{
  "best_global_step": 945,
  "best_metric": 0.9151147682979645,
  "best_model_checkpoint": "./deberta_multilabel_small\\checkpoint-945",
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 945,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0031746031746031746,
      "grad_norm": 1.707863211631775,
      "learning_rate": 1.9978835978835978e-05,
      "loss": 0.6884,
      "step": 1
    },
    {
      "epoch": 0.006349206349206349,
      "grad_norm": 1.665425181388855,
      "learning_rate": 1.995767195767196e-05,
      "loss": 0.6733,
      "step": 2
    },
    {
      "epoch": 0.009523809523809525,
      "grad_norm": 1.608183741569519,
      "learning_rate": 1.9936507936507938e-05,
      "loss": 0.6686,
      "step": 3
    },
    {
      "epoch": 0.012698412698412698,
      "grad_norm": 1.5483275651931763,
      "learning_rate": 1.9915343915343918e-05,
      "loss": 0.6651,
      "step": 4
    },
    {
      "epoch": 0.015873015873015872,
      "grad_norm": 1.5428708791732788,
      "learning_rate": 1.9894179894179895e-05,
      "loss": 0.6525,
      "step": 5
    },
    {
      "epoch": 0.01904761904761905,
      "grad_norm": 1.5983384847640991,
      "learning_rate": 1.9873015873015875e-05,
      "loss": 0.6515,
      "step": 6
    },
    {
      "epoch": 0.022222222222222223,
      "grad_norm": 1.3994489908218384,
      "learning_rate": 1.9851851851851855e-05,
      "loss": 0.6514,
      "step": 7
    },
    {
      "epoch": 0.025396825396825397,
      "grad_norm": 1.5558671951293945,
      "learning_rate": 1.983068783068783e-05,
      "loss": 0.6413,
      "step": 8
    },
    {
      "epoch": 0.02857142857142857,
      "grad_norm": 1.499327301979065,
      "learning_rate": 1.980952380952381e-05,
      "loss": 0.6362,
      "step": 9
    },
    {
      "epoch": 0.031746031746031744,
      "grad_norm": 1.3703057765960693,
      "learning_rate": 1.978835978835979e-05,
      "loss": 0.634,
      "step": 10
    },
    {
      "epoch": 0.03492063492063492,
      "grad_norm": 1.3919744491577148,
      "learning_rate": 1.9767195767195768e-05,
      "loss": 0.6228,
      "step": 11
    },
    {
      "epoch": 0.0380952380952381,
      "grad_norm": 1.4118250608444214,
      "learning_rate": 1.9746031746031748e-05,
      "loss": 0.6124,
      "step": 12
    },
    {
      "epoch": 0.04126984126984127,
      "grad_norm": 1.4763649702072144,
      "learning_rate": 1.9724867724867728e-05,
      "loss": 0.6078,
      "step": 13
    },
    {
      "epoch": 0.044444444444444446,
      "grad_norm": 1.3768792152404785,
      "learning_rate": 1.9703703703703704e-05,
      "loss": 0.5958,
      "step": 14
    },
    {
      "epoch": 0.047619047619047616,
      "grad_norm": 1.3956913948059082,
      "learning_rate": 1.9682539682539684e-05,
      "loss": 0.59,
      "step": 15
    },
    {
      "epoch": 0.050793650793650794,
      "grad_norm": 1.4166003465652466,
      "learning_rate": 1.9661375661375664e-05,
      "loss": 0.5857,
      "step": 16
    },
    {
      "epoch": 0.05396825396825397,
      "grad_norm": 1.4099714756011963,
      "learning_rate": 1.964021164021164e-05,
      "loss": 0.5631,
      "step": 17
    },
    {
      "epoch": 0.05714285714285714,
      "grad_norm": 1.303537130355835,
      "learning_rate": 1.961904761904762e-05,
      "loss": 0.5693,
      "step": 18
    },
    {
      "epoch": 0.06031746031746032,
      "grad_norm": 1.3324437141418457,
      "learning_rate": 1.95978835978836e-05,
      "loss": 0.5485,
      "step": 19
    },
    {
      "epoch": 0.06349206349206349,
      "grad_norm": 1.2637282609939575,
      "learning_rate": 1.9576719576719577e-05,
      "loss": 0.5416,
      "step": 20
    },
    {
      "epoch": 0.06666666666666667,
      "grad_norm": 1.4056634902954102,
      "learning_rate": 1.9555555555555557e-05,
      "loss": 0.514,
      "step": 21
    },
    {
      "epoch": 0.06984126984126984,
      "grad_norm": 1.3349767923355103,
      "learning_rate": 1.9534391534391537e-05,
      "loss": 0.5071,
      "step": 22
    },
    {
      "epoch": 0.07301587301587302,
      "grad_norm": 1.2290291786193848,
      "learning_rate": 1.9513227513227514e-05,
      "loss": 0.4885,
      "step": 23
    },
    {
      "epoch": 0.0761904761904762,
      "grad_norm": 1.0852007865905762,
      "learning_rate": 1.9492063492063494e-05,
      "loss": 0.4851,
      "step": 24
    },
    {
      "epoch": 0.07936507936507936,
      "grad_norm": 1.142454743385315,
      "learning_rate": 1.947089947089947e-05,
      "loss": 0.4523,
      "step": 25
    },
    {
      "epoch": 0.08253968253968254,
      "grad_norm": 1.1460373401641846,
      "learning_rate": 1.944973544973545e-05,
      "loss": 0.4371,
      "step": 26
    },
    {
      "epoch": 0.08571428571428572,
      "grad_norm": 1.2177430391311646,
      "learning_rate": 1.942857142857143e-05,
      "loss": 0.4482,
      "step": 27
    },
    {
      "epoch": 0.08888888888888889,
      "grad_norm": 1.0766550302505493,
      "learning_rate": 1.9407407407407407e-05,
      "loss": 0.4337,
      "step": 28
    },
    {
      "epoch": 0.09206349206349207,
      "grad_norm": 0.816282331943512,
      "learning_rate": 1.9386243386243387e-05,
      "loss": 0.4553,
      "step": 29
    },
    {
      "epoch": 0.09523809523809523,
      "grad_norm": 1.2499115467071533,
      "learning_rate": 1.9365079365079367e-05,
      "loss": 0.3698,
      "step": 30
    },
    {
      "epoch": 0.09841269841269841,
      "grad_norm": 0.9140530228614807,
      "learning_rate": 1.9343915343915343e-05,
      "loss": 0.4344,
      "step": 31
    },
    {
      "epoch": 0.10158730158730159,
      "grad_norm": 0.9753704071044922,
      "learning_rate": 1.9322751322751327e-05,
      "loss": 0.3812,
      "step": 32
    },
    {
      "epoch": 0.10476190476190476,
      "grad_norm": 0.6696608662605286,
      "learning_rate": 1.9301587301587303e-05,
      "loss": 0.4313,
      "step": 33
    },
    {
      "epoch": 0.10793650793650794,
      "grad_norm": 0.6708428263664246,
      "learning_rate": 1.928042328042328e-05,
      "loss": 0.3997,
      "step": 34
    },
    {
      "epoch": 0.1111111111111111,
      "grad_norm": 0.8106448650360107,
      "learning_rate": 1.925925925925926e-05,
      "loss": 0.3623,
      "step": 35
    },
    {
      "epoch": 0.11428571428571428,
      "grad_norm": 0.8407467603683472,
      "learning_rate": 1.923809523809524e-05,
      "loss": 0.3523,
      "step": 36
    },
    {
      "epoch": 0.11746031746031746,
      "grad_norm": 0.6817750930786133,
      "learning_rate": 1.921693121693122e-05,
      "loss": 0.3944,
      "step": 37
    },
    {
      "epoch": 0.12063492063492064,
      "grad_norm": 0.7455909848213196,
      "learning_rate": 1.9195767195767196e-05,
      "loss": 0.3371,
      "step": 38
    },
    {
      "epoch": 0.12380952380952381,
      "grad_norm": 0.789802074432373,
      "learning_rate": 1.9174603174603176e-05,
      "loss": 0.335,
      "step": 39
    },
    {
      "epoch": 0.12698412698412698,
      "grad_norm": 0.8360641002655029,
      "learning_rate": 1.9153439153439156e-05,
      "loss": 0.3151,
      "step": 40
    },
    {
      "epoch": 0.13015873015873017,
      "grad_norm": 0.7439279556274414,
      "learning_rate": 1.9132275132275133e-05,
      "loss": 0.3414,
      "step": 41
    },
    {
      "epoch": 0.13333333333333333,
      "grad_norm": 0.7024272680282593,
      "learning_rate": 1.9111111111111113e-05,
      "loss": 0.3169,
      "step": 42
    },
    {
      "epoch": 0.1365079365079365,
      "grad_norm": 0.6402228474617004,
      "learning_rate": 1.9089947089947093e-05,
      "loss": 0.3123,
      "step": 43
    },
    {
      "epoch": 0.13968253968253969,
      "grad_norm": 0.5698471665382385,
      "learning_rate": 1.906878306878307e-05,
      "loss": 0.3861,
      "step": 44
    },
    {
      "epoch": 0.14285714285714285,
      "grad_norm": 0.5489242076873779,
      "learning_rate": 1.904761904761905e-05,
      "loss": 0.3165,
      "step": 45
    },
    {
      "epoch": 0.14603174603174604,
      "grad_norm": 0.5373966097831726,
      "learning_rate": 1.902645502645503e-05,
      "loss": 0.3778,
      "step": 46
    },
    {
      "epoch": 0.1492063492063492,
      "grad_norm": 0.4751914441585541,
      "learning_rate": 1.9005291005291006e-05,
      "loss": 0.3835,
      "step": 47
    },
    {
      "epoch": 0.1523809523809524,
      "grad_norm": 0.7408629655838013,
      "learning_rate": 1.8984126984126986e-05,
      "loss": 0.2486,
      "step": 48
    },
    {
      "epoch": 0.15555555555555556,
      "grad_norm": 0.6370362639427185,
      "learning_rate": 1.8962962962962966e-05,
      "loss": 0.4077,
      "step": 49
    },
    {
      "epoch": 0.15873015873015872,
      "grad_norm": 0.7648457288742065,
      "learning_rate": 1.8941798941798943e-05,
      "loss": 0.3274,
      "step": 50
    },
    {
      "epoch": 0.1619047619047619,
      "grad_norm": 0.491093248128891,
      "learning_rate": 1.8920634920634923e-05,
      "loss": 0.3391,
      "step": 51
    },
    {
      "epoch": 0.16507936507936508,
      "grad_norm": 0.3735712468624115,
      "learning_rate": 1.8899470899470903e-05,
      "loss": 0.3465,
      "step": 52
    },
    {
      "epoch": 0.16825396825396827,
      "grad_norm": 0.47635459899902344,
      "learning_rate": 1.887830687830688e-05,
      "loss": 0.3113,
      "step": 53
    },
    {
      "epoch": 0.17142857142857143,
      "grad_norm": 0.49882033467292786,
      "learning_rate": 1.885714285714286e-05,
      "loss": 0.304,
      "step": 54
    },
    {
      "epoch": 0.1746031746031746,
      "grad_norm": 0.3629976212978363,
      "learning_rate": 1.8835978835978836e-05,
      "loss": 0.3925,
      "step": 55
    },
    {
      "epoch": 0.17777777777777778,
      "grad_norm": 0.5941588878631592,
      "learning_rate": 1.8814814814814816e-05,
      "loss": 0.2731,
      "step": 56
    },
    {
      "epoch": 0.18095238095238095,
      "grad_norm": 0.5410904884338379,
      "learning_rate": 1.8793650793650796e-05,
      "loss": 0.3428,
      "step": 57
    },
    {
      "epoch": 0.18412698412698414,
      "grad_norm": 0.36627745628356934,
      "learning_rate": 1.8772486772486772e-05,
      "loss": 0.3423,
      "step": 58
    },
    {
      "epoch": 0.1873015873015873,
      "grad_norm": 0.3826347887516022,
      "learning_rate": 1.8751322751322752e-05,
      "loss": 0.3116,
      "step": 59
    },
    {
      "epoch": 0.19047619047619047,
      "grad_norm": 0.5324172377586365,
      "learning_rate": 1.8730158730158732e-05,
      "loss": 0.338,
      "step": 60
    },
    {
      "epoch": 0.19365079365079366,
      "grad_norm": 0.5456912517547607,
      "learning_rate": 1.870899470899471e-05,
      "loss": 0.3208,
      "step": 61
    },
    {
      "epoch": 0.19682539682539682,
      "grad_norm": 0.43206119537353516,
      "learning_rate": 1.868783068783069e-05,
      "loss": 0.2893,
      "step": 62
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.41829076409339905,
      "learning_rate": 1.866666666666667e-05,
      "loss": 0.2758,
      "step": 63
    },
    {
      "epoch": 0.20317460317460317,
      "grad_norm": 0.4344908595085144,
      "learning_rate": 1.8645502645502645e-05,
      "loss": 0.3275,
      "step": 64
    },
    {
      "epoch": 0.20634920634920634,
      "grad_norm": 0.6365790963172913,
      "learning_rate": 1.8624338624338625e-05,
      "loss": 0.2502,
      "step": 65
    },
    {
      "epoch": 0.20952380952380953,
      "grad_norm": 0.5701781511306763,
      "learning_rate": 1.8603174603174605e-05,
      "loss": 0.3441,
      "step": 66
    },
    {
      "epoch": 0.2126984126984127,
      "grad_norm": 0.5187179446220398,
      "learning_rate": 1.8582010582010582e-05,
      "loss": 0.3109,
      "step": 67
    },
    {
      "epoch": 0.21587301587301588,
      "grad_norm": 0.5761861801147461,
      "learning_rate": 1.8560846560846562e-05,
      "loss": 0.2585,
      "step": 68
    },
    {
      "epoch": 0.21904761904761905,
      "grad_norm": 0.6790960431098938,
      "learning_rate": 1.853968253968254e-05,
      "loss": 0.207,
      "step": 69
    },
    {
      "epoch": 0.2222222222222222,
      "grad_norm": 0.5540696978569031,
      "learning_rate": 1.851851851851852e-05,
      "loss": 0.2745,
      "step": 70
    },
    {
      "epoch": 0.2253968253968254,
      "grad_norm": 0.4661020338535309,
      "learning_rate": 1.8497354497354498e-05,
      "loss": 0.3119,
      "step": 71
    },
    {
      "epoch": 0.22857142857142856,
      "grad_norm": 0.6018322110176086,
      "learning_rate": 1.8476190476190478e-05,
      "loss": 0.2835,
      "step": 72
    },
    {
      "epoch": 0.23174603174603176,
      "grad_norm": 0.5966388583183289,
      "learning_rate": 1.8455026455026458e-05,
      "loss": 0.2873,
      "step": 73
    },
    {
      "epoch": 0.23492063492063492,
      "grad_norm": 0.7091224193572998,
      "learning_rate": 1.8433862433862435e-05,
      "loss": 0.2938,
      "step": 74
    },
    {
      "epoch": 0.23809523809523808,
      "grad_norm": 0.4926207363605499,
      "learning_rate": 1.8412698412698415e-05,
      "loss": 0.2958,
      "step": 75
    },
    {
      "epoch": 0.24126984126984127,
      "grad_norm": 0.6199020147323608,
      "learning_rate": 1.8391534391534395e-05,
      "loss": 0.26,
      "step": 76
    },
    {
      "epoch": 0.24444444444444444,
      "grad_norm": 0.7603418827056885,
      "learning_rate": 1.837037037037037e-05,
      "loss": 0.2605,
      "step": 77
    },
    {
      "epoch": 0.24761904761904763,
      "grad_norm": 0.6452163457870483,
      "learning_rate": 1.834920634920635e-05,
      "loss": 0.3695,
      "step": 78
    },
    {
      "epoch": 0.2507936507936508,
      "grad_norm": 0.676623523235321,
      "learning_rate": 1.832804232804233e-05,
      "loss": 0.2754,
      "step": 79
    },
    {
      "epoch": 0.25396825396825395,
      "grad_norm": 0.8160414099693298,
      "learning_rate": 1.8306878306878308e-05,
      "loss": 0.2952,
      "step": 80
    },
    {
      "epoch": 0.2571428571428571,
      "grad_norm": 0.6845346093177795,
      "learning_rate": 1.8285714285714288e-05,
      "loss": 0.2527,
      "step": 81
    },
    {
      "epoch": 0.26031746031746034,
      "grad_norm": 0.7733173966407776,
      "learning_rate": 1.8264550264550268e-05,
      "loss": 0.3084,
      "step": 82
    },
    {
      "epoch": 0.2634920634920635,
      "grad_norm": 0.7036414742469788,
      "learning_rate": 1.8243386243386244e-05,
      "loss": 0.2358,
      "step": 83
    },
    {
      "epoch": 0.26666666666666666,
      "grad_norm": 0.6889075636863708,
      "learning_rate": 1.8222222222222224e-05,
      "loss": 0.2223,
      "step": 84
    },
    {
      "epoch": 0.2698412698412698,
      "grad_norm": 0.8003445267677307,
      "learning_rate": 1.82010582010582e-05,
      "loss": 0.2291,
      "step": 85
    },
    {
      "epoch": 0.273015873015873,
      "grad_norm": 0.9161721467971802,
      "learning_rate": 1.817989417989418e-05,
      "loss": 0.2632,
      "step": 86
    },
    {
      "epoch": 0.2761904761904762,
      "grad_norm": 0.5593780875205994,
      "learning_rate": 1.815873015873016e-05,
      "loss": 0.217,
      "step": 87
    },
    {
      "epoch": 0.27936507936507937,
      "grad_norm": 0.5470061302185059,
      "learning_rate": 1.8137566137566137e-05,
      "loss": 0.2378,
      "step": 88
    },
    {
      "epoch": 0.28253968253968254,
      "grad_norm": 0.6262850761413574,
      "learning_rate": 1.8116402116402117e-05,
      "loss": 0.187,
      "step": 89
    },
    {
      "epoch": 0.2857142857142857,
      "grad_norm": 0.6546247005462646,
      "learning_rate": 1.8095238095238097e-05,
      "loss": 0.1688,
      "step": 90
    },
    {
      "epoch": 0.28888888888888886,
      "grad_norm": 0.5408976674079895,
      "learning_rate": 1.8074074074074074e-05,
      "loss": 0.2043,
      "step": 91
    },
    {
      "epoch": 0.2920634920634921,
      "grad_norm": 0.8207212686538696,
      "learning_rate": 1.8052910052910054e-05,
      "loss": 0.2302,
      "step": 92
    },
    {
      "epoch": 0.29523809523809524,
      "grad_norm": 1.2199501991271973,
      "learning_rate": 1.8031746031746034e-05,
      "loss": 0.2449,
      "step": 93
    },
    {
      "epoch": 0.2984126984126984,
      "grad_norm": 0.7664761543273926,
      "learning_rate": 1.801058201058201e-05,
      "loss": 0.2934,
      "step": 94
    },
    {
      "epoch": 0.30158730158730157,
      "grad_norm": 0.6092225313186646,
      "learning_rate": 1.798941798941799e-05,
      "loss": 0.1762,
      "step": 95
    },
    {
      "epoch": 0.3047619047619048,
      "grad_norm": 1.1074532270431519,
      "learning_rate": 1.796825396825397e-05,
      "loss": 0.2298,
      "step": 96
    },
    {
      "epoch": 0.30793650793650795,
      "grad_norm": 1.01616370677948,
      "learning_rate": 1.7947089947089947e-05,
      "loss": 0.2145,
      "step": 97
    },
    {
      "epoch": 0.3111111111111111,
      "grad_norm": 1.5490742921829224,
      "learning_rate": 1.7925925925925927e-05,
      "loss": 0.2588,
      "step": 98
    },
    {
      "epoch": 0.3142857142857143,
      "grad_norm": 0.9028508067131042,
      "learning_rate": 1.7904761904761907e-05,
      "loss": 0.1811,
      "step": 99
    },
    {
      "epoch": 0.31746031746031744,
      "grad_norm": 0.7514470219612122,
      "learning_rate": 1.7883597883597884e-05,
      "loss": 0.2491,
      "step": 100
    },
    {
      "epoch": 0.32063492063492066,
      "grad_norm": 0.8829514384269714,
      "learning_rate": 1.7862433862433863e-05,
      "loss": 0.1813,
      "step": 101
    },
    {
      "epoch": 0.3238095238095238,
      "grad_norm": 1.1838492155075073,
      "learning_rate": 1.7841269841269843e-05,
      "loss": 0.2201,
      "step": 102
    },
    {
      "epoch": 0.326984126984127,
      "grad_norm": 1.3106063604354858,
      "learning_rate": 1.7820105820105823e-05,
      "loss": 0.2627,
      "step": 103
    },
    {
      "epoch": 0.33015873015873015,
      "grad_norm": 1.0679289102554321,
      "learning_rate": 1.77989417989418e-05,
      "loss": 0.2771,
      "step": 104
    },
    {
      "epoch": 0.3333333333333333,
      "grad_norm": 0.6278952360153198,
      "learning_rate": 1.7777777777777777e-05,
      "loss": 0.2189,
      "step": 105
    },
    {
      "epoch": 0.33650793650793653,
      "grad_norm": 0.7160807251930237,
      "learning_rate": 1.775661375661376e-05,
      "loss": 0.2276,
      "step": 106
    },
    {
      "epoch": 0.3396825396825397,
      "grad_norm": 0.7008633017539978,
      "learning_rate": 1.7735449735449737e-05,
      "loss": 0.2189,
      "step": 107
    },
    {
      "epoch": 0.34285714285714286,
      "grad_norm": 0.7340114712715149,
      "learning_rate": 1.7714285714285717e-05,
      "loss": 0.2723,
      "step": 108
    },
    {
      "epoch": 0.346031746031746,
      "grad_norm": 1.446215033531189,
      "learning_rate": 1.7693121693121696e-05,
      "loss": 0.2161,
      "step": 109
    },
    {
      "epoch": 0.3492063492063492,
      "grad_norm": 0.9089925289154053,
      "learning_rate": 1.7671957671957673e-05,
      "loss": 0.2249,
      "step": 110
    },
    {
      "epoch": 0.3523809523809524,
      "grad_norm": 0.9569251537322998,
      "learning_rate": 1.7650793650793653e-05,
      "loss": 0.1869,
      "step": 111
    },
    {
      "epoch": 0.35555555555555557,
      "grad_norm": 0.7527865171432495,
      "learning_rate": 1.7629629629629633e-05,
      "loss": 0.1953,
      "step": 112
    },
    {
      "epoch": 0.35873015873015873,
      "grad_norm": 0.9081427454948425,
      "learning_rate": 1.760846560846561e-05,
      "loss": 0.2284,
      "step": 113
    },
    {
      "epoch": 0.3619047619047619,
      "grad_norm": 0.6875194311141968,
      "learning_rate": 1.758730158730159e-05,
      "loss": 0.2107,
      "step": 114
    },
    {
      "epoch": 0.36507936507936506,
      "grad_norm": 1.0420364141464233,
      "learning_rate": 1.7566137566137566e-05,
      "loss": 0.1911,
      "step": 115
    },
    {
      "epoch": 0.3682539682539683,
      "grad_norm": 0.5223270654678345,
      "learning_rate": 1.7544973544973546e-05,
      "loss": 0.2173,
      "step": 116
    },
    {
      "epoch": 0.37142857142857144,
      "grad_norm": 1.264967441558838,
      "learning_rate": 1.7523809523809526e-05,
      "loss": 0.2817,
      "step": 117
    },
    {
      "epoch": 0.3746031746031746,
      "grad_norm": 1.184549331665039,
      "learning_rate": 1.7502645502645503e-05,
      "loss": 0.2155,
      "step": 118
    },
    {
      "epoch": 0.37777777777777777,
      "grad_norm": 0.6578064560890198,
      "learning_rate": 1.7481481481481483e-05,
      "loss": 0.2081,
      "step": 119
    },
    {
      "epoch": 0.38095238095238093,
      "grad_norm": 0.5839540362358093,
      "learning_rate": 1.7460317460317463e-05,
      "loss": 0.1765,
      "step": 120
    },
    {
      "epoch": 0.38412698412698415,
      "grad_norm": 1.2461695671081543,
      "learning_rate": 1.743915343915344e-05,
      "loss": 0.2419,
      "step": 121
    },
    {
      "epoch": 0.3873015873015873,
      "grad_norm": 0.6531723141670227,
      "learning_rate": 1.741798941798942e-05,
      "loss": 0.2196,
      "step": 122
    },
    {
      "epoch": 0.3904761904761905,
      "grad_norm": 0.9668866991996765,
      "learning_rate": 1.73968253968254e-05,
      "loss": 0.202,
      "step": 123
    },
    {
      "epoch": 0.39365079365079364,
      "grad_norm": 1.4050712585449219,
      "learning_rate": 1.7375661375661376e-05,
      "loss": 0.1671,
      "step": 124
    },
    {
      "epoch": 0.3968253968253968,
      "grad_norm": 1.0694444179534912,
      "learning_rate": 1.7354497354497356e-05,
      "loss": 0.206,
      "step": 125
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.283082127571106,
      "learning_rate": 1.7333333333333336e-05,
      "loss": 0.159,
      "step": 126
    },
    {
      "epoch": 0.4031746031746032,
      "grad_norm": 0.9682832360267639,
      "learning_rate": 1.7312169312169312e-05,
      "loss": 0.2354,
      "step": 127
    },
    {
      "epoch": 0.40634920634920635,
      "grad_norm": 0.9270000457763672,
      "learning_rate": 1.7291005291005292e-05,
      "loss": 0.2486,
      "step": 128
    },
    {
      "epoch": 0.4095238095238095,
      "grad_norm": 0.6798192262649536,
      "learning_rate": 1.7269841269841272e-05,
      "loss": 0.1792,
      "step": 129
    },
    {
      "epoch": 0.4126984126984127,
      "grad_norm": 0.9379510283470154,
      "learning_rate": 1.724867724867725e-05,
      "loss": 0.2353,
      "step": 130
    },
    {
      "epoch": 0.4158730158730159,
      "grad_norm": 0.6720793843269348,
      "learning_rate": 1.722751322751323e-05,
      "loss": 0.2095,
      "step": 131
    },
    {
      "epoch": 0.41904761904761906,
      "grad_norm": 0.553537905216217,
      "learning_rate": 1.720634920634921e-05,
      "loss": 0.2164,
      "step": 132
    },
    {
      "epoch": 0.4222222222222222,
      "grad_norm": 0.7756928205490112,
      "learning_rate": 1.7185185185185185e-05,
      "loss": 0.1583,
      "step": 133
    },
    {
      "epoch": 0.4253968253968254,
      "grad_norm": 0.5201077461242676,
      "learning_rate": 1.7164021164021165e-05,
      "loss": 0.1415,
      "step": 134
    },
    {
      "epoch": 0.42857142857142855,
      "grad_norm": 0.7277474999427795,
      "learning_rate": 1.7142857142857142e-05,
      "loss": 0.1154,
      "step": 135
    },
    {
      "epoch": 0.43174603174603177,
      "grad_norm": 0.6321858167648315,
      "learning_rate": 1.7121693121693125e-05,
      "loss": 0.149,
      "step": 136
    },
    {
      "epoch": 0.43492063492063493,
      "grad_norm": 0.8540474772453308,
      "learning_rate": 1.7100529100529102e-05,
      "loss": 0.2297,
      "step": 137
    },
    {
      "epoch": 0.4380952380952381,
      "grad_norm": 0.8365871906280518,
      "learning_rate": 1.707936507936508e-05,
      "loss": 0.1952,
      "step": 138
    },
    {
      "epoch": 0.44126984126984126,
      "grad_norm": 0.465338796377182,
      "learning_rate": 1.7058201058201062e-05,
      "loss": 0.1941,
      "step": 139
    },
    {
      "epoch": 0.4444444444444444,
      "grad_norm": 0.538543164730072,
      "learning_rate": 1.7037037037037038e-05,
      "loss": 0.1236,
      "step": 140
    },
    {
      "epoch": 0.44761904761904764,
      "grad_norm": 1.3244738578796387,
      "learning_rate": 1.7015873015873018e-05,
      "loss": 0.204,
      "step": 141
    },
    {
      "epoch": 0.4507936507936508,
      "grad_norm": 0.7624632716178894,
      "learning_rate": 1.6994708994708998e-05,
      "loss": 0.1989,
      "step": 142
    },
    {
      "epoch": 0.45396825396825397,
      "grad_norm": 0.7170157432556152,
      "learning_rate": 1.6973544973544975e-05,
      "loss": 0.1486,
      "step": 143
    },
    {
      "epoch": 0.45714285714285713,
      "grad_norm": 0.6547853350639343,
      "learning_rate": 1.6952380952380955e-05,
      "loss": 0.1412,
      "step": 144
    },
    {
      "epoch": 0.4603174603174603,
      "grad_norm": 1.1090824604034424,
      "learning_rate": 1.693121693121693e-05,
      "loss": 0.1471,
      "step": 145
    },
    {
      "epoch": 0.4634920634920635,
      "grad_norm": 0.5632514357566833,
      "learning_rate": 1.691005291005291e-05,
      "loss": 0.1809,
      "step": 146
    },
    {
      "epoch": 0.4666666666666667,
      "grad_norm": 1.2325406074523926,
      "learning_rate": 1.688888888888889e-05,
      "loss": 0.1445,
      "step": 147
    },
    {
      "epoch": 0.46984126984126984,
      "grad_norm": 1.233332633972168,
      "learning_rate": 1.6867724867724868e-05,
      "loss": 0.1592,
      "step": 148
    },
    {
      "epoch": 0.473015873015873,
      "grad_norm": 0.8157974481582642,
      "learning_rate": 1.6846560846560848e-05,
      "loss": 0.14,
      "step": 149
    },
    {
      "epoch": 0.47619047619047616,
      "grad_norm": 1.1565017700195312,
      "learning_rate": 1.6825396825396828e-05,
      "loss": 0.2513,
      "step": 150
    },
    {
      "epoch": 0.4793650793650794,
      "grad_norm": 1.364281415939331,
      "learning_rate": 1.6804232804232804e-05,
      "loss": 0.1664,
      "step": 151
    },
    {
      "epoch": 0.48253968253968255,
      "grad_norm": 0.6618668437004089,
      "learning_rate": 1.6783068783068784e-05,
      "loss": 0.2194,
      "step": 152
    },
    {
      "epoch": 0.4857142857142857,
      "grad_norm": 0.45250260829925537,
      "learning_rate": 1.6761904761904764e-05,
      "loss": 0.1681,
      "step": 153
    },
    {
      "epoch": 0.4888888888888889,
      "grad_norm": 1.1851006746292114,
      "learning_rate": 1.674074074074074e-05,
      "loss": 0.1895,
      "step": 154
    },
    {
      "epoch": 0.49206349206349204,
      "grad_norm": 0.46531859040260315,
      "learning_rate": 1.671957671957672e-05,
      "loss": 0.1427,
      "step": 155
    },
    {
      "epoch": 0.49523809523809526,
      "grad_norm": 0.7176091074943542,
      "learning_rate": 1.66984126984127e-05,
      "loss": 0.1459,
      "step": 156
    },
    {
      "epoch": 0.4984126984126984,
      "grad_norm": 1.1034573316574097,
      "learning_rate": 1.6677248677248677e-05,
      "loss": 0.2319,
      "step": 157
    },
    {
      "epoch": 0.5015873015873016,
      "grad_norm": 0.9433768391609192,
      "learning_rate": 1.6656084656084657e-05,
      "loss": 0.1828,
      "step": 158
    },
    {
      "epoch": 0.5047619047619047,
      "grad_norm": 0.8536099195480347,
      "learning_rate": 1.6634920634920637e-05,
      "loss": 0.1028,
      "step": 159
    },
    {
      "epoch": 0.5079365079365079,
      "grad_norm": 1.0035890340805054,
      "learning_rate": 1.6613756613756614e-05,
      "loss": 0.2181,
      "step": 160
    },
    {
      "epoch": 0.5111111111111111,
      "grad_norm": 1.4087523221969604,
      "learning_rate": 1.6592592592592594e-05,
      "loss": 0.1921,
      "step": 161
    },
    {
      "epoch": 0.5142857142857142,
      "grad_norm": 0.6502156257629395,
      "learning_rate": 1.6571428571428574e-05,
      "loss": 0.1491,
      "step": 162
    },
    {
      "epoch": 0.5174603174603175,
      "grad_norm": 1.1924118995666504,
      "learning_rate": 1.655026455026455e-05,
      "loss": 0.2415,
      "step": 163
    },
    {
      "epoch": 0.5206349206349207,
      "grad_norm": 1.7833726406097412,
      "learning_rate": 1.652910052910053e-05,
      "loss": 0.22,
      "step": 164
    },
    {
      "epoch": 0.5238095238095238,
      "grad_norm": 1.983986735343933,
      "learning_rate": 1.6507936507936507e-05,
      "loss": 0.2131,
      "step": 165
    },
    {
      "epoch": 0.526984126984127,
      "grad_norm": 0.8444494605064392,
      "learning_rate": 1.648677248677249e-05,
      "loss": 0.1036,
      "step": 166
    },
    {
      "epoch": 0.5301587301587302,
      "grad_norm": 1.197389006614685,
      "learning_rate": 1.6465608465608467e-05,
      "loss": 0.2068,
      "step": 167
    },
    {
      "epoch": 0.5333333333333333,
      "grad_norm": 0.575753390789032,
      "learning_rate": 1.6444444444444444e-05,
      "loss": 0.1807,
      "step": 168
    },
    {
      "epoch": 0.5365079365079365,
      "grad_norm": 1.3008365631103516,
      "learning_rate": 1.6423280423280427e-05,
      "loss": 0.2061,
      "step": 169
    },
    {
      "epoch": 0.5396825396825397,
      "grad_norm": 0.8197752833366394,
      "learning_rate": 1.6402116402116404e-05,
      "loss": 0.1564,
      "step": 170
    },
    {
      "epoch": 0.5428571428571428,
      "grad_norm": 0.9654321670532227,
      "learning_rate": 1.6380952380952384e-05,
      "loss": 0.1142,
      "step": 171
    },
    {
      "epoch": 0.546031746031746,
      "grad_norm": 1.2690932750701904,
      "learning_rate": 1.6359788359788363e-05,
      "loss": 0.1573,
      "step": 172
    },
    {
      "epoch": 0.5492063492063493,
      "grad_norm": 0.5396488308906555,
      "learning_rate": 1.633862433862434e-05,
      "loss": 0.1383,
      "step": 173
    },
    {
      "epoch": 0.5523809523809524,
      "grad_norm": 0.7310698628425598,
      "learning_rate": 1.631746031746032e-05,
      "loss": 0.1367,
      "step": 174
    },
    {
      "epoch": 0.5555555555555556,
      "grad_norm": 0.8129367828369141,
      "learning_rate": 1.6296296296296297e-05,
      "loss": 0.2145,
      "step": 175
    },
    {
      "epoch": 0.5587301587301587,
      "grad_norm": 0.8666160106658936,
      "learning_rate": 1.6275132275132277e-05,
      "loss": 0.1734,
      "step": 176
    },
    {
      "epoch": 0.5619047619047619,
      "grad_norm": 0.8468620777130127,
      "learning_rate": 1.6253968253968257e-05,
      "loss": 0.1788,
      "step": 177
    },
    {
      "epoch": 0.5650793650793651,
      "grad_norm": 0.7338481545448303,
      "learning_rate": 1.6232804232804233e-05,
      "loss": 0.1803,
      "step": 178
    },
    {
      "epoch": 0.5682539682539682,
      "grad_norm": 0.9112880825996399,
      "learning_rate": 1.6211640211640213e-05,
      "loss": 0.2009,
      "step": 179
    },
    {
      "epoch": 0.5714285714285714,
      "grad_norm": 1.0125837326049805,
      "learning_rate": 1.6190476190476193e-05,
      "loss": 0.1426,
      "step": 180
    },
    {
      "epoch": 0.5746031746031746,
      "grad_norm": 0.9464571475982666,
      "learning_rate": 1.616931216931217e-05,
      "loss": 0.1551,
      "step": 181
    },
    {
      "epoch": 0.5777777777777777,
      "grad_norm": 0.5625197291374207,
      "learning_rate": 1.614814814814815e-05,
      "loss": 0.1481,
      "step": 182
    },
    {
      "epoch": 0.580952380952381,
      "grad_norm": 0.7718688249588013,
      "learning_rate": 1.612698412698413e-05,
      "loss": 0.1608,
      "step": 183
    },
    {
      "epoch": 0.5841269841269842,
      "grad_norm": 1.238206386566162,
      "learning_rate": 1.6105820105820106e-05,
      "loss": 0.1397,
      "step": 184
    },
    {
      "epoch": 0.5873015873015873,
      "grad_norm": 1.1525496244430542,
      "learning_rate": 1.6084656084656086e-05,
      "loss": 0.1527,
      "step": 185
    },
    {
      "epoch": 0.5904761904761905,
      "grad_norm": 0.752475917339325,
      "learning_rate": 1.6063492063492066e-05,
      "loss": 0.1644,
      "step": 186
    },
    {
      "epoch": 0.5936507936507937,
      "grad_norm": 0.4704437851905823,
      "learning_rate": 1.6042328042328043e-05,
      "loss": 0.1578,
      "step": 187
    },
    {
      "epoch": 0.5968253968253968,
      "grad_norm": 1.0864458084106445,
      "learning_rate": 1.6021164021164023e-05,
      "loss": 0.1494,
      "step": 188
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7002252340316772,
      "learning_rate": 1.6000000000000003e-05,
      "loss": 0.1503,
      "step": 189
    },
    {
      "epoch": 0.6031746031746031,
      "grad_norm": 0.7413320541381836,
      "learning_rate": 1.597883597883598e-05,
      "loss": 0.1386,
      "step": 190
    },
    {
      "epoch": 0.6063492063492063,
      "grad_norm": 0.5265030860900879,
      "learning_rate": 1.595767195767196e-05,
      "loss": 0.164,
      "step": 191
    },
    {
      "epoch": 0.6095238095238096,
      "grad_norm": 0.6609658002853394,
      "learning_rate": 1.5936507936507936e-05,
      "loss": 0.1272,
      "step": 192
    },
    {
      "epoch": 0.6126984126984127,
      "grad_norm": 0.87735915184021,
      "learning_rate": 1.5915343915343916e-05,
      "loss": 0.2081,
      "step": 193
    },
    {
      "epoch": 0.6158730158730159,
      "grad_norm": 0.6933319568634033,
      "learning_rate": 1.5894179894179896e-05,
      "loss": 0.2144,
      "step": 194
    },
    {
      "epoch": 0.6190476190476191,
      "grad_norm": 0.8761839270591736,
      "learning_rate": 1.5873015873015872e-05,
      "loss": 0.1235,
      "step": 195
    },
    {
      "epoch": 0.6222222222222222,
      "grad_norm": 0.6609085202217102,
      "learning_rate": 1.5851851851851852e-05,
      "loss": 0.1895,
      "step": 196
    },
    {
      "epoch": 0.6253968253968254,
      "grad_norm": 0.6925118565559387,
      "learning_rate": 1.5830687830687832e-05,
      "loss": 0.1519,
      "step": 197
    },
    {
      "epoch": 0.6285714285714286,
      "grad_norm": 1.0692561864852905,
      "learning_rate": 1.580952380952381e-05,
      "loss": 0.1486,
      "step": 198
    },
    {
      "epoch": 0.6317460317460317,
      "grad_norm": 1.0505307912826538,
      "learning_rate": 1.5788359788359792e-05,
      "loss": 0.1478,
      "step": 199
    },
    {
      "epoch": 0.6349206349206349,
      "grad_norm": 0.9300786852836609,
      "learning_rate": 1.576719576719577e-05,
      "loss": 0.1357,
      "step": 200
    },
    {
      "epoch": 0.638095238095238,
      "grad_norm": 1.086264967918396,
      "learning_rate": 1.5746031746031745e-05,
      "loss": 0.1744,
      "step": 201
    },
    {
      "epoch": 0.6412698412698413,
      "grad_norm": 1.6160521507263184,
      "learning_rate": 1.5724867724867725e-05,
      "loss": 0.2014,
      "step": 202
    },
    {
      "epoch": 0.6444444444444445,
      "grad_norm": 0.5533894896507263,
      "learning_rate": 1.5703703703703705e-05,
      "loss": 0.1457,
      "step": 203
    },
    {
      "epoch": 0.6476190476190476,
      "grad_norm": 0.45417118072509766,
      "learning_rate": 1.5682539682539685e-05,
      "loss": 0.1059,
      "step": 204
    },
    {
      "epoch": 0.6507936507936508,
      "grad_norm": 0.6732428073883057,
      "learning_rate": 1.5661375661375662e-05,
      "loss": 0.1807,
      "step": 205
    },
    {
      "epoch": 0.653968253968254,
      "grad_norm": 1.3881568908691406,
      "learning_rate": 1.5640211640211642e-05,
      "loss": 0.14,
      "step": 206
    },
    {
      "epoch": 0.6571428571428571,
      "grad_norm": 1.7982734441757202,
      "learning_rate": 1.5619047619047622e-05,
      "loss": 0.203,
      "step": 207
    },
    {
      "epoch": 0.6603174603174603,
      "grad_norm": 0.9197050333023071,
      "learning_rate": 1.55978835978836e-05,
      "loss": 0.1466,
      "step": 208
    },
    {
      "epoch": 0.6634920634920635,
      "grad_norm": 0.40359485149383545,
      "learning_rate": 1.557671957671958e-05,
      "loss": 0.0849,
      "step": 209
    },
    {
      "epoch": 0.6666666666666666,
      "grad_norm": 0.5555000901222229,
      "learning_rate": 1.555555555555556e-05,
      "loss": 0.1751,
      "step": 210
    },
    {
      "epoch": 0.6698412698412698,
      "grad_norm": 0.9116174578666687,
      "learning_rate": 1.5534391534391535e-05,
      "loss": 0.212,
      "step": 211
    },
    {
      "epoch": 0.6730158730158731,
      "grad_norm": 1.1145563125610352,
      "learning_rate": 1.5513227513227515e-05,
      "loss": 0.161,
      "step": 212
    },
    {
      "epoch": 0.6761904761904762,
      "grad_norm": 0.6321357488632202,
      "learning_rate": 1.5492063492063495e-05,
      "loss": 0.1598,
      "step": 213
    },
    {
      "epoch": 0.6793650793650794,
      "grad_norm": 0.6547977328300476,
      "learning_rate": 1.547089947089947e-05,
      "loss": 0.1672,
      "step": 214
    },
    {
      "epoch": 0.6825396825396826,
      "grad_norm": 0.83708256483078,
      "learning_rate": 1.544973544973545e-05,
      "loss": 0.1774,
      "step": 215
    },
    {
      "epoch": 0.6857142857142857,
      "grad_norm": 1.1288692951202393,
      "learning_rate": 1.542857142857143e-05,
      "loss": 0.1316,
      "step": 216
    },
    {
      "epoch": 0.6888888888888889,
      "grad_norm": 1.0221855640411377,
      "learning_rate": 1.5407407407407408e-05,
      "loss": 0.2006,
      "step": 217
    },
    {
      "epoch": 0.692063492063492,
      "grad_norm": 0.7739736437797546,
      "learning_rate": 1.5386243386243388e-05,
      "loss": 0.1217,
      "step": 218
    },
    {
      "epoch": 0.6952380952380952,
      "grad_norm": 0.6374626755714417,
      "learning_rate": 1.5365079365079368e-05,
      "loss": 0.1408,
      "step": 219
    },
    {
      "epoch": 0.6984126984126984,
      "grad_norm": 1.109535574913025,
      "learning_rate": 1.5343915343915344e-05,
      "loss": 0.1638,
      "step": 220
    },
    {
      "epoch": 0.7015873015873015,
      "grad_norm": 1.166182518005371,
      "learning_rate": 1.5322751322751324e-05,
      "loss": 0.1348,
      "step": 221
    },
    {
      "epoch": 0.7047619047619048,
      "grad_norm": 0.8725109100341797,
      "learning_rate": 1.53015873015873e-05,
      "loss": 0.1731,
      "step": 222
    },
    {
      "epoch": 0.707936507936508,
      "grad_norm": 1.2718509435653687,
      "learning_rate": 1.528042328042328e-05,
      "loss": 0.1517,
      "step": 223
    },
    {
      "epoch": 0.7111111111111111,
      "grad_norm": 0.5993607640266418,
      "learning_rate": 1.525925925925926e-05,
      "loss": 0.1826,
      "step": 224
    },
    {
      "epoch": 0.7142857142857143,
      "grad_norm": 1.0330419540405273,
      "learning_rate": 1.523809523809524e-05,
      "loss": 0.1152,
      "step": 225
    },
    {
      "epoch": 0.7174603174603175,
      "grad_norm": 0.7561971545219421,
      "learning_rate": 1.521693121693122e-05,
      "loss": 0.1498,
      "step": 226
    },
    {
      "epoch": 0.7206349206349206,
      "grad_norm": 1.4977335929870605,
      "learning_rate": 1.5195767195767198e-05,
      "loss": 0.1501,
      "step": 227
    },
    {
      "epoch": 0.7238095238095238,
      "grad_norm": 0.9320580363273621,
      "learning_rate": 1.5174603174603176e-05,
      "loss": 0.0936,
      "step": 228
    },
    {
      "epoch": 0.726984126984127,
      "grad_norm": 1.293817162513733,
      "learning_rate": 1.5153439153439156e-05,
      "loss": 0.1541,
      "step": 229
    },
    {
      "epoch": 0.7301587301587301,
      "grad_norm": 1.3387163877487183,
      "learning_rate": 1.5132275132275134e-05,
      "loss": 0.1543,
      "step": 230
    },
    {
      "epoch": 0.7333333333333333,
      "grad_norm": 1.3477325439453125,
      "learning_rate": 1.5111111111111112e-05,
      "loss": 0.1922,
      "step": 231
    },
    {
      "epoch": 0.7365079365079366,
      "grad_norm": 1.4394477605819702,
      "learning_rate": 1.508994708994709e-05,
      "loss": 0.152,
      "step": 232
    },
    {
      "epoch": 0.7396825396825397,
      "grad_norm": 0.7048148512840271,
      "learning_rate": 1.506878306878307e-05,
      "loss": 0.1343,
      "step": 233
    },
    {
      "epoch": 0.7428571428571429,
      "grad_norm": 0.810907244682312,
      "learning_rate": 1.5047619047619049e-05,
      "loss": 0.1382,
      "step": 234
    },
    {
      "epoch": 0.746031746031746,
      "grad_norm": 0.9021286964416504,
      "learning_rate": 1.5026455026455027e-05,
      "loss": 0.2357,
      "step": 235
    },
    {
      "epoch": 0.7492063492063492,
      "grad_norm": 1.0414401292800903,
      "learning_rate": 1.5005291005291007e-05,
      "loss": 0.1848,
      "step": 236
    },
    {
      "epoch": 0.7523809523809524,
      "grad_norm": 0.7497075796127319,
      "learning_rate": 1.4984126984126985e-05,
      "loss": 0.1037,
      "step": 237
    },
    {
      "epoch": 0.7555555555555555,
      "grad_norm": 1.3421279191970825,
      "learning_rate": 1.4962962962962964e-05,
      "loss": 0.2382,
      "step": 238
    },
    {
      "epoch": 0.7587301587301587,
      "grad_norm": 0.9856558442115784,
      "learning_rate": 1.4941798941798944e-05,
      "loss": 0.1687,
      "step": 239
    },
    {
      "epoch": 0.7619047619047619,
      "grad_norm": 0.5265440940856934,
      "learning_rate": 1.4920634920634922e-05,
      "loss": 0.1232,
      "step": 240
    },
    {
      "epoch": 0.765079365079365,
      "grad_norm": 1.1311218738555908,
      "learning_rate": 1.48994708994709e-05,
      "loss": 0.2079,
      "step": 241
    },
    {
      "epoch": 0.7682539682539683,
      "grad_norm": 0.5580438375473022,
      "learning_rate": 1.4878306878306878e-05,
      "loss": 0.1154,
      "step": 242
    },
    {
      "epoch": 0.7714285714285715,
      "grad_norm": 0.5071591138839722,
      "learning_rate": 1.4857142857142858e-05,
      "loss": 0.0639,
      "step": 243
    },
    {
      "epoch": 0.7746031746031746,
      "grad_norm": 1.2320207357406616,
      "learning_rate": 1.4835978835978837e-05,
      "loss": 0.1997,
      "step": 244
    },
    {
      "epoch": 0.7777777777777778,
      "grad_norm": 0.8571259379386902,
      "learning_rate": 1.4814814814814815e-05,
      "loss": 0.1807,
      "step": 245
    },
    {
      "epoch": 0.780952380952381,
      "grad_norm": 0.7246306538581848,
      "learning_rate": 1.4793650793650795e-05,
      "loss": 0.2177,
      "step": 246
    },
    {
      "epoch": 0.7841269841269841,
      "grad_norm": 0.417015939950943,
      "learning_rate": 1.4772486772486773e-05,
      "loss": 0.1443,
      "step": 247
    },
    {
      "epoch": 0.7873015873015873,
      "grad_norm": 1.1055175065994263,
      "learning_rate": 1.4751322751322751e-05,
      "loss": 0.177,
      "step": 248
    },
    {
      "epoch": 0.7904761904761904,
      "grad_norm": 0.6800684928894043,
      "learning_rate": 1.4730158730158733e-05,
      "loss": 0.1207,
      "step": 249
    },
    {
      "epoch": 0.7936507936507936,
      "grad_norm": 0.862838089466095,
      "learning_rate": 1.470899470899471e-05,
      "loss": 0.0951,
      "step": 250
    },
    {
      "epoch": 0.7968253968253968,
      "grad_norm": 0.608191728591919,
      "learning_rate": 1.4687830687830688e-05,
      "loss": 0.1463,
      "step": 251
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7509791851043701,
      "learning_rate": 1.4666666666666666e-05,
      "loss": 0.1085,
      "step": 252
    },
    {
      "epoch": 0.8031746031746032,
      "grad_norm": 0.6635032296180725,
      "learning_rate": 1.4645502645502648e-05,
      "loss": 0.1681,
      "step": 253
    },
    {
      "epoch": 0.8063492063492064,
      "grad_norm": 0.882626473903656,
      "learning_rate": 1.4624338624338626e-05,
      "loss": 0.1442,
      "step": 254
    },
    {
      "epoch": 0.8095238095238095,
      "grad_norm": 1.0488476753234863,
      "learning_rate": 1.4603174603174603e-05,
      "loss": 0.2235,
      "step": 255
    },
    {
      "epoch": 0.8126984126984127,
      "grad_norm": 0.700793445110321,
      "learning_rate": 1.4582010582010584e-05,
      "loss": 0.1424,
      "step": 256
    },
    {
      "epoch": 0.8158730158730159,
      "grad_norm": 0.5714010000228882,
      "learning_rate": 1.4560846560846563e-05,
      "loss": 0.1589,
      "step": 257
    },
    {
      "epoch": 0.819047619047619,
      "grad_norm": 1.558276653289795,
      "learning_rate": 1.4539682539682541e-05,
      "loss": 0.1576,
      "step": 258
    },
    {
      "epoch": 0.8222222222222222,
      "grad_norm": 0.8912820816040039,
      "learning_rate": 1.4518518518518521e-05,
      "loss": 0.1878,
      "step": 259
    },
    {
      "epoch": 0.8253968253968254,
      "grad_norm": 1.2163937091827393,
      "learning_rate": 1.44973544973545e-05,
      "loss": 0.143,
      "step": 260
    },
    {
      "epoch": 0.8285714285714286,
      "grad_norm": 0.5732966065406799,
      "learning_rate": 1.4476190476190478e-05,
      "loss": 0.1343,
      "step": 261
    },
    {
      "epoch": 0.8317460317460318,
      "grad_norm": 0.8898940682411194,
      "learning_rate": 1.4455026455026456e-05,
      "loss": 0.1164,
      "step": 262
    },
    {
      "epoch": 0.834920634920635,
      "grad_norm": 0.7961615324020386,
      "learning_rate": 1.4433862433862436e-05,
      "loss": 0.1117,
      "step": 263
    },
    {
      "epoch": 0.8380952380952381,
      "grad_norm": 1.3982429504394531,
      "learning_rate": 1.4412698412698414e-05,
      "loss": 0.1047,
      "step": 264
    },
    {
      "epoch": 0.8412698412698413,
      "grad_norm": 1.5879848003387451,
      "learning_rate": 1.4391534391534392e-05,
      "loss": 0.1484,
      "step": 265
    },
    {
      "epoch": 0.8444444444444444,
      "grad_norm": 0.8436241745948792,
      "learning_rate": 1.4370370370370372e-05,
      "loss": 0.1359,
      "step": 266
    },
    {
      "epoch": 0.8476190476190476,
      "grad_norm": 0.5601047873497009,
      "learning_rate": 1.434920634920635e-05,
      "loss": 0.097,
      "step": 267
    },
    {
      "epoch": 0.8507936507936508,
      "grad_norm": 0.9039320945739746,
      "learning_rate": 1.4328042328042329e-05,
      "loss": 0.1719,
      "step": 268
    },
    {
      "epoch": 0.8539682539682539,
      "grad_norm": 0.8469903469085693,
      "learning_rate": 1.4306878306878309e-05,
      "loss": 0.173,
      "step": 269
    },
    {
      "epoch": 0.8571428571428571,
      "grad_norm": 0.6634661555290222,
      "learning_rate": 1.4285714285714287e-05,
      "loss": 0.1379,
      "step": 270
    },
    {
      "epoch": 0.8603174603174604,
      "grad_norm": 0.9670260548591614,
      "learning_rate": 1.4264550264550265e-05,
      "loss": 0.1777,
      "step": 271
    },
    {
      "epoch": 0.8634920634920635,
      "grad_norm": 0.7979556918144226,
      "learning_rate": 1.4243386243386244e-05,
      "loss": 0.1345,
      "step": 272
    },
    {
      "epoch": 0.8666666666666667,
      "grad_norm": 1.0763016939163208,
      "learning_rate": 1.4222222222222224e-05,
      "loss": 0.1413,
      "step": 273
    },
    {
      "epoch": 0.8698412698412699,
      "grad_norm": 2.051616668701172,
      "learning_rate": 1.4201058201058202e-05,
      "loss": 0.14,
      "step": 274
    },
    {
      "epoch": 0.873015873015873,
      "grad_norm": 2.323857545852661,
      "learning_rate": 1.417989417989418e-05,
      "loss": 0.2062,
      "step": 275
    },
    {
      "epoch": 0.8761904761904762,
      "grad_norm": 0.8823941349983215,
      "learning_rate": 1.415873015873016e-05,
      "loss": 0.101,
      "step": 276
    },
    {
      "epoch": 0.8793650793650793,
      "grad_norm": 1.145979642868042,
      "learning_rate": 1.4137566137566138e-05,
      "loss": 0.1229,
      "step": 277
    },
    {
      "epoch": 0.8825396825396825,
      "grad_norm": 1.6212157011032104,
      "learning_rate": 1.4116402116402117e-05,
      "loss": 0.1559,
      "step": 278
    },
    {
      "epoch": 0.8857142857142857,
      "grad_norm": 0.7980738282203674,
      "learning_rate": 1.4095238095238097e-05,
      "loss": 0.1897,
      "step": 279
    },
    {
      "epoch": 0.8888888888888888,
      "grad_norm": 0.4689996838569641,
      "learning_rate": 1.4074074074074075e-05,
      "loss": 0.1162,
      "step": 280
    },
    {
      "epoch": 0.8920634920634921,
      "grad_norm": 0.8861905336380005,
      "learning_rate": 1.4052910052910053e-05,
      "loss": 0.1599,
      "step": 281
    },
    {
      "epoch": 0.8952380952380953,
      "grad_norm": 0.9779180288314819,
      "learning_rate": 1.4031746031746032e-05,
      "loss": 0.0794,
      "step": 282
    },
    {
      "epoch": 0.8984126984126984,
      "grad_norm": 1.2002872228622437,
      "learning_rate": 1.4010582010582011e-05,
      "loss": 0.1379,
      "step": 283
    },
    {
      "epoch": 0.9015873015873016,
      "grad_norm": 1.263126254081726,
      "learning_rate": 1.398941798941799e-05,
      "loss": 0.1528,
      "step": 284
    },
    {
      "epoch": 0.9047619047619048,
      "grad_norm": 1.6818698644638062,
      "learning_rate": 1.3968253968253968e-05,
      "loss": 0.1855,
      "step": 285
    },
    {
      "epoch": 0.9079365079365079,
      "grad_norm": 0.8238258361816406,
      "learning_rate": 1.394708994708995e-05,
      "loss": 0.1301,
      "step": 286
    },
    {
      "epoch": 0.9111111111111111,
      "grad_norm": 0.9428375959396362,
      "learning_rate": 1.3925925925925928e-05,
      "loss": 0.1158,
      "step": 287
    },
    {
      "epoch": 0.9142857142857143,
      "grad_norm": 0.4689411520957947,
      "learning_rate": 1.3904761904761905e-05,
      "loss": 0.0725,
      "step": 288
    },
    {
      "epoch": 0.9174603174603174,
      "grad_norm": 1.0714961290359497,
      "learning_rate": 1.3883597883597886e-05,
      "loss": 0.1101,
      "step": 289
    },
    {
      "epoch": 0.9206349206349206,
      "grad_norm": 0.8055447340011597,
      "learning_rate": 1.3862433862433865e-05,
      "loss": 0.1315,
      "step": 290
    },
    {
      "epoch": 0.9238095238095239,
      "grad_norm": 1.4988516569137573,
      "learning_rate": 1.3841269841269843e-05,
      "loss": 0.1396,
      "step": 291
    },
    {
      "epoch": 0.926984126984127,
      "grad_norm": 1.7930117845535278,
      "learning_rate": 1.3820105820105821e-05,
      "loss": 0.1854,
      "step": 292
    },
    {
      "epoch": 0.9301587301587302,
      "grad_norm": 1.1544146537780762,
      "learning_rate": 1.3798941798941801e-05,
      "loss": 0.1343,
      "step": 293
    },
    {
      "epoch": 0.9333333333333333,
      "grad_norm": 0.8714034557342529,
      "learning_rate": 1.377777777777778e-05,
      "loss": 0.1306,
      "step": 294
    },
    {
      "epoch": 0.9365079365079365,
      "grad_norm": 0.860656201839447,
      "learning_rate": 1.3756613756613758e-05,
      "loss": 0.1702,
      "step": 295
    },
    {
      "epoch": 0.9396825396825397,
      "grad_norm": 0.7115416526794434,
      "learning_rate": 1.3735449735449738e-05,
      "loss": 0.1636,
      "step": 296
    },
    {
      "epoch": 0.9428571428571428,
      "grad_norm": 0.7452229857444763,
      "learning_rate": 1.3714285714285716e-05,
      "loss": 0.102,
      "step": 297
    },
    {
      "epoch": 0.946031746031746,
      "grad_norm": 0.6584372520446777,
      "learning_rate": 1.3693121693121694e-05,
      "loss": 0.0981,
      "step": 298
    },
    {
      "epoch": 0.9492063492063492,
      "grad_norm": 0.6651356816291809,
      "learning_rate": 1.3671957671957674e-05,
      "loss": 0.1248,
      "step": 299
    },
    {
      "epoch": 0.9523809523809523,
      "grad_norm": 0.72747403383255,
      "learning_rate": 1.3650793650793652e-05,
      "loss": 0.138,
      "step": 300
    },
    {
      "epoch": 0.9555555555555556,
      "grad_norm": 0.8110100030899048,
      "learning_rate": 1.362962962962963e-05,
      "loss": 0.0988,
      "step": 301
    },
    {
      "epoch": 0.9587301587301588,
      "grad_norm": 0.7712112069129944,
      "learning_rate": 1.3608465608465609e-05,
      "loss": 0.0913,
      "step": 302
    },
    {
      "epoch": 0.9619047619047619,
      "grad_norm": 1.1628859043121338,
      "learning_rate": 1.3587301587301589e-05,
      "loss": 0.1735,
      "step": 303
    },
    {
      "epoch": 0.9650793650793651,
      "grad_norm": 0.8472080230712891,
      "learning_rate": 1.3566137566137567e-05,
      "loss": 0.1679,
      "step": 304
    },
    {
      "epoch": 0.9682539682539683,
      "grad_norm": 0.7450751662254333,
      "learning_rate": 1.3544973544973545e-05,
      "loss": 0.1657,
      "step": 305
    },
    {
      "epoch": 0.9714285714285714,
      "grad_norm": 1.093201756477356,
      "learning_rate": 1.3523809523809525e-05,
      "loss": 0.2043,
      "step": 306
    },
    {
      "epoch": 0.9746031746031746,
      "grad_norm": 0.7437242269515991,
      "learning_rate": 1.3502645502645504e-05,
      "loss": 0.1547,
      "step": 307
    },
    {
      "epoch": 0.9777777777777777,
      "grad_norm": 1.1371376514434814,
      "learning_rate": 1.3481481481481482e-05,
      "loss": 0.1746,
      "step": 308
    },
    {
      "epoch": 0.9809523809523809,
      "grad_norm": 1.2805427312850952,
      "learning_rate": 1.3460317460317462e-05,
      "loss": 0.1757,
      "step": 309
    },
    {
      "epoch": 0.9841269841269841,
      "grad_norm": 1.074521541595459,
      "learning_rate": 1.343915343915344e-05,
      "loss": 0.1396,
      "step": 310
    },
    {
      "epoch": 0.9873015873015873,
      "grad_norm": 1.0821629762649536,
      "learning_rate": 1.3417989417989418e-05,
      "loss": 0.1213,
      "step": 311
    },
    {
      "epoch": 0.9904761904761905,
      "grad_norm": 1.1837350130081177,
      "learning_rate": 1.3396825396825397e-05,
      "loss": 0.1047,
      "step": 312
    },
    {
      "epoch": 0.9936507936507937,
      "grad_norm": 0.8410152792930603,
      "learning_rate": 1.3375661375661377e-05,
      "loss": 0.1178,
      "step": 313
    },
    {
      "epoch": 0.9968253968253968,
      "grad_norm": 0.8698068857192993,
      "learning_rate": 1.3354497354497355e-05,
      "loss": 0.1613,
      "step": 314
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9205747842788696,
      "learning_rate": 1.3333333333333333e-05,
      "loss": 0.0891,
      "step": 315
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.7526785714285714,
      "eval_f1": 0.8946796959826276,
      "eval_loss": 0.13787341117858887,
      "eval_runtime": 1.7538,
      "eval_samples_per_second": 638.61,
      "eval_steps_per_second": 19.957,
      "step": 315
    },
    {
      "epoch": 1.0031746031746032,
      "grad_norm": 1.1484298706054688,
      "learning_rate": 1.3312169312169313e-05,
      "loss": 0.1792,
      "step": 316
    },
    {
      "epoch": 1.0063492063492063,
      "grad_norm": 0.7768347263336182,
      "learning_rate": 1.3291005291005292e-05,
      "loss": 0.1215,
      "step": 317
    },
    {
      "epoch": 1.0095238095238095,
      "grad_norm": 0.6903197169303894,
      "learning_rate": 1.326984126984127e-05,
      "loss": 0.0928,
      "step": 318
    },
    {
      "epoch": 1.0126984126984127,
      "grad_norm": 0.7531824707984924,
      "learning_rate": 1.3248677248677251e-05,
      "loss": 0.1429,
      "step": 319
    },
    {
      "epoch": 1.0158730158730158,
      "grad_norm": 0.8707253336906433,
      "learning_rate": 1.322751322751323e-05,
      "loss": 0.1154,
      "step": 320
    },
    {
      "epoch": 1.019047619047619,
      "grad_norm": 0.8431862592697144,
      "learning_rate": 1.3206349206349206e-05,
      "loss": 0.1524,
      "step": 321
    },
    {
      "epoch": 1.0222222222222221,
      "grad_norm": 0.6184837222099304,
      "learning_rate": 1.3185185185185185e-05,
      "loss": 0.1387,
      "step": 322
    },
    {
      "epoch": 1.0253968253968253,
      "grad_norm": 0.6718246340751648,
      "learning_rate": 1.3164021164021166e-05,
      "loss": 0.1116,
      "step": 323
    },
    {
      "epoch": 1.0285714285714285,
      "grad_norm": 1.084627628326416,
      "learning_rate": 1.3142857142857145e-05,
      "loss": 0.1022,
      "step": 324
    },
    {
      "epoch": 1.0317460317460316,
      "grad_norm": 0.9810977578163147,
      "learning_rate": 1.3121693121693123e-05,
      "loss": 0.0961,
      "step": 325
    },
    {
      "epoch": 1.034920634920635,
      "grad_norm": 1.8372925519943237,
      "learning_rate": 1.3100529100529103e-05,
      "loss": 0.155,
      "step": 326
    },
    {
      "epoch": 1.0380952380952382,
      "grad_norm": 0.7125292420387268,
      "learning_rate": 1.3079365079365081e-05,
      "loss": 0.1295,
      "step": 327
    },
    {
      "epoch": 1.0412698412698413,
      "grad_norm": 0.7733258605003357,
      "learning_rate": 1.305820105820106e-05,
      "loss": 0.1504,
      "step": 328
    },
    {
      "epoch": 1.0444444444444445,
      "grad_norm": 0.7995859384536743,
      "learning_rate": 1.303703703703704e-05,
      "loss": 0.1945,
      "step": 329
    },
    {
      "epoch": 1.0476190476190477,
      "grad_norm": 0.6571292877197266,
      "learning_rate": 1.3015873015873018e-05,
      "loss": 0.0837,
      "step": 330
    },
    {
      "epoch": 1.0507936507936508,
      "grad_norm": 0.8413328528404236,
      "learning_rate": 1.2994708994708996e-05,
      "loss": 0.1184,
      "step": 331
    },
    {
      "epoch": 1.053968253968254,
      "grad_norm": 0.7271382808685303,
      "learning_rate": 1.2973544973544974e-05,
      "loss": 0.1363,
      "step": 332
    },
    {
      "epoch": 1.0571428571428572,
      "grad_norm": 1.2771875858306885,
      "learning_rate": 1.2952380952380954e-05,
      "loss": 0.1716,
      "step": 333
    },
    {
      "epoch": 1.0603174603174603,
      "grad_norm": 1.4482518434524536,
      "learning_rate": 1.2931216931216932e-05,
      "loss": 0.1438,
      "step": 334
    },
    {
      "epoch": 1.0634920634920635,
      "grad_norm": 0.4712583124637604,
      "learning_rate": 1.291005291005291e-05,
      "loss": 0.117,
      "step": 335
    },
    {
      "epoch": 1.0666666666666667,
      "grad_norm": 0.38839513063430786,
      "learning_rate": 1.288888888888889e-05,
      "loss": 0.0544,
      "step": 336
    },
    {
      "epoch": 1.0698412698412698,
      "grad_norm": 0.8053770065307617,
      "learning_rate": 1.2867724867724869e-05,
      "loss": 0.096,
      "step": 337
    },
    {
      "epoch": 1.073015873015873,
      "grad_norm": 0.79161137342453,
      "learning_rate": 1.2846560846560847e-05,
      "loss": 0.1406,
      "step": 338
    },
    {
      "epoch": 1.0761904761904761,
      "grad_norm": 0.9772685170173645,
      "learning_rate": 1.2825396825396827e-05,
      "loss": 0.2538,
      "step": 339
    },
    {
      "epoch": 1.0793650793650793,
      "grad_norm": 0.8453724980354309,
      "learning_rate": 1.2804232804232805e-05,
      "loss": 0.1497,
      "step": 340
    },
    {
      "epoch": 1.0825396825396825,
      "grad_norm": 0.8952115178108215,
      "learning_rate": 1.2783068783068784e-05,
      "loss": 0.1116,
      "step": 341
    },
    {
      "epoch": 1.0857142857142856,
      "grad_norm": 0.630199670791626,
      "learning_rate": 1.2761904761904762e-05,
      "loss": 0.1158,
      "step": 342
    },
    {
      "epoch": 1.0888888888888888,
      "grad_norm": 0.7082343101501465,
      "learning_rate": 1.2740740740740742e-05,
      "loss": 0.0821,
      "step": 343
    },
    {
      "epoch": 1.0920634920634922,
      "grad_norm": 0.8591349124908447,
      "learning_rate": 1.271957671957672e-05,
      "loss": 0.1518,
      "step": 344
    },
    {
      "epoch": 1.0952380952380953,
      "grad_norm": 0.5619752407073975,
      "learning_rate": 1.2698412698412699e-05,
      "loss": 0.0767,
      "step": 345
    },
    {
      "epoch": 1.0984126984126985,
      "grad_norm": 0.6850156188011169,
      "learning_rate": 1.2677248677248679e-05,
      "loss": 0.1137,
      "step": 346
    },
    {
      "epoch": 1.1015873015873017,
      "grad_norm": 1.0435181856155396,
      "learning_rate": 1.2656084656084657e-05,
      "loss": 0.0977,
      "step": 347
    },
    {
      "epoch": 1.1047619047619048,
      "grad_norm": 0.5554229617118835,
      "learning_rate": 1.2634920634920635e-05,
      "loss": 0.0637,
      "step": 348
    },
    {
      "epoch": 1.107936507936508,
      "grad_norm": 0.6769399046897888,
      "learning_rate": 1.2613756613756615e-05,
      "loss": 0.1269,
      "step": 349
    },
    {
      "epoch": 1.1111111111111112,
      "grad_norm": 0.32357361912727356,
      "learning_rate": 1.2592592592592593e-05,
      "loss": 0.0831,
      "step": 350
    },
    {
      "epoch": 1.1142857142857143,
      "grad_norm": 0.8532891273498535,
      "learning_rate": 1.2571428571428572e-05,
      "loss": 0.1913,
      "step": 351
    },
    {
      "epoch": 1.1174603174603175,
      "grad_norm": 0.7476623058319092,
      "learning_rate": 1.255026455026455e-05,
      "loss": 0.1531,
      "step": 352
    },
    {
      "epoch": 1.1206349206349207,
      "grad_norm": 0.45784443616867065,
      "learning_rate": 1.2529100529100532e-05,
      "loss": 0.0888,
      "step": 353
    },
    {
      "epoch": 1.1238095238095238,
      "grad_norm": 1.1472291946411133,
      "learning_rate": 1.2507936507936508e-05,
      "loss": 0.1199,
      "step": 354
    },
    {
      "epoch": 1.126984126984127,
      "grad_norm": 0.31491222977638245,
      "learning_rate": 1.2486772486772486e-05,
      "loss": 0.0942,
      "step": 355
    },
    {
      "epoch": 1.1301587301587301,
      "grad_norm": 0.7360396385192871,
      "learning_rate": 1.2465608465608468e-05,
      "loss": 0.1013,
      "step": 356
    },
    {
      "epoch": 1.1333333333333333,
      "grad_norm": 0.5503537058830261,
      "learning_rate": 1.2444444444444446e-05,
      "loss": 0.1601,
      "step": 357
    },
    {
      "epoch": 1.1365079365079365,
      "grad_norm": 1.0685306787490845,
      "learning_rate": 1.2423280423280425e-05,
      "loss": 0.1059,
      "step": 358
    },
    {
      "epoch": 1.1396825396825396,
      "grad_norm": 0.6969165802001953,
      "learning_rate": 1.2402116402116405e-05,
      "loss": 0.0942,
      "step": 359
    },
    {
      "epoch": 1.1428571428571428,
      "grad_norm": 0.9813324809074402,
      "learning_rate": 1.2380952380952383e-05,
      "loss": 0.2214,
      "step": 360
    },
    {
      "epoch": 1.146031746031746,
      "grad_norm": 1.0207619667053223,
      "learning_rate": 1.2359788359788361e-05,
      "loss": 0.1116,
      "step": 361
    },
    {
      "epoch": 1.1492063492063491,
      "grad_norm": 1.174749732017517,
      "learning_rate": 1.233862433862434e-05,
      "loss": 0.119,
      "step": 362
    },
    {
      "epoch": 1.1523809523809523,
      "grad_norm": 0.813128650188446,
      "learning_rate": 1.231746031746032e-05,
      "loss": 0.1814,
      "step": 363
    },
    {
      "epoch": 1.1555555555555554,
      "grad_norm": 0.7182943224906921,
      "learning_rate": 1.2296296296296298e-05,
      "loss": 0.0938,
      "step": 364
    },
    {
      "epoch": 1.1587301587301586,
      "grad_norm": 0.5025944709777832,
      "learning_rate": 1.2275132275132276e-05,
      "loss": 0.1199,
      "step": 365
    },
    {
      "epoch": 1.161904761904762,
      "grad_norm": 1.526428461074829,
      "learning_rate": 1.2253968253968256e-05,
      "loss": 0.1669,
      "step": 366
    },
    {
      "epoch": 1.1650793650793652,
      "grad_norm": 1.493025302886963,
      "learning_rate": 1.2232804232804234e-05,
      "loss": 0.1519,
      "step": 367
    },
    {
      "epoch": 1.1682539682539683,
      "grad_norm": 1.0965336561203003,
      "learning_rate": 1.2211640211640212e-05,
      "loss": 0.1781,
      "step": 368
    },
    {
      "epoch": 1.1714285714285715,
      "grad_norm": 0.6048644185066223,
      "learning_rate": 1.2190476190476192e-05,
      "loss": 0.1451,
      "step": 369
    },
    {
      "epoch": 1.1746031746031746,
      "grad_norm": 0.9804893136024475,
      "learning_rate": 1.216931216931217e-05,
      "loss": 0.1313,
      "step": 370
    },
    {
      "epoch": 1.1777777777777778,
      "grad_norm": 1.2321230173110962,
      "learning_rate": 1.2148148148148149e-05,
      "loss": 0.1108,
      "step": 371
    },
    {
      "epoch": 1.180952380952381,
      "grad_norm": 0.6345534324645996,
      "learning_rate": 1.2126984126984127e-05,
      "loss": 0.1168,
      "step": 372
    },
    {
      "epoch": 1.1841269841269841,
      "grad_norm": 0.8624942302703857,
      "learning_rate": 1.2105820105820107e-05,
      "loss": 0.134,
      "step": 373
    },
    {
      "epoch": 1.1873015873015873,
      "grad_norm": 0.5653309226036072,
      "learning_rate": 1.2084656084656086e-05,
      "loss": 0.1298,
      "step": 374
    },
    {
      "epoch": 1.1904761904761905,
      "grad_norm": 0.561185896396637,
      "learning_rate": 1.2063492063492064e-05,
      "loss": 0.1382,
      "step": 375
    },
    {
      "epoch": 1.1936507936507936,
      "grad_norm": 0.6253190040588379,
      "learning_rate": 1.2042328042328044e-05,
      "loss": 0.1322,
      "step": 376
    },
    {
      "epoch": 1.1968253968253968,
      "grad_norm": 0.7884150743484497,
      "learning_rate": 1.2021164021164022e-05,
      "loss": 0.1457,
      "step": 377
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9688888192176819,
      "learning_rate": 1.2e-05,
      "loss": 0.1056,
      "step": 378
    },
    {
      "epoch": 1.2031746031746031,
      "grad_norm": 0.5485055446624756,
      "learning_rate": 1.1978835978835979e-05,
      "loss": 0.1313,
      "step": 379
    },
    {
      "epoch": 1.2063492063492063,
      "grad_norm": 0.8263285756111145,
      "learning_rate": 1.1957671957671959e-05,
      "loss": 0.1287,
      "step": 380
    },
    {
      "epoch": 1.2095238095238094,
      "grad_norm": 0.6682865619659424,
      "learning_rate": 1.1936507936507937e-05,
      "loss": 0.1126,
      "step": 381
    },
    {
      "epoch": 1.2126984126984126,
      "grad_norm": 0.7351716160774231,
      "learning_rate": 1.1915343915343915e-05,
      "loss": 0.1312,
      "step": 382
    },
    {
      "epoch": 1.215873015873016,
      "grad_norm": 0.7624906897544861,
      "learning_rate": 1.1894179894179895e-05,
      "loss": 0.1099,
      "step": 383
    },
    {
      "epoch": 1.2190476190476192,
      "grad_norm": 1.174400806427002,
      "learning_rate": 1.1873015873015873e-05,
      "loss": 0.1696,
      "step": 384
    },
    {
      "epoch": 1.2222222222222223,
      "grad_norm": 0.9188776016235352,
      "learning_rate": 1.1851851851851852e-05,
      "loss": 0.0892,
      "step": 385
    },
    {
      "epoch": 1.2253968253968255,
      "grad_norm": 0.8369996547698975,
      "learning_rate": 1.1830687830687833e-05,
      "loss": 0.1275,
      "step": 386
    },
    {
      "epoch": 1.2285714285714286,
      "grad_norm": 0.9498575329780579,
      "learning_rate": 1.180952380952381e-05,
      "loss": 0.0801,
      "step": 387
    },
    {
      "epoch": 1.2317460317460318,
      "grad_norm": 0.9700520038604736,
      "learning_rate": 1.1788359788359788e-05,
      "loss": 0.1674,
      "step": 388
    },
    {
      "epoch": 1.234920634920635,
      "grad_norm": 2.180983304977417,
      "learning_rate": 1.1767195767195766e-05,
      "loss": 0.142,
      "step": 389
    },
    {
      "epoch": 1.2380952380952381,
      "grad_norm": 1.365975260734558,
      "learning_rate": 1.1746031746031748e-05,
      "loss": 0.1416,
      "step": 390
    },
    {
      "epoch": 1.2412698412698413,
      "grad_norm": 0.8028814792633057,
      "learning_rate": 1.1724867724867726e-05,
      "loss": 0.1218,
      "step": 391
    },
    {
      "epoch": 1.2444444444444445,
      "grad_norm": 0.5584167242050171,
      "learning_rate": 1.1703703703703703e-05,
      "loss": 0.1399,
      "step": 392
    },
    {
      "epoch": 1.2476190476190476,
      "grad_norm": 2.2481939792633057,
      "learning_rate": 1.1682539682539685e-05,
      "loss": 0.1633,
      "step": 393
    },
    {
      "epoch": 1.2507936507936508,
      "grad_norm": 0.882199227809906,
      "learning_rate": 1.1661375661375663e-05,
      "loss": 0.1537,
      "step": 394
    },
    {
      "epoch": 1.253968253968254,
      "grad_norm": 1.5113340616226196,
      "learning_rate": 1.1640211640211641e-05,
      "loss": 0.2221,
      "step": 395
    },
    {
      "epoch": 1.2571428571428571,
      "grad_norm": 0.42928898334503174,
      "learning_rate": 1.1619047619047621e-05,
      "loss": 0.0991,
      "step": 396
    },
    {
      "epoch": 1.2603174603174603,
      "grad_norm": 1.0709649324417114,
      "learning_rate": 1.15978835978836e-05,
      "loss": 0.1671,
      "step": 397
    },
    {
      "epoch": 1.2634920634920634,
      "grad_norm": 0.8396578431129456,
      "learning_rate": 1.1576719576719578e-05,
      "loss": 0.1541,
      "step": 398
    },
    {
      "epoch": 1.2666666666666666,
      "grad_norm": 0.6324909329414368,
      "learning_rate": 1.1555555555555556e-05,
      "loss": 0.088,
      "step": 399
    },
    {
      "epoch": 1.2698412698412698,
      "grad_norm": 0.5754315853118896,
      "learning_rate": 1.1534391534391536e-05,
      "loss": 0.111,
      "step": 400
    },
    {
      "epoch": 1.273015873015873,
      "grad_norm": 1.0489622354507446,
      "learning_rate": 1.1513227513227514e-05,
      "loss": 0.1047,
      "step": 401
    },
    {
      "epoch": 1.276190476190476,
      "grad_norm": 0.8893101215362549,
      "learning_rate": 1.1492063492063492e-05,
      "loss": 0.1209,
      "step": 402
    },
    {
      "epoch": 1.2793650793650793,
      "grad_norm": 0.9546168446540833,
      "learning_rate": 1.1470899470899472e-05,
      "loss": 0.1244,
      "step": 403
    },
    {
      "epoch": 1.2825396825396824,
      "grad_norm": 1.0668022632598877,
      "learning_rate": 1.144973544973545e-05,
      "loss": 0.1584,
      "step": 404
    },
    {
      "epoch": 1.2857142857142856,
      "grad_norm": 0.7670823931694031,
      "learning_rate": 1.1428571428571429e-05,
      "loss": 0.1262,
      "step": 405
    },
    {
      "epoch": 1.2888888888888888,
      "grad_norm": 0.4818229377269745,
      "learning_rate": 1.1407407407407409e-05,
      "loss": 0.0887,
      "step": 406
    },
    {
      "epoch": 1.2920634920634921,
      "grad_norm": 1.0843604803085327,
      "learning_rate": 1.1386243386243387e-05,
      "loss": 0.1524,
      "step": 407
    },
    {
      "epoch": 1.2952380952380953,
      "grad_norm": 0.47771039605140686,
      "learning_rate": 1.1365079365079366e-05,
      "loss": 0.0963,
      "step": 408
    },
    {
      "epoch": 1.2984126984126985,
      "grad_norm": 0.8418334126472473,
      "learning_rate": 1.1343915343915344e-05,
      "loss": 0.1089,
      "step": 409
    },
    {
      "epoch": 1.3015873015873016,
      "grad_norm": 1.0364947319030762,
      "learning_rate": 1.1322751322751324e-05,
      "loss": 0.1417,
      "step": 410
    },
    {
      "epoch": 1.3047619047619048,
      "grad_norm": 0.5752162337303162,
      "learning_rate": 1.1301587301587302e-05,
      "loss": 0.0636,
      "step": 411
    },
    {
      "epoch": 1.307936507936508,
      "grad_norm": 0.883034884929657,
      "learning_rate": 1.128042328042328e-05,
      "loss": 0.1372,
      "step": 412
    },
    {
      "epoch": 1.3111111111111111,
      "grad_norm": 0.7002597451210022,
      "learning_rate": 1.125925925925926e-05,
      "loss": 0.1602,
      "step": 413
    },
    {
      "epoch": 1.3142857142857143,
      "grad_norm": 0.44680556654930115,
      "learning_rate": 1.1238095238095239e-05,
      "loss": 0.093,
      "step": 414
    },
    {
      "epoch": 1.3174603174603174,
      "grad_norm": 1.561047911643982,
      "learning_rate": 1.1216931216931217e-05,
      "loss": 0.1506,
      "step": 415
    },
    {
      "epoch": 1.3206349206349206,
      "grad_norm": 1.0878902673721313,
      "learning_rate": 1.1195767195767197e-05,
      "loss": 0.1151,
      "step": 416
    },
    {
      "epoch": 1.3238095238095238,
      "grad_norm": 0.42074644565582275,
      "learning_rate": 1.1174603174603175e-05,
      "loss": 0.0718,
      "step": 417
    },
    {
      "epoch": 1.326984126984127,
      "grad_norm": 0.5986080169677734,
      "learning_rate": 1.1153439153439153e-05,
      "loss": 0.0857,
      "step": 418
    },
    {
      "epoch": 1.33015873015873,
      "grad_norm": 0.5089951157569885,
      "learning_rate": 1.1132275132275132e-05,
      "loss": 0.109,
      "step": 419
    },
    {
      "epoch": 1.3333333333333333,
      "grad_norm": 1.1861498355865479,
      "learning_rate": 1.1111111111111113e-05,
      "loss": 0.1404,
      "step": 420
    },
    {
      "epoch": 1.3365079365079366,
      "grad_norm": 0.48356208205223083,
      "learning_rate": 1.108994708994709e-05,
      "loss": 0.0771,
      "step": 421
    },
    {
      "epoch": 1.3396825396825398,
      "grad_norm": 0.5442419052124023,
      "learning_rate": 1.1068783068783068e-05,
      "loss": 0.0717,
      "step": 422
    },
    {
      "epoch": 1.342857142857143,
      "grad_norm": 0.8858305811882019,
      "learning_rate": 1.104761904761905e-05,
      "loss": 0.142,
      "step": 423
    },
    {
      "epoch": 1.3460317460317461,
      "grad_norm": 0.5416662096977234,
      "learning_rate": 1.1026455026455028e-05,
      "loss": 0.0831,
      "step": 424
    },
    {
      "epoch": 1.3492063492063493,
      "grad_norm": 1.2472801208496094,
      "learning_rate": 1.1005291005291006e-05,
      "loss": 0.1309,
      "step": 425
    },
    {
      "epoch": 1.3523809523809525,
      "grad_norm": 1.3547828197479248,
      "learning_rate": 1.0984126984126986e-05,
      "loss": 0.1147,
      "step": 426
    },
    {
      "epoch": 1.3555555555555556,
      "grad_norm": 0.3764377236366272,
      "learning_rate": 1.0962962962962965e-05,
      "loss": 0.0656,
      "step": 427
    },
    {
      "epoch": 1.3587301587301588,
      "grad_norm": 0.9181296825408936,
      "learning_rate": 1.0941798941798943e-05,
      "loss": 0.111,
      "step": 428
    },
    {
      "epoch": 1.361904761904762,
      "grad_norm": 0.768558919429779,
      "learning_rate": 1.0920634920634921e-05,
      "loss": 0.0961,
      "step": 429
    },
    {
      "epoch": 1.3650793650793651,
      "grad_norm": 0.7232595682144165,
      "learning_rate": 1.0899470899470901e-05,
      "loss": 0.1384,
      "step": 430
    },
    {
      "epoch": 1.3682539682539683,
      "grad_norm": 0.9489431381225586,
      "learning_rate": 1.087830687830688e-05,
      "loss": 0.1402,
      "step": 431
    },
    {
      "epoch": 1.3714285714285714,
      "grad_norm": 0.4146624207496643,
      "learning_rate": 1.0857142857142858e-05,
      "loss": 0.0422,
      "step": 432
    },
    {
      "epoch": 1.3746031746031746,
      "grad_norm": 0.8513931035995483,
      "learning_rate": 1.0835978835978838e-05,
      "loss": 0.1086,
      "step": 433
    },
    {
      "epoch": 1.3777777777777778,
      "grad_norm": 1.0045173168182373,
      "learning_rate": 1.0814814814814816e-05,
      "loss": 0.2016,
      "step": 434
    },
    {
      "epoch": 1.380952380952381,
      "grad_norm": 0.7364886403083801,
      "learning_rate": 1.0793650793650794e-05,
      "loss": 0.1337,
      "step": 435
    },
    {
      "epoch": 1.384126984126984,
      "grad_norm": 1.0422242879867554,
      "learning_rate": 1.0772486772486774e-05,
      "loss": 0.1257,
      "step": 436
    },
    {
      "epoch": 1.3873015873015873,
      "grad_norm": 3.243945837020874,
      "learning_rate": 1.0751322751322753e-05,
      "loss": 0.2538,
      "step": 437
    },
    {
      "epoch": 1.3904761904761904,
      "grad_norm": 0.8591335415840149,
      "learning_rate": 1.073015873015873e-05,
      "loss": 0.1001,
      "step": 438
    },
    {
      "epoch": 1.3936507936507936,
      "grad_norm": 1.002109408378601,
      "learning_rate": 1.0708994708994709e-05,
      "loss": 0.0755,
      "step": 439
    },
    {
      "epoch": 1.3968253968253967,
      "grad_norm": 0.9193840622901917,
      "learning_rate": 1.0687830687830689e-05,
      "loss": 0.0575,
      "step": 440
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.6536378860473633,
      "learning_rate": 1.0666666666666667e-05,
      "loss": 0.1013,
      "step": 441
    },
    {
      "epoch": 1.403174603174603,
      "grad_norm": 0.9502435326576233,
      "learning_rate": 1.0645502645502646e-05,
      "loss": 0.1423,
      "step": 442
    },
    {
      "epoch": 1.4063492063492062,
      "grad_norm": 1.0123416185379028,
      "learning_rate": 1.0624338624338626e-05,
      "loss": 0.1411,
      "step": 443
    },
    {
      "epoch": 1.4095238095238094,
      "grad_norm": 1.6801286935806274,
      "learning_rate": 1.0603174603174604e-05,
      "loss": 0.1101,
      "step": 444
    },
    {
      "epoch": 1.4126984126984126,
      "grad_norm": 0.9140623211860657,
      "learning_rate": 1.0582010582010582e-05,
      "loss": 0.1266,
      "step": 445
    },
    {
      "epoch": 1.415873015873016,
      "grad_norm": 0.835747241973877,
      "learning_rate": 1.0560846560846562e-05,
      "loss": 0.148,
      "step": 446
    },
    {
      "epoch": 1.4190476190476191,
      "grad_norm": 0.49627140164375305,
      "learning_rate": 1.053968253968254e-05,
      "loss": 0.1441,
      "step": 447
    },
    {
      "epoch": 1.4222222222222223,
      "grad_norm": 0.9323590993881226,
      "learning_rate": 1.0518518518518519e-05,
      "loss": 0.1535,
      "step": 448
    },
    {
      "epoch": 1.4253968253968254,
      "grad_norm": 0.6182277202606201,
      "learning_rate": 1.0497354497354497e-05,
      "loss": 0.0785,
      "step": 449
    },
    {
      "epoch": 1.4285714285714286,
      "grad_norm": 0.6031960844993591,
      "learning_rate": 1.0476190476190477e-05,
      "loss": 0.106,
      "step": 450
    },
    {
      "epoch": 1.4317460317460318,
      "grad_norm": 1.0345180034637451,
      "learning_rate": 1.0455026455026455e-05,
      "loss": 0.1437,
      "step": 451
    },
    {
      "epoch": 1.434920634920635,
      "grad_norm": 1.7959545850753784,
      "learning_rate": 1.0433862433862433e-05,
      "loss": 0.0751,
      "step": 452
    },
    {
      "epoch": 1.438095238095238,
      "grad_norm": 1.4171229600906372,
      "learning_rate": 1.0412698412698415e-05,
      "loss": 0.1188,
      "step": 453
    },
    {
      "epoch": 1.4412698412698413,
      "grad_norm": 1.0954288244247437,
      "learning_rate": 1.0391534391534392e-05,
      "loss": 0.1259,
      "step": 454
    },
    {
      "epoch": 1.4444444444444444,
      "grad_norm": 0.6407185792922974,
      "learning_rate": 1.037037037037037e-05,
      "loss": 0.1041,
      "step": 455
    },
    {
      "epoch": 1.4476190476190476,
      "grad_norm": 0.5068238377571106,
      "learning_rate": 1.0349206349206352e-05,
      "loss": 0.0894,
      "step": 456
    },
    {
      "epoch": 1.4507936507936507,
      "grad_norm": 0.754906952381134,
      "learning_rate": 1.032804232804233e-05,
      "loss": 0.099,
      "step": 457
    },
    {
      "epoch": 1.453968253968254,
      "grad_norm": 0.5085781812667847,
      "learning_rate": 1.0306878306878308e-05,
      "loss": 0.0636,
      "step": 458
    },
    {
      "epoch": 1.457142857142857,
      "grad_norm": 0.7322441935539246,
      "learning_rate": 1.0285714285714285e-05,
      "loss": 0.1222,
      "step": 459
    },
    {
      "epoch": 1.4603174603174602,
      "grad_norm": 0.6341411471366882,
      "learning_rate": 1.0264550264550266e-05,
      "loss": 0.134,
      "step": 460
    },
    {
      "epoch": 1.4634920634920636,
      "grad_norm": 0.9442792534828186,
      "learning_rate": 1.0243386243386245e-05,
      "loss": 0.1049,
      "step": 461
    },
    {
      "epoch": 1.4666666666666668,
      "grad_norm": 1.2521939277648926,
      "learning_rate": 1.0222222222222223e-05,
      "loss": 0.1157,
      "step": 462
    },
    {
      "epoch": 1.46984126984127,
      "grad_norm": 0.6030839681625366,
      "learning_rate": 1.0201058201058203e-05,
      "loss": 0.158,
      "step": 463
    },
    {
      "epoch": 1.4730158730158731,
      "grad_norm": 1.1996594667434692,
      "learning_rate": 1.0179894179894181e-05,
      "loss": 0.1324,
      "step": 464
    },
    {
      "epoch": 1.4761904761904763,
      "grad_norm": 0.629490077495575,
      "learning_rate": 1.015873015873016e-05,
      "loss": 0.1301,
      "step": 465
    },
    {
      "epoch": 1.4793650793650794,
      "grad_norm": 1.0028965473175049,
      "learning_rate": 1.013756613756614e-05,
      "loss": 0.1091,
      "step": 466
    },
    {
      "epoch": 1.4825396825396826,
      "grad_norm": 0.5149552226066589,
      "learning_rate": 1.0116402116402118e-05,
      "loss": 0.0701,
      "step": 467
    },
    {
      "epoch": 1.4857142857142858,
      "grad_norm": 0.971042811870575,
      "learning_rate": 1.0095238095238096e-05,
      "loss": 0.1023,
      "step": 468
    },
    {
      "epoch": 1.488888888888889,
      "grad_norm": 0.41423165798187256,
      "learning_rate": 1.0074074074074074e-05,
      "loss": 0.0722,
      "step": 469
    },
    {
      "epoch": 1.492063492063492,
      "grad_norm": 0.8377816081047058,
      "learning_rate": 1.0052910052910054e-05,
      "loss": 0.1919,
      "step": 470
    },
    {
      "epoch": 1.4952380952380953,
      "grad_norm": 1.606553554534912,
      "learning_rate": 1.0031746031746033e-05,
      "loss": 0.1165,
      "step": 471
    },
    {
      "epoch": 1.4984126984126984,
      "grad_norm": 0.5874265432357788,
      "learning_rate": 1.001058201058201e-05,
      "loss": 0.0976,
      "step": 472
    },
    {
      "epoch": 1.5015873015873016,
      "grad_norm": 1.1042734384536743,
      "learning_rate": 9.989417989417989e-06,
      "loss": 0.1813,
      "step": 473
    },
    {
      "epoch": 1.5047619047619047,
      "grad_norm": 1.1558371782302856,
      "learning_rate": 9.968253968253969e-06,
      "loss": 0.1781,
      "step": 474
    },
    {
      "epoch": 1.507936507936508,
      "grad_norm": 1.2665374279022217,
      "learning_rate": 9.947089947089947e-06,
      "loss": 0.1717,
      "step": 475
    },
    {
      "epoch": 1.511111111111111,
      "grad_norm": 1.099094033241272,
      "learning_rate": 9.925925925925927e-06,
      "loss": 0.0751,
      "step": 476
    },
    {
      "epoch": 1.5142857142857142,
      "grad_norm": 0.669716477394104,
      "learning_rate": 9.904761904761906e-06,
      "loss": 0.1271,
      "step": 477
    },
    {
      "epoch": 1.5174603174603174,
      "grad_norm": 0.4861948788166046,
      "learning_rate": 9.883597883597884e-06,
      "loss": 0.083,
      "step": 478
    },
    {
      "epoch": 1.5206349206349206,
      "grad_norm": 1.1369725465774536,
      "learning_rate": 9.862433862433864e-06,
      "loss": 0.1287,
      "step": 479
    },
    {
      "epoch": 1.5238095238095237,
      "grad_norm": 0.8680781722068787,
      "learning_rate": 9.841269841269842e-06,
      "loss": 0.171,
      "step": 480
    },
    {
      "epoch": 1.5269841269841269,
      "grad_norm": 0.7066832184791565,
      "learning_rate": 9.82010582010582e-06,
      "loss": 0.163,
      "step": 481
    },
    {
      "epoch": 1.53015873015873,
      "grad_norm": 0.4178638756275177,
      "learning_rate": 9.7989417989418e-06,
      "loss": 0.1025,
      "step": 482
    },
    {
      "epoch": 1.5333333333333332,
      "grad_norm": 0.8113085031509399,
      "learning_rate": 9.777777777777779e-06,
      "loss": 0.1282,
      "step": 483
    },
    {
      "epoch": 1.5365079365079364,
      "grad_norm": 0.43854859471321106,
      "learning_rate": 9.756613756613757e-06,
      "loss": 0.0737,
      "step": 484
    },
    {
      "epoch": 1.5396825396825395,
      "grad_norm": 0.5710107088088989,
      "learning_rate": 9.735449735449735e-06,
      "loss": 0.0757,
      "step": 485
    },
    {
      "epoch": 1.5428571428571427,
      "grad_norm": 0.648597002029419,
      "learning_rate": 9.714285714285715e-06,
      "loss": 0.1247,
      "step": 486
    },
    {
      "epoch": 1.5460317460317459,
      "grad_norm": 0.6734611988067627,
      "learning_rate": 9.693121693121693e-06,
      "loss": 0.1283,
      "step": 487
    },
    {
      "epoch": 1.5492063492063493,
      "grad_norm": 0.9539926648139954,
      "learning_rate": 9.671957671957672e-06,
      "loss": 0.099,
      "step": 488
    },
    {
      "epoch": 1.5523809523809524,
      "grad_norm": 0.5695520639419556,
      "learning_rate": 9.650793650793652e-06,
      "loss": 0.1106,
      "step": 489
    },
    {
      "epoch": 1.5555555555555556,
      "grad_norm": 0.8400272727012634,
      "learning_rate": 9.62962962962963e-06,
      "loss": 0.1532,
      "step": 490
    },
    {
      "epoch": 1.5587301587301587,
      "grad_norm": 1.0214587450027466,
      "learning_rate": 9.60846560846561e-06,
      "loss": 0.1158,
      "step": 491
    },
    {
      "epoch": 1.561904761904762,
      "grad_norm": 0.9745279550552368,
      "learning_rate": 9.587301587301588e-06,
      "loss": 0.146,
      "step": 492
    },
    {
      "epoch": 1.565079365079365,
      "grad_norm": 0.6575698256492615,
      "learning_rate": 9.566137566137567e-06,
      "loss": 0.0779,
      "step": 493
    },
    {
      "epoch": 1.5682539682539682,
      "grad_norm": 2.248106002807617,
      "learning_rate": 9.544973544973546e-06,
      "loss": 0.2046,
      "step": 494
    },
    {
      "epoch": 1.5714285714285714,
      "grad_norm": 0.35883384943008423,
      "learning_rate": 9.523809523809525e-06,
      "loss": 0.0506,
      "step": 495
    },
    {
      "epoch": 1.5746031746031746,
      "grad_norm": 0.7438737154006958,
      "learning_rate": 9.502645502645503e-06,
      "loss": 0.0998,
      "step": 496
    },
    {
      "epoch": 1.5777777777777777,
      "grad_norm": 0.7771816849708557,
      "learning_rate": 9.481481481481483e-06,
      "loss": 0.1545,
      "step": 497
    },
    {
      "epoch": 1.580952380952381,
      "grad_norm": 1.4896730184555054,
      "learning_rate": 9.460317460317461e-06,
      "loss": 0.1694,
      "step": 498
    },
    {
      "epoch": 1.5841269841269843,
      "grad_norm": 0.5341618657112122,
      "learning_rate": 9.43915343915344e-06,
      "loss": 0.0701,
      "step": 499
    },
    {
      "epoch": 1.5873015873015874,
      "grad_norm": 0.6365366578102112,
      "learning_rate": 9.417989417989418e-06,
      "loss": 0.1173,
      "step": 500
    },
    {
      "epoch": 1.5904761904761906,
      "grad_norm": 0.9295374155044556,
      "learning_rate": 9.396825396825398e-06,
      "loss": 0.0994,
      "step": 501
    },
    {
      "epoch": 1.5936507936507938,
      "grad_norm": 0.7063966989517212,
      "learning_rate": 9.375661375661376e-06,
      "loss": 0.0868,
      "step": 502
    },
    {
      "epoch": 1.596825396825397,
      "grad_norm": 1.7500840425491333,
      "learning_rate": 9.354497354497354e-06,
      "loss": 0.1348,
      "step": 503
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.6251238584518433,
      "learning_rate": 9.333333333333334e-06,
      "loss": 0.072,
      "step": 504
    },
    {
      "epoch": 1.6031746031746033,
      "grad_norm": 0.5500174760818481,
      "learning_rate": 9.312169312169313e-06,
      "loss": 0.0974,
      "step": 505
    },
    {
      "epoch": 1.6063492063492064,
      "grad_norm": 0.6557893753051758,
      "learning_rate": 9.291005291005291e-06,
      "loss": 0.0752,
      "step": 506
    },
    {
      "epoch": 1.6095238095238096,
      "grad_norm": 0.4798218607902527,
      "learning_rate": 9.26984126984127e-06,
      "loss": 0.1283,
      "step": 507
    },
    {
      "epoch": 1.6126984126984127,
      "grad_norm": 0.6837607622146606,
      "learning_rate": 9.248677248677249e-06,
      "loss": 0.0791,
      "step": 508
    },
    {
      "epoch": 1.615873015873016,
      "grad_norm": 1.7913333177566528,
      "learning_rate": 9.227513227513229e-06,
      "loss": 0.1687,
      "step": 509
    },
    {
      "epoch": 1.619047619047619,
      "grad_norm": 1.1361005306243896,
      "learning_rate": 9.206349206349207e-06,
      "loss": 0.1057,
      "step": 510
    },
    {
      "epoch": 1.6222222222222222,
      "grad_norm": 1.6567833423614502,
      "learning_rate": 9.185185185185186e-06,
      "loss": 0.1278,
      "step": 511
    },
    {
      "epoch": 1.6253968253968254,
      "grad_norm": 1.154585599899292,
      "learning_rate": 9.164021164021166e-06,
      "loss": 0.1208,
      "step": 512
    },
    {
      "epoch": 1.6285714285714286,
      "grad_norm": 1.131409764289856,
      "learning_rate": 9.142857142857144e-06,
      "loss": 0.0973,
      "step": 513
    },
    {
      "epoch": 1.6317460317460317,
      "grad_norm": 0.991510272026062,
      "learning_rate": 9.121693121693122e-06,
      "loss": 0.0861,
      "step": 514
    },
    {
      "epoch": 1.6349206349206349,
      "grad_norm": 0.7155593633651733,
      "learning_rate": 9.1005291005291e-06,
      "loss": 0.106,
      "step": 515
    },
    {
      "epoch": 1.638095238095238,
      "grad_norm": 1.5030924081802368,
      "learning_rate": 9.07936507936508e-06,
      "loss": 0.1214,
      "step": 516
    },
    {
      "epoch": 1.6412698412698412,
      "grad_norm": 0.8777892589569092,
      "learning_rate": 9.058201058201059e-06,
      "loss": 0.1157,
      "step": 517
    },
    {
      "epoch": 1.6444444444444444,
      "grad_norm": 1.2287359237670898,
      "learning_rate": 9.037037037037037e-06,
      "loss": 0.0879,
      "step": 518
    },
    {
      "epoch": 1.6476190476190475,
      "grad_norm": 0.45202577114105225,
      "learning_rate": 9.015873015873017e-06,
      "loss": 0.1006,
      "step": 519
    },
    {
      "epoch": 1.6507936507936507,
      "grad_norm": 0.46383896470069885,
      "learning_rate": 8.994708994708995e-06,
      "loss": 0.1152,
      "step": 520
    },
    {
      "epoch": 1.6539682539682539,
      "grad_norm": 0.6837895512580872,
      "learning_rate": 8.973544973544973e-06,
      "loss": 0.1063,
      "step": 521
    },
    {
      "epoch": 1.657142857142857,
      "grad_norm": 0.9806374311447144,
      "learning_rate": 8.952380952380953e-06,
      "loss": 0.1182,
      "step": 522
    },
    {
      "epoch": 1.6603174603174602,
      "grad_norm": 1.8979694843292236,
      "learning_rate": 8.931216931216932e-06,
      "loss": 0.1465,
      "step": 523
    },
    {
      "epoch": 1.6634920634920634,
      "grad_norm": 0.8033770322799683,
      "learning_rate": 8.910052910052912e-06,
      "loss": 0.1605,
      "step": 524
    },
    {
      "epoch": 1.6666666666666665,
      "grad_norm": 1.2472139596939087,
      "learning_rate": 8.888888888888888e-06,
      "loss": 0.1135,
      "step": 525
    },
    {
      "epoch": 1.6698412698412697,
      "grad_norm": 0.8076494336128235,
      "learning_rate": 8.867724867724868e-06,
      "loss": 0.173,
      "step": 526
    },
    {
      "epoch": 1.673015873015873,
      "grad_norm": 1.030251383781433,
      "learning_rate": 8.846560846560848e-06,
      "loss": 0.1176,
      "step": 527
    },
    {
      "epoch": 1.6761904761904762,
      "grad_norm": 0.6217860579490662,
      "learning_rate": 8.825396825396827e-06,
      "loss": 0.0803,
      "step": 528
    },
    {
      "epoch": 1.6793650793650794,
      "grad_norm": 0.4145120084285736,
      "learning_rate": 8.804232804232805e-06,
      "loss": 0.0651,
      "step": 529
    },
    {
      "epoch": 1.6825396825396826,
      "grad_norm": 2.2825121879577637,
      "learning_rate": 8.783068783068783e-06,
      "loss": 0.173,
      "step": 530
    },
    {
      "epoch": 1.6857142857142857,
      "grad_norm": 0.7827410101890564,
      "learning_rate": 8.761904761904763e-06,
      "loss": 0.1484,
      "step": 531
    },
    {
      "epoch": 1.6888888888888889,
      "grad_norm": 1.9165403842926025,
      "learning_rate": 8.740740740740741e-06,
      "loss": 0.1681,
      "step": 532
    },
    {
      "epoch": 1.692063492063492,
      "grad_norm": 1.80486261844635,
      "learning_rate": 8.71957671957672e-06,
      "loss": 0.1342,
      "step": 533
    },
    {
      "epoch": 1.6952380952380952,
      "grad_norm": 0.8068901896476746,
      "learning_rate": 8.6984126984127e-06,
      "loss": 0.0774,
      "step": 534
    },
    {
      "epoch": 1.6984126984126984,
      "grad_norm": 1.1465476751327515,
      "learning_rate": 8.677248677248678e-06,
      "loss": 0.0734,
      "step": 535
    },
    {
      "epoch": 1.7015873015873015,
      "grad_norm": 1.4191033840179443,
      "learning_rate": 8.656084656084656e-06,
      "loss": 0.0878,
      "step": 536
    },
    {
      "epoch": 1.704761904761905,
      "grad_norm": 0.8128553628921509,
      "learning_rate": 8.634920634920636e-06,
      "loss": 0.1779,
      "step": 537
    },
    {
      "epoch": 1.707936507936508,
      "grad_norm": 1.0794931650161743,
      "learning_rate": 8.613756613756614e-06,
      "loss": 0.1264,
      "step": 538
    },
    {
      "epoch": 1.7111111111111112,
      "grad_norm": 0.6890401244163513,
      "learning_rate": 8.592592592592593e-06,
      "loss": 0.0868,
      "step": 539
    },
    {
      "epoch": 1.7142857142857144,
      "grad_norm": 0.7165282964706421,
      "learning_rate": 8.571428571428571e-06,
      "loss": 0.161,
      "step": 540
    },
    {
      "epoch": 1.7174603174603176,
      "grad_norm": 0.7967664003372192,
      "learning_rate": 8.550264550264551e-06,
      "loss": 0.1021,
      "step": 541
    },
    {
      "epoch": 1.7206349206349207,
      "grad_norm": 0.6971952319145203,
      "learning_rate": 8.529100529100531e-06,
      "loss": 0.0845,
      "step": 542
    },
    {
      "epoch": 1.723809523809524,
      "grad_norm": 0.46871671080589294,
      "learning_rate": 8.507936507936509e-06,
      "loss": 0.0621,
      "step": 543
    },
    {
      "epoch": 1.726984126984127,
      "grad_norm": 1.7202202081680298,
      "learning_rate": 8.486772486772487e-06,
      "loss": 0.1626,
      "step": 544
    },
    {
      "epoch": 1.7301587301587302,
      "grad_norm": 0.6655100584030151,
      "learning_rate": 8.465608465608466e-06,
      "loss": 0.1451,
      "step": 545
    },
    {
      "epoch": 1.7333333333333334,
      "grad_norm": 0.38534989953041077,
      "learning_rate": 8.444444444444446e-06,
      "loss": 0.0853,
      "step": 546
    },
    {
      "epoch": 1.7365079365079366,
      "grad_norm": 0.5553216934204102,
      "learning_rate": 8.423280423280424e-06,
      "loss": 0.0769,
      "step": 547
    },
    {
      "epoch": 1.7396825396825397,
      "grad_norm": 1.5887691974639893,
      "learning_rate": 8.402116402116402e-06,
      "loss": 0.0727,
      "step": 548
    },
    {
      "epoch": 1.7428571428571429,
      "grad_norm": 1.2850600481033325,
      "learning_rate": 8.380952380952382e-06,
      "loss": 0.1493,
      "step": 549
    },
    {
      "epoch": 1.746031746031746,
      "grad_norm": 0.9276643991470337,
      "learning_rate": 8.35978835978836e-06,
      "loss": 0.1196,
      "step": 550
    },
    {
      "epoch": 1.7492063492063492,
      "grad_norm": 0.9015336632728577,
      "learning_rate": 8.338624338624339e-06,
      "loss": 0.0984,
      "step": 551
    },
    {
      "epoch": 1.7523809523809524,
      "grad_norm": 0.7600472569465637,
      "learning_rate": 8.317460317460319e-06,
      "loss": 0.1071,
      "step": 552
    },
    {
      "epoch": 1.7555555555555555,
      "grad_norm": 0.9144738912582397,
      "learning_rate": 8.296296296296297e-06,
      "loss": 0.087,
      "step": 553
    },
    {
      "epoch": 1.7587301587301587,
      "grad_norm": 1.0788496732711792,
      "learning_rate": 8.275132275132275e-06,
      "loss": 0.0952,
      "step": 554
    },
    {
      "epoch": 1.7619047619047619,
      "grad_norm": 1.1808207035064697,
      "learning_rate": 8.253968253968254e-06,
      "loss": 0.1143,
      "step": 555
    },
    {
      "epoch": 1.765079365079365,
      "grad_norm": 0.5436956882476807,
      "learning_rate": 8.232804232804234e-06,
      "loss": 0.1108,
      "step": 556
    },
    {
      "epoch": 1.7682539682539682,
      "grad_norm": 0.3755543529987335,
      "learning_rate": 8.211640211640213e-06,
      "loss": 0.0947,
      "step": 557
    },
    {
      "epoch": 1.7714285714285714,
      "grad_norm": 2.325793743133545,
      "learning_rate": 8.190476190476192e-06,
      "loss": 0.1716,
      "step": 558
    },
    {
      "epoch": 1.7746031746031745,
      "grad_norm": 0.8447362184524536,
      "learning_rate": 8.16931216931217e-06,
      "loss": 0.108,
      "step": 559
    },
    {
      "epoch": 1.7777777777777777,
      "grad_norm": 0.947551429271698,
      "learning_rate": 8.148148148148148e-06,
      "loss": 0.094,
      "step": 560
    },
    {
      "epoch": 1.7809523809523808,
      "grad_norm": 1.8139454126358032,
      "learning_rate": 8.126984126984128e-06,
      "loss": 0.0897,
      "step": 561
    },
    {
      "epoch": 1.784126984126984,
      "grad_norm": 2.12823486328125,
      "learning_rate": 8.105820105820107e-06,
      "loss": 0.1166,
      "step": 562
    },
    {
      "epoch": 1.7873015873015872,
      "grad_norm": 0.8317103981971741,
      "learning_rate": 8.084656084656085e-06,
      "loss": 0.1167,
      "step": 563
    },
    {
      "epoch": 1.7904761904761903,
      "grad_norm": 0.797480046749115,
      "learning_rate": 8.063492063492065e-06,
      "loss": 0.1526,
      "step": 564
    },
    {
      "epoch": 1.7936507936507935,
      "grad_norm": 0.4185366630554199,
      "learning_rate": 8.042328042328043e-06,
      "loss": 0.0834,
      "step": 565
    },
    {
      "epoch": 1.7968253968253967,
      "grad_norm": 0.8348795175552368,
      "learning_rate": 8.021164021164021e-06,
      "loss": 0.1133,
      "step": 566
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.4025468826293945,
      "learning_rate": 8.000000000000001e-06,
      "loss": 0.1262,
      "step": 567
    },
    {
      "epoch": 1.8031746031746032,
      "grad_norm": 0.4260364770889282,
      "learning_rate": 7.97883597883598e-06,
      "loss": 0.0449,
      "step": 568
    },
    {
      "epoch": 1.8063492063492064,
      "grad_norm": 0.8140082955360413,
      "learning_rate": 7.957671957671958e-06,
      "loss": 0.1455,
      "step": 569
    },
    {
      "epoch": 1.8095238095238095,
      "grad_norm": 0.8299204111099243,
      "learning_rate": 7.936507936507936e-06,
      "loss": 0.1659,
      "step": 570
    },
    {
      "epoch": 1.8126984126984127,
      "grad_norm": 0.6794671416282654,
      "learning_rate": 7.915343915343916e-06,
      "loss": 0.1042,
      "step": 571
    },
    {
      "epoch": 1.8158730158730159,
      "grad_norm": 0.5644881725311279,
      "learning_rate": 7.894179894179896e-06,
      "loss": 0.0796,
      "step": 572
    },
    {
      "epoch": 1.819047619047619,
      "grad_norm": 1.543426513671875,
      "learning_rate": 7.873015873015873e-06,
      "loss": 0.1012,
      "step": 573
    },
    {
      "epoch": 1.8222222222222222,
      "grad_norm": 0.5663846731185913,
      "learning_rate": 7.851851851851853e-06,
      "loss": 0.1157,
      "step": 574
    },
    {
      "epoch": 1.8253968253968254,
      "grad_norm": 1.3915711641311646,
      "learning_rate": 7.830687830687831e-06,
      "loss": 0.1278,
      "step": 575
    },
    {
      "epoch": 1.8285714285714287,
      "grad_norm": 1.3186984062194824,
      "learning_rate": 7.809523809523811e-06,
      "loss": 0.0999,
      "step": 576
    },
    {
      "epoch": 1.831746031746032,
      "grad_norm": 1.1411281824111938,
      "learning_rate": 7.78835978835979e-06,
      "loss": 0.1124,
      "step": 577
    },
    {
      "epoch": 1.834920634920635,
      "grad_norm": 1.359994888305664,
      "learning_rate": 7.767195767195767e-06,
      "loss": 0.1503,
      "step": 578
    },
    {
      "epoch": 1.8380952380952382,
      "grad_norm": 1.3508297204971313,
      "learning_rate": 7.746031746031747e-06,
      "loss": 0.1875,
      "step": 579
    },
    {
      "epoch": 1.8412698412698414,
      "grad_norm": 0.40984153747558594,
      "learning_rate": 7.724867724867726e-06,
      "loss": 0.0895,
      "step": 580
    },
    {
      "epoch": 1.8444444444444446,
      "grad_norm": 0.8926004767417908,
      "learning_rate": 7.703703703703704e-06,
      "loss": 0.1123,
      "step": 581
    },
    {
      "epoch": 1.8476190476190477,
      "grad_norm": 0.8729626536369324,
      "learning_rate": 7.682539682539684e-06,
      "loss": 0.0852,
      "step": 582
    },
    {
      "epoch": 1.8507936507936509,
      "grad_norm": 1.0873273611068726,
      "learning_rate": 7.661375661375662e-06,
      "loss": 0.1193,
      "step": 583
    },
    {
      "epoch": 1.853968253968254,
      "grad_norm": 0.8294671177864075,
      "learning_rate": 7.64021164021164e-06,
      "loss": 0.0961,
      "step": 584
    },
    {
      "epoch": 1.8571428571428572,
      "grad_norm": 0.868336021900177,
      "learning_rate": 7.61904761904762e-06,
      "loss": 0.1158,
      "step": 585
    },
    {
      "epoch": 1.8603174603174604,
      "grad_norm": 1.773681879043579,
      "learning_rate": 7.597883597883599e-06,
      "loss": 0.1108,
      "step": 586
    },
    {
      "epoch": 1.8634920634920635,
      "grad_norm": 0.7267278432846069,
      "learning_rate": 7.576719576719578e-06,
      "loss": 0.0864,
      "step": 587
    },
    {
      "epoch": 1.8666666666666667,
      "grad_norm": 1.9358700513839722,
      "learning_rate": 7.555555555555556e-06,
      "loss": 0.1447,
      "step": 588
    },
    {
      "epoch": 1.8698412698412699,
      "grad_norm": 1.2288655042648315,
      "learning_rate": 7.534391534391535e-06,
      "loss": 0.1123,
      "step": 589
    },
    {
      "epoch": 1.873015873015873,
      "grad_norm": 0.5252324342727661,
      "learning_rate": 7.5132275132275136e-06,
      "loss": 0.0837,
      "step": 590
    },
    {
      "epoch": 1.8761904761904762,
      "grad_norm": 2.124035120010376,
      "learning_rate": 7.492063492063493e-06,
      "loss": 0.1675,
      "step": 591
    },
    {
      "epoch": 1.8793650793650793,
      "grad_norm": 0.3034132122993469,
      "learning_rate": 7.470899470899472e-06,
      "loss": 0.0749,
      "step": 592
    },
    {
      "epoch": 1.8825396825396825,
      "grad_norm": 0.8175469040870667,
      "learning_rate": 7.44973544973545e-06,
      "loss": 0.1146,
      "step": 593
    },
    {
      "epoch": 1.8857142857142857,
      "grad_norm": 1.0407495498657227,
      "learning_rate": 7.428571428571429e-06,
      "loss": 0.1496,
      "step": 594
    },
    {
      "epoch": 1.8888888888888888,
      "grad_norm": 0.7908574342727661,
      "learning_rate": 7.4074074074074075e-06,
      "loss": 0.1351,
      "step": 595
    },
    {
      "epoch": 1.892063492063492,
      "grad_norm": 0.5789360404014587,
      "learning_rate": 7.386243386243387e-06,
      "loss": 0.1128,
      "step": 596
    },
    {
      "epoch": 1.8952380952380952,
      "grad_norm": 0.6722256541252136,
      "learning_rate": 7.3650793650793666e-06,
      "loss": 0.1215,
      "step": 597
    },
    {
      "epoch": 1.8984126984126983,
      "grad_norm": 1.1160768270492554,
      "learning_rate": 7.343915343915344e-06,
      "loss": 0.1323,
      "step": 598
    },
    {
      "epoch": 1.9015873015873015,
      "grad_norm": 1.629210352897644,
      "learning_rate": 7.322751322751324e-06,
      "loss": 0.146,
      "step": 599
    },
    {
      "epoch": 1.9047619047619047,
      "grad_norm": 1.4182788133621216,
      "learning_rate": 7.301587301587301e-06,
      "loss": 0.1113,
      "step": 600
    },
    {
      "epoch": 1.9079365079365078,
      "grad_norm": 0.38122087717056274,
      "learning_rate": 7.280423280423281e-06,
      "loss": 0.0736,
      "step": 601
    },
    {
      "epoch": 1.911111111111111,
      "grad_norm": 0.6037584543228149,
      "learning_rate": 7.2592592592592605e-06,
      "loss": 0.1171,
      "step": 602
    },
    {
      "epoch": 1.9142857142857141,
      "grad_norm": 0.6923559308052063,
      "learning_rate": 7.238095238095239e-06,
      "loss": 0.0879,
      "step": 603
    },
    {
      "epoch": 1.9174603174603173,
      "grad_norm": 1.0224826335906982,
      "learning_rate": 7.216931216931218e-06,
      "loss": 0.1489,
      "step": 604
    },
    {
      "epoch": 1.9206349206349205,
      "grad_norm": 0.39160704612731934,
      "learning_rate": 7.195767195767196e-06,
      "loss": 0.0534,
      "step": 605
    },
    {
      "epoch": 1.9238095238095239,
      "grad_norm": 1.5186269283294678,
      "learning_rate": 7.174603174603175e-06,
      "loss": 0.1025,
      "step": 606
    },
    {
      "epoch": 1.926984126984127,
      "grad_norm": 1.641759991645813,
      "learning_rate": 7.1534391534391544e-06,
      "loss": 0.1511,
      "step": 607
    },
    {
      "epoch": 1.9301587301587302,
      "grad_norm": 1.1003851890563965,
      "learning_rate": 7.132275132275133e-06,
      "loss": 0.1253,
      "step": 608
    },
    {
      "epoch": 1.9333333333333333,
      "grad_norm": 0.45853930711746216,
      "learning_rate": 7.111111111111112e-06,
      "loss": 0.0819,
      "step": 609
    },
    {
      "epoch": 1.9365079365079365,
      "grad_norm": 1.1490768194198608,
      "learning_rate": 7.08994708994709e-06,
      "loss": 0.1103,
      "step": 610
    },
    {
      "epoch": 1.9396825396825397,
      "grad_norm": 0.9988818168640137,
      "learning_rate": 7.068783068783069e-06,
      "loss": 0.0441,
      "step": 611
    },
    {
      "epoch": 1.9428571428571428,
      "grad_norm": 0.8255701661109924,
      "learning_rate": 7.047619047619048e-06,
      "loss": 0.1868,
      "step": 612
    },
    {
      "epoch": 1.946031746031746,
      "grad_norm": 0.6808100938796997,
      "learning_rate": 7.026455026455027e-06,
      "loss": 0.0796,
      "step": 613
    },
    {
      "epoch": 1.9492063492063492,
      "grad_norm": 0.839131772518158,
      "learning_rate": 7.005291005291006e-06,
      "loss": 0.0812,
      "step": 614
    },
    {
      "epoch": 1.9523809523809523,
      "grad_norm": 1.2730712890625,
      "learning_rate": 6.984126984126984e-06,
      "loss": 0.1331,
      "step": 615
    },
    {
      "epoch": 1.9555555555555557,
      "grad_norm": 0.6310842037200928,
      "learning_rate": 6.962962962962964e-06,
      "loss": 0.1158,
      "step": 616
    },
    {
      "epoch": 1.9587301587301589,
      "grad_norm": 0.9359753727912903,
      "learning_rate": 6.941798941798943e-06,
      "loss": 0.1069,
      "step": 617
    },
    {
      "epoch": 1.961904761904762,
      "grad_norm": 1.2897759675979614,
      "learning_rate": 6.920634920634921e-06,
      "loss": 0.0955,
      "step": 618
    },
    {
      "epoch": 1.9650793650793652,
      "grad_norm": 0.4904727637767792,
      "learning_rate": 6.8994708994709005e-06,
      "loss": 0.093,
      "step": 619
    },
    {
      "epoch": 1.9682539682539684,
      "grad_norm": 0.8107464909553528,
      "learning_rate": 6.878306878306879e-06,
      "loss": 0.0924,
      "step": 620
    },
    {
      "epoch": 1.9714285714285715,
      "grad_norm": 0.6122804880142212,
      "learning_rate": 6.857142857142858e-06,
      "loss": 0.0941,
      "step": 621
    },
    {
      "epoch": 1.9746031746031747,
      "grad_norm": 0.8837588429450989,
      "learning_rate": 6.835978835978837e-06,
      "loss": 0.1106,
      "step": 622
    },
    {
      "epoch": 1.9777777777777779,
      "grad_norm": 0.7891656160354614,
      "learning_rate": 6.814814814814815e-06,
      "loss": 0.0748,
      "step": 623
    },
    {
      "epoch": 1.980952380952381,
      "grad_norm": 0.5850039124488831,
      "learning_rate": 6.7936507936507944e-06,
      "loss": 0.0618,
      "step": 624
    },
    {
      "epoch": 1.9841269841269842,
      "grad_norm": 0.7541988492012024,
      "learning_rate": 6.772486772486773e-06,
      "loss": 0.0578,
      "step": 625
    },
    {
      "epoch": 1.9873015873015873,
      "grad_norm": 1.2322847843170166,
      "learning_rate": 6.751322751322752e-06,
      "loss": 0.0935,
      "step": 626
    },
    {
      "epoch": 1.9904761904761905,
      "grad_norm": 1.4187729358673096,
      "learning_rate": 6.730158730158731e-06,
      "loss": 0.1235,
      "step": 627
    },
    {
      "epoch": 1.9936507936507937,
      "grad_norm": 0.8365176916122437,
      "learning_rate": 6.708994708994709e-06,
      "loss": 0.1006,
      "step": 628
    },
    {
      "epoch": 1.9968253968253968,
      "grad_norm": 0.5329151749610901,
      "learning_rate": 6.687830687830688e-06,
      "loss": 0.1063,
      "step": 629
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.6751351952552795,
      "learning_rate": 6.666666666666667e-06,
      "loss": 0.0676,
      "step": 630
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.7705357142857143,
      "eval_f1": 0.9055459272097055,
      "eval_loss": 0.1228988915681839,
      "eval_runtime": 1.7929,
      "eval_samples_per_second": 624.692,
      "eval_steps_per_second": 19.522,
      "step": 630
    },
    {
      "epoch": 2.003174603174603,
      "grad_norm": 0.5677501559257507,
      "learning_rate": 6.645502645502646e-06,
      "loss": 0.1069,
      "step": 631
    },
    {
      "epoch": 2.0063492063492063,
      "grad_norm": 0.5953556299209595,
      "learning_rate": 6.624338624338626e-06,
      "loss": 0.0618,
      "step": 632
    },
    {
      "epoch": 2.0095238095238095,
      "grad_norm": 1.0832258462905884,
      "learning_rate": 6.603174603174603e-06,
      "loss": 0.1192,
      "step": 633
    },
    {
      "epoch": 2.0126984126984127,
      "grad_norm": 1.079674482345581,
      "learning_rate": 6.582010582010583e-06,
      "loss": 0.115,
      "step": 634
    },
    {
      "epoch": 2.015873015873016,
      "grad_norm": 0.911268949508667,
      "learning_rate": 6.560846560846561e-06,
      "loss": 0.1321,
      "step": 635
    },
    {
      "epoch": 2.019047619047619,
      "grad_norm": 1.3121854066848755,
      "learning_rate": 6.5396825396825405e-06,
      "loss": 0.1029,
      "step": 636
    },
    {
      "epoch": 2.022222222222222,
      "grad_norm": 1.6950548887252808,
      "learning_rate": 6.51851851851852e-06,
      "loss": 0.1825,
      "step": 637
    },
    {
      "epoch": 2.0253968253968253,
      "grad_norm": 0.5674888491630554,
      "learning_rate": 6.497354497354498e-06,
      "loss": 0.1008,
      "step": 638
    },
    {
      "epoch": 2.0285714285714285,
      "grad_norm": 1.5443412065505981,
      "learning_rate": 6.476190476190477e-06,
      "loss": 0.1776,
      "step": 639
    },
    {
      "epoch": 2.0317460317460316,
      "grad_norm": 0.5738489627838135,
      "learning_rate": 6.455026455026455e-06,
      "loss": 0.0542,
      "step": 640
    },
    {
      "epoch": 2.034920634920635,
      "grad_norm": 1.294780969619751,
      "learning_rate": 6.4338624338624345e-06,
      "loss": 0.1492,
      "step": 641
    },
    {
      "epoch": 2.038095238095238,
      "grad_norm": 0.8680015206336975,
      "learning_rate": 6.412698412698414e-06,
      "loss": 0.1061,
      "step": 642
    },
    {
      "epoch": 2.041269841269841,
      "grad_norm": 0.7214539051055908,
      "learning_rate": 6.391534391534392e-06,
      "loss": 0.0913,
      "step": 643
    },
    {
      "epoch": 2.0444444444444443,
      "grad_norm": 0.9868113398551941,
      "learning_rate": 6.370370370370371e-06,
      "loss": 0.1301,
      "step": 644
    },
    {
      "epoch": 2.0476190476190474,
      "grad_norm": 0.8200106620788574,
      "learning_rate": 6.349206349206349e-06,
      "loss": 0.0704,
      "step": 645
    },
    {
      "epoch": 2.0507936507936506,
      "grad_norm": 0.5085193514823914,
      "learning_rate": 6.328042328042328e-06,
      "loss": 0.0569,
      "step": 646
    },
    {
      "epoch": 2.0539682539682538,
      "grad_norm": 0.7045906186103821,
      "learning_rate": 6.3068783068783075e-06,
      "loss": 0.1047,
      "step": 647
    },
    {
      "epoch": 2.057142857142857,
      "grad_norm": 0.6057743430137634,
      "learning_rate": 6.285714285714286e-06,
      "loss": 0.1143,
      "step": 648
    },
    {
      "epoch": 2.06031746031746,
      "grad_norm": 0.4814365804195404,
      "learning_rate": 6.264550264550266e-06,
      "loss": 0.0552,
      "step": 649
    },
    {
      "epoch": 2.0634920634920633,
      "grad_norm": 0.9408270716667175,
      "learning_rate": 6.243386243386243e-06,
      "loss": 0.1447,
      "step": 650
    },
    {
      "epoch": 2.066666666666667,
      "grad_norm": 0.4177621901035309,
      "learning_rate": 6.222222222222223e-06,
      "loss": 0.0956,
      "step": 651
    },
    {
      "epoch": 2.06984126984127,
      "grad_norm": 1.1701905727386475,
      "learning_rate": 6.201058201058202e-06,
      "loss": 0.1479,
      "step": 652
    },
    {
      "epoch": 2.073015873015873,
      "grad_norm": 0.6085675358772278,
      "learning_rate": 6.1798941798941806e-06,
      "loss": 0.1103,
      "step": 653
    },
    {
      "epoch": 2.0761904761904764,
      "grad_norm": 0.7830865979194641,
      "learning_rate": 6.15873015873016e-06,
      "loss": 0.097,
      "step": 654
    },
    {
      "epoch": 2.0793650793650795,
      "grad_norm": 0.5818350315093994,
      "learning_rate": 6.137566137566138e-06,
      "loss": 0.1056,
      "step": 655
    },
    {
      "epoch": 2.0825396825396827,
      "grad_norm": 1.025743007659912,
      "learning_rate": 6.116402116402117e-06,
      "loss": 0.2149,
      "step": 656
    },
    {
      "epoch": 2.085714285714286,
      "grad_norm": 1.5567277669906616,
      "learning_rate": 6.095238095238096e-06,
      "loss": 0.1273,
      "step": 657
    },
    {
      "epoch": 2.088888888888889,
      "grad_norm": 0.5851786732673645,
      "learning_rate": 6.0740740740740745e-06,
      "loss": 0.0992,
      "step": 658
    },
    {
      "epoch": 2.092063492063492,
      "grad_norm": 1.044635534286499,
      "learning_rate": 6.052910052910054e-06,
      "loss": 0.0941,
      "step": 659
    },
    {
      "epoch": 2.0952380952380953,
      "grad_norm": 0.6114885210990906,
      "learning_rate": 6.031746031746032e-06,
      "loss": 0.1036,
      "step": 660
    },
    {
      "epoch": 2.0984126984126985,
      "grad_norm": 0.49471500515937805,
      "learning_rate": 6.010582010582011e-06,
      "loss": 0.0674,
      "step": 661
    },
    {
      "epoch": 2.1015873015873017,
      "grad_norm": 0.6662114262580872,
      "learning_rate": 5.989417989417989e-06,
      "loss": 0.0596,
      "step": 662
    },
    {
      "epoch": 2.104761904761905,
      "grad_norm": 1.3240714073181152,
      "learning_rate": 5.968253968253968e-06,
      "loss": 0.1146,
      "step": 663
    },
    {
      "epoch": 2.107936507936508,
      "grad_norm": 0.7612365484237671,
      "learning_rate": 5.9470899470899475e-06,
      "loss": 0.0663,
      "step": 664
    },
    {
      "epoch": 2.111111111111111,
      "grad_norm": 0.4063294529914856,
      "learning_rate": 5.925925925925926e-06,
      "loss": 0.0719,
      "step": 665
    },
    {
      "epoch": 2.1142857142857143,
      "grad_norm": 0.5315088033676147,
      "learning_rate": 5.904761904761905e-06,
      "loss": 0.1343,
      "step": 666
    },
    {
      "epoch": 2.1174603174603175,
      "grad_norm": 0.976586103439331,
      "learning_rate": 5.883597883597883e-06,
      "loss": 0.114,
      "step": 667
    },
    {
      "epoch": 2.1206349206349207,
      "grad_norm": 0.8407710194587708,
      "learning_rate": 5.862433862433863e-06,
      "loss": 0.0897,
      "step": 668
    },
    {
      "epoch": 2.123809523809524,
      "grad_norm": 1.1143242120742798,
      "learning_rate": 5.841269841269842e-06,
      "loss": 0.0834,
      "step": 669
    },
    {
      "epoch": 2.126984126984127,
      "grad_norm": 1.2708462476730347,
      "learning_rate": 5.820105820105821e-06,
      "loss": 0.1344,
      "step": 670
    },
    {
      "epoch": 2.13015873015873,
      "grad_norm": 1.386818289756775,
      "learning_rate": 5.7989417989418e-06,
      "loss": 0.108,
      "step": 671
    },
    {
      "epoch": 2.1333333333333333,
      "grad_norm": 0.9245710372924805,
      "learning_rate": 5.777777777777778e-06,
      "loss": 0.0969,
      "step": 672
    },
    {
      "epoch": 2.1365079365079365,
      "grad_norm": 1.0209077596664429,
      "learning_rate": 5.756613756613757e-06,
      "loss": 0.1246,
      "step": 673
    },
    {
      "epoch": 2.1396825396825396,
      "grad_norm": 0.9524020552635193,
      "learning_rate": 5.735449735449736e-06,
      "loss": 0.1045,
      "step": 674
    },
    {
      "epoch": 2.142857142857143,
      "grad_norm": 1.3763508796691895,
      "learning_rate": 5.7142857142857145e-06,
      "loss": 0.1577,
      "step": 675
    },
    {
      "epoch": 2.146031746031746,
      "grad_norm": 0.7689756155014038,
      "learning_rate": 5.693121693121694e-06,
      "loss": 0.1152,
      "step": 676
    },
    {
      "epoch": 2.149206349206349,
      "grad_norm": 0.8806107044219971,
      "learning_rate": 5.671957671957672e-06,
      "loss": 0.1004,
      "step": 677
    },
    {
      "epoch": 2.1523809523809523,
      "grad_norm": 0.8259665369987488,
      "learning_rate": 5.650793650793651e-06,
      "loss": 0.1373,
      "step": 678
    },
    {
      "epoch": 2.1555555555555554,
      "grad_norm": 0.8434117436408997,
      "learning_rate": 5.62962962962963e-06,
      "loss": 0.0842,
      "step": 679
    },
    {
      "epoch": 2.1587301587301586,
      "grad_norm": 1.112399697303772,
      "learning_rate": 5.6084656084656084e-06,
      "loss": 0.1326,
      "step": 680
    },
    {
      "epoch": 2.1619047619047618,
      "grad_norm": 0.7231675982475281,
      "learning_rate": 5.5873015873015876e-06,
      "loss": 0.1018,
      "step": 681
    },
    {
      "epoch": 2.165079365079365,
      "grad_norm": 0.9234125018119812,
      "learning_rate": 5.566137566137566e-06,
      "loss": 0.1641,
      "step": 682
    },
    {
      "epoch": 2.168253968253968,
      "grad_norm": 0.9991856217384338,
      "learning_rate": 5.544973544973545e-06,
      "loss": 0.1093,
      "step": 683
    },
    {
      "epoch": 2.1714285714285713,
      "grad_norm": 0.8652076721191406,
      "learning_rate": 5.523809523809525e-06,
      "loss": 0.1004,
      "step": 684
    },
    {
      "epoch": 2.1746031746031744,
      "grad_norm": 0.7533403635025024,
      "learning_rate": 5.502645502645503e-06,
      "loss": 0.1094,
      "step": 685
    },
    {
      "epoch": 2.1777777777777776,
      "grad_norm": 1.3620589971542358,
      "learning_rate": 5.481481481481482e-06,
      "loss": 0.1544,
      "step": 686
    },
    {
      "epoch": 2.1809523809523808,
      "grad_norm": 0.6398664116859436,
      "learning_rate": 5.460317460317461e-06,
      "loss": 0.0666,
      "step": 687
    },
    {
      "epoch": 2.1841269841269844,
      "grad_norm": 0.6695507168769836,
      "learning_rate": 5.43915343915344e-06,
      "loss": 0.0903,
      "step": 688
    },
    {
      "epoch": 2.1873015873015875,
      "grad_norm": 1.1547316312789917,
      "learning_rate": 5.417989417989419e-06,
      "loss": 0.1217,
      "step": 689
    },
    {
      "epoch": 2.1904761904761907,
      "grad_norm": 0.8324596285820007,
      "learning_rate": 5.396825396825397e-06,
      "loss": 0.1296,
      "step": 690
    },
    {
      "epoch": 2.193650793650794,
      "grad_norm": 0.584938108921051,
      "learning_rate": 5.375661375661376e-06,
      "loss": 0.0851,
      "step": 691
    },
    {
      "epoch": 2.196825396825397,
      "grad_norm": 0.8484377861022949,
      "learning_rate": 5.3544973544973545e-06,
      "loss": 0.0974,
      "step": 692
    },
    {
      "epoch": 2.2,
      "grad_norm": 0.7113614082336426,
      "learning_rate": 5.333333333333334e-06,
      "loss": 0.0579,
      "step": 693
    },
    {
      "epoch": 2.2031746031746033,
      "grad_norm": 0.8446152806282043,
      "learning_rate": 5.312169312169313e-06,
      "loss": 0.1067,
      "step": 694
    },
    {
      "epoch": 2.2063492063492065,
      "grad_norm": 0.8121873736381531,
      "learning_rate": 5.291005291005291e-06,
      "loss": 0.0915,
      "step": 695
    },
    {
      "epoch": 2.2095238095238097,
      "grad_norm": 0.9365905523300171,
      "learning_rate": 5.26984126984127e-06,
      "loss": 0.1019,
      "step": 696
    },
    {
      "epoch": 2.212698412698413,
      "grad_norm": 0.9007267951965332,
      "learning_rate": 5.2486772486772485e-06,
      "loss": 0.0816,
      "step": 697
    },
    {
      "epoch": 2.215873015873016,
      "grad_norm": 0.5663630366325378,
      "learning_rate": 5.227513227513228e-06,
      "loss": 0.0724,
      "step": 698
    },
    {
      "epoch": 2.219047619047619,
      "grad_norm": 0.8520302176475525,
      "learning_rate": 5.2063492063492076e-06,
      "loss": 0.0651,
      "step": 699
    },
    {
      "epoch": 2.2222222222222223,
      "grad_norm": 0.6091964244842529,
      "learning_rate": 5.185185185185185e-06,
      "loss": 0.0661,
      "step": 700
    },
    {
      "epoch": 2.2253968253968255,
      "grad_norm": 0.8232845664024353,
      "learning_rate": 5.164021164021165e-06,
      "loss": 0.097,
      "step": 701
    },
    {
      "epoch": 2.2285714285714286,
      "grad_norm": 0.5729593634605408,
      "learning_rate": 5.142857142857142e-06,
      "loss": 0.0673,
      "step": 702
    },
    {
      "epoch": 2.231746031746032,
      "grad_norm": 0.8750647306442261,
      "learning_rate": 5.121693121693122e-06,
      "loss": 0.1167,
      "step": 703
    },
    {
      "epoch": 2.234920634920635,
      "grad_norm": 0.6703619956970215,
      "learning_rate": 5.1005291005291015e-06,
      "loss": 0.1041,
      "step": 704
    },
    {
      "epoch": 2.238095238095238,
      "grad_norm": 0.5414075255393982,
      "learning_rate": 5.07936507936508e-06,
      "loss": 0.0736,
      "step": 705
    },
    {
      "epoch": 2.2412698412698413,
      "grad_norm": 0.8254547119140625,
      "learning_rate": 5.058201058201059e-06,
      "loss": 0.1507,
      "step": 706
    },
    {
      "epoch": 2.2444444444444445,
      "grad_norm": 0.923128604888916,
      "learning_rate": 5.037037037037037e-06,
      "loss": 0.0844,
      "step": 707
    },
    {
      "epoch": 2.2476190476190476,
      "grad_norm": 0.5627334713935852,
      "learning_rate": 5.015873015873016e-06,
      "loss": 0.0836,
      "step": 708
    },
    {
      "epoch": 2.250793650793651,
      "grad_norm": 0.6246154308319092,
      "learning_rate": 4.9947089947089946e-06,
      "loss": 0.083,
      "step": 709
    },
    {
      "epoch": 2.253968253968254,
      "grad_norm": 0.5700716376304626,
      "learning_rate": 4.973544973544974e-06,
      "loss": 0.0589,
      "step": 710
    },
    {
      "epoch": 2.257142857142857,
      "grad_norm": 0.675677478313446,
      "learning_rate": 4.952380952380953e-06,
      "loss": 0.0905,
      "step": 711
    },
    {
      "epoch": 2.2603174603174603,
      "grad_norm": 1.1091946363449097,
      "learning_rate": 4.931216931216932e-06,
      "loss": 0.0932,
      "step": 712
    },
    {
      "epoch": 2.2634920634920634,
      "grad_norm": 0.7644129395484924,
      "learning_rate": 4.91005291005291e-06,
      "loss": 0.0695,
      "step": 713
    },
    {
      "epoch": 2.2666666666666666,
      "grad_norm": 0.5853104591369629,
      "learning_rate": 4.888888888888889e-06,
      "loss": 0.0838,
      "step": 714
    },
    {
      "epoch": 2.2698412698412698,
      "grad_norm": 0.45854365825653076,
      "learning_rate": 4.867724867724868e-06,
      "loss": 0.0631,
      "step": 715
    },
    {
      "epoch": 2.273015873015873,
      "grad_norm": 1.552655577659607,
      "learning_rate": 4.846560846560847e-06,
      "loss": 0.0879,
      "step": 716
    },
    {
      "epoch": 2.276190476190476,
      "grad_norm": 0.5847281217575073,
      "learning_rate": 4.825396825396826e-06,
      "loss": 0.0856,
      "step": 717
    },
    {
      "epoch": 2.2793650793650793,
      "grad_norm": 0.8583001494407654,
      "learning_rate": 4.804232804232805e-06,
      "loss": 0.1018,
      "step": 718
    },
    {
      "epoch": 2.2825396825396824,
      "grad_norm": 0.48612359166145325,
      "learning_rate": 4.783068783068783e-06,
      "loss": 0.0628,
      "step": 719
    },
    {
      "epoch": 2.2857142857142856,
      "grad_norm": 0.6603057384490967,
      "learning_rate": 4.761904761904762e-06,
      "loss": 0.101,
      "step": 720
    },
    {
      "epoch": 2.2888888888888888,
      "grad_norm": 0.7552214860916138,
      "learning_rate": 4.7407407407407415e-06,
      "loss": 0.0551,
      "step": 721
    },
    {
      "epoch": 2.292063492063492,
      "grad_norm": 0.5127883553504944,
      "learning_rate": 4.71957671957672e-06,
      "loss": 0.0555,
      "step": 722
    },
    {
      "epoch": 2.295238095238095,
      "grad_norm": 1.3932037353515625,
      "learning_rate": 4.698412698412699e-06,
      "loss": 0.122,
      "step": 723
    },
    {
      "epoch": 2.2984126984126982,
      "grad_norm": 0.6060371398925781,
      "learning_rate": 4.677248677248677e-06,
      "loss": 0.093,
      "step": 724
    },
    {
      "epoch": 2.3015873015873014,
      "grad_norm": 1.4660332202911377,
      "learning_rate": 4.656084656084656e-06,
      "loss": 0.1068,
      "step": 725
    },
    {
      "epoch": 2.3047619047619046,
      "grad_norm": 0.5299437642097473,
      "learning_rate": 4.634920634920635e-06,
      "loss": 0.1051,
      "step": 726
    },
    {
      "epoch": 2.3079365079365077,
      "grad_norm": 1.433962106704712,
      "learning_rate": 4.6137566137566145e-06,
      "loss": 0.1299,
      "step": 727
    },
    {
      "epoch": 2.311111111111111,
      "grad_norm": 0.5282876491546631,
      "learning_rate": 4.592592592592593e-06,
      "loss": 0.0735,
      "step": 728
    },
    {
      "epoch": 2.314285714285714,
      "grad_norm": 1.0122612714767456,
      "learning_rate": 4.571428571428572e-06,
      "loss": 0.114,
      "step": 729
    },
    {
      "epoch": 2.317460317460317,
      "grad_norm": 0.675662100315094,
      "learning_rate": 4.55026455026455e-06,
      "loss": 0.0905,
      "step": 730
    },
    {
      "epoch": 2.320634920634921,
      "grad_norm": 0.5909380912780762,
      "learning_rate": 4.529100529100529e-06,
      "loss": 0.0841,
      "step": 731
    },
    {
      "epoch": 2.323809523809524,
      "grad_norm": 0.5422948598861694,
      "learning_rate": 4.5079365079365085e-06,
      "loss": 0.0702,
      "step": 732
    },
    {
      "epoch": 2.326984126984127,
      "grad_norm": 0.4477941691875458,
      "learning_rate": 4.486772486772487e-06,
      "loss": 0.0907,
      "step": 733
    },
    {
      "epoch": 2.3301587301587303,
      "grad_norm": 0.5739128589630127,
      "learning_rate": 4.465608465608466e-06,
      "loss": 0.1033,
      "step": 734
    },
    {
      "epoch": 2.3333333333333335,
      "grad_norm": 1.242101788520813,
      "learning_rate": 4.444444444444444e-06,
      "loss": 0.0966,
      "step": 735
    },
    {
      "epoch": 2.3365079365079366,
      "grad_norm": 1.1147149801254272,
      "learning_rate": 4.423280423280424e-06,
      "loss": 0.1284,
      "step": 736
    },
    {
      "epoch": 2.33968253968254,
      "grad_norm": 0.9129058122634888,
      "learning_rate": 4.402116402116402e-06,
      "loss": 0.1166,
      "step": 737
    },
    {
      "epoch": 2.342857142857143,
      "grad_norm": 1.0708431005477905,
      "learning_rate": 4.3809523809523815e-06,
      "loss": 0.1033,
      "step": 738
    },
    {
      "epoch": 2.346031746031746,
      "grad_norm": 1.3865693807601929,
      "learning_rate": 4.35978835978836e-06,
      "loss": 0.0963,
      "step": 739
    },
    {
      "epoch": 2.3492063492063493,
      "grad_norm": 1.9295120239257812,
      "learning_rate": 4.338624338624339e-06,
      "loss": 0.1189,
      "step": 740
    },
    {
      "epoch": 2.3523809523809525,
      "grad_norm": 0.8500393629074097,
      "learning_rate": 4.317460317460318e-06,
      "loss": 0.1147,
      "step": 741
    },
    {
      "epoch": 2.3555555555555556,
      "grad_norm": 1.174766182899475,
      "learning_rate": 4.296296296296296e-06,
      "loss": 0.0759,
      "step": 742
    },
    {
      "epoch": 2.358730158730159,
      "grad_norm": 0.7025968432426453,
      "learning_rate": 4.2751322751322754e-06,
      "loss": 0.0836,
      "step": 743
    },
    {
      "epoch": 2.361904761904762,
      "grad_norm": 0.6774269938468933,
      "learning_rate": 4.2539682539682546e-06,
      "loss": 0.0921,
      "step": 744
    },
    {
      "epoch": 2.365079365079365,
      "grad_norm": 0.6932780146598816,
      "learning_rate": 4.232804232804233e-06,
      "loss": 0.0855,
      "step": 745
    },
    {
      "epoch": 2.3682539682539683,
      "grad_norm": 0.9605720043182373,
      "learning_rate": 4.211640211640212e-06,
      "loss": 0.1366,
      "step": 746
    },
    {
      "epoch": 2.3714285714285714,
      "grad_norm": 0.8600250482559204,
      "learning_rate": 4.190476190476191e-06,
      "loss": 0.1426,
      "step": 747
    },
    {
      "epoch": 2.3746031746031746,
      "grad_norm": 0.6296926736831665,
      "learning_rate": 4.169312169312169e-06,
      "loss": 0.0921,
      "step": 748
    },
    {
      "epoch": 2.3777777777777778,
      "grad_norm": 1.0476994514465332,
      "learning_rate": 4.1481481481481485e-06,
      "loss": 0.0623,
      "step": 749
    },
    {
      "epoch": 2.380952380952381,
      "grad_norm": 1.8777590990066528,
      "learning_rate": 4.126984126984127e-06,
      "loss": 0.0875,
      "step": 750
    },
    {
      "epoch": 2.384126984126984,
      "grad_norm": 1.5180679559707642,
      "learning_rate": 4.105820105820107e-06,
      "loss": 0.1284,
      "step": 751
    },
    {
      "epoch": 2.3873015873015873,
      "grad_norm": 0.8465520739555359,
      "learning_rate": 4.084656084656085e-06,
      "loss": 0.1098,
      "step": 752
    },
    {
      "epoch": 2.3904761904761904,
      "grad_norm": 1.4953113794326782,
      "learning_rate": 4.063492063492064e-06,
      "loss": 0.1436,
      "step": 753
    },
    {
      "epoch": 2.3936507936507936,
      "grad_norm": 0.8960604071617126,
      "learning_rate": 4.042328042328042e-06,
      "loss": 0.0868,
      "step": 754
    },
    {
      "epoch": 2.3968253968253967,
      "grad_norm": 1.3076069355010986,
      "learning_rate": 4.0211640211640215e-06,
      "loss": 0.1564,
      "step": 755
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2682548761367798,
      "learning_rate": 4.000000000000001e-06,
      "loss": 0.1189,
      "step": 756
    },
    {
      "epoch": 2.403174603174603,
      "grad_norm": 0.352100133895874,
      "learning_rate": 3.978835978835979e-06,
      "loss": 0.0874,
      "step": 757
    },
    {
      "epoch": 2.4063492063492062,
      "grad_norm": 1.2029849290847778,
      "learning_rate": 3.957671957671958e-06,
      "loss": 0.1624,
      "step": 758
    },
    {
      "epoch": 2.4095238095238094,
      "grad_norm": 0.5802411437034607,
      "learning_rate": 3.936507936507936e-06,
      "loss": 0.0646,
      "step": 759
    },
    {
      "epoch": 2.4126984126984126,
      "grad_norm": 0.6743428111076355,
      "learning_rate": 3.9153439153439155e-06,
      "loss": 0.0764,
      "step": 760
    },
    {
      "epoch": 2.4158730158730157,
      "grad_norm": 1.734810709953308,
      "learning_rate": 3.894179894179895e-06,
      "loss": 0.1334,
      "step": 761
    },
    {
      "epoch": 2.419047619047619,
      "grad_norm": 0.6183326840400696,
      "learning_rate": 3.873015873015874e-06,
      "loss": 0.0794,
      "step": 762
    },
    {
      "epoch": 2.422222222222222,
      "grad_norm": 3.1293373107910156,
      "learning_rate": 3.851851851851852e-06,
      "loss": 0.1701,
      "step": 763
    },
    {
      "epoch": 2.425396825396825,
      "grad_norm": 0.7868879437446594,
      "learning_rate": 3.830687830687831e-06,
      "loss": 0.0794,
      "step": 764
    },
    {
      "epoch": 2.4285714285714284,
      "grad_norm": 2.1806819438934326,
      "learning_rate": 3.80952380952381e-06,
      "loss": 0.1565,
      "step": 765
    },
    {
      "epoch": 2.431746031746032,
      "grad_norm": 1.0895839929580688,
      "learning_rate": 3.788359788359789e-06,
      "loss": 0.109,
      "step": 766
    },
    {
      "epoch": 2.434920634920635,
      "grad_norm": 0.694896936416626,
      "learning_rate": 3.7671957671957676e-06,
      "loss": 0.091,
      "step": 767
    },
    {
      "epoch": 2.4380952380952383,
      "grad_norm": 0.9746938347816467,
      "learning_rate": 3.7460317460317463e-06,
      "loss": 0.1086,
      "step": 768
    },
    {
      "epoch": 2.4412698412698415,
      "grad_norm": 0.8791543841362,
      "learning_rate": 3.724867724867725e-06,
      "loss": 0.1382,
      "step": 769
    },
    {
      "epoch": 2.4444444444444446,
      "grad_norm": 0.9889678955078125,
      "learning_rate": 3.7037037037037037e-06,
      "loss": 0.0702,
      "step": 770
    },
    {
      "epoch": 2.447619047619048,
      "grad_norm": 0.7683724761009216,
      "learning_rate": 3.6825396825396833e-06,
      "loss": 0.0741,
      "step": 771
    },
    {
      "epoch": 2.450793650793651,
      "grad_norm": 0.618735671043396,
      "learning_rate": 3.661375661375662e-06,
      "loss": 0.0924,
      "step": 772
    },
    {
      "epoch": 2.453968253968254,
      "grad_norm": 0.6444486379623413,
      "learning_rate": 3.6402116402116407e-06,
      "loss": 0.106,
      "step": 773
    },
    {
      "epoch": 2.4571428571428573,
      "grad_norm": 1.0320513248443604,
      "learning_rate": 3.6190476190476194e-06,
      "loss": 0.1248,
      "step": 774
    },
    {
      "epoch": 2.4603174603174605,
      "grad_norm": 1.5131829977035522,
      "learning_rate": 3.597883597883598e-06,
      "loss": 0.1331,
      "step": 775
    },
    {
      "epoch": 2.4634920634920636,
      "grad_norm": 0.4671209752559662,
      "learning_rate": 3.5767195767195772e-06,
      "loss": 0.0834,
      "step": 776
    },
    {
      "epoch": 2.466666666666667,
      "grad_norm": 1.0072500705718994,
      "learning_rate": 3.555555555555556e-06,
      "loss": 0.0917,
      "step": 777
    },
    {
      "epoch": 2.46984126984127,
      "grad_norm": 1.027008295059204,
      "learning_rate": 3.5343915343915346e-06,
      "loss": 0.125,
      "step": 778
    },
    {
      "epoch": 2.473015873015873,
      "grad_norm": 1.1031194925308228,
      "learning_rate": 3.5132275132275133e-06,
      "loss": 0.1286,
      "step": 779
    },
    {
      "epoch": 2.4761904761904763,
      "grad_norm": 0.6343787908554077,
      "learning_rate": 3.492063492063492e-06,
      "loss": 0.1128,
      "step": 780
    },
    {
      "epoch": 2.4793650793650794,
      "grad_norm": 1.582716464996338,
      "learning_rate": 3.4708994708994716e-06,
      "loss": 0.0961,
      "step": 781
    },
    {
      "epoch": 2.4825396825396826,
      "grad_norm": 0.62384033203125,
      "learning_rate": 3.4497354497354503e-06,
      "loss": 0.0904,
      "step": 782
    },
    {
      "epoch": 2.4857142857142858,
      "grad_norm": 1.096556544303894,
      "learning_rate": 3.428571428571429e-06,
      "loss": 0.1013,
      "step": 783
    },
    {
      "epoch": 2.488888888888889,
      "grad_norm": 0.8943139314651489,
      "learning_rate": 3.4074074074074077e-06,
      "loss": 0.1228,
      "step": 784
    },
    {
      "epoch": 2.492063492063492,
      "grad_norm": 0.5017927289009094,
      "learning_rate": 3.3862433862433864e-06,
      "loss": 0.0686,
      "step": 785
    },
    {
      "epoch": 2.4952380952380953,
      "grad_norm": 0.38915061950683594,
      "learning_rate": 3.3650793650793655e-06,
      "loss": 0.0476,
      "step": 786
    },
    {
      "epoch": 2.4984126984126984,
      "grad_norm": 0.9441356062889099,
      "learning_rate": 3.343915343915344e-06,
      "loss": 0.0688,
      "step": 787
    },
    {
      "epoch": 2.5015873015873016,
      "grad_norm": 0.8325197696685791,
      "learning_rate": 3.322751322751323e-06,
      "loss": 0.0923,
      "step": 788
    },
    {
      "epoch": 2.5047619047619047,
      "grad_norm": 0.961479902267456,
      "learning_rate": 3.3015873015873016e-06,
      "loss": 0.0737,
      "step": 789
    },
    {
      "epoch": 2.507936507936508,
      "grad_norm": 1.0913158655166626,
      "learning_rate": 3.2804232804232807e-06,
      "loss": 0.1042,
      "step": 790
    },
    {
      "epoch": 2.511111111111111,
      "grad_norm": 1.1050466299057007,
      "learning_rate": 3.25925925925926e-06,
      "loss": 0.0727,
      "step": 791
    },
    {
      "epoch": 2.5142857142857142,
      "grad_norm": 1.5482778549194336,
      "learning_rate": 3.2380952380952385e-06,
      "loss": 0.1491,
      "step": 792
    },
    {
      "epoch": 2.5174603174603174,
      "grad_norm": 0.8196762800216675,
      "learning_rate": 3.2169312169312172e-06,
      "loss": 0.0748,
      "step": 793
    },
    {
      "epoch": 2.5206349206349206,
      "grad_norm": 0.8683902621269226,
      "learning_rate": 3.195767195767196e-06,
      "loss": 0.1027,
      "step": 794
    },
    {
      "epoch": 2.5238095238095237,
      "grad_norm": 1.0352352857589722,
      "learning_rate": 3.1746031746031746e-06,
      "loss": 0.0837,
      "step": 795
    },
    {
      "epoch": 2.526984126984127,
      "grad_norm": 0.9635798931121826,
      "learning_rate": 3.1534391534391538e-06,
      "loss": 0.1109,
      "step": 796
    },
    {
      "epoch": 2.53015873015873,
      "grad_norm": 0.46996819972991943,
      "learning_rate": 3.132275132275133e-06,
      "loss": 0.0756,
      "step": 797
    },
    {
      "epoch": 2.533333333333333,
      "grad_norm": 0.6797614693641663,
      "learning_rate": 3.1111111111111116e-06,
      "loss": 0.0819,
      "step": 798
    },
    {
      "epoch": 2.5365079365079364,
      "grad_norm": 0.3735337555408478,
      "learning_rate": 3.0899470899470903e-06,
      "loss": 0.0689,
      "step": 799
    },
    {
      "epoch": 2.5396825396825395,
      "grad_norm": 1.0652881860733032,
      "learning_rate": 3.068783068783069e-06,
      "loss": 0.1504,
      "step": 800
    },
    {
      "epoch": 2.5428571428571427,
      "grad_norm": 1.6396489143371582,
      "learning_rate": 3.047619047619048e-06,
      "loss": 0.1607,
      "step": 801
    },
    {
      "epoch": 2.546031746031746,
      "grad_norm": 1.7293888330459595,
      "learning_rate": 3.026455026455027e-06,
      "loss": 0.1336,
      "step": 802
    },
    {
      "epoch": 2.549206349206349,
      "grad_norm": 0.6983936429023743,
      "learning_rate": 3.0052910052910055e-06,
      "loss": 0.0461,
      "step": 803
    },
    {
      "epoch": 2.552380952380952,
      "grad_norm": 2.2259507179260254,
      "learning_rate": 2.984126984126984e-06,
      "loss": 0.0545,
      "step": 804
    },
    {
      "epoch": 2.5555555555555554,
      "grad_norm": 0.8321781754493713,
      "learning_rate": 2.962962962962963e-06,
      "loss": 0.1078,
      "step": 805
    },
    {
      "epoch": 2.5587301587301585,
      "grad_norm": 0.5068649649620056,
      "learning_rate": 2.9417989417989416e-06,
      "loss": 0.0777,
      "step": 806
    },
    {
      "epoch": 2.5619047619047617,
      "grad_norm": 1.6705137491226196,
      "learning_rate": 2.920634920634921e-06,
      "loss": 0.0915,
      "step": 807
    },
    {
      "epoch": 2.565079365079365,
      "grad_norm": 1.85332190990448,
      "learning_rate": 2.8994708994709e-06,
      "loss": 0.1291,
      "step": 808
    },
    {
      "epoch": 2.568253968253968,
      "grad_norm": 0.6907798647880554,
      "learning_rate": 2.8783068783068786e-06,
      "loss": 0.1108,
      "step": 809
    },
    {
      "epoch": 2.571428571428571,
      "grad_norm": 0.7357982993125916,
      "learning_rate": 2.8571428571428573e-06,
      "loss": 0.0968,
      "step": 810
    },
    {
      "epoch": 2.5746031746031743,
      "grad_norm": 0.5089612007141113,
      "learning_rate": 2.835978835978836e-06,
      "loss": 0.0572,
      "step": 811
    },
    {
      "epoch": 2.5777777777777775,
      "grad_norm": 1.1260141134262085,
      "learning_rate": 2.814814814814815e-06,
      "loss": 0.0876,
      "step": 812
    },
    {
      "epoch": 2.580952380952381,
      "grad_norm": 0.9506105780601501,
      "learning_rate": 2.7936507936507938e-06,
      "loss": 0.0931,
      "step": 813
    },
    {
      "epoch": 2.5841269841269843,
      "grad_norm": 0.7848398089408875,
      "learning_rate": 2.7724867724867725e-06,
      "loss": 0.0879,
      "step": 814
    },
    {
      "epoch": 2.5873015873015874,
      "grad_norm": 2.5763306617736816,
      "learning_rate": 2.7513227513227516e-06,
      "loss": 0.1192,
      "step": 815
    },
    {
      "epoch": 2.5904761904761906,
      "grad_norm": 0.5476844906806946,
      "learning_rate": 2.7301587301587303e-06,
      "loss": 0.0759,
      "step": 816
    },
    {
      "epoch": 2.5936507936507938,
      "grad_norm": 0.5420606136322021,
      "learning_rate": 2.7089947089947094e-06,
      "loss": 0.0422,
      "step": 817
    },
    {
      "epoch": 2.596825396825397,
      "grad_norm": 0.9260907769203186,
      "learning_rate": 2.687830687830688e-06,
      "loss": 0.104,
      "step": 818
    },
    {
      "epoch": 2.6,
      "grad_norm": 0.7739430069923401,
      "learning_rate": 2.666666666666667e-06,
      "loss": 0.1028,
      "step": 819
    },
    {
      "epoch": 2.6031746031746033,
      "grad_norm": 0.542879045009613,
      "learning_rate": 2.6455026455026455e-06,
      "loss": 0.0682,
      "step": 820
    },
    {
      "epoch": 2.6063492063492064,
      "grad_norm": 0.41392824053764343,
      "learning_rate": 2.6243386243386242e-06,
      "loss": 0.0502,
      "step": 821
    },
    {
      "epoch": 2.6095238095238096,
      "grad_norm": 0.5193356871604919,
      "learning_rate": 2.6031746031746038e-06,
      "loss": 0.1151,
      "step": 822
    },
    {
      "epoch": 2.6126984126984127,
      "grad_norm": 0.7275385856628418,
      "learning_rate": 2.5820105820105825e-06,
      "loss": 0.0841,
      "step": 823
    },
    {
      "epoch": 2.615873015873016,
      "grad_norm": 0.9508132934570312,
      "learning_rate": 2.560846560846561e-06,
      "loss": 0.0758,
      "step": 824
    },
    {
      "epoch": 2.619047619047619,
      "grad_norm": 0.6445814967155457,
      "learning_rate": 2.53968253968254e-06,
      "loss": 0.0979,
      "step": 825
    },
    {
      "epoch": 2.6222222222222222,
      "grad_norm": 0.8983713388442993,
      "learning_rate": 2.5185185185185186e-06,
      "loss": 0.0975,
      "step": 826
    },
    {
      "epoch": 2.6253968253968254,
      "grad_norm": 1.419110655784607,
      "learning_rate": 2.4973544973544973e-06,
      "loss": 0.0831,
      "step": 827
    },
    {
      "epoch": 2.6285714285714286,
      "grad_norm": 0.7938588857650757,
      "learning_rate": 2.4761904761904764e-06,
      "loss": 0.0959,
      "step": 828
    },
    {
      "epoch": 2.6317460317460317,
      "grad_norm": 0.7394862771034241,
      "learning_rate": 2.455026455026455e-06,
      "loss": 0.1038,
      "step": 829
    },
    {
      "epoch": 2.634920634920635,
      "grad_norm": 0.26812636852264404,
      "learning_rate": 2.433862433862434e-06,
      "loss": 0.0393,
      "step": 830
    },
    {
      "epoch": 2.638095238095238,
      "grad_norm": 0.5576487183570862,
      "learning_rate": 2.412698412698413e-06,
      "loss": 0.0768,
      "step": 831
    },
    {
      "epoch": 2.641269841269841,
      "grad_norm": 0.42976438999176025,
      "learning_rate": 2.3915343915343916e-06,
      "loss": 0.0573,
      "step": 832
    },
    {
      "epoch": 2.6444444444444444,
      "grad_norm": 0.3815513551235199,
      "learning_rate": 2.3703703703703707e-06,
      "loss": 0.0486,
      "step": 833
    },
    {
      "epoch": 2.6476190476190475,
      "grad_norm": 0.8436024188995361,
      "learning_rate": 2.3492063492063494e-06,
      "loss": 0.1032,
      "step": 834
    },
    {
      "epoch": 2.6507936507936507,
      "grad_norm": 0.4033740162849426,
      "learning_rate": 2.328042328042328e-06,
      "loss": 0.0628,
      "step": 835
    },
    {
      "epoch": 2.653968253968254,
      "grad_norm": 0.8692321181297302,
      "learning_rate": 2.3068783068783073e-06,
      "loss": 0.0786,
      "step": 836
    },
    {
      "epoch": 2.657142857142857,
      "grad_norm": 0.546276330947876,
      "learning_rate": 2.285714285714286e-06,
      "loss": 0.0617,
      "step": 837
    },
    {
      "epoch": 2.66031746031746,
      "grad_norm": 0.8689720630645752,
      "learning_rate": 2.2645502645502647e-06,
      "loss": 0.1203,
      "step": 838
    },
    {
      "epoch": 2.6634920634920634,
      "grad_norm": 0.6834908723831177,
      "learning_rate": 2.2433862433862434e-06,
      "loss": 0.0566,
      "step": 839
    },
    {
      "epoch": 2.6666666666666665,
      "grad_norm": 1.067679762840271,
      "learning_rate": 2.222222222222222e-06,
      "loss": 0.127,
      "step": 840
    },
    {
      "epoch": 2.6698412698412697,
      "grad_norm": 0.7603820562362671,
      "learning_rate": 2.201058201058201e-06,
      "loss": 0.0862,
      "step": 841
    },
    {
      "epoch": 2.6730158730158733,
      "grad_norm": 1.0284106731414795,
      "learning_rate": 2.17989417989418e-06,
      "loss": 0.1265,
      "step": 842
    },
    {
      "epoch": 2.6761904761904765,
      "grad_norm": 0.9925879836082458,
      "learning_rate": 2.158730158730159e-06,
      "loss": 0.113,
      "step": 843
    },
    {
      "epoch": 2.6793650793650796,
      "grad_norm": 0.9208048582077026,
      "learning_rate": 2.1375661375661377e-06,
      "loss": 0.0728,
      "step": 844
    },
    {
      "epoch": 2.682539682539683,
      "grad_norm": 0.5659469962120056,
      "learning_rate": 2.1164021164021164e-06,
      "loss": 0.052,
      "step": 845
    },
    {
      "epoch": 2.685714285714286,
      "grad_norm": 0.6194931268692017,
      "learning_rate": 2.0952380952380955e-06,
      "loss": 0.1076,
      "step": 846
    },
    {
      "epoch": 2.688888888888889,
      "grad_norm": 0.9001226425170898,
      "learning_rate": 2.0740740740740742e-06,
      "loss": 0.0694,
      "step": 847
    },
    {
      "epoch": 2.6920634920634923,
      "grad_norm": 0.5578948259353638,
      "learning_rate": 2.0529100529100534e-06,
      "loss": 0.1106,
      "step": 848
    },
    {
      "epoch": 2.6952380952380954,
      "grad_norm": 1.2801584005355835,
      "learning_rate": 2.031746031746032e-06,
      "loss": 0.126,
      "step": 849
    },
    {
      "epoch": 2.6984126984126986,
      "grad_norm": 0.7589452266693115,
      "learning_rate": 2.0105820105820108e-06,
      "loss": 0.1036,
      "step": 850
    },
    {
      "epoch": 2.7015873015873018,
      "grad_norm": 0.7427812814712524,
      "learning_rate": 1.9894179894179895e-06,
      "loss": 0.0955,
      "step": 851
    },
    {
      "epoch": 2.704761904761905,
      "grad_norm": 0.40698403120040894,
      "learning_rate": 1.968253968253968e-06,
      "loss": 0.0458,
      "step": 852
    },
    {
      "epoch": 2.707936507936508,
      "grad_norm": 2.27262020111084,
      "learning_rate": 1.9470899470899473e-06,
      "loss": 0.2436,
      "step": 853
    },
    {
      "epoch": 2.7111111111111112,
      "grad_norm": 0.7904677987098694,
      "learning_rate": 1.925925925925926e-06,
      "loss": 0.07,
      "step": 854
    },
    {
      "epoch": 2.7142857142857144,
      "grad_norm": 0.9789160490036011,
      "learning_rate": 1.904761904761905e-06,
      "loss": 0.0926,
      "step": 855
    },
    {
      "epoch": 2.7174603174603176,
      "grad_norm": 0.5903559923171997,
      "learning_rate": 1.8835978835978838e-06,
      "loss": 0.0857,
      "step": 856
    },
    {
      "epoch": 2.7206349206349207,
      "grad_norm": 0.6822364926338196,
      "learning_rate": 1.8624338624338625e-06,
      "loss": 0.0799,
      "step": 857
    },
    {
      "epoch": 2.723809523809524,
      "grad_norm": 0.7783228158950806,
      "learning_rate": 1.8412698412698416e-06,
      "loss": 0.136,
      "step": 858
    },
    {
      "epoch": 2.726984126984127,
      "grad_norm": 1.053891897201538,
      "learning_rate": 1.8201058201058203e-06,
      "loss": 0.1244,
      "step": 859
    },
    {
      "epoch": 2.7301587301587302,
      "grad_norm": 0.8601414561271667,
      "learning_rate": 1.798941798941799e-06,
      "loss": 0.1102,
      "step": 860
    },
    {
      "epoch": 2.7333333333333334,
      "grad_norm": 1.1214914321899414,
      "learning_rate": 1.777777777777778e-06,
      "loss": 0.104,
      "step": 861
    },
    {
      "epoch": 2.7365079365079366,
      "grad_norm": 0.6030844449996948,
      "learning_rate": 1.7566137566137567e-06,
      "loss": 0.0685,
      "step": 862
    },
    {
      "epoch": 2.7396825396825397,
      "grad_norm": 1.2531230449676514,
      "learning_rate": 1.7354497354497358e-06,
      "loss": 0.1287,
      "step": 863
    },
    {
      "epoch": 2.742857142857143,
      "grad_norm": 0.6852943897247314,
      "learning_rate": 1.7142857142857145e-06,
      "loss": 0.0978,
      "step": 864
    },
    {
      "epoch": 2.746031746031746,
      "grad_norm": 0.6224417090415955,
      "learning_rate": 1.6931216931216932e-06,
      "loss": 0.1132,
      "step": 865
    },
    {
      "epoch": 2.749206349206349,
      "grad_norm": 0.7010740041732788,
      "learning_rate": 1.671957671957672e-06,
      "loss": 0.0944,
      "step": 866
    },
    {
      "epoch": 2.7523809523809524,
      "grad_norm": 1.4900541305541992,
      "learning_rate": 1.6507936507936508e-06,
      "loss": 0.0891,
      "step": 867
    },
    {
      "epoch": 2.7555555555555555,
      "grad_norm": 0.6522367000579834,
      "learning_rate": 1.62962962962963e-06,
      "loss": 0.061,
      "step": 868
    },
    {
      "epoch": 2.7587301587301587,
      "grad_norm": 0.9777023196220398,
      "learning_rate": 1.6084656084656086e-06,
      "loss": 0.0685,
      "step": 869
    },
    {
      "epoch": 2.761904761904762,
      "grad_norm": 1.239327311515808,
      "learning_rate": 1.5873015873015873e-06,
      "loss": 0.1412,
      "step": 870
    },
    {
      "epoch": 2.765079365079365,
      "grad_norm": 0.5872946977615356,
      "learning_rate": 1.5661375661375664e-06,
      "loss": 0.095,
      "step": 871
    },
    {
      "epoch": 2.768253968253968,
      "grad_norm": 0.8251181840896606,
      "learning_rate": 1.5449735449735451e-06,
      "loss": 0.0892,
      "step": 872
    },
    {
      "epoch": 2.7714285714285714,
      "grad_norm": 0.8245250582695007,
      "learning_rate": 1.523809523809524e-06,
      "loss": 0.0588,
      "step": 873
    },
    {
      "epoch": 2.7746031746031745,
      "grad_norm": 0.4351988434791565,
      "learning_rate": 1.5026455026455028e-06,
      "loss": 0.0456,
      "step": 874
    },
    {
      "epoch": 2.7777777777777777,
      "grad_norm": 1.1531660556793213,
      "learning_rate": 1.4814814814814815e-06,
      "loss": 0.1317,
      "step": 875
    },
    {
      "epoch": 2.780952380952381,
      "grad_norm": 1.0304726362228394,
      "learning_rate": 1.4603174603174606e-06,
      "loss": 0.1003,
      "step": 876
    },
    {
      "epoch": 2.784126984126984,
      "grad_norm": 0.49166861176490784,
      "learning_rate": 1.4391534391534393e-06,
      "loss": 0.0902,
      "step": 877
    },
    {
      "epoch": 2.787301587301587,
      "grad_norm": 1.9814692735671997,
      "learning_rate": 1.417989417989418e-06,
      "loss": 0.192,
      "step": 878
    },
    {
      "epoch": 2.7904761904761903,
      "grad_norm": 1.3572824001312256,
      "learning_rate": 1.3968253968253969e-06,
      "loss": 0.0904,
      "step": 879
    },
    {
      "epoch": 2.7936507936507935,
      "grad_norm": 0.6977200508117676,
      "learning_rate": 1.3756613756613758e-06,
      "loss": 0.0722,
      "step": 880
    },
    {
      "epoch": 2.7968253968253967,
      "grad_norm": 0.8476284742355347,
      "learning_rate": 1.3544973544973547e-06,
      "loss": 0.129,
      "step": 881
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.2630301713943481,
      "learning_rate": 1.3333333333333334e-06,
      "loss": 0.099,
      "step": 882
    },
    {
      "epoch": 2.803174603174603,
      "grad_norm": 1.754107117652893,
      "learning_rate": 1.3121693121693121e-06,
      "loss": 0.0951,
      "step": 883
    },
    {
      "epoch": 2.806349206349206,
      "grad_norm": 0.4020839333534241,
      "learning_rate": 1.2910052910052912e-06,
      "loss": 0.1031,
      "step": 884
    },
    {
      "epoch": 2.8095238095238093,
      "grad_norm": 0.9032473564147949,
      "learning_rate": 1.26984126984127e-06,
      "loss": 0.1029,
      "step": 885
    },
    {
      "epoch": 2.8126984126984125,
      "grad_norm": 1.1078736782073975,
      "learning_rate": 1.2486772486772486e-06,
      "loss": 0.1219,
      "step": 886
    },
    {
      "epoch": 2.8158730158730156,
      "grad_norm": 0.8839858174324036,
      "learning_rate": 1.2275132275132276e-06,
      "loss": 0.0956,
      "step": 887
    },
    {
      "epoch": 2.819047619047619,
      "grad_norm": 0.621709406375885,
      "learning_rate": 1.2063492063492065e-06,
      "loss": 0.052,
      "step": 888
    },
    {
      "epoch": 2.822222222222222,
      "grad_norm": 2.1563985347747803,
      "learning_rate": 1.1851851851851854e-06,
      "loss": 0.139,
      "step": 889
    },
    {
      "epoch": 2.825396825396825,
      "grad_norm": 1.4044126272201538,
      "learning_rate": 1.164021164021164e-06,
      "loss": 0.1162,
      "step": 890
    },
    {
      "epoch": 2.8285714285714287,
      "grad_norm": 1.3765867948532104,
      "learning_rate": 1.142857142857143e-06,
      "loss": 0.139,
      "step": 891
    },
    {
      "epoch": 2.831746031746032,
      "grad_norm": 0.7476480603218079,
      "learning_rate": 1.1216931216931217e-06,
      "loss": 0.0822,
      "step": 892
    },
    {
      "epoch": 2.834920634920635,
      "grad_norm": 0.9987853169441223,
      "learning_rate": 1.1005291005291006e-06,
      "loss": 0.1842,
      "step": 893
    },
    {
      "epoch": 2.8380952380952382,
      "grad_norm": 0.8177071213722229,
      "learning_rate": 1.0793650793650795e-06,
      "loss": 0.084,
      "step": 894
    },
    {
      "epoch": 2.8412698412698414,
      "grad_norm": 1.5093634128570557,
      "learning_rate": 1.0582010582010582e-06,
      "loss": 0.1296,
      "step": 895
    },
    {
      "epoch": 2.8444444444444446,
      "grad_norm": 0.8142589926719666,
      "learning_rate": 1.0370370370370371e-06,
      "loss": 0.0665,
      "step": 896
    },
    {
      "epoch": 2.8476190476190477,
      "grad_norm": 0.5820671319961548,
      "learning_rate": 1.015873015873016e-06,
      "loss": 0.0535,
      "step": 897
    },
    {
      "epoch": 2.850793650793651,
      "grad_norm": 0.6321248412132263,
      "learning_rate": 9.947089947089947e-07,
      "loss": 0.0891,
      "step": 898
    },
    {
      "epoch": 2.853968253968254,
      "grad_norm": 0.6557403802871704,
      "learning_rate": 9.735449735449736e-07,
      "loss": 0.083,
      "step": 899
    },
    {
      "epoch": 2.857142857142857,
      "grad_norm": 0.5062572956085205,
      "learning_rate": 9.523809523809525e-07,
      "loss": 0.0515,
      "step": 900
    },
    {
      "epoch": 2.8603174603174604,
      "grad_norm": 1.0809507369995117,
      "learning_rate": 9.312169312169313e-07,
      "loss": 0.0975,
      "step": 901
    },
    {
      "epoch": 2.8634920634920635,
      "grad_norm": 0.7125364542007446,
      "learning_rate": 9.100529100529102e-07,
      "loss": 0.0926,
      "step": 902
    },
    {
      "epoch": 2.8666666666666667,
      "grad_norm": 0.6580918431282043,
      "learning_rate": 8.88888888888889e-07,
      "loss": 0.1289,
      "step": 903
    },
    {
      "epoch": 2.86984126984127,
      "grad_norm": 0.730147659778595,
      "learning_rate": 8.677248677248679e-07,
      "loss": 0.0903,
      "step": 904
    },
    {
      "epoch": 2.873015873015873,
      "grad_norm": 1.3870775699615479,
      "learning_rate": 8.465608465608466e-07,
      "loss": 0.1131,
      "step": 905
    },
    {
      "epoch": 2.876190476190476,
      "grad_norm": 0.6383364200592041,
      "learning_rate": 8.253968253968254e-07,
      "loss": 0.0681,
      "step": 906
    },
    {
      "epoch": 2.8793650793650793,
      "grad_norm": 0.5035264492034912,
      "learning_rate": 8.042328042328043e-07,
      "loss": 0.0557,
      "step": 907
    },
    {
      "epoch": 2.8825396825396825,
      "grad_norm": 1.1113923788070679,
      "learning_rate": 7.830687830687832e-07,
      "loss": 0.1193,
      "step": 908
    },
    {
      "epoch": 2.8857142857142857,
      "grad_norm": 0.7754139304161072,
      "learning_rate": 7.61904761904762e-07,
      "loss": 0.0881,
      "step": 909
    },
    {
      "epoch": 2.888888888888889,
      "grad_norm": 0.786375880241394,
      "learning_rate": 7.407407407407407e-07,
      "loss": 0.1275,
      "step": 910
    },
    {
      "epoch": 2.892063492063492,
      "grad_norm": 0.7526279091835022,
      "learning_rate": 7.195767195767196e-07,
      "loss": 0.1165,
      "step": 911
    },
    {
      "epoch": 2.895238095238095,
      "grad_norm": 0.851228654384613,
      "learning_rate": 6.984126984126984e-07,
      "loss": 0.0849,
      "step": 912
    },
    {
      "epoch": 2.8984126984126983,
      "grad_norm": 2.135032892227173,
      "learning_rate": 6.772486772486774e-07,
      "loss": 0.1099,
      "step": 913
    },
    {
      "epoch": 2.9015873015873015,
      "grad_norm": 0.9407389760017395,
      "learning_rate": 6.560846560846561e-07,
      "loss": 0.1306,
      "step": 914
    },
    {
      "epoch": 2.9047619047619047,
      "grad_norm": 0.5231833457946777,
      "learning_rate": 6.34920634920635e-07,
      "loss": 0.0828,
      "step": 915
    },
    {
      "epoch": 2.907936507936508,
      "grad_norm": 1.1373553276062012,
      "learning_rate": 6.137566137566138e-07,
      "loss": 0.1373,
      "step": 916
    },
    {
      "epoch": 2.911111111111111,
      "grad_norm": 0.9961684346199036,
      "learning_rate": 5.925925925925927e-07,
      "loss": 0.0728,
      "step": 917
    },
    {
      "epoch": 2.914285714285714,
      "grad_norm": 0.9935572743415833,
      "learning_rate": 5.714285714285715e-07,
      "loss": 0.1011,
      "step": 918
    },
    {
      "epoch": 2.9174603174603173,
      "grad_norm": 1.7094495296478271,
      "learning_rate": 5.502645502645503e-07,
      "loss": 0.1714,
      "step": 919
    },
    {
      "epoch": 2.9206349206349205,
      "grad_norm": 1.0301735401153564,
      "learning_rate": 5.291005291005291e-07,
      "loss": 0.068,
      "step": 920
    },
    {
      "epoch": 2.923809523809524,
      "grad_norm": 1.4113563299179077,
      "learning_rate": 5.07936507936508e-07,
      "loss": 0.2197,
      "step": 921
    },
    {
      "epoch": 2.9269841269841272,
      "grad_norm": 0.47605448961257935,
      "learning_rate": 4.867724867724868e-07,
      "loss": 0.0476,
      "step": 922
    },
    {
      "epoch": 2.9301587301587304,
      "grad_norm": 0.9930559992790222,
      "learning_rate": 4.6560846560846563e-07,
      "loss": 0.0676,
      "step": 923
    },
    {
      "epoch": 2.9333333333333336,
      "grad_norm": 0.7487044930458069,
      "learning_rate": 4.444444444444445e-07,
      "loss": 0.0697,
      "step": 924
    },
    {
      "epoch": 2.9365079365079367,
      "grad_norm": 0.47881636023521423,
      "learning_rate": 4.232804232804233e-07,
      "loss": 0.0658,
      "step": 925
    },
    {
      "epoch": 2.93968253968254,
      "grad_norm": 0.8862658739089966,
      "learning_rate": 4.0211640211640215e-07,
      "loss": 0.0696,
      "step": 926
    },
    {
      "epoch": 2.942857142857143,
      "grad_norm": 0.748858630657196,
      "learning_rate": 3.80952380952381e-07,
      "loss": 0.0669,
      "step": 927
    },
    {
      "epoch": 2.9460317460317462,
      "grad_norm": 0.49354302883148193,
      "learning_rate": 3.597883597883598e-07,
      "loss": 0.0805,
      "step": 928
    },
    {
      "epoch": 2.9492063492063494,
      "grad_norm": 0.7746992111206055,
      "learning_rate": 3.386243386243387e-07,
      "loss": 0.0605,
      "step": 929
    },
    {
      "epoch": 2.9523809523809526,
      "grad_norm": 0.8144791126251221,
      "learning_rate": 3.174603174603175e-07,
      "loss": 0.1749,
      "step": 930
    },
    {
      "epoch": 2.9555555555555557,
      "grad_norm": 0.606460690498352,
      "learning_rate": 2.9629629629629634e-07,
      "loss": 0.0823,
      "step": 931
    },
    {
      "epoch": 2.958730158730159,
      "grad_norm": 1.1171824932098389,
      "learning_rate": 2.7513227513227515e-07,
      "loss": 0.1041,
      "step": 932
    },
    {
      "epoch": 2.961904761904762,
      "grad_norm": 0.6133437156677246,
      "learning_rate": 2.53968253968254e-07,
      "loss": 0.1352,
      "step": 933
    },
    {
      "epoch": 2.965079365079365,
      "grad_norm": 0.9138501286506653,
      "learning_rate": 2.3280423280423281e-07,
      "loss": 0.1343,
      "step": 934
    },
    {
      "epoch": 2.9682539682539684,
      "grad_norm": 0.47824618220329285,
      "learning_rate": 2.1164021164021165e-07,
      "loss": 0.1017,
      "step": 935
    },
    {
      "epoch": 2.9714285714285715,
      "grad_norm": 0.3149285316467285,
      "learning_rate": 1.904761904761905e-07,
      "loss": 0.0406,
      "step": 936
    },
    {
      "epoch": 2.9746031746031747,
      "grad_norm": 0.653850257396698,
      "learning_rate": 1.6931216931216934e-07,
      "loss": 0.0485,
      "step": 937
    },
    {
      "epoch": 2.977777777777778,
      "grad_norm": 1.3658442497253418,
      "learning_rate": 1.4814814814814817e-07,
      "loss": 0.2061,
      "step": 938
    },
    {
      "epoch": 2.980952380952381,
      "grad_norm": 1.2120649814605713,
      "learning_rate": 1.26984126984127e-07,
      "loss": 0.1029,
      "step": 939
    },
    {
      "epoch": 2.984126984126984,
      "grad_norm": 0.46431148052215576,
      "learning_rate": 1.0582010582010582e-07,
      "loss": 0.0924,
      "step": 940
    },
    {
      "epoch": 2.9873015873015873,
      "grad_norm": 0.5093331336975098,
      "learning_rate": 8.465608465608467e-08,
      "loss": 0.0679,
      "step": 941
    },
    {
      "epoch": 2.9904761904761905,
      "grad_norm": 0.9145987033843994,
      "learning_rate": 6.34920634920635e-08,
      "loss": 0.1186,
      "step": 942
    },
    {
      "epoch": 2.9936507936507937,
      "grad_norm": 0.9311897158622742,
      "learning_rate": 4.2328042328042335e-08,
      "loss": 0.1147,
      "step": 943
    },
    {
      "epoch": 2.996825396825397,
      "grad_norm": 0.8751983642578125,
      "learning_rate": 2.1164021164021167e-08,
      "loss": 0.1416,
      "step": 944
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.8466974496841431,
      "learning_rate": 0.0,
      "loss": 0.1305,
      "step": 945
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.7830357142857143,
      "eval_f1": 0.9151147682979645,
      "eval_loss": 0.11422336101531982,
      "eval_runtime": 1.7246,
      "eval_samples_per_second": 649.415,
      "eval_steps_per_second": 20.294,
      "step": 945
    }
  ],
  "logging_steps": 1,
  "max_steps": 945,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1001614222909440.0,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": null
}