.999798877915404e-05,
      "loss": 2.9284,
      "step": 63
    },
    {
      "epoch": 0.012462757774618815,
      "grad_norm": 3.984375,
      "learning_rate": 3.99978100917161e-05,
      "loss": 2.9504,
      "step": 64
    },
    {
      "epoch": 0.012657488364847234,
      "grad_norm": 3.921875,
      "learning_rate": 3.999762380112403e-05,
      "loss": 2.927,
      "step": 65
    },
    {
      "epoch": 0.012852218955075653,
      "grad_norm": 2.0625,
      "learning_rate": 3.999742990744864e-05,
      "loss": 2.9422,
      "step": 66
    },
    {
      "epoch": 0.013046949545304071,
      "grad_norm": 2.140625,
      "learning_rate": 3.9997228410763675e-05,
      "loss": 2.9264,
      "step": 67
    },
    {
      "epoch": 0.01324168013553249,
      "grad_norm": 3.15625,
      "learning_rate": 3.9997019311145724e-05,
      "loss": 2.9341,
      "step": 68
    },
    {
      "epoch": 0.013436410725760909,
      "grad_norm": 2.6875,
      "learning_rate": 3.999680260867429e-05,
      "loss": 2.9183,
      "step": 69
    },
    {
      "epoch": 0.01363114131598933,
      "grad_norm": 3.96875,
      "learning_rate": 3.9996578303431775e-05,
      "loss": 2.9261,
      "step": 70
    },
    {
      "epoch": 0.013825871906217748,
      "grad_norm": 3.484375,
      "learning_rate": 3.999634639550344e-05,
      "loss": 2.9119,
      "step": 71
    },
    {
      "epoch": 0.014020602496446167,
      "grad_norm": 2.765625,
      "learning_rate": 3.999610688497748e-05,
      "loss": 2.9234,
      "step": 72
    },
    {
      "epoch": 0.014215333086674586,
      "grad_norm": 2.546875,
      "learning_rate": 3.9995859771944936e-05,
      "loss": 2.9181,
      "step": 73
    },
    {
      "epoch": 0.014410063676903005,
      "grad_norm": 3.3125,
      "learning_rate": 3.9995605056499775e-05,
      "loss": 2.8889,
      "step": 74
    },
    {
      "epoch": 0.014604794267131424,
      "grad_norm": 3.0625,
      "learning_rate": 3.9995342738738824e-05,
      "loss": 2.9338,
      "step": 75
    },
    {
      "epoch": 0.014799524857359843,
      "grad_norm": 3.265625,
      "learning_rate": 3.9995072818761836e-05,
      "loss": 2.9161,
      "step": 76
    },
    {
      "epoch": 0.014994255447588261,
      "grad_norm": 3.328125,
      "learning_rate": 3.999479529667142e-05,
      "loss": 2.9041,
      "step": 77
    },
    {
      "epoch": 0.01518898603781668,
      "grad_norm": 1.84375,
      "learning_rate": 3.999451017257309e-05,
      "loss": 2.9224,
      "step": 78
    },
    {
      "epoch": 0.015383716628045099,
      "grad_norm": 1.6484375,
      "learning_rate": 3.999421744657525e-05,
      "loss": 2.9374,
      "step": 79
    },
    {
      "epoch": 0.015578447218273518,
      "grad_norm": 3.484375,
      "learning_rate": 3.9993917118789205e-05,
      "loss": 2.9133,
      "step": 80
    },
    {
      "epoch": 0.01577317780850194,
      "grad_norm": 2.5625,
      "learning_rate": 3.999360918932913e-05,
      "loss": 2.9206,
      "step": 81
    },
    {
      "epoch": 0.015967908398730357,
      "grad_norm": 4.125,
      "learning_rate": 3.99932936583121e-05,
      "loss": 2.9109,
      "step": 82
    },
    {
      "epoch": 0.016162638988958776,
      "grad_norm": 4.125,
      "learning_rate": 3.9992970525858084e-05,
      "loss": 2.8969,
      "step": 83
    },
    {
      "epoch": 0.016357369579187195,
      "grad_norm": 1.2578125,
      "learning_rate": 3.999263979208993e-05,
      "loss": 2.899,
      "step": 84
    },
    {
      "epoch": 0.016552100169415614,
      "grad_norm": 2.484375,
      "learning_rate": 3.9992301457133384e-05,
      "loss": 2.9096,
      "step": 85
    },
    {
      "epoch": 0.016746830759644032,
      "grad_norm": 1.796875,
      "learning_rate": 3.999195552111709e-05,
      "loss": 2.894,
      "step": 86
    },
    {
      "epoch": 0.01694156134987245,
      "grad_norm": 1.578125,
      "learning_rate": 3.999160198417256e-05,
      "loss": 2.8972,
      "step": 87
    },
    {
      "epoch": 0.01713629194010087,
      "grad_norm": 2.671875,
      "learning_rate": 3.999124084643422e-05,
      "loss": 2.9006,
      "step": 88
    },
    {
      "epoch": 0.01733102253032929,
      "grad_norm": 1.4453125,
      "learning_rate": 3.9990872108039364e-05,
      "loss": 2.8865,
      "step": 89
    },
    {
      "epoch": 0.017525753120557708,
      "grad_norm": 4.65625,
      "learning_rate": 3.999049576912819e-05,
      "loss": 2.929,
      "step": 90
    },
    {
      "epoch": 0.017720483710786127,
      "grad_norm": 4.09375,
      "learning_rate": 3.9990111829843795e-05,
      "loss": 2.9034,
      "step": 91
    },
    {
      "epoch": 0.017915214301014545,
      "grad_norm": 3.046875,
      "learning_rate": 3.9989720290332126e-05,
      "loss": 2.9342,
      "step": 92
    },
    {
      "epoch": 0.018109944891242964,
      "grad_norm": 3.0,
      "learning_rate": 3.998932115074207e-05,
      "loss": 2.9049,
      "step": 93
    },
    {
      "epoch": 0.018304675481471383,
      "grad_norm": 2.625,
      "learning_rate": 3.9988914411225366e-05,
      "loss": 2.8866,
      "step": 94
    },
    {
      "epoch": 0.018499406071699802,
      "grad_norm": 2.375,
      "learning_rate": 3.998850007193666e-05,
      "loss": 2.9089,
      "step": 95
    },
    {
      "epoch": 0.01869413666192822,
      "grad_norm": 2.921875,
      "learning_rate": 3.998807813303348e-05,
      "loss": 2.9015,
      "step": 96
    },
    {
      "epoch": 0.018888867252156643,
      "grad_norm": 2.40625,
      "learning_rate": 3.9987648594676246e-05,
      "loss": 2.8994,
      "step": 97
    },
    {
      "epoch": 0.019083597842385062,
      "grad_norm": 3.546875,
      "learning_rate": 3.998721145702829e-05,
      "loss": 2.9017,
      "step": 98
    },
    {
      "epoch": 0.01927832843261348,
      "grad_norm": 3.09375,
      "learning_rate": 3.9986766720255786e-05,
      "loss": 2.9012,
      "step": 99
    },
    {
      "epoch": 0.0194730590228419,
      "grad_norm": 2.65625,
      "learning_rate": 3.9986314384527823e-05,
      "loss": 2.8928,
      "step": 100
    },
    {
      "epoch": 0.019667789613070318,
      "grad_norm": 2.53125,
      "learning_rate": 3.998585445001639e-05,
      "loss": 2.8996,
      "step": 101
    },
    {
      "epoch": 0.019862520203298737,
      "grad_norm": 2.546875,
      "learning_rate": 3.9985386916896354e-05,
      "loss": 2.9069,
      "step": 102
    },
    {
      "epoch": 0.020057250793527156,
      "grad_norm": 2.203125,
      "learning_rate": 3.998491178534546e-05,
      "loss": 2.9112,
      "step": 103
    },
    {
      "epoch": 0.020251981383755575,
      "grad_norm": 3.09375,
      "learning_rate": 3.998442905554437e-05,
      "loss": 2.8826,
      "step": 104
    },
    {
      "epoch": 0.020446711973983994,
      "grad_norm": 2.78125,
      "learning_rate": 3.9983938727676605e-05,
      "loss": 2.9088,
      "step": 105
    },
    {
      "epoch": 0.020641442564212412,
      "grad_norm": 2.59375,
      "learning_rate": 3.998344080192858e-05,
      "loss": 2.878,
      "step": 106
    },
    {
      "epoch": 0.02083617315444083,
      "grad_norm": 2.46875,
      "learning_rate": 3.998293527848962e-05,
      "loss": 2.901,
      "step": 107
    },
    {
      "epoch": 0.02103090374466925,
      "grad_norm": 2.359375,
      "learning_rate": 3.998242215755193e-05,
      "loss": 2.8678,
      "step": 108
    },
    {
      "epoch": 0.02122563433489767,
      "grad_norm": 2.0625,
      "learning_rate": 3.9981901439310575e-05,
      "loss": 2.8947,
      "step": 109
    },
    {
      "epoch": 0.021420364925126088,
      "grad_norm": 3.0,
      "learning_rate": 3.998137312396355e-05,
      "loss": 2.8832,
      "step": 110
    },
    {
      "epoch": 0.021615095515354506,
      "grad_norm": 2.5,
      "learning_rate": 3.998083721171172e-05,
      "loss": 2.8863,
      "step": 111
    },
    {
      "epoch": 0.021809826105582925,
      "grad_norm": 3.0625,
      "learning_rate": 3.998029370275883e-05,
      "loss": 2.8958,
      "step": 112
    },
    {
      "epoch": 0.022004556695811344,
      "grad_norm": 2.859375,
      "learning_rate": 3.997974259731153e-05,
      "loss": 2.8797,
      "step": 113
    },
    {
      "epoch": 0.022199287286039763,
      "grad_norm": 2.359375,
      "learning_rate": 3.997918389557933e-05,
      "loss": 2.8945,
      "step": 114
    },
    {
      "epoch": 0.02239401787626818,
      "grad_norm": 2.125,
      "learning_rate": 3.997861759777468e-05,
      "loss": 2.8896,
      "step": 115
    },
    {
      "epoch": 0.0225887484664966,
      "grad_norm": 2.875,
      "learning_rate": 3.997804370411286e-05,
      "loss": 2.8889,
      "step": 116
    },
    {
      "epoch": 0.02278347905672502,
      "grad_norm": 2.578125,
      "learning_rate": 3.997746221481208e-05,
      "loss": 2.9048,
      "step": 117
    },
    {
      "epoch": 0.02297820964695344,
      "grad_norm": 2.734375,
      "learning_rate": 3.997687313009341e-05,
      "loss": 2.8841,
      "step": 118
    },
    {
      "epoch": 0.02317294023718186,
      "grad_norm": 2.515625,
      "learning_rate": 3.9976276450180833e-05,
      "loss": 2.8869,
      "step": 119
    },
    {
      "epoch": 0.02336767082741028,
      "grad_norm": 2.640625,
      "learning_rate": 3.9975672175301194e-05,
      "loss": 2.8733,
      "step": 120
    },
    {
      "epoch": 0.023562401417638698,
      "grad_norm": 2.390625,
      "learning_rate": 3.997506030568424e-05,
      "loss": 2.891,
      "step": 121
    },
    {
      "epoch": 0.023757132007867117,
      "grad_norm": 2.859375,
      "learning_rate": 3.997444084156261e-05,
      "loss": 2.8863,
      "step": 122
    },
    {
      "epoch": 0.023951862598095536,
      "grad_norm": 2.671875,
      "learning_rate": 3.9973813783171815e-05,
      "loss": 2.8925,
      "step": 123
    },
    {
      "epoch": 0.024146593188323955,
      "grad_norm": 2.625,
      "learning_rate": 3.997317913075027e-05,
      "loss": 2.8833,
      "step": 124
    },
    {
      "epoch": 0.024341323778552373,
      "grad_norm": 2.359375,
      "learning_rate": 3.997253688453927e-05,
      "loss": 2.8847,
      "step": 125
    },
    {
      "epoch": 0.024536054368780792,
      "grad_norm": 2.578125,
      "learning_rate": 3.9971887044782985e-05,
      "loss": 2.888,
      "step": 126
    },
    {
      "epoch": 0.02473078495900921,
      "grad_norm": 2.125,
      "learning_rate": 3.997122961172849e-05,
      "loss": 2.8671,
      "step": 127
    },
    {
      "epoch": 0.02492551554923763,
      "grad_norm": 3.015625,
      "learning_rate": 3.9970564585625754e-05,
      "loss": 2.8693,
      "step": 128
    },
    {
      "epoch": 0.02512024613946605,
      "grad_norm": 2.65625,
      "learning_rate": 3.9969891966727593e-05,
      "loss": 2.8651,
      "step": 129
    },
    {
      "epoch": 0.025314976729694467,
      "grad_norm": 2.640625,
      "learning_rate": 3.996921175528976e-05,
      "loss": 2.8721,
      "step": 130
    },
    {
      "epoch": 0.025509707319922886,
      "grad_norm": 2.40625,
      "learning_rate": 3.996852395157086e-05,
      "loss": 2.8592,
      "step": 131
    },
    {
      "epoch": 0.025704437910151305,
      "grad_norm": 2.609375,
      "learning_rate": 3.9967828555832395e-05,
      "loss": 2.8701,
      "step": 132
    },
    {
      "epoch": 0.025899168500379724,
      "grad_norm": 2.34375,
      "learning_rate": 3.996712556833876e-05,
      "loss": 2.8901,
      "step": 133
    },
    {
      "epoch": 0.026093899090608143,
      "grad_norm": 2.5625,
      "learning_rate": 3.9966414989357226e-05,
      "loss": 2.8688,
      "step": 134
    },
    {
      "epoch": 0.02628862968083656,
      "grad_norm": 2.25,
      "learning_rate": 3.996569681915795e-05,
      "loss": 2.8651,
      "step": 135
    },
    {
      "epoch": 0.02648336027106498,
      "grad_norm": 2.46875,
      "learning_rate": 3.9964971058013986e-05,
      "loss": 2.872,
      "step": 136
    },
    {
      "epoch": 0.0266780908612934,
      "grad_norm": 2.125,
      "learning_rate": 3.996423770620127e-05,
      "loss": 2.8867,
      "step": 137
    },
    {
      "epoch": 0.026872821451521818,
      "grad_norm": 2.671875,
      "learning_rate": 3.996349676399862e-05,
      "loss": 2.9028,
      "step": 138
    },
    {
      "epoch": 0.027067552041750237,
      "grad_norm": 2.265625,
      "learning_rate": 3.996274823168774e-05,
      "loss": 2.8654,
      "step": 139
    },
    {
      "epoch": 0.02726228263197866,
      "grad_norm": 2.734375,
      "learning_rate": 3.996199210955322e-05,
      "loss": 2.8561,
      "step": 140
    },
    {
      "epoch": 0.027457013222207078,
      "grad_norm": 2.453125,
      "learning_rate": 3.996122839788254e-05,
      "loss": 2.8661,
      "step": 141
    },
    {
      "epoch": 0.027651743812435497,
      "grad_norm": 2.40625,
      "learning_rate": 3.996045709696606e-05,
      "loss": 2.8655,
      "step": 142
    },
    {
      "epoch": 0.027846474402663916,
      "grad_norm": 2.203125,
      "learning_rate": 3.995967820709704e-05,
      "loss": 2.8586,
      "step": 143
    },
    {
      "epoch": 0.028041204992892334,
      "grad_norm": 2.421875,
      "learning_rate": 3.9958891728571585e-05,
      "loss": 2.8472,
      "step": 144
    },
    {
      "epoch": 0.028235935583120753,
      "grad_norm": 2.15625,
      "learning_rate": 3.9958097661688746e-05,
      "loss": 2.8924,
      "step": 145
    },
    {
      "epoch": 0.028430666173349172,
      "grad_norm": 2.390625,
      "learning_rate": 3.9957296006750406e-05,
      "loss": 2.8624,
      "step": 146
    },
    {
      "epoch": 0.02862539676357759,
      "grad_norm": 2.09375,
      "learning_rate": 3.9956486764061365e-05,
      "loss": 2.8657,
      "step": 147
    },
    {
      "epoch": 0.02882012735380601,
      "grad_norm": 2.484375,
      "learning_rate": 3.9955669933929294e-05,
      "loss": 2.8625,
      "step": 148
    },
    {
      "epoch": 0.02901485794403443,
      "grad_norm": 2.171875,
      "learning_rate": 3.9954845516664736e-05,
      "loss": 2.8548,
      "step": 149
    },
    {
      "epoch": 0.029209588534262847,
      "grad_norm": 2.5625,
      "learning_rate": 3.995401351258117e-05,
      "loss": 2.8624,
      "step": 150
    },
    {
      "epoch": 0.029404319124491266,
      "grad_norm": 2.453125,
      "learning_rate": 3.995317392199488e-05,
      "loss": 2.8623,
      "step": 151
    },
    {
      "epoch": 0.029599049714719685,
      "grad_norm": 2.453125,
      "learning_rate": 3.99523267452251e-05,
      "loss": 2.8664,
      "step": 152
    },
    {
      "epoch": 0.029793780304948104,
      "grad_norm": 2.328125,
      "learning_rate": 3.9951471982593936e-05,
      "loss": 2.8531,
      "step": 153
    },
    {
      "epoch": 0.029988510895176523,
      "grad_norm": 2.21875,
      "learning_rate": 3.995060963442635e-05,
      "loss": 2.8745,
      "step": 154
    },
    {
      "epoch": 0.03018324148540494,
      "grad_norm": 2.09375,
      "learning_rate": 3.994973970105021e-05,
      "loss": 2.8563,
      "step": 155
    },
    {
      "epoch": 0.03037797207563336,
      "grad_norm": 1.953125,
      "learning_rate": 3.994886218279627e-05,
      "loss": 2.8517,
      "step": 156
    },
    {
      "epoch": 0.03057270266586178,
      "grad_norm": 2.15625,
      "learning_rate": 3.994797707999816e-05,
      "loss": 2.8422,
      "step": 157
    },
    {
      "epoch": 0.030767433256090198,
      "grad_norm": 1.765625,
      "learning_rate": 3.9947084392992386e-05,
      "loss": 2.8434,
      "step": 158
    },
    {
      "epoch": 0.030962163846318617,
      "grad_norm": 3.125,
      "learning_rate": 3.994618412211836e-05,
      "loss": 2.8664,
      "step": 159
    },
    {
      "epoch": 0.031156894436547036,
      "grad_norm": 2.71875,
      "learning_rate": 3.994527626771836e-05,
      "loss": 2.859,
      "step": 160
    },
    {
      "epoch": 0.03135162502677546,
      "grad_norm": 2.484375,
      "learning_rate": 3.9944360830137554e-05,
      "loss": 2.8459,
      "step": 161
    },
    {
      "epoch": 0.03154635561700388,
      "grad_norm": 2.421875,
      "learning_rate": 3.994343780972399e-05,
      "loss": 2.8534,
      "step": 162
    },
    {
      "epoch": 0.031741086207232296,
      "grad_norm": 2.046875,
      "learning_rate": 3.994250720682859e-05,
      "loss": 2.8518,
      "step": 163
    },
    {
      "epoch": 0.031935816797460714,
      "grad_norm": 1.59375,
      "learning_rate": 3.994156902180518e-05,
      "loss": 2.8317,
      "step": 164
    },
    {
      "epoch": 0.03213054738768913,
      "grad_norm": 2.171875,
      "learning_rate": 3.9940623255010454e-05,
      "loss": 2.8486,
      "step": 165
    },
    {
      "epoch": 0.03232527797791755,
      "grad_norm": 1.34375,
      "learning_rate": 3.993966990680399e-05,
      "loss": 2.8575,
      "step": 166
    },
    {
      "epoch": 0.03252000856814597,
      "grad_norm": 2.984375,
      "learning_rate": 3.9938708977548256e-05,
      "loss": 2.8517,
      "step": 167
    },
    {
      "epoch": 0.03271473915837439,
      "grad_norm": 2.421875,
      "learning_rate": 3.993774046760859e-05,
      "loss": 2.8867,
      "step": 168
    },
    {
      "epoch": 0.03290946974860281,
      "grad_norm": 2.984375,
      "learning_rate": 3.993676437735322e-05,
      "loss": 2.8349,
      "step": 169
    },
    {
      "epoch": 0.03310420033883123,
      "grad_norm": 2.875,
      "learning_rate": 3.993578070715326e-05,
      "loss": 2.8626,
      "step": 170
    },
    {
      "epoch": 0.033298930929059646,
      "grad_norm": 1.8359375,
      "learning_rate": 3.993478945738269e-05,
      "loss": 2.8534,
      "step": 171
    },
    {
      "epoch": 0.033493661519288065,
      "grad_norm": 2.046875,
      "learning_rate": 3.99337906284184e-05,
      "loss": 2.8419,
      "step": 172
    },
    {
      "epoch": 0.033688392109516484,
      "grad_norm": 1.703125,
      "learning_rate": 3.993278422064012e-05,
      "loss": 2.8403,
      "step": 173
    },
    {
      "epoch": 0.0338831226997449,
      "grad_norm": 1.8359375,
      "learning_rate": 3.9931770234430504e-05,
      "loss": 2.8521,
      "step": 174
    },
    {
      "epoch": 0.03407785328997332,
      "grad_norm": 1.3515625,
      "learning_rate": 3.993074867017507e-05,
      "loss": 2.848,
      "step": 175
    },
    {
      "epoch": 0.03427258388020174,
      "grad_norm": 2.546875,
      "learning_rate": 3.9929719528262205e-05,
      "loss": 2.8589,
      "step": 176
    },
    {
      "epoch": 0.03446731447043016,
      "grad_norm": 2.046875,
      "learning_rate": 3.9928682809083195e-05,
      "loss": 2.8713,
      "step": 177
    },
    {
      "epoch": 0.03466204506065858,
      "grad_norm": 2.375,
      "learning_rate": 3.99276385130322e-05,
      "loss": 2.8502,
      "step": 178
    },
    {
      "epoch": 0.034856775650887,
      "grad_norm": 2.390625,
      "learning_rate": 3.992658664050626e-05,
      "loss": 2.8415,
      "step": 179
    },
    {
      "epoch": 0.035051506241115415,
      "grad_norm": 1.7109375,
      "learning_rate": 3.9925527191905285e-05,
      "loss": 2.8511,
      "step": 180
    },
    {
      "epoch": 0.035246236831343834,
      "grad_norm": 2.09375,
      "learning_rate": 3.9924460167632094e-05,
      "loss": 2.8261,
      "step": 181
    },
    {
      "epoch": 0.03544096742157225,
      "grad_norm": 1.515625,
      "learning_rate": 3.9923385568092364e-05,
      "loss": 2.8649,
      "step": 182
    },
    {
      "epoch": 0.03563569801180067,
      "grad_norm": 2.75,
      "learning_rate": 3.992230339369465e-05,
      "loss": 2.8341,
      "step": 183
    },
    {
      "epoch": 0.03583042860202909,
      "grad_norm": 2.328125,
      "learning_rate": 3.99212136448504e-05,
      "loss": 2.8325,
      "step": 184
    },
    {
      "epoch": 0.03602515919225751,
      "grad_norm": 2.609375,
      "learning_rate": 3.9920116321973935e-05,
      "loss": 2.8139,
      "step": 185
    },
    {
      "epoch": 0.03621988978248593,
      "grad_norm": 2.03125,
      "learning_rate": 3.9919011425482456e-05,
      "loss": 2.8263,
      "step": 186
    },
    {
      "epoch": 0.03641462037271435,
      "grad_norm": 2.453125,
      "learning_rate": 3.9917898955796044e-05,
      "loss": 2.8474,
      "step": 187
    },
    {
      "epoch": 0.036609350962942766,
      "grad_norm": 1.5,
      "learning_rate": 3.991677891333765e-05,
      "loss": 2.8471,
      "step": 188
    },
    {
      "epoch": 0.036804081553171185,
      "grad_norm": 3.203125,
      "learning_rate": 3.991565129853314e-05,
      "loss": 2.848,
      "step": 189
    },
    {
      "epoch": 0.036998812143399604,
      "grad_norm": 2.671875,
      "learning_rate": 3.9914516111811204e-05,
      "loss": 2.8344,
      "step": 190
    },
    {
      "epoch": 0.03719354273362802,
      "grad_norm": 3.140625,
      "learning_rate": 3.991337335360345e-05,
      "loss": 2.8606,
      "step": 191
    },
    {
      "epoch": 0.03738827332385644,
      "grad_norm": 3.09375,
      "learning_rate": 3.9912223024344354e-05,
      "loss": 2.8294,
      "step": 192
    },
    {
      "epoch": 0.03758300391408486,
      "grad_norm": 2.046875,
      "learning_rate": 3.991106512447127e-05,
      "loss": 2.8365,
      "step": 193
    },
    {
      "epoch": 0.037777734504313286,
      "grad_norm": 2.25,
      "learning_rate": 3.990989965442444e-05,
      "loss": 2.8336,
      "step": 194
    },
    {
      "epoch": 0.037972465094541705,
      "grad_norm": 1.6953125,
      "learning_rate": 3.9908726614646955e-05,
      "loss": 2.8225,
      "step": 195
    },
    {
      "epoch": 0.038167195684770124,
      "grad_norm": 2.546875,
      "learning_rate": 3.990754600558482e-05,
      "loss": 2.845,
      "step": 196
    },
    {
      "epoch": 0.03836192627499854,
      "grad_norm": 1.9609375,
      "learning_rate": 3.9906357827686895e-05,
      "loss": 2.8078,
      "step": 197
    },
    {
      "epoch": 0.03855665686522696,
      "grad_norm": 2.953125,
      "learning_rate": 3.990516208140493e-05,
      "loss": 2.8399,
      "step": 198
    },
    {
      "epoch": 0.03875138745545538,
      "grad_norm": 2.765625,
      "learning_rate": 3.990395876719354e-05,
      "loss": 2.8332,
      "step": 199
    },
    {
      "epoch": 0.0389461180456838,
      "grad_norm": 2.421875,
      "learning_rate": 3.990274788551023e-05,
      "loss": 2.8359,
      "step": 200
    },
    {
      "epoch": 0.03914084863591222,
      "grad_norm": 2.390625,
      "learning_rate": 3.990152943681537e-05,
      "loss": 2.8253,
      "step": 201
    },
    {
      "epoch": 0.039335579226140636,
      "grad_norm": 2.640625,
      "learning_rate": 3.990030342157222e-05,
      "loss": 2.8394,
      "step": 202
    },
    {
      "epoch": 0.039530309816369055,
      "grad_norm": 1.8671875,
      "learning_rate": 3.9899069840246906e-05,
      "loss": 2.8302,
      "step": 203
    },
    {
      "epoch": 0.039725040406597474,
      "grad_norm": 2.546875,
      "learning_rate": 3.9897828693308434e-05,
      "loss": 2.8443,
      "step": 204
    },
    {
      "epoch": 0.03991977099682589,
      "grad_norm": 1.71875,
      "learning_rate": 3.989657998122869e-05,
      "loss": 2.8171,
      "step": 205
    },
    {
      "epoch": 0.04011450158705431,
      "grad_norm": 2.96875,
      "learning_rate": 3.989532370448244e-05,
      "loss": 2.8084,
      "step": 206
    },
    {
      "epoch": 0.04030923217728273,
      "grad_norm": 2.40625,
      "learning_rate": 3.98940598635473e-05,
      "loss": 2.8457,
      "step": 207
    },
    {
      "epoch": 0.04050396276751115,
      "grad_norm": 2.609375,
      "learning_rate": 3.989278845890381e-05,
      "loss": 2.8221,
      "step": 208
    },
    {
      "epoch": 0.04069869335773957,
      "grad_norm": 2.40625,
      "learning_rate": 3.9891509491035334e-05,
      "loss": 2.8287,
      "step": 209
    },
    {
      "epoch": 0.04089342394796799,
      "grad_norm": 2.28125,
      "learning_rate": 3.989022296042814e-05,
      "loss": 2.7986,
      "step": 210
    },
    {
      "epoch": 0.041088154538196406,
      "grad_norm": 2.03125,
      "learning_rate": 3.988892886757137e-05,
      "loss": 2.8082,
      "step": 211
    },
    {
      "epoch": 0.041282885128424825,
      "grad_norm": 2.71875,
      "learning_rate": 3.988762721295703e-05,
      "loss": 2.8256,
      "step": 212
    },
    {
      "epoch": 0.041477615718653243,
      "grad_norm": 2.265625,
      "learning_rate": 3.988631799708002e-05,
      "loss": 2.8246,
      "step": 213
    },
    {
      "epoch": 0.04167234630888166,
      "grad_norm": 2.9375,
      "learning_rate": 3.988500122043809e-05,
      "loss": 2.8291,
      "step": 214
    },
    {
      "epoch": 0.04186707689911008,
      "grad_norm": 2.8125,
      "learning_rate": 3.9883676883531895e-05,
      "loss": 2.8105,
      "step": 215
    },
    {
      "epoch": 0.0420618074893385,
      "grad_norm": 1.9375,
      "learning_rate": 3.988234498686493e-05,
      "loss": 2.7985,
      "step": 216
    },
    {
      "epoch": 0.04225653807956692,
      "grad_norm": 1.9296875,
      "learning_rate": 3.988100553094358e-05,
      "loss": 2.8031,
      "step": 217
    },
    {
      "epoch": 0.04245126866979534,
      "grad_norm": 2.203125,
      "learning_rate": 3.987965851627713e-05,
      "loss": 2.8259,
      "step": 218
    },
    {
      "epoch": 0.042645999260023756,
      "grad_norm": 1.8046875,
      "learning_rate": 3.987830394337768e-05,
      "loss": 2.7989,
      "step": 219
    },
    {
      "epoch": 0.042840729850252175,
      "grad_norm": 2.9375,
      "learning_rate": 3.987694181276027e-05,
      "loss": 2.811,
      "step": 220
    },
    {
      "epoch": 0.043035460440480594,
      "grad_norm": 2.796875,
      "learning_rate": 3.987557212494275e-05,
      "loss": 2.8235,
      "step": 221
    },
    {
      "epoch": 0.04323019103070901,
      "grad_norm": 1.609375,
      "learning_rate": 3.98741948804459e-05,
      "loss": 2.8029,
      "step": 222
    },
    {
      "epoch": 0.04342492162093743,
      "grad_norm": 1.796875,
      "learning_rate": 3.987281007979333e-05,
      "loss": 2.8179,
      "step": 223
    },
    {
      "epoch": 0.04361965221116585,
      "grad_norm": 1.8671875,
      "learning_rate": 3.9871417723511555e-05,
      "loss": 2.8174,
      "step": 224
    },
    {
      "epoch": 0.04381438280139427,
      "grad_norm": 1.2578125,
      "learning_rate": 3.987001781212994e-05,
      "loss": 2.8341,
      "step": 225
    },
    {
      "epoch": 0.04400911339162269,
      "grad_norm": 3.1875,
      "learning_rate": 3.9868610346180726e-05,
      "loss": 2.8261,
      "step": 226
    },
    {
      "epoch": 0.04420384398185111,
      "grad_norm": 2.6875,
      "learning_rate": 3.986719532619904e-05,
      "loss": 2.814,
      "step": 227
    },
    {
      "epoch": 0.044398574572079526,
      "grad_norm": 2.40625,
      "learning_rate": 3.9865772752722866e-05,
      "loss": 2.8147,
      "step": 228
    },
    {
      "epoch": 0.044593305162307945,
      "grad_norm": 2.375,
      "learning_rate": 3.986434262629307e-05,
      "loss": 2.8104,
      "step": 229
    },
    {
      "epoch": 0.04478803575253636,
      "grad_norm": 1.96875,
      "learning_rate": 3.986290494745338e-05,
      "loss": 2.8118,
      "step": 230
    },
    {
      "epoch": 0.04498276634276478,
      "grad_norm": 1.8046875,
      "learning_rate": 3.98614597167504e-05,
      "loss": 2.8133,
      "step": 231
    },
    {
      "epoch": 0.0451774969329932,
      "grad_norm": 2.546875,
      "learning_rate": 3.9860006934733605e-05,
      "loss": 2.8165,
      "step": 232
    },
    {
      "epoch": 0.04537222752322162,
      "grad_norm": 2.265625,
      "learning_rate": 3.985854660195535e-05,
      "loss": 2.8021,
      "step": 233
    },
    {
      "epoch": 0.04556695811345004,
      "grad_norm": 2.53125,
      "learning_rate": 3.985707871897085e-05,
      "loss": 2.8255,
      "step": 234
    },
    {
      "epoch": 0.04576168870367846,
      "grad_norm": 2.5625,
      "learning_rate": 3.985560328633819e-05,
      "loss": 2.8181,
      "step": 235
    },
    {
      "epoch": 0.04595641929390688,
      "grad_norm": 1.8671875,
      "learning_rate": 3.9854120304618327e-05,
      "loss": 2.8204,
      "step": 236
    },
    {
      "epoch": 0.0461511498841353,
      "grad_norm": 2.0625,
      "learning_rate": 3.985262977437509e-05,
      "loss": 2.7901,
      "step": 237
    },
    {
      "epoch": 0.04634588047436372,
      "grad_norm": 1.859375,
      "learning_rate": 3.985113169617518e-05,
      "loss": 2.7858,
      "step": 238
    },
    {
      "epoch": 0.04654061106459214,
      "grad_norm": 1.6015625,
      "learning_rate": 3.9849626070588165e-05,
      "loss": 2.8177,
      "step": 239
    },
    {
      "epoch": 0.04673534165482056,
      "grad_norm": 2.8125,
      "learning_rate": 3.984811289818649e-05,
      "loss": 2.813,
      "step": 240
    },
    {
      "epoch": 0.04693007224504898,
      "grad_norm": 2.46875,
      "learning_rate": 3.984659217954545e-05,
      "loss": 2.7943,
      "step": 241
    },
    {
      "epoch": 0.047124802835277396,
      "grad_norm": 2.234375,
      "learning_rate": 3.9845063915243224e-05,
      "loss": 2.8103,
      "step": 242
    },
    {
      "epoch": 0.047319533425505815,
      "grad_norm": 2.234375,
      "learning_rate": 3.984352810586086e-05,
      "loss": 2.8177,
      "step": 243
    },
    {
      "epoch": 0.047514264015734234,
      "grad_norm": 2.0625,
      "learning_rate": 3.984198475198227e-05,
      "loss": 2.7931,
      "step": 244
    },
    {
      "epoch": 0.04770899460596265,
      "grad_norm": 1.7578125,
      "learning_rate": 3.9840433854194234e-05,
      "loss": 2.8071,
      "step": 245
    },
    {
      "epoch": 0.04790372519619107,
      "grad_norm": 2.328125,
      "learning_rate": 3.983887541308641e-05,
      "loss": 2.8016,
      "step": 246
    },
    {
      "epoch": 0.04809845578641949,
      "grad_norm": 1.96875,
      "learning_rate": 3.983730942925131e-05,
      "loss": 2.81,
      "step": 247
    },
    {
      "epoch": 0.04829318637664791,
      "grad_norm": 2.453125,
      "learning_rate": 3.983573590328432e-05,
      "loss": 2.807,
      "step": 248
    },
    {
      "epoch": 0.04848791696687633,
      "grad_norm": 2.125,
      "learning_rate": 3.98341548357837e-05,
      "loss": 2.8085,
      "step": 249
    },
    {
      "epoch": 0.04868264755710475,
      "grad_norm": 2.25,
      "learning_rate": 3.9832566227350564e-05,
      "loss": 2.788,
      "step": 250
    },
    {
      "epoch": 0.048877378147333166,
      "grad_norm": 2.015625,
      "learning_rate": 3.98309700785889e-05,
      "loss": 2.786,
      "step": 251
    },
    {
      "epoch": 0.049072108737561584,
      "grad_norm": 2.390625,
      "learning_rate": 3.9829366390105566e-05,
      "loss": 2.808,
      "step": 252
    },
    {
      "epoch": 0.04926683932779,
      "grad_norm": 2.0,
      "learning_rate": 3.982775516251028e-05,
      "loss": 2.807,
      "step": 253
    },
    {
      "epoch": 0.04946156991801842,
      "grad_norm": 2.171875,
      "learning_rate": 3.982613639641564e-05,
      "loss": 2.7798,
      "step": 254
    },
    {
      "epoch": 0.04965630050824684,
      "grad_norm": 1.921875,
      "learning_rate": 3.982451009243708e-05,
      "loss": 2.8126,
      "step": 255
    },
    {
      "epoch": 0.04985103109847526,
      "grad_norm": 2.25,
      "learning_rate": 3.982287625119293e-05,
      "loss": 2.8028,
      "step": 256
    },
    {
      "epoch": 0.05004576168870368,
      "grad_norm": 1.8125,
      "learning_rate": 3.982123487330438e-05,
      "loss": 2.8078,
      "step": 257
    },
    {
      "epoch": 0.0502404922789321,
      "grad_norm": 1.9296875,
      "learning_rate": 3.9819585959395486e-05,
      "loss": 2.7725,
      "step": 258
    },
    {
      "epoch": 0.050435222869160516,
      "grad_norm": 1.578125,
      "learning_rate": 3.9817929510093144e-05,
      "loss": 2.8028,
      "step": 259
    },
    {
      "epoch": 0.050629953459388935,
      "grad_norm": 2.140625,
      "learning_rate": 3.981626552602715e-05,
      "loss": 2.7998,
      "step": 260
    },
    {
      "epoch": 0.050824684049617354,
      "grad_norm": 1.6953125,
      "learning_rate": 3.981459400783014e-05,
      "loss": 2.7972,
      "step": 261
    },
    {
      "epoch": 0.05101941463984577,
      "grad_norm": 2.796875,
      "learning_rate": 3.981291495613764e-05,
      "loss": 2.809,
      "step": 262
    },
    {
      "epoch": 0.05121414523007419,
      "grad_norm": 2.71875,
      "learning_rate": 3.981122837158801e-05,
      "loss": 2.7838,
      "step": 263
    },
    {
      "epoch": 0.05140887582030261,
      "grad_norm": 1.4765625,
      "learning_rate": 3.980953425482248e-05,
      "loss": 2.7956,
      "step": 264
    },
    {
      "epoch": 0.05160360641053103,
      "grad_norm": 1.8125,
      "learning_rate": 3.980783260648518e-05,
      "loss": 2.7875,
      "step": 265
    },
    {
      "epoch": 0.05179833700075945,
      "grad_norm": 1.6484375,
      "learning_rate": 3.980612342722306e-05,
      "loss": 2.775,
      "step": 266
    },
    {
      "epoch": 0.05199306759098787,
      "grad_norm": 1.421875,
      "learning_rate": 3.980440671768594e-05,
      "loss": 2.7914,
      "step": 267
    },
    {
      "epoch": 0.052187798181216286,
      "grad_norm": 2.546875,
      "learning_rate": 3.980268247852653e-05,
      "loss": 2.7803,
      "step": 268
    },
    {
      "epoch": 0.052382528771444704,
      "grad_norm": 2.015625,
      "learning_rate": 3.980095071040037e-05,
      "loss": 2.7797,
      "step": 269
    },
    {
      "epoch": 0.05257725936167312,
      "grad_norm": 2.796875,
      "learning_rate": 3.979921141396588e-05,
      "loss": 2.7891,
      "step": 270
    },
    {
      "epoch": 0.05277198995190154,
      "grad_norm": 2.609375,
      "learning_rate": 3.9797464589884344e-05,
      "loss": 2.7975,
      "step": 271
    },
    {
      "epoch": 0.05296672054212996,
      "grad_norm": 1.6796875,
      "learning_rate": 3.97957102388199e-05,
      "loss": 2.7746,
      "step": 272
    },
    {
      "epoch": 0.05316145113235838,
      "grad_norm": 1.6484375,
      "learning_rate": 3.979394836143955e-05,
      "loss": 2.7739,
      "step": 273
    },
    {
      "epoch": 0.0533561817225868,
      "grad_norm": 1.4609375,
      "learning_rate": 3.979217895841317e-05,
      "loss": 2.79,
      "step": 274
    },
    {
      "epoch": 0.05355091231281522,
      "grad_norm": 1.3515625,
      "learning_rate": 3.979040203041347e-05,
      "loss": 2.7767,
      "step": 275
    },
    {
      "epoch": 0.053745642903043636,
      "grad_norm": 1.7578125,
      "learning_rate": 3.978861757811604e-05,
      "loss": 2.7691,
      "step": 276
    },
    {
      "epoch": 0.053940373493272055,
      "grad_norm": 1.25,
      "learning_rate": 3.978682560219934e-05,
      "loss": 2.7946,
      "step": 277
    },
    {
      "epoch": 0.054135104083500474,
      "grad_norm": 2.171875,
      "learning_rate": 3.9785026103344664e-05,
      "loss": 2.7638,
      "step": 278
    },
    {
      "epoch": 0.0543298346737289,
      "grad_norm": 1.546875,
      "learning_rate": 3.978321908223619e-05,
      "loss": 2.7949,
      "step": 279
    },
    {
      "epoch": 0.05452456526395732,
      "grad_norm": 3.4375,
      "learning_rate": 3.978140453956093e-05,
      "loss": 2.8131,
      "step": 280
    },
    {
      "epoch": 0.05471929585418574,
      "grad_norm": 3.328125,
      "learning_rate": 3.977958247600879e-05,
      "loss": 2.7778,
      "step": 281
    },
    {
      "epoch": 0.054914026444414156,
      "grad_norm": 1.4609375,
      "learning_rate": 3.9777752892272515e-05,
      "loss": 2.7709,
      "step": 282
    },
    {
      "epoch": 0.055108757034642575,
      "grad_norm": 2.8125,
      "learning_rate": 3.97759157890477e-05,
      "loss": 2.7963,
      "step": 283
    },
    {
      "epoch": 0.055303487624870994,
      "grad_norm": 2.234375,
      "learning_rate": 3.977407116703283e-05,
      "loss": 2.7975,
      "step": 284
    },
    {
      "epoch": 0.05549821821509941,
      "grad_norm": 2.3125,
      "learning_rate": 3.977221902692921e-05,
      "loss": 2.7718,
      "step": 285
    },
    {
      "epoch": 0.05569294880532783,
      "grad_norm": 2.390625,
      "learning_rate": 3.977035936944102e-05,
      "loss": 2.7864,
      "step": 286
    },
    {
      "epoch": 0.05588767939555625,
      "grad_norm": 1.84375,
      "learning_rate": 3.976849219527532e-05,
      "loss": 2.7984,
      "step": 287
    },
    {
      "epoch": 0.05608240998578467,
      "grad_norm": 2.09375,
      "learning_rate": 3.9766617505141996e-05,
      "loss": 2.7621,
      "step": 288
    },
    {
      "epoch": 0.05627714057601309,
      "grad_norm": 1.609375,
      "learning_rate": 3.9764735299753804e-05,
      "loss": 2.78,
      "step": 289
    },
    {
      "epoch": 0.05647187116624151,
      "grad_norm": 2.296875,
      "learning_rate": 3.976284557982636e-05,
      "loss": 2.7936,
      "step": 290
    },
    {
      "epoch": 0.056666601756469925,
      "grad_norm": 1.796875,
      "learning_rate": 3.976094834607814e-05,
      "loss": 2.7811,
      "step": 291
    },
    {
      "epoch": 0.056861332346698344,
      "grad_norm": 2.703125,
      "learning_rate": 3.9759043599230464e-05,
      "loss": 2.7803,
      "step": 292
    },
    {
      "epoch": 0.05705606293692676,
      "grad_norm": 2.140625,
      "learning_rate": 3.975713134000751e-05,
      "loss": 2.7814,
      "step": 293
    },
    {
      "epoch": 0.05725079352715518,
      "grad_norm": 2.453125,
      "learning_rate": 3.975521156913633e-05,
      "loss": 2.7846,
      "step": 294
    },
    {
      "epoch": 0.0574455241173836,
      "grad_norm": 1.8046875,
      "learning_rate": 3.975328428734681e-05,
      "loss": 2.748,
      "step": 295
    },
    {
      "epoch": 0.05764025470761202,
      "grad_norm": 2.59375,
      "learning_rate": 3.97513494953717e-05,
      "loss": 2.7715,
      "step": 296
    },
    {
      "epoch": 0.05783498529784044,
      "grad_norm": 1.8984375,
      "learning_rate": 3.974940719394662e-05,
      "loss": 2.7667,
      "step": 297
    },
    {
      "epoch": 0.05802971588806886,
      "grad_norm": 3.078125,
      "learning_rate": 3.9747457383810016e-05,
      "loss": 2.7667,
      "step": 298
    },
    {
      "epoch": 0.058224446478297276,
      "grad_norm": 2.78125,
      "learning_rate": 3.9745500065703215e-05,
      "loss": 2.7784,
      "step": 299
    },
    {
      "epoch": 0.058419177068525695,
      "grad_norm": 2.28125,
      "learning_rate": 3.9743535240370385e-05,
      "loss": 2.7495,
      "step": 300
    },
    {
      "epoch": 0.058613907658754114,
      "grad_norm": 2.25,
      "learning_rate": 3.974156290855855e-05,
      "loss": 2.7728,
      "step": 301
    },
    {
      "epoch": 0.05880863824898253,
      "grad_norm": 2.453125,
      "learning_rate": 3.973958307101759e-05,
      "loss": 2.7698,
      "step": 302
    },
    {
      "epoch": 0.05900336883921095,
      "grad_norm": 2.15625,
      "learning_rate": 3.973759572850024e-05,
      "loss": 2.7699,
      "step": 303
    },
    {
      "epoch": 0.05919809942943937,
      "grad_norm": 2.53125,
      "learning_rate": 3.973560088176209e-05,
      "loss": 2.7987,
      "step": 304
    },
    {
      "epoch": 0.05939283001966779,
      "grad_norm": 2.3125,
      "learning_rate": 3.973359853156156e-05,
      "loss": 2.7683,
      "step": 305
    },
    {
      "epoch": 0.05958756060989621,
      "grad_norm": 2.15625,
      "learning_rate": 3.9731588678659966e-05,
      "loss": 2.746,
      "step": 306
    },
    {
      "epoch": 0.059782291200124626,
      "grad_norm": 1.9921875,
      "learning_rate": 3.972957132382144e-05,
      "loss": 2.7646,
      "step": 307
    },
    {
      "epoch": 0.059977021790353045,
      "grad_norm": 2.546875,
      "learning_rate": 3.9727546467812984e-05,
      "loss": 2.7698,
      "step": 308
    },
    {
      "epoch": 0.060171752380581464,
      "grad_norm": 2.328125,
      "learning_rate": 3.972551411140444e-05,
      "loss": 2.7469,
      "step": 309
    },
    {
      "epoch": 0.06036648297080988,
      "grad_norm": 2.453125,
      "learning_rate": 3.9723474255368516e-05,
      "loss": 2.7641,
      "step": 310
    },
    {
      "epoch": 0.0605612135610383,
      "grad_norm": 2.234375,
      "learning_rate": 3.972142690048077e-05,
      "loss": 2.7547,
      "step": 311
    },
    {
      "epoch": 0.06075594415126672,
      "grad_norm": 2.171875,
      "learning_rate": 3.971937204751958e-05,
      "loss": 2.7617,
      "step": 312
    },
    {
      "epoch": 0.06095067474149514,
      "grad_norm": 2.046875,
      "learning_rate": 3.9717309697266224e-05,
      "loss": 2.7612,
      "step": 313
    },
    {
      "epoch": 0.06114540533172356,
      "grad_norm": 2.359375,
      "learning_rate": 3.9715239850504795e-05,
      "loss": 2.7504,
      "step": 314
    },
    {
      "epoch": 0.06134013592195198,
      "grad_norm": 2.15625,
      "learning_rate": 3.9713162508022254e-05,
      "loss": 2.7592,
      "step": 315
    },
    {
      "epoch": 0.061534866512180396,
      "grad_norm": 2.15625,
      "learning_rate": 3.97110776706084e-05,
      "loss": 2.763,
      "step": 316
    },
    {
      "epoch": 0.061729597102408815,
      "grad_norm": 2.0625,
      "learning_rate": 3.970898533905589e-05,
      "loss": 2.7518,
      "step": 317
    },
    {
      "epoch": 0.061924327692637234,
      "grad_norm": 2.203125,
      "learning_rate": 3.9706885514160224e-05,
      "loss": 2.745,
      "step": 318
    },
    {
      "epoch": 0.06211905828286565,
      "grad_norm": 1.9609375,
      "learning_rate": 3.9704778196719756e-05,
      "loss": 2.7815,
      "step": 319
    },
    {
      "epoch": 0.06231378887309407,
      "grad_norm": 2.3125,
      "learning_rate": 3.970266338753569e-05,
      "loss": 2.7587,
      "step": 320
    },
    {
      "epoch": 0.06250851946332249,
      "grad_norm": 1.9921875,
      "learning_rate": 3.970054108741206e-05,
      "loss": 2.7711,
      "step": 321
    },
    {
      "epoch": 0.06270325005355092,
      "grad_norm": 2.203125,
      "learning_rate": 3.969841129715579e-05,
      "loss": 2.7447,
      "step": 322
    },
    {
      "epoch": 0.06289798064377933,
      "grad_norm": 2.015625,
      "learning_rate": 3.9696274017576595e-05,
      "loss": 2.7521,
      "step": 323
    },
    {
      "epoch": 0.06309271123400775,
      "grad_norm": 2.28125,
      "learning_rate": 3.969412924948709e-05,
      "loss": 2.7688,
      "step": 324
    },
    {
      "epoch": 0.06328744182423617,
      "grad_norm": 2.25,
      "learning_rate": 3.9691976993702706e-05,
      "loss": 2.7484,
      "step": 325
    },
    {
      "epoch": 0.06348217241446459,
      "grad_norm": 1.984375,
      "learning_rate": 3.968981725104172e-05,
      "loss": 2.7565,
      "step": 326
    },
    {
      "epoch": 0.063676903004693,
      "grad_norm": 1.875,
      "learning_rate": 3.968765002232529e-05,
      "loss": 2.7532,
      "step": 327
    },
    {
      "epoch": 0.06387163359492143,
      "grad_norm": 2.234375,
      "learning_rate": 3.9685475308377364e-05,
      "loss": 2.7582,
      "step": 328
    },
    {
      "epoch": 0.06406636418514984,
      "grad_norm": 1.984375,
      "learning_rate": 3.968329311002479e-05,
      "loss": 2.7563,
      "step": 329
    },
    {
      "epoch": 0.06426109477537827,
      "grad_norm": 2.265625,
      "learning_rate": 3.9681103428097225e-05,
      "loss": 2.7577,
      "step": 330
    },
    {
      "epoch": 0.06445582536560668,
      "grad_norm": 2.234375,
      "learning_rate": 3.967890626342719e-05,
      "loss": 2.7584,
      "step": 331
    },
    {
      "epoch": 0.0646505559558351,
      "grad_norm": 2.03125,
      "learning_rate": 3.9676701616850045e-05,
      "loss": 2.754,
      "step": 332
    },
    {
      "epoch": 0.06484528654606352,
      "grad_norm": 1.90625,
      "learning_rate": 3.9674489489204e-05,
      "loss": 2.7401,
      "step": 333
    },
    {
      "epoch": 0.06504001713629194,
      "grad_norm": 2.078125,
      "learning_rate": 3.9672269881330095e-05,
      "loss": 2.76,
      "step": 334
    },
    {
      "epoch": 0.06523474772652035,
      "grad_norm": 1.859375,
      "learning_rate": 3.967004279407223e-05,
      "loss": 2.7586,
      "step": 335
    },
    {
      "epoch": 0.06542947831674878,
      "grad_norm": 2.265625,
      "learning_rate": 3.966780822827714e-05,
      "loss": 2.7424,
      "step": 336
    },
    {
      "epoch": 0.06562420890697719,
      "grad_norm": 2.109375,
      "learning_rate": 3.966556618479441e-05,
      "loss": 2.7741,
      "step": 337
    },
    {
      "epoch": 0.06581893949720562,
      "grad_norm": 2.109375,
      "learning_rate": 3.966331666447645e-05,
      "loss": 2.7579,
      "step": 338
    },
    {
      "epoch": 0.06601367008743403,
      "grad_norm": 1.9609375,
      "learning_rate": 3.9661059668178546e-05,
      "loss": 2.7644,
      "step": 339
    },
    {
      "epoch": 0.06620840067766245,
      "grad_norm": 2.0625,
      "learning_rate": 3.965879519675879e-05,
      "loss": 2.7143,
      "step": 340
    },
    {
      "epoch": 0.06640313126789087,
      "grad_norm": 1.828125,
      "learning_rate": 3.9656523251078135e-05,
      "loss": 2.7482,
      "step": 341
    },
    {
      "epoch": 0.06659786185811929,
      "grad_norm": 2.125,
      "learning_rate": 3.9654243832000384e-05,
      "loss": 2.7462,
      "step": 342
    },
    {
      "epoch": 0.0667925924483477,
      "grad_norm": 1.9453125,
      "learning_rate": 3.965195694039216e-05,
      "loss": 2.7609,
      "step": 343
    },
    {
      "epoch": 0.06698732303857613,
      "grad_norm": 2.171875,
      "learning_rate": 3.964966257712294e-05,
      "loss": 2.7451,
      "step": 344
    },
    {
      "epoch": 0.06718205362880456,
      "grad_norm": 1.96875,
      "learning_rate": 3.964736074306504e-05,
      "loss": 2.7562,
      "step": 345
    },
    {
      "epoch": 0.06737678421903297,
      "grad_norm": 2.21875,
      "learning_rate": 3.9645051439093616e-05,
      "loss": 2.739,
      "step": 346
    },
    {
      "epoch": 0.0675715148092614,
      "grad_norm": 1.984375,
      "learning_rate": 3.9642734666086664e-05,
      "loss": 2.7597,
      "step": 347
    },
    {
      "epoch": 0.0677662453994898,
      "grad_norm": 2.109375,
      "learning_rate": 3.964041042492502e-05,
      "loss": 2.7549,
      "step": 348
    },
    {
      "epoch": 0.06796097598971823,
      "grad_norm": 1.8671875,
      "learning_rate": 3.963807871649236e-05,
      "loss": 2.7629,
      "step": 349
    },
    {
      "epoch": 0.06815570657994664,
      "grad_norm": 2.21875,
      "learning_rate": 3.9635739541675195e-05,
      "loss": 2.7625,
      "step": 350
    },
    {
      "epoch": 0.06835043717017507,
      "grad_norm": 2.0,
      "learning_rate": 3.9633392901362884e-05,
      "loss": 2.761,
      "step": 351
    },
    {
      "epoch": 0.06854516776040348,
      "grad_norm": 2.34375,
      "learning_rate": 3.96310387964476e-05,
      "loss": 2.7469,
      "step": 352
    },
    {
      "epoch": 0.0687398983506319,
      "grad_norm": 2.1875,
      "learning_rate": 3.96286772278244e-05,
      "loss": 2.7396,
      "step": 353
    },
    {
      "epoch": 0.06893462894086032,
      "grad_norm": 1.9609375,
      "learning_rate": 3.962630819639112e-05,
      "loss": 2.735,
      "step": 354
    },
    {
      "epoch": 0.06912935953108874,
      "grad_norm": 1.7890625,
      "learning_rate": 3.9623931703048497e-05,
      "loss": 2.761,
      "step": 355
    },
    {
      "epoch": 0.06932409012131716,
      "grad_norm": 2.34375,
      "learning_rate": 3.962154774870005e-05,
      "loss": 2.7588,
      "step": 356
    },
    {
      "epoch": 0.06951882071154558,
      "grad_norm": 2.015625,
      "learning_rate": 3.961915633425216e-05,
      "loss": 2.7265,
      "step": 357
    },
    {
      "epoch": 0.069713551301774,
      "grad_norm": 2.46875,
      "learning_rate": 3.961675746061405e-05,
      "loss": 2.7473,
      "step": 358
    },
    {
      "epoch": 0.06990828189200242,
      "grad_norm": 2.21875,
      "learning_rate": 3.961435112869775e-05,
      "loss": 2.7514,
      "step": 359
    },
    {
      "epoch": 0.07010301248223083,
      "grad_norm": 2.265625,
      "learning_rate": 3.9611937339418164e-05,
      "loss": 2.7634,
      "step": 360
    },
    {
      "epoch": 0.07029774307245926,
      "grad_norm": 2.015625,
      "learning_rate": 3.9609516093693016e-05,
      "loss": 2.7656,
      "step": 361
    },
    {
      "epoch": 0.07049247366268767,
      "grad_norm": 2.328125,
      "learning_rate": 3.960708739244285e-05,
      "loss": 2.766,
      "step": 362
    },
    {
      "epoch": 0.0706872042529161,
      "grad_norm": 2.015625,
      "learning_rate": 3.9604651236591064e-05,
      "loss": 2.7408,
      "step": 363
    },
    {
      "epoch": 0.0708819348431445,
      "grad_norm": 2.5625,
      "learning_rate": 3.9602207627063875e-05,
      "loss": 2.7465,
      "step": 364
    },
    {
      "epoch": 0.07107666543337293,
      "grad_norm": 2.25,
      "learning_rate": 3.959975656479036e-05,
      "loss": 2.7491,
      "step": 365
    },
    {
      "epoch": 0.07127139602360134,
      "grad_norm": 2.4375,
      "learning_rate": 3.9597298050702385e-05,
      "loss": 2.7458,
      "step": 366
    },
    {
      "epoch": 0.07146612661382977,
      "grad_norm": 2.234375,
      "learning_rate": 3.959483208573469e-05,
      "loss": 2.7252,
      "step": 367
    },
    {
      "epoch": 0.07166085720405818,
      "grad_norm": 2.421875,
      "learning_rate": 3.9592358670824835e-05,
      "loss": 2.7255,
      "step": 368
    },
    {
      "epoch": 0.07185558779428661,
      "grad_norm": 2.203125,
      "learning_rate": 3.958987780691321e-05,
      "loss": 2.7454,
      "step": 369
    },
    {
      "epoch": 0.07205031838451502,
      "grad_norm": 2.578125,
      "learning_rate": 3.958738949494303e-05,
      "loss": 2.7425,
      "step": 370
    },
    {
      "epoch": 0.07224504897474344,
      "grad_norm": 2.421875,
      "learning_rate": 3.9584893735860356e-05,
      "loss": 2.7341,
      "step": 371
    },
    {
      "epoch": 0.07243977956497186,
      "grad_norm": 2.203125,
      "learning_rate": 3.958239053061407e-05,
      "loss": 2.7321,
      "step": 372
    },
    {
      "epoch": 0.07263451015520028,
      "grad_norm": 2.015625,
      "learning_rate": 3.95798798801559e-05,
      "loss": 2.7465,
      "step": 373
    },
    {
      "epoch": 0.0728292407454287,
      "grad_norm": 2.3125,
      "learning_rate": 3.9577361785440375e-05,
      "loss": 2.75,
      "step": 374
    },
    {
      "epoch": 0.07302397133565712,
      "grad_norm": 2.03125,
      "learning_rate": 3.957483624742489e-05,
      "loss": 2.7682,
      "step": 375
    },
    {
      "epoch": 0.07321870192588553,
      "grad_norm": 2.5,
      "learning_rate": 3.9572303267069646e-05,
      "loss": 2.7491,
      "step": 376
    },
    {
      "epoch": 0.07341343251611396,
      "grad_norm": 2.453125,
      "learning_rate": 3.956976284533767e-05,
      "loss": 2.7305,
      "step": 377
    },
    {
      "epoch": 0.07360816310634237,
      "grad_norm": 1.890625,
      "learning_rate": 3.956721498319486e-05,
      "loss": 2.7461,
      "step": 378
    },
    {
      "epoch": 0.0738028936965708,
      "grad_norm": 1.890625,
      "learning_rate": 3.956465968160988e-05,
      "loss": 2.7662,
      "step": 379
    },
    {
      "epoch": 0.07399762428679921,
      "grad_norm": 1.921875,
      "learning_rate": 3.956209694155426e-05,
      "loss": 2.7107,
      "step": 380
    },
    {
      "epoch": 0.07419235487702763,
      "grad_norm": 1.7578125,
      "learning_rate": 3.9559526764002364e-05,
      "loss": 2.7418,
      "step": 381
    },
    {
      "epoch": 0.07438708546725604,
      "grad_norm": 2.4375,
      "learning_rate": 3.955694914993136e-05,
      "loss": 2.7114,
      "step": 382
    },
    {
      "epoch": 0.07458181605748447,
      "grad_norm": 2.171875,
      "learning_rate": 3.955436410032126e-05,
      "loss": 2.7266,
      "step": 383
    },
    {
      "epoch": 0.07477654664771288,
      "grad_norm": 2.0,
      "learning_rate": 3.9551771616154904e-05,
      "loss": 2.7181,
      "step": 384
    },
    {
      "epoch": 0.07497127723794131,
      "grad_norm": 2.015625,
      "learning_rate": 3.954917169841794e-05,
      "loss": 2.7378,
      "step": 385
    },
    {
      "epoch": 0.07516600782816972,
      "grad_norm": 1.8671875,
      "learning_rate": 3.954656434809886e-05,
      "loss": 2.7383,
      "step": 386
    },
    {
      "epoch": 0.07536073841839815,
      "grad_norm": 1.6328125,
      "learning_rate": 3.954394956618899e-05,
      "loss": 2.7201,
      "step": 387
    },
    {
      "epoch": 0.07555546900862657,
      "grad_norm": 2.328125,
      "learning_rate": 3.954132735368244e-05,
      "loss": 2.7214,
      "step": 388
    },
    {
      "epoch": 0.07575019959885498,
      "grad_norm": 2.15625,
      "learning_rate": 3.9538697711576204e-05,
      "loss": 2.7281,
      "step": 389
    },
    {
      "epoch": 0.07594493018908341,
      "grad_norm": 2.109375,
      "learning_rate": 3.9536060640870054e-05,
      "loss": 2.7301,
      "step": 390
    },
    {
      "epoch": 0.07613966077931182,
      "grad_norm": 2.078125,
      "learning_rate": 3.95334161425666e-05,
      "loss": 2.7327,
      "step": 391
    },
    {
      "epoch": 0.07633439136954025,
      "grad_norm": 1.9453125,
      "learning_rate": 3.953076421767129e-05,
      "loss": 2.7518,
      "step": 392
    },
    {
      "epoch": 0.07652912195976866,
      "grad_norm": 1.7421875,
      "learning_rate": 3.9528104867192375e-05,
      "loss": 2.7392,
      "step": 393
    },
    {
      "epoch": 0.07672385254999708,
      "grad_norm": 2.09375,
      "learning_rate": 3.952543809214094e-05,
      "loss": 2.738,
      "step": 394
    },
    {
      "epoch": 0.0769185831402255,
      "grad_norm": 1.7578125,
      "learning_rate": 3.952276389353089e-05,
      "loss": 2.7102,
      "step": 395
    },
    {
      "epoch": 0.07711331373045392,
      "grad_norm": 2.3125,
      "learning_rate": 3.952008227237896e-05,
      "loss": 2.7204,
      "step": 396
    },
    {
      "epoch": 0.07730804432068233,
      "grad_norm": 2.21875,
      "learning_rate": 3.951739322970469e-05,
      "loss": 2.7353,
      "step": 397
    },
    {
      "epoch": 0.07750277491091076,
      "grad_norm": 1.8046875,
      "learning_rate": 3.9514696766530474e-05,
      "loss": 2.721,
      "step": 398
    },
    {
      "epoch": 0.07769750550113917,
      "grad_norm": 1.5546875,
      "learning_rate": 3.951199288388147e-05,
      "loss": 2.7347,
      "step": 399
    },
    {
      "epoch": 0.0778922360913676,
      "grad_norm": 2.28125,
      "learning_rate": 3.9509281582785725e-05,
      "loss": 2.7236,
      "step": 400
    },
    {
      "epoch": 0.07808696668159601,
      "grad_norm": 2.046875,
      "learning_rate": 3.950656286427406e-05,
      "loss": 2.7279,
      "step": 401
    },
    {
      "epoch": 0.07828169727182444,
      "grad_norm": 2.109375,
      "learning_rate": 3.950383672938013e-05,
      "loss": 2.7401,
      "step": 402
    },
    {
      "epoch": 0.07847642786205285,
      "grad_norm": 1.96875,
      "learning_rate": 3.950110317914041e-05,
      "loss": 2.7379,
      "step": 403
    },
    {
      "epoch": 0.07867115845228127,
      "grad_norm": 2.015625,
      "learning_rate": 3.949836221459419e-05,
      "loss": 2.7411,
      "step": 404
    },
    {
      "epoch": 0.07886588904250968,
      "grad_norm": 1.90625,
      "learning_rate": 3.9495613836783595e-05,
      "loss": 2.7416,
      "step": 405
    },
    {
      "epoch": 0.07906061963273811,
      "grad_norm": 2.046875,
      "learning_rate": 3.949285804675354e-05,
      "loss": 2.7151,
      "step": 406
    },
    {
      "epoch": 0.07925535022296652,
      "grad_norm": 1.96875,
      "learning_rate": 3.94900948455518e-05,
      "loss": 2.732,
      "step": 407
    },
    {
      "epoch": 0.07945008081319495,
      "grad_norm": 2.0,
      "learning_rate": 3.948732423422891e-05,
      "loss": 2.7233,
      "step": 408
    },
    {
      "epoch": 0.07964481140342336,
      "grad_norm": 1.765625,
      "learning_rate": 3.9484546213838276e-05,
      "loss": 2.7343,
      "step": 409
    },
    {
      "epoch": 0.07983954199365179,
      "grad_norm": 2.203125,
      "learning_rate": 3.94817607854361e-05,
      "loss": 2.7015,
      "step": 410
    },
    {
      "epoch": 0.0800342725838802,
      "grad_norm": 1.984375,
      "learning_rate": 3.9478967950081386e-05,
      "loss": 2.7251,
      "step": 411
    },
    {
      "epoch": 0.08022900317410862,
      "grad_norm": 1.984375,
      "learning_rate": 3.947616770883598e-05,
      "loss": 2.7163,
      "step": 412
    },
    {
      "epoch": 0.08042373376433704,
      "grad_norm": 2.03125,
      "learning_rate": 3.947336006276452e-05,
      "loss": 2.7153,
      "step": 413
    },
    {
      "epoch": 0.08061846435456546,
      "grad_norm": 1.8125,
      "learning_rate": 3.9470545012934495e-05,
      "loss": 2.7189,
      "step": 414
    },
    {
      "epoch": 0.08081319494479387,
      "grad_norm": 1.53125,
      "learning_rate": 3.9467722560416156e-05,
      "loss": 2.7194,
      "step": 415
    },
    {
      "epoch": 0.0810079255350223,
      "grad_norm": 2.3125,
      "learning_rate": 3.946489270628262e-05,
      "loss": 2.7202,
      "step": 416
    },
    {
      "epoch": 0.08120265612525071,
      "grad_norm": 1.953125,
      "learning_rate": 3.946205545160978e-05,
      "loss": 2.7067,
      "step": 417
    },
    {
      "epoch": 0.08139738671547914,
      "grad_norm": 2.265625,
      "learning_rate": 3.945921079747637e-05,
      "loss": 2.7156,
      "step": 418
    },
    {
      "epoch": 0.08159211730570755,
      "grad_norm": 2.28125,
      "learning_rate": 3.9456358744963914e-05,
      "loss": 2.7257,
      "step": 419
    },
    {
      "epoch": 0.08178684789593597,
      "grad_norm": 1.5078125,
      "learning_rate": 3.945349929515678e-05,
      "loss": 2.7084,
      "step": 420
    },
    {
      "epoch": 0.08198157848616439,
      "grad_norm": 1.3984375,
      "learning_rate": 3.9450632449142106e-05,
      "loss": 2.7142,
      "step": 421
    },
    {
      "epoch": 0.08217630907639281,
      "grad_norm": 2.0625,
      "learning_rate": 3.9447758208009876e-05,
      "loss": 2.7337,
      "step": 422
    },
    {
      "epoch": 0.08237103966662122,
      "grad_norm": 1.65625,
      "learning_rate": 3.944487657285287e-05,
      "loss": 2.7216,
      "step": 423
    },
    {
      "epoch": 0.08256577025684965,
      "grad_norm": 2.78125,
      "learning_rate": 3.944198754476669e-05,
      "loss": 2.7214,
      "step": 424
    },
    {
      "epoch": 0.08276050084707806,
      "grad_norm": 2.703125,
      "learning_rate": 3.9439091124849745e-05,
      "loss": 2.72,
      "step": 425
    },
    {
      "epoch": 0.08295523143730649,
      "grad_norm": 1.265625,
      "learning_rate": 3.943618731420324e-05,
      "loss": 2.714,
      "step": 426
    },
    {
      "epoch": 0.0831499620275349,
      "grad_norm": 1.484375,
      "learning_rate": 3.943327611393121e-05,
      "loss": 2.7319,
      "step": 427
    },
    {
      "epoch": 0.08334469261776332,
      "grad_norm": 1.734375,
      "learning_rate": 3.9430357525140495e-05,
      "loss": 2.7207,
      "step": 428
    },
    {
      "epoch": 0.08353942320799174,
      "grad_norm": 1.1484375,
      "learning_rate": 3.9427431548940734e-05,
      "loss": 2.7368,
      "step": 429
    },
    {
      "epoch": 0.08373415379822016,
      "grad_norm": 2.96875,
      "learning_rate": 3.9424498186444376e-05,
      "loss": 2.7301,
      "step": 430
    },
    {
      "epoch": 0.08392888438844859,
      "grad_norm": 2.578125,
      "learning_rate": 3.9421557438766697e-05,
      "loss": 2.7317,
      "step": 431
    },
    {
      "epoch": 0.084123614978677,
      "grad_norm": 2.046875,
      "learning_rate": 3.941860930702576e-05,
      "loss": 2.722,
      "step": 432
    },
    {
      "epoch": 0.08431834556890543,
      "grad_norm": 2.265625,
      "learning_rate": 3.941565379234244e-05,
      "loss": 2.7366,
      "step": 433
    },
    {
      "epoch": 0.08451307615913384,
      "grad_norm": 1.2421875,
      "learning_rate": 3.941269089584043e-05,
      "loss": 2.7072,
      "step": 434
    },
    {
      "epoch": 0.08470780674936226,
      "grad_norm": 1.5,
      "learning_rate": 3.9409720618646205e-05,
      "loss": 2.7394,
      "step": 435
    },
    {
      "epoch": 0.08490253733959068,
      "grad_norm": 1.6328125,
      "learning_rate": 3.940674296188909e-05,
      "loss": 2.7141,
      "step": 436
    },
    {
      "epoch": 0.0850972679298191,
      "grad_norm": 1.09375,
      "learning_rate": 3.940375792670116e-05,
      "loss": 2.7222,
      "step": 437
    },
    {
      "epoch": 0.08529199852004751,
      "grad_norm": 2.984375,
      "learning_rate": 3.940076551421734e-05,
      "loss": 2.7358,
      "step": 438
    },
    {
      "epoch": 0.08548672911027594,
      "grad_norm": 2.578125,
      "learning_rate": 3.9397765725575336e-05,
      "loss": 2.6998,
      "step": 439
    },
    {
      "epoch": 0.08568145970050435,
      "grad_norm": 2.140625,
      "learning_rate": 3.939475856191567e-05,
      "loss": 2.7203,
      "step": 440
    },
    {
      "epoch": 0.08587619029073278,
      "grad_norm": 2.15625,
      "learning_rate": 3.9391744024381665e-05,
      "loss": 2.7161,
      "step": 441
    },
    {
      "epoch": 0.08607092088096119,
      "grad_norm": 1.578125,
      "learning_rate": 3.938872211411944e-05,
      "loss": 2.7172,
      "step": 442
    },
    {
      "epoch": 0.08626565147118961,
      "grad_norm": 1.453125,
      "learning_rate": 3.938569283227793e-05,
      "loss": 2.7161,
      "step": 443
    },
    {
      "epoch": 0.08646038206141803,
      "grad_norm": 2.125,
      "learning_rate": 3.938265618000885e-05,
      "loss": 2.7215,
      "step": 444
    },
    {
      "epoch": 0.08665511265164645,
      "grad_norm": 1.7421875,
      "learning_rate": 3.937961215846676e-05,
      "loss": 2.704,
      "step": 445
    },
    {
      "epoch": 0.08684984324187486,
      "grad_norm": 2.34375,
      "learning_rate": 3.937656076880897e-05,
      "loss": 2.7228,
      "step": 446
    },
    {
      "epoch": 0.08704457383210329,
      "grad_norm": 2.140625,
      "learning_rate": 3.937350201219563e-05,
      "loss": 2.7281,
      "step": 447
    },
    {
      "epoch": 0.0872393044223317,
      "grad_norm": 1.7421875,
      "learning_rate": 3.9370435889789674e-05,
      "loss": 2.7229,
      "step": 448
    },
    {
      "epoch": 0.08743403501256013,
      "grad_norm": 1.6796875,
      "learning_rate": 3.9367362402756836e-05,
      "loss": 2.7229,
      "step": 449
    },
    {
      "epoch": 0.08762876560278854,
      "grad_norm": 1.8359375,
      "learning_rate": 3.9364281552265664e-05,
      "loss": 2.7102,
      "step": 450
    },
    {
      "epoch": 0.08782349619301696,
      "grad_norm": 1.6015625,
      "learning_rate": 3.9361193339487484e-05,
      "loss": 2.7025,
      "step": 451
    },
    {
      "epoch": 0.08801822678324538,
      "grad_norm": 2.125,
      "learning_rate": 3.9358097765596445e-05,
      "loss": 2.7212,
      "step": 452
    },
    {
      "epoch": 0.0882129573734738,
      "grad_norm": 1.796875,
      "learning_rate": 3.935499483176947e-05,
      "loss": 2.7282,
      "step": 453
    },
    {
      "epoch": 0.08840768796370221,
      "grad_norm": 2.078125,
      "learning_rate": 3.935188453918629e-05,
      "loss": 2.7163,
      "step": 454
    },
    {
      "epoch": 0.08860241855393064,
      "grad_norm": 1.8984375,
      "learning_rate": 3.934876688902945e-05,
      "loss": 2.7122,
      "step": 455
    },
    {
      "epoch": 0.08879714914415905,
      "grad_norm": 1.84375,
      "learning_rate": 3.934564188248427e-05,
      "loss": 2.7304,
      "step": 456
    },
    {
      "epoch": 0.08899187973438748,
      "grad_norm": 1.671875,
      "learning_rate": 3.934250952073887e-05,
      "loss": 2.7093,
      "step": 457
    },
    {
      "epoch": 0.08918661032461589,
      "grad_norm": 2.0,
      "learning_rate": 3.933936980498419e-05,
      "loss": 2.6912,
      "step": 458
    },
    {
      "epoch": 0.08938134091484431,
      "grad_norm": 1.6875,
      "learning_rate": 3.933622273641393e-05,
      "loss": 2.6993,
      "step": 459
    },
    {
      "epoch": 0.08957607150507273,
      "grad_norm": 2.09375,
      "learning_rate": 3.933306831622461e-05,
      "loss": 2.7167,
      "step": 460
    },
    {
      "epoch": 0.08977080209530115,
      "grad_norm": 1.8671875,
      "learning_rate": 3.932990654561554e-05,
      "loss": 2.7129,
      "step": 461
    },
    {
      "epoch": 0.08996553268552956,
      "grad_norm": 1.8046875,
      "learning_rate": 3.9326737425788824e-05,
      "loss": 2.6996,
      "step": 462
    },
    {
      "epoch": 0.09016026327575799,
      "grad_norm": 1.6953125,
      "learning_rate": 3.932356095794935e-05,
      "loss": 2.7033,
      "step": 463
    },
    {
      "epoch": 0.0903549938659864,
      "grad_norm": 1.953125,
      "learning_rate": 3.9320377143304816e-05,
      "loss": 2.7025,
      "step": 464
    },
    {
      "epoch": 0.09054972445621483,
      "grad_norm": 1.8125,
      "learning_rate": 3.931718598306571e-05,
      "loss": 2.7123,
      "step": 465
    },
    {
      "epoch": 0.09074445504644324,
      "grad_norm": 1.9375,
      "learning_rate": 3.93139874784453e-05,
      "loss": 2.697,
      "step": 466
    },
    {
      "epoch": 0.09093918563667167,
      "grad_norm": 1.640625,
      "learning_rate": 3.9310781630659666e-05,
      "loss": 2.7097,
      "step": 467
    },
    {
      "epoch": 0.09113391622690008,
      "grad_norm": 2.0625,
      "learning_rate": 3.9307568440927665e-05,
      "loss": 2.6997,
      "step": 468
    },
    {
      "epoch": 0.0913286468171285,
      "grad_norm": 1.828125,
      "learning_rate": 3.930434791047094e-05,
      "loss": 2.7047,
      "step": 469
    },
    {
      "epoch": 0.09152337740735692,
      "grad_norm": 2.015625,
      "learning_rate": 3.930112004051395e-05,
      "loss": 2.7126,
      "step": 470
    },
    {
      "epoch": 0.09171810799758534,
      "grad_norm": 1.7109375,
      "learning_rate": 3.929788483228392e-05,
      "loss": 2.7118,
      "step": 471
    },
    {
      "epoch": 0.09191283858781377,
      "grad_norm": 2.03125,
      "learning_rate": 3.929464228701087e-05,
      "loss": 2.7257,
      "step": 472
    },
    {
      "epoch": 0.09210756917804218,
      "grad_norm": 1.6640625,
      "learning_rate": 3.9291392405927624e-05,
      "loss": 2.7023,
      "step": 473
    },
    {
      "epoch": 0.0923022997682706,
      "grad_norm": 2.046875,
      "learning_rate": 3.928813519026978e-05,
      "loss": 2.6788,
      "step": 474
    },
    {
      "epoch": 0.09249703035849902,
      "grad_norm": 1.7109375,
      "learning_rate": 3.928487064127573e-05,
      "loss": 2.7217,
      "step": 475
    },
    {
      "epoch": 0.09269176094872744,
      "grad_norm": 2.296875,
      "learning_rate": 3.928159876018665e-05,
      "loss": 2.7107,
      "step": 476
    },
    {
      "epoch": 0.09288649153895585,
      "grad_norm": 1.953125,
      "learning_rate": 3.9278319548246505e-05,
      "loss": 2.7205,
      "step": 477
    },
    {
      "epoch": 0.09308122212918428,
      "grad_norm": 2.078125,
      "learning_rate": 3.927503300670206e-05,
      "loss": 2.7072,
      "step": 478
    },
    {
      "epoch": 0.09327595271941269,
      "grad_norm": 1.8671875,
      "learning_rate": 3.927173913680285e-05,
      "loss": 2.6888,
      "step": 479
    },
    {
      "epoch": 0.09347068330964112,
      "grad_norm": 2.078125,
      "learning_rate": 3.9268437939801195e-05,
      "loss": 2.6925,
      "step": 480
    },
    {
      "epoch": 0.09366541389986953,
      "grad_norm": 1.640625,
      "learning_rate": 3.9265129416952216e-05,
      "loss": 2.6936,
      "step": 481
    },
    {
      "epoch": 0.09386014449009795,
      "grad_norm": 2.421875,
      "learning_rate": 3.9261813569513816e-05,
      "loss": 2.7304,
      "step": 482
    },
    {
      "epoch": 0.09405487508032637,
      "grad_norm": 2.0625,
      "learning_rate": 3.9258490398746665e-05,
      "loss": 2.6862,
      "step": 483
    },
    {
      "epoch": 0.09424960567055479,
      "grad_norm": 2.5,
      "learning_rate": 3.9255159905914244e-05,
      "loss": 2.6904,
      "step": 484
    },
    {
      "epoch": 0.0944443362607832,
      "grad_norm": 2.203125,
      "learning_rate": 3.925182209228279e-05,
      "loss": 2.7077,
      "step": 485
    },
    {
      "epoch": 0.09463906685101163,
      "grad_norm": 2.25,
      "learning_rate": 3.924847695912135e-05,
      "loss": 2.7149,
      "step": 486
    },
    {
      "epoch": 0.09483379744124004,
      "grad_norm": 2.125,
      "learning_rate": 3.924512450770173e-05,
      "loss": 2.7006,
      "step": 487
    },
    {
      "epoch": 0.09502852803146847,
      "grad_norm": 2.078125,
      "learning_rate": 3.924176473929854e-05,
      "loss": 2.7121,
      "step": 488
    },
    {
      "epoch": 0.09522325862169688,
      "grad_norm": 1.890625,
      "learning_rate": 3.9238397655189165e-05,
      "loss": 2.6932,
      "step": 489
    },
    {
      "epoch": 0.0954179892119253,
      "grad_norm": 2.28125,
      "learning_rate": 3.923502325665375e-05,
      "loss": 2.7218,
      "step": 490
    },
    {
      "epoch": 0.09561271980215372,
      "grad_norm": 1.9140625,
      "learning_rate": 3.923164154497525e-05,
      "loss": 2.6892,
      "step": 491
    },
    {
      "epoch": 0.09580745039238214,
      "grad_norm": 2.421875,
      "learning_rate": 3.922825252143939e-05,
      "loss": 2.691,
      "step": 492
    },
    {
      "epoch": 0.09600218098261055,
      "grad_norm": 2.203125,
      "learning_rate": 3.922485618733468e-05,
      "loss": 2.699,
      "step": 493
    },
    {
      "epoch": 0.09619691157283898,
      "grad_norm": 2.0,
      "learning_rate": 3.92214525439524e-05,
      "loss": 2.6965,
      "step": 494
    },
    {
      "epoch": 0.09639164216306739,
      "grad_norm": 1.75,
      "learning_rate": 3.92180415925866e-05,
      "loss": 2.7039,
      "step": 495
    },
    {
      "epoch": 0.09658637275329582,
      "grad_norm": 2.34375,
      "learning_rate": 3.9214623334534136e-05,
      "loss": 2.6724,
      "step": 496
    },
    {
      "epoch": 0.09678110334352423,
      "grad_norm": 2.046875,
      "learning_rate": 3.921119777109463e-05,
      "loss": 2.6996,
      "step": 497
    },
    {
      "epoch": 0.09697583393375266,
      "grad_norm": 2.34375,
      "learning_rate": 3.920776490357047e-05,
      "loss": 2.6998,
      "step": 498
    },
    {
      "epoch": 0.09717056452398107,
      "grad_norm": 2.234375,
      "learning_rate": 3.920432473326683e-05,
      "loss": 2.6903,
      "step": 499
    },
    {
      "epoch": 0.0973652951142095,
      "grad_norm": 1.8359375,
      "learning_rate": 3.9200877261491664e-05,
      "loss": 2.687,
      "step": 500
    },
    {
      "epoch": 0.0975600257044379,
      "grad_norm": 1.671875,
      "learning_rate": 3.9197422489555694e-05,
      "loss": 2.6841,
      "step": 501
    },
    {
      "epoch": 0.09775475629466633,
      "grad_norm": 2.265625,
      "learning_rate": 3.9193960418772434e-05,
      "loss": 2.6863,
      "step": 502
    },
    {
      "epoch": 0.09794948688489474,
      "grad_norm": 2.328125,
      "learning_rate": 3.919049105045815e-05,
      "loss": 2.6962,
      "step": 503
    },
    {
      "epoch": 0.09814421747512317,
      "grad_norm": 1.6171875,
      "learning_rate": 3.9187014385931895e-05,
      "loss": 2.7143,
      "step": 504
    },
    {
      "epoch": 0.09833894806535158,
      "grad_norm": 1.3515625,
      "learning_rate": 3.918353042651549e-05,
      "loss": 2.6927,
      "step": 505
    },
    {
      "epoch": 0.09853367865558,
      "grad_norm": 2.609375,
      "learning_rate": 3.918003917353355e-05,
      "loss": 2.7081,
      "step": 506
    },
    {
      "epoch": 0.09872840924580842,
      "grad_norm": 2.4375,
      "learning_rate": 3.917654062831343e-05,
      "loss": 2.6944,
      "step": 507
    },
    {
      "epoch": 0.09892313983603684,
      "grad_norm": 2.015625,
      "learning_rate": 3.9173034792185286e-05,
      "loss": 2.6908,
      "step": 508
    },
    {
      "epoch": 0.09911787042626526,
      "grad_norm": 1.953125,
      "learning_rate": 3.916952166648202e-05,
      "loss": 2.6684,
      "step": 509
    },
    {
      "epoch": 0.09931260101649368,
      "grad_norm": 1.640625,
      "learning_rate": 3.916600125253934e-05,
      "loss": 2.6886,
      "step": 510
    },
    {
      "epoch": 0.0995073316067221,
      "grad_norm": 1.5,
      "learning_rate": 3.9162473551695695e-05,
      "loss": 2.6996,
      "step": 511
    },
    {
      "epoch": 0.09970206219695052,
      "grad_norm": 4.53125,
      "learning_rate": 3.915893856529231e-05,
      "loss": 2.7046,
      "step": 512
    },
    {
      "epoch": 0.09989679278717893,
      "grad_norm": 3.40625,
      "learning_rate": 3.915539629467319e-05,
      "loss": 2.7045,
      "step": 513
    },
    {
      "epoch": 0.10009152337740736,
      "grad_norm": 2.984375,
      "learning_rate": 3.915184674118511e-05,
      "loss": 2.7074,
      "step": 514
    },
    {
      "epoch": 0.10028625396763578,
      "grad_norm": 2.578125,
      "learning_rate": 3.9148289906177595e-05,
      "loss": 2.6798,
      "step": 515
    },
    {
      "epoch": 0.1004809845578642,
      "grad_norm": 2.46875,
      "learning_rate": 3.914472579100296e-05,
      "loss": 2.6958,
      "step": 516
    },
    {
      "epoch": 0.10067571514809262,
      "grad_norm": 1.90625,
      "learning_rate": 3.9141154397016273e-05,
      "loss": 2.6877,
      "step": 517
    },
    {
      "epoch": 0.10087044573832103,
      "grad_norm": 1.9921875,
      "learning_rate": 3.9137575725575376e-05,
      "loss": 2.6889,
      "step": 518
    },
    {
      "epoch": 0.10106517632854946,
      "grad_norm": 1.0,
      "learning_rate": 3.9133989778040885e-05,
      "loss": 2.692,
      "step": 519
    },
    {
      "epoch": 0.10125990691877787,
      "grad_norm": 2.046875,
      "learning_rate": 3.913039655577617e-05,
      "loss": 2.7028,
      "step": 520
    },
    {
      "epoch": 0.1014546375090063,
      "grad_norm": 1.21875,
      "learning_rate": 3.912679606014737e-05,
      "loss": 2.705,
      "step": 521
    },
    {
      "epoch": 0.10164936809923471,
      "grad_norm": 2.484375,
      "learning_rate": 3.9123188292523396e-05,
      "loss": 2.6904,
      "step": 522
    },
    {
      "epoch": 0.10184409868946313,
      "grad_norm": 1.8515625,
      "learning_rate": 3.911957325427592e-05,
      "loss": 2.6709,
      "step": 523
    },
    {
      "epoch": 0.10203882927969155,
      "grad_norm": 2.84375,
      "learning_rate": 3.911595094677937e-05,
      "loss": 2.6937,
      "step": 524
    },
    {
      "epoch": 0.10223355986991997,
      "grad_norm": 2.734375,
      "learning_rate": 3.911232137141094e-05,
      "loss": 2.6945,
      "step": 525
    },
    {
      "epoch": 0.10242829046014838,
      "grad_norm": 1.71875,
      "learning_rate": 3.910868452955061e-05,
      "loss": 2.6891,
      "step": 526
    },
    {
      "epoch": 0.10262302105037681,
      "grad_norm": 1.78125,
      "learning_rate": 3.91050404225811e-05,
      "loss": 2.7044,
      "step": 527
    },
    {
      "epoch": 0.10281775164060522,
      "grad_norm": 1.625,
      "learning_rate": 3.9101389051887884e-05,
      "loss": 2.6946,
      "step": 528
    },
    {
      "epoch": 0.10301248223083365,
      "grad_norm": 1.453125,
      "learning_rate": 3.9097730418859214e-05,
      "loss": 2.6905,
      "step": 529
    },
    {
      "epoch": 0.10320721282106206,
      "grad_norm": 1.9453125,
      "learning_rate": 3.909406452488611e-05,
      "loss": 2.6865,
      "step": 530
    },
    {
      "epoch": 0.10340194341129048,
      "grad_norm": 1.3515625,
      "learning_rate": 3.909039137136234e-05,
      "loss": 2.6897,
      "step": 531
    },
    {
      "epoch": 0.1035966740015189,
      "grad_norm": 2.8125,
      "learning_rate": 3.908671095968443e-05,
      "loss": 2.7116,
      "step": 532
    },
    {
      "epoch": 0.10379140459174732,
      "grad_norm": 2.4375,
      "learning_rate": 3.908302329125168e-05,
      "loss": 2.6837,
      "step": 533
    },
    {
      "epoch": 0.10398613518197573,
      "grad_norm": 2.125,
      "learning_rate": 3.9079328367466115e-05,
      "loss": 2.7053,
      "step": 534
    },
    {
      "epoch": 0.10418086577220416,
      "grad_norm": 2.0625,
      "learning_rate": 3.907562618973257e-05,
      "loss": 2.679,
      "step": 535
    },
    {
      "epoch": 0.10437559636243257,
      "grad_norm": 1.84375,
      "learning_rate": 3.9071916759458595e-05,
      "loss": 2.7068,
      "step": 536
    },
    {
      "epoch": 0.104570326952661,
      "grad_norm": 1.609375,
      "learning_rate": 3.906820007805451e-05,
      "loss": 2.6759,
      "step": 537
    },
    {
      "epoch": 0.10476505754288941,
      "grad_norm": 2.0625,
      "learning_rate": 3.906447614693341e-05,
      "loss": 2.6758,
      "step": 538
    },
    {
      "epoch": 0.10495978813311783,
      "grad_norm": 1.7421875,
      "learning_rate": 3.906074496751112e-05,
      "loss": 2.6989,
      "step": 539
    },
    {
      "epoch": 0.10515451872334625,
      "grad_norm": 2.046875,
      "learning_rate": 3.905700654120624e-05,
      "loss": 2.6617,
      "step": 540
    },
    {
      "epoch": 0.10534924931357467,
      "grad_norm": 1.7890625,
      "learning_rate": 3.905326086944011e-05,
      "loss": 2.6789,
      "step": 541
    },
    {
      "epoch": 0.10554397990380308,
      "grad_norm": 1.890625,
      "learning_rate": 3.904950795363683e-05,
      "loss": 2.6862,
      "step": 542
    },
    {
      "epoch": 0.10573871049403151,
      "grad_norm": 1.546875,
      "learning_rate": 3.904574779522326e-05,
      "loss": 2.6735,
      "step": 543
    },
    {
      "epoch": 0.10593344108425992,
      "grad_norm": 2.109375,
      "learning_rate": 3.904198039562902e-05,
      "loss": 2.6881,
      "step": 544
    },
    {
      "epoch": 0.10612817167448835,
      "grad_norm": 1.609375,
      "learning_rate": 3.9038205756286454e-05,
      "loss": 2.6647,
      "step": 545
    },
    {
      "epoch": 0.10632290226471676,
      "grad_norm": 2.65625,
      "learning_rate": 3.903442387863069e-05,
      "loss": 2.6868,
      "step": 546
    },
    {
      "epoch": 0.10651763285494519,
      "grad_norm": 2.46875,
      "learning_rate": 3.9030634764099584e-05,
      "loss": 2.6766,
      "step": 547
    },
    {
      "epoch": 0.1067123634451736,
      "grad_norm": 1.859375,
      "learning_rate": 3.902683841413377e-05,
      "loss": 2.686,
      "step": 548
    },
    {
      "epoch": 0.10690709403540202,
      "grad_norm": 1.875,
      "learning_rate": 3.902303483017661e-05,
      "loss": 2.6768,
      "step": 549
    },
    {
      "epoch": 0.10710182462563043,
      "grad_norm": 1.640625,
      "learning_rate": 3.9019224013674215e-05,
      "loss": 2.6748,
      "step": 550
    },
    {
      "epoch": 0.10729655521585886,
      "grad_norm": 1.4140625,
      "learning_rate": 3.901540596607547e-05,
      "loss": 2.6757,
      "step": 551
    },
    {
      "epoch": 0.10749128580608727,
      "grad_norm": 2.078125,
      "learning_rate": 3.901158068883199e-05,
      "loss": 2.7015,
      "step": 552
    },
    {
      "epoch": 0.1076860163963157,
      "grad_norm": 1.671875,
      "learning_rate": 3.9007748183398145e-05,
      "loss": 2.671,
      "step": 553
    },
    {
      "epoch": 0.10788074698654411,
      "grad_norm": 2.40625,
      "learning_rate": 3.9003908451231036e-05,
      "loss": 2.6531,
      "step": 554
    },
    {
      "epoch": 0.10807547757677254,
      "grad_norm": 2.3125,
      "learning_rate": 3.900006149379054e-05,
      "loss": 2.6771,
      "step": 555
    },
    {
      "epoch": 0.10827020816700095,
      "grad_norm": 1.5625,
      "learning_rate": 3.899620731253927e-05,
      "loss": 2.6793,
      "step": 556
    },
    {
      "epoch": 0.10846493875722937,
      "grad_norm": 1.6640625,
      "learning_rate": 3.899234590894258e-05,
      "loss": 2.696,
      "step": 557
    },
    {
      "epoch": 0.1086596693474578,
      "grad_norm": 1.71875,
      "learning_rate": 3.8988477284468566e-05,
      "loss": 2.6835,
      "step": 558
    },
    {
      "epoch": 0.10885439993768621,
      "grad_norm": 1.3671875,
      "learning_rate": 3.8984601440588086e-05,
      "loss": 2.6724,
      "step": 559
    },
    {
      "epoch": 0.10904913052791464,
      "grad_norm": 2.21875,
      "learning_rate": 3.898071837877472e-05,
      "loss": 2.6852,
      "step": 560
    },
    {
      "epoch": 0.10924386111814305,
      "grad_norm": 1.890625,
      "learning_rate": 3.897682810050483e-05,
      "loss": 2.6818,
      "step": 561
    },
    {
      "epoch": 0.10943859170837147,
      "grad_norm": 2.078125,
      "learning_rate": 3.897293060725747e-05,
      "loss": 2.6863,
      "step": 562
    },
    {
      "epoch": 0.10963332229859989,
      "grad_norm": 1.984375,
      "learning_rate": 3.8969025900514486e-05,
      "loss": 2.6638,
      "step": 563
    },
    {
      "epoch": 0.10982805288882831,
      "grad_norm": 1.4140625,
      "learning_rate": 3.896511398176043e-05,
      "loss": 2.679,
      "step": 564
    },
    {
      "epoch": 0.11002278347905672,
      "grad_norm": 1.4453125,
      "learning_rate": 3.8961194852482614e-05,
      "loss": 2.6811,
      "step": 565
    },
    {
      "epoch": 0.11021751406928515,
      "grad_norm": 1.84375,
      "learning_rate": 3.8957268514171096e-05,
      "loss": 2.6766,
      "step": 566
    },
    {
      "epoch": 0.11041224465951356,
      "grad_norm": 1.4296875,
      "learning_rate": 3.895333496831866e-05,
      "loss": 2.6819,
      "step": 567
    },
    {
      "epoch": 0.11060697524974199,
      "grad_norm": 2.515625,
      "learning_rate": 3.8949394216420835e-05,
      "loss": 2.6757,
      "step": 568
    },
    {
      "epoch": 0.1108017058399704,
      "grad_norm": 2.359375,
      "learning_rate": 3.89454462599759e-05,
      "loss": 2.6869,
      "step": 569
    },
    {
      "epoch": 0.11099643643019882,
      "grad_norm": 1.484375,
      "learning_rate": 3.8941491100484855e-05,
      "loss": 2.683,
      "step": 570
    },
    {
      "epoch": 0.11119116702042724,
      "grad_norm": 1.6484375,
      "learning_rate": 3.893752873945147e-05,
      "loss": 2.6797,
      "step": 571
    },
    {
      "epoch": 0.11138589761065566,
      "grad_norm": 1.5859375,
      "learning_rate": 3.893355917838221e-05,
      "loss": 2.6708,
      "step": 572
    },
    {
      "epoch": 0.11158062820088407,
      "grad_norm": 1.265625,
      "learning_rate": 3.892958241878631e-05,
      "loss": 2.6459,
      "step": 573
    },
    {
      "epoch": 0.1117753587911125,
      "grad_norm": 2.40625,
      "learning_rate": 3.892559846217572e-05,
      "loss": 2.6645,
      "step": 574
    },
    {
      "epoch": 0.11197008938134091,
      "grad_norm": 2.125,
      "learning_rate": 3.8921607310065154e-05,
      "loss": 2.6589,
      "step": 575
    },
    {
      "epoch": 0.11216481997156934,
      "grad_norm": 1.7734375,
      "learning_rate": 3.891760896397204e-05,
      "loss": 2.6633,
      "step": 576
    },
    {
      "epoch": 0.11235955056179775,
      "grad_norm": 1.765625,
      "learning_rate": 3.8913603425416546e-05,
      "loss": 2.6846,
      "step": 577
    },
    {
      "epoch": 0.11255428115202618,
      "grad_norm": 1.46875,
      "learning_rate": 3.890959069592157e-05,
      "loss": 2.6624,
      "step": 578
    },
    {
      "epoch": 0.11274901174225459,
      "grad_norm": 1.25,
      "learning_rate": 3.890557077701275e-05,
      "loss": 2.6788,
      "step": 579
    },
    {
      "epoch": 0.11294374233248301,
      "grad_norm": 2.0625,
      "learning_rate": 3.890154367021846e-05,
      "loss": 2.6805,
      "step": 580
    },
    {
      "epoch": 0.11313847292271142,
      "grad_norm": 1.71875,
      "learning_rate": 3.8897509377069804e-05,
      "loss": 2.6679,
      "step": 581
    },
    {
      "epoch": 0.11333320351293985,
      "grad_norm": 2.1875,
      "learning_rate": 3.889346789910062e-05,
      "loss": 2.6729,
      "step": 582
    },
    {
      "epoch": 0.11352793410316826,
      "grad_norm": 2.0625,
      "learning_rate": 3.888941923784747e-05,
      "loss": 2.6516,
      "step": 583
    },
    {
      "epoch": 0.11372266469339669,
      "grad_norm": 1.390625,
      "learning_rate": 3.888536339484965e-05,
      "loss": 2.6774,
      "step": 584
    },
    {
      "epoch": 0.1139173952836251,
      "grad_norm": 1.40625,
      "learning_rate": 3.888130037164919e-05,
      "loss": 2.646,
      "step": 585
    },
    {
      "epoch": 0.11411212587385353,
      "grad_norm": 1.625,
      "learning_rate": 3.887723016979086e-05,
      "loss": 2.673,
      "step": 586
    },
    {
      "epoch": 0.11430685646408194,
      "grad_norm": 1.2578125,
      "learning_rate": 3.887315279082213e-05,
      "loss": 2.6507,
      "step": 587
    },
    {
      "epoch": 0.11450158705431036,
      "grad_norm": 2.46875,
      "learning_rate": 3.886906823629323e-05,
      "loss": 2.6773,
      "step": 588
    },
    {
      "epoch": 0.11469631764453878,
      "grad_norm": 2.296875,
      "learning_rate": 3.88649765077571e-05,
      "loss": 2.6679,
      "step": 589
    },
    {
      "epoch": 0.1148910482347672,
      "grad_norm": 1.53125,
      "learning_rate": 3.886087760676942e-05,
      "loss": 2.6876,
      "step": 590
    },
    {
      "epoch": 0.11508577882499561,
      "grad_norm": 1.734375,
      "learning_rate": 3.885677153488858e-05,
      "loss": 2.687,
      "step": 591
    },
    {
      "epoch": 0.11528050941522404,
      "grad_norm": 1.4296875,
      "learning_rate": 3.88526582936757e-05,
      "loss": 2.6523,
      "step": 592
    },
    {
      "epoch": 0.11547524000545245,
      "grad_norm": 1.2421875,
      "learning_rate": 3.884853788469465e-05,
      "loss": 2.6543,
      "step": 593
    },
    {
      "epoch": 0.11566997059568088,
      "grad_norm": 1.984375,
      "learning_rate": 3.884441030951199e-05,
      "loss": 2.6719,
      "step": 594
    },
    {
      "epoch": 0.11586470118590929,
      "grad_norm": 1.6015625,
      "learning_rate": 3.884027556969703e-05,
      "loss": 2.6796,
      "step": 595
    },
    {
      "epoch": 0.11605943177613771,
      "grad_norm": 2.328125,
      "learning_rate": 3.88361336668218e-05,
      "loss": 2.6492,
      "step": 596
    },
    {
      "epoch": 0.11625416236636613,
      "grad_norm": 2.171875,
      "learning_rate": 3.883198460246104e-05,
      "loss": 2.679,
      "step": 597
    },
    {
      "epoch": 0.11644889295659455,
      "grad_norm": 1.3984375,
      "learning_rate": 3.882782837819222e-05,
      "loss": 2.6618,
      "step": 598
    },
    {
      "epoch": 0.11664362354682296,
      "grad_norm": 1.390625,
      "learning_rate": 3.882366499559554e-05,
      "loss": 2.6688,
      "step": 599
    },
    {
      "epoch": 0.11683835413705139,
      "grad_norm": 1.5546875,
      "learning_rate": 3.881949445625391e-05,
      "loss": 2.6594,
      "step": 600
    },
    {
      "epoch": 0.11703308472727982,
      "grad_norm": 1.171875,
      "learning_rate": 3.881531676175297e-05,
      "loss": 2.6838,
      "step": 601
    },
    {
      "epoch": 0.11722781531750823,
      "grad_norm": 2.28125,
      "learning_rate": 3.881113191368108e-05,
      "loss": 2.661,
      "step": 602
    },
    {
      "epoch": 0.11742254590773665,
      "grad_norm": 2.046875,
      "learning_rate": 3.880693991362931e-05,
      "loss": 2.6572,
      "step": 603
    },
    {
      "epoch": 0.11761727649796506,
      "grad_norm": 1.703125,
      "learning_rate": 3.880274076319146e-05,
      "loss": 2.6652,
      "step": 604
    },
    {
      "epoch": 0.11781200708819349,
      "grad_norm": 1.8125,
      "learning_rate": 3.8798534463964036e-05,
      "loss": 2.6715,
      "step": 605
    },
    {
      "epoch": 0.1180067376784219,
      "grad_norm": 1.421875,
      "learning_rate": 3.879432101754629e-05,
      "loss": 2.663,
      "step": 606
    },
    {
      "epoch": 0.11820146826865033,
      "grad_norm": 1.2578125,
      "learning_rate": 3.8790100425540144e-05,
      "loss": 2.6672,
      "step": 607
    },
    {
      "epoch": 0.11839619885887874,
      "grad_norm": 1.796875,
      "learning_rate": 3.878587268955028e-05,
      "loss": 2.6639,
      "step": 608
    },
    {
      "epoch": 0.11859092944910717,
      "grad_norm": 1.3515625,
      "learning_rate": 3.878163781118408e-05,
      "loss": 2.6562,
      "step": 609
    },
    {
      "epoch": 0.11878566003933558,
      "grad_norm": 2.140625,
      "learning_rate": 3.8777395792051644e-05,
      "loss": 2.6381,
      "step": 610
    },
    {
      "epoch": 0.118980390629564,
      "grad_norm": 1.953125,
      "learning_rate": 3.877314663376578e-05,
      "loss": 2.6648,
      "step": 611
    },
    {
      "epoch": 0.11917512121979242,
      "grad_norm": 1.6171875,
      "learning_rate": 3.8768890337942016e-05,
      "loss": 2.6649,
      "step": 612
    },
    {
      "epoch": 0.11936985181002084,
      "grad_norm": 1.5,
      "learning_rate": 3.8764626906198594e-05,
      "loss": 2.6466,
      "step": 613
    },
    {
      "epoch": 0.11956458240024925,
      "grad_norm": 1.859375,
      "learning_rate": 3.876035634015647e-05,
      "loss": 2.6616,
      "step": 614
    },
    {
      "epoch": 0.11975931299047768,
      "grad_norm": 1.515625,
      "learning_rate": 3.875607864143931e-05,
      "loss": 2.6442,
      "step": 615
    },
    {
      "epoch": 0.11995404358070609,
      "grad_norm": 2.15625,
      "learning_rate": 3.875179381167348e-05,
      "loss": 2.6782,
      "step": 616
    },
    {
      "epoch": 0.12014877417093452,
      "grad_norm": 1.8984375,
      "learning_rate": 3.874750185248808e-05,
      "loss": 2.6581,
      "step": 617
    },
    {
      "epoch": 0.12034350476116293,
      "grad_norm": 1.6875,
      "learning_rate": 3.8743202765514914e-05,
      "loss": 2.6498,
      "step": 618
    },
    {
      "epoch": 0.12053823535139135,
      "grad_norm": 1.4765625,
      "learning_rate": 3.873889655238848e-05,
      "loss": 2.6659,
      "step": 619
    },
    {
      "epoch": 0.12073296594161977,
      "grad_norm": 1.625,
      "learning_rate": 3.873458321474602e-05,
      "loss": 2.6682,
      "step": 620
    },
    {
      "epoch": 0.12092769653184819,
      "grad_norm": 1.3046875,
      "learning_rate": 3.873026275422744e-05,
      "loss": 2.644,
      "step": 621
    },
    {
      "epoch": 0.1211224271220766,
      "grad_norm": 2.046875,
      "learning_rate": 3.872593517247538e-05,
      "loss": 2.6594,
      "step": 622
    },
    {
      "epoch": 0.12131715771230503,
      "grad_norm": 1.796875,
      "learning_rate": 3.872160047113519e-05,
      "loss": 2.6616,
      "step": 623
    },
    {
      "epoch": 0.12151188830253344,
      "grad_norm": 1.890625,
      "learning_rate": 3.8717258651854914e-05,
      "loss": 2.6474,
      "step": 624
    },
    {
      "epoch": 0.12170661889276187,
      "grad_norm": 1.734375,
      "learning_rate": 3.871290971628532e-05,
      "loss": 2.6511,
      "step": 625
    },
    {
      "epoch": 0.12190134948299028,
      "grad_norm": 1.7265625,
      "learning_rate": 3.870855366607986e-05,
      "loss": 2.6804,
      "step": 626
    },
    {
      "epoch": 0.1220960800732187,
      "grad_norm": 1.5390625,
      "learning_rate": 3.8704190502894696e-05,
      "loss": 2.6632,
      "step": 627
    },
    {
      "epoch": 0.12229081066344712,
      "grad_norm": 1.8515625,
      "learning_rate": 3.869982022838871e-05,
      "loss": 2.6615,
      "step": 628
    },
    {
      "epoch": 0.12248554125367554,
      "grad_norm": 1.609375,
      "learning_rate": 3.8695442844223485e-05,
      "loss": 2.6748,
      "step": 629
    },
    {
      "epoch": 0.12268027184390395,
      "grad_norm": 2.140625,
      "learning_rate": 3.869105835206328e-05,
      "loss": 2.6571,
      "step": 630
    },
    {
      "epoch": 0.12287500243413238,
      "grad_norm": 1.8828125,
      "learning_rate": 3.868666675357509e-05,
      "loss": 2.6524,
      "step": 631
    },
    {
      "epoch": 0.12306973302436079,
      "grad_norm": 1.8359375,
      "learning_rate": 3.8682268050428594e-05,
      "loss": 2.6533,
      "step": 632
    },
    {
      "epoch": 0.12326446361458922,
      "grad_norm": 1.609375,
      "learning_rate": 3.8677862244296175e-05,
      "loss": 2.655,
      "step": 633
    },
    {
      "epoch": 0.12345919420481763,
      "grad_norm": 1.7578125,
      "learning_rate": 3.8673449336852916e-05,
      "loss": 2.6527,
      "step": 634
    },
    {
      "epoch": 0.12365392479504606,
      "grad_norm": 1.4375,
      "learning_rate": 3.866902932977661e-05,
      "loss": 2.6763,
      "step": 635
    },
    {
      "epoch": 0.12384865538527447,
      "grad_norm": 2.046875,
      "learning_rate": 3.866460222474773e-05,
      "loss": 2.6608,
      "step": 636
    },
    {
      "epoch": 0.12404338597550289,
      "grad_norm": 1.6484375,
      "learning_rate": 3.866016802344946e-05,
      "loss": 2.6479,
      "step": 637
    },
    {
      "epoch": 0.1242381165657313,
      "grad_norm": 2.25,
      "learning_rate": 3.865572672756769e-05,
      "loss": 2.656,
      "step": 638
    },
    {
      "epoch": 0.12443284715595973,
      "grad_norm": 1.9375,
      "learning_rate": 3.865127833879099e-05,
      "loss": 2.6697,
      "step": 639
    },
    {
      "epoch": 0.12462757774618814,
      "grad_norm": 1.96875,
      "learning_rate": 3.864682285881063e-05,
      "loss": 2.6664,
      "step": 640
    },
    {
      "epoch": 0.12482230833641657,
      "grad_norm": 1.7265625,
      "learning_rate": 3.864236028932059e-05,
      "loss": 2.662,
      "step": 641
    },
    {
      "epoch": 0.12501703892664498,
      "grad_norm": 2.015625,
      "learning_rate": 3.8637890632017534e-05,
      "loss": 2.6386,
      "step": 642
    },
    {
      "epoch": 0.1252117695168734,
      "grad_norm": 1.640625,
      "learning_rate": 3.8633413888600815e-05,
      "loss": 2.6466,
      "step": 643
    },
    {
      "epoch": 0.12540650010710183,
      "grad_norm": 2.3125,
      "learning_rate": 3.86289300607725e-05,
      "loss": 2.665,
      "step": 644
    },
    {
      "epoch": 0.12560123069733026,
      "grad_norm": 1.875,
      "learning_rate": 3.862443915023733e-05,
      "loss": 2.6505,
      "step": 645
    },
    {
      "epoch": 0.12579596128755866,
      "grad_norm": 2.34375,
      "learning_rate": 3.861994115870274e-05,
      "loss": 2.6715,
      "step": 646
    },
    {
      "epoch": 0.12599069187778708,
      "grad_norm": 2.265625,
      "learning_rate": 3.861543608787888e-05,
      "loss": 2.6575,
      "step": 647
    },
    {
      "epoch": 0.1261854224680155,
      "grad_norm": 1.71875,
      "learning_rate": 3.861092393947856e-05,
      "loss": 2.6604,
      "step": 648
    },
    {
      "epoch": 0.12638015305824393,
      "grad_norm": 1.6484375,
      "learning_rate": 3.86064047152173e-05,
      "loss": 2.6351,
      "step": 649
    },
    {
      "epoch": 0.12657488364847233,
      "grad_norm": 1.8359375,
      "learning_rate": 3.860187841681331e-05,
      "loss": 2.6526,
      "step": 650
    },
    {
      "epoch": 0.12676961423870076,
      "grad_norm": 1.46875,
      "learning_rate": 3.8597345045987486e-05,
      "loss": 2.664,
      "step": 651
    },
    {
      "epoch": 0.12696434482892918,
      "grad_norm": 2.46875,
      "learning_rate": 3.8592804604463404e-05,
      "loss": 2.6433,
      "step": 652
    },
    {
      "epoch": 0.1271590754191576,
      "grad_norm": 2.28125,
      "learning_rate": 3.858825709396735e-05,
      "loss": 2.6552,
      "step": 653
    },
    {
      "epoch": 0.127353806009386,
      "grad_norm": 1.796875,
      "learning_rate": 3.858370251622827e-05,
      "loss": 2.6661,
      "step": 654
    },
    {
      "epoch": 0.12754853659961443,
      "grad_norm": 1.8203125,
      "learning_rate": 3.857914087297782e-05,
      "loss": 2.6458,
      "step": 655
    },
    {
      "epoch": 0.12774326718984286,
      "grad_norm": 1.828125,
      "learning_rate": 3.857457216595033e-05,
      "loss": 2.6577,
      "step": 656
    },
    {
      "epoch": 0.12793799778007128,
      "grad_norm": 1.5703125,
      "learning_rate": 3.856999639688283e-05,
      "loss": 2.6389,
      "step": 657
    },
    {
      "epoch": 0.12813272837029968,
      "grad_norm": 2.109375,
      "learning_rate": 3.8565413567515006e-05,
      "loss": 2.6449,
      "step": 658
    },
    {
      "epoch": 0.1283274589605281,
      "grad_norm": 1.9375,
      "learning_rate": 3.856082367958927e-05,
      "loss": 2.6369,
      "step": 659
    },
    {
      "epoch": 0.12852218955075653,
      "grad_norm": 1.953125,
      "learning_rate": 3.855622673485066e-05,
      "loss": 2.632,
      "step": 660
    },
    {
      "epoch": 0.12871692014098496,
      "grad_norm": 1.859375,
      "learning_rate": 3.855162273504697e-05,
      "loss": 2.6554,
      "step": 661
    },
    {
      "epoch": 0.12891165073121336,
      "grad_norm": 1.6875,
      "learning_rate": 3.854701168192862e-05,
      "loss": 2.6512,
      "step": 662
    },
    {
      "epoch": 0.12910638132144178,
      "grad_norm": 1.59375,
      "learning_rate": 3.854239357724872e-05,
      "loss": 2.6367,
      "step": 663
    },
    {
      "epoch": 0.1293011119116702,
      "grad_norm": 1.9140625,
      "learning_rate": 3.853776842276308e-05,
      "loss": 2.6361,
      "step": 664
    },
    {
      "epoch": 0.12949584250189863,
      "grad_norm": 1.546875,
      "learning_rate": 3.853313622023019e-05,
      "loss": 2.6443,
      "step": 665
    },
    {
      "epoch": 0.12969057309212703,
      "grad_norm": 2.03125,
      "learning_rate": 3.8528496971411195e-05,
      "loss": 2.6386,
      "step": 666
    },
    {
      "epoch": 0.12988530368235546,
      "grad_norm": 1.8984375,
      "learning_rate": 3.852385067806994e-05,
      "loss": 2.6532,
      "step": 667
    },
    {
      "epoch": 0.13008003427258388,
      "grad_norm": 1.625,
      "learning_rate": 3.851919734197294e-05,
      "loss": 2.626,
      "step": 668
    },
    {
      "epoch": 0.1302747648628123,
      "grad_norm": 1.59375,
      "learning_rate": 3.851453696488939e-05,
      "loss": 2.6581,
      "step": 669
    },
    {
      "epoch": 0.1304694954530407,
      "grad_norm": 1.8046875,
      "learning_rate": 3.850986954859118e-05,
      "loss": 2.6416,
      "step": 670
    },
    {
      "epoch": 0.13066422604326913,
      "grad_norm": 1.40625,
      "learning_rate": 3.850519509485283e-05,
      "loss": 2.6329,
      "step": 671
    },
    {
      "epoch": 0.13085895663349756,
      "grad_norm": 2.390625,
      "learning_rate": 3.850051360545158e-05,
      "loss": 2.6494,
      "step": 672
    },
    {
      "epoch": 0.13105368722372598,
      "grad_norm": 2.0625,
      "learning_rate": 3.849582508216733e-05,
      "loss": 2.6599,
      "step": 673
    },
    {
      "epoch": 0.13124841781395438,
      "grad_norm": 1.8203125,
      "learning_rate": 3.849112952678265e-05,
      "loss": 2.6392,
      "step": 674
    },
    {
      "epoch": 0.1314431484041828,
      "grad_norm": 1.78125,
      "learning_rate": 3.848642694108279e-05,
      "loss": 2.6531,
      "step": 675
    },
    {
      "epoch": 0.13163787899441123,
      "grad_norm": 1.5234375,
      "learning_rate": 3.848171732685566e-05,
      "loss": 2.6283,
      "step": 676
    },
    {
      "epoch": 0.13183260958463966,
      "grad_norm": 1.28125,
      "learning_rate": 3.8477000685891866e-05,
      "loss": 2.6424,
      "step": 677
    },
    {
      "epoch": 0.13202734017486806,
      "grad_norm": 2.203125,
      "learning_rate": 3.847227701998467e-05,
      "loss": 2.653,
      "step": 678
    },
    {
      "epoch": 0.13222207076509648,
      "grad_norm": 1.859375,
      "learning_rate": 3.8467546330929995e-05,
      "loss": 2.6495,
      "step": 679
    },
    {
      "epoch": 0.1324168013553249,
      "grad_norm": 2.265625,
      "learning_rate": 3.846280862052646e-05,
      "loss": 2.6455,
      "step": 680
    },
    {
      "epoch": 0.13261153194555333,
      "grad_norm": 2.265625,
      "learning_rate": 3.845806389057532e-05,
      "loss": 2.6616,
      "step": 681
    },
    {
      "epoch": 0.13280626253578173,
      "grad_norm": 1.2421875,
      "learning_rate": 3.845331214288054e-05,
      "loss": 2.6401,
      "step": 682
    },
    {
      "epoch": 0.13300099312601016,
      "grad_norm": 1.359375,
      "learning_rate": 3.8448553379248724e-05,
      "loss": 2.6653,
      "step": 683
    },
    {
      "epoch": 0.13319572371623858,
      "grad_norm": 1.6875,
      "learning_rate": 3.8443787601489145e-05,
      "loss": 2.6463,
      "step": 684
    },
    {
      "epoch": 0.133390454306467,
      "grad_norm": 1.25,
      "learning_rate": 3.843901481141375e-05,
      "loss": 2.63,
      "step": 685
    },
    {
      "epoch": 0.1335851848966954,
      "grad_norm": 2.84375,
      "learning_rate": 3.843423501083715e-05,
      "loss": 2.6326,
      "step": 686
    },
    {
      "epoch": 0.13377991548692383,
      "grad_norm": 2.8125,
      "learning_rate": 3.842944820157663e-05,
      "loss": 2.6498,
      "step": 687
    },
    {
      "epoch": 0.13397464607715226,
      "grad_norm": 1.03125,
      "learning_rate": 3.842465438545213e-05,
      "loss": 2.6426,
      "step": 688
    },
    {
      "epoch": 0.13416937666738069,
      "grad_norm": 1.6875,
      "learning_rate": 3.841985356428625e-05,
      "loss": 2.651,
      "step": 689
    },
    {
      "epoch": 0.1343641072576091,
      "grad_norm": 1.109375,
      "learning_rate": 3.841504573990426e-05,
      "loss": 2.6384,
      "step": 690
    },
    {
      "epoch": 0.1345588378478375,
      "grad_norm": 1.3671875,
      "learning_rate": 3.841023091413409e-05,
      "loss": 2.6358,
      "step": 691
    },
    {
      "epoch": 0.13475356843806593,
      "grad_norm": 1.28125,
      "learning_rate": 3.840540908880634e-05,
      "loss": 2.6452,
      "step": 692
    },
    {
      "epoch": 0.13494829902829436,
      "grad_norm": 1.1171875,
      "learning_rate": 3.840058026575426e-05,
      "loss": 2.6144,
      "step": 693
    },
    {
      "epoch": 0.1351430296185228,
      "grad_norm": 1.40625,
      "learning_rate": 3.8395744446813765e-05,
      "loss": 2.6419,
      "step": 694
    },
    {
      "epoch": 0.13533776020875118,
      "grad_norm": 0.99609375,
      "learning_rate": 3.839090163382344e-05,
      "loss": 2.6616,
      "step": 695
    },
    {
      "epoch": 0.1355324907989796,
      "grad_norm": 1.8046875,
      "learning_rate": 3.83860518286245e-05,
      "loss": 2.6595,
      "step": 696
    },
    {
      "epoch": 0.13572722138920804,
      "grad_norm": 1.1484375,
      "learning_rate": 3.838119503306086e-05,
      "loss": 2.663,
      "step": 697
    },
    {
      "epoch": 0.13592195197943646,
      "grad_norm": 1.90625,
      "learning_rate": 3.837633124897905e-05,
      "loss": 2.6043,
      "step": 698
    },
    {
      "epoch": 0.13611668256966486,
      "grad_norm": 1.5078125,
      "learning_rate": 3.837146047822829e-05,
      "loss": 2.649,
      "step": 699
    },
    {
      "epoch": 0.13631141315989329,
      "grad_norm": 1.9609375,
      "learning_rate": 3.836658272266044e-05,
      "loss": 2.6512,
      "step": 700
    },
    {
      "epoch": 0.1365061437501217,
      "grad_norm": 1.625,
      "learning_rate": 3.836169798413002e-05,
      "loss": 2.6409,
      "step": 701
    },
    {
      "epoch": 0.13670087434035014,
      "grad_norm": 1.9296875,
      "learning_rate": 3.8356806264494205e-05,
      "loss": 2.6307,
      "step": 702
    },
    {
      "epoch": 0.13689560493057853,
      "grad_norm": 1.546875,
      "learning_rate": 3.8351907565612824e-05,
      "loss": 2.6392,
      "step": 703
    },
    {
      "epoch": 0.13709033552080696,
      "grad_norm": 1.8203125,
      "learning_rate": 3.834700188934836e-05,
      "loss": 2.6623,
      "step": 704
    },
    {
      "epoch": 0.1372850661110354,
      "grad_norm": 1.46875,
      "learning_rate": 3.834208923756594e-05,
      "loss": 2.6541,
      "step": 705
    },
    {
      "epoch": 0.1374797967012638,
      "grad_norm": 1.6640625,
      "learning_rate": 3.8337169612133355e-05,
      "loss": 2.64,
      "step": 706
    },
    {
      "epoch": 0.1376745272914922,
      "grad_norm": 1.4296875,
      "learning_rate": 3.833224301492105e-05,
      "loss": 2.6577,
      "step": 707
    },
    {
      "epoch": 0.13786925788172064,
      "grad_norm": 1.859375,
      "learning_rate": 3.8327309447802114e-05,
      "loss": 2.6638,
      "step": 708
    },
    {
      "epoch": 0.13806398847194906,
      "grad_norm": 1.3984375,
      "learning_rate": 3.8322368912652275e-05,
      "loss": 2.626,
      "step": 709
    },
    {
      "epoch": 0.1382587190621775,
      "grad_norm": 2.125,
      "learning_rate": 3.831742141134993e-05,
      "loss": 2.6703,
      "step": 710
    },
    {
      "epoch": 0.13845344965240589,
      "grad_norm": 1.84375,
      "learning_rate": 3.831246694577611e-05,
      "loss": 2.6508,
      "step": 711
    },
    {
      "epoch": 0.1386481802426343,
      "grad_norm": 2.0,
      "learning_rate": 3.83075055178145e-05,
      "loss": 2.6474,
      "step": 712
    },
    {
      "epoch": 0.13884291083286274,
      "grad_norm": 1.8046875,
      "learning_rate": 3.830253712935144e-05,
      "loss": 2.6394,
      "step": 713
    },
    {
      "epoch": 0.13903764142309116,
      "grad_norm": 1.6875,
      "learning_rate": 3.8297561782275893e-05,
      "loss": 2.6607,
      "step": 714
    },
    {
      "epoch": 0.13923237201331956,
      "grad_norm": 1.5,
      "learning_rate": 3.82925794784795e-05,
      "loss": 2.6444,
      "step": 715
    },
    {
      "epoch": 0.139427102603548,
      "grad_norm": 1.53125,
      "learning_rate": 3.8287590219856524e-05,
      "loss": 2.641,
      "step": 716
    },
    {
      "epoch": 0.1396218331937764,
      "grad_norm": 1.203125,
      "learning_rate": 3.8282594008303866e-05,
      "loss": 2.6586,
      "step": 717
    },
    {
      "epoch": 0.13981656378400484,
      "grad_norm": 1.4609375,
      "learning_rate": 3.82775908457211e-05,
      "loss": 2.6224,
      "step": 718
    },
    {
      "epoch": 0.14001129437423324,
      "grad_norm": 1.2265625,
      "learning_rate": 3.827258073401041e-05,
      "loss": 2.6356,
      "step": 719
    },
    {
      "epoch": 0.14020602496446166,
      "grad_norm": 1.546875,
      "learning_rate": 3.8267563675076644e-05,
      "loss": 2.6399,
      "step": 720
    },
    {
      "epoch": 0.1404007555546901,
      "grad_norm": 1.0859375,
      "learning_rate": 3.826253967082729e-05,
      "loss": 2.6377,
      "step": 721
    },
    {
      "epoch": 0.1405954861449185,
      "grad_norm": 1.9375,
      "learning_rate": 3.8257508723172464e-05,
      "loss": 2.6365,
      "step": 722
    },
    {
      "epoch": 0.1407902167351469,
      "grad_norm": 1.5625,
      "learning_rate": 3.8252470834024925e-05,
      "loss": 2.6582,
      "step": 723
    },
    {
      "epoch": 0.14098494732537534,
      "grad_norm": 2.0,
      "learning_rate": 3.824742600530008e-05,
      "loss": 2.633,
      "step": 724
    },
    {
      "epoch": 0.14117967791560376,
      "grad_norm": 1.84375,
      "learning_rate": 3.824237423891598e-05,
      "loss": 2.6319,
      "step": 725
    },
    {
      "epoch": 0.1413744085058322,
      "grad_norm": 1.5859375,
      "learning_rate": 3.823731553679328e-05,
      "loss": 2.6185,
      "step": 726
    },
    {
      "epoch": 0.1415691390960606,
      "grad_norm": 1.5625,
      "learning_rate": 3.8232249900855315e-05,
      "loss": 2.6416,
      "step": 727
    },
    {
      "epoch": 0.141763869686289,
      "grad_norm": 1.359375,
      "learning_rate": 3.8227177333028035e-05,
      "loss": 2.6204,
      "step": 728
    },
    {
      "epoch": 0.14195860027651744,
      "grad_norm": 1.40625,
      "learning_rate": 3.822209783524001e-05,
      "loss": 2.6509,
      "step": 729
    },
    {
      "epoch": 0.14215333086674586,
      "grad_norm": 1.328125,
      "learning_rate": 3.821701140942248e-05,
      "loss": 2.623,
      "step": 730
    },
    {
      "epoch": 0.1423480614569743,
      "grad_norm": 1.1875,
      "learning_rate": 3.821191805750929e-05,
      "loss": 2.6383,
      "step": 731
    },
    {
      "epoch": 0.1425427920472027,
      "grad_norm": 1.40625,
      "learning_rate": 3.820681778143694e-05,
      "loss": 2.6333,
      "step": 732
    },
    {
      "epoch": 0.1427375226374311,
      "grad_norm": 1.1640625,
      "learning_rate": 3.820171058314454e-05,
      "loss": 2.6126,
      "step": 733
    },
    {
      "epoch": 0.14293225322765954,
      "grad_norm": 1.484375,
      "learning_rate": 3.8196596464573857e-05,
      "loss": 2.6236,
      "step": 734
    },
    {
      "epoch": 0.14312698381788796,
      "grad_norm": 1.2109375,
      "learning_rate": 3.819147542766925e-05,
      "loss": 2.6436,
      "step": 735
    },
    {
      "epoch": 0.14332171440811636,
      "grad_norm": 1.5078125,
      "learning_rate": 3.8186347474377766e-05,
      "loss": 2.6375,
      "step": 736
    },
    {
      "epoch": 0.1435164449983448,
      "grad_norm": 1.171875,
      "learning_rate": 3.818121260664903e-05,
      "loss": 2.6299,
      "step": 737
    },
    {
      "epoch": 0.14371117558857321,
      "grad_norm": 1.5234375,
      "learning_rate": 3.8176070826435315e-05,
      "loss": 2.6281,
      "step": 738
    },
    {
      "epoch": 0.14390590617880164,
      "grad_norm": 1.4296875,
      "learning_rate": 3.8170922135691525e-05,
      "loss": 2.6311,
      "step": 739
    },
    {
      "epoch": 0.14410063676903004,
      "grad_norm": 1.40625,
      "learning_rate": 3.816576653637519e-05,
      "loss": 2.6281,
      "step": 740
    },
    {
      "epoch": 0.14429536735925846,
      "grad_norm": 1.2734375,
      "learning_rate": 3.816060403044647e-05,
      "loss": 2.6514,
      "step": 741
    },
    {
      "epoch": 0.1444900979494869,
      "grad_norm": 1.3125,
      "learning_rate": 3.815543461986814e-05,
      "loss": 2.657,
      "step": 742
    },
    {
      "epoch": 0.14468482853971532,
      "grad_norm": 1.15625,
      "learning_rate": 3.8150258306605616e-05,
      "loss": 2.6295,
      "step": 743
    },
    {
      "epoch": 0.1448795591299437,
      "grad_norm": 1.5390625,
      "learning_rate": 3.814507509262691e-05,
      "loss": 2.6369,
      "step": 744
    },
    {
      "epoch": 0.14507428972017214,
      "grad_norm": 1.171875,
      "learning_rate": 3.8139884979902696e-05,
      "loss": 2.6159,
      "step": 745
    },
    {
      "epoch": 0.14526902031040057,
      "grad_norm": 1.7109375,
      "learning_rate": 3.813468797040624e-05,
      "loss": 2.6359,
      "step": 746
    },
    {
      "epoch": 0.145463750900629,
      "grad_norm": 1.5390625,
      "learning_rate": 3.812948406611345e-05,
      "loss": 2.6319,
      "step": 747
    },
    {
      "epoch": 0.1456584814908574,
      "grad_norm": 1.421875,
      "learning_rate": 3.8124273269002834e-05,
      "loss": 2.6277,
      "step": 748
    },
    {
      "epoch": 0.14585321208108581,
      "grad_norm": 1.28125,
      "learning_rate": 3.8119055581055545e-05,
      "loss": 2.6151,
      "step": 749
    },
    {
      "epoch": 0.14604794267131424,
      "grad_norm": 1.0859375,
      "learning_rate": 3.8113831004255343e-05,
      "loss": 2.6108,
      "step": 750
    },
    {
      "epoch": 0.14624267326154267,
      "grad_norm": 1.2421875,
      "learning_rate": 3.8108599540588604e-05,
      "loss": 2.6297,
      "step": 751
    },
    {
      "epoch": 0.14643740385177106,
      "grad_norm": 0.95703125,
      "learning_rate": 3.810336119204433e-05,
      "loss": 2.6323,
      "step": 752
    },
    {
      "epoch": 0.1466321344419995,
      "grad_norm": 1.25,
      "learning_rate": 3.809811596061414e-05,
      "loss": 2.6296,
      "step": 753
    },
    {
      "epoch": 0.14682686503222792,
      "grad_norm": 1.1640625,
      "learning_rate": 3.8092863848292256e-05,
      "loss": 2.6193,
      "step": 754
    },
    {
      "epoch": 0.14702159562245634,
      "grad_norm": 1.4296875,
      "learning_rate": 3.808760485707555e-05,
      "loss": 2.6339,
      "step": 755
    },
    {
      "epoch": 0.14721632621268474,
      "grad_norm": 1.0625,
      "learning_rate": 3.8082338988963465e-05,
      "loss": 2.6367,
      "step": 756
    },
    {
      "epoch": 0.14741105680291317,
      "grad_norm": 1.359375,
      "learning_rate": 3.807706624595809e-05,
      "loss": 2.6058,
      "step": 757
    },
    {
      "epoch": 0.1476057873931416,
      "grad_norm": 1.0625,
      "learning_rate": 3.8071786630064115e-05,
      "loss": 2.6202,
      "step": 758
    },
    {
      "epoch": 0.14780051798337002,
      "grad_norm": 1.53125,
      "learning_rate": 3.8066500143288856e-05,
      "loss": 2.636,
      "step": 759
    },
    {
      "epoch": 0.14799524857359841,
      "grad_norm": 1.265625,
      "learning_rate": 3.806120678764222e-05,
      "loss": 2.6268,
      "step": 760
    },
    {
      "epoch": 0.14818997916382684,
      "grad_norm": 1.4609375,
      "learning_rate": 3.805590656513674e-05,
      "loss": 2.6347,
      "step": 761
    },
    {
      "epoch": 0.14838470975405527,
      "grad_norm": 1.2578125,
      "learning_rate": 3.8050599477787564e-05,
      "loss": 2.6117,
      "step": 762
    },
    {
      "epoch": 0.1485794403442837,
      "grad_norm": 1.34375,
      "learning_rate": 3.804528552761244e-05,
      "loss": 2.6346,
      "step": 763
    },
    {
      "epoch": 0.1487741709345121,
      "grad_norm": 1.2421875,
      "learning_rate": 3.803996471663171e-05,
      "loss": 2.6181,
      "step": 764
    },
    {
      "epoch": 0.14896890152474052,
      "grad_norm": 1.0546875,
      "learning_rate": 3.803463704686838e-05,
      "loss": 2.6196,
      "step": 765
    },
    {
      "epoch": 0.14916363211496894,
      "grad_norm": 1.3125,
      "learning_rate": 3.8029302520348e-05,
      "loss": 2.6245,
      "step": 766
    },
    {
      "epoch": 0.14935836270519737,
      "grad_norm": 1.15625,
      "learning_rate": 3.802396113909875e-05,
      "loss": 2.6178,
      "step": 767
    },
    {
      "epoch": 0.14955309329542577,
      "grad_norm": 1.21875,
      "learning_rate": 3.801861290515144e-05,
      "loss": 2.6309,
      "step": 768
    },
    {
      "epoch": 0.1497478238856542,
      "grad_norm": 0.94921875,
      "learning_rate": 3.801325782053945e-05,
      "loss": 2.6408,
      "step": 769
    },
    {
      "epoch": 0.14994255447588262,
      "grad_norm": 1.34375,
      "learning_rate": 3.800789588729878e-05,
      "loss": 2.6137,
      "step": 770
    },
    {
      "epoch": 0.15013728506611104,
      "grad_norm": 1.0859375,
      "learning_rate": 3.800252710746803e-05,
      "loss": 2.6187,
      "step": 771
    },
    {
      "epoch": 0.15033201565633944,
      "grad_norm": 1.2890625,
      "learning_rate": 3.7997151483088425e-05,
      "loss": 2.6183,
      "step": 772
    },
    {
      "epoch": 0.15052674624656787,
      "grad_norm": 1.0390625,
      "learning_rate": 3.799176901620377e-05,
      "loss": 2.6231,
      "step": 773
    },
    {
      "epoch": 0.1507214768367963,
      "grad_norm": 1.5,
      "learning_rate": 3.798637970886045e-05,
      "loss": 2.6136,
      "step": 774
    },
    {
      "epoch": 0.15091620742702472,
      "grad_norm": 1.234375,
      "learning_rate": 3.7980983563107505e-05,
      "loss": 2.6364,
      "step": 775
    },
    {
      "epoch": 0.15111093801725314,
      "grad_norm": 1.4453125,
      "learning_rate": 3.797558058099653e-05,
      "loss": 2.6123,
      "step": 776
    },
    {
      "epoch": 0.15130566860748154,
      "grad_norm": 1.265625,
      "learning_rate": 3.797017076458174e-05,
      "loss": 2.6155,
      "step": 777
    },
    {
      "epoch": 0.15150039919770997,
      "grad_norm": 1.3515625,
      "learning_rate": 3.796475411591995e-05,
      "loss": 2.6195,
      "step": 778
    },
    {
      "epoch": 0.1516951297879384,
      "grad_norm": 1.2265625,
      "learning_rate": 3.7959330637070545e-05,
      "loss": 2.6241,
      "step": 779
    },
    {
      "epoch": 0.15188986037816682,
      "grad_norm": 1.1640625,
      "learning_rate": 3.795390033009556e-05,
      "loss": 2.6329,
      "step": 780
    },
    {
      "epoch": 0.15208459096839522,
      "grad_norm": 1.046875,
      "learning_rate": 3.794846319705957e-05,
      "loss": 2.6221,
      "step": 781
    },
    {
      "epoch": 0.15227932155862364,
      "grad_norm": 1.2109375,
      "learning_rate": 3.794301924002977e-05,
      "loss": 2.6052,
      "step": 782
    },
    {
      "epoch": 0.15247405214885207,
      "grad_norm": 1.1015625,
      "learning_rate": 3.793756846107596e-05,
      "loss": 2.6014,
      "step": 783
    },
    {
      "epoch": 0.1526687827390805,
      "grad_norm": 1.3671875,
      "learning_rate": 3.793211086227052e-05,
      "loss": 2.6186,
      "step": 784
    },
    {
      "epoch": 0.1528635133293089,
      "grad_norm": 0.9296875,
      "learning_rate": 3.792664644568842e-05,
      "loss": 2.6166,
      "step": 785
    },
    {
      "epoch": 0.15305824391953732,
      "grad_norm": 1.5625,
      "learning_rate": 3.792117521340723e-05,
      "loss": 2.5918,
      "step": 786
    },
    {
      "epoch": 0.15325297450976574,
      "grad_norm": 1.328125,
      "learning_rate": 3.791569716750711e-05,
      "loss": 2.6242,
      "step": 787
    },
    {
      "epoch": 0.15344770509999417,
      "grad_norm": 1.1640625,
      "learning_rate": 3.791021231007082e-05,
      "loss": 2.6234,
      "step": 788
    },
    {
      "epoch": 0.15364243569022257,
      "grad_norm": 1.4453125,
      "learning_rate": 3.790472064318367e-05,
      "loss": 2.6173,
      "step": 789
    },
    {
      "epoch": 0.153837166280451,
      "grad_norm": 0.8828125,
      "learning_rate": 3.789922216893362e-05,
      "loss": 2.5962,
      "step": 790
    },
    {
      "epoch": 0.15403189687067942,
      "grad_norm": 1.390625,
      "learning_rate": 3.789371688941117e-05,
      "loss": 2.5859,
      "step": 791
    },
    {
      "epoch": 0.15422662746090784,
      "grad_norm": 0.90625,
      "learning_rate": 3.788820480670942e-05,
      "loss": 2.6311,
      "step": 792
    },
    {
      "epoch": 0.15442135805113624,
      "grad_norm": 1.3671875,
      "learning_rate": 3.788268592292407e-05,
      "loss": 2.6127,
      "step": 793
    },
    {
      "epoch": 0.15461608864136467,
      "grad_norm": 1.171875,
      "learning_rate": 3.787716024015339e-05,
      "loss": 2.6194,
      "step": 794
    },
    {
      "epoch": 0.1548108192315931,
      "grad_norm": 0.9765625,
      "learning_rate": 3.787162776049825e-05,
      "loss": 2.625,
      "step": 795
    },
    {
      "epoch": 0.15500554982182152,
      "grad_norm": 1.265625,
      "learning_rate": 3.786608848606208e-05,
      "loss": 2.6314,
      "step": 796
    },
    {
      "epoch": 0.15520028041204992,
      "grad_norm": 0.95703125,
      "learning_rate": 3.786054241895092e-05,
      "loss": 2.6172,
      "step": 797
    },
    {
      "epoch": 0.15539501100227834,
      "grad_norm": 1.3828125,
      "learning_rate": 3.785498956127338e-05,
      "loss": 2.6323,
      "step": 798
    },
    {
      "epoch": 0.15558974159250677,
      "grad_norm": 0.94140625,
      "learning_rate": 3.784942991514065e-05,
      "loss": 2.6157,
      "step": 799
    },
    {
      "epoch": 0.1557844721827352,
      "grad_norm": 1.3125,
      "learning_rate": 3.78438634826665e-05,
      "loss": 2.6031,
      "step": 800
    },
    {
      "epoch": 0.1559792027729636,
      "grad_norm": 1.171875,
      "learning_rate": 3.7838290265967294e-05,
      "loss": 2.6035,
      "step": 801
    },
    {
      "epoch": 0.15617393336319202,
      "grad_norm": 1.2265625,
      "learning_rate": 3.783271026716196e-05,
      "loss": 2.6212,
      "step": 802
    },
    {
      "epoch": 0.15636866395342044,
      "grad_norm": 1.3671875,
      "learning_rate": 3.7827123488372e-05,
      "loss": 2.6063,
      "step": 803
    },
    {
      "epoch": 0.15656339454364887,
      "grad_norm": 1.0703125,
      "learning_rate": 3.782152993172153e-05,
      "loss": 2.6018,
      "step": 804
    },
    {
      "epoch": 0.15675812513387727,
      "grad_norm": 1.96875,
      "learning_rate": 3.781592959933719e-05,
      "loss": 2.5956,
      "step": 805
    },
    {
      "epoch": 0.1569528557241057,
      "grad_norm": 1.359375,
      "learning_rate": 3.781032249334823e-05,
      "loss": 2.5966,
      "step": 806
    },
    {
      "epoch": 0.15714758631433412,
      "grad_norm": 2.40625,
      "learning_rate": 3.7804708615886475e-05,
      "loss": 2.6398,
      "step": 807
    },
    {
      "epoch": 0.15734231690456255,
      "grad_norm": 1.953125,
      "learning_rate": 3.77990879690863e-05,
      "loss": 2.6086,
      "step": 808
    },
    {
      "epoch": 0.15753704749479094,
      "grad_norm": 2.28125,
      "learning_rate": 3.779346055508469e-05,
      "loss": 2.6323,
      "step": 809
    },
    {
      "epoch": 0.15773177808501937,
      "grad_norm": 2.140625,
      "learning_rate": 3.778782637602119e-05,
      "loss": 2.6206,
      "step": 810
    },
    {
      "epoch": 0.1579265086752478,
      "grad_norm": 1.6953125,
      "learning_rate": 3.778218543403789e-05,
      "loss": 2.6005,
      "step": 811
    },
    {
      "epoch": 0.15812123926547622,
      "grad_norm": 1.59375,
      "learning_rate": 3.7776537731279486e-05,
      "loss": 2.6496,
      "step": 812
    },
    {
      "epoch": 0.15831596985570462,
      "grad_norm": 1.84375,
      "learning_rate": 3.777088326989322e-05,
      "loss": 2.602,
      "step": 813
    },
    {
      "epoch": 0.15851070044593304,
      "grad_norm": 1.484375,
      "learning_rate": 3.776522205202892e-05,
      "loss": 2.6059,
      "step": 814
    },
    {
      "epoch": 0.15870543103616147,
      "grad_norm": 2.203125,
      "learning_rate": 3.7759554079838985e-05,
      "loss": 2.6009,
      "step": 815
    },
    {
      "epoch": 0.1589001616263899,
      "grad_norm": 2.0625,
      "learning_rate": 3.775387935547836e-05,
      "loss": 2.6057,
      "step": 816
    },
    {
      "epoch": 0.15909489221661832,
      "grad_norm": 1.828125,
      "learning_rate": 3.774819788110459e-05,
      "loss": 2.603,
      "step": 817
    },
    {
      "epoch": 0.15928962280684672,
      "grad_norm": 1.78125,
      "learning_rate": 3.774250965887775e-05,
      "loss": 2.5996,
      "step": 818
    },
    {
      "epoch": 0.15948435339707515,
      "grad_norm": 1.5390625,
      "learning_rate": 3.77368146909605e-05,
      "loss": 2.5989,
      "step": 819
    },
    {
      "epoch": 0.15967908398730357,
      "grad_norm": 1.5,
      "learning_rate": 3.773111297951808e-05,
      "loss": 2.5956,
      "step": 820
    },
    {
      "epoch": 0.159873814577532,
      "grad_norm": 1.828125,
      "learning_rate": 3.772540452671825e-05,
      "loss": 2.6217,
      "step": 821
    },
    {
      "epoch": 0.1600685451677604,
      "grad_norm": 1.4765625,
      "learning_rate": 3.771968933473138e-05,
      "loss": 2.6071,
      "step": 822
    },
    {
      "epoch": 0.16026327575798882,
      "grad_norm": 2.03125,
      "learning_rate": 3.771396740573038e-05,
      "loss": 2.6055,
      "step": 823
    },
    {
      "epoch": 0.16045800634821725,
      "grad_norm": 1.8828125,
      "learning_rate": 3.7708238741890724e-05,
      "loss": 2.6055,
      "step": 824
    },
    {
      "epoch": 0.16065273693844567,
      "grad_norm": 1.59375,
      "learning_rate": 3.770250334539044e-05,
      "loss": 2.5986,
      "step": 825
    },
    {
      "epoch": 0.16084746752867407,
      "grad_norm": 1.40625,
      "learning_rate": 3.769676121841012e-05,
      "loss": 2.6104,
      "step": 826
    },
    {
      "epoch": 0.1610421981189025,
      "grad_norm": 1.90625,
      "learning_rate": 3.7691012363132926e-05,
      "loss": 2.5941,
      "step": 827
    },
    {
      "epoch": 0.16123692870913092,
      "grad_norm": 1.6953125,
      "learning_rate": 3.7685256781744565e-05,
      "loss": 2.6057,
      "step": 828
    },
    {
      "epoch": 0.16143165929935935,
      "grad_norm": 1.921875,
      "learning_rate": 3.7679494476433306e-05,
      "loss": 2.5887,
      "step": 829
    },
    {
      "epoch": 0.16162638988958775,
      "grad_norm": 1.859375,
      "learning_rate": 3.767372544938996e-05,
      "loss": 2.5898,
      "step": 830
    },
    {
      "epoch": 0.16182112047981617,
      "grad_norm": 1.421875,
      "learning_rate": 3.7667949702807935e-05,
      "loss": 2.5778,
      "step": 831
    },
    {
      "epoch": 0.1620158510700446,
      "grad_norm": 1.2890625,
      "learning_rate": 3.766216723888315e-05,
      "loss": 2.6034,
      "step": 832
    },
    {
      "epoch": 0.16221058166027302,
      "grad_norm": 1.9765625,
      "learning_rate": 3.7656378059814086e-05,
      "loss": 2.6007,
      "step": 833
    },
    {
      "epoch": 0.16240531225050142,
      "grad_norm": 1.75,
      "learning_rate": 3.76505821678018e-05,
      "loss": 2.6121,
      "step": 834
    },
    {
      "epoch": 0.16260004284072985,
      "grad_norm": 1.8984375,
      "learning_rate": 3.764477956504988e-05,
      "loss": 2.5991,
      "step": 835
    },
    {
      "epoch": 0.16279477343095827,
      "grad_norm": 1.8671875,
      "learning_rate": 3.763897025376447e-05,
      "loss": 2.5936,
      "step": 836
    },
    {
      "epoch": 0.1629895040211867,
      "grad_norm": 1.3125,
      "learning_rate": 3.763315423615427e-05,
      "loss": 2.5897,
      "step": 837
    },
    {
      "epoch": 0.1631842346114151,
      "grad_norm": 1.203125,
      "learning_rate": 3.762733151443052e-05,
      "loss": 2.5903,
      "step": 838
    },
    {
      "epoch": 0.16337896520164352,
      "grad_norm": 1.890625,
      "learning_rate": 3.762150209080702e-05,
      "loss": 2.6236,
      "step": 839
    },
    {
      "epoch": 0.16357369579187195,
      "grad_norm": 1.5234375,
      "learning_rate": 3.761566596750011e-05,
      "loss": 2.5881,
      "step": 840
    },
    {
      "epoch": 0.16376842638210037,
      "grad_norm": 2.15625,
      "learning_rate": 3.760982314672869e-05,
      "loss": 2.5709,
      "step": 841
    },
    {
      "epoch": 0.16396315697232877,
      "grad_norm": 2.078125,
      "learning_rate": 3.7603973630714184e-05,
      "loss": 2.581,
      "step": 842
    },
    {
      "epoch": 0.1641578875625572,
      "grad_norm": 1.171875,
      "learning_rate": 3.759811742168058e-05,
      "loss": 2.5867,
      "step": 843
    },
    {
      "epoch": 0.16435261815278562,
      "grad_norm": 1.2421875,
      "learning_rate": 3.75922545218544e-05,
      "loss": 2.5865,
      "step": 844
    },
    {
      "epoch": 0.16454734874301405,
      "grad_norm": 1.84375,
      "learning_rate": 3.758638493346472e-05,
      "loss": 2.5918,
      "step": 845
    },
    {
      "epoch": 0.16474207933324245,
      "grad_norm": 1.515625,
      "learning_rate": 3.758050865874315e-05,
      "loss": 2.5979,
      "step": 846
    },
    {
      "epoch": 0.16493680992347087,
      "grad_norm": 2.125,
      "learning_rate": 3.7574625699923856e-05,
      "loss": 2.6103,
      "step": 847
    },
    {
      "epoch": 0.1651315405136993,
      "grad_norm": 2.03125,
      "learning_rate": 3.756873605924353e-05,
      "loss": 2.5925,
      "step": 848
    },
    {
      "epoch": 0.16532627110392772,
      "grad_norm": 1.203125,
      "learning_rate": 3.756283973894139e-05,
      "loss": 2.5903,
      "step": 849
    },
    {
      "epoch": 0.16552100169415612,
      "grad_norm": 1.171875,
      "learning_rate": 3.755693674125924e-05,
      "loss": 2.5953,
      "step": 850
    },
    {
      "epoch": 0.16571573228438455,
      "grad_norm": 1.703125,
      "learning_rate": 3.755102706844138e-05,
      "loss": 2.5923,
      "step": 851
    },
    {
      "epoch": 0.16591046287461297,
      "grad_norm": 1.2734375,
      "learning_rate": 3.754511072273467e-05,
      "loss": 2.5877,
      "step": 852
    },
    {
      "epoch": 0.1661051934648414,
      "grad_norm": 2.5625,
      "learning_rate": 3.7539187706388506e-05,
      "loss": 2.5853,
      "step": 853
    },
    {
      "epoch": 0.1662999240550698,
      "grad_norm": 2.5,
      "learning_rate": 3.7533258021654804e-05,
      "loss": 2.6113,
      "step": 854
    },
    {
      "epoch": 0.16649465464529822,
      "grad_norm": 0.8515625,
      "learning_rate": 3.7527321670788023e-05,
      "loss": 2.5966,
      "step": 855
    },
    {
      "epoch": 0.16668938523552665,
      "grad_norm": 1.734375,
      "learning_rate": 3.7521378656045174e-05,
      "loss": 2.5832,
      "step": 856
    },
    {
      "epoch": 0.16688411582575507,
      "grad_norm": 0.84765625,
      "learning_rate": 3.751542897968579e-05,
      "loss": 2.587,
      "step": 857
    },
    {
      "epoch": 0.16707884641598347,
      "grad_norm": 1.734375,
      "learning_rate": 3.750947264397191e-05,
      "loss": 2.6037,
      "step": 858
    },
    {
      "epoch": 0.1672735770062119,
      "grad_norm": 1.1015625,
      "learning_rate": 3.7503509651168154e-05,
      "loss": 2.5926,
      "step": 859
    },
    {
      "epoch": 0.16746830759644032,
      "grad_norm": 2.359375,
      "learning_rate": 3.749754000354163e-05,
      "loss": 2.6014,
      "step": 860
    },
    {
      "epoch": 0.16766303818666875,
      "grad_norm": 2.046875,
      "learning_rate": 3.7491563703362e-05,
      "loss": 2.6108,
      "step": 861
    },
    {
      "epoch": 0.16785776877689718,
      "grad_norm": 1.6484375,
      "learning_rate": 3.7485580752901455e-05,
      "loss": 2.5987,
      "step": 862
    },
    {
      "epoch": 0.16805249936712557,
      "grad_norm": 1.6171875,
      "learning_rate": 3.747959115443471e-05,
      "loss": 2.6041,
      "step": 863
    },
    {
      "epoch": 0.168247229957354,
      "grad_norm": 1.3125,
      "learning_rate": 3.747359491023899e-05,
      "loss": 2.6048,
      "step": 864
    },
    {
      "epoch": 0.16844196054758243,
      "grad_norm": 1.2890625,
      "learning_rate": 3.746759202259408e-05,
      "loss": 2.5816,
      "step": 865
    },
    {
      "epoch": 0.16863669113781085,
      "grad_norm": 1.359375,
      "learning_rate": 3.7461582493782266e-05,
      "loss": 2.5977,
      "step": 866
    },
    {
      "epoch": 0.16883142172803925,
      "grad_norm": 1.0625,
      "learning_rate": 3.7455566326088374e-05,
      "loss": 2.5649,
      "step": 867
    },
    {
      "epoch": 0.16902615231826768,
      "grad_norm": 1.1875,
      "learning_rate": 3.744954352179973e-05,
      "loss": 2.5808,
      "step": 868
    },
    {
      "epoch": 0.1692208829084961,
      "grad_norm": 1.0859375,
      "learning_rate": 3.744351408320622e-05,
      "loss": 2.5943,
      "step": 869
    },
    {
      "epoch": 0.16941561349872453,
      "grad_norm": 1.015625,
      "learning_rate": 3.743747801260023e-05,
      "loss": 2.5988,
      "step": 870
    },
    {
      "epoch": 0.16961034408895292,
      "grad_norm": 1.2890625,
      "learning_rate": 3.743143531227666e-05,
      "loss": 2.6178,
      "step": 871
    },
    {
      "epoch": 0.16980507467918135,
      "grad_norm": 0.9140625,
      "learning_rate": 3.742538598453294e-05,
      "loss": 2.5826,
      "step": 872
    },
    {
      "epoch": 0.16999980526940978,
      "grad_norm": 1.28125,
      "learning_rate": 3.741933003166903e-05,
      "loss": 2.5879,
      "step": 873
    },
    {
      "epoch": 0.1701945358596382,
      "grad_norm": 0.98046875,
      "learning_rate": 3.741326745598739e-05,
      "loss": 2.591,
      "step": 874
    },
    {
      "epoch": 0.1703892664498666,
      "grad_norm": 1.296875,
      "learning_rate": 3.740719825979302e-05,
      "loss": 2.5926,
      "step": 875
    },
    {
      "epoch": 0.17058399704009503,
      "grad_norm": 1.1015625,
      "learning_rate": 3.740112244539341e-05,
      "loss": 2.6006,
      "step": 876
    },
    {
      "epoch": 0.17077872763032345,
      "grad_norm": 1.234375,
      "learning_rate": 3.739504001509859e-05,
      "loss": 2.5969,
      "step": 877
    },
    {
      "epoch": 0.17097345822055188,
      "grad_norm": 1.1484375,
      "learning_rate": 3.738895097122109e-05,
      "loss": 2.5747,
      "step": 878
    },
    {
      "epoch": 0.17116818881078028,
      "grad_norm": 1.3515625,
      "learning_rate": 3.738285531607597e-05,
      "loss": 2.5874,
      "step": 879
    },
    {
      "epoch": 0.1713629194010087,
      "grad_norm": 1.0546875,
      "learning_rate": 3.737675305198078e-05,
      "loss": 2.5885,
      "step": 880
    },
    {
      "epoch": 0.17155764999123713,
      "grad_norm": 1.6484375,
      "learning_rate": 3.7370644181255616e-05,
      "loss": 2.599,
      "step": 881
    },
    {
      "epoch": 0.17175238058146555,
      "grad_norm": 1.171875,
      "learning_rate": 3.7364528706223045e-05,
      "loss": 2.5876,
      "step": 882
    },
    {
      "epoch": 0.17194711117169395,
      "grad_norm": 1.796875,
      "learning_rate": 3.735840662920818e-05,
      "loss": 2.6105,
      "step": 883
    },
    {
      "epoch": 0.17214184176192238,
      "grad_norm": 1.4921875,
      "learning_rate": 3.735227795253862e-05,
      "loss": 2.5754,
      "step": 884
    },
    {
      "epoch": 0.1723365723521508,
      "grad_norm": 1.65625,
      "learning_rate": 3.7346142678544494e-05,
      "loss": 2.6086,
      "step": 885
    },
    {
      "epoch": 0.17253130294237923,
      "grad_norm": 1.421875,
      "learning_rate": 3.734000080955843e-05,
      "loss": 2.5988,
      "step": 886
    },
    {
      "epoch": 0.17272603353260763,
      "grad_norm": 1.6484375,
      "learning_rate": 3.733385234791556e-05,
      "loss": 2.5978,
      "step": 887
    },
    {
      "epoch": 0.17292076412283605,
      "grad_norm": 1.21875,
      "learning_rate": 3.732769729595352e-05,
      "loss": 2.6004,
      "step": 888
    },
    {
      "epoch": 0.17311549471306448,
      "grad_norm": 2.03125,
      "learning_rate": 3.7321535656012456e-05,
      "loss": 2.5987,
      "step": 889
    },
    {
      "epoch": 0.1733102253032929,
      "grad_norm": 1.6328125,
      "learning_rate": 3.731536743043503e-05,
      "loss": 2.61,
      "step": 890
    },
    {
      "epoch": 0.1735049558935213,
      "grad_norm": 1.9453125,
      "learning_rate": 3.7309192621566394e-05,
      "loss": 2.5962,
      "step": 891
    },
    {
      "epoch": 0.17369968648374973,
      "grad_norm": 1.8984375,
      "learning_rate": 3.730301123175419e-05,
      "loss": 2.6114,
      "step": 892
    },
    {
      "epoch": 0.17389441707397815,
      "grad_norm": 1.3984375,
      "learning_rate": 3.72968232633486e-05,
      "loss": 2.5827,
      "step": 893
    },
    {
      "epoch": 0.17408914766420658,
      "grad_norm": 1.3125,
      "learning_rate": 3.729062871870228e-05,
      "loss": 2.5964,
      "step": 894
    },
    {
      "epoch": 0.17428387825443498,
      "grad_norm": 1.5078125,
      "learning_rate": 3.7284427600170384e-05,
      "loss": 2.579,
      "step": 895
    },
    {
      "epoch": 0.1744786088446634,
      "grad_norm": 1.21875,
      "learning_rate": 3.727821991011058e-05,
      "loss": 2.597,
      "step": 896
    },
    {
      "epoch": 0.17467333943489183,
      "grad_norm": 2.125,
      "learning_rate": 3.7272005650883026e-05,
      "loss": 2.5936,
      "step": 897
    },
    {
      "epoch": 0.17486807002512025,
      "grad_norm": 1.859375,
      "learning_rate": 3.7265784824850375e-05,
      "loss": 2.5859,
      "step": 898
    },
    {
      "epoch": 0.17506280061534865,
      "grad_norm": 1.6015625,
      "learning_rate": 3.725955743437779e-05,
      "loss": 2.5946,
      "step": 899
    },
    {
      "epoch": 0.17525753120557708,
      "grad_norm": 1.578125,
      "learning_rate": 3.725332348183291e-05,
      "loss": 2.5893,
      "step": 900
    },
    {
      "epoch": 0.1754522617958055,
      "grad_norm": 1.390625,
      "learning_rate": 3.7247082969585884e-05,
      "loss": 2.5967,
      "step": 901
    },
    {
      "epoch": 0.17564699238603393,
      "grad_norm": 1.2734375,
      "learning_rate": 3.7240835900009356e-05,
      "loss": 2.5908,
      "step": 902
    },
    {
      "epoch": 0.17584172297626235,
      "grad_norm": 1.5234375,
      "learning_rate": 3.7234582275478454e-05,
      "loss": 2.5986,
      "step": 903
    },
    {
      "epoch": 0.17603645356649075,
      "grad_norm": 1.296875,
      "learning_rate": 3.72283220983708e-05,
      "loss": 2.5938,
      "step": 904
    },
    {
      "epoch": 0.17623118415671918,
      "grad_norm": 1.7734375,
      "learning_rate": 3.722205537106651e-05,
      "loss": 2.5849,
      "step": 905
    },
    {
      "epoch": 0.1764259147469476,
      "grad_norm": 1.3828125,
      "learning_rate": 3.7215782095948194e-05,
      "loss": 2.5781,
      "step": 906
    },
    {
      "epoch": 0.17662064533717603,
      "grad_norm": 2.109375,
      "learning_rate": 3.7209502275400934e-05,
      "loss": 2.5674,
      "step": 907
    },
    {
      "epoch": 0.17681537592740443,
      "grad_norm": 1.8828125,
      "learning_rate": 3.7203215911812324e-05,
      "loss": 2.5952,
      "step": 908
    },
    {
      "epoch": 0.17701010651763285,
      "grad_norm": 1.8046875,
      "learning_rate": 3.719692300757244e-05,
      "loss": 2.5735,
      "step": 909
    },
    {
      "epoch": 0.17720483710786128,
      "grad_norm": 1.8125,
      "learning_rate": 3.719062356507383e-05,
      "loss": 2.582,
      "step": 910
    },
    {
      "epoch": 0.1773995676980897,
      "grad_norm": 1.5703125,
      "learning_rate": 3.718431758671154e-05,
      "loss": 2.5931,
      "step": 911
    },
    {
      "epoch": 0.1775942982883181,
      "grad_norm": 1.53125,
      "learning_rate": 3.717800507488311e-05,
      "loss": 2.5936,
      "step": 912
    },
    {
      "epoch": 0.17778902887854653,
      "grad_norm": 1.7578125,
      "learning_rate": 3.7171686031988535e-05,
      "loss": 2.5788,
      "step": 913
    },
    {
      "epoch": 0.17798375946877495,
      "grad_norm": 1.4609375,
      "learning_rate": 3.716536046043032e-05,
      "loss": 2.5871,
      "step": 914
    },
    {
      "epoch": 0.17817849005900338,
      "grad_norm": 1.9921875,
      "learning_rate": 3.715902836261344e-05,
      "loss": 2.5852,
      "step": 915
    },
    {
      "epoch": 0.17837322064923178,
      "grad_norm": 1.8515625,
      "learning_rate": 3.715268974094536e-05,
      "loss": 2.6022,
      "step": 916
    },
    {
      "epoch": 0.1785679512394602,
      "grad_norm": 1.3828125,
      "learning_rate": 3.714634459783602e-05,
      "loss": 2.572,
      "step": 917
    },
    {
      "epoch": 0.17876268182968863,
      "grad_norm": 1.484375,
      "learning_rate": 3.7139992935697834e-05,
      "loss": 2.577,
      "step": 918
    },
    {
      "epoch": 0.17895741241991706,
      "grad_norm": 1.1171875,
      "learning_rate": 3.71336347569457e-05,
      "loss": 2.5989,
      "step": 919
    },
    {
      "epoch": 0.17915214301014545,
      "grad_norm": 1.1484375,
      "learning_rate": 3.7127270063996996e-05,
      "loss": 2.5796,
      "step": 920
    },
    {
      "epoch": 0.17934687360037388,
      "grad_norm": 1.3671875,
      "learning_rate": 3.712089885927157e-05,
      "loss": 2.5922,
      "step": 921
    },
    {
      "epoch": 0.1795416041906023,
      "grad_norm": 1.0078125,
      "learning_rate": 3.711452114519175e-05,
      "loss": 2.5922,
      "step": 922
    },
    {
      "epoch": 0.17973633478083073,
      "grad_norm": 1.8671875,
      "learning_rate": 3.710813692418235e-05,
      "loss": 2.5812,
      "step": 923
    },
    {
      "epoch": 0.17993106537105913,
      "grad_norm": 1.4453125,
      "learning_rate": 3.710174619867063e-05,
      "loss": 2.5583,
      "step": 924
    },
    {
      "epoch": 0.18012579596128755,
      "grad_norm": 1.9296875,
      "learning_rate": 3.709534897108635e-05,
      "loss": 2.5884,
      "step": 925
    },
    {
      "epoch": 0.18032052655151598,
      "grad_norm": 1.7421875,
      "learning_rate": 3.708894524386174e-05,
      "loss": 2.5698,
      "step": 926
    },
    {
      "epoch": 0.1805152571417444,
      "grad_norm": 1.4140625,
      "learning_rate": 3.708253501943147e-05,
      "loss": 2.5647,
      "step": 927
    },
    {
      "epoch": 0.1807099877319728,
      "grad_norm": 1.34375,
      "learning_rate": 3.707611830023272e-05,
      "loss": 2.5576,
      "step": 928
    },
    {
      "epoch": 0.18090471832220123,
      "grad_norm": 1.46875,
      "learning_rate": 3.7069695088705114e-05,
      "loss": 2.5785,
      "step": 929
    },
    {
      "epoch": 0.18109944891242966,
      "grad_norm": 1.125,
      "learning_rate": 3.706326538729076e-05,
      "loss": 2.5638,
      "step": 930
    },
    {
      "epoch": 0.18129417950265808,
      "grad_norm": 2.078125,
      "learning_rate": 3.705682919843422e-05,
      "loss": 2.5727,
      "step": 931
    },
    {
      "epoch": 0.18148891009288648,
      "grad_norm": 1.7890625,
      "learning_rate": 3.705038652458253e-05,
      "loss": 2.5993,
      "step": 932
    },
    {
      "epoch": 0.1816836406831149,
      "grad_norm": 1.6796875,
      "learning_rate": 3.704393736818519e-05,
      "loss": 2.5928,
      "step": 933
    },
    {
      "epoch": 0.18187837127334333,
      "grad_norm": 1.640625,
      "learning_rate": 3.703748173169417e-05,
      "loss": 2.5917,
      "step": 934
    },
    {
      "epoch": 0.18207310186357176,
      "grad_norm": 1.4296875,
      "learning_rate": 3.70310196175639e-05,
      "loss": 2.5711,
      "step": 935
    },
    {
      "epoch": 0.18226783245380015,
      "grad_norm": 1.1953125,
      "learning_rate": 3.7024551028251256e-05,
      "loss": 2.5911,
      "step": 936
    },
    {
      "epoch": 0.18246256304402858,
      "grad_norm": 1.671875,
      "learning_rate": 3.701807596621561e-05,
      "loss": 2.5751,
      "step": 937
    },
    {
      "epoch": 0.182657293634257,
      "grad_norm": 1.28125,
      "learning_rate": 3.701159443391876e-05,
      "loss": 2.5744,
      "step": 938
    },
    {
      "epoch": 0.18285202422448543,
      "grad_norm": 2.09375,
      "learning_rate": 3.7005106433825e-05,
      "loss": 2.5894,
      "step": 939
    },
    {
      "epoch": 0.18304675481471383,
      "grad_norm": 1.859375,
      "learning_rate": 3.6998611968401044e-05,
      "loss": 2.5525,
      "step": 940
    },
    {
      "epoch": 0.18324148540494226,
      "grad_norm": 1.5625,
      "learning_rate": 3.6992111040116095e-05,
      "loss": 2.5748,
      "step": 941
    },
    {
      "epoch": 0.18343621599517068,
      "grad_norm": 1.5234375,
      "learning_rate": 3.69856036514418e-05,
      "loss": 2.5566,
      "step": 942
    },
    {
      "epoch": 0.1836309465853991,
      "grad_norm": 1.4609375,
      "learning_rate": 3.6979089804852254e-05,
      "loss": 2.5643,
      "step": 943
    },
    {
      "epoch": 0.18382567717562753,
      "grad_norm": 1.3359375,
      "learning_rate": 3.6972569502824034e-05,
      "loss": 2.5686,
      "step": 944
    },
    {
      "epoch": 0.18402040776585593,
      "grad_norm": 1.671875,
      "learning_rate": 3.696604274783614e-05,
      "loss": 2.5501,
      "step": 945
    },
    {
      "epoch": 0.18421513835608436,
      "grad_norm": 1.40625,
      "learning_rate": 3.6959509542370045e-05,
      "loss": 2.6066,
      "step": 946
    },
    {
      "epoch": 0.18440986894631278,
      "grad_norm": 1.7734375,
      "learning_rate": 3.695296988890968e-05,
      "loss": 2.5824,
      "step": 947
    },
    {
      "epoch": 0.1846045995365412,
      "grad_norm": 1.5703125,
      "learning_rate": 3.6946423789941394e-05,
      "loss": 2.5888,
      "step": 948
    },
    {
      "epoch": 0.1847993301267696,
      "grad_norm": 1.625,
      "learning_rate": 3.6939871247954024e-05,
      "loss": 2.5632,
      "step": 949
    },
    {
      "epoch": 0.18499406071699803,
      "grad_norm": 1.3359375,
      "learning_rate": 3.693331226543885e-05,
      "loss": 2.5799,
      "step": 950
    },
    {
      "epoch": 0.18518879130722646,
      "grad_norm": 1.7421875,
      "learning_rate": 3.6926746844889574e-05,
      "loss": 2.5812,
      "step": 951
    },
    {
      "epoch": 0.18538352189745488,
      "grad_norm": 1.4453125,
      "learning_rate": 3.692017498880238e-05,
      "loss": 2.5825,
      "step": 952
    },
    {
      "epoch": 0.18557825248768328,
      "grad_norm": 1.8515625,
      "learning_rate": 3.6913596699675866e-05,
      "loss": 2.5754,
      "step": 953
    },
    {
      "epoch": 0.1857729830779117,
      "grad_norm": 1.65625,
      "learning_rate": 3.690701198001112e-05,
      "loss": 2.5989,
      "step": 954
    },
    {
      "epoch": 0.18596771366814013,
      "grad_norm": 1.6171875,
      "learning_rate": 3.690042083231162e-05,
      "loss": 2.5748,
      "step": 955
    },
    {
      "epoch": 0.18616244425836856,
      "grad_norm": 1.3671875,
      "learning_rate": 3.6893823259083335e-05,
      "loss": 2.5648,
      "step": 956
    },
    {
      "epoch": 0.18635717484859696,
      "grad_norm": 1.7890625,
      "learning_rate": 3.688721926283465e-05,
      "loss": 2.5634,
      "step": 957
    },
    {
      "epoch": 0.18655190543882538,
      "grad_norm": 1.4140625,
      "learning_rate": 3.68806088460764e-05,
      "loss": 2.5693,
      "step": 958
    },
    {
      "epoch": 0.1867466360290538,
      "grad_norm": 2.078125,
      "learning_rate": 3.687399201132186e-05,
      "loss": 2.5696,
      "step": 959
    },
    {
      "epoch": 0.18694136661928223,
      "grad_norm": 1.984375,
      "learning_rate": 3.6867368761086755e-05,
      "loss": 2.5833,
      "step": 960
    },
    {
      "epoch": 0.18713609720951063,
      "grad_norm": 1.484375,
      "learning_rate": 3.686073909788923e-05,
      "loss": 2.5639,
      "step": 961
    },
    {
      "epoch": 0.18733082779973906,
      "grad_norm": 1.421875,
      "learning_rate": 3.685410302424988e-05,
      "loss": 2.5853,
      "step": 962
    },
    {
      "epoch": 0.18752555838996748,
      "grad_norm": 1.7421875,
      "learning_rate": 3.684746054269174e-05,
      "loss": 2.5727,
      "step": 963
    },
    {
      "epoch": 0.1877202889801959,
      "grad_norm": 1.4375,
      "learning_rate": 3.684081165574028e-05,
      "loss": 2.5907,
      "step": 964
    },
    {
      "epoch": 0.1879150195704243,
      "grad_norm": 2.234375,
      "learning_rate": 3.683415636592339e-05,
      "loss": 2.5791,
      "step": 965
    },
    {
      "epoch": 0.18810975016065273,
      "grad_norm": 2.140625,
      "learning_rate": 3.682749467577142e-05,
      "loss": 2.5647,
      "step": 966
    },
    {
      "epoch": 0.18830448075088116,
      "grad_norm": 1.1171875,
      "learning_rate": 3.682082658781713e-05,
      "loss": 2.5944,
      "step": 967
    },
    {
      "epoch": 0.18849921134110958,
      "grad_norm": 1.2421875,
      "learning_rate": 3.681415210459573e-05,
      "loss": 2.5656,
      "step": 968
    },
    {
      "epoch": 0.18869394193133798,
      "grad_norm": 1.7265625,
      "learning_rate": 3.680747122864485e-05,
      "loss": 2.5786,
      "step": 969
    },
    {
      "epoch": 0.1888886725215664,
      "grad_norm": 1.3046875,
      "learning_rate": 3.680078396250456e-05,
      "loss": 2.5806,
      "step": 970
    },
    {
      "epoch": 0.18908340311179483,
      "grad_norm": 2.5,
      "learning_rate": 3.679409030871735e-05,
      "loss": 2.5476,
      "step": 971
    },
    {
      "epoch": 0.18927813370202326,
      "grad_norm": 2.5,
      "learning_rate": 3.678739026982814e-05,
      "loss": 2.5604,
      "step": 972
    },
    {
      "epoch": 0.18947286429225166,
      "grad_norm": 0.96875,
      "learning_rate": 3.678068384838428e-05,
      "loss": 2.575,
      "step": 973
    },
    {
      "epoch": 0.18966759488248008,
      "grad_norm": 1.65625,
      "learning_rate": 3.677397104693556e-05,
      "loss": 2.5489,
      "step": 974
    },
    {
      "epoch": 0.1898623254727085,
      "grad_norm": 0.99609375,
      "learning_rate": 3.676725186803416e-05,
      "loss": 2.5664,
      "step": 975
    },
    {
      "epoch": 0.19005705606293694,
      "grad_norm": 1.453125,
      "learning_rate": 3.6760526314234735e-05,
      "loss": 2.5796,
      "step": 976
    },
    {
      "epoch": 0.19025178665316533,
      "grad_norm": 1.1328125,
      "learning_rate": 3.6753794388094305e-05,
      "loss": 2.5721,
      "step": 977
    },
    {
      "epoch": 0.19044651724339376,
      "grad_norm": 1.2421875,
      "learning_rate": 3.6747056092172376e-05,
      "loss": 2.5646,
      "step": 978
    },
    {
      "epoch": 0.19064124783362218,
      "grad_norm": 1.265625,
      "learning_rate": 3.674031142903082e-05,
      "loss": 2.5701,
      "step": 979
    },
    {
      "epoch": 0.1908359784238506,
      "grad_norm": 1.0703125,
      "learning_rate": 3.673356040123396e-05,
      "loss": 2.5769,
      "step": 980
    },
    {
      "epoch": 0.191030709014079,
      "grad_norm": 1.390625,
      "learning_rate": 3.672680301134853e-05,
      "loss": 2.5766,
      "step": 981
    },
    {
      "epoch": 0.19122543960430743,
      "grad_norm": 1.0703125,
      "learning_rate": 3.672003926194369e-05,
      "loss": 2.593,
      "step": 982
    },
    {
      "epoch": 0.19142017019453586,
      "grad_norm": 1.3203125,
      "learning_rate": 3.671326915559102e-05,
      "loss": 2.5453,
      "step": 983
    },
    {
      "epoch": 0.19161490078476429,
      "grad_norm": 1.0703125,
      "learning_rate": 3.67064926948645e-05,
      "loss": 2.562,
      "step": 984
    },
    {
      "epoch": 0.19180963137499268,
      "grad_norm": 1.421875,
      "learning_rate": 3.6699709882340535e-05,
      "loss": 2.5703,
      "step": 985
    },
    {
      "epoch": 0.1920043619652211,
      "grad_norm": 0.96875,
      "learning_rate": 3.669292072059794e-05,
      "loss": 2.5428,
      "step": 986
    },
    {
      "epoch": 0.19219909255544954,
      "grad_norm": 1.6953125,
      "learning_rate": 3.668612521221797e-05,
      "loss": 2.566,
      "step": 987
    },
    {
      "epoch": 0.19239382314567796,
      "grad_norm": 1.265625,
      "learning_rate": 3.6679323359784254e-05,
      "loss": 2.5726,
      "step": 988
    },
    {
      "epoch": 0.1925885537359064,
      "grad_norm": 1.9296875,
      "learning_rate": 3.6672515165882864e-05,
      "loss": 2.5719,
      "step": 989
    },
    {
      "epoch": 0.19278328432613479,
      "grad_norm": 1.6171875,
      "learning_rate": 3.6665700633102264e-05,
      "loss": 2.5755,
      "step": 990
    },
    {
      "epoch": 0.1929780149163632,
      "grad_norm": 1.7109375,
      "learning_rate": 3.665887976403333e-05,
      "loss": 2.5924,
      "step": 991
    },
    {
      "epoch": 0.19317274550659164,
      "grad_norm": 1.421875,
      "learning_rate": 3.6652052561269364e-05,
      "loss": 2.587,
      "step": 992
    },
    {
      "epoch": 0.19336747609682006,
      "grad_norm": 1.6796875,
      "learning_rate": 3.664521902740606e-05,
      "loss": 2.5606,
      "step": 993
    },
    {
      "epoch": 0.19356220668704846,
      "grad_norm": 1.390625,
      "learning_rate": 3.663837916504152e-05,
      "loss": 2.5867,
      "step": 994
    },
    {
      "epoch": 0.19375693727727689,
      "grad_norm": 1.734375,
      "learning_rate": 3.6631532976776254e-05,
      "loss": 2.5751,
      "step": 995
    },
    {
      "epoch": 0.1939516678675053,
      "grad_norm": 1.4296875,
      "learning_rate": 3.662468046521318e-05,
      "loss": 2.5722,
      "step": 996
    },
    {
      "epoch": 0.19414639845773374,
      "grad_norm": 1.9140625,
      "learning_rate": 3.6617821632957626e-05,
      "loss": 2.5813,
      "step": 997
    },
    {
      "epoch": 0.19434112904796214,
      "grad_norm": 1.59375,
      "learning_rate": 3.66109564826173e-05,
      "loss": 2.5675,
      "step": 998
    },
    {
      "epoch": 0.19453585963819056,
      "grad_norm": 1.9375,
      "learning_rate": 3.660408501680234e-05,
      "loss": 2.5711,
      "step": 999
    },
    {
      "epoch": 0.194730590228419,
      "grad_norm": 1.765625,
      "learning_rate": 3.659720723812527e-05,
      "loss": 2.5406,
      "step": 1000
    },
    {
      "epoch": 0.1949253208186474,
      "grad_norm": 1.6484375,
      "learning_rate": 3.6590323149201015e-05,
      "loss": 2.5652,
      "step": 1001
    },
    {
      "epoch": 0.1951200514088758,
      "grad_norm": 1.5234375,
      "learning_rate": 3.65834327526469e-05,
      "loss": 2.5701,
      "step": 1002
    },
    {
      "epoch": 0.19531478199910424,
      "grad_norm": 1.3515625,
      "learning_rate": 3.6576536051082656e-05,
      "loss": 2.5672,
      "step": 1003
    },
    {
      "epoch": 0.19550951258933266,
      "grad_norm": 1.265625,
      "learning_rate": 3.656963304713039e-05,
      "loss": 2.5627,
      "step": 1004
    },
    {
      "epoch": 0.1957042431795611,
      "grad_norm": 1.5078125,
      "learning_rate": 3.656272374341463e-05,
      "loss": 2.5577,
      "step": 1005
    },
    {
      "epoch": 0.1958989737697895,
      "grad_norm": 1.171875,
      "learning_rate": 3.6555808142562294e-05,
      "loss": 2.5666,
      "step": 1006
    },
    {
      "epoch": 0.1960937043600179,
      "grad_norm": 2.109375,
      "learning_rate": 3.654888624720268e-05,
      "loss": 2.5885,
      "step": 1007
    },
    {
      "epoch": 0.19628843495024634,
      "grad_norm": 1.75,
      "learning_rate": 3.6541958059967495e-05,
      "loss": 2.565,
      "step": 1008
    },
    {
      "epoch": 0.19648316554047476,
      "grad_norm": 1.859375,
      "learning_rate": 3.653502358349082e-05,
      "loss": 2.5759,
      "step": 1009
    },
    {
      "epoch": 0.19667789613070316,
      "grad_norm": 1.7265625,
      "learning_rate": 3.652808282040915e-05,
      "loss": 2.5707,
      "step": 1010
    },
    {
      "epoch": 0.1968726267209316,
      "grad_norm": 1.4296875,
      "learning_rate": 3.652113577336135e-05,
      "loss": 2.5651,
      "step": 1011
    },
    {
      "epoch": 0.19706735731116,
      "grad_norm": 1.4609375,
      "learning_rate": 3.6514182444988685e-05,
      "loss": 2.5589,
      "step": 1012
    },
    {
      "epoch": 0.19726208790138844,
      "grad_norm": 1.515625,
      "learning_rate": 3.6507222837934806e-05,
      "loss": 2.5647,
      "step": 1013
    },
    {
      "epoch": 0.19745681849161684,
      "grad_norm": 1.3515625,
      "learning_rate": 3.6500256954845754e-05,
      "loss": 2.5722,
      "step": 1014
    },
    {
      "epoch": 0.19765154908184526,
      "grad_norm": 1.7734375,
      "learning_rate": 3.649328479836996e-05,
      "loss": 2.5644,
      "step": 1015
    },
    {
      "epoch": 0.1978462796720737,
      "grad_norm": 1.6640625,
      "learning_rate": 3.648630637115821e-05,
      "loss": 2.5615,
      "step": 1016
    },
    {
      "epoch": 0.19804101026230211,
      "grad_norm": 1.453125,
      "learning_rate": 3.6479321675863715e-05,
      "loss": 2.5346,
      "step": 1017
    },
    {
      "epoch": 0.1982357408525305,
      "grad_norm": 1.40625,
      "learning_rate": 3.647233071514205e-05,
      "loss": 2.5652,
      "step": 1018
    },
    {
      "epoch": 0.19843047144275894,
      "grad_norm": 1.578125,
      "learning_rate": 3.6465333491651175e-05,
      "loss": 2.5512,
      "step": 1019
    },
    {
      "epoch": 0.19862520203298736,
      "grad_norm": 1.3828125,
      "learning_rate": 3.6458330008051426e-05,
      "loss": 2.57,
      "step": 1020
    },
    {
      "epoch": 0.1988199326232158,
      "grad_norm": 1.921875,
      "learning_rate": 3.645132026700552e-05,
      "loss": 2.5618,
      "step": 1021
    },
    {
      "epoch": 0.1990146632134442,
      "grad_norm": 1.71875,
      "learning_rate": 3.644430427117857e-05,
      "loss": 2.5616,
      "step": 1022
    },
    {
      "epoch": 0.1992093938036726,
      "grad_norm": 1.5078125,
      "learning_rate": 3.6437282023238025e-05,
      "loss": 2.5624,
      "step": 1023
    },
    {
      "epoch": 0.19940412439390104,
      "grad_norm": 1.40625,
      "learning_rate": 3.643025352585377e-05,
      "loss": 2.5812,
      "step": 1024
    },
    {
      "epoch": 0.19959885498412946,
      "grad_norm": 1.5703125,
      "learning_rate": 3.642321878169801e-05,
      "loss": 2.5584,
      "step": 1025
    },
    {
      "epoch": 0.19979358557435786,
      "grad_norm": 1.3984375,
      "learning_rate": 3.641617779344537e-05,
      "loss": 2.5681,
      "step": 1026
    },
    {
      "epoch": 0.1999883161645863,
      "grad_norm": 1.7109375,
      "learning_rate": 3.640913056377282e-05,
      "loss": 2.5672,
      "step": 1027
    },
    {
      "epoch": 0.20018304675481471,
      "grad_norm": 1.5546875,
      "learning_rate": 3.640207709535971e-05,
      "loss": 2.5459,
      "step": 1028
    },
    {
      "epoch": 0.20037777734504314,
      "grad_norm": 1.5859375,
      "learning_rate": 3.6395017390887775e-05,
      "loss": 2.5611,
      "step": 1029
    },
    {
      "epoch": 0.20057250793527157,
      "grad_norm": 1.5,
      "learning_rate": 3.638795145304109e-05,
      "loss": 2.5566,
      "step": 1030
    },
    {
      "epoch": 0.20076723852549996,
      "grad_norm": 1.515625,
      "learning_rate": 3.6380879284506134e-05,
      "loss": 2.5664,
      "step": 1031
    },
    {
      "epoch": 0.2009619691157284,
      "grad_norm": 1.3984375,
      "learning_rate": 3.6373800887971744e-05,
      "loss": 2.5627,
      "step": 1032
    },
    {
      "epoch": 0.20115669970595682,
      "grad_norm": 1.765625,
      "learning_rate": 3.636671626612911e-05,
      "loss": 2.5553,
      "step": 1033
    },
    {
      "epoch": 0.20135143029618524,
      "grad_norm": 1.703125,
      "learning_rate": 3.635962542167181e-05,
      "loss": 2.5773,
      "step": 1034
    },
    {
      "epoch": 0.20154616088641364,
      "grad_norm": 1.421875,
      "learning_rate": 3.635252835729577e-05,
      "loss": 2.5612,
      "step": 1035
    },
    {
      "epoch": 0.20174089147664206,
      "grad_norm": 1.5,
      "learning_rate": 3.6345425075699295e-05,
      "loss": 2.5646,
      "step": 1036
    },
    {
      "epoch": 0.2019356220668705,
      "grad_norm": 1.4765625,
      "learning_rate": 3.633831557958305e-05,
      "loss": 2.5552,
      "step": 1037
    },
    {
      "epoch": 0.20213035265709892,
      "grad_norm": 1.2578125,
      "learning_rate": 3.6331199871650055e-05,
      "loss": 2.5649,
      "step": 1038
    },
    {
      "epoch": 0.20232508324732731,
      "grad_norm": 1.921875,
      "learning_rate": 3.63240779546057e-05,
      "loss": 2.5576,
      "step": 1039
    },
    {
      "epoch": 0.20251981383755574,
      "grad_norm": 1.765625,
      "learning_rate": 3.631694983115773e-05,
      "loss": 2.5617,
      "step": 1040
    },
    {
      "epoch": 0.20271454442778417,
      "grad_norm": 1.4921875,
      "learning_rate": 3.630981550401627e-05,
      "loss": 2.5611,
      "step": 1041
    },
    {
      "epoch": 0.2029092750180126,
      "grad_norm": 1.40625,
      "learning_rate": 3.630267497589376e-05,
      "loss": 2.5549,
      "step": 1042
    },
    {
      "epoch": 0.203104005608241,
      "grad_norm": 1.546875,
      "learning_rate": 3.629552824950504e-05,
      "loss": 2.5857,
      "step": 1043
    },
    {
      "epoch": 0.20329873619846942,
      "grad_norm": 1.3671875,
      "learning_rate": 3.6288375327567286e-05,
      "loss": 2.5728,
      "step": 1044
    },
    {
      "epoch": 0.20349346678869784,
      "grad_norm": 1.875,
      "learning_rate": 3.628121621280004e-05,
      "loss": 2.5613,
      "step": 1045
    },
    {
      "epoch": 0.20368819737892627,
      "grad_norm": 1.7734375,
      "learning_rate": 3.6274050907925186e-05,
      "loss": 2.5439,
      "step": 1046
    },
    {
      "epoch": 0.20388292796915466,
      "grad_norm": 1.296875,
      "learning_rate": 3.6266879415666964e-05,
      "loss": 2.5526,
      "step": 1047
    },
    {
      "epoch": 0.2040776585593831,
      "grad_norm": 1.28125,
      "learning_rate": 3.625970173875198e-05,
      "loss": 2.5684,
      "step": 1048
    },
    {
      "epoch": 0.20427238914961152,
      "grad_norm": 1.6015625,
      "learning_rate": 3.6252517879909175e-05,
      "loss": 2.5858,
      "step": 1049
    },
    {
      "epoch": 0.20446711973983994,
      "grad_norm": 1.4453125,
      "learning_rate": 3.624532784186986e-05,
      "loss": 2.5783,
      "step": 1050
    },
    {
      "epoch": 0.20466185033006834,
      "grad_norm": 1.84375,
      "learning_rate": 3.623813162736767e-05,
      "loss": 2.5656,
      "step": 1051
    },
    {
      "epoch": 0.20485658092029677,
      "grad_norm": 1.7890625,
      "learning_rate": 3.62309292391386e-05,
      "loss": 2.5531,
      "step": 1052
    },
    {
      "epoch": 0.2050513115105252,
      "grad_norm": 1.2578125,
      "learning_rate": 3.6223720679921e-05,
      "loss": 2.5686,
      "step": 1053
    },
    {
      "epoch": 0.20524604210075362,
      "grad_norm": 1.234375,
      "learning_rate": 3.6216505952455564e-05,
      "loss": 2.5447,
      "step": 1054
    },
    {
      "epoch": 0.20544077269098202,
      "grad_norm": 1.625,
      "learning_rate": 3.620928505948531e-05,
      "loss": 2.5511,
      "step": 1055
    },
    {
      "epoch": 0.20563550328121044,
      "grad_norm": 1.390625,
      "learning_rate": 3.620205800375563e-05,
      "loss": 2.5359,
      "step": 1056
    },
    {
      "epoch": 0.20583023387143887,
      "grad_norm": 2.015625,
      "learning_rate": 3.619482478801425e-05,
      "loss": 2.5701,
      "step": 1057
    },
    {
      "epoch": 0.2060249644616673,
      "grad_norm": 1.8359375,
      "learning_rate": 3.6187585415011224e-05,
      "loss": 2.5427,
      "step": 1058
    },
    {
      "epoch": 0.2062196950518957,
      "grad_norm": 1.3203125,
      "learning_rate": 3.6180339887498953e-05,
      "loss": 2.5706,
      "step": 1059
    },
    {
      "epoch": 0.20641442564212412,
      "grad_norm": 1.3515625,
      "learning_rate": 3.617308820823219e-05,
      "loss": 2.5526,
      "step": 1060
    },
    {
      "epoch": 0.20660915623235254,
      "grad_norm": 1.578125,
      "learning_rate": 3.616583037996802e-05,
      "loss": 2.5572,
      "step": 1061
    },
    {
      "epoch": 0.20680388682258097,
      "grad_norm": 1.3828125,
      "learning_rate": 3.615856640546586e-05,
      "loss": 2.5355,
      "step": 1062
    },
    {
      "epoch": 0.20699861741280937,
      "grad_norm": 1.734375,
      "learning_rate": 3.6151296287487464e-05,
      "loss": 2.5614,
      "step": 1063
    },
    {
      "epoch": 0.2071933480030378,
      "grad_norm": 1.6640625,
      "learning_rate": 3.614402002879693e-05,
      "loss": 2.5574,
      "step": 1064
    },
    {
      "epoch": 0.20738807859326622,
      "grad_norm": 1.40625,
      "learning_rate": 3.6136737632160696e-05,
      "loss": 2.5609,
      "step": 1065
    },
    {
      "epoch": 0.20758280918349464,
      "grad_norm": 1.3359375,
      "learning_rate": 3.612944910034751e-05,
      "loss": 2.5538,
      "step": 1066
    },
    {
      "epoch": 0.20777753977372304,
      "grad_norm": 1.5703125,
      "learning_rate": 3.612215443612849e-05,
      "loss": 2.566,
      "step": 1067
    },
    {
      "epoch": 0.20797227036395147,
      "grad_norm": 1.375,
      "learning_rate": 3.6114853642277024e-05,
      "loss": 2.5535,
      "step": 1068
    },
    {
      "epoch": 0.2081670009541799,
      "grad_norm": 1.765625,
      "learning_rate": 3.610754672156891e-05,
      "loss": 2.5758,
      "step": 1069
    },
    {
      "epoch": 0.20836173154440832,
      "grad_norm": 1.6875,
      "learning_rate": 3.61002336767822e-05,
      "loss": 2.5433,
      "step": 1070
    },
    {
      "epoch": 0.20855646213463672,
      "grad_norm": 1.4375,
      "learning_rate": 3.609291451069733e-05,
      "loss": 2.5467,
      "step": 1071
    },
    {
      "epoch": 0.20875119272486514,
      "grad_norm": 1.390625,
      "learning_rate": 3.608558922609704e-05,
      "loss": 2.5798,
      "step": 1072
    },
    {
      "epoch": 0.20894592331509357,
      "grad_norm": 1.5390625,
      "learning_rate": 3.607825782576639e-05,
      "loss": 2.5621,
      "step": 1073
    },
    {
      "epoch": 0.209140653905322,
      "grad_norm": 1.359375,
      "learning_rate": 3.607092031249278e-05,
      "loss": 2.5388,
      "step": 1074
    },
    {
      "epoch": 0.20933538449555042,
      "grad_norm": 1.7109375,
      "learning_rate": 3.606357668906592e-05,
      "loss": 2.5432,
      "step": 1075
    },
    {
      "epoch": 0.20953011508577882,
      "grad_norm": 1.46875,
      "learning_rate": 3.6056226958277863e-05,
      "loss": 2.5396,
      "step": 1076
    },
    {
      "epoch": 0.20972484567600724,
      "grad_norm": 1.640625,
      "learning_rate": 3.604887112292297e-05,
      "loss": 2.5672,
      "step": 1077
    },
    {
      "epoch": 0.20991957626623567,
      "grad_norm": 1.5234375,
      "learning_rate": 3.604150918579791e-05,
      "loss": 2.556,
      "step": 1078
    },
    {
      "epoch": 0.2101143068564641,
      "grad_norm": 1.59375,
      "learning_rate": 3.6034141149701696e-05,
      "loss": 2.5513,
      "step": 1079
    },
    {
      "epoch": 0.2103090374466925,
      "grad_norm": 1.4609375,
      "learning_rate": 3.602676701743565e-05,
      "loss": 2.5325,
      "step": 1080
    },
    {
      "epoch": 0.21050376803692092,
      "grad_norm": 1.5703125,
      "learning_rate": 3.601938679180343e-05,
      "loss": 2.5287,
      "step": 1081
    },
    {
      "epoch": 0.21069849862714934,
      "grad_norm": 1.390625,
      "learning_rate": 3.6012000475610965e-05,
      "loss": 2.5526,
      "step": 1082
    },
    {
      "epoch": 0.21089322921737777,
      "grad_norm": 1.6015625,
      "learning_rate": 3.600460807166653e-05,
      "loss": 2.5346,
      "step": 1083
    },
    {
      "epoch": 0.21108795980760617,
      "grad_norm": 1.5078125,
      "learning_rate": 3.599720958278073e-05,
      "loss": 2.5446,
      "step": 1084
    },
    {
      "epoch": 0.2112826903978346,
      "grad_norm": 1.609375,
      "learning_rate": 3.598980501176646e-05,
      "loss": 2.5594,
      "step": 1085
    },
    {
      "epoch": 0.21147742098806302,
      "grad_norm": 1.53125,
      "learning_rate": 3.5982394361438926e-05,
      "loss": 2.5424,
      "step": 1086
    },
    {
      "epoch": 0.21167215157829145,
      "grad_norm": 1.484375,
      "learning_rate": 3.597497763461565e-05,
      "loss": 2.5478,
      "step": 1087
    },
    {
      "epoch": 0.21186688216851984,
      "grad_norm": 1.359375,
      "learning_rate": 3.596755483411648e-05,
      "loss": 2.5446,
      "step": 1088
    },
    {
      "epoch": 0.21206161275874827,
      "grad_norm": 1.5546875,
      "learning_rate": 3.596012596276355e-05,
      "loss": 2.5494,
      "step": 1089
    },
    {
      "epoch": 0.2122563433489767,
      "grad_norm": 1.3671875,
      "learning_rate": 3.595269102338131e-05,
      "loss": 2.5268,
      "step": 1090
    },
    {
      "epoch": 0.21245107393920512,
      "grad_norm": 1.796875,
      "learning_rate": 3.594525001879652e-05,
      "loss": 2.5424,
      "step": 1091
    },
    {
      "epoch": 0.21264580452943352,
      "grad_norm": 1.703125,
      "learning_rate": 3.593780295183826e-05,
      "loss": 2.5377,
      "step": 1092
    },
    {
      "epoch": 0.21284053511966194,
      "grad_norm": 1.328125,
      "learning_rate": 3.5930349825337883e-05,
      "loss": 2.5588,
      "step": 1093
    },
    {
      "epoch": 0.21303526570989037,
      "grad_norm": 1.3046875,
      "learning_rate": 3.5922890642129064e-05,
      "loss": 2.5528,
      "step": 1094
    },
    {
      "epoch": 0.2132299963001188,
      "grad_norm": 1.578125,
      "learning_rate": 3.591542540504779e-05,
      "loss": 2.5519,
      "step": 1095
    },
    {
      "epoch": 0.2134247268903472,
      "grad_norm": 1.4140625,
      "learning_rate": 3.5907954116932326e-05,
      "loss": 2.5769,
      "step": 1096
    },
    {
      "epoch": 0.21361945748057562,
      "grad_norm": 1.7265625,
      "learning_rate": 3.590047678062326e-05,
      "loss": 2.5394,
      "step": 1097
    },
    {
      "epoch": 0.21381418807080405,
      "grad_norm": 1.7109375,
      "learning_rate": 3.5892993398963466e-05,
      "loss": 2.5832,
      "step": 1098
    },
    {
      "epoch": 0.21400891866103247,
      "grad_norm": 1.3515625,
      "learning_rate": 3.5885503974798125e-05,
      "loss": 2.5689,
      "step": 1099
    },
    {
      "epoch": 0.21420364925126087,
      "grad_norm": 1.3203125,
      "learning_rate": 3.5878008510974705e-05,
      "loss": 2.5566,
      "step": 1100
    },
    {
      "epoch": 0.2143983798414893,
      "grad_norm": 1.453125,
      "learning_rate": 3.5870507010342984e-05,
      "loss": 2.5382,
      "step": 1101
    },
    {
      "epoch": 0.21459311043171772,
      "grad_norm": 1.2421875,
      "learning_rate": 3.5862999475755024e-05,
      "loss": 2.5539,
      "step": 1102
    },
    {
      "epoch": 0.21478784102194615,
      "grad_norm": 1.984375,
      "learning_rate": 3.5855485910065177e-05,
      "loss": 2.5547,
      "step": 1103
    },
    {
      "epoch": 0.21498257161217454,
      "grad_norm": 1.8515625,
      "learning_rate": 3.584796631613011e-05,
      "loss": 2.5699,
      "step": 1104
    },
    {
      "epoch": 0.21517730220240297,
      "grad_norm": 1.25,
      "learning_rate": 3.584044069680876e-05,
      "loss": 2.5673,
      "step": 1105
    },
    {
      "epoch": 0.2153720327926314,
      "grad_norm": 1.1328125,
      "learning_rate": 3.583290905496236e-05,
      "loss": 2.5436,
      "step": 1106
    },
    {
      "epoch": 0.21556676338285982,
      "grad_norm": 1.5234375,
      "learning_rate": 3.582537139345444e-05,
      "loss": 2.5652,
      "step": 1107
    },
    {
      "epoch": 0.21576149397308822,
      "grad_norm": 1.1484375,
      "learning_rate": 3.5817827715150814e-05,
      "loss": 2.5557,
      "step": 1108
    },
    {
      "epoch": 0.21595622456331665,
      "grad_norm": 2.046875,
      "learning_rate": 3.581027802291957e-05,
      "loss": 2.5513,
      "step": 1109
    },
    {
      "epoch": 0.21615095515354507,
      "grad_norm": 1.90625,
      "learning_rate": 3.580272231963112e-05,
      "loss": 2.545,
      "step": 1110
    },
    {
      "epoch": 0.2163456857437735,
      "grad_norm": 1.2578125,
      "learning_rate": 3.579516060815811e-05,
      "loss": 2.5484,
      "step": 1111
    },
    {
      "epoch": 0.2165404163340019,
      "grad_norm": 1.1875,
      "learning_rate": 3.578759289137553e-05,
      "loss": 2.5365,
      "step": 1112
    },
    {
      "epoch": 0.21673514692423032,
      "grad_norm": 1.609375,
      "learning_rate": 3.57800191721606e-05,
      "loss": 2.5459,
      "step": 1113
    },
    {
      "epoch": 0.21692987751445875,
      "grad_norm": 1.234375,
      "learning_rate": 3.577243945339284e-05,
      "loss": 2.5694,
      "step": 1114
    },
    {
      "epoch": 0.21712460810468717,
      "grad_norm": 2.015625,
      "learning_rate": 3.576485373795406e-05,
      "loss": 2.5564,
      "step": 1115
    },
    {
      "epoch": 0.2173193386949156,
      "grad_norm": 1.921875,
      "learning_rate": 3.575726202872835e-05,
      "loss": 2.5484,
      "step": 1116
    },
    {
      "epoch": 0.217514069285144,
      "grad_norm": 1.203125,
      "learning_rate": 3.5749664328602064e-05,
      "loss": 2.5619,
      "step": 1117
    },
    {
      "epoch": 0.21770879987537242,
      "grad_norm": 1.2265625,
      "learning_rate": 3.574206064046384e-05,
      "loss": 2.5377,
      "step": 1118
    },
    {
      "epoch": 0.21790353046560085,
      "grad_norm": 1.34375,
      "learning_rate": 3.5734450967204606e-05,
      "loss": 2.5396,
      "step": 1119
    },
    {
      "epoch": 0.21809826105582927,
      "grad_norm": 1.1171875,
      "learning_rate": 3.572683531171755e-05,
      "loss": 2.5416,
      "step": 1120
    },
    {
      "epoch": 0.21829299164605767,
      "grad_norm": 2.015625,
      "learning_rate": 3.5719213676898144e-05,
      "loss": 2.5407,
      "step": 1121
    },
    {
      "epoch": 0.2184877222362861,
      "grad_norm": 1.859375,
      "learning_rate": 3.571158606564412e-05,
      "loss": 2.5647,
      "step": 1122
    },
    {
      "epoch": 0.21868245282651452,
      "grad_norm": 1.4609375,
      "learning_rate": 3.570395248085549e-05,
      "loss": 2.5505,
      "step": 1123
    },
    {
      "epoch": 0.21887718341674295,
      "grad_norm": 1.421875,
      "learning_rate": 3.5696312925434546e-05,
      "loss": 2.5591,
      "step": 1124
    },
    {
      "epoch": 0.21907191400697135,
      "grad_norm": 1.28125,
      "learning_rate": 3.568866740228584e-05,
      "loss": 2.5515,
      "step": 1125
    },
    {
      "epoch": 0.21926664459719977,
      "grad_norm": 1.1953125,
      "learning_rate": 3.568101591431619e-05,
      "loss": 2.5624,
      "step": 1126
    },
    {
      "epoch": 0.2194613751874282,
      "grad_norm": 1.7890625,
      "learning_rate": 3.567335846443469e-05,
      "loss": 2.5403,
      "step": 1127
    },
    {
      "epoch": 0.21965610577765662,
      "grad_norm": 1.484375,
      "learning_rate": 3.5665695055552696e-05,
      "loss": 2.5614,
      "step": 1128
    },
    {
      "epoch": 0.21985083636788502,
      "grad_norm": 1.7421875,
      "learning_rate": 3.565802569058383e-05,
      "loss": 2.5646,
      "step": 1129
    },
    {
      "epoch": 0.22004556695811345,
      "grad_norm": 1.625,
      "learning_rate": 3.565035037244398e-05,
      "loss": 2.5536,
      "step": 1130
    },
    {
      "epoch": 0.22024029754834187,
      "grad_norm": 1.3671875,
      "learning_rate": 3.56426691040513e-05,
      "loss": 2.5502,
      "step": 1131
    },
    {
      "epoch": 0.2204350281385703,
      "grad_norm": 1.375,
      "learning_rate": 3.563498188832621e-05,
      "loss": 2.5323,
      "step": 1132
    },
    {
      "epoch": 0.2206297587287987,
      "grad_norm": 1.453125,
      "learning_rate": 3.562728872819136e-05,
      "loss": 2.5473,
      "step": 1133
    },
    {
      "epoch": 0.22082448931902712,
      "grad_norm": 1.2421875,
      "learning_rate": 3.561958962657171e-05,
      "loss": 2.558,
      "step": 1134
    },
    {
      "epoch": 0.22101921990925555,
      "grad_norm": 1.6953125,
      "learning_rate": 3.5611884586394445e-05,
      "loss": 2.5398,
      "step": 1135
    },
    {
      "epoch": 0.22121395049948397,
      "grad_norm": 1.515625,
      "learning_rate": 3.5604173610589006e-05,
      "loss": 2.5471,
      "step": 1136
    },
    {
      "epoch": 0.22140868108971237,
      "grad_norm": 1.6328125,
      "learning_rate": 3.5596456702087114e-05,
      "loss": 2.5735,
      "step": 1137
    },
    {
      "epoch": 0.2216034116799408,
      "grad_norm": 1.546875,
      "learning_rate": 3.558873386382273e-05,
      "loss": 2.5515,
      "step": 1138
    },
    {
      "epoch": 0.22179814227016922,
      "grad_norm": 1.40625,
      "learning_rate": 3.558100509873207e-05,
      "loss": 2.5508,
      "step": 1139
    },
    {
      "epoch": 0.22199287286039765,
      "grad_norm": 1.4921875,
      "learning_rate": 3.55732704097536e-05,
      "loss": 2.5598,
      "step": 1140
    },
    {
      "epoch": 0.22218760345062605,
      "grad_norm": 1.53125,
      "learning_rate": 3.556552979982805e-05,
      "loss": 2.5425,
      "step": 1141
    },
    {
      "epoch": 0.22238233404085447,
      "grad_norm": 1.3984375,
      "learning_rate": 3.555778327189839e-05,
      "loss": 2.5531,
      "step": 1142
    },
    {
      "epoch": 0.2225770646310829,
      "grad_norm": 1.53125,
      "learning_rate": 3.5550030828909856e-05,
      "loss": 2.5707,
      "step": 1143
    },
    {
      "epoch": 0.22277179522131133,
      "grad_norm": 1.4453125,
      "learning_rate": 3.554227247380991e-05,
      "loss": 2.5315,
      "step": 1144
    },
    {
      "epoch": 0.22296652581153972,
      "grad_norm": 1.4375,
      "learning_rate": 3.553450820954828e-05,
      "loss": 2.5588,
      "step": 1145
    },
    {
      "epoch": 0.22316125640176815,
      "grad_norm": 1.3359375,
      "learning_rate": 3.5526738039076934e-05,
      "loss": 2.5611,
      "step": 1146
    },
    {
      "epoch": 0.22335598699199657,
      "grad_norm": 1.609375,
      "learning_rate": 3.551896196535008e-05,
      "loss": 2.5491,
      "step": 1147
    },
    {
      "epoch": 0.223550717582225,
      "grad_norm": 1.5078125,
      "learning_rate": 3.551117999132418e-05,
      "loss": 2.5559,
      "step": 1148
    },
    {
      "epoch": 0.2237454481724534,
      "grad_norm": 1.5859375,
      "learning_rate": 3.550339211995793e-05,
      "loss": 2.5626,
      "step": 1149
    },
    {
      "epoch": 0.22394017876268182,
      "grad_norm": 1.4609375,
      "learning_rate": 3.549559835421228e-05,
      "loss": 2.5366,
      "step": 1150
    },
    {
      "epoch": 0.22413490935291025,
      "grad_norm": 1.375,
      "learning_rate": 3.548779869705041e-05,
      "loss": 2.5544,
      "step": 1151
    },
    {
      "epoch": 0.22432963994313868,
      "grad_norm": 1.265625,
      "learning_rate": 3.5479993151437754e-05,
      "loss": 2.5494,
      "step": 1152
    },
    {
      "epoch": 0.22452437053336707,
      "grad_norm": 1.4921875,
      "learning_rate": 3.547218172034196e-05,
      "loss": 2.5546,
      "step": 1153
    },
    {
      "epoch": 0.2247191011235955,
      "grad_norm": 1.328125,
      "learning_rate": 3.546436440673294e-05,
      "loss": 2.5212,
      "step": 1154
    },
    {
      "epoch": 0.22491383171382393,
      "grad_norm": 1.6640625,
      "learning_rate": 3.545654121358284e-05,
      "loss": 2.547,
      "step": 1155
    },
    {
      "epoch": 0.22510856230405235,
      "grad_norm": 1.5234375,
      "learning_rate": 3.5448712143866016e-05,
      "loss": 2.5549,
      "step": 1156
    },
    {
      "epoch": 0.22530329289428075,
      "grad_norm": 1.4453125,
      "learning_rate": 3.544087720055908e-05,
      "loss": 2.535,
      "step": 1157
    },
    {
      "epoch": 0.22549802348450917,
      "grad_norm": 1.40625,
      "learning_rate": 3.543303638664087e-05,
      "loss": 2.533,
      "step": 1158
    },
    {
      "epoch": 0.2256927540747376,
      "grad_norm": 1.546875,
      "learning_rate": 3.542518970509246e-05,
      "loss": 2.5299,
      "step": 1159
    },
    {
      "epoch": 0.22588748466496603,
      "grad_norm": 1.34375,
      "learning_rate": 3.541733715889716e-05,
      "loss": 2.5469,
      "step": 1160
    },
    {
      "epoch": 0.22608221525519445,
      "grad_norm": 1.6875,
      "learning_rate": 3.540947875104049e-05,
      "loss": 2.5253,
      "step": 1161
    },
    {
      "epoch": 0.22627694584542285,
      "grad_norm": 1.484375,
      "learning_rate": 3.540161448451023e-05,
      "loss": 2.5393,
      "step": 1162
    },
    {
      "epoch": 0.22647167643565128,
      "grad_norm": 1.453125,
      "learning_rate": 3.539374436229635e-05,
      "loss": 2.5538,
      "step": 1163
    },
    {
      "epoch": 0.2266664070258797,
      "grad_norm": 1.34375,
      "learning_rate": 3.5385868387391076e-05,
      "loss": 2.5387,
      "step": 1164
    },
    {
      "epoch": 0.22686113761610813,
      "grad_norm": 1.3671875,
      "learning_rate": 3.5377986562788844e-05,
      "loss": 2.5283,
      "step": 1165
    },
    {
      "epoch": 0.22705586820633653,
      "grad_norm": 1.296875,
      "learning_rate": 3.537009889148632e-05,
      "loss": 2.5525,
      "step": 1166
    },
    {
      "epoch": 0.22725059879656495,
      "grad_norm": 1.625,
      "learning_rate": 3.5362205376482386e-05,
      "loss": 2.5341,
      "step": 1167
    },
    {
      "epoch": 0.22744532938679338,
      "grad_norm": 1.46875,
      "learning_rate": 3.535430602077816e-05,
      "loss": 2.5464,
      "step": 1168
    },
    {
      "epoch": 0.2276400599770218,
      "grad_norm": 1.5078125,
      "learning_rate": 3.534640082737697e-05,
      "loss": 2.5389,
      "step": 1169
    },
    {
      "epoch": 0.2278347905672502,
      "grad_norm": 1.375,
      "learning_rate": 3.533848979928436e-05,
      "loss": 2.5453,
      "step": 1170
    },
    {
      "epoch": 0.22802952115747863,
      "grad_norm": 1.34375,
      "learning_rate": 3.533057293950811e-05,
      "loss": 2.5227,
      "step": 1171
    },
    {
      "epoch": 0.22822425174770705,
      "grad_norm": 1.171875,
      "learning_rate": 3.532265025105819e-05,
      "loss": 2.5617,
      "step": 1172
    },
    {
      "epoch": 0.22841898233793548,
      "grad_norm": 1.671875,
      "learning_rate": 3.53147217369468e-05,
      "loss": 2.526,
      "step": 1173
    },
    {
      "epoch": 0.22861371292816388,
      "grad_norm": 1.453125,
      "learning_rate": 3.5306787400188376e-05,
      "loss": 2.5293,
      "step": 1174
    },
    {
      "epoch": 0.2288084435183923,
      "grad_norm": 1.578125,
      "learning_rate": 3.5298847243799525e-05,
      "loss": 2.5378,
      "step": 1175
    },
    {
      "epoch": 0.22900317410862073,
      "grad_norm": 1.453125,
      "learning_rate": 3.5290901270799104e-05,
      "loss": 2.5319,
      "step": 1176
    },
    {
      "epoch": 0.22919790469884915,
      "grad_norm": 1.3359375,
      "learning_rate": 3.528294948420816e-05,
      "loss": 2.5647,
      "step": 1177
    },
    {
      "epoch": 0.22939263528907755,
      "grad_norm": 1.21875,
      "learning_rate": 3.5274991887049954e-05,
      "loss": 2.5349,
      "step": 1178
    },
    {
      "epoch": 0.22958736587930598,
      "grad_norm": 1.5,
      "learning_rate": 3.526702848234996e-05,
      "loss": 2.5412,
      "step": 1179
    },
    {
      "epoch": 0.2297820964695344,
      "grad_norm": 1.4453125,
      "learning_rate": 3.5259059273135874e-05,
      "loss": 2.5493,
      "step": 1180
    },
    {
      "epoch": 0.22997682705976283,
      "grad_norm": 1.46875,
      "learning_rate": 3.525108426243757e-05,
      "loss": 2.534,
      "step": 1181
    },
    {
      "epoch": 0.23017155764999123,
      "grad_norm": 1.4609375,
      "learning_rate": 3.524310345328713e-05,
      "loss": 2.5267,
      "step": 1182
    },
    {
      "epoch": 0.23036628824021965,
      "grad_norm": 1.390625,
      "learning_rate": 3.523511684871888e-05,
      "loss": 2.5381,
      "step": 1183
    },
    {
      "epoch": 0.23056101883044808,
      "grad_norm": 1.25,
      "learning_rate": 3.522712445176931e-05,
      "loss": 2.5353,
      "step": 1184
    },
    {
      "epoch": 0.2307557494206765,
      "grad_norm": 1.578125,
      "learning_rate": 3.521912626547712e-05,
      "loss": 2.5254,
      "step": 1185
    },
    {
      "epoch": 0.2309504800109049,
      "grad_norm": 1.390625,
      "learning_rate": 3.5211122292883216e-05,
      "loss": 2.5589,
      "step": 1186
    },
    {
      "epoch": 0.23114521060113333,
      "grad_norm": 1.5546875,
      "learning_rate": 3.52031125370307e-05,
      "loss": 2.5619,
      "step": 1187
    },
    {
      "epoch": 0.23133994119136175,
      "grad_norm": 1.453125,
      "learning_rate": 3.519509700096489e-05,
      "loss": 2.5369,
      "step": 1188
    },
    {
      "epoch": 0.23153467178159018,
      "grad_norm": 1.4375,
      "learning_rate": 3.518707568773327e-05,
      "loss": 2.5586,
      "step": 1189
    },
    {
      "epoch": 0.23172940237181858,
      "grad_norm": 1.3515625,
      "learning_rate": 3.517904860038555e-05,
      "loss": 2.5442,
      "step": 1190
    },
    {
      "epoch": 0.231924132962047,
      "grad_norm": 1.3515625,
      "learning_rate": 3.5171015741973615e-05,
      "loss": 2.5302,
      "step": 1191
    },
    {
      "epoch": 0.23211886355227543,
      "grad_norm": 1.25,
      "learning_rate": 3.516297711555156e-05,
      "loss": 2.5362,
      "step": 1192
    },
    {
      "epoch": 0.23231359414250385,
      "grad_norm": 1.5625,
      "learning_rate": 3.515493272417566e-05,
      "loss": 2.5433,
      "step": 1193
    },
    {
      "epoch": 0.23250832473273225,
      "grad_norm": 1.390625,
      "learning_rate": 3.514688257090439e-05,
      "loss": 2.547,
      "step": 1194
    },
    {
      "epoch": 0.23270305532296068,
      "grad_norm": 1.671875,
      "learning_rate": 3.513882665879842e-05,
      "loss": 2.5369,
      "step": 1195
    },
    {
      "epoch": 0.2328977859131891,
      "grad_norm": 1.578125,
      "learning_rate": 3.513076499092058e-05,
      "loss": 2.535,
      "step": 1196
    },
    {
      "epoch": 0.23309251650341753,
      "grad_norm": 1.1953125,
      "learning_rate": 3.512269757033594e-05,
      "loss": 2.5515,
      "step": 1197
    },
    {
      "epoch": 0.23328724709364593,
      "grad_norm": 1.1875,
      "learning_rate": 3.511462440011171e-05,
      "loss": 2.544,
      "step": 1198
    },
    {
      "epoch": 0.23348197768387435,
      "grad_norm": 1.3671875,
      "learning_rate": 3.51065454833173e-05,
      "loss": 2.5418,
      "step": 1199
    },
    {
      "epoch": 0.23367670827410278,
      "grad_norm": 1.1015625,
      "learning_rate": 3.509846082302432e-05,
      "loss": 2.5445,
      "step": 1200
    },
    {
      "epoch": 0.2338714388643312,
      "grad_norm": 1.8046875,
      "learning_rate": 3.5090370422306555e-05,
      "loss": 2.537,
      "step": 1201
    },
    {
      "epoch": 0.23406616945455963,
      "grad_norm": 1.578125,
      "learning_rate": 3.508227428423996e-05,
      "loss": 2.5526,
      "step": 1202
    },
    {
      "epoch": 0.23426090004478803,
      "grad_norm": 1.3515625,
      "learning_rate": 3.5074172411902677e-05,
      "loss": 2.5334,
      "step": 1203
    },
    {
      "epoch": 0.23445563063501645,
      "grad_norm": 1.359375,
      "learning_rate": 3.506606480837505e-05,
      "loss": 2.551,
      "step": 1204
    },
    {
      "epoch": 0.23465036122524488,
      "grad_norm": 1.453125,
      "learning_rate": 3.505795147673957e-05,
      "loss": 2.5456,
      "step": 1205
    },
    {
      "epoch": 0.2348450918154733,
      "grad_norm": 1.359375,
      "learning_rate": 3.5049832420080934e-05,
      "loss": 2.5417,
      "step": 1206
    },
    {
      "epoch": 0.2350398224057017,
      "grad_norm": 1.3984375,
      "learning_rate": 3.504170764148598e-05,
      "loss": 2.5351,
      "step": 1207
    },
    {
      "epoch": 0.23523455299593013,
      "grad_norm": 1.296875,
      "learning_rate": 3.503357714404376e-05,
      "loss": 2.5344,
      "step": 1208
    },
    {
      "epoch": 0.23542928358615856,
      "grad_norm": 1.3984375,
      "learning_rate": 3.502544093084549e-05,
      "loss": 2.5403,
      "step": 1209
    },
    {
      "epoch": 0.23562401417638698,
      "grad_norm": 1.25,
      "learning_rate": 3.5017299004984536e-05,
      "loss": 2.5389,
      "step": 1210
    },
    {
      "epoch": 0.23581874476661538,
      "grad_norm": 1.59375,
      "learning_rate": 3.500915136955646e-05,
      "loss": 2.5383,
      "step": 1211
    },
    {
      "epoch": 0.2360134753568438,
      "grad_norm": 1.3671875,
      "learning_rate": 3.500099802765899e-05,
      "loss": 2.5413,
      "step": 1212
    },
    {
      "epoch": 0.23620820594707223,
      "grad_norm": 1.515625,
      "learning_rate": 3.4992838982392016e-05,
      "loss": 2.5292,
      "step": 1213
    },
    {
      "epoch": 0.23640293653730066,
      "grad_norm": 1.5234375,
      "learning_rate": 3.4984674236857605e-05,
      "loss": 2.538,
      "step": 1214
    },
    {
      "epoch": 0.23659766712752905,
      "grad_norm": 1.1953125,
      "learning_rate": 3.497650379415999e-05,
      "loss": 2.5321,
      "step": 1215
    },
    {
      "epoch": 0.23679239771775748,
      "grad_norm": 1.140625,
      "learning_rate": 3.496832765740556e-05,
      "loss": 2.5356,
      "step": 1216
    },
    {
      "epoch": 0.2369871283079859,
      "grad_norm": 1.4296875,
      "learning_rate": 3.496014582970289e-05,
      "loss": 2.5271,
      "step": 1217
    },
    {
      "epoch": 0.23718185889821433,
      "grad_norm": 1.1796875,
      "learning_rate": 3.4951958314162696e-05,
      "loss": 2.5406,
      "step": 1218
    },
    {
      "epoch": 0.23737658948844273,
      "grad_norm": 1.8046875,
      "learning_rate": 3.494376511389786e-05,
      "loss": 2.5507,
      "step": 1219
    },
    {
      "epoch": 0.23757132007867116,
      "grad_norm": 1.6953125,
      "learning_rate": 3.4935566232023444e-05,
      "loss": 2.5472,
      "step": 1220
    },
    {
      "epoch": 0.23776605066889958,
      "grad_norm": 1.15625,
      "learning_rate": 3.492736167165665e-05,
      "loss": 2.5426,
      "step": 1221
    },
    {
      "epoch": 0.237960781259128,
      "grad_norm": 1.3359375,
      "learning_rate": 3.491915143591685e-05,
      "loss": 2.5365,
      "step": 1222
    },
    {
      "epoch": 0.2381555118493564,
      "grad_norm": 1.1953125,
      "learning_rate": 3.491093552792557e-05,
      "loss": 2.5415,
      "step": 1223
    },
    {
      "epoch": 0.23835024243958483,
      "grad_norm": 0.9921875,
      "learning_rate": 3.4902713950806494e-05,
      "loss": 2.5406,
      "step": 1224
    },
    {
      "epoch": 0.23854497302981326,
      "grad_norm": 1.6640625,
      "learning_rate": 3.4894486707685454e-05,
      "loss": 2.5249,
      "step": 1225
    },
    {
      "epoch": 0.23873970362004168,
      "grad_norm": 1.46875,
      "learning_rate": 3.488625380169044e-05,
      "loss": 2.5296,
      "step": 1226
    },
    {
      "epoch": 0.23893443421027008,
      "grad_norm": 1.640625,
      "learning_rate": 3.4878015235951614e-05,
      "loss": 2.5197,
      "step": 1227
    },
    {
      "epoch": 0.2391291648004985,
      "grad_norm": 1.640625,
      "learning_rate": 3.486977101360126e-05,
      "loss": 2.5379,
      "step": 1228
    },
    {
      "epoch": 0.23932389539072693,
      "grad_norm": 1.15625,
      "learning_rate": 3.486152113777383e-05,
      "loss": 2.553,
      "step": 1229
    },
    {
      "epoch": 0.23951862598095536,
      "grad_norm": 1.15625,
      "learning_rate": 3.485326561160591e-05,
      "loss": 2.5288,
      "step": 1230
    },
    {
      "epoch": 0.23971335657118376,
      "grad_norm": 1.2578125,
      "learning_rate": 3.484500443823627e-05,
      "loss": 2.5397,
      "step": 1231
    },
    {
      "epoch": 0.23990808716141218,
      "grad_norm": 1.0625,
      "learning_rate": 3.4836737620805774e-05,
      "loss": 2.5417,
      "step": 1232
    },
    {
      "epoch": 0.2401028177516406,
      "grad_norm": 1.7109375,
      "learning_rate": 3.482846516245749e-05,
      "loss": 2.5443,
      "step": 1233
    },
    {
      "epoch": 0.24029754834186903,
      "grad_norm": 1.421875,
      "learning_rate": 3.482018706633657e-05,
      "loss": 2.5086,
      "step": 1234
    },
    {
      "epoch": 0.24049227893209743,
      "grad_norm": 1.546875,
      "learning_rate": 3.481190333559037e-05,
      "loss": 2.526,
      "step": 1235
    },
    {
      "epoch": 0.24068700952232586,
      "grad_norm": 1.4453125,
      "learning_rate": 3.4803613973368334e-05,
      "loss": 2.5334,
      "step": 1236
    },
    {
      "epoch": 0.24088174011255428,
      "grad_norm": 1.2265625,
      "learning_rate": 3.4795318982822084e-05,
      "loss": 2.532,
      "step": 1237
    },
    {
      "epoch": 0.2410764707027827,
      "grad_norm": 1.1484375,
      "learning_rate": 3.4787018367105374e-05,
      "loss": 2.5517,
      "step": 1238
    },
    {
      "epoch": 0.2412712012930111,
      "grad_norm": 1.453125,
      "learning_rate": 3.477871212937409e-05,
      "loss": 2.5459,
      "step": 1239
    },
    {
      "epoch": 0.24146593188323953,
      "grad_norm": 1.203125,
      "learning_rate": 3.477040027278626e-05,
      "loss": 2.5402,
      "step": 1240
    },
    {
      "epoch": 0.24166066247346796,
      "grad_norm": 1.6015625,
      "learning_rate": 3.4762082800502034e-05,
      "loss": 2.5474,
      "step": 1241
    },
    {
      "epoch": 0.24185539306369638,
      "grad_norm": 1.4375,
      "learning_rate": 3.475375971568372e-05,
      "loss": 2.5225,
      "step": 1242
    },
    {
      "epoch": 0.24205012365392478,
      "grad_norm": 1.4453125,
      "learning_rate": 3.4745431021495744e-05,
      "loss": 2.5604,
      "step": 1243
    },
    {
      "epoch": 0.2422448542441532,
      "grad_norm": 1.3828125,
      "learning_rate": 3.473709672110467e-05,
      "loss": 2.5455,
      "step": 1244
    },
    {
      "epoch": 0.24243958483438163,
      "grad_norm": 1.3125,
      "learning_rate": 3.472875681767921e-05,
      "loss": 2.5193,
      "step": 1245
    },
    {
      "epoch": 0.24263431542461006,
      "grad_norm": 1.09375,
      "learning_rate": 3.472041131439016e-05,
      "loss": 2.5284,
      "step": 1246
    },
    {
      "epoch": 0.24282904601483848,
      "grad_norm": 1.4765625,
      "learning_rate": 3.471206021441051e-05,
      "loss": 2.5134,
      "step": 1247
    },
    {
      "epoch": 0.24302377660506688,
      "grad_norm": 1.2734375,
      "learning_rate": 3.4703703520915314e-05,
      "loss": 2.5306,
      "step": 1248
    },
    {
      "epoch": 0.2432185071952953,
      "grad_norm": 1.5859375,
      "learning_rate": 3.469534123708178e-05,
      "loss": 2.5214,
      "step": 1249
    },
    {
      "epoch": 0.24341323778552373,
      "grad_norm": 1.46875,
      "learning_rate": 3.4686973366089274e-05,
      "loss": 2.536,
      "step": 1250
    },
    {
      "epoch": 0.24360796837575216,
      "grad_norm": 1.21875,
      "learning_rate": 3.4678599911119224e-05,
      "loss": 2.5316,
      "step": 1251
    },
    {
      "epoch": 0.24380269896598056,
      "grad_norm": 1.1953125,
      "learning_rate": 3.467022087535522e-05,
      "loss": 2.526,
      "step": 1252
    },
    {
      "epoch": 0.24399742955620898,
      "grad_norm": 1.2109375,
      "learning_rate": 3.466183626198298e-05,
      "loss": 2.5186,
      "step": 1253
    },
    {
      "epoch": 0.2441921601464374,
      "grad_norm": 1.1015625,
      "learning_rate": 3.465344607419031e-05,
      "loss": 2.5346,
      "step": 1254
    },
    {
      "epoch": 0.24438689073666583,
      "grad_norm": 1.515625,
      "learning_rate": 3.4645050315167164e-05,
      "loss": 2.5325,
      "step": 1255
    },
    {
      "epoch": 0.24458162132689423,
      "grad_norm": 1.21875,
      "learning_rate": 3.46366489881056e-05,
      "loss": 2.5536,
      "step": 1256
    },
    {
      "epoch": 0.24477635191712266,
      "grad_norm": 1.6328125,
      "learning_rate": 3.46282420961998e-05,
      "loss": 2.5552,
      "step": 1257
    },
    {
      "epoch": 0.24497108250735108,
      "grad_norm": 1.5234375,
      "learning_rate": 3.4619829642646055e-05,
      "loss": 2.5468,
      "step": 1258
    },
    {
      "epoch": 0.2451658130975795,
      "grad_norm": 1.234375,
      "learning_rate": 3.4611411630642764e-05,
      "loss": 2.5476,
      "step": 1259
    },
    {
      "epoch": 0.2453605436878079,
      "grad_norm": 1.2265625,
      "learning_rate": 3.4602988063390476e-05,
      "loss": 2.5348,
      "step": 1260
    },
    {
      "epoch": 0.24555527427803633,
      "grad_norm": 1.2421875,
      "learning_rate": 3.459455894409181e-05,
      "loss": 2.5245,
      "step": 1261
    },
    {
      "epoch": 0.24575000486826476,
      "grad_norm": 0.99609375,
      "learning_rate": 3.458612427595151e-05,
      "loss": 2.5464,
      "step": 1262
    },
    {
      "epoch": 0.24594473545849319,
      "grad_norm": 1.484375,
      "learning_rate": 3.4577684062176435e-05,
      "loss": 2.521,
      "step": 1263
    },
    {
      "epoch": 0.24613946604872158,
      "grad_norm": 1.21875,
      "learning_rate": 3.456923830597555e-05,
      "loss": 2.513,
      "step": 1264
    },
    {
      "epoch": 0.24633419663895,
      "grad_norm": 1.7734375,
      "learning_rate": 3.456078701055992e-05,
      "loss": 2.5242,
      "step": 1265
    },
    {
      "epoch": 0.24652892722917844,
      "grad_norm": 1.703125,
      "learning_rate": 3.4552330179142725e-05,
      "loss": 2.5337,
      "step": 1266
    },
    {
      "epoch": 0.24672365781940686,
      "grad_norm": 1.0078125,
      "learning_rate": 3.454386781493925e-05,
      "loss": 2.5396,
      "step": 1267
    },
    {
      "epoch": 0.24691838840963526,
      "grad_norm": 1.140625,
      "learning_rate": 3.453539992116688e-05,
      "loss": 2.526,
      "step": 1268
    },
    {
      "epoch": 0.24711311899986368,
      "grad_norm": 1.03125,
      "learning_rate": 3.45269265010451e-05,
      "loss": 2.515,
      "step": 1269
    },
    {
      "epoch": 0.2473078495900921,
      "grad_norm": 0.9375,
      "learning_rate": 3.4518447557795495e-05,
      "loss": 2.5503,
      "step": 1270
    },
    {
      "epoch": 0.24750258018032054,
      "grad_norm": 1.2265625,
      "learning_rate": 3.450996309464176e-05,
      "loss": 2.5379,
      "step": 1271
    },
    {
      "epoch": 0.24769731077054893,
      "grad_norm": 0.87890625,
      "learning_rate": 3.450147311480968e-05,
      "loss": 2.5201,
      "step": 1272
    },
    {
      "epoch": 0.24789204136077736,
      "grad_norm": 1.7109375,
      "learning_rate": 3.4492977621527136e-05,
      "loss": 2.5526,
      "step": 1273
    },
    {
      "epoch": 0.24808677195100579,
      "grad_norm": 1.3359375,
      "learning_rate": 3.448447661802411e-05,
      "loss": 2.5307,
      "step": 1274
    },
    {
      "epoch": 0.2482815025412342,
      "grad_norm": 1.859375,
      "learning_rate": 3.447597010753269e-05,
      "loss": 2.5325,
      "step": 1275
    },
    {
      "epoch": 0.2484762331314626,
      "grad_norm": 1.8984375,
      "learning_rate": 3.446745809328703e-05,
      "loss": 2.5213,
      "step": 1276
    },
    {
      "epoch": 0.24867096372169104,
      "grad_norm": 0.85546875,
      "learning_rate": 3.4458940578523394e-05,
      "loss": 2.5148,
      "step": 1277
    },
    {
      "epoch": 0.24886569431191946,
      "grad_norm": 1.59375,
      "learning_rate": 3.445041756648014e-05,
      "loss": 2.5224,
      "step": 1278
    },
    {
      "epoch": 0.2490604249021479,
      "grad_norm": 0.90234375,
      "learning_rate": 3.444188906039772e-05,
      "loss": 2.5328,
      "step": 1279
    },
    {
      "epoch": 0.24925515549237628,
      "grad_norm": 1.9296875,
      "learning_rate": 3.443335506351865e-05,
      "loss": 2.541,
      "step": 1280
    },
    {
      "epoch": 0.2494498860826047,
      "grad_norm": 1.8203125,
      "learning_rate": 3.442481557908757e-05,
      "loss": 2.5433,
      "step": 1281
    },
    {
      "epoch": 0.24964461667283314,
      "grad_norm": 1.046875,
      "learning_rate": 3.441627061035116e-05,
      "loss": 2.5227,
      "step": 1282
    },
    {
      "epoch": 0.24983934726306156,
      "grad_norm": 1.5546875,
      "learning_rate": 3.4407720160558226e-05,
      "loss": 2.5454,
      "step": 1283
    },
    {
      "epoch": 0.25003407785328996,
      "grad_norm": 0.93359375,
      "learning_rate": 3.439916423295964e-05,
      "loss": 2.5341,
      "step": 1284
    },
    {
      "epoch": 0.2502288084435184,
      "grad_norm": 2.03125,
      "learning_rate": 3.439060283080838e-05,
      "loss": 2.5394,
      "step": 1285
    },
    {
      "epoch": 0.2504235390337468,
      "grad_norm": 1.6328125,
      "learning_rate": 3.438203595735945e-05,
      "loss": 2.5486,
      "step": 1286
    },
    {
      "epoch": 0.2506182696239752,
      "grad_norm": 1.7421875,
      "learning_rate": 3.4373463615869996e-05,
      "loss": 2.5318,
      "step": 1287
    },
    {
      "epoch": 0.25081300021420366,
      "grad_norm": 1.578125,
      "learning_rate": 3.43648858095992e-05,
      "loss": 2.5437,
      "step": 1288
    },
    {
      "epoch": 0.25100773080443206,
      "grad_norm": 1.53125,
      "learning_rate": 3.435630254180835e-05,
      "loss": 2.5391,
      "step": 1289
    },
    {
      "epoch": 0.2512024613946605,
      "grad_norm": 1.171875,
      "learning_rate": 3.43477138157608e-05,
      "loss": 2.53,
      "step": 1290
    },
    {
      "epoch": 0.2513971919848889,
      "grad_norm": 1.40625,
      "learning_rate": 3.4339119634721974e-05,
      "loss": 2.5319,
      "step": 1291
    },
    {
      "epoch": 0.2515919225751173,
      "grad_norm": 0.94140625,
      "learning_rate": 3.4330520001959364e-05,
      "loss": 2.5236,
      "step": 1292
    },
    {
      "epoch": 0.25178665316534576,
      "grad_norm": 1.734375,
      "learning_rate": 3.432191492074256e-05,
      "loss": 2.5267,
      "step": 1293
    },
    {
      "epoch": 0.25198138375557416,
      "grad_norm": 1.0234375,
      "learning_rate": 3.43133043943432e-05,
      "loss": 2.5402,
      "step": 1294
    },
    {
      "epoch": 0.25217611434580256,
      "grad_norm": 2.328125,
      "learning_rate": 3.4304688426035005e-05,
      "loss": 2.5507,
      "step": 1295
    },
    {
      "epoch": 0.252370844936031,
      "grad_norm": 2.0,
      "learning_rate": 3.429606701909375e-05,
      "loss": 2.5209,
      "step": 1296
    },
    {
      "epoch": 0.2525655755262594,
      "grad_norm": 1.7578125,
      "learning_rate": 3.42874401767973e-05,
      "loss": 2.5277,
      "step": 1297
    },
    {
      "epoch": 0.25276030611648787,
      "grad_norm": 1.7421875,
      "learning_rate": 3.427880790242557e-05,
      "loss": 2.5414,
      "step": 1298
    },
    {
      "epoch": 0.25295503670671626,
      "grad_norm": 1.5546875,
      "learning_rate": 3.427017019926055e-05,
      "loss": 2.5516,
      "step": 1299
    },
    {
      "epoch": 0.25314976729694466,
      "grad_norm": 1.375,
      "learning_rate": 3.426152707058628e-05,
      "loss": 2.5162,
      "step": 1300
    },
    {
      "epoch": 0.2533444978871731,
      "grad_norm": 1.5390625,
      "learning_rate": 3.425287851968888e-05,
      "loss": 2.5242,
      "step": 1301
    },
    {
      "epoch": 0.2535392284774015,
      "grad_norm": 1.3046875,
      "learning_rate": 3.424422454985651e-05,
      "loss": 2.5425,
      "step": 1302
    },
    {
      "epoch": 0.2537339590676299,
      "grad_norm": 1.578125,
      "learning_rate": 3.423556516437943e-05,
      "loss": 2.5286,
      "step": 1303
    },
    {
      "epoch": 0.25392868965785836,
      "grad_norm": 1.3671875,
      "learning_rate": 3.4226900366549906e-05,
      "loss": 2.5128,
      "step": 1304
    },
    {
      "epoch": 0.25412342024808676,
      "grad_norm": 1.5546875,
      "learning_rate": 3.4218230159662294e-05,
      "loss": 2.5208,
      "step": 1305
    },
    {
      "epoch": 0.2543181508383152,
      "grad_norm": 1.390625,
      "learning_rate": 3.420955454701301e-05,
      "loss": 2.5447,
      "step": 1306
    },
    {
      "epoch": 0.2545128814285436,
      "grad_norm": 1.609375,
      "learning_rate": 3.420087353190051e-05,
      "loss": 2.5309,
      "step": 1307
    },
    {
      "epoch": 0.254707612018772,
      "grad_norm": 1.4921875,
      "learning_rate": 3.4192187117625316e-05,
      "loss": 2.5426,
      "step": 1308
    },
    {
      "epoch": 0.25490234260900047,
      "grad_norm": 1.6015625,
      "learning_rate": 3.4183495307489984e-05,
      "loss": 2.5283,
      "step": 1309
    },
    {
      "epoch": 0.25509707319922886,
      "grad_norm": 1.4296875,
      "learning_rate": 3.417479810479915e-05,
      "loss": 2.5168,
      "step": 1310
    },
    {
      "epoch": 0.25529180378945726,
      "grad_norm": 1.6796875,
      "learning_rate": 3.4166095512859474e-05,
      "loss": 2.5224,
      "step": 1311
    },
    {
      "epoch": 0.2554865343796857,
      "grad_norm": 1.421875,
      "learning_rate": 3.415738753497968e-05,
      "loss": 2.5223,
      "step": 1312
    },
    {
      "epoch": 0.2556812649699141,
      "grad_norm": 1.625,
      "learning_rate": 3.4148674174470535e-05,
      "loss": 2.5325,
      "step": 1313
    },
    {
      "epoch": 0.25587599556014257,
      "grad_norm": 1.484375,
      "learning_rate": 3.413995543464486e-05,
      "loss": 2.5209,
      "step": 1314
    },
    {
      "epoch": 0.25607072615037096,
      "grad_norm": 1.640625,
      "learning_rate": 3.41312313188175e-05,
      "loss": 2.5103,
      "step": 1315
    },
    {
      "epoch": 0.25626545674059936,
      "grad_norm": 1.421875,
      "learning_rate": 3.412250183030536e-05,
      "loss": 2.5204,
      "step": 1316
    },
    {
      "epoch": 0.2564601873308278,
      "grad_norm": 1.5390625,
      "learning_rate": 3.411376697242739e-05,
      "loss": 2.5198,
      "step": 1317
    },
    {
      "epoch": 0.2566549179210562,
      "grad_norm": 1.453125,
      "learning_rate": 3.410502674850459e-05,
      "loss": 2.4981,
      "step": 1318
    },
    {
      "epoch": 0.2568496485112846,
      "grad_norm": 1.46875,
      "learning_rate": 3.409628116185996e-05,
      "loss": 2.5237,
      "step": 1319
    },
    {
      "epoch": 0.25704437910151307,
      "grad_norm": 1.3359375,
      "learning_rate": 3.408753021581859e-05,
      "loss": 2.5228,
      "step": 1320
    },
    {
      "epoch": 0.25723910969174146,
      "grad_norm": 1.5234375,
      "learning_rate": 3.4078773913707575e-05,
      "loss": 2.5192,
      "step": 1321
    },
    {
      "epoch": 0.2574338402819699,
      "grad_norm": 1.40625,
      "learning_rate": 3.407001225885606e-05,
      "loss": 2.5046,
      "step": 1322
    },
    {
      "epoch": 0.2576285708721983,
      "grad_norm": 1.421875,
      "learning_rate": 3.406124525459521e-05,
      "loss": 2.5172,
      "step": 1323
    },
    {
      "epoch": 0.2578233014624267,
      "grad_norm": 1.28125,
      "learning_rate": 3.405247290425823e-05,
      "loss": 2.5406,
      "step": 1324
    },
    {
      "epoch": 0.25801803205265517,
      "grad_norm": 1.4296875,
      "learning_rate": 3.4043695211180384e-05,
      "loss": 2.5208,
      "step": 1325
    },
    {
      "epoch": 0.25821276264288356,
      "grad_norm": 1.2890625,
      "learning_rate": 3.4034912178698935e-05,
      "loss": 2.5175,
      "step": 1326
    },
    {
      "epoch": 0.25840749323311196,
      "grad_norm": 1.5078125,
      "learning_rate": 3.402612381015318e-05,
      "loss": 2.5231,
      "step": 1327
    },
    {
      "epoch": 0.2586022238233404,
      "grad_norm": 1.296875,
      "learning_rate": 3.401733010888445e-05,
      "loss": 2.5194,
      "step": 1328
    },
    {
      "epoch": 0.2587969544135688,
      "grad_norm": 1.4140625,
      "learning_rate": 3.400853107823611e-05,
      "loss": 2.5223,
      "step": 1329
    },
    {
      "epoch": 0.25899168500379727,
      "grad_norm": 1.265625,
      "learning_rate": 3.399972672155356e-05,
      "loss": 2.5201,
      "step": 1330
    },
    {
      "epoch": 0.25918641559402567,
      "grad_norm": 1.375,
      "learning_rate": 3.399091704218417e-05,
      "loss": 2.5181,
      "step": 1331
    },
    {
      "epoch": 0.25938114618425406,
      "grad_norm": 1.3125,
      "learning_rate": 3.398210204347742e-05,
      "loss": 2.5382,
      "step": 1332
    },
    {
      "epoch": 0.2595758767744825,
      "grad_norm": 1.453125,
      "learning_rate": 3.397328172878473e-05,
      "loss": 2.5277,
      "step": 1333
    },
    {
      "epoch": 0.2597706073647109,
      "grad_norm": 1.375,
      "learning_rate": 3.396445610145961e-05,
      "loss": 2.5216,
      "step": 1334
    },
    {
      "epoch": 0.25996533795493937,
      "grad_norm": 1.3671875,
      "learning_rate": 3.3955625164857534e-05,
      "loss": 2.5368,
      "step": 1335
    },
    {
      "epoch": 0.26016006854516777,
      "grad_norm": 1.3046875,
      "learning_rate": 3.3946788922336025e-05,
      "loss": 2.5166,
      "step": 1336
    },
    {
      "epoch": 0.26035479913539616,
      "grad_norm": 1.2578125,
      "learning_rate": 3.3937947377254625e-05,
      "loss": 2.5293,
      "step": 1337
    },
    {
      "epoch": 0.2605495297256246,
      "grad_norm": 1.125,
      "learning_rate": 3.3929100532974876e-05,
      "loss": 2.5237,
      "step": 1338
    },
    {
      "epoch": 0.260744260315853,
      "grad_norm": 1.4140625,
      "learning_rate": 3.392024839286033e-05,
      "loss": 2.5181,
      "step": 1339
    },
    {
      "epoch": 0.2609389909060814,
      "grad_norm": 1.21875,
      "learning_rate": 3.391139096027659e-05,
      "loss": 2.5172,
      "step": 1340
    },
    {
      "epoch": 0.26113372149630987,
      "grad_norm": 1.6484375,
      "learning_rate": 3.390252823859123e-05,
      "loss": 2.5352,
      "step": 1341
    },
    {
      "epoch": 0.26132845208653827,
      "grad_norm": 1.5546875,
      "learning_rate": 3.3893660231173846e-05,
      "loss": 2.5215,
      "step": 1342
    },
    {
      "epoch": 0.2615231826767667,
      "grad_norm": 1.1484375,
      "learning_rate": 3.388478694139606e-05,
      "loss": 2.5187,
      "step": 1343
    },
    {
      "epoch": 0.2617179132669951,
      "grad_norm": 1.234375,
      "learning_rate": 3.387590837263149e-05,
      "loss": 2.5145,
      "step": 1344
    },
    {
      "epoch": 0.2619126438572235,
      "grad_norm": 1.046875,
      "learning_rate": 3.386702452825576e-05,
      "loss": 2.5053,
      "step": 1345
    },
    {
      "epoch": 0.26210737444745197,
      "grad_norm": 0.97265625,
      "learning_rate": 3.385813541164651e-05,
      "loss": 2.5197,
      "step": 1346
    },
    {
      "epoch": 0.26230210503768037,
      "grad_norm": 1.0546875,
      "learning_rate": 3.3849241026183354e-05,
      "loss": 2.5188,
      "step": 1347
    },
    {
      "epoch": 0.26249683562790876,
      "grad_norm": 0.80078125,
      "learning_rate": 3.384034137524795e-05,
      "loss": 2.5039,
      "step": 1348
    },
    {
      "epoch": 0.2626915662181372,
      "grad_norm": 0.92578125,
      "learning_rate": 3.3831436462223936e-05,
      "loss": 2.5147,
      "step": 1349
    },
    {
      "epoch": 0.2628862968083656,
      "grad_norm": 0.82421875,
      "learning_rate": 3.382252629049695e-05,
      "loss": 2.5333,
      "step": 1350
    },
    {
      "epoch": 0.26308102739859407,
      "grad_norm": 0.82421875,
      "learning_rate": 3.381361086345464e-05,
      "loss": 2.5165,
      "step": 1351
    },
    {
      "epoch": 0.26327575798882247,
      "grad_norm": 0.85546875,
      "learning_rate": 3.380469018448663e-05,
      "loss": 2.5211,
      "step": 1352
    },
    {
      "epoch": 0.26347048857905087,
      "grad_norm": 0.91796875,
      "learning_rate": 3.379576425698458e-05,
      "loss": 2.5172,
      "step": 1353
    },
    {
      "epoch": 0.2636652191692793,
      "grad_norm": 0.765625,
      "learning_rate": 3.37868330843421e-05,
      "loss": 2.5153,
      "step": 1354
    },
    {
      "epoch": 0.2638599497595077,
      "grad_norm": 0.91796875,
      "learning_rate": 3.377789666995483e-05,
      "loss": 2.5338,
      "step": 1355
    },
    {
      "epoch": 0.2640546803497361,
      "grad_norm": 0.8828125,
      "learning_rate": 3.376895501722039e-05,
      "loss": 2.5032,
      "step": 1356
    },
    {
      "epoch": 0.26424941093996457,
      "grad_norm": 0.890625,
      "learning_rate": 3.376000812953837e-05,
      "loss": 2.5331,
      "step": 1357
    },
    {
      "epoch": 0.26444414153019297,
      "grad_norm": 1.171875,
      "learning_rate": 3.375105601031039e-05,
      "loss": 2.5193,
      "step": 1358
    },
    {
      "epoch": 0.2646388721204214,
      "grad_norm": 0.74609375,
      "learning_rate": 3.3742098662940044e-05,
      "loss": 2.5191,
      "step": 1359
    },
    {
      "epoch": 0.2648336027106498,
      "grad_norm": 0.83203125,
      "learning_rate": 3.3733136090832895e-05,
      "loss": 2.5245,
      "step": 1360
    },
    {
      "epoch": 0.2650283333008782,
      "grad_norm": 0.88671875,
      "learning_rate": 3.372416829739651e-05,
      "loss": 2.5178,
      "step": 1361
    },
    {
      "epoch": 0.26522306389110667,
      "grad_norm": 0.96875,
      "learning_rate": 3.371519528604044e-05,
      "loss": 2.5353,
      "step": 1362
    },
    {
      "epoch": 0.26541779448133507,
      "grad_norm": 0.88671875,
      "learning_rate": 3.370621706017622e-05,
      "loss": 2.5245,
      "step": 1363
    },
    {
      "epoch": 0.26561252507156347,
      "grad_norm": 0.80078125,
      "learning_rate": 3.369723362321736e-05,
      "loss": 2.5339,
      "step": 1364
    },
    {
      "epoch": 0.2658072556617919,
      "grad_norm": 0.84765625,
      "learning_rate": 3.368824497857937e-05,
      "loss": 2.5104,
      "step": 1365
    },
    {
      "epoch": 0.2660019862520203,
      "grad_norm": 0.890625,
      "learning_rate": 3.3679251129679715e-05,
      "loss": 2.5147,
      "step": 1366
    },
    {
      "epoch": 0.26619671684224877,
      "grad_norm": 0.78515625,
      "learning_rate": 3.367025207993785e-05,
      "loss": 2.5338,
      "step": 1367
    },
    {
      "epoch": 0.26639144743247717,
      "grad_norm": 0.72265625,
      "learning_rate": 3.3661247832775205e-05,
      "loss": 2.5138,
      "step": 1368
    },
    {
      "epoch": 0.26658617802270557,
      "grad_norm": 0.69921875,
      "learning_rate": 3.365223839161521e-05,
      "loss": 2.5462,
      "step": 1369
    },
    {
      "epoch": 0.266780908612934,
      "grad_norm": 0.7421875,
      "learning_rate": 3.3643223759883224e-05,
      "loss": 2.5304,
      "step": 1370
    },
    {
      "epoch": 0.2669756392031624,
      "grad_norm": 0.78515625,
      "learning_rate": 3.36342039410066e-05,
      "loss": 2.5402,
      "step": 1371
    },
    {
      "epoch": 0.2671703697933908,
      "grad_norm": 0.6875,
      "learning_rate": 3.36251789384147e-05,
      "loss": 2.5278,
      "step": 1372
    },
    {
      "epoch": 0.26736510038361927,
      "grad_norm": 0.7265625,
      "learning_rate": 3.3616148755538796e-05,
      "loss": 2.537,
      "step": 1373
    },
    {
      "epoch": 0.26755983097384767,
      "grad_norm": 0.82421875,
      "learning_rate": 3.360711339581217e-05,
      "loss": 2.5378,
      "step": 1374
    },
    {
      "epoch": 0.2677545615640761,
      "grad_norm": 0.73046875,
      "learning_rate": 3.359807286267005e-05,
      "loss": 2.5323,
      "step": 1375
    },
    {
      "epoch": 0.2679492921543045,
      "grad_norm": 0.83203125,
      "learning_rate": 3.358902715954965e-05,
      "loss": 2.5158,
      "step": 1376
    },
    {
      "epoch": 0.2681440227445329,
      "grad_norm": 1.015625,
      "learning_rate": 3.357997628989013e-05,
      "loss": 2.5322,
      "step": 1377
    },
    {
      "epoch": 0.26833875333476137,
      "grad_norm": 1.0078125,
      "learning_rate": 3.3570920257132635e-05,
      "loss": 2.5121,
      "step": 1378
    },
    {
      "epoch": 0.26853348392498977,
      "grad_norm": 0.8828125,
      "learning_rate": 3.356185906472026e-05,
      "loss": 2.5086,
      "step": 1379
    },
    {
      "epoch": 0.2687282145152182,
      "grad_norm": 0.81640625,
      "learning_rate": 3.3552792716098066e-05,
      "loss": 2.5292,
      "step": 1380
    },
    {
      "epoch": 0.2689229451054466,
      "grad_norm": 0.8359375,
      "learning_rate": 3.354372121471306e-05,
      "loss": 2.5161,
      "step": 1381
    },
    {
      "epoch": 0.269117675695675,
      "grad_norm": 0.76953125,
      "learning_rate": 3.3534644564014245e-05,
      "loss": 2.5252,
      "step": 1382
    },
    {
      "epoch": 0.26931240628590347,
      "grad_norm": 0.83984375,
      "learning_rate": 3.3525562767452534e-05,
      "loss": 2.5215,
      "step": 1383
    },
    {
      "epoch": 0.26950713687613187,
      "grad_norm": 0.84375,
      "learning_rate": 3.351647582848083e-05,
      "loss": 2.5048,
      "step": 1384
    },
    {
      "epoch": 0.26970186746636027,
      "grad_norm": 0.81640625,
      "learning_rate": 3.3507383750553974e-05,
      "loss": 2.519,
      "step": 1385
    },
    {
      "epoch": 0.2698965980565887,
      "grad_norm": 0.7578125,
      "learning_rate": 3.3498286537128785e-05,
      "loss": 2.5113,
      "step": 1386
    },
    {
      "epoch": 0.2700913286468171,
      "grad_norm": 0.67578125,
      "learning_rate": 3.3489184191664e-05,
      "loss": 2.518,
      "step": 1387
    },
    {
      "epoch": 0.2702860592370456,
      "grad_norm": 0.8359375,
      "learning_rate": 3.348007671762034e-05,
      "loss": 2.5237,
      "step": 1388
    },
    {
      "epoch": 0.27048078982727397,
      "grad_norm": 0.828125,
      "learning_rate": 3.3470964118460444e-05,
      "loss": 2.5113,
      "step": 1389
    },
    {
      "epoch": 0.27067552041750237,
      "grad_norm": 0.8515625,
      "learning_rate": 3.346184639764893e-05,
      "loss": 2.5083,
      "step": 1390
    },
    {
      "epoch": 0.2708702510077308,
      "grad_norm": 0.7109375,
      "learning_rate": 3.345272355865234e-05,
      "loss": 2.5148,
      "step": 1391
    },
    {
      "epoch": 0.2710649815979592,
      "grad_norm": 0.69140625,
      "learning_rate": 3.344359560493918e-05,
      "loss": 2.531,
      "step": 1392
    },
    {
      "epoch": 0.2712597121881876,
      "grad_norm": 0.81640625,
      "learning_rate": 3.343446253997988e-05,
      "loss": 2.5303,
      "step": 1393
    },
    {
      "epoch": 0.27145444277841607,
      "grad_norm": 0.75390625,
      "learning_rate": 3.3425324367246834e-05,
      "loss": 2.5288,
      "step": 1394
    },
    {
      "epoch": 0.27164917336864447,
      "grad_norm": 0.703125,
      "learning_rate": 3.341618109021438e-05,
      "loss": 2.5294,
      "step": 1395
    },
    {
      "epoch": 0.2718439039588729,
      "grad_norm": 0.76171875,
      "learning_rate": 3.3407032712358755e-05,
      "loss": 2.5038,
      "step": 1396
    },
    {
      "epoch": 0.2720386345491013,
      "grad_norm": 0.70703125,
      "learning_rate": 3.33978792371582e-05,
      "loss": 2.5252,
      "step": 1397
    },
    {
      "epoch": 0.2722333651393297,
      "grad_norm": 0.71484375,
      "learning_rate": 3.338872066809283e-05,
      "loss": 2.5118,
      "step": 1398
    },
    {
      "epoch": 0.2724280957295582,
      "grad_norm": 0.73828125,
      "learning_rate": 3.3379557008644754e-05,
      "loss": 2.5309,
      "step": 1399
    },
    {
      "epoch": 0.27262282631978657,
      "grad_norm": 0.75390625,
      "learning_rate": 3.337038826229797e-05,
      "loss": 2.5421,
      "step": 1400
    },
    {
      "epoch": 0.27281755691001497,
      "grad_norm": 0.7265625,
      "learning_rate": 3.336121443253844e-05,
      "loss": 2.5019,
      "step": 1401
    },
    {
      "epoch": 0.2730122875002434,
      "grad_norm": 0.75390625,
      "learning_rate": 3.335203552285404e-05,
      "loss": 2.5362,
      "step": 1402
    },
    {
      "epoch": 0.2732070180904718,
      "grad_norm": 0.6953125,
      "learning_rate": 3.334285153673459e-05,
      "loss": 2.52,
      "step": 1403
    },
    {
      "epoch": 0.2734017486807003,
      "grad_norm": 0.7109375,
      "learning_rate": 3.3333662477671835e-05,
      "loss": 2.5193,
      "step": 1404
    },
    {
      "epoch": 0.27359647927092867,
      "grad_norm": 0.81640625,
      "learning_rate": 3.332446834915946e-05,
      "loss": 2.509,
      "step": 1405
    },
    {
      "epoch": 0.27379120986115707,
      "grad_norm": 0.63671875,
      "learning_rate": 3.331526915469304e-05,
      "loss": 2.4967,
      "step": 1406
    },
    {
      "epoch": 0.2739859404513855,
      "grad_norm": 0.7109375,
      "learning_rate": 3.3306064897770124e-05,
      "loss": 2.5192,
      "step": 1407
    },
    {
      "epoch": 0.2741806710416139,
      "grad_norm": 0.68359375,
      "learning_rate": 3.329685558189016e-05,
      "loss": 2.509,
      "step": 1408
    },
    {
      "epoch": 0.2743754016318423,
      "grad_norm": 0.6796875,
      "learning_rate": 3.3287641210554523e-05,
      "loss": 2.5219,
      "step": 1409
    },
    {
      "epoch": 0.2745701322220708,
      "grad_norm": 0.75,
      "learning_rate": 3.327842178726651e-05,
      "loss": 2.5255,
      "step": 1410
    },
    {
      "epoch": 0.27476486281229917,
      "grad_norm": 0.6171875,
      "learning_rate": 3.326919731553134e-05,
      "loss": 2.5107,
      "step": 1411
    },
    {
      "epoch": 0.2749595934025276,
      "grad_norm": 0.71875,
      "learning_rate": 3.325996779885615e-05,
      "loss": 2.5255,
      "step": 1412
    },
    {
      "epoch": 0.275154323992756,
      "grad_norm": 0.65234375,
      "learning_rate": 3.3250733240750006e-05,
      "loss": 2.5171,
      "step": 1413
    },
    {
      "epoch": 0.2753490545829844,
      "grad_norm": 0.765625,
      "learning_rate": 3.324149364472386e-05,
      "loss": 2.5173,
      "step": 1414
    },
    {
      "epoch": 0.2755437851732129,
      "grad_norm": 0.73046875,
      "learning_rate": 3.323224901429062e-05,
      "loss": 2.4991,
      "step": 1415
    },
    {
      "epoch": 0.27573851576344127,
      "grad_norm": 0.7109375,
      "learning_rate": 3.3222999352965074e-05,
      "loss": 2.5214,
      "step": 1416
    },
    {
      "epoch": 0.2759332463536697,
      "grad_norm": 0.76171875,
      "learning_rate": 3.321374466426395e-05,
      "loss": 2.504,
      "step": 1417
    },
    {
      "epoch": 0.2761279769438981,
      "grad_norm": 0.71484375,
      "learning_rate": 3.3204484951705864e-05,
      "loss": 2.4994,
      "step": 1418
    },
    {
      "epoch": 0.2763227075341265,
      "grad_norm": 0.73046875,
      "learning_rate": 3.3195220218811356e-05,
      "loss": 2.5214,
      "step": 1419
    },
    {
      "epoch": 0.276517438124355,
      "grad_norm": 0.61328125,
      "learning_rate": 3.318595046910288e-05,
      "loss": 2.517,
      "step": 1420
    },
    {
      "epoch": 0.2767121687145834,
      "grad_norm": 0.734375,
      "learning_rate": 3.317667570610477e-05,
      "loss": 2.5069,
      "step": 1421
    },
    {
      "epoch": 0.27690689930481177,
      "grad_norm": 0.77734375,
      "learning_rate": 3.3167395933343304e-05,
      "loss": 2.5204,
      "step": 1422
    },
    {
      "epoch": 0.2771016298950402,
      "grad_norm": 0.68359375,
      "learning_rate": 3.3158111154346635e-05,
      "loss": 2.5154,
      "step": 1423
    },
    {
      "epoch": 0.2772963604852686,
      "grad_norm": 0.7890625,
      "learning_rate": 3.314882137264483e-05,
      "loss": 2.5086,
      "step": 1424
    },
    {
      "epoch": 0.2774910910754971,
      "grad_norm": 0.74609375,
      "learning_rate": 3.313952659176986e-05,
      "loss": 2.5278,
      "step": 1425
    },
    {
      "epoch": 0.2776858216657255,
      "grad_norm": 0.6484375,
      "learning_rate": 3.31302268152556e-05,
      "loss": 2.5336,
      "step": 1426
    },
    {
      "epoch": 0.27788055225595387,
      "grad_norm": 0.7421875,
      "learning_rate": 3.31209220466378e-05,
      "loss": 2.5174,
      "step": 1427
    },
    {
      "epoch": 0.2780752828461823,
      "grad_norm": 0.70703125,
      "learning_rate": 3.311161228945415e-05,
      "loss": 2.5108,
      "step": 1428
    },
    {
      "epoch": 0.2782700134364107,
      "grad_norm": 0.60546875,
      "learning_rate": 3.3102297547244204e-05,
      "loss": 2.507,
      "step": 1429
    },
    {
      "epoch": 0.2784647440266391,
      "grad_norm": 0.69140625,
      "learning_rate": 3.309297782354942e-05,
      "loss": 2.5045,
      "step": 1430
    },
    {
      "epoch": 0.2786594746168676,
      "grad_norm": 0.703125,
      "learning_rate": 3.308365312191315e-05,
      "loss": 2.5147,
      "step": 1431
    },
    {
      "epoch": 0.278854205207096,
      "grad_norm": 0.78515625,
      "learning_rate": 3.3074323445880636e-05,
      "loss": 2.5291,
      "step": 1432
    },
    {
      "epoch": 0.2790489357973244,
      "grad_norm": 0.734375,
      "learning_rate": 3.306498879899902e-05,
      "loss": 2.5039,
      "step": 1433
    },
    {
      "epoch": 0.2792436663875528,
      "grad_norm": 0.62890625,
      "learning_rate": 3.3055649184817334e-05,
      "loss": 2.5414,
      "step": 1434
    },
    {
      "epoch": 0.2794383969777812,
      "grad_norm": 0.70703125,
      "learning_rate": 3.3046304606886476e-05,
      "loss": 2.5103,
      "step": 1435
    },
    {
      "epoch": 0.2796331275680097,
      "grad_norm": 0.71484375,
      "learning_rate": 3.3036955068759274e-05,
      "loss": 2.5089,
      "step": 1436
    },
    {
      "epoch": 0.2798278581582381,
      "grad_norm": 0.6796875,
      "learning_rate": 3.302760057399039e-05,
      "loss": 2.5126,
      "step": 1437
    },
    {
      "epoch": 0.28002258874846647,
      "grad_norm": 0.76171875,
      "learning_rate": 3.301824112613641e-05,
      "loss": 2.5227,
      "step": 1438
    },
    {
      "epoch": 0.2802173193386949,
      "grad_norm": 0.734375,
      "learning_rate": 3.300887672875579e-05,
      "loss": 2.5204,
      "step": 1439
    },
    {
      "epoch": 0.2804120499289233,
      "grad_norm": 0.6796875,
      "learning_rate": 3.299950738540886e-05,
      "loss": 2.5151,
      "step": 1440
    },
    {
      "epoch": 0.2806067805191518,
      "grad_norm": 0.6875,
      "learning_rate": 3.2990133099657856e-05,
      "loss": 2.5106,
      "step": 1441
    },
    {
      "epoch": 0.2808015111093802,
      "grad_norm": 0.76953125,
      "learning_rate": 3.298075387506685e-05,
      "loss": 2.5214,
      "step": 1442
    },
    {
      "epoch": 0.2809962416996086,
      "grad_norm": 0.73046875,
      "learning_rate": 3.297136971520184e-05,
      "loss": 2.5067,
      "step": 1443
    },
    {
      "epoch": 0.281190972289837,
      "grad_norm": 0.7421875,
      "learning_rate": 3.2961980623630656e-05,
      "loss": 2.5185,
      "step": 1444
    },
    {
      "epoch": 0.2813857028800654,
      "grad_norm": 0.640625,
      "learning_rate": 3.295258660392304e-05,
      "loss": 2.4911,
      "step": 1445
    },
    {
      "epoch": 0.2815804334702938,
      "grad_norm": 0.796875,
      "learning_rate": 3.2943187659650595e-05,
      "loss": 2.5035,
      "step": 1446
    },
    {
      "epoch": 0.2817751640605223,
      "grad_norm": 0.6875,
      "learning_rate": 3.293378379438678e-05,
      "loss": 2.4933,
      "step": 1447
    },
    {
      "epoch": 0.2819698946507507,
      "grad_norm": 0.6484375,
      "learning_rate": 3.292437501170694e-05,
      "loss": 2.5205,
      "step": 1448
    },
    {
      "epoch": 0.28216462524097913,
      "grad_norm": 0.67578125,
      "learning_rate": 3.29149613151883e-05,
      "loss": 2.4911,
      "step": 1449
    },
    {
      "epoch": 0.2823593558312075,
      "grad_norm": 0.6171875,
      "learning_rate": 3.290554270840993e-05,
      "loss": 2.4942,
      "step": 1450
    },
    {
      "epoch": 0.2825540864214359,
      "grad_norm": 0.63671875,
      "learning_rate": 3.289611919495278e-05,
      "loss": 2.5014,
      "step": 1451
    },
    {
      "epoch": 0.2827488170116644,
      "grad_norm": 0.6015625,
      "learning_rate": 3.288669077839966e-05,
      "loss": 2.5213,
      "step": 1452
    },
    {
      "epoch": 0.2829435476018928,
      "grad_norm": 0.671875,
      "learning_rate": 3.287725746233527e-05,
      "loss": 2.4945,
      "step": 1453
    },
    {
      "epoch": 0.2831382781921212,
      "grad_norm": 0.65625,
      "learning_rate": 3.286781925034612e-05,
      "loss": 2.4997,
      "step": 1454
    },
    {
      "epoch": 0.2833330087823496,
      "grad_norm": 0.7421875,
      "learning_rate": 3.2858376146020624e-05,
      "loss": 2.509,
      "step": 1455
    },
    {
      "epoch": 0.283527739372578,
      "grad_norm": 0.66015625,
      "learning_rate": 3.284892815294905e-05,
      "loss": 2.5077,
      "step": 1456
    },
    {
      "epoch": 0.2837224699628065,
      "grad_norm": 0.6953125,
      "learning_rate": 3.2839475274723516e-05,
      "loss": 2.513,
      "step": 1457
    },
    {
      "epoch": 0.2839172005530349,
      "grad_norm": 0.75390625,
      "learning_rate": 3.2830017514937996e-05,
      "loss": 2.5095,
      "step": 1458
    },
    {
      "epoch": 0.2841119311432633,
      "grad_norm": 0.6796875,
      "learning_rate": 3.2820554877188326e-05,
      "loss": 2.5093,
      "step": 1459
    },
    {
      "epoch": 0.28430666173349173,
      "grad_norm": 0.73828125,
      "learning_rate": 3.281108736507219e-05,
      "loss": 2.4989,
      "step": 1460
    },
    {
      "epoch": 0.2845013923237201,
      "grad_norm": 0.80859375,
      "learning_rate": 3.280161498218914e-05,
      "loss": 2.5049,
      "step": 1461
    },
    {
      "epoch": 0.2846961229139486,
      "grad_norm": 0.6953125,
      "learning_rate": 3.279213773214056e-05,
      "loss": 2.4879,
      "step": 1462
    },
    {
      "epoch": 0.284890853504177,
      "grad_norm": 0.71875,
      "learning_rate": 3.2782655618529707e-05,
      "loss": 2.5252,
      "step": 1463
    },
    {
      "epoch": 0.2850855840944054,
      "grad_norm": 0.74609375,
      "learning_rate": 3.2773168644961656e-05,
      "loss": 2.5177,
      "step": 1464
    },
    {
      "epoch": 0.28528031468463383,
      "grad_norm": 0.6875,
      "learning_rate": 3.2763676815043366e-05,
      "loss": 2.5225,
      "step": 1465
    },
    {
      "epoch": 0.2854750452748622,
      "grad_norm": 0.76953125,
      "learning_rate": 3.2754180132383604e-05,
      "loss": 2.5187,
      "step": 1466
    },
    {
      "epoch": 0.2856697758650906,
      "grad_norm": 0.70703125,
      "learning_rate": 3.2744678600593026e-05,
      "loss": 2.5111,
      "step": 1467
    },
    {
      "epoch": 0.2858645064553191,
      "grad_norm": 0.6484375,
      "learning_rate": 3.2735172223284087e-05,
      "loss": 2.4721,
      "step": 1468
    },
    {
      "epoch": 0.2860592370455475,
      "grad_norm": 0.734375,
      "learning_rate": 3.272566100407112e-05,
      "loss": 2.5104,
      "step": 1469
    },
    {
      "epoch": 0.28625396763577593,
      "grad_norm": 0.6953125,
      "learning_rate": 3.271614494657028e-05,
      "loss": 2.4985,
      "step": 1470
    },
    {
      "epoch": 0.28644869822600433,
      "grad_norm": 0.6796875,
      "learning_rate": 3.270662405439957e-05,
      "loss": 2.5256,
      "step": 1471
    },
    {
      "epoch": 0.2866434288162327,
      "grad_norm": 0.73828125,
      "learning_rate": 3.2697098331178814e-05,
      "loss": 2.4843,
      "step": 1472
    },
    {
      "epoch": 0.2868381594064612,
      "grad_norm": 0.67578125,
      "learning_rate": 3.268756778052969e-05,
      "loss": 2.505,
      "step": 1473
    },
    {
      "epoch": 0.2870328899966896,
      "grad_norm": 0.73046875,
      "learning_rate": 3.267803240607572e-05,
      "loss": 2.5107,
      "step": 1474
    },
    {
      "epoch": 0.287227620586918,
      "grad_norm": 0.7265625,
      "learning_rate": 3.266849221144223e-05,
      "loss": 2.5136,
      "step": 1475
    },
    {
      "epoch": 0.28742235117714643,
      "grad_norm": 0.765625,
      "learning_rate": 3.265894720025641e-05,
      "loss": 2.508,
      "step": 1476
    },
    {
      "epoch": 0.2876170817673748,
      "grad_norm": 0.6640625,
      "learning_rate": 3.264939737614726e-05,
      "loss": 2.5259,
      "step": 1477
    },
    {
      "epoch": 0.2878118123576033,
      "grad_norm": 0.7421875,
      "learning_rate": 3.2639842742745614e-05,
      "loss": 2.4963,
      "step": 1478
    },
    {
      "epoch": 0.2880065429478317,
      "grad_norm": 0.76171875,
      "learning_rate": 3.263028330368414e-05,
      "loss": 2.5111,
      "step": 1479
    },
    {
      "epoch": 0.2882012735380601,
      "grad_norm": 0.69921875,
      "learning_rate": 3.262071906259732e-05,
      "loss": 2.4866,
      "step": 1480
    },
    {
      "epoch": 0.28839600412828853,
      "grad_norm": 0.83984375,
      "learning_rate": 3.26111500231215e-05,
      "loss": 2.5036,
      "step": 1481
    },
    {
      "epoch": 0.28859073471851693,
      "grad_norm": 0.7890625,
      "learning_rate": 3.26015761888948e-05,
      "loss": 2.5162,
      "step": 1482
    },
    {
      "epoch": 0.2887854653087453,
      "grad_norm": 0.75,
      "learning_rate": 3.2591997563557184e-05,
      "loss": 2.503,
      "step": 1483
    },
    {
      "epoch": 0.2889801958989738,
      "grad_norm": 1.0,
      "learning_rate": 3.258241415075046e-05,
      "loss": 2.4913,
      "step": 1484
    },
    {
      "epoch": 0.2891749264892022,
      "grad_norm": 0.79296875,
      "learning_rate": 3.25728259541182e-05,
      "loss": 2.4995,
      "step": 1485
    },
    {
      "epoch": 0.28936965707943063,
      "grad_norm": 0.68359375,
      "learning_rate": 3.256323297730587e-05,
      "loss": 2.5083,
      "step": 1486
    },
    {
      "epoch": 0.28956438766965903,
      "grad_norm": 0.69921875,
      "learning_rate": 3.2553635223960683e-05,
      "loss": 2.4892,
      "step": 1487
    },
    {
      "epoch": 0.2897591182598874,
      "grad_norm": 0.703125,
      "learning_rate": 3.2544032697731714e-05,
      "loss": 2.499,
      "step": 1488
    },
    {
      "epoch": 0.2899538488501159,
      "grad_norm": 0.6875,
      "learning_rate": 3.253442540226983e-05,
      "loss": 2.4948,
      "step": 1489
    },
    {
      "epoch": 0.2901485794403443,
      "grad_norm": 0.6875,
      "learning_rate": 3.252481334122773e-05,
      "loss": 2.5079,
      "step": 1490
    },
    {
      "epoch": 0.2903433100305727,
      "grad_norm": 0.66015625,
      "learning_rate": 3.25151965182599e-05,
      "loss": 2.5014,
      "step": 1491
    },
    {
      "epoch": 0.29053804062080113,
      "grad_norm": 0.67578125,
      "learning_rate": 3.250557493702265e-05,
      "loss": 2.5138,
      "step": 1492
    },
    {
      "epoch": 0.29073277121102953,
      "grad_norm": 0.6640625,
      "learning_rate": 3.2495948601174106e-05,
      "loss": 2.4943,
      "step": 1493
    },
    {
      "epoch": 0.290927501801258,
      "grad_norm": 0.71484375,
      "learning_rate": 3.248631751437419e-05,
      "loss": 2.5035,
      "step": 1494
    },
    {
      "epoch": 0.2911222323914864,
      "grad_norm": 0.69140625,
      "learning_rate": 3.2476681680284633e-05,
      "loss": 2.4936,
      "step": 1495
    },
    {
      "epoch": 0.2913169629817148,
      "grad_norm": 0.66015625,
      "learning_rate": 3.2467041102568986e-05,
      "loss": 2.5013,
      "step": 1496
    },
    {
      "epoch": 0.29151169357194323,
      "grad_norm": 0.68359375,
      "learning_rate": 3.245739578489257e-05,
      "loss": 2.5022,
      "step": 1497
    },
    {
      "epoch": 0.29170642416217163,
      "grad_norm": 0.67578125,
      "learning_rate": 3.2447745730922546e-05,
      "loss": 2.4883,
      "step": 1498
    },
    {
      "epoch": 0.2919011547524,
      "grad_norm": 0.796875,
      "learning_rate": 3.2438090944327844e-05,
      "loss": 2.5088,
      "step": 1499
    },
    {
      "epoch": 0.2920958853426285,
      "grad_norm": 0.6875,
      "learning_rate": 3.242843142877922e-05,
      "loss": 2.5134,
      "step": 1500
    },
    {
      "epoch": 0.2922906159328569,
      "grad_norm": 0.71484375,
      "learning_rate": 3.24187671879492e-05,
      "loss": 2.4969,
      "step": 1501
    },
    {
      "epoch": 0.29248534652308533,
      "grad_norm": 0.6796875,
      "learning_rate": 3.240909822551214e-05,
      "loss": 2.4935,
      "step": 1502
    },
    {
      "epoch": 0.29268007711331373,
      "grad_norm": 0.76953125,
      "learning_rate": 3.239942454514417e-05,
      "loss": 2.504,
      "step": 1503
    },
    {
      "epoch": 0.29287480770354213,
      "grad_norm": 0.70703125,
      "learning_rate": 3.23897461505232e-05,
      "loss": 2.5054,
      "step": 1504
    },
    {
      "epoch": 0.2930695382937706,
      "grad_norm": 0.828125,
      "learning_rate": 3.2380063045328965e-05,
      "loss": 2.4854,
      "step": 1505
    },
    {
      "epoch": 0.293264268883999,
      "grad_norm": 0.95703125,
      "learning_rate": 3.237037523324297e-05,
      "loss": 2.4904,
      "step": 1506
    },
    {
      "epoch": 0.29345899947422743,
      "grad_norm": 0.85546875,
      "learning_rate": 3.236068271794852e-05,
      "loss": 2.4965,
      "step": 1507
    },
    {
      "epoch": 0.29365373006445583,
      "grad_norm": 0.7578125,
      "learning_rate": 3.2350985503130694e-05,
      "loss": 2.506,
      "step": 1508
    },
    {
      "epoch": 0.29384846065468423,
      "grad_norm": 0.9765625,
      "learning_rate": 3.2341283592476373e-05,
      "loss": 2.4938,
      "step": 1509
    },
    {
      "epoch": 0.2940431912449127,
      "grad_norm": 0.9609375,
      "learning_rate": 3.233157698967421e-05,
      "loss": 2.4953,
      "step": 1510
    },
    {
      "epoch": 0.2942379218351411,
      "grad_norm": 0.67578125,
      "learning_rate": 3.2321865698414665e-05,
      "loss": 2.501,
      "step": 1511
    },
    {
      "epoch": 0.2944326524253695,
      "grad_norm": 0.6953125,
      "learning_rate": 3.231214972238995e-05,
      "loss": 2.4985,
      "step": 1512
    },
    {
      "epoch": 0.29462738301559793,
      "grad_norm": 0.7421875,
      "learning_rate": 3.230242906529407e-05,
      "loss": 2.521,
      "step": 1513
    },
    {
      "epoch": 0.29482211360582633,
      "grad_norm": 0.625,
      "learning_rate": 3.2292703730822824e-05,
      "loss": 2.508,
      "step": 1514
    },
    {
      "epoch": 0.2950168441960548,
      "grad_norm": 0.69921875,
      "learning_rate": 3.2282973722673774e-05,
      "loss": 2.5022,
      "step": 1515
    },
    {
      "epoch": 0.2952115747862832,
      "grad_norm": 0.76171875,
      "learning_rate": 3.227323904454626e-05,
      "loss": 2.4977,
      "step": 1516
    },
    {
      "epoch": 0.2954063053765116,
      "grad_norm": 0.66015625,
      "learning_rate": 3.226349970014139e-05,
      "loss": 2.4912,
      "step": 1517
    },
    {
      "epoch": 0.29560103596674003,
      "grad_norm": 0.6953125,
      "learning_rate": 3.225375569316208e-05,
      "loss": 2.4949,
      "step": 1518
    },
    {
      "epoch": 0.29579576655696843,
      "grad_norm": 0.7890625,
      "learning_rate": 3.2244007027312975e-05,
      "loss": 2.522,
      "step": 1519
    },
    {
      "epoch": 0.29599049714719683,
      "grad_norm": 0.6484375,
      "learning_rate": 3.2234253706300526e-05,
      "loss": 2.475,
      "step": 1520
    },
    {
      "epoch": 0.2961852277374253,
      "grad_norm": 0.65625,
      "learning_rate": 3.2224495733832926e-05,
      "loss": 2.503,
      "step": 1521
    },
    {
      "epoch": 0.2963799583276537,
      "grad_norm": 0.68359375,
      "learning_rate": 3.221473311362016e-05,
      "loss": 2.5028,
      "step": 1522
    },
    {
      "epoch": 0.29657468891788213,
      "grad_norm": 0.71875,
      "learning_rate": 3.220496584937396e-05,
      "loss": 2.4968,
      "step": 1523
    },
    {
      "epoch": 0.29676941950811053,
      "grad_norm": 0.67578125,
      "learning_rate": 3.2195193944807845e-05,
      "loss": 2.5296,
      "step": 1524
    },
    {
      "epoch": 0.29696415009833893,
      "grad_norm": 0.7265625,
      "learning_rate": 3.2185417403637076e-05,
      "loss": 2.5014,
      "step": 1525
    },
    {
      "epoch": 0.2971588806885674,
      "grad_norm": 0.6484375,
      "learning_rate": 3.2175636229578695e-05,
      "loss": 2.483,
      "step": 1526
    },
    {
      "epoch": 0.2973536112787958,
      "grad_norm": 0.6953125,
      "learning_rate": 3.216585042635149e-05,
      "loss": 2.49,
      "step": 1527
    },
    {
      "epoch": 0.2975483418690242,
      "grad_norm": 0.71484375,
      "learning_rate": 3.2156059997676025e-05,
      "loss": 2.4916,
      "step": 1528
    },
    {
      "epoch": 0.29774307245925263,
      "grad_norm": 0.734375,
      "learning_rate": 3.214626494727461e-05,
      "loss": 2.4823,
      "step": 1529
    },
    {
      "epoch": 0.29793780304948103,
      "grad_norm": 0.6875,
      "learning_rate": 3.2136465278871324e-05,
      "loss": 2.5144,
      "step": 1530
    },
    {
      "epoch": 0.2981325336397095,
      "grad_norm": 0.69921875,
      "learning_rate": 3.212666099619198e-05,
      "loss": 2.4676,
      "step": 1531
    },
    {
      "epoch": 0.2983272642299379,
      "grad_norm": 0.67578125,
      "learning_rate": 3.211685210296417e-05,
      "loss": 2.5013,
      "step": 1532
    },
    {
      "epoch": 0.2985219948201663,
      "grad_norm": 0.7265625,
      "learning_rate": 3.2107038602917225e-05,
      "loss": 2.4797,
      "step": 1533
    },
    {
      "epoch": 0.29871672541039473,
      "grad_norm": 0.69921875,
      "learning_rate": 3.209722049978223e-05,
      "loss": 2.4905,
      "step": 1534
    },
    {
      "epoch": 0.29891145600062313,
      "grad_norm": 0.625,
      "learning_rate": 3.2087397797292034e-05,
      "loss": 2.4984,
      "step": 1535
    },
    {
      "epoch": 0.29910618659085153,
      "grad_norm": 0.7734375,
      "learning_rate": 3.2077570499181206e-05,
      "loss": 2.5095,
      "step": 1536
    },
    {
      "epoch": 0.29930091718108,
      "grad_norm": 0.7890625,
      "learning_rate": 3.206773860918609e-05,
      "loss": 2.4874,
      "step": 1537
    },
    {
      "epoch": 0.2994956477713084,
      "grad_norm": 0.6171875,
      "learning_rate": 3.2057902131044754e-05,
      "loss": 2.5007,
      "step": 1538
    },
    {
      "epoch": 0.29969037836153684,
      "grad_norm": 0.7734375,
      "learning_rate": 3.2048061068497016e-05,
      "loss": 2.4804,
      "step": 1539
    },
    {
      "epoch": 0.29988510895176523,
      "grad_norm": 0.78125,
      "learning_rate": 3.203821542528446e-05,
      "loss": 2.4805,
      "step": 1540
    },
    {
      "epoch": 0.30007983954199363,
      "grad_norm": 0.6328125,
      "learning_rate": 3.202836520515039e-05,
      "loss": 2.4885,
      "step": 1541
    },
    {
      "epoch": 0.3002745701322221,
      "grad_norm": 0.64453125,
      "learning_rate": 3.201851041183983e-05,
      "loss": 2.5019,
      "step": 1542
    },
    {
      "epoch": 0.3004693007224505,
      "grad_norm": 0.66015625,
      "learning_rate": 3.2008651049099596e-05,
      "loss": 2.5132,
      "step": 1543
    },
    {
      "epoch": 0.3006640313126789,
      "grad_norm": 0.66015625,
      "learning_rate": 3.1998787120678194e-05,
      "loss": 2.5079,
      "step": 1544
    },
    {
      "epoch": 0.30085876190290733,
      "grad_norm": 0.671875,
      "learning_rate": 3.198891863032589e-05,
      "loss": 2.5112,
      "step": 1545
    },
    {
      "epoch": 0.30105349249313573,
      "grad_norm": 0.60546875,
      "learning_rate": 3.197904558179466e-05,
      "loss": 2.4738,
      "step": 1546
    },
    {
      "epoch": 0.3012482230833642,
      "grad_norm": 0.6953125,
      "learning_rate": 3.1969167978838255e-05,
      "loss": 2.4986,
      "step": 1547
    },
    {
      "epoch": 0.3014429536735926,
      "grad_norm": 0.65625,
      "learning_rate": 3.195928582521212e-05,
      "loss": 2.4822,
      "step": 1548
    },
    {
      "epoch": 0.301637684263821,
      "grad_norm": 0.62890625,
      "learning_rate": 3.194939912467345e-05,
      "loss": 2.5068,
      "step": 1549
    },
    {
      "epoch": 0.30183241485404944,
      "grad_norm": 0.7109375,
      "learning_rate": 3.1939507880981154e-05,
      "loss": 2.4818,
      "step": 1550
    },
    {
      "epoch": 0.30202714544427783,
      "grad_norm": 0.7890625,
      "learning_rate": 3.192961209789588e-05,
      "loss": 2.4996,
      "step": 1551
    },
    {
      "epoch": 0.3022218760345063,
      "grad_norm": 0.6953125,
      "learning_rate": 3.191971177918e-05,
      "loss": 2.4843,
      "step": 1552
    },
    {
      "epoch": 0.3024166066247347,
      "grad_norm": 0.69140625,
      "learning_rate": 3.1909806928597606e-05,
      "loss": 2.5024,
      "step": 1553
    },
    {
      "epoch": 0.3026113372149631,
      "grad_norm": 0.69921875,
      "learning_rate": 3.189989754991452e-05,
      "loss": 2.504,
      "step": 1554
    },
    {
      "epoch": 0.30280606780519154,
      "grad_norm": 0.67578125,
      "learning_rate": 3.188998364689827e-05,
      "loss": 2.4912,
      "step": 1555
    },
    {
      "epoch": 0.30300079839541993,
      "grad_norm": 0.69140625,
      "learning_rate": 3.1880065223318127e-05,
      "loss": 2.4878,
      "step": 1556
    },
    {
      "epoch": 0.30319552898564833,
      "grad_norm": 0.71484375,
      "learning_rate": 3.187014228294507e-05,
      "loss": 2.4857,
      "step": 1557
    },
    {
      "epoch": 0.3033902595758768,
      "grad_norm": 0.66015625,
      "learning_rate": 3.186021482955179e-05,
      "loss": 2.5205,
      "step": 1558
    },
    {
      "epoch": 0.3035849901661052,
      "grad_norm": 0.78125,
      "learning_rate": 3.185028286691269e-05,
      "loss": 2.4663,
      "step": 1559
    },
    {
      "epoch": 0.30377972075633364,
      "grad_norm": 0.71484375,
      "learning_rate": 3.18403463988039e-05,
      "loss": 2.4911,
      "step": 1560
    },
    {
      "epoch": 0.30397445134656204,
      "grad_norm": 0.6796875,
      "learning_rate": 3.1830405429003275e-05,
      "loss": 2.5018,
      "step": 1561
    },
    {
      "epoch": 0.30416918193679043,
      "grad_norm": 0.72265625,
      "learning_rate": 3.182045996129034e-05,
      "loss": 2.4719,
      "step": 1562
    },
    {
      "epoch": 0.3043639125270189,
      "grad_norm": 0.64453125,
      "learning_rate": 3.1810509999446376e-05,
      "loss": 2.5045,
      "step": 1563
    },
    {
      "epoch": 0.3045586431172473,
      "grad_norm": 0.6640625,
      "learning_rate": 3.180055554725434e-05,
      "loss": 2.4907,
      "step": 1564
    },
    {
      "epoch": 0.3047533737074757,
      "grad_norm": 0.6328125,
      "learning_rate": 3.179059660849892e-05,
      "loss": 2.491,
      "step": 1565
    },
    {
      "epoch": 0.30494810429770414,
      "grad_norm": 0.64453125,
      "learning_rate": 3.178063318696648e-05,
      "loss": 2.4899,
      "step": 1566
    },
    {
      "epoch": 0.30514283488793253,
      "grad_norm": 0.70703125,
      "learning_rate": 3.177066528644512e-05,
      "loss": 2.4991,
      "step": 1567
    },
    {
      "epoch": 0.305337565478161,
      "grad_norm": 0.7265625,
      "learning_rate": 3.1760692910724614e-05,
      "loss": 2.4874,
      "step": 1568
    },
    {
      "epoch": 0.3055322960683894,
      "grad_norm": 0.6875,
      "learning_rate": 3.175071606359647e-05,
      "loss": 2.4998,
      "step": 1569
    },
    {
      "epoch": 0.3057270266586178,
      "grad_norm": 0.62890625,
      "learning_rate": 3.174073474885387e-05,
      "loss": 2.4891,
      "step": 1570
    },
    {
      "epoch": 0.30592175724884624,
      "grad_norm": 0.73046875,
      "learning_rate": 3.1730748970291714e-05,
      "loss": 2.4866,
      "step": 1571
    },
    {
      "epoch": 0.30611648783907464,
      "grad_norm": 0.69140625,
      "learning_rate": 3.1720758731706574e-05,
      "loss": 2.4887,
      "step": 1572
    },
    {
      "epoch": 0.30631121842930303,
      "grad_norm": 0.65234375,
      "learning_rate": 3.171076403689673e-05,
      "loss": 2.4978,
      "step": 1573
    },
    {
      "epoch": 0.3065059490195315,
      "grad_norm": 0.66796875,
      "learning_rate": 3.170076488966218e-05,
      "loss": 2.511,
      "step": 1574
    },
    {
      "epoch": 0.3067006796097599,
      "grad_norm": 0.62890625,
      "learning_rate": 3.169076129380457e-05,
      "loss": 2.4795,
      "step": 1575
    },
    {
      "epoch": 0.30689541019998834,
      "grad_norm": 0.6875,
      "learning_rate": 3.168075325312727e-05,
      "loss": 2.4913,
      "step": 1576
    },
    {
      "epoch": 0.30709014079021674,
      "grad_norm": 0.734375,
      "learning_rate": 3.1670740771435336e-05,
      "loss": 2.4861,
      "step": 1577
    },
    {
      "epoch": 0.30728487138044513,
      "grad_norm": 0.65234375,
      "learning_rate": 3.166072385253549e-05,
      "loss": 2.4824,
      "step": 1578
    },
    {
      "epoch": 0.3074796019706736,
      "grad_norm": 0.6875,
      "learning_rate": 3.165070250023618e-05,
      "loss": 2.5042,
      "step": 1579
    },
    {
      "epoch": 0.307674332560902,
      "grad_norm": 0.6640625,
      "learning_rate": 3.1640676718347505e-05,
      "loss": 2.4955,
      "step": 1580
    },
    {
      "epoch": 0.3078690631511304,
      "grad_norm": 0.66796875,
      "learning_rate": 3.163064651068126e-05,
      "loss": 2.4949,
      "step": 1581
    },
    {
      "epoch": 0.30806379374135884,
      "grad_norm": 0.64453125,
      "learning_rate": 3.162061188105092e-05,
      "loss": 2.4907,
      "step": 1582
    },
    {
      "epoch": 0.30825852433158724,
      "grad_norm": 0.65625,
      "learning_rate": 3.161057283327166e-05,
      "loss": 2.487,
      "step": 1583
    },
    {
      "epoch": 0.3084532549218157,
      "grad_norm": 0.640625,
      "learning_rate": 3.1600529371160306e-05,
      "loss": 2.4963,
      "step": 1584
    },
    {
      "epoch": 0.3086479855120441,
      "grad_norm": 0.69140625,
      "learning_rate": 3.159048149853537e-05,
      "loss": 2.4919,
      "step": 1585
    },
    {
      "epoch": 0.3088427161022725,
      "grad_norm": 0.6640625,
      "learning_rate": 3.1580429219217066e-05,
      "loss": 2.4813,
      "step": 1586
    },
    {
      "epoch": 0.30903744669250094,
      "grad_norm": 0.61328125,
      "learning_rate": 3.157037253702725e-05,
      "loss": 2.471,
      "step": 1587
    },
    {
      "epoch": 0.30923217728272934,
      "grad_norm": 0.73046875,
      "learning_rate": 3.156031145578947e-05,
      "loss": 2.4664,
      "step": 1588
    },
    {
      "epoch": 0.3094269078729578,
      "grad_norm": 0.6875,
      "learning_rate": 3.1550245979328934e-05,
      "loss": 2.479,
      "step": 1589
    },
    {
      "epoch": 0.3096216384631862,
      "grad_norm": 0.671875,
      "learning_rate": 3.1540176111472546e-05,
      "loss": 2.4904,
      "step": 1590
    },
    {
      "epoch": 0.3098163690534146,
      "grad_norm": 0.71484375,
      "learning_rate": 3.153010185604885e-05,
      "loss": 2.4714,
      "step": 1591
    },
    {
      "epoch": 0.31001109964364304,
      "grad_norm": 0.6171875,
      "learning_rate": 3.152002321688808e-05,
      "loss": 2.4854,
      "step": 1592
    },
    {
      "epoch": 0.31020583023387144,
      "grad_norm": 0.69921875,
      "learning_rate": 3.150994019782212e-05,
      "loss": 2.4907,
      "step": 1593
    },
    {
      "epoch": 0.31040056082409984,
      "grad_norm": 0.62890625,
      "learning_rate": 3.149985280268453e-05,
      "loss": 2.5,
      "step": 1594
    },
    {
      "epoch": 0.3105952914143283,
      "grad_norm": 0.62890625,
      "learning_rate": 3.148976103531053e-05,
      "loss": 2.4979,
      "step": 1595
    },
    {
      "epoch": 0.3107900220045567,
      "grad_norm": 0.6953125,
      "learning_rate": 3.147966489953701e-05,
      "loss": 2.4844,
      "step": 1596
    },
    {
      "epoch": 0.31098475259478514,
      "grad_norm": 0.71875,
      "learning_rate": 3.1469564399202504e-05,
      "loss": 2.486,
      "step": 1597
    },
    {
      "epoch": 0.31117948318501354,
      "grad_norm": 0.69140625,
      "learning_rate": 3.145945953814722e-05,
      "loss": 2.4762,
      "step": 1598
    },
    {
      "epoch": 0.31137421377524194,
      "grad_norm": 0.5859375,
      "learning_rate": 3.144935032021302e-05,
      "loss": 2.4789,
      "step": 1599
    },
    {
      "epoch": 0.3115689443654704,
      "grad_norm": 0.7890625,
      "learning_rate": 3.143923674924343e-05,
      "loss": 2.4745,
      "step": 1600
    },
    {
      "epoch": 0.3117636749556988,
      "grad_norm": 0.8125,
      "learning_rate": 3.1429118829083604e-05,
      "loss": 2.4839,
      "step": 1601
    },
    {
      "epoch": 0.3119584055459272,
      "grad_norm": 0.65625,
      "learning_rate": 3.141899656358038e-05,
      "loss": 2.4847,
      "step": 1602
    },
    {
      "epoch": 0.31215313613615564,
      "grad_norm": 0.67578125,
      "learning_rate": 3.140886995658224e-05,
      "loss": 2.4935,
      "step": 1603
    },
    {
      "epoch": 0.31234786672638404,
      "grad_norm": 0.6796875,
      "learning_rate": 3.139873901193931e-05,
      "loss": 2.4899,
      "step": 1604
    },
    {
      "epoch": 0.3125425973166125,
      "grad_norm": 0.62890625,
      "learning_rate": 3.138860373350337e-05,
      "loss": 2.4868,
      "step": 1605
    },
    {
      "epoch": 0.3127373279068409,
      "grad_norm": 0.61328125,
      "learning_rate": 3.137846412512783e-05,
      "loss": 2.4684,
      "step": 1606
    },
    {
      "epoch": 0.3129320584970693,
      "grad_norm": 0.63671875,
      "learning_rate": 3.1368320190667784e-05,
      "loss": 2.4962,
      "step": 1607
    },
    {
      "epoch": 0.31312678908729774,
      "grad_norm": 0.67578125,
      "learning_rate": 3.1358171933979946e-05,
      "loss": 2.4719,
      "step": 1608
    },
    {
      "epoch": 0.31332151967752614,
      "grad_norm": 0.63671875,
      "learning_rate": 3.134801935892267e-05,
      "loss": 2.4903,
      "step": 1609
    },
    {
      "epoch": 0.31351625026775454,
      "grad_norm": 0.74609375,
      "learning_rate": 3.133786246935596e-05,
      "loss": 2.4912,
      "step": 1610
    },
    {
      "epoch": 0.313710980857983,
      "grad_norm": 0.67578125,
      "learning_rate": 3.132770126914145e-05,
      "loss": 2.484,
      "step": 1611
    },
    {
      "epoch": 0.3139057114482114,
      "grad_norm": 0.65234375,
      "learning_rate": 3.131753576214245e-05,
      "loss": 2.4927,
      "step": 1612
    },
    {
      "epoch": 0.31410044203843984,
      "grad_norm": 0.69140625,
      "learning_rate": 3.1307365952223837e-05,
      "loss": 2.4849,
      "step": 1613
    },
    {
      "epoch": 0.31429517262866824,
      "grad_norm": 0.625,
      "learning_rate": 3.1297191843252197e-05,
      "loss": 2.477,
      "step": 1614
    },
    {
      "epoch": 0.31448990321889664,
      "grad_norm": 0.68359375,
      "learning_rate": 3.128701343909571e-05,
      "loss": 2.4828,
      "step": 1615
    },
    {
      "epoch": 0.3146846338091251,
      "grad_norm": 0.65625,
      "learning_rate": 3.1276830743624204e-05,
      "loss": 2.4821,
      "step": 1616
    },
    {
      "epoch": 0.3148793643993535,
      "grad_norm": 0.703125,
      "learning_rate": 3.1266643760709123e-05,
      "loss": 2.4883,
      "step": 1617
    },
    {
      "epoch": 0.3150740949895819,
      "grad_norm": 0.6875,
      "learning_rate": 3.125645249422356e-05,
      "loss": 2.4834,
      "step": 1618
    },
    {
      "epoch": 0.31526882557981034,
      "grad_norm": 0.6328125,
      "learning_rate": 3.124625694804222e-05,
      "loss": 2.4721,
      "step": 1619
    },
    {
      "epoch": 0.31546355617003874,
      "grad_norm": 0.703125,
      "learning_rate": 3.123605712604145e-05,
      "loss": 2.4981,
      "step": 1620
    },
    {
      "epoch": 0.3156582867602672,
      "grad_norm": 0.62890625,
      "learning_rate": 3.1225853032099215e-05,
      "loss": 2.4897,
      "step": 1621
    },
    {
      "epoch": 0.3158530173504956,
      "grad_norm": 0.73046875,
      "learning_rate": 3.12156446700951e-05,
      "loss": 2.4873,
      "step": 1622
    },
    {
      "epoch": 0.316047747940724,
      "grad_norm": 0.640625,
      "learning_rate": 3.120543204391032e-05,
      "loss": 2.4897,
      "step": 1623
    },
    {
      "epoch": 0.31624247853095244,
      "grad_norm": 0.765625,
      "learning_rate": 3.1195215157427704e-05,
      "loss": 2.5026,
      "step": 1624
    },
    {
      "epoch": 0.31643720912118084,
      "grad_norm": 0.71875,
      "learning_rate": 3.1184994014531715e-05,
      "loss": 2.4743,
      "step": 1625
    },
    {
      "epoch": 0.31663193971140924,
      "grad_norm": 0.69921875,
      "learning_rate": 3.117476861910843e-05,
      "loss": 2.4953,
      "step": 1626
    },
    {
      "epoch": 0.3168266703016377,
      "grad_norm": 0.7265625,
      "learning_rate": 3.116453897504551e-05,
      "loss": 2.4923,
      "step": 1627
    },
    {
      "epoch": 0.3170214008918661,
      "grad_norm": 0.66015625,
      "learning_rate": 3.115430508623228e-05,
      "loss": 2.4642,
      "step": 1628
    },
    {
      "epoch": 0.31721613148209454,
      "grad_norm": 0.75,
      "learning_rate": 3.114406695655966e-05,
      "loss": 2.476,
      "step": 1629
    },
    {
      "epoch": 0.31741086207232294,
      "grad_norm": 0.83984375,
      "learning_rate": 3.1133824589920164e-05,
      "loss": 2.4849,
      "step": 1630
    },
    {
      "epoch": 0.31760559266255134,
      "grad_norm": 0.7109375,
      "learning_rate": 3.1123577990207955e-05,
      "loss": 2.4861,
      "step": 1631
    },
    {
      "epoch": 0.3178003232527798,
      "grad_norm": 0.72265625,
      "learning_rate": 3.1113327161318766e-05,
      "loss": 2.5122,
      "step": 1632
    },
    {
      "epoch": 0.3179950538430082,
      "grad_norm": 0.76171875,
      "learning_rate": 3.110307210714996e-05,
      "loss": 2.4778,
      "step": 1633
    },
    {
      "epoch": 0.31818978443323664,
      "grad_norm": 0.6640625,
      "learning_rate": 3.10928128316005e-05,
      "loss": 2.4849,
      "step": 1634
    },
    {
      "epoch": 0.31838451502346504,
      "grad_norm": 0.86328125,
      "learning_rate": 3.108254933857096e-05,
      "loss": 2.4875,
      "step": 1635
    },
    {
      "epoch": 0.31857924561369344,
      "grad_norm": 0.77734375,
      "learning_rate": 3.107228163196352e-05,
      "loss": 2.4752,
      "step": 1636
    },
    {
      "epoch": 0.3187739762039219,
      "grad_norm": 0.7734375,
      "learning_rate": 3.106200971568194e-05,
      "loss": 2.4756,
      "step": 1637
    },
    {
      "epoch": 0.3189687067941503,
      "grad_norm": 0.76953125,
      "learning_rate": 3.1051733593631606e-05,
      "loss": 2.495,
      "step": 1638
    },
    {
      "epoch": 0.3191634373843787,
      "grad_norm": 0.7890625,
      "learning_rate": 3.104145326971949e-05,
      "loss": 2.4788,
      "step": 1639
    },
    {
      "epoch": 0.31935816797460714,
      "grad_norm": 0.7421875,
      "learning_rate": 3.103116874785416e-05,
      "loss": 2.493,
      "step": 1640
    },
    {
      "epoch": 0.31955289856483554,
      "grad_norm": 0.7890625,
      "learning_rate": 3.1020880031945785e-05,
      "loss": 2.4906,
      "step": 1641
    },
    {
      "epoch": 0.319747629155064,
      "grad_norm": 0.6953125,
      "learning_rate": 3.101058712590614e-05,
      "loss": 2.4966,
      "step": 1642
    },
    {
      "epoch": 0.3199423597452924,
      "grad_norm": 0.76953125,
      "learning_rate": 3.100029003364856e-05,
      "loss": 2.4838,
      "step": 1643
    },
    {
      "epoch": 0.3201370903355208,
      "grad_norm": 0.6953125,
      "learning_rate": 3.098998875908802e-05,
      "loss": 2.483,
      "step": 1644
    },
    {
      "epoch": 0.32033182092574924,
      "grad_norm": 0.67578125,
      "learning_rate": 3.0979683306141035e-05,
      "loss": 2.4734,
      "step": 1645
    },
    {
      "epoch": 0.32052655151597764,
      "grad_norm": 0.734375,
      "learning_rate": 3.0969373678725734e-05,
      "loss": 2.4849,
      "step": 1646
    },
    {
      "epoch": 0.32072128210620604,
      "grad_norm": 0.67578125,
      "learning_rate": 3.095905988076184e-05,
      "loss": 2.4686,
      "step": 1647
    },
    {
      "epoch": 0.3209160126964345,
      "grad_norm": 0.640625,
      "learning_rate": 3.094874191617065e-05,
      "loss": 2.4796,
      "step": 1648
    },
    {
      "epoch": 0.3211107432866629,
      "grad_norm": 0.7109375,
      "learning_rate": 3.093841978887504e-05,
      "loss": 2.4958,
      "step": 1649
    },
    {
      "epoch": 0.32130547387689135,
      "grad_norm": 0.65234375,
      "learning_rate": 3.092809350279949e-05,
      "loss": 2.4789,
      "step": 1650
    },
    {
      "epoch": 0.32150020446711974,
      "grad_norm": 0.65234375,
      "learning_rate": 3.091776306187003e-05,
      "loss": 2.4818,
      "step": 1651
    },
    {
      "epoch": 0.32169493505734814,
      "grad_norm": 0.7265625,
      "learning_rate": 3.09074284700143e-05,
      "loss": 2.4837,
      "step": 1652
    },
    {
      "epoch": 0.3218896656475766,
      "grad_norm": 0.62890625,
      "learning_rate": 3.08970897311615e-05,
      "loss": 2.467,
      "step": 1653
    },
    {
      "epoch": 0.322084396237805,
      "grad_norm": 0.73828125,
      "learning_rate": 3.0886746849242405e-05,
      "loss": 2.4797,
      "step": 1654
    },
    {
      "epoch": 0.3222791268280334,
      "grad_norm": 0.93359375,
      "learning_rate": 3.0876399828189377e-05,
      "loss": 2.4995,
      "step": 1655
    },
    {
      "epoch": 0.32247385741826184,
      "grad_norm": 0.73046875,
      "learning_rate": 3.086604867193636e-05,
      "loss": 2.4876,
      "step": 1656
    },
    {
      "epoch": 0.32266858800849024,
      "grad_norm": 0.70703125,
      "learning_rate": 3.085569338441883e-05,
      "loss": 2.4691,
      "step": 1657
    },
    {
      "epoch": 0.3228633185987187,
      "grad_norm": 0.734375,
      "learning_rate": 3.0845333969573886e-05,
      "loss": 2.4851,
      "step": 1658
    },
    {
      "epoch": 0.3230580491889471,
      "grad_norm": 0.72265625,
      "learning_rate": 3.0834970431340154e-05,
      "loss": 2.4675,
      "step": 1659
    },
    {
      "epoch": 0.3232527797791755,
      "grad_norm": 0.6640625,
      "learning_rate": 3.082460277365785e-05,
      "loss": 2.4886,
      "step": 1660
    },
    {
      "epoch": 0.32344751036940395,
      "grad_norm": 0.73046875,
      "learning_rate": 3.0814231000468753e-05,
      "loss": 2.448,
      "step": 1661
    },
    {
      "epoch": 0.32364224095963234,
      "grad_norm": 0.8359375,
      "learning_rate": 3.0803855115716195e-05,
      "loss": 2.4879,
      "step": 1662
    },
    {
      "epoch": 0.32383697154986074,
      "grad_norm": 0.609375,
      "learning_rate": 3.0793475123345094e-05,
      "loss": 2.4935,
      "step": 1663
    },
    {
      "epoch": 0.3240317021400892,
      "grad_norm": 0.70703125,
      "learning_rate": 3.078309102730191e-05,
      "loss": 2.4868,
      "step": 1664
    },
    {
      "epoch": 0.3242264327303176,
      "grad_norm": 0.71875,
      "learning_rate": 3.077270283153467e-05,
      "loss": 2.5109,
      "step": 1665
    },
    {
      "epoch": 0.32442116332054605,
      "grad_norm": 0.6328125,
      "learning_rate": 3.076231053999296e-05,
      "loss": 2.4769,
      "step": 1666
    },
    {
      "epoch": 0.32461589391077444,
      "grad_norm": 0.64453125,
      "learning_rate": 3.075191415662791e-05,
      "loss": 2.4838,
      "step": 1667
    },
    {
      "epoch": 0.32481062450100284,
      "grad_norm": 0.7578125,
      "learning_rate": 3.074151368539224e-05,
      "loss": 2.4826,
      "step": 1668
    },
    {
      "epoch": 0.3250053550912313,
      "grad_norm": 0.73046875,
      "learning_rate": 3.0731109130240185e-05,
      "loss": 2.4773,
      "step": 1669
    },
    {
      "epoch": 0.3252000856814597,
      "grad_norm": 0.671875,
      "learning_rate": 3.072070049512755e-05,
      "loss": 2.4802,
      "step": 1670
    },
    {
      "epoch": 0.3253948162716881,
      "grad_norm": 0.68359375,
      "learning_rate": 3.0710287784011705e-05,
      "loss": 2.4827,
      "step": 1671
    },
    {
      "epoch": 0.32558954686191655,
      "grad_norm": 0.765625,
      "learning_rate": 3.069987100085155e-05,
      "loss": 2.5038,
      "step": 1672
    },
    {
      "epoch": 0.32578427745214494,
      "grad_norm": 0.62890625,
      "learning_rate": 3.0689450149607535e-05,
      "loss": 2.4626,
      "step": 1673
    },
    {
      "epoch": 0.3259790080423734,
      "grad_norm": 0.76953125,
      "learning_rate": 3.067902523424165e-05,
      "loss": 2.472,
      "step": 1674
    },
    {
      "epoch": 0.3261737386326018,
      "grad_norm": 0.73828125,
      "learning_rate": 3.066859625871747e-05,
      "loss": 2.4741,
      "step": 1675
    },
    {
      "epoch": 0.3263684692228302,
      "grad_norm": 0.7421875,
      "learning_rate": 3.065816322700006e-05,
      "loss": 2.4638,
      "step": 1676
    },
    {
      "epoch": 0.32656319981305865,
      "grad_norm": 0.76953125,
      "learning_rate": 3.0647726143056064e-05,
      "loss": 2.4833,
      "step": 1677
    },
    {
      "epoch": 0.32675793040328704,
      "grad_norm": 0.69140625,
      "learning_rate": 3.063728501085365e-05,
      "loss": 2.4652,
      "step": 1678
    },
    {
      "epoch": 0.3269526609935155,
      "grad_norm": 0.8359375,
      "learning_rate": 3.062683983436252e-05,
      "loss": 2.4649,
      "step": 1679
    },
    {
      "epoch": 0.3271473915837439,
      "grad_norm": 0.73828125,
      "learning_rate": 3.061639061755394e-05,
      "loss": 2.4801,
      "step": 1680
    },
    {
      "epoch": 0.3273421221739723,
      "grad_norm": 0.80859375,
      "learning_rate": 3.06059373644007e-05,
      "loss": 2.4854,
      "step": 1681
    },
    {
      "epoch": 0.32753685276420075,
      "grad_norm": 0.890625,
      "learning_rate": 3.059548007887709e-05,
      "loss": 2.473,
      "step": 1682
    },
    {
      "epoch": 0.32773158335442915,
      "grad_norm": 0.68359375,
      "learning_rate": 3.0585018764958994e-05,
      "loss": 2.4827,
      "step": 1683
    },
    {
      "epoch": 0.32792631394465754,
      "grad_norm": 0.95703125,
      "learning_rate": 3.057455342662378e-05,
      "loss": 2.4865,
      "step": 1684
    },
    {
      "epoch": 0.328121044534886,
      "grad_norm": 0.97265625,
      "learning_rate": 3.0564084067850365e-05,
      "loss": 2.4937,
      "step": 1685
    },
    {
      "epoch": 0.3283157751251144,
      "grad_norm": 0.59765625,
      "learning_rate": 3.055361069261919e-05,
      "loss": 2.4818,
      "step": 1686
    },
    {
      "epoch": 0.32851050571534285,
      "grad_norm": 0.96484375,
      "learning_rate": 3.0543133304912236e-05,
      "loss": 2.4614,
      "step": 1687
    },
    {
      "epoch": 0.32870523630557125,
      "grad_norm": 0.8671875,
      "learning_rate": 3.0532651908712984e-05,
      "loss": 2.4969,
      "step": 1688
    },
    {
      "epoch": 0.32889996689579964,
      "grad_norm": 0.625,
      "learning_rate": 3.052216650800647e-05,
      "loss": 2.4798,
      "step": 1689
    },
    {
      "epoch": 0.3290946974860281,
      "grad_norm": 0.7734375,
      "learning_rate": 3.0511677106779225e-05,
      "loss": 2.4942,
      "step": 1690
    },
    {
      "epoch": 0.3292894280762565,
      "grad_norm": 0.671875,
      "learning_rate": 3.0501183709019312e-05,
      "loss": 2.4708,
      "step": 1691
    },
    {
      "epoch": 0.3294841586664849,
      "grad_norm": 0.703125,
      "learning_rate": 3.0490686318716324e-05,
      "loss": 2.4787,
      "step": 1692
    },
    {
      "epoch": 0.32967888925671335,
      "grad_norm": 0.63671875,
      "learning_rate": 3.0480184939861347e-05,
      "loss": 2.4891,
      "step": 1693
    },
    {
      "epoch": 0.32987361984694175,
      "grad_norm": 0.7109375,
      "learning_rate": 3.0469679576447015e-05,
      "loss": 2.4929,
      "step": 1694
    },
    {
      "epoch": 0.3300683504371702,
      "grad_norm": 0.62109375,
      "learning_rate": 3.045917023246745e-05,
      "loss": 2.4802,
      "step": 1695
    },
    {
      "epoch": 0.3302630810273986,
      "grad_norm": 0.6484375,
      "learning_rate": 3.0448656911918295e-05,
      "loss": 2.4926,
      "step": 1696
    },
    {
      "epoch": 0.330457811617627,
      "grad_norm": 0.625,
      "learning_rate": 3.0438139618796714e-05,
      "loss": 2.4752,
      "step": 1697
    },
    {
      "epoch": 0.33065254220785545,
      "grad_norm": 0.61328125,
      "learning_rate": 3.0427618357101375e-05,
      "loss": 2.4901,
      "step": 1698
    },
    {
      "epoch": 0.33084727279808385,
      "grad_norm": 0.6484375,
      "learning_rate": 3.041709313083245e-05,
      "loss": 2.482,
      "step": 1699
    },
    {
      "epoch": 0.33104200338831224,
      "grad_norm": 0.59765625,
      "learning_rate": 3.0406563943991626e-05,
      "loss": 2.4888,
      "step": 1700
    }
  ],
  "logging_steps": 1,
  "max_steps": 5135,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 50,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 3.5490820797130342e+19,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}