{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9998533079067038,
  "eval_steps": 100,
  "global_step": 3408,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 7.766990291262136e-07,
      "loss": 3.4245,
      "step": 1
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.5533980582524272e-06,
      "loss": 3.1989,
      "step": 2
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.330097087378641e-06,
      "loss": 3.2308,
      "step": 3
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.1067961165048544e-06,
      "loss": 2.9591,
      "step": 4
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.883495145631068e-06,
      "loss": 2.6008,
      "step": 5
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.660194174757282e-06,
      "loss": 2.5077,
      "step": 6
    },
    {
      "epoch": 0.0,
      "learning_rate": 5.436893203883496e-06,
      "loss": 2.3392,
      "step": 7
    },
    {
      "epoch": 0.0,
      "learning_rate": 6.213592233009709e-06,
      "loss": 2.2419,
      "step": 8
    },
    {
      "epoch": 0.0,
      "learning_rate": 6.990291262135923e-06,
      "loss": 2.0988,
      "step": 9
    },
    {
      "epoch": 0.0,
      "learning_rate": 7.766990291262136e-06,
      "loss": 2.0581,
      "step": 10
    },
    {
      "epoch": 0.0,
      "learning_rate": 8.54368932038835e-06,
      "loss": 1.9165,
      "step": 11
    },
    {
      "epoch": 0.0,
      "learning_rate": 9.320388349514565e-06,
      "loss": 2.0559,
      "step": 12
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.0097087378640778e-05,
      "loss": 1.8834,
      "step": 13
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.0873786407766991e-05,
      "loss": 1.8474,
      "step": 14
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.1650485436893204e-05,
      "loss": 1.8903,
      "step": 15
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.2427184466019418e-05,
      "loss": 1.8739,
      "step": 16
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.3203883495145633e-05,
      "loss": 1.8753,
      "step": 17
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.3980582524271846e-05,
      "loss": 1.8335,
      "step": 18
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.475728155339806e-05,
      "loss": 1.848,
      "step": 19
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.5533980582524273e-05,
      "loss": 1.7487,
      "step": 20
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.6310679611650486e-05,
      "loss": 1.8451,
      "step": 21
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.70873786407767e-05,
      "loss": 1.742,
      "step": 22
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.7864077669902916e-05,
      "loss": 1.8671,
      "step": 23
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.864077669902913e-05,
      "loss": 1.7642,
      "step": 24
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.9417475728155343e-05,
      "loss": 1.7682,
      "step": 25
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.0194174757281556e-05,
      "loss": 1.7926,
      "step": 26
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.097087378640777e-05,
      "loss": 1.7951,
      "step": 27
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.1747572815533982e-05,
      "loss": 1.7663,
      "step": 28
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.2524271844660196e-05,
      "loss": 1.7114,
      "step": 29
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.330097087378641e-05,
      "loss": 1.7552,
      "step": 30
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.4077669902912622e-05,
      "loss": 1.7745,
      "step": 31
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.4854368932038836e-05,
      "loss": 1.7002,
      "step": 32
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.5631067961165052e-05,
      "loss": 1.6888,
      "step": 33
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.6407766990291266e-05,
      "loss": 1.7667,
      "step": 34
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.718446601941748e-05,
      "loss": 1.6512,
      "step": 35
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.7961165048543692e-05,
      "loss": 1.6581,
      "step": 36
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.8737864077669905e-05,
      "loss": 1.7076,
      "step": 37
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.951456310679612e-05,
      "loss": 1.7243,
      "step": 38
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.0291262135922332e-05,
      "loss": 1.6582,
      "step": 39
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.1067961165048545e-05,
      "loss": 1.692,
      "step": 40
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.184466019417476e-05,
      "loss": 1.6946,
      "step": 41
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.262135922330097e-05,
      "loss": 1.7463,
      "step": 42
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.339805825242719e-05,
      "loss": 1.6669,
      "step": 43
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.41747572815534e-05,
      "loss": 1.6517,
      "step": 44
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.4951456310679615e-05,
      "loss": 1.6559,
      "step": 45
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.572815533980583e-05,
      "loss": 1.692,
      "step": 46
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.650485436893204e-05,
      "loss": 1.6341,
      "step": 47
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.728155339805826e-05,
      "loss": 1.6815,
      "step": 48
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.805825242718447e-05,
      "loss": 1.7242,
      "step": 49
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.8834951456310685e-05,
      "loss": 1.6661,
      "step": 50
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.9611650485436895e-05,
      "loss": 1.7046,
      "step": 51
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.038834951456311e-05,
      "loss": 1.683,
      "step": 52
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.116504854368932e-05,
      "loss": 1.682,
      "step": 53
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.194174757281554e-05,
      "loss": 1.6719,
      "step": 54
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.271844660194175e-05,
      "loss": 1.6538,
      "step": 55
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.3495145631067965e-05,
      "loss": 1.6646,
      "step": 56
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.4271844660194175e-05,
      "loss": 1.6805,
      "step": 57
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.504854368932039e-05,
      "loss": 1.6531,
      "step": 58
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.58252427184466e-05,
      "loss": 1.6359,
      "step": 59
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.660194174757282e-05,
      "loss": 1.6408,
      "step": 60
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.737864077669903e-05,
      "loss": 1.6991,
      "step": 61
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8155339805825245e-05,
      "loss": 1.5819,
      "step": 62
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8932038834951454e-05,
      "loss": 1.6714,
      "step": 63
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.970873786407767e-05,
      "loss": 1.6898,
      "step": 64
    },
    {
      "epoch": 0.02,
      "learning_rate": 5.0485436893203895e-05,
      "loss": 1.6105,
      "step": 65
    },
    {
      "epoch": 0.02,
      "learning_rate": 5.1262135922330105e-05,
      "loss": 1.6153,
      "step": 66
    },
    {
      "epoch": 0.02,
      "learning_rate": 5.203883495145632e-05,
      "loss": 1.708,
      "step": 67
    },
    {
      "epoch": 0.02,
      "learning_rate": 5.281553398058253e-05,
      "loss": 1.6854,
      "step": 68
    },
    {
      "epoch": 0.02,
      "learning_rate": 5.359223300970875e-05,
      "loss": 1.6446,
      "step": 69
    },
    {
      "epoch": 0.02,
      "learning_rate": 5.436893203883496e-05,
      "loss": 1.5945,
      "step": 70
    },
    {
      "epoch": 0.02,
      "learning_rate": 5.5145631067961174e-05,
      "loss": 1.6984,
      "step": 71
    },
    {
      "epoch": 0.02,
      "learning_rate": 5.5922330097087384e-05,
      "loss": 1.6914,
      "step": 72
    },
    {
      "epoch": 0.02,
      "learning_rate": 5.66990291262136e-05,
      "loss": 1.693,
      "step": 73
    },
    {
      "epoch": 0.02,
      "learning_rate": 5.747572815533981e-05,
      "loss": 1.642,
      "step": 74
    },
    {
      "epoch": 0.02,
      "learning_rate": 5.825242718446603e-05,
      "loss": 1.61,
      "step": 75
    },
    {
      "epoch": 0.02,
      "learning_rate": 5.902912621359224e-05,
      "loss": 1.7104,
      "step": 76
    },
    {
      "epoch": 0.02,
      "learning_rate": 5.9805825242718454e-05,
      "loss": 1.6252,
      "step": 77
    },
    {
      "epoch": 0.02,
      "learning_rate": 6.0582524271844664e-05,
      "loss": 1.6818,
      "step": 78
    },
    {
      "epoch": 0.02,
      "learning_rate": 6.135922330097087e-05,
      "loss": 1.6772,
      "step": 79
    },
    {
      "epoch": 0.02,
      "learning_rate": 6.213592233009709e-05,
      "loss": 1.699,
      "step": 80
    },
    {
      "epoch": 0.02,
      "learning_rate": 6.291262135922331e-05,
      "loss": 1.6542,
      "step": 81
    },
    {
      "epoch": 0.02,
      "learning_rate": 6.368932038834952e-05,
      "loss": 1.6459,
      "step": 82
    },
    {
      "epoch": 0.02,
      "learning_rate": 6.446601941747573e-05,
      "loss": 1.6477,
      "step": 83
    },
    {
      "epoch": 0.02,
      "learning_rate": 6.524271844660194e-05,
      "loss": 1.6493,
      "step": 84
    },
    {
      "epoch": 0.02,
      "learning_rate": 6.601941747572816e-05,
      "loss": 1.6539,
      "step": 85
    },
    {
      "epoch": 0.03,
      "learning_rate": 6.679611650485438e-05,
      "loss": 1.6467,
      "step": 86
    },
    {
      "epoch": 0.03,
      "learning_rate": 6.757281553398058e-05,
      "loss": 1.7471,
      "step": 87
    },
    {
      "epoch": 0.03,
      "learning_rate": 6.83495145631068e-05,
      "loss": 1.638,
      "step": 88
    },
    {
      "epoch": 0.03,
      "learning_rate": 6.912621359223301e-05,
      "loss": 1.6656,
      "step": 89
    },
    {
      "epoch": 0.03,
      "learning_rate": 6.990291262135923e-05,
      "loss": 1.6356,
      "step": 90
    },
    {
      "epoch": 0.03,
      "learning_rate": 7.067961165048545e-05,
      "loss": 1.7034,
      "step": 91
    },
    {
      "epoch": 0.03,
      "learning_rate": 7.145631067961166e-05,
      "loss": 1.7101,
      "step": 92
    },
    {
      "epoch": 0.03,
      "learning_rate": 7.223300970873787e-05,
      "loss": 1.6649,
      "step": 93
    },
    {
      "epoch": 0.03,
      "learning_rate": 7.300970873786408e-05,
      "loss": 1.7123,
      "step": 94
    },
    {
      "epoch": 0.03,
      "learning_rate": 7.37864077669903e-05,
      "loss": 1.6884,
      "step": 95
    },
    {
      "epoch": 0.03,
      "learning_rate": 7.456310679611652e-05,
      "loss": 1.7021,
      "step": 96
    },
    {
      "epoch": 0.03,
      "learning_rate": 7.533980582524272e-05,
      "loss": 1.6138,
      "step": 97
    },
    {
      "epoch": 0.03,
      "learning_rate": 7.611650485436894e-05,
      "loss": 1.7511,
      "step": 98
    },
    {
      "epoch": 0.03,
      "learning_rate": 7.689320388349515e-05,
      "loss": 1.6645,
      "step": 99
    },
    {
      "epoch": 0.03,
      "learning_rate": 7.766990291262137e-05,
      "loss": 1.6774,
      "step": 100
    },
    {
      "epoch": 0.03,
      "eval_loss": 0.18088601529598236,
      "eval_runtime": 25.5251,
      "eval_samples_per_second": 78.354,
      "eval_steps_per_second": 0.627,
      "step": 100
    },
    {
      "epoch": 0.03,
      "learning_rate": 7.844660194174757e-05,
      "loss": 1.642,
      "step": 101
    },
    {
      "epoch": 0.03,
      "learning_rate": 7.922330097087379e-05,
      "loss": 1.7632,
      "step": 102
    },
    {
      "epoch": 0.03,
      "learning_rate": 8e-05,
      "loss": 1.7431,
      "step": 103
    },
    {
      "epoch": 0.03,
      "learning_rate": 7.999998192880882e-05,
      "loss": 1.6428,
      "step": 104
    },
    {
      "epoch": 0.03,
      "learning_rate": 7.999992771525157e-05,
      "loss": 1.675,
      "step": 105
    },
    {
      "epoch": 0.03,
      "learning_rate": 7.999983735937726e-05,
      "loss": 1.685,
      "step": 106
    },
    {
      "epoch": 0.03,
      "learning_rate": 7.999971086126752e-05,
      "loss": 1.6642,
      "step": 107
    },
    {
      "epoch": 0.03,
      "learning_rate": 7.999954822103665e-05,
      "loss": 1.6388,
      "step": 108
    },
    {
      "epoch": 0.03,
      "learning_rate": 7.999934943883161e-05,
      "loss": 1.711,
      "step": 109
    },
    {
      "epoch": 0.03,
      "learning_rate": 7.9999114514832e-05,
      "loss": 1.683,
      "step": 110
    },
    {
      "epoch": 0.03,
      "learning_rate": 7.999884344925012e-05,
      "loss": 1.6607,
      "step": 111
    },
    {
      "epoch": 0.03,
      "learning_rate": 7.999853624233086e-05,
      "loss": 1.7063,
      "step": 112
    },
    {
      "epoch": 0.03,
      "learning_rate": 7.999819289435179e-05,
      "loss": 1.6994,
      "step": 113
    },
    {
      "epoch": 0.03,
      "learning_rate": 7.999781340562318e-05,
      "loss": 1.6633,
      "step": 114
    },
    {
      "epoch": 0.03,
      "learning_rate": 7.999739777648792e-05,
      "loss": 1.656,
      "step": 115
    },
    {
      "epoch": 0.03,
      "learning_rate": 7.999694600732152e-05,
      "loss": 1.6876,
      "step": 116
    },
    {
      "epoch": 0.03,
      "learning_rate": 7.999645809853222e-05,
      "loss": 1.7541,
      "step": 117
    },
    {
      "epoch": 0.03,
      "learning_rate": 7.999593405056084e-05,
      "loss": 1.65,
      "step": 118
    },
    {
      "epoch": 0.03,
      "learning_rate": 7.999537386388092e-05,
      "loss": 1.6745,
      "step": 119
    },
    {
      "epoch": 0.04,
      "learning_rate": 7.999477753899861e-05,
      "loss": 1.7118,
      "step": 120
    },
    {
      "epoch": 0.04,
      "learning_rate": 7.999414507645272e-05,
      "loss": 1.732,
      "step": 121
    },
    {
      "epoch": 0.04,
      "learning_rate": 7.99934764768147e-05,
      "loss": 1.6339,
      "step": 122
    },
    {
      "epoch": 0.04,
      "learning_rate": 7.999277174068872e-05,
      "loss": 1.62,
      "step": 123
    },
    {
      "epoch": 0.04,
      "learning_rate": 7.999203086871149e-05,
      "loss": 1.6943,
      "step": 124
    },
    {
      "epoch": 0.04,
      "learning_rate": 7.999125386155248e-05,
      "loss": 1.6972,
      "step": 125
    },
    {
      "epoch": 0.04,
      "learning_rate": 7.999044071991375e-05,
      "loss": 1.6469,
      "step": 126
    },
    {
      "epoch": 0.04,
      "learning_rate": 7.998959144453001e-05,
      "loss": 1.6318,
      "step": 127
    },
    {
      "epoch": 0.04,
      "learning_rate": 7.998870603616864e-05,
      "loss": 1.7085,
      "step": 128
    },
    {
      "epoch": 0.04,
      "learning_rate": 7.998778449562965e-05,
      "loss": 1.6849,
      "step": 129
    },
    {
      "epoch": 0.04,
      "learning_rate": 7.998682682374572e-05,
      "loss": 1.6529,
      "step": 130
    },
    {
      "epoch": 0.04,
      "learning_rate": 7.998583302138217e-05,
      "loss": 1.6256,
      "step": 131
    },
    {
      "epoch": 0.04,
      "learning_rate": 7.998480308943692e-05,
      "loss": 1.6608,
      "step": 132
    },
    {
      "epoch": 0.04,
      "learning_rate": 7.998373702884062e-05,
      "loss": 1.6488,
      "step": 133
    },
    {
      "epoch": 0.04,
      "learning_rate": 7.998263484055649e-05,
      "loss": 1.6508,
      "step": 134
    },
    {
      "epoch": 0.04,
      "learning_rate": 7.998149652558045e-05,
      "loss": 1.6393,
      "step": 135
    },
    {
      "epoch": 0.04,
      "learning_rate": 7.998032208494101e-05,
      "loss": 1.6887,
      "step": 136
    },
    {
      "epoch": 0.04,
      "learning_rate": 7.997911151969936e-05,
      "loss": 1.6903,
      "step": 137
    },
    {
      "epoch": 0.04,
      "learning_rate": 7.997786483094931e-05,
      "loss": 1.6407,
      "step": 138
    },
    {
      "epoch": 0.04,
      "learning_rate": 7.997658201981732e-05,
      "loss": 1.6433,
      "step": 139
    },
    {
      "epoch": 0.04,
      "learning_rate": 7.997526308746249e-05,
      "loss": 1.6846,
      "step": 140
    },
    {
      "epoch": 0.04,
      "learning_rate": 7.997390803507655e-05,
      "loss": 1.6547,
      "step": 141
    },
    {
      "epoch": 0.04,
      "learning_rate": 7.997251686388387e-05,
      "loss": 1.666,
      "step": 142
    },
    {
      "epoch": 0.04,
      "learning_rate": 7.997108957514146e-05,
      "loss": 1.6367,
      "step": 143
    },
    {
      "epoch": 0.04,
      "learning_rate": 7.996962617013897e-05,
      "loss": 1.6775,
      "step": 144
    },
    {
      "epoch": 0.04,
      "learning_rate": 7.996812665019865e-05,
      "loss": 1.6688,
      "step": 145
    },
    {
      "epoch": 0.04,
      "learning_rate": 7.996659101667542e-05,
      "loss": 1.6512,
      "step": 146
    },
    {
      "epoch": 0.04,
      "learning_rate": 7.996501927095682e-05,
      "loss": 1.6536,
      "step": 147
    },
    {
      "epoch": 0.04,
      "learning_rate": 7.9963411414463e-05,
      "loss": 1.615,
      "step": 148
    },
    {
      "epoch": 0.04,
      "learning_rate": 7.996176744864678e-05,
      "loss": 1.6131,
      "step": 149
    },
    {
      "epoch": 0.04,
      "learning_rate": 7.996008737499354e-05,
      "loss": 1.6378,
      "step": 150
    },
    {
      "epoch": 0.04,
      "learning_rate": 7.995837119502137e-05,
      "loss": 1.6612,
      "step": 151
    },
    {
      "epoch": 0.04,
      "learning_rate": 7.995661891028093e-05,
      "loss": 1.5754,
      "step": 152
    },
    {
      "epoch": 0.04,
      "learning_rate": 7.99548305223555e-05,
      "loss": 1.634,
      "step": 153
    },
    {
      "epoch": 0.05,
      "learning_rate": 7.995300603286099e-05,
      "loss": 1.6674,
      "step": 154
    },
    {
      "epoch": 0.05,
      "learning_rate": 7.995114544344596e-05,
      "loss": 1.6133,
      "step": 155
    },
    {
      "epoch": 0.05,
      "learning_rate": 7.994924875579154e-05,
      "loss": 1.662,
      "step": 156
    },
    {
      "epoch": 0.05,
      "learning_rate": 7.994731597161152e-05,
      "loss": 1.6249,
      "step": 157
    },
    {
      "epoch": 0.05,
      "learning_rate": 7.994534709265226e-05,
      "loss": 1.6465,
      "step": 158
    },
    {
      "epoch": 0.05,
      "learning_rate": 7.994334212069278e-05,
      "loss": 1.6165,
      "step": 159
    },
    {
      "epoch": 0.05,
      "learning_rate": 7.99413010575447e-05,
      "loss": 1.6191,
      "step": 160
    },
    {
      "epoch": 0.05,
      "learning_rate": 7.993922390505221e-05,
      "loss": 1.5617,
      "step": 161
    },
    {
      "epoch": 0.05,
      "learning_rate": 7.993711066509218e-05,
      "loss": 1.6773,
      "step": 162
    },
    {
      "epoch": 0.05,
      "learning_rate": 7.993496133957401e-05,
      "loss": 1.6366,
      "step": 163
    },
    {
      "epoch": 0.05,
      "learning_rate": 7.993277593043977e-05,
      "loss": 1.6121,
      "step": 164
    },
    {
      "epoch": 0.05,
      "learning_rate": 7.99305544396641e-05,
      "loss": 1.6534,
      "step": 165
    },
    {
      "epoch": 0.05,
      "learning_rate": 7.992829686925425e-05,
      "loss": 1.6495,
      "step": 166
    },
    {
      "epoch": 0.05,
      "learning_rate": 7.992600322125006e-05,
      "loss": 1.6239,
      "step": 167
    },
    {
      "epoch": 0.05,
      "learning_rate": 7.992367349772398e-05,
      "loss": 1.6081,
      "step": 168
    },
    {
      "epoch": 0.05,
      "learning_rate": 7.992130770078107e-05,
      "loss": 1.6434,
      "step": 169
    },
    {
      "epoch": 0.05,
      "learning_rate": 7.991890583255895e-05,
      "loss": 1.6174,
      "step": 170
    },
    {
      "epoch": 0.05,
      "learning_rate": 7.991646789522786e-05,
      "loss": 1.6288,
      "step": 171
    },
    {
      "epoch": 0.05,
      "learning_rate": 7.991399389099062e-05,
      "loss": 1.5779,
      "step": 172
    },
    {
      "epoch": 0.05,
      "learning_rate": 7.991148382208265e-05,
      "loss": 1.6689,
      "step": 173
    },
    {
      "epoch": 0.05,
      "learning_rate": 7.990893769077192e-05,
      "loss": 1.5706,
      "step": 174
    },
    {
      "epoch": 0.05,
      "learning_rate": 7.990635549935905e-05,
      "loss": 1.6077,
      "step": 175
    },
    {
      "epoch": 0.05,
      "learning_rate": 7.990373725017716e-05,
      "loss": 1.6109,
      "step": 176
    },
    {
      "epoch": 0.05,
      "learning_rate": 7.990108294559202e-05,
      "loss": 1.6423,
      "step": 177
    },
    {
      "epoch": 0.05,
      "learning_rate": 7.989839258800196e-05,
      "loss": 1.6463,
      "step": 178
    },
    {
      "epoch": 0.05,
      "learning_rate": 7.989566617983786e-05,
      "loss": 1.6117,
      "step": 179
    },
    {
      "epoch": 0.05,
      "learning_rate": 7.989290372356319e-05,
      "loss": 1.6587,
      "step": 180
    },
    {
      "epoch": 0.05,
      "learning_rate": 7.989010522167401e-05,
      "loss": 1.6205,
      "step": 181
    },
    {
      "epoch": 0.05,
      "learning_rate": 7.988727067669893e-05,
      "loss": 1.5736,
      "step": 182
    },
    {
      "epoch": 0.05,
      "learning_rate": 7.988440009119911e-05,
      "loss": 1.6257,
      "step": 183
    },
    {
      "epoch": 0.05,
      "learning_rate": 7.988149346776833e-05,
      "loss": 1.6637,
      "step": 184
    },
    {
      "epoch": 0.05,
      "learning_rate": 7.987855080903286e-05,
      "loss": 1.563,
      "step": 185
    },
    {
      "epoch": 0.05,
      "learning_rate": 7.987557211765161e-05,
      "loss": 1.5749,
      "step": 186
    },
    {
      "epoch": 0.05,
      "learning_rate": 7.987255739631596e-05,
      "loss": 1.5997,
      "step": 187
    },
    {
      "epoch": 0.06,
      "learning_rate": 7.986950664774992e-05,
      "loss": 1.5899,
      "step": 188
    },
    {
      "epoch": 0.06,
      "learning_rate": 7.986641987471e-05,
      "loss": 1.6201,
      "step": 189
    },
    {
      "epoch": 0.06,
      "learning_rate": 7.986329707998532e-05,
      "loss": 1.5987,
      "step": 190
    },
    {
      "epoch": 0.06,
      "learning_rate": 7.986013826639747e-05,
      "loss": 1.6092,
      "step": 191
    },
    {
      "epoch": 0.06,
      "learning_rate": 7.985694343680065e-05,
      "loss": 1.6173,
      "step": 192
    },
    {
      "epoch": 0.06,
      "learning_rate": 7.985371259408157e-05,
      "loss": 1.5692,
      "step": 193
    },
    {
      "epoch": 0.06,
      "learning_rate": 7.985044574115951e-05,
      "loss": 1.5751,
      "step": 194
    },
    {
      "epoch": 0.06,
      "learning_rate": 7.984714288098623e-05,
      "loss": 1.596,
      "step": 195
    },
    {
      "epoch": 0.06,
      "learning_rate": 7.984380401654608e-05,
      "loss": 1.6791,
      "step": 196
    },
    {
      "epoch": 0.06,
      "learning_rate": 7.984042915085592e-05,
      "loss": 1.608,
      "step": 197
    },
    {
      "epoch": 0.06,
      "learning_rate": 7.983701828696515e-05,
      "loss": 1.5983,
      "step": 198
    },
    {
      "epoch": 0.06,
      "learning_rate": 7.983357142795569e-05,
      "loss": 1.586,
      "step": 199
    },
    {
      "epoch": 0.06,
      "learning_rate": 7.983008857694196e-05,
      "loss": 1.5633,
      "step": 200
    },
    {
      "epoch": 0.06,
      "eval_loss": 0.17678602039813995,
      "eval_runtime": 24.7483,
      "eval_samples_per_second": 80.814,
      "eval_steps_per_second": 0.647,
      "step": 200
    },
    {
      "epoch": 0.06,
      "learning_rate": 7.982656973707095e-05,
      "loss": 1.5357,
      "step": 201
    },
    {
      "epoch": 0.06,
      "learning_rate": 7.982301491152213e-05,
      "loss": 1.5821,
      "step": 202
    },
    {
      "epoch": 0.06,
      "learning_rate": 7.98194241035075e-05,
      "loss": 1.5463,
      "step": 203
    },
    {
      "epoch": 0.06,
      "learning_rate": 7.981579731627155e-05,
      "loss": 1.5106,
      "step": 204
    },
    {
      "epoch": 0.06,
      "learning_rate": 7.981213455309131e-05,
      "loss": 1.5734,
      "step": 205
    },
    {
      "epoch": 0.06,
      "learning_rate": 7.980843581727633e-05,
      "loss": 1.5619,
      "step": 206
    },
    {
      "epoch": 0.06,
      "learning_rate": 7.98047011121686e-05,
      "loss": 1.5589,
      "step": 207
    },
    {
      "epoch": 0.06,
      "learning_rate": 7.980093044114269e-05,
      "loss": 1.543,
      "step": 208
    },
    {
      "epoch": 0.06,
      "learning_rate": 7.979712380760557e-05,
      "loss": 1.508,
      "step": 209
    },
    {
      "epoch": 0.06,
      "learning_rate": 7.97932812149968e-05,
      "loss": 1.5922,
      "step": 210
    },
    {
      "epoch": 0.06,
      "learning_rate": 7.97894026667884e-05,
      "loss": 1.5705,
      "step": 211
    },
    {
      "epoch": 0.06,
      "learning_rate": 7.978548816648484e-05,
      "loss": 1.5767,
      "step": 212
    },
    {
      "epoch": 0.06,
      "learning_rate": 7.978153771762311e-05,
      "loss": 1.5894,
      "step": 213
    },
    {
      "epoch": 0.06,
      "learning_rate": 7.977755132377269e-05,
      "loss": 1.592,
      "step": 214
    },
    {
      "epoch": 0.06,
      "learning_rate": 7.977352898853551e-05,
      "loss": 1.5524,
      "step": 215
    },
    {
      "epoch": 0.06,
      "learning_rate": 7.9769470715546e-05,
      "loss": 1.5463,
      "step": 216
    },
    {
      "epoch": 0.06,
      "learning_rate": 7.976537650847104e-05,
      "loss": 1.4902,
      "step": 217
    },
    {
      "epoch": 0.06,
      "learning_rate": 7.976124637101e-05,
      "loss": 1.5343,
      "step": 218
    },
    {
      "epoch": 0.06,
      "learning_rate": 7.975708030689471e-05,
      "loss": 1.5716,
      "step": 219
    },
    {
      "epoch": 0.06,
      "learning_rate": 7.975287831988944e-05,
      "loss": 1.5758,
      "step": 220
    },
    {
      "epoch": 0.06,
      "learning_rate": 7.974864041379095e-05,
      "loss": 1.5188,
      "step": 221
    },
    {
      "epoch": 0.07,
      "learning_rate": 7.974436659242843e-05,
      "loss": 1.5565,
      "step": 222
    },
    {
      "epoch": 0.07,
      "learning_rate": 7.974005685966354e-05,
      "loss": 1.508,
      "step": 223
    },
    {
      "epoch": 0.07,
      "learning_rate": 7.973571121939037e-05,
      "loss": 1.5189,
      "step": 224
    },
    {
      "epoch": 0.07,
      "learning_rate": 7.973132967553549e-05,
      "loss": 1.4511,
      "step": 225
    },
    {
      "epoch": 0.07,
      "learning_rate": 7.972691223205784e-05,
      "loss": 1.5976,
      "step": 226
    },
    {
      "epoch": 0.07,
      "learning_rate": 7.97224588929489e-05,
      "loss": 1.5234,
      "step": 227
    },
    {
      "epoch": 0.07,
      "learning_rate": 7.971796966223248e-05,
      "loss": 1.5315,
      "step": 228
    },
    {
      "epoch": 0.07,
      "learning_rate": 7.97134445439649e-05,
      "loss": 1.5066,
      "step": 229
    },
    {
      "epoch": 0.07,
      "learning_rate": 7.970888354223485e-05,
      "loss": 1.5541,
      "step": 230
    },
    {
      "epoch": 0.07,
      "learning_rate": 7.970428666116348e-05,
      "loss": 1.5383,
      "step": 231
    },
    {
      "epoch": 0.07,
      "learning_rate": 7.969965390490434e-05,
      "loss": 1.5633,
      "step": 232
    },
    {
      "epoch": 0.07,
      "learning_rate": 7.969498527764341e-05,
      "loss": 1.5309,
      "step": 233
    },
    {
      "epoch": 0.07,
      "learning_rate": 7.969028078359905e-05,
      "loss": 1.5301,
      "step": 234
    },
    {
      "epoch": 0.07,
      "learning_rate": 7.968554042702208e-05,
      "loss": 1.5394,
      "step": 235
    },
    {
      "epoch": 0.07,
      "learning_rate": 7.968076421219568e-05,
      "loss": 1.4884,
      "step": 236
    },
    {
      "epoch": 0.07,
      "learning_rate": 7.967595214343544e-05,
      "loss": 1.5066,
      "step": 237
    },
    {
      "epoch": 0.07,
      "learning_rate": 7.967110422508936e-05,
      "loss": 1.455,
      "step": 238
    },
    {
      "epoch": 0.07,
      "learning_rate": 7.966622046153783e-05,
      "loss": 1.5569,
      "step": 239
    },
    {
      "epoch": 0.07,
      "learning_rate": 7.966130085719359e-05,
      "loss": 1.4966,
      "step": 240
    },
    {
      "epoch": 0.07,
      "learning_rate": 7.965634541650182e-05,
      "loss": 1.5731,
      "step": 241
    },
    {
      "epoch": 0.07,
      "learning_rate": 7.965135414394007e-05,
      "loss": 1.4655,
      "step": 242
    },
    {
      "epoch": 0.07,
      "learning_rate": 7.964632704401823e-05,
      "loss": 1.4646,
      "step": 243
    },
    {
      "epoch": 0.07,
      "learning_rate": 7.964126412127858e-05,
      "loss": 1.5041,
      "step": 244
    },
    {
      "epoch": 0.07,
      "learning_rate": 7.963616538029578e-05,
      "loss": 1.5773,
      "step": 245
    },
    {
      "epoch": 0.07,
      "learning_rate": 7.963103082567685e-05,
      "loss": 1.5009,
      "step": 246
    },
    {
      "epoch": 0.07,
      "learning_rate": 7.962586046206117e-05,
      "loss": 1.4912,
      "step": 247
    },
    {
      "epoch": 0.07,
      "learning_rate": 7.962065429412046e-05,
      "loss": 1.5504,
      "step": 248
    },
    {
      "epoch": 0.07,
      "learning_rate": 7.961541232655879e-05,
      "loss": 1.4792,
      "step": 249
    },
    {
      "epoch": 0.07,
      "learning_rate": 7.961013456411262e-05,
      "loss": 1.5081,
      "step": 250
    },
    {
      "epoch": 0.07,
      "learning_rate": 7.960482101155072e-05,
      "loss": 1.472,
      "step": 251
    },
    {
      "epoch": 0.07,
      "learning_rate": 7.959947167367417e-05,
      "loss": 1.5423,
      "step": 252
    },
    {
      "epoch": 0.07,
      "learning_rate": 7.959408655531646e-05,
      "loss": 1.4912,
      "step": 253
    },
    {
      "epoch": 0.07,
      "learning_rate": 7.958866566134332e-05,
      "loss": 1.5218,
      "step": 254
    },
    {
      "epoch": 0.07,
      "learning_rate": 7.958320899665287e-05,
      "loss": 1.5307,
      "step": 255
    },
    {
      "epoch": 0.08,
      "learning_rate": 7.957771656617555e-05,
      "loss": 1.5089,
      "step": 256
    },
    {
      "epoch": 0.08,
      "learning_rate": 7.957218837487408e-05,
      "loss": 1.5274,
      "step": 257
    },
    {
      "epoch": 0.08,
      "learning_rate": 7.956662442774351e-05,
      "loss": 1.4868,
      "step": 258
    },
    {
      "epoch": 0.08,
      "learning_rate": 7.95610247298112e-05,
      "loss": 1.5192,
      "step": 259
    },
    {
      "epoch": 0.08,
      "learning_rate": 7.95553892861368e-05,
      "loss": 1.5363,
      "step": 260
    },
    {
      "epoch": 0.08,
      "learning_rate": 7.954971810181229e-05,
      "loss": 1.4791,
      "step": 261
    },
    {
      "epoch": 0.08,
      "learning_rate": 7.95440111819619e-05,
      "loss": 1.5371,
      "step": 262
    },
    {
      "epoch": 0.08,
      "learning_rate": 7.953826853174218e-05,
      "loss": 1.4994,
      "step": 263
    },
    {
      "epoch": 0.08,
      "learning_rate": 7.953249015634198e-05,
      "loss": 1.5022,
      "step": 264
    },
    {
      "epoch": 0.08,
      "learning_rate": 7.952667606098237e-05,
      "loss": 1.4916,
      "step": 265
    },
    {
      "epoch": 0.08,
      "learning_rate": 7.952082625091676e-05,
      "loss": 1.4575,
      "step": 266
    },
    {
      "epoch": 0.08,
      "learning_rate": 7.951494073143078e-05,
      "loss": 1.4916,
      "step": 267
    },
    {
      "epoch": 0.08,
      "learning_rate": 7.950901950784236e-05,
      "loss": 1.4733,
      "step": 268
    },
    {
      "epoch": 0.08,
      "learning_rate": 7.950306258550168e-05,
      "loss": 1.4767,
      "step": 269
    },
    {
      "epoch": 0.08,
      "learning_rate": 7.949706996979115e-05,
      "loss": 1.5004,
      "step": 270
    },
    {
      "epoch": 0.08,
      "learning_rate": 7.949104166612551e-05,
      "loss": 1.5292,
      "step": 271
    },
    {
      "epoch": 0.08,
      "learning_rate": 7.948497767995163e-05,
      "loss": 1.4958,
      "step": 272
    },
    {
      "epoch": 0.08,
      "learning_rate": 7.947887801674872e-05,
      "loss": 1.4727,
      "step": 273
    },
    {
      "epoch": 0.08,
      "learning_rate": 7.947274268202817e-05,
      "loss": 1.5495,
      "step": 274
    },
    {
      "epoch": 0.08,
      "learning_rate": 7.946657168133363e-05,
      "loss": 1.4684,
      "step": 275
    },
    {
      "epoch": 0.08,
      "learning_rate": 7.946036502024098e-05,
      "loss": 1.4778,
      "step": 276
    },
    {
      "epoch": 0.08,
      "learning_rate": 7.945412270435826e-05,
      "loss": 1.4795,
      "step": 277
    },
    {
      "epoch": 0.08,
      "learning_rate": 7.944784473932583e-05,
      "loss": 1.517,
      "step": 278
    },
    {
      "epoch": 0.08,
      "learning_rate": 7.944153113081618e-05,
      "loss": 1.4645,
      "step": 279
    },
    {
      "epoch": 0.08,
      "learning_rate": 7.943518188453403e-05,
      "loss": 1.5227,
      "step": 280
    },
    {
      "epoch": 0.08,
      "learning_rate": 7.94287970062163e-05,
      "loss": 1.5244,
      "step": 281
    },
    {
      "epoch": 0.08,
      "learning_rate": 7.942237650163213e-05,
      "loss": 1.5063,
      "step": 282
    },
    {
      "epoch": 0.08,
      "learning_rate": 7.941592037658279e-05,
      "loss": 1.4413,
      "step": 283
    },
    {
      "epoch": 0.08,
      "learning_rate": 7.940942863690181e-05,
      "loss": 1.5263,
      "step": 284
    },
    {
      "epoch": 0.08,
      "learning_rate": 7.940290128845485e-05,
      "loss": 1.4906,
      "step": 285
    },
    {
      "epoch": 0.08,
      "learning_rate": 7.939633833713975e-05,
      "loss": 1.5155,
      "step": 286
    },
    {
      "epoch": 0.08,
      "learning_rate": 7.938973978888654e-05,
      "loss": 1.5025,
      "step": 287
    },
    {
      "epoch": 0.08,
      "learning_rate": 7.93831056496574e-05,
      "loss": 1.4087,
      "step": 288
    },
    {
      "epoch": 0.08,
      "learning_rate": 7.937643592544665e-05,
      "loss": 1.5133,
      "step": 289
    },
    {
      "epoch": 0.09,
      "learning_rate": 7.936973062228081e-05,
      "loss": 1.5485,
      "step": 290
    },
    {
      "epoch": 0.09,
      "learning_rate": 7.93629897462185e-05,
      "loss": 1.4673,
      "step": 291
    },
    {
      "epoch": 0.09,
      "learning_rate": 7.935621330335052e-05,
      "loss": 1.4585,
      "step": 292
    },
    {
      "epoch": 0.09,
      "learning_rate": 7.934940129979979e-05,
      "loss": 1.5047,
      "step": 293
    },
    {
      "epoch": 0.09,
      "learning_rate": 7.934255374172134e-05,
      "loss": 1.5027,
      "step": 294
    },
    {
      "epoch": 0.09,
      "learning_rate": 7.933567063530236e-05,
      "loss": 1.4654,
      "step": 295
    },
    {
      "epoch": 0.09,
      "learning_rate": 7.932875198676215e-05,
      "loss": 1.4796,
      "step": 296
    },
    {
      "epoch": 0.09,
      "learning_rate": 7.932179780235212e-05,
      "loss": 1.5474,
      "step": 297
    },
    {
      "epoch": 0.09,
      "learning_rate": 7.931480808835577e-05,
      "loss": 1.4751,
      "step": 298
    },
    {
      "epoch": 0.09,
      "learning_rate": 7.930778285108876e-05,
      "loss": 1.4563,
      "step": 299
    },
    {
      "epoch": 0.09,
      "learning_rate": 7.93007220968988e-05,
      "loss": 1.4894,
      "step": 300
    },
    {
      "epoch": 0.09,
      "eval_loss": 0.16949589550495148,
      "eval_runtime": 24.7827,
      "eval_samples_per_second": 80.701,
      "eval_steps_per_second": 0.646,
      "step": 300
    },
    {
      "epoch": 0.09,
      "learning_rate": 7.929362583216566e-05,
      "loss": 1.5054,
      "step": 301
    },
    {
      "epoch": 0.09,
      "learning_rate": 7.92864940633013e-05,
      "loss": 1.4639,
      "step": 302
    },
    {
      "epoch": 0.09,
      "learning_rate": 7.927932679674964e-05,
      "loss": 1.4206,
      "step": 303
    },
    {
      "epoch": 0.09,
      "learning_rate": 7.927212403898677e-05,
      "loss": 1.4521,
      "step": 304
    },
    {
      "epoch": 0.09,
      "learning_rate": 7.926488579652081e-05,
      "loss": 1.5274,
      "step": 305
    },
    {
      "epoch": 0.09,
      "learning_rate": 7.925761207589193e-05,
      "loss": 1.4917,
      "step": 306
    },
    {
      "epoch": 0.09,
      "learning_rate": 7.925030288367238e-05,
      "loss": 1.4868,
      "step": 307
    },
    {
      "epoch": 0.09,
      "learning_rate": 7.924295822646643e-05,
      "loss": 1.5159,
      "step": 308
    },
    {
      "epoch": 0.09,
      "learning_rate": 7.923557811091044e-05,
      "loss": 1.5021,
      "step": 309
    },
    {
      "epoch": 0.09,
      "learning_rate": 7.922816254367277e-05,
      "loss": 1.4571,
      "step": 310
    },
    {
      "epoch": 0.09,
      "learning_rate": 7.922071153145384e-05,
      "loss": 1.4389,
      "step": 311
    },
    {
      "epoch": 0.09,
      "learning_rate": 7.921322508098605e-05,
      "loss": 1.5344,
      "step": 312
    },
    {
      "epoch": 0.09,
      "learning_rate": 7.92057031990339e-05,
      "loss": 1.4675,
      "step": 313
    },
    {
      "epoch": 0.09,
      "learning_rate": 7.919814589239382e-05,
      "loss": 1.4391,
      "step": 314
    },
    {
      "epoch": 0.09,
      "learning_rate": 7.919055316789431e-05,
      "loss": 1.483,
      "step": 315
    },
    {
      "epoch": 0.09,
      "learning_rate": 7.918292503239584e-05,
      "loss": 1.4582,
      "step": 316
    },
    {
      "epoch": 0.09,
      "learning_rate": 7.91752614927909e-05,
      "loss": 1.4664,
      "step": 317
    },
    {
      "epoch": 0.09,
      "learning_rate": 7.91675625560039e-05,
      "loss": 1.4992,
      "step": 318
    },
    {
      "epoch": 0.09,
      "learning_rate": 7.915982822899136e-05,
      "loss": 1.4297,
      "step": 319
    },
    {
      "epoch": 0.09,
      "learning_rate": 7.915205851874168e-05,
      "loss": 1.5146,
      "step": 320
    },
    {
      "epoch": 0.09,
      "learning_rate": 7.914425343227523e-05,
      "loss": 1.4252,
      "step": 321
    },
    {
      "epoch": 0.09,
      "learning_rate": 7.91364129766444e-05,
      "loss": 1.3957,
      "step": 322
    },
    {
      "epoch": 0.09,
      "learning_rate": 7.91285371589335e-05,
      "loss": 1.4833,
      "step": 323
    },
    {
      "epoch": 0.1,
      "learning_rate": 7.912062598625879e-05,
      "loss": 1.4306,
      "step": 324
    },
    {
      "epoch": 0.1,
      "learning_rate": 7.91126794657685e-05,
      "loss": 1.4544,
      "step": 325
    },
    {
      "epoch": 0.1,
      "learning_rate": 7.910469760464277e-05,
      "loss": 1.442,
      "step": 326
    },
    {
      "epoch": 0.1,
      "learning_rate": 7.909668041009372e-05,
      "loss": 1.5333,
      "step": 327
    },
    {
      "epoch": 0.1,
      "learning_rate": 7.908862788936532e-05,
      "loss": 1.4602,
      "step": 328
    },
    {
      "epoch": 0.1,
      "learning_rate": 7.908054004973352e-05,
      "loss": 1.4749,
      "step": 329
    },
    {
      "epoch": 0.1,
      "learning_rate": 7.907241689850617e-05,
      "loss": 1.4597,
      "step": 330
    },
    {
      "epoch": 0.1,
      "learning_rate": 7.906425844302302e-05,
      "loss": 1.5203,
      "step": 331
    },
    {
      "epoch": 0.1,
      "learning_rate": 7.905606469065572e-05,
      "loss": 1.4625,
      "step": 332
    },
    {
      "epoch": 0.1,
      "learning_rate": 7.904783564880779e-05,
      "loss": 1.4267,
      "step": 333
    },
    {
      "epoch": 0.1,
      "learning_rate": 7.90395713249147e-05,
      "loss": 1.4752,
      "step": 334
    },
    {
      "epoch": 0.1,
      "learning_rate": 7.903127172644374e-05,
      "loss": 1.5467,
      "step": 335
    },
    {
      "epoch": 0.1,
      "learning_rate": 7.902293686089407e-05,
      "loss": 1.4317,
      "step": 336
    },
    {
      "epoch": 0.1,
      "learning_rate": 7.901456673579677e-05,
      "loss": 1.4016,
      "step": 337
    },
    {
      "epoch": 0.1,
      "learning_rate": 7.900616135871474e-05,
      "loss": 1.5466,
      "step": 338
    },
    {
      "epoch": 0.1,
      "learning_rate": 7.899772073724272e-05,
      "loss": 1.4207,
      "step": 339
    },
    {
      "epoch": 0.1,
      "learning_rate": 7.898924487900734e-05,
      "loss": 1.4678,
      "step": 340
    },
    {
      "epoch": 0.1,
      "learning_rate": 7.898073379166703e-05,
      "loss": 1.458,
      "step": 341
    },
    {
      "epoch": 0.1,
      "learning_rate": 7.897218748291206e-05,
      "loss": 1.5305,
      "step": 342
    },
    {
      "epoch": 0.1,
      "learning_rate": 7.896360596046453e-05,
      "loss": 1.4003,
      "step": 343
    },
    {
      "epoch": 0.1,
      "learning_rate": 7.895498923207836e-05,
      "loss": 1.505,
      "step": 344
    },
    {
      "epoch": 0.1,
      "learning_rate": 7.894633730553928e-05,
      "loss": 1.4764,
      "step": 345
    },
    {
      "epoch": 0.1,
      "learning_rate": 7.893765018866482e-05,
      "loss": 1.4739,
      "step": 346
    },
    {
      "epoch": 0.1,
      "learning_rate": 7.89289278893043e-05,
      "loss": 1.4403,
      "step": 347
    },
    {
      "epoch": 0.1,
      "learning_rate": 7.892017041533886e-05,
      "loss": 1.4891,
      "step": 348
    },
    {
      "epoch": 0.1,
      "learning_rate": 7.891137777468136e-05,
      "loss": 1.4556,
      "step": 349
    },
    {
      "epoch": 0.1,
      "learning_rate": 7.89025499752765e-05,
      "loss": 1.5635,
      "step": 350
    },
    {
      "epoch": 0.1,
      "learning_rate": 7.889368702510073e-05,
      "loss": 1.5482,
      "step": 351
    },
    {
      "epoch": 0.1,
      "learning_rate": 7.888478893216223e-05,
      "loss": 1.4714,
      "step": 352
    },
    {
      "epoch": 0.1,
      "learning_rate": 7.887585570450098e-05,
      "loss": 1.5668,
      "step": 353
    },
    {
      "epoch": 0.1,
      "learning_rate": 7.886688735018866e-05,
      "loss": 1.6016,
      "step": 354
    },
    {
      "epoch": 0.1,
      "learning_rate": 7.885788387732874e-05,
      "loss": 1.6025,
      "step": 355
    },
    {
      "epoch": 0.1,
      "learning_rate": 7.884884529405636e-05,
      "loss": 1.5715,
      "step": 356
    },
    {
      "epoch": 0.1,
      "learning_rate": 7.883977160853847e-05,
      "loss": 1.6736,
      "step": 357
    },
    {
      "epoch": 0.11,
      "learning_rate": 7.883066282897362e-05,
      "loss": 1.61,
      "step": 358
    },
    {
      "epoch": 0.11,
      "learning_rate": 7.882151896359218e-05,
      "loss": 1.5429,
      "step": 359
    },
    {
      "epoch": 0.11,
      "learning_rate": 7.881234002065617e-05,
      "loss": 1.5019,
      "step": 360
    },
    {
      "epoch": 0.11,
      "learning_rate": 7.880312600845928e-05,
      "loss": 1.4937,
      "step": 361
    },
    {
      "epoch": 0.11,
      "learning_rate": 7.879387693532697e-05,
      "loss": 1.4798,
      "step": 362
    },
    {
      "epoch": 0.11,
      "learning_rate": 7.878459280961629e-05,
      "loss": 1.4661,
      "step": 363
    },
    {
      "epoch": 0.11,
      "learning_rate": 7.877527363971602e-05,
      "loss": 1.529,
      "step": 364
    },
    {
      "epoch": 0.11,
      "learning_rate": 7.876591943404658e-05,
      "loss": 1.5507,
      "step": 365
    },
    {
      "epoch": 0.11,
      "learning_rate": 7.875653020106006e-05,
      "loss": 1.499,
      "step": 366
    },
    {
      "epoch": 0.11,
      "learning_rate": 7.874710594924017e-05,
      "loss": 1.4649,
      "step": 367
    },
    {
      "epoch": 0.11,
      "learning_rate": 7.873764668710228e-05,
      "loss": 1.5681,
      "step": 368
    },
    {
      "epoch": 0.11,
      "learning_rate": 7.872815242319343e-05,
      "loss": 1.4876,
      "step": 369
    },
    {
      "epoch": 0.11,
      "learning_rate": 7.871862316609224e-05,
      "loss": 1.5506,
      "step": 370
    },
    {
      "epoch": 0.11,
      "learning_rate": 7.870905892440895e-05,
      "loss": 1.5359,
      "step": 371
    },
    {
      "epoch": 0.11,
      "learning_rate": 7.869945970678541e-05,
      "loss": 1.5306,
      "step": 372
    },
    {
      "epoch": 0.11,
      "learning_rate": 7.868982552189514e-05,
      "loss": 1.489,
      "step": 373
    },
    {
      "epoch": 0.11,
      "learning_rate": 7.868015637844313e-05,
      "loss": 1.5582,
      "step": 374
    },
    {
      "epoch": 0.11,
      "learning_rate": 7.867045228516607e-05,
      "loss": 1.5321,
      "step": 375
    },
    {
      "epoch": 0.11,
      "learning_rate": 7.866071325083217e-05,
      "loss": 1.5471,
      "step": 376
    },
    {
      "epoch": 0.11,
      "learning_rate": 7.865093928424123e-05,
      "loss": 1.4685,
      "step": 377
    },
    {
      "epoch": 0.11,
      "learning_rate": 7.864113039422464e-05,
      "loss": 1.5347,
      "step": 378
    },
    {
      "epoch": 0.11,
      "learning_rate": 7.863128658964527e-05,
      "loss": 1.5319,
      "step": 379
    },
    {
      "epoch": 0.11,
      "learning_rate": 7.862140787939761e-05,
      "loss": 1.5242,
      "step": 380
    },
    {
      "epoch": 0.11,
      "learning_rate": 7.861149427240765e-05,
      "loss": 1.4791,
      "step": 381
    },
    {
      "epoch": 0.11,
      "learning_rate": 7.860154577763293e-05,
      "loss": 1.4532,
      "step": 382
    },
    {
      "epoch": 0.11,
      "learning_rate": 7.859156240406252e-05,
      "loss": 1.4927,
      "step": 383
    },
    {
      "epoch": 0.11,
      "learning_rate": 7.858154416071697e-05,
      "loss": 1.5222,
      "step": 384
    },
    {
      "epoch": 0.11,
      "learning_rate": 7.857149105664839e-05,
      "loss": 1.4811,
      "step": 385
    },
    {
      "epoch": 0.11,
      "learning_rate": 7.856140310094033e-05,
      "loss": 1.5013,
      "step": 386
    },
    {
      "epoch": 0.11,
      "learning_rate": 7.855128030270786e-05,
      "loss": 1.4733,
      "step": 387
    },
    {
      "epoch": 0.11,
      "learning_rate": 7.854112267109756e-05,
      "loss": 1.5002,
      "step": 388
    },
    {
      "epoch": 0.11,
      "learning_rate": 7.853093021528742e-05,
      "loss": 1.4687,
      "step": 389
    },
    {
      "epoch": 0.11,
      "learning_rate": 7.852070294448693e-05,
      "loss": 1.5572,
      "step": 390
    },
    {
      "epoch": 0.11,
      "learning_rate": 7.851044086793709e-05,
      "loss": 1.5006,
      "step": 391
    },
    {
      "epoch": 0.12,
      "learning_rate": 7.850014399491024e-05,
      "loss": 1.5307,
      "step": 392
    },
    {
      "epoch": 0.12,
      "learning_rate": 7.848981233471024e-05,
      "loss": 1.4129,
      "step": 393
    },
    {
      "epoch": 0.12,
      "learning_rate": 7.847944589667236e-05,
      "loss": 1.5944,
      "step": 394
    },
    {
      "epoch": 0.12,
      "learning_rate": 7.846904469016328e-05,
      "loss": 1.5001,
      "step": 395
    },
    {
      "epoch": 0.12,
      "learning_rate": 7.845860872458116e-05,
      "loss": 1.4848,
      "step": 396
    },
    {
      "epoch": 0.12,
      "learning_rate": 7.844813800935544e-05,
      "loss": 1.4631,
      "step": 397
    },
    {
      "epoch": 0.12,
      "learning_rate": 7.843763255394711e-05,
      "loss": 1.5068,
      "step": 398
    },
    {
      "epoch": 0.12,
      "learning_rate": 7.842709236784842e-05,
      "loss": 1.4764,
      "step": 399
    },
    {
      "epoch": 0.12,
      "learning_rate": 7.841651746058308e-05,
      "loss": 1.5033,
      "step": 400
    },
    {
      "epoch": 0.12,
      "eval_loss": 0.1591062694787979,
      "eval_runtime": 24.8751,
      "eval_samples_per_second": 80.402,
      "eval_steps_per_second": 0.643,
      "step": 400
    },
    {
      "epoch": 0.12,
      "learning_rate": 7.840590784170614e-05,
      "loss": 1.5365,
      "step": 401
    },
    {
      "epoch": 0.12,
      "learning_rate": 7.839526352080403e-05,
      "loss": 1.5073,
      "step": 402
    },
    {
      "epoch": 0.12,
      "learning_rate": 7.838458450749452e-05,
      "loss": 1.429,
      "step": 403
    },
    {
      "epoch": 0.12,
      "learning_rate": 7.837387081142675e-05,
      "loss": 1.3831,
      "step": 404
    },
    {
      "epoch": 0.12,
      "learning_rate": 7.836312244228115e-05,
      "loss": 1.4639,
      "step": 405
    },
    {
      "epoch": 0.12,
      "learning_rate": 7.835233940976955e-05,
      "loss": 1.5145,
      "step": 406
    },
    {
      "epoch": 0.12,
      "learning_rate": 7.834152172363505e-05,
      "loss": 1.4166,
      "step": 407
    },
    {
      "epoch": 0.12,
      "learning_rate": 7.833066939365206e-05,
      "loss": 1.3999,
      "step": 408
    },
    {
      "epoch": 0.12,
      "learning_rate": 7.831978242962632e-05,
      "loss": 1.514,
      "step": 409
    },
    {
      "epoch": 0.12,
      "learning_rate": 7.830886084139483e-05,
      "loss": 1.4822,
      "step": 410
    },
    {
      "epoch": 0.12,
      "learning_rate": 7.829790463882593e-05,
      "loss": 1.4782,
      "step": 411
    },
    {
      "epoch": 0.12,
      "learning_rate": 7.828691383181918e-05,
      "loss": 1.5076,
      "step": 412
    },
    {
      "epoch": 0.12,
      "learning_rate": 7.827588843030543e-05,
      "loss": 1.4506,
      "step": 413
    },
    {
      "epoch": 0.12,
      "learning_rate": 7.826482844424679e-05,
      "loss": 1.4893,
      "step": 414
    },
    {
      "epoch": 0.12,
      "learning_rate": 7.825373388363661e-05,
      "loss": 1.4542,
      "step": 415
    },
    {
      "epoch": 0.12,
      "learning_rate": 7.824260475849949e-05,
      "loss": 1.4955,
      "step": 416
    },
    {
      "epoch": 0.12,
      "learning_rate": 7.823144107889126e-05,
      "loss": 1.5095,
      "step": 417
    },
    {
      "epoch": 0.12,
      "learning_rate": 7.822024285489896e-05,
      "loss": 1.5355,
      "step": 418
    },
    {
      "epoch": 0.12,
      "learning_rate": 7.820901009664089e-05,
      "loss": 1.4864,
      "step": 419
    },
    {
      "epoch": 0.12,
      "learning_rate": 7.819774281426645e-05,
      "loss": 1.4763,
      "step": 420
    },
    {
      "epoch": 0.12,
      "learning_rate": 7.818644101795635e-05,
      "loss": 1.4513,
      "step": 421
    },
    {
      "epoch": 0.12,
      "learning_rate": 7.817510471792243e-05,
      "loss": 1.4477,
      "step": 422
    },
    {
      "epoch": 0.12,
      "learning_rate": 7.81637339244077e-05,
      "loss": 1.4506,
      "step": 423
    },
    {
      "epoch": 0.12,
      "learning_rate": 7.815232864768636e-05,
      "loss": 1.4667,
      "step": 424
    },
    {
      "epoch": 0.12,
      "learning_rate": 7.814088889806374e-05,
      "loss": 1.509,
      "step": 425
    },
    {
      "epoch": 0.12,
      "learning_rate": 7.812941468587636e-05,
      "loss": 1.5293,
      "step": 426
    },
    {
      "epoch": 0.13,
      "learning_rate": 7.811790602149182e-05,
      "loss": 1.4266,
      "step": 427
    },
    {
      "epoch": 0.13,
      "learning_rate": 7.810636291530893e-05,
      "loss": 1.4618,
      "step": 428
    },
    {
      "epoch": 0.13,
      "learning_rate": 7.809478537775753e-05,
      "loss": 1.4584,
      "step": 429
    },
    {
      "epoch": 0.13,
      "learning_rate": 7.808317341929863e-05,
      "loss": 1.544,
      "step": 430
    },
    {
      "epoch": 0.13,
      "learning_rate": 7.807152705042435e-05,
      "loss": 1.4677,
      "step": 431
    },
    {
      "epoch": 0.13,
      "learning_rate": 7.805984628165785e-05,
      "loss": 1.43,
      "step": 432
    },
    {
      "epoch": 0.13,
      "learning_rate": 7.804813112355339e-05,
      "loss": 1.4429,
      "step": 433
    },
    {
      "epoch": 0.13,
      "learning_rate": 7.803638158669635e-05,
      "loss": 1.4925,
      "step": 434
    },
    {
      "epoch": 0.13,
      "learning_rate": 7.80245976817031e-05,
      "loss": 1.455,
      "step": 435
    },
    {
      "epoch": 0.13,
      "learning_rate": 7.801277941922114e-05,
      "loss": 1.4856,
      "step": 436
    },
    {
      "epoch": 0.13,
      "learning_rate": 7.800092680992893e-05,
      "loss": 1.436,
      "step": 437
    },
    {
      "epoch": 0.13,
      "learning_rate": 7.798903986453603e-05,
      "loss": 1.4624,
      "step": 438
    },
    {
      "epoch": 0.13,
      "learning_rate": 7.797711859378302e-05,
      "loss": 1.4458,
      "step": 439
    },
    {
      "epoch": 0.13,
      "learning_rate": 7.796516300844144e-05,
      "loss": 1.4932,
      "step": 440
    },
    {
      "epoch": 0.13,
      "learning_rate": 7.79531731193139e-05,
      "loss": 1.4725,
      "step": 441
    },
    {
      "epoch": 0.13,
      "learning_rate": 7.794114893723398e-05,
      "loss": 1.4608,
      "step": 442
    },
    {
      "epoch": 0.13,
      "learning_rate": 7.792909047306623e-05,
      "loss": 1.5006,
      "step": 443
    },
    {
      "epoch": 0.13,
      "learning_rate": 7.79169977377062e-05,
      "loss": 1.4568,
      "step": 444
    },
    {
      "epoch": 0.13,
      "learning_rate": 7.79048707420804e-05,
      "loss": 1.4616,
      "step": 445
    },
    {
      "epoch": 0.13,
      "learning_rate": 7.789270949714629e-05,
      "loss": 1.4262,
      "step": 446
    },
    {
      "epoch": 0.13,
      "learning_rate": 7.788051401389226e-05,
      "loss": 1.5236,
      "step": 447
    },
    {
      "epoch": 0.13,
      "learning_rate": 7.786828430333769e-05,
      "loss": 1.4592,
      "step": 448
    },
    {
      "epoch": 0.13,
      "learning_rate": 7.785602037653283e-05,
      "loss": 1.4754,
      "step": 449
    },
    {
      "epoch": 0.13,
      "learning_rate": 7.784372224455886e-05,
      "loss": 1.4002,
      "step": 450
    },
    {
      "epoch": 0.13,
      "learning_rate": 7.78313899185279e-05,
      "loss": 1.4484,
      "step": 451
    },
    {
      "epoch": 0.13,
      "learning_rate": 7.781902340958291e-05,
      "loss": 1.448,
      "step": 452
    },
    {
      "epoch": 0.13,
      "learning_rate": 7.78066227288978e-05,
      "loss": 1.335,
      "step": 453
    },
    {
      "epoch": 0.13,
      "learning_rate": 7.779418788767732e-05,
      "loss": 1.476,
      "step": 454
    },
    {
      "epoch": 0.13,
      "learning_rate": 7.778171889715706e-05,
      "loss": 1.5081,
      "step": 455
    },
    {
      "epoch": 0.13,
      "learning_rate": 7.776921576860351e-05,
      "loss": 1.4232,
      "step": 456
    },
    {
      "epoch": 0.13,
      "learning_rate": 7.775667851331402e-05,
      "loss": 1.4606,
      "step": 457
    },
    {
      "epoch": 0.13,
      "learning_rate": 7.77441071426167e-05,
      "loss": 1.4605,
      "step": 458
    },
    {
      "epoch": 0.13,
      "learning_rate": 7.773150166787057e-05,
      "loss": 1.4318,
      "step": 459
    },
    {
      "epoch": 0.13,
      "learning_rate": 7.771886210046541e-05,
      "loss": 1.4087,
      "step": 460
    },
    {
      "epoch": 0.14,
      "learning_rate": 7.770618845182182e-05,
      "loss": 1.4923,
      "step": 461
    },
    {
      "epoch": 0.14,
      "learning_rate": 7.76934807333912e-05,
      "loss": 1.4581,
      "step": 462
    },
    {
      "epoch": 0.14,
      "learning_rate": 7.768073895665573e-05,
      "loss": 1.4013,
      "step": 463
    },
    {
      "epoch": 0.14,
      "learning_rate": 7.766796313312836e-05,
      "loss": 1.4862,
      "step": 464
    },
    {
      "epoch": 0.14,
      "learning_rate": 7.765515327435282e-05,
      "loss": 1.457,
      "step": 465
    },
    {
      "epoch": 0.14,
      "learning_rate": 7.764230939190358e-05,
      "loss": 1.4757,
      "step": 466
    },
    {
      "epoch": 0.14,
      "learning_rate": 7.762943149738583e-05,
      "loss": 1.458,
      "step": 467
    },
    {
      "epoch": 0.14,
      "learning_rate": 7.761651960243554e-05,
      "loss": 1.4541,
      "step": 468
    },
    {
      "epoch": 0.14,
      "learning_rate": 7.760357371871935e-05,
      "loss": 1.4809,
      "step": 469
    },
    {
      "epoch": 0.14,
      "learning_rate": 7.759059385793466e-05,
      "loss": 1.4987,
      "step": 470
    },
    {
      "epoch": 0.14,
      "learning_rate": 7.757758003180956e-05,
      "loss": 1.4491,
      "step": 471
    },
    {
      "epoch": 0.14,
      "learning_rate": 7.756453225210279e-05,
      "loss": 1.39,
      "step": 472
    },
    {
      "epoch": 0.14,
      "learning_rate": 7.755145053060378e-05,
      "loss": 1.4782,
      "step": 473
    },
    {
      "epoch": 0.14,
      "learning_rate": 7.753833487913269e-05,
      "loss": 1.4619,
      "step": 474
    },
    {
      "epoch": 0.14,
      "learning_rate": 7.752518530954027e-05,
      "loss": 1.4314,
      "step": 475
    },
    {
      "epoch": 0.14,
      "learning_rate": 7.751200183370792e-05,
      "loss": 1.4804,
      "step": 476
    },
    {
      "epoch": 0.14,
      "learning_rate": 7.749878446354775e-05,
      "loss": 1.5127,
      "step": 477
    },
    {
      "epoch": 0.14,
      "learning_rate": 7.748553321100238e-05,
      "loss": 1.4101,
      "step": 478
    },
    {
      "epoch": 0.14,
      "learning_rate": 7.747224808804514e-05,
      "loss": 1.4358,
      "step": 479
    },
    {
      "epoch": 0.14,
      "learning_rate": 7.745892910667991e-05,
      "loss": 1.4502,
      "step": 480
    },
    {
      "epoch": 0.14,
      "learning_rate": 7.744557627894122e-05,
      "loss": 1.4834,
      "step": 481
    },
    {
      "epoch": 0.14,
      "learning_rate": 7.74321896168941e-05,
      "loss": 1.4163,
      "step": 482
    },
    {
      "epoch": 0.14,
      "learning_rate": 7.741876913263422e-05,
      "loss": 1.4211,
      "step": 483
    },
    {
      "epoch": 0.14,
      "learning_rate": 7.740531483828779e-05,
      "loss": 1.4433,
      "step": 484
    },
    {
      "epoch": 0.14,
      "learning_rate": 7.739182674601155e-05,
      "loss": 1.5171,
      "step": 485
    },
    {
      "epoch": 0.14,
      "learning_rate": 7.737830486799283e-05,
      "loss": 1.4501,
      "step": 486
    },
    {
      "epoch": 0.14,
      "learning_rate": 7.73647492164494e-05,
      "loss": 1.4429,
      "step": 487
    },
    {
      "epoch": 0.14,
      "learning_rate": 7.735115980362964e-05,
      "loss": 1.4322,
      "step": 488
    },
    {
      "epoch": 0.14,
      "learning_rate": 7.733753664181238e-05,
      "loss": 1.5037,
      "step": 489
    },
    {
      "epoch": 0.14,
      "learning_rate": 7.732387974330695e-05,
      "loss": 1.3909,
      "step": 490
    },
    {
      "epoch": 0.14,
      "learning_rate": 7.731018912045317e-05,
      "loss": 1.3948,
      "step": 491
    },
    {
      "epoch": 0.14,
      "learning_rate": 7.729646478562135e-05,
      "loss": 1.4809,
      "step": 492
    },
    {
      "epoch": 0.14,
      "learning_rate": 7.728270675121224e-05,
      "loss": 1.4446,
      "step": 493
    },
    {
      "epoch": 0.14,
      "learning_rate": 7.726891502965703e-05,
      "loss": 1.4451,
      "step": 494
    },
    {
      "epoch": 0.15,
      "learning_rate": 7.725508963341738e-05,
      "loss": 1.4505,
      "step": 495
    },
    {
      "epoch": 0.15,
      "learning_rate": 7.724123057498535e-05,
      "loss": 1.4686,
      "step": 496
    },
    {
      "epoch": 0.15,
      "learning_rate": 7.722733786688342e-05,
      "loss": 1.4688,
      "step": 497
    },
    {
      "epoch": 0.15,
      "learning_rate": 7.721341152166448e-05,
      "loss": 1.4049,
      "step": 498
    },
    {
      "epoch": 0.15,
      "learning_rate": 7.719945155191182e-05,
      "loss": 1.4108,
      "step": 499
    },
    {
      "epoch": 0.15,
      "learning_rate": 7.718545797023911e-05,
      "loss": 1.4831,
      "step": 500
    },
    {
      "epoch": 0.15,
      "eval_loss": 0.15852396190166473,
      "eval_runtime": 24.8605,
      "eval_samples_per_second": 80.449,
      "eval_steps_per_second": 0.644,
      "step": 500
    },
    {
      "epoch": 0.15,
      "learning_rate": 7.717143078929036e-05,
      "loss": 1.3533,
      "step": 501
    },
    {
      "epoch": 0.15,
      "learning_rate": 7.715737002174e-05,
      "loss": 1.4558,
      "step": 502
    },
    {
      "epoch": 0.15,
      "learning_rate": 7.714327568029272e-05,
      "loss": 1.4789,
      "step": 503
    },
    {
      "epoch": 0.15,
      "learning_rate": 7.712914777768363e-05,
      "loss": 1.4575,
      "step": 504
    },
    {
      "epoch": 0.15,
      "learning_rate": 7.711498632667815e-05,
      "loss": 1.444,
      "step": 505
    },
    {
      "epoch": 0.15,
      "learning_rate": 7.710079134007195e-05,
      "loss": 1.4473,
      "step": 506
    },
    {
      "epoch": 0.15,
      "learning_rate": 7.708656283069108e-05,
      "loss": 1.4423,
      "step": 507
    },
    {
      "epoch": 0.15,
      "learning_rate": 7.707230081139184e-05,
      "loss": 1.4372,
      "step": 508
    },
    {
      "epoch": 0.15,
      "learning_rate": 7.705800529506078e-05,
      "loss": 1.4089,
      "step": 509
    },
    {
      "epoch": 0.15,
      "learning_rate": 7.70436762946148e-05,
      "loss": 1.4312,
      "step": 510
    },
    {
      "epoch": 0.15,
      "learning_rate": 7.702931382300099e-05,
      "loss": 1.4583,
      "step": 511
    },
    {
      "epoch": 0.15,
      "learning_rate": 7.701491789319668e-05,
      "loss": 1.4665,
      "step": 512
    },
    {
      "epoch": 0.15,
      "learning_rate": 7.700048851820946e-05,
      "loss": 1.4332,
      "step": 513
    },
    {
      "epoch": 0.15,
      "learning_rate": 7.698602571107715e-05,
      "loss": 1.4066,
      "step": 514
    },
    {
      "epoch": 0.15,
      "learning_rate": 7.697152948486771e-05,
      "loss": 1.4632,
      "step": 515
    },
    {
      "epoch": 0.15,
      "learning_rate": 7.695699985267938e-05,
      "loss": 1.4606,
      "step": 516
    },
    {
      "epoch": 0.15,
      "learning_rate": 7.694243682764054e-05,
      "loss": 1.4023,
      "step": 517
    },
    {
      "epoch": 0.15,
      "learning_rate": 7.692784042290976e-05,
      "loss": 1.3851,
      "step": 518
    },
    {
      "epoch": 0.15,
      "learning_rate": 7.691321065167575e-05,
      "loss": 1.5098,
      "step": 519
    },
    {
      "epoch": 0.15,
      "learning_rate": 7.689854752715738e-05,
      "loss": 1.4204,
      "step": 520
    },
    {
      "epoch": 0.15,
      "learning_rate": 7.688385106260367e-05,
      "loss": 1.4667,
      "step": 521
    },
    {
      "epoch": 0.15,
      "learning_rate": 7.686912127129373e-05,
      "loss": 1.3986,
      "step": 522
    },
    {
      "epoch": 0.15,
      "learning_rate": 7.685435816653681e-05,
      "loss": 1.4379,
      "step": 523
    },
    {
      "epoch": 0.15,
      "learning_rate": 7.683956176167227e-05,
      "loss": 1.4095,
      "step": 524
    },
    {
      "epoch": 0.15,
      "learning_rate": 7.682473207006952e-05,
      "loss": 1.3978,
      "step": 525
    },
    {
      "epoch": 0.15,
      "learning_rate": 7.680986910512808e-05,
      "loss": 1.4799,
      "step": 526
    },
    {
      "epoch": 0.15,
      "learning_rate": 7.679497288027752e-05,
      "loss": 1.3983,
      "step": 527
    },
    {
      "epoch": 0.15,
      "learning_rate": 7.678004340897747e-05,
      "loss": 1.4198,
      "step": 528
    },
    {
      "epoch": 0.16,
      "learning_rate": 7.676508070471761e-05,
      "loss": 1.4455,
      "step": 529
    },
    {
      "epoch": 0.16,
      "learning_rate": 7.675008478101761e-05,
      "loss": 1.4528,
      "step": 530
    },
    {
      "epoch": 0.16,
      "learning_rate": 7.673505565142721e-05,
      "loss": 1.3844,
      "step": 531
    },
    {
      "epoch": 0.16,
      "learning_rate": 7.671999332952609e-05,
      "loss": 1.5005,
      "step": 532
    },
    {
      "epoch": 0.16,
      "learning_rate": 7.670489782892396e-05,
      "loss": 1.3887,
      "step": 533
    },
    {
      "epoch": 0.16,
      "learning_rate": 7.668976916326055e-05,
      "loss": 1.507,
      "step": 534
    },
    {
      "epoch": 0.16,
      "learning_rate": 7.667460734620545e-05,
      "loss": 1.4026,
      "step": 535
    },
    {
      "epoch": 0.16,
      "learning_rate": 7.665941239145829e-05,
      "loss": 1.4413,
      "step": 536
    },
    {
      "epoch": 0.16,
      "learning_rate": 7.664418431274861e-05,
      "loss": 1.4717,
      "step": 537
    },
    {
      "epoch": 0.16,
      "learning_rate": 7.662892312383592e-05,
      "loss": 1.4155,
      "step": 538
    },
    {
      "epoch": 0.16,
      "learning_rate": 7.661362883850955e-05,
      "loss": 1.3942,
      "step": 539
    },
    {
      "epoch": 0.16,
      "learning_rate": 7.659830147058885e-05,
      "loss": 1.4519,
      "step": 540
    },
    {
      "epoch": 0.16,
      "learning_rate": 7.658294103392299e-05,
      "loss": 1.4495,
      "step": 541
    },
    {
      "epoch": 0.16,
      "learning_rate": 7.656754754239105e-05,
      "loss": 1.4263,
      "step": 542
    },
    {
      "epoch": 0.16,
      "learning_rate": 7.655212100990195e-05,
      "loss": 1.4372,
      "step": 543
    },
    {
      "epoch": 0.16,
      "learning_rate": 7.65366614503945e-05,
      "loss": 1.416,
      "step": 544
    },
    {
      "epoch": 0.16,
      "learning_rate": 7.652116887783731e-05,
      "loss": 1.4372,
      "step": 545
    },
    {
      "epoch": 0.16,
      "learning_rate": 7.650564330622886e-05,
      "loss": 1.422,
      "step": 546
    },
    {
      "epoch": 0.16,
      "learning_rate": 7.649008474959743e-05,
      "loss": 1.4067,
      "step": 547
    },
    {
      "epoch": 0.16,
      "learning_rate": 7.647449322200108e-05,
      "loss": 1.4077,
      "step": 548
    },
    {
      "epoch": 0.16,
      "learning_rate": 7.645886873752771e-05,
      "loss": 1.406,
      "step": 549
    },
    {
      "epoch": 0.16,
      "learning_rate": 7.644321131029497e-05,
      "loss": 1.3987,
      "step": 550
    },
    {
      "epoch": 0.16,
      "learning_rate": 7.642752095445027e-05,
      "loss": 1.4265,
      "step": 551
    },
    {
      "epoch": 0.16,
      "learning_rate": 7.641179768417076e-05,
      "loss": 1.4959,
      "step": 552
    },
    {
      "epoch": 0.16,
      "learning_rate": 7.639604151366339e-05,
      "loss": 1.3929,
      "step": 553
    },
    {
      "epoch": 0.16,
      "learning_rate": 7.638025245716478e-05,
      "loss": 1.3676,
      "step": 554
    },
    {
      "epoch": 0.16,
      "learning_rate": 7.636443052894126e-05,
      "loss": 1.3983,
      "step": 555
    },
    {
      "epoch": 0.16,
      "learning_rate": 7.634857574328892e-05,
      "loss": 1.4391,
      "step": 556
    },
    {
      "epoch": 0.16,
      "learning_rate": 7.633268811453351e-05,
      "loss": 1.4234,
      "step": 557
    },
    {
      "epoch": 0.16,
      "learning_rate": 7.631676765703042e-05,
      "loss": 1.4307,
      "step": 558
    },
    {
      "epoch": 0.16,
      "learning_rate": 7.630081438516474e-05,
      "loss": 1.4127,
      "step": 559
    },
    {
      "epoch": 0.16,
      "learning_rate": 7.628482831335121e-05,
      "loss": 1.4341,
      "step": 560
    },
    {
      "epoch": 0.16,
      "learning_rate": 7.626880945603418e-05,
      "loss": 1.3882,
      "step": 561
    },
    {
      "epoch": 0.16,
      "learning_rate": 7.625275782768765e-05,
      "loss": 1.3935,
      "step": 562
    },
    {
      "epoch": 0.17,
      "learning_rate": 7.623667344281522e-05,
      "loss": 1.4045,
      "step": 563
    },
    {
      "epoch": 0.17,
      "learning_rate": 7.62205563159501e-05,
      "loss": 1.439,
      "step": 564
    },
    {
      "epoch": 0.17,
      "learning_rate": 7.620440646165508e-05,
      "loss": 1.3988,
      "step": 565
    },
    {
      "epoch": 0.17,
      "learning_rate": 7.61882238945225e-05,
      "loss": 1.434,
      "step": 566
    },
    {
      "epoch": 0.17,
      "learning_rate": 7.617200862917427e-05,
      "loss": 1.4462,
      "step": 567
    },
    {
      "epoch": 0.17,
      "learning_rate": 7.615576068026187e-05,
      "loss": 1.4164,
      "step": 568
    },
    {
      "epoch": 0.17,
      "learning_rate": 7.613948006246626e-05,
      "loss": 1.391,
      "step": 569
    },
    {
      "epoch": 0.17,
      "learning_rate": 7.612316679049796e-05,
      "loss": 1.4269,
      "step": 570
    },
    {
      "epoch": 0.17,
      "learning_rate": 7.610682087909699e-05,
      "loss": 1.501,
      "step": 571
    },
    {
      "epoch": 0.17,
      "learning_rate": 7.609044234303286e-05,
      "loss": 1.3566,
      "step": 572
    },
    {
      "epoch": 0.17,
      "learning_rate": 7.607403119710453e-05,
      "loss": 1.4112,
      "step": 573
    },
    {
      "epoch": 0.17,
      "learning_rate": 7.605758745614046e-05,
      "loss": 1.3702,
      "step": 574
    },
    {
      "epoch": 0.17,
      "learning_rate": 7.604111113499856e-05,
      "loss": 1.47,
      "step": 575
    },
    {
      "epoch": 0.17,
      "learning_rate": 7.602460224856615e-05,
      "loss": 1.3678,
      "step": 576
    },
    {
      "epoch": 0.17,
      "learning_rate": 7.600806081176e-05,
      "loss": 1.3715,
      "step": 577
    },
    {
      "epoch": 0.17,
      "learning_rate": 7.599148683952628e-05,
      "loss": 1.4038,
      "step": 578
    },
    {
      "epoch": 0.17,
      "learning_rate": 7.597488034684058e-05,
      "loss": 1.3995,
      "step": 579
    },
    {
      "epoch": 0.17,
      "learning_rate": 7.595824134870782e-05,
      "loss": 1.4314,
      "step": 580
    },
    {
      "epoch": 0.17,
      "learning_rate": 7.594156986016235e-05,
      "loss": 1.4516,
      "step": 581
    },
    {
      "epoch": 0.17,
      "learning_rate": 7.592486589626784e-05,
      "loss": 1.4193,
      "step": 582
    },
    {
      "epoch": 0.17,
      "learning_rate": 7.590812947211733e-05,
      "loss": 1.3927,
      "step": 583
    },
    {
      "epoch": 0.17,
      "learning_rate": 7.589136060283318e-05,
      "loss": 1.4211,
      "step": 584
    },
    {
      "epoch": 0.17,
      "learning_rate": 7.587455930356704e-05,
      "loss": 1.4091,
      "step": 585
    },
    {
      "epoch": 0.17,
      "learning_rate": 7.585772558949989e-05,
      "loss": 1.485,
      "step": 586
    },
    {
      "epoch": 0.17,
      "learning_rate": 7.584085947584202e-05,
      "loss": 1.4084,
      "step": 587
    },
    {
      "epoch": 0.17,
      "learning_rate": 7.582396097783294e-05,
      "loss": 1.4219,
      "step": 588
    },
    {
      "epoch": 0.17,
      "learning_rate": 7.580703011074147e-05,
      "loss": 1.4331,
      "step": 589
    },
    {
      "epoch": 0.17,
      "learning_rate": 7.579006688986563e-05,
      "loss": 1.4127,
      "step": 590
    },
    {
      "epoch": 0.17,
      "learning_rate": 7.57730713305327e-05,
      "loss": 1.4424,
      "step": 591
    },
    {
      "epoch": 0.17,
      "learning_rate": 7.575604344809921e-05,
      "loss": 1.4335,
      "step": 592
    },
    {
      "epoch": 0.17,
      "learning_rate": 7.573898325795086e-05,
      "loss": 1.4293,
      "step": 593
    },
    {
      "epoch": 0.17,
      "learning_rate": 7.572189077550253e-05,
      "loss": 1.4775,
      "step": 594
    },
    {
      "epoch": 0.17,
      "learning_rate": 7.57047660161983e-05,
      "loss": 1.3645,
      "step": 595
    },
    {
      "epoch": 0.17,
      "learning_rate": 7.568760899551142e-05,
      "loss": 1.4359,
      "step": 596
    },
    {
      "epoch": 0.18,
      "learning_rate": 7.567041972894428e-05,
      "loss": 1.4445,
      "step": 597
    },
    {
      "epoch": 0.18,
      "learning_rate": 7.565319823202838e-05,
      "loss": 1.4056,
      "step": 598
    },
    {
      "epoch": 0.18,
      "learning_rate": 7.563594452032441e-05,
      "loss": 1.399,
      "step": 599
    },
    {
      "epoch": 0.18,
      "learning_rate": 7.561865860942209e-05,
      "loss": 1.3824,
      "step": 600
    },
    {
      "epoch": 0.18,
      "eval_loss": 0.14922283589839935,
      "eval_runtime": 24.4837,
      "eval_samples_per_second": 81.687,
      "eval_steps_per_second": 0.653,
      "step": 600
    },
    {
      "epoch": 0.18,
      "learning_rate": 7.560134051494031e-05,
      "loss": 1.4521,
      "step": 601
    },
    {
      "epoch": 0.18,
      "learning_rate": 7.558399025252695e-05,
      "loss": 1.4126,
      "step": 602
    },
    {
      "epoch": 0.18,
      "learning_rate": 7.556660783785904e-05,
      "loss": 1.4326,
      "step": 603
    },
    {
      "epoch": 0.18,
      "learning_rate": 7.554919328664262e-05,
      "loss": 1.3813,
      "step": 604
    },
    {
      "epoch": 0.18,
      "learning_rate": 7.553174661461276e-05,
      "loss": 1.4193,
      "step": 605
    },
    {
      "epoch": 0.18,
      "learning_rate": 7.551426783753359e-05,
      "loss": 1.3702,
      "step": 606
    },
    {
      "epoch": 0.18,
      "learning_rate": 7.549675697119822e-05,
      "loss": 1.4451,
      "step": 607
    },
    {
      "epoch": 0.18,
      "learning_rate": 7.547921403142874e-05,
      "loss": 1.4015,
      "step": 608
    },
    {
      "epoch": 0.18,
      "learning_rate": 7.546163903407627e-05,
      "loss": 1.4294,
      "step": 609
    },
    {
      "epoch": 0.18,
      "learning_rate": 7.544403199502084e-05,
      "loss": 1.3551,
      "step": 610
    },
    {
      "epoch": 0.18,
      "learning_rate": 7.542639293017148e-05,
      "loss": 1.389,
      "step": 611
    },
    {
      "epoch": 0.18,
      "learning_rate": 7.540872185546614e-05,
      "loss": 1.4239,
      "step": 612
    },
    {
      "epoch": 0.18,
      "learning_rate": 7.539101878687165e-05,
      "loss": 1.4485,
      "step": 613
    },
    {
      "epoch": 0.18,
      "learning_rate": 7.537328374038383e-05,
      "loss": 1.4248,
      "step": 614
    },
    {
      "epoch": 0.18,
      "learning_rate": 7.535551673202734e-05,
      "loss": 1.3853,
      "step": 615
    },
    {
      "epoch": 0.18,
      "learning_rate": 7.533771777785572e-05,
      "loss": 1.4486,
      "step": 616
    },
    {
      "epoch": 0.18,
      "learning_rate": 7.531988689395137e-05,
      "loss": 1.3802,
      "step": 617
    },
    {
      "epoch": 0.18,
      "learning_rate": 7.53020240964256e-05,
      "loss": 1.4025,
      "step": 618
    },
    {
      "epoch": 0.18,
      "learning_rate": 7.528412940141847e-05,
      "loss": 1.4193,
      "step": 619
    },
    {
      "epoch": 0.18,
      "learning_rate": 7.526620282509891e-05,
      "loss": 1.4409,
      "step": 620
    },
    {
      "epoch": 0.18,
      "learning_rate": 7.524824438366468e-05,
      "loss": 1.4012,
      "step": 621
    },
    {
      "epoch": 0.18,
      "learning_rate": 7.523025409334227e-05,
      "loss": 1.3932,
      "step": 622
    },
    {
      "epoch": 0.18,
      "learning_rate": 7.5212231970387e-05,
      "loss": 1.4654,
      "step": 623
    },
    {
      "epoch": 0.18,
      "learning_rate": 7.51941780310829e-05,
      "loss": 1.4722,
      "step": 624
    },
    {
      "epoch": 0.18,
      "learning_rate": 7.517609229174281e-05,
      "loss": 1.4342,
      "step": 625
    },
    {
      "epoch": 0.18,
      "learning_rate": 7.515797476870827e-05,
      "loss": 1.3967,
      "step": 626
    },
    {
      "epoch": 0.18,
      "learning_rate": 7.513982547834952e-05,
      "loss": 1.4339,
      "step": 627
    },
    {
      "epoch": 0.18,
      "learning_rate": 7.512164443706555e-05,
      "loss": 1.4651,
      "step": 628
    },
    {
      "epoch": 0.18,
      "learning_rate": 7.510343166128399e-05,
      "loss": 1.4477,
      "step": 629
    },
    {
      "epoch": 0.18,
      "learning_rate": 7.508518716746118e-05,
      "loss": 1.4284,
      "step": 630
    },
    {
      "epoch": 0.19,
      "learning_rate": 7.506691097208211e-05,
      "loss": 1.4734,
      "step": 631
    },
    {
      "epoch": 0.19,
      "learning_rate": 7.504860309166042e-05,
      "loss": 1.3807,
      "step": 632
    },
    {
      "epoch": 0.19,
      "learning_rate": 7.503026354273834e-05,
      "loss": 1.414,
      "step": 633
    },
    {
      "epoch": 0.19,
      "learning_rate": 7.501189234188677e-05,
      "loss": 1.398,
      "step": 634
    },
    {
      "epoch": 0.19,
      "learning_rate": 7.499348950570518e-05,
      "loss": 1.4523,
      "step": 635
    },
    {
      "epoch": 0.19,
      "learning_rate": 7.497505505082161e-05,
      "loss": 1.3659,
      "step": 636
    },
    {
      "epoch": 0.19,
      "learning_rate": 7.495658899389271e-05,
      "loss": 1.4372,
      "step": 637
    },
    {
      "epoch": 0.19,
      "learning_rate": 7.493809135160367e-05,
      "loss": 1.381,
      "step": 638
    },
    {
      "epoch": 0.19,
      "learning_rate": 7.491956214066818e-05,
      "loss": 1.4524,
      "step": 639
    },
    {
      "epoch": 0.19,
      "learning_rate": 7.490100137782851e-05,
      "loss": 1.3724,
      "step": 640
    },
    {
      "epoch": 0.19,
      "learning_rate": 7.48824090798554e-05,
      "loss": 1.4382,
      "step": 641
    },
    {
      "epoch": 0.19,
      "learning_rate": 7.486378526354812e-05,
      "loss": 1.4203,
      "step": 642
    },
    {
      "epoch": 0.19,
      "learning_rate": 7.484512994573438e-05,
      "loss": 1.3743,
      "step": 643
    },
    {
      "epoch": 0.19,
      "learning_rate": 7.482644314327036e-05,
      "loss": 1.419,
      "step": 644
    },
    {
      "epoch": 0.19,
      "learning_rate": 7.480772487304072e-05,
      "loss": 1.4332,
      "step": 645
    },
    {
      "epoch": 0.19,
      "learning_rate": 7.478897515195852e-05,
      "loss": 1.4577,
      "step": 646
    },
    {
      "epoch": 0.19,
      "learning_rate": 7.477019399696528e-05,
      "loss": 1.3823,
      "step": 647
    },
    {
      "epoch": 0.19,
      "learning_rate": 7.475138142503083e-05,
      "loss": 1.4017,
      "step": 648
    },
    {
      "epoch": 0.19,
      "learning_rate": 7.473253745315352e-05,
      "loss": 1.4017,
      "step": 649
    },
    {
      "epoch": 0.19,
      "learning_rate": 7.471366209835994e-05,
      "loss": 1.4157,
      "step": 650
    },
    {
      "epoch": 0.19,
      "learning_rate": 7.469475537770513e-05,
      "loss": 1.373,
      "step": 651
    },
    {
      "epoch": 0.19,
      "learning_rate": 7.467581730827244e-05,
      "loss": 1.4197,
      "step": 652
    },
    {
      "epoch": 0.19,
      "learning_rate": 7.465684790717354e-05,
      "loss": 1.3704,
      "step": 653
    },
    {
      "epoch": 0.19,
      "learning_rate": 7.46378471915484e-05,
      "loss": 1.3811,
      "step": 654
    },
    {
      "epoch": 0.19,
      "learning_rate": 7.46188151785653e-05,
      "loss": 1.3896,
      "step": 655
    },
    {
      "epoch": 0.19,
      "learning_rate": 7.459975188542082e-05,
      "loss": 1.38,
      "step": 656
    },
    {
      "epoch": 0.19,
      "learning_rate": 7.458065732933976e-05,
      "loss": 1.4459,
      "step": 657
    },
    {
      "epoch": 0.19,
      "learning_rate": 7.45615315275752e-05,
      "loss": 1.4214,
      "step": 658
    },
    {
      "epoch": 0.19,
      "learning_rate": 7.454237449740843e-05,
      "loss": 1.3855,
      "step": 659
    },
    {
      "epoch": 0.19,
      "learning_rate": 7.452318625614898e-05,
      "loss": 1.3763,
      "step": 660
    },
    {
      "epoch": 0.19,
      "learning_rate": 7.450396682113453e-05,
      "loss": 1.4183,
      "step": 661
    },
    {
      "epoch": 0.19,
      "learning_rate": 7.448471620973106e-05,
      "loss": 1.4542,
      "step": 662
    },
    {
      "epoch": 0.19,
      "learning_rate": 7.446543443933258e-05,
      "loss": 1.3802,
      "step": 663
    },
    {
      "epoch": 0.19,
      "learning_rate": 7.444612152736135e-05,
      "loss": 1.4174,
      "step": 664
    },
    {
      "epoch": 0.2,
      "learning_rate": 7.44267774912677e-05,
      "loss": 1.4064,
      "step": 665
    },
    {
      "epoch": 0.2,
      "learning_rate": 7.440740234853017e-05,
      "loss": 1.4343,
      "step": 666
    },
    {
      "epoch": 0.2,
      "learning_rate": 7.438799611665531e-05,
      "loss": 1.3878,
      "step": 667
    },
    {
      "epoch": 0.2,
      "learning_rate": 7.436855881317784e-05,
      "loss": 1.3862,
      "step": 668
    },
    {
      "epoch": 0.2,
      "learning_rate": 7.43490904556605e-05,
      "loss": 1.4194,
      "step": 669
    },
    {
      "epoch": 0.2,
      "learning_rate": 7.432959106169412e-05,
      "loss": 1.3693,
      "step": 670
    },
    {
      "epoch": 0.2,
      "learning_rate": 7.431006064889755e-05,
      "loss": 1.3839,
      "step": 671
    },
    {
      "epoch": 0.2,
      "learning_rate": 7.42904992349177e-05,
      "loss": 1.4427,
      "step": 672
    },
    {
      "epoch": 0.2,
      "learning_rate": 7.427090683742947e-05,
      "loss": 1.4084,
      "step": 673
    },
    {
      "epoch": 0.2,
      "learning_rate": 7.425128347413574e-05,
      "loss": 1.4009,
      "step": 674
    },
    {
      "epoch": 0.2,
      "learning_rate": 7.42316291627674e-05,
      "loss": 1.4115,
      "step": 675
    },
    {
      "epoch": 0.2,
      "learning_rate": 7.42119439210833e-05,
      "loss": 1.4218,
      "step": 676
    },
    {
      "epoch": 0.2,
      "learning_rate": 7.419222776687022e-05,
      "loss": 1.4334,
      "step": 677
    },
    {
      "epoch": 0.2,
      "learning_rate": 7.417248071794288e-05,
      "loss": 1.4145,
      "step": 678
    },
    {
      "epoch": 0.2,
      "learning_rate": 7.415270279214392e-05,
      "loss": 1.3816,
      "step": 679
    },
    {
      "epoch": 0.2,
      "learning_rate": 7.413289400734385e-05,
      "loss": 1.413,
      "step": 680
    },
    {
      "epoch": 0.2,
      "learning_rate": 7.411305438144111e-05,
      "loss": 1.3879,
      "step": 681
    },
    {
      "epoch": 0.2,
      "learning_rate": 7.409318393236199e-05,
      "loss": 1.3713,
      "step": 682
    },
    {
      "epoch": 0.2,
      "learning_rate": 7.40732826780606e-05,
      "loss": 1.4331,
      "step": 683
    },
    {
      "epoch": 0.2,
      "learning_rate": 7.405335063651893e-05,
      "loss": 1.4204,
      "step": 684
    },
    {
      "epoch": 0.2,
      "learning_rate": 7.403338782574674e-05,
      "loss": 1.4173,
      "step": 685
    },
    {
      "epoch": 0.2,
      "learning_rate": 7.401339426378165e-05,
      "loss": 1.3759,
      "step": 686
    },
    {
      "epoch": 0.2,
      "learning_rate": 7.399336996868902e-05,
      "loss": 1.4106,
      "step": 687
    },
    {
      "epoch": 0.2,
      "learning_rate": 7.397331495856199e-05,
      "loss": 1.3805,
      "step": 688
    },
    {
      "epoch": 0.2,
      "learning_rate": 7.395322925152145e-05,
      "loss": 1.4235,
      "step": 689
    },
    {
      "epoch": 0.2,
      "learning_rate": 7.393311286571607e-05,
      "loss": 1.3466,
      "step": 690
    },
    {
      "epoch": 0.2,
      "learning_rate": 7.391296581932216e-05,
      "loss": 1.4122,
      "step": 691
    },
    {
      "epoch": 0.2,
      "learning_rate": 7.389278813054378e-05,
      "loss": 1.3957,
      "step": 692
    },
    {
      "epoch": 0.2,
      "learning_rate": 7.38725798176127e-05,
      "loss": 1.3724,
      "step": 693
    },
    {
      "epoch": 0.2,
      "learning_rate": 7.385234089878831e-05,
      "loss": 1.4261,
      "step": 694
    },
    {
      "epoch": 0.2,
      "learning_rate": 7.383207139235769e-05,
      "loss": 1.4567,
      "step": 695
    },
    {
      "epoch": 0.2,
      "learning_rate": 7.381177131663555e-05,
      "loss": 1.4201,
      "step": 696
    },
    {
      "epoch": 0.2,
      "learning_rate": 7.37914406899642e-05,
      "loss": 1.4183,
      "step": 697
    },
    {
      "epoch": 0.2,
      "learning_rate": 7.37710795307136e-05,
      "loss": 1.3535,
      "step": 698
    },
    {
      "epoch": 0.21,
      "learning_rate": 7.375068785728124e-05,
      "loss": 1.4394,
      "step": 699
    },
    {
      "epoch": 0.21,
      "learning_rate": 7.373026568809224e-05,
      "loss": 1.4229,
      "step": 700
    },
    {
      "epoch": 0.21,
      "eval_loss": 0.14763599634170532,
      "eval_runtime": 25.4163,
      "eval_samples_per_second": 78.69,
      "eval_steps_per_second": 0.63,
      "step": 700
    },
    {
      "epoch": 0.21,
      "learning_rate": 7.370981304159921e-05,
      "loss": 1.3756,
      "step": 701
    },
    {
      "epoch": 0.21,
      "learning_rate": 7.368932993628237e-05,
      "loss": 1.4517,
      "step": 702
    },
    {
      "epoch": 0.21,
      "learning_rate": 7.36688163906494e-05,
      "loss": 1.3883,
      "step": 703
    },
    {
      "epoch": 0.21,
      "learning_rate": 7.364827242323554e-05,
      "loss": 1.4049,
      "step": 704
    },
    {
      "epoch": 0.21,
      "learning_rate": 7.362769805260345e-05,
      "loss": 1.3641,
      "step": 705
    },
    {
      "epoch": 0.21,
      "learning_rate": 7.360709329734332e-05,
      "loss": 1.4187,
      "step": 706
    },
    {
      "epoch": 0.21,
      "learning_rate": 7.358645817607279e-05,
      "loss": 1.3369,
      "step": 707
    },
    {
      "epoch": 0.21,
      "learning_rate": 7.356579270743689e-05,
      "loss": 1.3943,
      "step": 708
    },
    {
      "epoch": 0.21,
      "learning_rate": 7.354509691010813e-05,
      "loss": 1.3244,
      "step": 709
    },
    {
      "epoch": 0.21,
      "learning_rate": 7.352437080278637e-05,
      "loss": 1.429,
      "step": 710
    },
    {
      "epoch": 0.21,
      "learning_rate": 7.35036144041989e-05,
      "loss": 1.3726,
      "step": 711
    },
    {
      "epoch": 0.21,
      "learning_rate": 7.348282773310035e-05,
      "loss": 1.4076,
      "step": 712
    },
    {
      "epoch": 0.21,
      "learning_rate": 7.346201080827272e-05,
      "loss": 1.3623,
      "step": 713
    },
    {
      "epoch": 0.21,
      "learning_rate": 7.344116364852534e-05,
      "loss": 1.4287,
      "step": 714
    },
    {
      "epoch": 0.21,
      "learning_rate": 7.342028627269485e-05,
      "loss": 1.369,
      "step": 715
    },
    {
      "epoch": 0.21,
      "learning_rate": 7.339937869964524e-05,
      "loss": 1.3475,
      "step": 716
    },
    {
      "epoch": 0.21,
      "learning_rate": 7.337844094826771e-05,
      "loss": 1.4277,
      "step": 717
    },
    {
      "epoch": 0.21,
      "learning_rate": 7.335747303748079e-05,
      "loss": 1.4137,
      "step": 718
    },
    {
      "epoch": 0.21,
      "learning_rate": 7.333647498623022e-05,
      "loss": 1.4134,
      "step": 719
    },
    {
      "epoch": 0.21,
      "learning_rate": 7.331544681348898e-05,
      "loss": 1.3573,
      "step": 720
    },
    {
      "epoch": 0.21,
      "learning_rate": 7.32943885382573e-05,
      "loss": 1.4284,
      "step": 721
    },
    {
      "epoch": 0.21,
      "learning_rate": 7.327330017956259e-05,
      "loss": 1.4261,
      "step": 722
    },
    {
      "epoch": 0.21,
      "learning_rate": 7.325218175645942e-05,
      "loss": 1.4344,
      "step": 723
    },
    {
      "epoch": 0.21,
      "learning_rate": 7.323103328802954e-05,
      "loss": 1.3898,
      "step": 724
    },
    {
      "epoch": 0.21,
      "learning_rate": 7.320985479338187e-05,
      "loss": 1.4311,
      "step": 725
    },
    {
      "epoch": 0.21,
      "learning_rate": 7.318864629165242e-05,
      "loss": 1.3822,
      "step": 726
    },
    {
      "epoch": 0.21,
      "learning_rate": 7.316740780200434e-05,
      "loss": 1.3852,
      "step": 727
    },
    {
      "epoch": 0.21,
      "learning_rate": 7.314613934362788e-05,
      "loss": 1.3979,
      "step": 728
    },
    {
      "epoch": 0.21,
      "learning_rate": 7.312484093574035e-05,
      "loss": 1.3993,
      "step": 729
    },
    {
      "epoch": 0.21,
      "learning_rate": 7.310351259758614e-05,
      "loss": 1.3642,
      "step": 730
    },
    {
      "epoch": 0.21,
      "learning_rate": 7.308215434843668e-05,
      "loss": 1.3717,
      "step": 731
    },
    {
      "epoch": 0.21,
      "learning_rate": 7.30607662075904e-05,
      "loss": 1.3792,
      "step": 732
    },
    {
      "epoch": 0.22,
      "learning_rate": 7.303934819437277e-05,
      "loss": 1.4255,
      "step": 733
    },
    {
      "epoch": 0.22,
      "learning_rate": 7.301790032813622e-05,
      "loss": 1.3832,
      "step": 734
    },
    {
      "epoch": 0.22,
      "learning_rate": 7.299642262826019e-05,
      "loss": 1.4055,
      "step": 735
    },
    {
      "epoch": 0.22,
      "learning_rate": 7.297491511415109e-05,
      "loss": 1.384,
      "step": 736
    },
    {
      "epoch": 0.22,
      "learning_rate": 7.29533778052422e-05,
      "loss": 1.4092,
      "step": 737
    },
    {
      "epoch": 0.22,
      "learning_rate": 7.293181072099377e-05,
      "loss": 1.4142,
      "step": 738
    },
    {
      "epoch": 0.22,
      "learning_rate": 7.291021388089295e-05,
      "loss": 1.3888,
      "step": 739
    },
    {
      "epoch": 0.22,
      "learning_rate": 7.288858730445378e-05,
      "loss": 1.3622,
      "step": 740
    },
    {
      "epoch": 0.22,
      "learning_rate": 7.286693101121714e-05,
      "loss": 1.3975,
      "step": 741
    },
    {
      "epoch": 0.22,
      "learning_rate": 7.28452450207508e-05,
      "loss": 1.3726,
      "step": 742
    },
    {
      "epoch": 0.22,
      "learning_rate": 7.282352935264934e-05,
      "loss": 1.419,
      "step": 743
    },
    {
      "epoch": 0.22,
      "learning_rate": 7.280178402653415e-05,
      "loss": 1.4451,
      "step": 744
    },
    {
      "epoch": 0.22,
      "learning_rate": 7.278000906205343e-05,
      "loss": 1.393,
      "step": 745
    },
    {
      "epoch": 0.22,
      "learning_rate": 7.275820447888216e-05,
      "loss": 1.3612,
      "step": 746
    },
    {
      "epoch": 0.22,
      "learning_rate": 7.273637029672208e-05,
      "loss": 1.341,
      "step": 747
    },
    {
      "epoch": 0.22,
      "learning_rate": 7.271450653530167e-05,
      "loss": 1.4138,
      "step": 748
    },
    {
      "epoch": 0.22,
      "learning_rate": 7.269261321437616e-05,
      "loss": 1.3845,
      "step": 749
    },
    {
      "epoch": 0.22,
      "learning_rate": 7.267069035372742e-05,
      "loss": 1.3907,
      "step": 750
    },
    {
      "epoch": 0.22,
      "learning_rate": 7.264873797316412e-05,
      "loss": 1.4135,
      "step": 751
    },
    {
      "epoch": 0.22,
      "learning_rate": 7.262675609252151e-05,
      "loss": 1.3303,
      "step": 752
    },
    {
      "epoch": 0.22,
      "learning_rate": 7.260474473166154e-05,
      "loss": 1.4013,
      "step": 753
    },
    {
      "epoch": 0.22,
      "learning_rate": 7.258270391047279e-05,
      "loss": 1.3951,
      "step": 754
    },
    {
      "epoch": 0.22,
      "learning_rate": 7.256063364887043e-05,
      "loss": 1.3926,
      "step": 755
    },
    {
      "epoch": 0.22,
      "learning_rate": 7.253853396679628e-05,
      "loss": 1.329,
      "step": 756
    },
    {
      "epoch": 0.22,
      "learning_rate": 7.25164048842187e-05,
      "loss": 1.394,
      "step": 757
    },
    {
      "epoch": 0.22,
      "learning_rate": 7.249424642113266e-05,
      "loss": 1.3518,
      "step": 758
    },
    {
      "epoch": 0.22,
      "learning_rate": 7.247205859755962e-05,
      "loss": 1.3932,
      "step": 759
    },
    {
      "epoch": 0.22,
      "learning_rate": 7.244984143354763e-05,
      "loss": 1.3802,
      "step": 760
    },
    {
      "epoch": 0.22,
      "learning_rate": 7.24275949491712e-05,
      "loss": 1.3144,
      "step": 761
    },
    {
      "epoch": 0.22,
      "learning_rate": 7.240531916453136e-05,
      "loss": 1.4218,
      "step": 762
    },
    {
      "epoch": 0.22,
      "learning_rate": 7.238301409975561e-05,
      "loss": 1.3826,
      "step": 763
    },
    {
      "epoch": 0.22,
      "learning_rate": 7.236067977499791e-05,
      "loss": 1.4198,
      "step": 764
    },
    {
      "epoch": 0.22,
      "learning_rate": 7.233831621043864e-05,
      "loss": 1.3776,
      "step": 765
    },
    {
      "epoch": 0.22,
      "learning_rate": 7.231592342628462e-05,
      "loss": 1.3987,
      "step": 766
    },
    {
      "epoch": 0.23,
      "learning_rate": 7.229350144276906e-05,
      "loss": 1.382,
      "step": 767
    },
    {
      "epoch": 0.23,
      "learning_rate": 7.227105028015156e-05,
      "loss": 1.3476,
      "step": 768
    },
    {
      "epoch": 0.23,
      "learning_rate": 7.224856995871808e-05,
      "loss": 1.3935,
      "step": 769
    },
    {
      "epoch": 0.23,
      "learning_rate": 7.222606049878092e-05,
      "loss": 1.4203,
      "step": 770
    },
    {
      "epoch": 0.23,
      "learning_rate": 7.220352192067875e-05,
      "loss": 1.4174,
      "step": 771
    },
    {
      "epoch": 0.23,
      "learning_rate": 7.218095424477649e-05,
      "loss": 1.4153,
      "step": 772
    },
    {
      "epoch": 0.23,
      "learning_rate": 7.21583574914654e-05,
      "loss": 1.4617,
      "step": 773
    },
    {
      "epoch": 0.23,
      "learning_rate": 7.213573168116295e-05,
      "loss": 1.4448,
      "step": 774
    },
    {
      "epoch": 0.23,
      "learning_rate": 7.211307683431295e-05,
      "loss": 1.3778,
      "step": 775
    },
    {
      "epoch": 0.23,
      "learning_rate": 7.209039297138541e-05,
      "loss": 1.4113,
      "step": 776
    },
    {
      "epoch": 0.23,
      "learning_rate": 7.206768011287651e-05,
      "loss": 1.3898,
      "step": 777
    },
    {
      "epoch": 0.23,
      "learning_rate": 7.204493827930869e-05,
      "loss": 1.4325,
      "step": 778
    },
    {
      "epoch": 0.23,
      "learning_rate": 7.202216749123057e-05,
      "loss": 1.3944,
      "step": 779
    },
    {
      "epoch": 0.23,
      "learning_rate": 7.199936776921688e-05,
      "loss": 1.3671,
      "step": 780
    },
    {
      "epoch": 0.23,
      "learning_rate": 7.197653913386855e-05,
      "loss": 1.4266,
      "step": 781
    },
    {
      "epoch": 0.23,
      "learning_rate": 7.19536816058126e-05,
      "loss": 1.3937,
      "step": 782
    },
    {
      "epoch": 0.23,
      "learning_rate": 7.193079520570217e-05,
      "loss": 1.4037,
      "step": 783
    },
    {
      "epoch": 0.23,
      "learning_rate": 7.19078799542165e-05,
      "loss": 1.3865,
      "step": 784
    },
    {
      "epoch": 0.23,
      "learning_rate": 7.188493587206087e-05,
      "loss": 1.4308,
      "step": 785
    },
    {
      "epoch": 0.23,
      "learning_rate": 7.186196297996661e-05,
      "loss": 1.3721,
      "step": 786
    },
    {
      "epoch": 0.23,
      "learning_rate": 7.183896129869112e-05,
      "loss": 1.4385,
      "step": 787
    },
    {
      "epoch": 0.23,
      "learning_rate": 7.18159308490178e-05,
      "loss": 1.3767,
      "step": 788
    },
    {
      "epoch": 0.23,
      "learning_rate": 7.179287165175599e-05,
      "loss": 1.4287,
      "step": 789
    },
    {
      "epoch": 0.23,
      "learning_rate": 7.17697837277411e-05,
      "loss": 1.3299,
      "step": 790
    },
    {
      "epoch": 0.23,
      "learning_rate": 7.174666709783438e-05,
      "loss": 1.419,
      "step": 791
    },
    {
      "epoch": 0.23,
      "learning_rate": 7.172352178292315e-05,
      "loss": 1.3639,
      "step": 792
    },
    {
      "epoch": 0.23,
      "learning_rate": 7.170034780392055e-05,
      "loss": 1.3968,
      "step": 793
    },
    {
      "epoch": 0.23,
      "learning_rate": 7.167714518176564e-05,
      "loss": 1.4081,
      "step": 794
    },
    {
      "epoch": 0.23,
      "learning_rate": 7.16539139374234e-05,
      "loss": 1.3629,
      "step": 795
    },
    {
      "epoch": 0.23,
      "learning_rate": 7.163065409188461e-05,
      "loss": 1.4389,
      "step": 796
    },
    {
      "epoch": 0.23,
      "learning_rate": 7.160736566616595e-05,
      "loss": 1.3337,
      "step": 797
    },
    {
      "epoch": 0.23,
      "learning_rate": 7.158404868130988e-05,
      "loss": 1.3868,
      "step": 798
    },
    {
      "epoch": 0.23,
      "learning_rate": 7.156070315838468e-05,
      "loss": 1.4085,
      "step": 799
    },
    {
      "epoch": 0.23,
      "learning_rate": 7.153732911848446e-05,
      "loss": 1.4004,
      "step": 800
    },
    {
      "epoch": 0.23,
      "eval_loss": 0.13538677990436554,
      "eval_runtime": 25.0737,
      "eval_samples_per_second": 79.765,
      "eval_steps_per_second": 0.638,
      "step": 800
    },
    {
      "epoch": 0.24,
      "learning_rate": 7.151392658272902e-05,
      "loss": 1.3815,
      "step": 801
    },
    {
      "epoch": 0.24,
      "learning_rate": 7.149049557226397e-05,
      "loss": 1.3802,
      "step": 802
    },
    {
      "epoch": 0.24,
      "learning_rate": 7.14670361082606e-05,
      "loss": 1.3954,
      "step": 803
    },
    {
      "epoch": 0.24,
      "learning_rate": 7.144354821191595e-05,
      "loss": 1.417,
      "step": 804
    },
    {
      "epoch": 0.24,
      "learning_rate": 7.142003190445272e-05,
      "loss": 1.4042,
      "step": 805
    },
    {
      "epoch": 0.24,
      "learning_rate": 7.139648720711931e-05,
      "loss": 1.3799,
      "step": 806
    },
    {
      "epoch": 0.24,
      "learning_rate": 7.137291414118974e-05,
      "loss": 1.3941,
      "step": 807
    },
    {
      "epoch": 0.24,
      "learning_rate": 7.134931272796368e-05,
      "loss": 1.4179,
      "step": 808
    },
    {
      "epoch": 0.24,
      "learning_rate": 7.132568298876643e-05,
      "loss": 1.3838,
      "step": 809
    },
    {
      "epoch": 0.24,
      "learning_rate": 7.130202494494886e-05,
      "loss": 1.3578,
      "step": 810
    },
    {
      "epoch": 0.24,
      "learning_rate": 7.127833861788741e-05,
      "loss": 1.4084,
      "step": 811
    },
    {
      "epoch": 0.24,
      "learning_rate": 7.12546240289841e-05,
      "loss": 1.3892,
      "step": 812
    },
    {
      "epoch": 0.24,
      "learning_rate": 7.123088119966647e-05,
      "loss": 1.3437,
      "step": 813
    },
    {
      "epoch": 0.24,
      "learning_rate": 7.120711015138756e-05,
      "loss": 1.3919,
      "step": 814
    },
    {
      "epoch": 0.24,
      "learning_rate": 7.118331090562598e-05,
      "loss": 1.3928,
      "step": 815
    },
    {
      "epoch": 0.24,
      "learning_rate": 7.11594834838857e-05,
      "loss": 1.3909,
      "step": 816
    },
    {
      "epoch": 0.24,
      "learning_rate": 7.113562790769626e-05,
      "loss": 1.3852,
      "step": 817
    },
    {
      "epoch": 0.24,
      "learning_rate": 7.11117441986126e-05,
      "loss": 1.409,
      "step": 818
    },
    {
      "epoch": 0.24,
      "learning_rate": 7.108783237821503e-05,
      "loss": 1.4075,
      "step": 819
    },
    {
      "epoch": 0.24,
      "learning_rate": 7.106389246810934e-05,
      "loss": 1.3625,
      "step": 820
    },
    {
      "epoch": 0.24,
      "learning_rate": 7.103992448992665e-05,
      "loss": 1.3513,
      "step": 821
    },
    {
      "epoch": 0.24,
      "learning_rate": 7.101592846532346e-05,
      "loss": 1.4093,
      "step": 822
    },
    {
      "epoch": 0.24,
      "learning_rate": 7.099190441598161e-05,
      "loss": 1.385,
      "step": 823
    },
    {
      "epoch": 0.24,
      "learning_rate": 7.096785236360826e-05,
      "loss": 1.3637,
      "step": 824
    },
    {
      "epoch": 0.24,
      "learning_rate": 7.094377232993586e-05,
      "loss": 1.3566,
      "step": 825
    },
    {
      "epoch": 0.24,
      "learning_rate": 7.091966433672217e-05,
      "loss": 1.3794,
      "step": 826
    },
    {
      "epoch": 0.24,
      "learning_rate": 7.089552840575018e-05,
      "loss": 1.3432,
      "step": 827
    },
    {
      "epoch": 0.24,
      "learning_rate": 7.087136455882816e-05,
      "loss": 1.3394,
      "step": 828
    },
    {
      "epoch": 0.24,
      "learning_rate": 7.084717281778957e-05,
      "loss": 1.3437,
      "step": 829
    },
    {
      "epoch": 0.24,
      "learning_rate": 7.08229532044931e-05,
      "loss": 1.3786,
      "step": 830
    },
    {
      "epoch": 0.24,
      "learning_rate": 7.07987057408226e-05,
      "loss": 1.3717,
      "step": 831
    },
    {
      "epoch": 0.24,
      "learning_rate": 7.077443044868712e-05,
      "loss": 1.3552,
      "step": 832
    },
    {
      "epoch": 0.24,
      "learning_rate": 7.07501273500208e-05,
      "loss": 1.3643,
      "step": 833
    },
    {
      "epoch": 0.24,
      "learning_rate": 7.072579646678297e-05,
      "loss": 1.3818,
      "step": 834
    },
    {
      "epoch": 0.24,
      "learning_rate": 7.070143782095802e-05,
      "loss": 1.4252,
      "step": 835
    },
    {
      "epoch": 0.25,
      "learning_rate": 7.067705143455543e-05,
      "loss": 1.3912,
      "step": 836
    },
    {
      "epoch": 0.25,
      "learning_rate": 7.065263732960976e-05,
      "loss": 1.3889,
      "step": 837
    },
    {
      "epoch": 0.25,
      "learning_rate": 7.06281955281806e-05,
      "loss": 1.3984,
      "step": 838
    },
    {
      "epoch": 0.25,
      "learning_rate": 7.060372605235259e-05,
      "loss": 1.3513,
      "step": 839
    },
    {
      "epoch": 0.25,
      "learning_rate": 7.057922892423534e-05,
      "loss": 1.3649,
      "step": 840
    },
    {
      "epoch": 0.25,
      "learning_rate": 7.055470416596347e-05,
      "loss": 1.413,
      "step": 841
    },
    {
      "epoch": 0.25,
      "learning_rate": 7.053015179969656e-05,
      "loss": 1.4038,
      "step": 842
    },
    {
      "epoch": 0.25,
      "learning_rate": 7.050557184761915e-05,
      "loss": 1.381,
      "step": 843
    },
    {
      "epoch": 0.25,
      "learning_rate": 7.048096433194067e-05,
      "loss": 1.3958,
      "step": 844
    },
    {
      "epoch": 0.25,
      "learning_rate": 7.045632927489548e-05,
      "loss": 1.4419,
      "step": 845
    },
    {
      "epoch": 0.25,
      "learning_rate": 7.043166669874282e-05,
      "loss": 1.3879,
      "step": 846
    },
    {
      "epoch": 0.25,
      "learning_rate": 7.04069766257668e-05,
      "loss": 1.3386,
      "step": 847
    },
    {
      "epoch": 0.25,
      "learning_rate": 7.038225907827639e-05,
      "loss": 1.3618,
      "step": 848
    },
    {
      "epoch": 0.25,
      "learning_rate": 7.035751407860533e-05,
      "loss": 1.4196,
      "step": 849
    },
    {
      "epoch": 0.25,
      "learning_rate": 7.033274164911223e-05,
      "loss": 1.3625,
      "step": 850
    },
    {
      "epoch": 0.25,
      "learning_rate": 7.030794181218045e-05,
      "loss": 1.3966,
      "step": 851
    },
    {
      "epoch": 0.25,
      "learning_rate": 7.028311459021812e-05,
      "loss": 1.3846,
      "step": 852
    },
    {
      "epoch": 0.25,
      "learning_rate": 7.02582600056581e-05,
      "loss": 1.4307,
      "step": 853
    },
    {
      "epoch": 0.25,
      "learning_rate": 7.023337808095798e-05,
      "loss": 1.3722,
      "step": 854
    },
    {
      "epoch": 0.25,
      "learning_rate": 7.020846883860011e-05,
      "loss": 1.3685,
      "step": 855
    },
    {
      "epoch": 0.25,
      "learning_rate": 7.018353230109142e-05,
      "loss": 1.3929,
      "step": 856
    },
    {
      "epoch": 0.25,
      "learning_rate": 7.015856849096357e-05,
      "loss": 1.436,
      "step": 857
    },
    {
      "epoch": 0.25,
      "learning_rate": 7.013357743077289e-05,
      "loss": 1.3745,
      "step": 858
    },
    {
      "epoch": 0.25,
      "learning_rate": 7.010855914310023e-05,
      "loss": 1.3611,
      "step": 859
    },
    {
      "epoch": 0.25,
      "learning_rate": 7.008351365055116e-05,
      "loss": 1.3971,
      "step": 860
    },
    {
      "epoch": 0.25,
      "learning_rate": 7.00584409757557e-05,
      "loss": 1.3832,
      "step": 861
    },
    {
      "epoch": 0.25,
      "learning_rate": 7.00333411413686e-05,
      "loss": 1.3349,
      "step": 862
    },
    {
      "epoch": 0.25,
      "learning_rate": 7.000821417006898e-05,
      "loss": 1.3833,
      "step": 863
    },
    {
      "epoch": 0.25,
      "learning_rate": 6.998306008456059e-05,
      "loss": 1.4131,
      "step": 864
    },
    {
      "epoch": 0.25,
      "learning_rate": 6.995787890757162e-05,
      "loss": 1.3692,
      "step": 865
    },
    {
      "epoch": 0.25,
      "learning_rate": 6.993267066185481e-05,
      "loss": 1.3741,
      "step": 866
    },
    {
      "epoch": 0.25,
      "learning_rate": 6.990743537018726e-05,
      "loss": 1.3803,
      "step": 867
    },
    {
      "epoch": 0.25,
      "learning_rate": 6.988217305537058e-05,
      "loss": 1.4036,
      "step": 868
    },
    {
      "epoch": 0.25,
      "learning_rate": 6.985688374023079e-05,
      "loss": 1.3663,
      "step": 869
    },
    {
      "epoch": 0.26,
      "learning_rate": 6.983156744761826e-05,
      "loss": 1.3626,
      "step": 870
    },
    {
      "epoch": 0.26,
      "learning_rate": 6.98062242004078e-05,
      "loss": 1.414,
      "step": 871
    },
    {
      "epoch": 0.26,
      "learning_rate": 6.978085402149853e-05,
      "loss": 1.3694,
      "step": 872
    },
    {
      "epoch": 0.26,
      "learning_rate": 6.97554569338139e-05,
      "loss": 1.3627,
      "step": 873
    },
    {
      "epoch": 0.26,
      "learning_rate": 6.973003296030172e-05,
      "loss": 1.338,
      "step": 874
    },
    {
      "epoch": 0.26,
      "learning_rate": 6.970458212393405e-05,
      "loss": 1.3856,
      "step": 875
    },
    {
      "epoch": 0.26,
      "learning_rate": 6.967910444770725e-05,
      "loss": 1.4076,
      "step": 876
    },
    {
      "epoch": 0.26,
      "learning_rate": 6.96535999546419e-05,
      "loss": 1.3819,
      "step": 877
    },
    {
      "epoch": 0.26,
      "learning_rate": 6.962806866778284e-05,
      "loss": 1.3942,
      "step": 878
    },
    {
      "epoch": 0.26,
      "learning_rate": 6.960251061019909e-05,
      "loss": 1.3432,
      "step": 879
    },
    {
      "epoch": 0.26,
      "learning_rate": 6.95769258049839e-05,
      "loss": 1.4081,
      "step": 880
    },
    {
      "epoch": 0.26,
      "learning_rate": 6.955131427525466e-05,
      "loss": 1.3791,
      "step": 881
    },
    {
      "epoch": 0.26,
      "learning_rate": 6.95256760441529e-05,
      "loss": 1.3528,
      "step": 882
    },
    {
      "epoch": 0.26,
      "learning_rate": 6.950001113484432e-05,
      "loss": 1.4087,
      "step": 883
    },
    {
      "epoch": 0.26,
      "learning_rate": 6.947431957051865e-05,
      "loss": 1.3472,
      "step": 884
    },
    {
      "epoch": 0.26,
      "learning_rate": 6.944860137438978e-05,
      "loss": 1.3455,
      "step": 885
    },
    {
      "epoch": 0.26,
      "learning_rate": 6.942285656969563e-05,
      "loss": 1.3966,
      "step": 886
    },
    {
      "epoch": 0.26,
      "learning_rate": 6.939708517969815e-05,
      "loss": 1.3807,
      "step": 887
    },
    {
      "epoch": 0.26,
      "learning_rate": 6.937128722768333e-05,
      "loss": 1.3226,
      "step": 888
    },
    {
      "epoch": 0.26,
      "learning_rate": 6.934546273696116e-05,
      "loss": 1.3624,
      "step": 889
    },
    {
      "epoch": 0.26,
      "learning_rate": 6.931961173086561e-05,
      "loss": 1.3662,
      "step": 890
    },
    {
      "epoch": 0.26,
      "learning_rate": 6.92937342327546e-05,
      "loss": 1.3635,
      "step": 891
    },
    {
      "epoch": 0.26,
      "learning_rate": 6.926783026600997e-05,
      "loss": 1.3248,
      "step": 892
    },
    {
      "epoch": 0.26,
      "learning_rate": 6.924189985403753e-05,
      "loss": 1.3703,
      "step": 893
    },
    {
      "epoch": 0.26,
      "learning_rate": 6.921594302026693e-05,
      "loss": 1.4017,
      "step": 894
    },
    {
      "epoch": 0.26,
      "learning_rate": 6.918995978815174e-05,
      "loss": 1.3598,
      "step": 895
    },
    {
      "epoch": 0.26,
      "learning_rate": 6.916395018116932e-05,
      "loss": 1.3817,
      "step": 896
    },
    {
      "epoch": 0.26,
      "learning_rate": 6.913791422282094e-05,
      "loss": 1.3994,
      "step": 897
    },
    {
      "epoch": 0.26,
      "learning_rate": 6.91118519366316e-05,
      "loss": 1.4117,
      "step": 898
    },
    {
      "epoch": 0.26,
      "learning_rate": 6.908576334615016e-05,
      "loss": 1.4131,
      "step": 899
    },
    {
      "epoch": 0.26,
      "learning_rate": 6.90596484749492e-05,
      "loss": 1.3987,
      "step": 900
    },
    {
      "epoch": 0.26,
      "eval_loss": 0.1372963935136795,
      "eval_runtime": 25.0998,
      "eval_samples_per_second": 79.682,
      "eval_steps_per_second": 0.637,
      "step": 900
    },
    {
      "epoch": 0.26,
      "learning_rate": 6.903350734662505e-05,
      "loss": 1.3955,
      "step": 901
    },
    {
      "epoch": 0.26,
      "learning_rate": 6.90073399847978e-05,
      "loss": 1.3236,
      "step": 902
    },
    {
      "epoch": 0.26,
      "learning_rate": 6.898114641311122e-05,
      "loss": 1.4109,
      "step": 903
    },
    {
      "epoch": 0.27,
      "learning_rate": 6.895492665523274e-05,
      "loss": 1.3628,
      "step": 904
    },
    {
      "epoch": 0.27,
      "learning_rate": 6.892868073485348e-05,
      "loss": 1.4298,
      "step": 905
    },
    {
      "epoch": 0.27,
      "learning_rate": 6.89024086756882e-05,
      "loss": 1.3471,
      "step": 906
    },
    {
      "epoch": 0.27,
      "learning_rate": 6.887611050147528e-05,
      "loss": 1.3855,
      "step": 907
    },
    {
      "epoch": 0.27,
      "learning_rate": 6.884978623597665e-05,
      "loss": 1.3755,
      "step": 908
    },
    {
      "epoch": 0.27,
      "learning_rate": 6.88234359029779e-05,
      "loss": 1.4083,
      "step": 909
    },
    {
      "epoch": 0.27,
      "learning_rate": 6.879705952628809e-05,
      "loss": 1.3681,
      "step": 910
    },
    {
      "epoch": 0.27,
      "learning_rate": 6.877065712973986e-05,
      "loss": 1.3067,
      "step": 911
    },
    {
      "epoch": 0.27,
      "learning_rate": 6.874422873718934e-05,
      "loss": 1.3881,
      "step": 912
    },
    {
      "epoch": 0.27,
      "learning_rate": 6.871777437251617e-05,
      "loss": 1.3749,
      "step": 913
    },
    {
      "epoch": 0.27,
      "learning_rate": 6.869129405962342e-05,
      "loss": 1.3276,
      "step": 914
    },
    {
      "epoch": 0.27,
      "learning_rate": 6.866478782243766e-05,
      "loss": 1.3876,
      "step": 915
    },
    {
      "epoch": 0.27,
      "learning_rate": 6.863825568490885e-05,
      "loss": 1.4264,
      "step": 916
    },
    {
      "epoch": 0.27,
      "learning_rate": 6.861169767101034e-05,
      "loss": 1.3915,
      "step": 917
    },
    {
      "epoch": 0.27,
      "learning_rate": 6.858511380473887e-05,
      "loss": 1.383,
      "step": 918
    },
    {
      "epoch": 0.27,
      "learning_rate": 6.855850411011459e-05,
      "loss": 1.3867,
      "step": 919
    },
    {
      "epoch": 0.27,
      "learning_rate": 6.853186861118089e-05,
      "loss": 1.419,
      "step": 920
    },
    {
      "epoch": 0.27,
      "learning_rate": 6.850520733200456e-05,
      "loss": 1.3738,
      "step": 921
    },
    {
      "epoch": 0.27,
      "learning_rate": 6.847852029667565e-05,
      "loss": 1.3506,
      "step": 922
    },
    {
      "epoch": 0.27,
      "learning_rate": 6.845180752930749e-05,
      "loss": 1.3391,
      "step": 923
    },
    {
      "epoch": 0.27,
      "learning_rate": 6.842506905403664e-05,
      "loss": 1.3604,
      "step": 924
    },
    {
      "epoch": 0.27,
      "learning_rate": 6.839830489502293e-05,
      "loss": 1.44,
      "step": 925
    },
    {
      "epoch": 0.27,
      "learning_rate": 6.837151507644935e-05,
      "loss": 1.3542,
      "step": 926
    },
    {
      "epoch": 0.27,
      "learning_rate": 6.83446996225221e-05,
      "loss": 1.3602,
      "step": 927
    },
    {
      "epoch": 0.27,
      "learning_rate": 6.831785855747054e-05,
      "loss": 1.4267,
      "step": 928
    },
    {
      "epoch": 0.27,
      "learning_rate": 6.829099190554717e-05,
      "loss": 1.3565,
      "step": 929
    },
    {
      "epoch": 0.27,
      "learning_rate": 6.826409969102763e-05,
      "loss": 1.3914,
      "step": 930
    },
    {
      "epoch": 0.27,
      "learning_rate": 6.823718193821061e-05,
      "loss": 1.3666,
      "step": 931
    },
    {
      "epoch": 0.27,
      "learning_rate": 6.821023867141791e-05,
      "loss": 1.3441,
      "step": 932
    },
    {
      "epoch": 0.27,
      "learning_rate": 6.81832699149944e-05,
      "loss": 1.3724,
      "step": 933
    },
    {
      "epoch": 0.27,
      "learning_rate": 6.815627569330793e-05,
      "loss": 1.3677,
      "step": 934
    },
    {
      "epoch": 0.27,
      "learning_rate": 6.81292560307494e-05,
      "loss": 1.3859,
      "step": 935
    },
    {
      "epoch": 0.27,
      "learning_rate": 6.810221095173268e-05,
      "loss": 1.3665,
      "step": 936
    },
    {
      "epoch": 0.27,
      "learning_rate": 6.807514048069461e-05,
      "loss": 1.334,
      "step": 937
    },
    {
      "epoch": 0.28,
      "learning_rate": 6.804804464209498e-05,
      "loss": 1.4157,
      "step": 938
    },
    {
      "epoch": 0.28,
      "learning_rate": 6.802092346041647e-05,
      "loss": 1.383,
      "step": 939
    },
    {
      "epoch": 0.28,
      "learning_rate": 6.799377696016471e-05,
      "loss": 1.3366,
      "step": 940
    },
    {
      "epoch": 0.28,
      "learning_rate": 6.796660516586817e-05,
      "loss": 1.4138,
      "step": 941
    },
    {
      "epoch": 0.28,
      "learning_rate": 6.793940810207819e-05,
      "loss": 1.3729,
      "step": 942
    },
    {
      "epoch": 0.28,
      "learning_rate": 6.791218579336891e-05,
      "loss": 1.3978,
      "step": 943
    },
    {
      "epoch": 0.28,
      "learning_rate": 6.788493826433735e-05,
      "loss": 1.3338,
      "step": 944
    },
    {
      "epoch": 0.28,
      "learning_rate": 6.785766553960324e-05,
      "loss": 1.3656,
      "step": 945
    },
    {
      "epoch": 0.28,
      "learning_rate": 6.783036764380913e-05,
      "loss": 1.3965,
      "step": 946
    },
    {
      "epoch": 0.28,
      "learning_rate": 6.780304460162029e-05,
      "loss": 1.3199,
      "step": 947
    },
    {
      "epoch": 0.28,
      "learning_rate": 6.777569643772472e-05,
      "loss": 1.4047,
      "step": 948
    },
    {
      "epoch": 0.28,
      "learning_rate": 6.77483231768331e-05,
      "loss": 1.4125,
      "step": 949
    },
    {
      "epoch": 0.28,
      "learning_rate": 6.772092484367882e-05,
      "loss": 1.3662,
      "step": 950
    },
    {
      "epoch": 0.28,
      "learning_rate": 6.76935014630179e-05,
      "loss": 1.3421,
      "step": 951
    },
    {
      "epoch": 0.28,
      "learning_rate": 6.7666053059629e-05,
      "loss": 1.3097,
      "step": 952
    },
    {
      "epoch": 0.28,
      "learning_rate": 6.763857965831337e-05,
      "loss": 1.3788,
      "step": 953
    },
    {
      "epoch": 0.28,
      "learning_rate": 6.761108128389488e-05,
      "loss": 1.4091,
      "step": 954
    },
    {
      "epoch": 0.28,
      "learning_rate": 6.758355796121994e-05,
      "loss": 1.3692,
      "step": 955
    },
    {
      "epoch": 0.28,
      "learning_rate": 6.755600971515753e-05,
      "loss": 1.3709,
      "step": 956
    },
    {
      "epoch": 0.28,
      "learning_rate": 6.752843657059911e-05,
      "loss": 1.3784,
      "step": 957
    },
    {
      "epoch": 0.28,
      "learning_rate": 6.750083855245869e-05,
      "loss": 1.4132,
      "step": 958
    },
    {
      "epoch": 0.28,
      "learning_rate": 6.747321568567267e-05,
      "loss": 1.3661,
      "step": 959
    },
    {
      "epoch": 0.28,
      "learning_rate": 6.74455679952e-05,
      "loss": 1.3735,
      "step": 960
    },
    {
      "epoch": 0.28,
      "learning_rate": 6.7417895506022e-05,
      "loss": 1.3725,
      "step": 961
    },
    {
      "epoch": 0.28,
      "learning_rate": 6.739019824314242e-05,
      "loss": 1.3935,
      "step": 962
    },
    {
      "epoch": 0.28,
      "learning_rate": 6.736247623158738e-05,
      "loss": 1.3863,
      "step": 963
    },
    {
      "epoch": 0.28,
      "learning_rate": 6.733472949640534e-05,
      "loss": 1.3645,
      "step": 964
    },
    {
      "epoch": 0.28,
      "learning_rate": 6.730695806266719e-05,
      "loss": 1.3497,
      "step": 965
    },
    {
      "epoch": 0.28,
      "learning_rate": 6.727916195546602e-05,
      "loss": 1.3656,
      "step": 966
    },
    {
      "epoch": 0.28,
      "learning_rate": 6.72513411999173e-05,
      "loss": 1.303,
      "step": 967
    },
    {
      "epoch": 0.28,
      "learning_rate": 6.722349582115872e-05,
      "loss": 1.3516,
      "step": 968
    },
    {
      "epoch": 0.28,
      "learning_rate": 6.719562584435026e-05,
      "loss": 1.4204,
      "step": 969
    },
    {
      "epoch": 0.28,
      "learning_rate": 6.716773129467406e-05,
      "loss": 1.3299,
      "step": 970
    },
    {
      "epoch": 0.28,
      "learning_rate": 6.713981219733457e-05,
      "loss": 1.4038,
      "step": 971
    },
    {
      "epoch": 0.29,
      "learning_rate": 6.711186857755831e-05,
      "loss": 1.349,
      "step": 972
    },
    {
      "epoch": 0.29,
      "learning_rate": 6.708390046059402e-05,
      "loss": 1.3678,
      "step": 973
    },
    {
      "epoch": 0.29,
      "learning_rate": 6.705590787171256e-05,
      "loss": 1.3807,
      "step": 974
    },
    {
      "epoch": 0.29,
      "learning_rate": 6.70278908362069e-05,
      "loss": 1.383,
      "step": 975
    },
    {
      "epoch": 0.29,
      "learning_rate": 6.69998493793921e-05,
      "loss": 1.361,
      "step": 976
    },
    {
      "epoch": 0.29,
      "learning_rate": 6.697178352660528e-05,
      "loss": 1.377,
      "step": 977
    },
    {
      "epoch": 0.29,
      "learning_rate": 6.694369330320563e-05,
      "loss": 1.3704,
      "step": 978
    },
    {
      "epoch": 0.29,
      "learning_rate": 6.691557873457432e-05,
      "loss": 1.3005,
      "step": 979
    },
    {
      "epoch": 0.29,
      "learning_rate": 6.688743984611455e-05,
      "loss": 1.4302,
      "step": 980
    },
    {
      "epoch": 0.29,
      "learning_rate": 6.685927666325146e-05,
      "loss": 1.3991,
      "step": 981
    },
    {
      "epoch": 0.29,
      "learning_rate": 6.68310892114322e-05,
      "loss": 1.3321,
      "step": 982
    },
    {
      "epoch": 0.29,
      "learning_rate": 6.680287751612576e-05,
      "loss": 1.3784,
      "step": 983
    },
    {
      "epoch": 0.29,
      "learning_rate": 6.677464160282315e-05,
      "loss": 1.4145,
      "step": 984
    },
    {
      "epoch": 0.29,
      "learning_rate": 6.674638149703714e-05,
      "loss": 1.398,
      "step": 985
    },
    {
      "epoch": 0.29,
      "learning_rate": 6.671809722430247e-05,
      "loss": 1.3176,
      "step": 986
    },
    {
      "epoch": 0.29,
      "learning_rate": 6.668978881017562e-05,
      "loss": 1.3949,
      "step": 987
    },
    {
      "epoch": 0.29,
      "learning_rate": 6.666145628023495e-05,
      "loss": 1.3917,
      "step": 988
    },
    {
      "epoch": 0.29,
      "learning_rate": 6.66330996600806e-05,
      "loss": 1.321,
      "step": 989
    },
    {
      "epoch": 0.29,
      "learning_rate": 6.660471897533444e-05,
      "loss": 1.3186,
      "step": 990
    },
    {
      "epoch": 0.29,
      "learning_rate": 6.657631425164013e-05,
      "loss": 1.3523,
      "step": 991
    },
    {
      "epoch": 0.29,
      "learning_rate": 6.6547885514663e-05,
      "loss": 1.3991,
      "step": 992
    },
    {
      "epoch": 0.29,
      "learning_rate": 6.651943279009015e-05,
      "loss": 1.365,
      "step": 993
    },
    {
      "epoch": 0.29,
      "learning_rate": 6.649095610363028e-05,
      "loss": 1.3931,
      "step": 994
    },
    {
      "epoch": 0.29,
      "learning_rate": 6.646245548101379e-05,
      "loss": 1.331,
      "step": 995
    },
    {
      "epoch": 0.29,
      "learning_rate": 6.643393094799265e-05,
      "loss": 1.3461,
      "step": 996
    },
    {
      "epoch": 0.29,
      "learning_rate": 6.640538253034055e-05,
      "loss": 1.3782,
      "step": 997
    },
    {
      "epoch": 0.29,
      "learning_rate": 6.637681025385261e-05,
      "loss": 1.3777,
      "step": 998
    },
    {
      "epoch": 0.29,
      "learning_rate": 6.634821414434563e-05,
      "loss": 1.4046,
      "step": 999
    },
    {
      "epoch": 0.29,
      "learning_rate": 6.631959422765787e-05,
      "loss": 1.3816,
      "step": 1000
    },
    {
      "epoch": 0.29,
      "eval_loss": 0.1353466957807541,
      "eval_runtime": 25.1181,
      "eval_samples_per_second": 79.624,
      "eval_steps_per_second": 0.637,
      "step": 1000
    },
    {
      "epoch": 0.29,
      "learning_rate": 6.629095052964915e-05,
      "loss": 1.3722,
      "step": 1001
    },
    {
      "epoch": 0.29,
      "learning_rate": 6.626228307620074e-05,
      "loss": 1.3547,
      "step": 1002
    },
    {
      "epoch": 0.29,
      "learning_rate": 6.62335918932154e-05,
      "loss": 1.386,
      "step": 1003
    },
    {
      "epoch": 0.29,
      "learning_rate": 6.620487700661734e-05,
      "loss": 1.3744,
      "step": 1004
    },
    {
      "epoch": 0.29,
      "learning_rate": 6.617613844235214e-05,
      "loss": 1.3426,
      "step": 1005
    },
    {
      "epoch": 0.3,
      "learning_rate": 6.614737622638684e-05,
      "loss": 1.3546,
      "step": 1006
    },
    {
      "epoch": 0.3,
      "learning_rate": 6.611859038470977e-05,
      "loss": 1.3962,
      "step": 1007
    },
    {
      "epoch": 0.3,
      "learning_rate": 6.608978094333068e-05,
      "loss": 1.3445,
      "step": 1008
    },
    {
      "epoch": 0.3,
      "learning_rate": 6.606094792828062e-05,
      "loss": 1.3404,
      "step": 1009
    },
    {
      "epoch": 0.3,
      "learning_rate": 6.603209136561192e-05,
      "loss": 1.4451,
      "step": 1010
    },
    {
      "epoch": 0.3,
      "learning_rate": 6.600321128139822e-05,
      "loss": 1.3401,
      "step": 1011
    },
    {
      "epoch": 0.3,
      "learning_rate": 6.597430770173437e-05,
      "loss": 1.3405,
      "step": 1012
    },
    {
      "epoch": 0.3,
      "learning_rate": 6.594538065273652e-05,
      "loss": 1.3704,
      "step": 1013
    },
    {
      "epoch": 0.3,
      "learning_rate": 6.591643016054195e-05,
      "loss": 1.4086,
      "step": 1014
    },
    {
      "epoch": 0.3,
      "learning_rate": 6.588745625130914e-05,
      "loss": 1.3891,
      "step": 1015
    },
    {
      "epoch": 0.3,
      "learning_rate": 6.585845895121777e-05,
      "loss": 1.3765,
      "step": 1016
    },
    {
      "epoch": 0.3,
      "learning_rate": 6.582943828646862e-05,
      "loss": 1.3589,
      "step": 1017
    },
    {
      "epoch": 0.3,
      "learning_rate": 6.58003942832836e-05,
      "loss": 1.3866,
      "step": 1018
    },
    {
      "epoch": 0.3,
      "learning_rate": 6.577132696790568e-05,
      "loss": 1.3543,
      "step": 1019
    },
    {
      "epoch": 0.3,
      "learning_rate": 6.574223636659891e-05,
      "loss": 1.3239,
      "step": 1020
    },
    {
      "epoch": 0.3,
      "learning_rate": 6.571312250564838e-05,
      "loss": 1.2979,
      "step": 1021
    },
    {
      "epoch": 0.3,
      "learning_rate": 6.568398541136022e-05,
      "loss": 1.3911,
      "step": 1022
    },
    {
      "epoch": 0.3,
      "learning_rate": 6.56548251100615e-05,
      "loss": 1.333,
      "step": 1023
    },
    {
      "epoch": 0.3,
      "learning_rate": 6.562564162810031e-05,
      "loss": 1.3779,
      "step": 1024
    },
    {
      "epoch": 0.3,
      "learning_rate": 6.559643499184566e-05,
      "loss": 1.403,
      "step": 1025
    },
    {
      "epoch": 0.3,
      "learning_rate": 6.556720522768748e-05,
      "loss": 1.4145,
      "step": 1026
    },
    {
      "epoch": 0.3,
      "learning_rate": 6.553795236203661e-05,
      "loss": 1.3439,
      "step": 1027
    },
    {
      "epoch": 0.3,
      "learning_rate": 6.550867642132474e-05,
      "loss": 1.3679,
      "step": 1028
    },
    {
      "epoch": 0.3,
      "learning_rate": 6.547937743200445e-05,
      "loss": 1.4203,
      "step": 1029
    },
    {
      "epoch": 0.3,
      "learning_rate": 6.54500554205491e-05,
      "loss": 1.3372,
      "step": 1030
    },
    {
      "epoch": 0.3,
      "learning_rate": 6.542071041345288e-05,
      "loss": 1.3788,
      "step": 1031
    },
    {
      "epoch": 0.3,
      "learning_rate": 6.539134243723078e-05,
      "loss": 1.3547,
      "step": 1032
    },
    {
      "epoch": 0.3,
      "learning_rate": 6.536195151841847e-05,
      "loss": 1.4116,
      "step": 1033
    },
    {
      "epoch": 0.3,
      "learning_rate": 6.533253768357243e-05,
      "loss": 1.3164,
      "step": 1034
    },
    {
      "epoch": 0.3,
      "learning_rate": 6.530310095926979e-05,
      "loss": 1.3799,
      "step": 1035
    },
    {
      "epoch": 0.3,
      "learning_rate": 6.527364137210839e-05,
      "loss": 1.3404,
      "step": 1036
    },
    {
      "epoch": 0.3,
      "learning_rate": 6.524415894870674e-05,
      "loss": 1.3918,
      "step": 1037
    },
    {
      "epoch": 0.3,
      "learning_rate": 6.521465371570393e-05,
      "loss": 1.3864,
      "step": 1038
    },
    {
      "epoch": 0.3,
      "learning_rate": 6.518512569975972e-05,
      "loss": 1.3427,
      "step": 1039
    },
    {
      "epoch": 0.31,
      "learning_rate": 6.515557492755441e-05,
      "loss": 1.4154,
      "step": 1040
    },
    {
      "epoch": 0.31,
      "learning_rate": 6.512600142578892e-05,
      "loss": 1.4025,
      "step": 1041
    },
    {
      "epoch": 0.31,
      "learning_rate": 6.509640522118465e-05,
      "loss": 1.3397,
      "step": 1042
    },
    {
      "epoch": 0.31,
      "learning_rate": 6.506678634048353e-05,
      "loss": 1.38,
      "step": 1043
    },
    {
      "epoch": 0.31,
      "learning_rate": 6.503714481044798e-05,
      "loss": 1.41,
      "step": 1044
    },
    {
      "epoch": 0.31,
      "learning_rate": 6.500748065786088e-05,
      "loss": 1.3764,
      "step": 1045
    },
    {
      "epoch": 0.31,
      "learning_rate": 6.497779390952559e-05,
      "loss": 1.39,
      "step": 1046
    },
    {
      "epoch": 0.31,
      "learning_rate": 6.494808459226584e-05,
      "loss": 1.3411,
      "step": 1047
    },
    {
      "epoch": 0.31,
      "learning_rate": 6.491835273292574e-05,
      "loss": 1.3865,
      "step": 1048
    },
    {
      "epoch": 0.31,
      "learning_rate": 6.488859835836984e-05,
      "loss": 1.3903,
      "step": 1049
    },
    {
      "epoch": 0.31,
      "learning_rate": 6.485882149548296e-05,
      "loss": 1.422,
      "step": 1050
    },
    {
      "epoch": 0.31,
      "learning_rate": 6.482902217117028e-05,
      "loss": 1.3057,
      "step": 1051
    },
    {
      "epoch": 0.31,
      "learning_rate": 6.479920041235727e-05,
      "loss": 1.3637,
      "step": 1052
    },
    {
      "epoch": 0.31,
      "learning_rate": 6.476935624598966e-05,
      "loss": 1.4102,
      "step": 1053
    },
    {
      "epoch": 0.31,
      "learning_rate": 6.473948969903342e-05,
      "loss": 1.3363,
      "step": 1054
    },
    {
      "epoch": 0.31,
      "learning_rate": 6.470960079847477e-05,
      "loss": 1.3814,
      "step": 1055
    },
    {
      "epoch": 0.31,
      "learning_rate": 6.467968957132011e-05,
      "loss": 1.3407,
      "step": 1056
    },
    {
      "epoch": 0.31,
      "learning_rate": 6.4649756044596e-05,
      "loss": 1.4138,
      "step": 1057
    },
    {
      "epoch": 0.31,
      "learning_rate": 6.461980024534918e-05,
      "loss": 1.3256,
      "step": 1058
    },
    {
      "epoch": 0.31,
      "learning_rate": 6.458982220064649e-05,
      "loss": 1.3617,
      "step": 1059
    },
    {
      "epoch": 0.31,
      "learning_rate": 6.455982193757489e-05,
      "loss": 1.3438,
      "step": 1060
    },
    {
      "epoch": 0.31,
      "learning_rate": 6.452979948324139e-05,
      "loss": 1.3553,
      "step": 1061
    },
    {
      "epoch": 0.31,
      "learning_rate": 6.449975486477309e-05,
      "loss": 1.3779,
      "step": 1062
    },
    {
      "epoch": 0.31,
      "learning_rate": 6.446968810931707e-05,
      "loss": 1.3494,
      "step": 1063
    },
    {
      "epoch": 0.31,
      "learning_rate": 6.443959924404041e-05,
      "loss": 1.3816,
      "step": 1064
    },
    {
      "epoch": 0.31,
      "learning_rate": 6.440948829613025e-05,
      "loss": 1.3517,
      "step": 1065
    },
    {
      "epoch": 0.31,
      "learning_rate": 6.43793552927936e-05,
      "loss": 1.3375,
      "step": 1066
    },
    {
      "epoch": 0.31,
      "learning_rate": 6.434920026125741e-05,
      "loss": 1.4003,
      "step": 1067
    },
    {
      "epoch": 0.31,
      "learning_rate": 6.431902322876855e-05,
      "loss": 1.3346,
      "step": 1068
    },
    {
      "epoch": 0.31,
      "learning_rate": 6.428882422259379e-05,
      "loss": 1.3349,
      "step": 1069
    },
    {
      "epoch": 0.31,
      "learning_rate": 6.42586032700197e-05,
      "loss": 1.3448,
      "step": 1070
    },
    {
      "epoch": 0.31,
      "learning_rate": 6.422836039835273e-05,
      "loss": 1.3875,
      "step": 1071
    },
    {
      "epoch": 0.31,
      "learning_rate": 6.41980956349191e-05,
      "loss": 1.4008,
      "step": 1072
    },
    {
      "epoch": 0.31,
      "learning_rate": 6.416780900706484e-05,
      "loss": 1.3485,
      "step": 1073
    },
    {
      "epoch": 0.32,
      "learning_rate": 6.413750054215574e-05,
      "loss": 1.4057,
      "step": 1074
    },
    {
      "epoch": 0.32,
      "learning_rate": 6.410717026757726e-05,
      "loss": 1.3803,
      "step": 1075
    },
    {
      "epoch": 0.32,
      "learning_rate": 6.407681821073463e-05,
      "loss": 1.3363,
      "step": 1076
    },
    {
      "epoch": 0.32,
      "learning_rate": 6.404644439905274e-05,
      "loss": 1.342,
      "step": 1077
    },
    {
      "epoch": 0.32,
      "learning_rate": 6.401604885997614e-05,
      "loss": 1.4685,
      "step": 1078
    },
    {
      "epoch": 0.32,
      "learning_rate": 6.398563162096903e-05,
      "loss": 1.3779,
      "step": 1079
    },
    {
      "epoch": 0.32,
      "learning_rate": 6.395519270951517e-05,
      "loss": 1.3877,
      "step": 1080
    },
    {
      "epoch": 0.32,
      "learning_rate": 6.392473215311792e-05,
      "loss": 1.3241,
      "step": 1081
    },
    {
      "epoch": 0.32,
      "learning_rate": 6.389424997930023e-05,
      "loss": 1.4014,
      "step": 1082
    },
    {
      "epoch": 0.32,
      "learning_rate": 6.386374621560455e-05,
      "loss": 1.3541,
      "step": 1083
    },
    {
      "epoch": 0.32,
      "learning_rate": 6.383322088959287e-05,
      "loss": 1.3849,
      "step": 1084
    },
    {
      "epoch": 0.32,
      "learning_rate": 6.38026740288466e-05,
      "loss": 1.3652,
      "step": 1085
    },
    {
      "epoch": 0.32,
      "learning_rate": 6.377210566096668e-05,
      "loss": 1.3465,
      "step": 1086
    },
    {
      "epoch": 0.32,
      "learning_rate": 6.374151581357343e-05,
      "loss": 1.4152,
      "step": 1087
    },
    {
      "epoch": 0.32,
      "learning_rate": 6.37109045143066e-05,
      "loss": 1.341,
      "step": 1088
    },
    {
      "epoch": 0.32,
      "learning_rate": 6.368027179082536e-05,
      "loss": 1.3858,
      "step": 1089
    },
    {
      "epoch": 0.32,
      "learning_rate": 6.364961767080816e-05,
      "loss": 1.3795,
      "step": 1090
    },
    {
      "epoch": 0.32,
      "learning_rate": 6.361894218195282e-05,
      "loss": 1.3392,
      "step": 1091
    },
    {
      "epoch": 0.32,
      "learning_rate": 6.35882453519765e-05,
      "loss": 1.3573,
      "step": 1092
    },
    {
      "epoch": 0.32,
      "learning_rate": 6.355752720861559e-05,
      "loss": 1.3841,
      "step": 1093
    },
    {
      "epoch": 0.32,
      "learning_rate": 6.352678777962578e-05,
      "loss": 1.3559,
      "step": 1094
    },
    {
      "epoch": 0.32,
      "learning_rate": 6.349602709278195e-05,
      "loss": 1.3413,
      "step": 1095
    },
    {
      "epoch": 0.32,
      "learning_rate": 6.346524517587825e-05,
      "loss": 1.3898,
      "step": 1096
    },
    {
      "epoch": 0.32,
      "learning_rate": 6.343444205672794e-05,
      "loss": 1.4117,
      "step": 1097
    },
    {
      "epoch": 0.32,
      "learning_rate": 6.340361776316349e-05,
      "loss": 1.3277,
      "step": 1098
    },
    {
      "epoch": 0.32,
      "learning_rate": 6.337277232303647e-05,
      "loss": 1.2878,
      "step": 1099
    },
    {
      "epoch": 0.32,
      "learning_rate": 6.33419057642176e-05,
      "loss": 1.3592,
      "step": 1100
    },
    {
      "epoch": 0.32,
      "eval_loss": 0.12910032272338867,
      "eval_runtime": 25.0654,
      "eval_samples_per_second": 79.791,
      "eval_steps_per_second": 0.638,
      "step": 1100
    },
    {
      "epoch": 0.32,
      "learning_rate": 6.331101811459663e-05,
      "loss": 1.3193,
      "step": 1101
    },
    {
      "epoch": 0.32,
      "learning_rate": 6.328010940208241e-05,
      "loss": 1.3701,
      "step": 1102
    },
    {
      "epoch": 0.32,
      "learning_rate": 6.324917965460279e-05,
      "loss": 1.3797,
      "step": 1103
    },
    {
      "epoch": 0.32,
      "learning_rate": 6.321822890010464e-05,
      "loss": 1.3794,
      "step": 1104
    },
    {
      "epoch": 0.32,
      "learning_rate": 6.318725716655382e-05,
      "loss": 1.3539,
      "step": 1105
    },
    {
      "epoch": 0.32,
      "learning_rate": 6.315626448193511e-05,
      "loss": 1.3542,
      "step": 1106
    },
    {
      "epoch": 0.32,
      "learning_rate": 6.312525087425227e-05,
      "loss": 1.3845,
      "step": 1107
    },
    {
      "epoch": 0.33,
      "learning_rate": 6.309421637152794e-05,
      "loss": 1.3844,
      "step": 1108
    },
    {
      "epoch": 0.33,
      "learning_rate": 6.306316100180363e-05,
      "loss": 1.3809,
      "step": 1109
    },
    {
      "epoch": 0.33,
      "learning_rate": 6.303208479313974e-05,
      "loss": 1.4075,
      "step": 1110
    },
    {
      "epoch": 0.33,
      "learning_rate": 6.300098777361545e-05,
      "loss": 1.3933,
      "step": 1111
    },
    {
      "epoch": 0.33,
      "learning_rate": 6.296986997132877e-05,
      "loss": 1.3927,
      "step": 1112
    },
    {
      "epoch": 0.33,
      "learning_rate": 6.29387314143965e-05,
      "loss": 1.3684,
      "step": 1113
    },
    {
      "epoch": 0.33,
      "learning_rate": 6.290757213095417e-05,
      "loss": 1.366,
      "step": 1114
    },
    {
      "epoch": 0.33,
      "learning_rate": 6.287639214915606e-05,
      "loss": 1.3546,
      "step": 1115
    },
    {
      "epoch": 0.33,
      "learning_rate": 6.284519149717514e-05,
      "loss": 1.3749,
      "step": 1116
    },
    {
      "epoch": 0.33,
      "learning_rate": 6.281397020320307e-05,
      "loss": 1.3547,
      "step": 1117
    },
    {
      "epoch": 0.33,
      "learning_rate": 6.278272829545011e-05,
      "loss": 1.3282,
      "step": 1118
    },
    {
      "epoch": 0.33,
      "learning_rate": 6.27514658021452e-05,
      "loss": 1.3842,
      "step": 1119
    },
    {
      "epoch": 0.33,
      "learning_rate": 6.272018275153588e-05,
      "loss": 1.3804,
      "step": 1120
    },
    {
      "epoch": 0.33,
      "learning_rate": 6.268887917188824e-05,
      "loss": 1.3177,
      "step": 1121
    },
    {
      "epoch": 0.33,
      "learning_rate": 6.265755509148692e-05,
      "loss": 1.3476,
      "step": 1122
    },
    {
      "epoch": 0.33,
      "learning_rate": 6.26262105386351e-05,
      "loss": 1.3697,
      "step": 1123
    },
    {
      "epoch": 0.33,
      "learning_rate": 6.259484554165446e-05,
      "loss": 1.3288,
      "step": 1124
    },
    {
      "epoch": 0.33,
      "learning_rate": 6.256346012888514e-05,
      "loss": 1.3559,
      "step": 1125
    },
    {
      "epoch": 0.33,
      "learning_rate": 6.253205432868573e-05,
      "loss": 1.3796,
      "step": 1126
    },
    {
      "epoch": 0.33,
      "learning_rate": 6.250062816943322e-05,
      "loss": 1.3589,
      "step": 1127
    },
    {
      "epoch": 0.33,
      "learning_rate": 6.246918167952304e-05,
      "loss": 1.3468,
      "step": 1128
    },
    {
      "epoch": 0.33,
      "learning_rate": 6.243771488736896e-05,
      "loss": 1.3535,
      "step": 1129
    },
    {
      "epoch": 0.33,
      "learning_rate": 6.240622782140308e-05,
      "loss": 1.3399,
      "step": 1130
    },
    {
      "epoch": 0.33,
      "learning_rate": 6.237472051007589e-05,
      "loss": 1.3423,
      "step": 1131
    },
    {
      "epoch": 0.33,
      "learning_rate": 6.234319298185606e-05,
      "loss": 1.3419,
      "step": 1132
    },
    {
      "epoch": 0.33,
      "learning_rate": 6.231164526523063e-05,
      "loss": 1.3864,
      "step": 1133
    },
    {
      "epoch": 0.33,
      "learning_rate": 6.228007738870484e-05,
      "loss": 1.3656,
      "step": 1134
    },
    {
      "epoch": 0.33,
      "learning_rate": 6.224848938080212e-05,
      "loss": 1.3756,
      "step": 1135
    },
    {
      "epoch": 0.33,
      "learning_rate": 6.221688127006415e-05,
      "loss": 1.4115,
      "step": 1136
    },
    {
      "epoch": 0.33,
      "learning_rate": 6.218525308505072e-05,
      "loss": 1.3585,
      "step": 1137
    },
    {
      "epoch": 0.33,
      "learning_rate": 6.21536048543398e-05,
      "loss": 1.3752,
      "step": 1138
    },
    {
      "epoch": 0.33,
      "learning_rate": 6.212193660652743e-05,
      "loss": 1.3487,
      "step": 1139
    },
    {
      "epoch": 0.33,
      "learning_rate": 6.209024837022774e-05,
      "loss": 1.3239,
      "step": 1140
    },
    {
      "epoch": 0.33,
      "learning_rate": 6.205854017407297e-05,
      "loss": 1.3535,
      "step": 1141
    },
    {
      "epoch": 0.34,
      "learning_rate": 6.202681204671338e-05,
      "loss": 1.3799,
      "step": 1142
    },
    {
      "epoch": 0.34,
      "learning_rate": 6.199506401681718e-05,
      "loss": 1.3443,
      "step": 1143
    },
    {
      "epoch": 0.34,
      "learning_rate": 6.196329611307062e-05,
      "loss": 1.3641,
      "step": 1144
    },
    {
      "epoch": 0.34,
      "learning_rate": 6.19315083641779e-05,
      "loss": 1.3903,
      "step": 1145
    },
    {
      "epoch": 0.34,
      "learning_rate": 6.189970079886115e-05,
      "loss": 1.3888,
      "step": 1146
    },
    {
      "epoch": 0.34,
      "learning_rate": 6.18678734458604e-05,
      "loss": 1.3682,
      "step": 1147
    },
    {
      "epoch": 0.34,
      "learning_rate": 6.183602633393352e-05,
      "loss": 1.3903,
      "step": 1148
    },
    {
      "epoch": 0.34,
      "learning_rate": 6.180415949185633e-05,
      "loss": 1.4066,
      "step": 1149
    },
    {
      "epoch": 0.34,
      "learning_rate": 6.17722729484224e-05,
      "loss": 1.3253,
      "step": 1150
    },
    {
      "epoch": 0.34,
      "learning_rate": 6.17403667324431e-05,
      "loss": 1.3374,
      "step": 1151
    },
    {
      "epoch": 0.34,
      "learning_rate": 6.170844087274761e-05,
      "loss": 1.3239,
      "step": 1152
    },
    {
      "epoch": 0.34,
      "learning_rate": 6.167649539818284e-05,
      "loss": 1.3604,
      "step": 1153
    },
    {
      "epoch": 0.34,
      "learning_rate": 6.164453033761345e-05,
      "loss": 1.3063,
      "step": 1154
    },
    {
      "epoch": 0.34,
      "learning_rate": 6.161254571992175e-05,
      "loss": 1.344,
      "step": 1155
    },
    {
      "epoch": 0.34,
      "learning_rate": 6.158054157400776e-05,
      "loss": 1.342,
      "step": 1156
    },
    {
      "epoch": 0.34,
      "learning_rate": 6.154851792878915e-05,
      "loss": 1.4195,
      "step": 1157
    },
    {
      "epoch": 0.34,
      "learning_rate": 6.151647481320114e-05,
      "loss": 1.3671,
      "step": 1158
    },
    {
      "epoch": 0.34,
      "learning_rate": 6.148441225619664e-05,
      "loss": 1.3076,
      "step": 1159
    },
    {
      "epoch": 0.34,
      "learning_rate": 6.145233028674605e-05,
      "loss": 1.3985,
      "step": 1160
    },
    {
      "epoch": 0.34,
      "learning_rate": 6.142022893383739e-05,
      "loss": 1.292,
      "step": 1161
    },
    {
      "epoch": 0.34,
      "learning_rate": 6.138810822647608e-05,
      "loss": 1.349,
      "step": 1162
    },
    {
      "epoch": 0.34,
      "learning_rate": 6.135596819368512e-05,
      "loss": 1.3283,
      "step": 1163
    },
    {
      "epoch": 0.34,
      "learning_rate": 6.132380886450497e-05,
      "loss": 1.3997,
      "step": 1164
    },
    {
      "epoch": 0.34,
      "learning_rate": 6.129163026799346e-05,
      "loss": 1.3227,
      "step": 1165
    },
    {
      "epoch": 0.34,
      "learning_rate": 6.125943243322588e-05,
      "loss": 1.3925,
      "step": 1166
    },
    {
      "epoch": 0.34,
      "learning_rate": 6.12272153892949e-05,
      "loss": 1.3542,
      "step": 1167
    },
    {
      "epoch": 0.34,
      "learning_rate": 6.119497916531053e-05,
      "loss": 1.3854,
      "step": 1168
    },
    {
      "epoch": 0.34,
      "learning_rate": 6.116272379040011e-05,
      "loss": 1.3879,
      "step": 1169
    },
    {
      "epoch": 0.34,
      "learning_rate": 6.113044929370832e-05,
      "loss": 1.361,
      "step": 1170
    },
    {
      "epoch": 0.34,
      "learning_rate": 6.109815570439707e-05,
      "loss": 1.3438,
      "step": 1171
    },
    {
      "epoch": 0.34,
      "learning_rate": 6.106584305164553e-05,
      "loss": 1.4087,
      "step": 1172
    },
    {
      "epoch": 0.34,
      "learning_rate": 6.103351136465014e-05,
      "loss": 1.3363,
      "step": 1173
    },
    {
      "epoch": 0.34,
      "learning_rate": 6.1001160672624464e-05,
      "loss": 1.3155,
      "step": 1174
    },
    {
      "epoch": 0.34,
      "learning_rate": 6.0968791004799316e-05,
      "loss": 1.3705,
      "step": 1175
    },
    {
      "epoch": 0.35,
      "learning_rate": 6.093640239042261e-05,
      "loss": 1.2851,
      "step": 1176
    },
    {
      "epoch": 0.35,
      "learning_rate": 6.090399485875937e-05,
      "loss": 1.3735,
      "step": 1177
    },
    {
      "epoch": 0.35,
      "learning_rate": 6.0871568439091743e-05,
      "loss": 1.3692,
      "step": 1178
    },
    {
      "epoch": 0.35,
      "learning_rate": 6.083912316071893e-05,
      "loss": 1.333,
      "step": 1179
    },
    {
      "epoch": 0.35,
      "learning_rate": 6.080665905295718e-05,
      "loss": 1.369,
      "step": 1180
    },
    {
      "epoch": 0.35,
      "learning_rate": 6.077417614513973e-05,
      "loss": 1.3416,
      "step": 1181
    },
    {
      "epoch": 0.35,
      "learning_rate": 6.0741674466616854e-05,
      "loss": 1.3586,
      "step": 1182
    },
    {
      "epoch": 0.35,
      "learning_rate": 6.070915404675571e-05,
      "loss": 1.4149,
      "step": 1183
    },
    {
      "epoch": 0.35,
      "learning_rate": 6.067661491494046e-05,
      "loss": 1.3259,
      "step": 1184
    },
    {
      "epoch": 0.35,
      "learning_rate": 6.064405710057215e-05,
      "loss": 1.3377,
      "step": 1185
    },
    {
      "epoch": 0.35,
      "learning_rate": 6.061148063306869e-05,
      "loss": 1.3291,
      "step": 1186
    },
    {
      "epoch": 0.35,
      "learning_rate": 6.0578885541864875e-05,
      "loss": 1.4016,
      "step": 1187
    },
    {
      "epoch": 0.35,
      "learning_rate": 6.0546271856412306e-05,
      "loss": 1.3682,
      "step": 1188
    },
    {
      "epoch": 0.35,
      "learning_rate": 6.051363960617938e-05,
      "loss": 1.3339,
      "step": 1189
    },
    {
      "epoch": 0.35,
      "learning_rate": 6.0480988820651306e-05,
      "loss": 1.3934,
      "step": 1190
    },
    {
      "epoch": 0.35,
      "learning_rate": 6.044831952932997e-05,
      "loss": 1.3039,
      "step": 1191
    },
    {
      "epoch": 0.35,
      "learning_rate": 6.041563176173406e-05,
      "loss": 1.2647,
      "step": 1192
    },
    {
      "epoch": 0.35,
      "learning_rate": 6.03829255473989e-05,
      "loss": 1.3321,
      "step": 1193
    },
    {
      "epoch": 0.35,
      "learning_rate": 6.035020091587651e-05,
      "loss": 1.3712,
      "step": 1194
    },
    {
      "epoch": 0.35,
      "learning_rate": 6.031745789673554e-05,
      "loss": 1.3317,
      "step": 1195
    },
    {
      "epoch": 0.35,
      "learning_rate": 6.028469651956128e-05,
      "loss": 1.3625,
      "step": 1196
    },
    {
      "epoch": 0.35,
      "learning_rate": 6.0251916813955554e-05,
      "loss": 1.3284,
      "step": 1197
    },
    {
      "epoch": 0.35,
      "learning_rate": 6.0219118809536794e-05,
      "loss": 1.3712,
      "step": 1198
    },
    {
      "epoch": 0.35,
      "learning_rate": 6.018630253593994e-05,
      "loss": 1.3322,
      "step": 1199
    },
    {
      "epoch": 0.35,
      "learning_rate": 6.015346802281647e-05,
      "loss": 1.368,
      "step": 1200
    },
    {
      "epoch": 0.35,
      "eval_loss": 0.12655386328697205,
      "eval_runtime": 24.7692,
      "eval_samples_per_second": 80.746,
      "eval_steps_per_second": 0.646,
      "step": 1200
    },
    {
      "epoch": 0.35,
      "learning_rate": 6.012061529983431e-05,
      "loss": 1.3421,
      "step": 1201
    },
    {
      "epoch": 0.35,
      "learning_rate": 6.008774439667786e-05,
      "loss": 1.3628,
      "step": 1202
    },
    {
      "epoch": 0.35,
      "learning_rate": 6.0054855343047914e-05,
      "loss": 1.3855,
      "step": 1203
    },
    {
      "epoch": 0.35,
      "learning_rate": 6.002194816866171e-05,
      "loss": 1.3642,
      "step": 1204
    },
    {
      "epoch": 0.35,
      "learning_rate": 5.998902290325283e-05,
      "loss": 1.3857,
      "step": 1205
    },
    {
      "epoch": 0.35,
      "learning_rate": 5.995607957657124e-05,
      "loss": 1.3355,
      "step": 1206
    },
    {
      "epoch": 0.35,
      "learning_rate": 5.992311821838316e-05,
      "loss": 1.3002,
      "step": 1207
    },
    {
      "epoch": 0.35,
      "learning_rate": 5.989013885847117e-05,
      "loss": 1.3445,
      "step": 1208
    },
    {
      "epoch": 0.35,
      "learning_rate": 5.985714152663406e-05,
      "loss": 1.3945,
      "step": 1209
    },
    {
      "epoch": 0.35,
      "learning_rate": 5.982412625268691e-05,
      "loss": 1.4002,
      "step": 1210
    },
    {
      "epoch": 0.36,
      "learning_rate": 5.979109306646097e-05,
      "loss": 1.3741,
      "step": 1211
    },
    {
      "epoch": 0.36,
      "learning_rate": 5.9758041997803685e-05,
      "loss": 1.341,
      "step": 1212
    },
    {
      "epoch": 0.36,
      "learning_rate": 5.972497307657869e-05,
      "loss": 1.4316,
      "step": 1213
    },
    {
      "epoch": 0.36,
      "learning_rate": 5.969188633266571e-05,
      "loss": 1.344,
      "step": 1214
    },
    {
      "epoch": 0.36,
      "learning_rate": 5.9658781795960574e-05,
      "loss": 1.307,
      "step": 1215
    },
    {
      "epoch": 0.36,
      "learning_rate": 5.9625659496375234e-05,
      "loss": 1.3425,
      "step": 1216
    },
    {
      "epoch": 0.36,
      "learning_rate": 5.959251946383763e-05,
      "loss": 1.3767,
      "step": 1217
    },
    {
      "epoch": 0.36,
      "learning_rate": 5.955936172829179e-05,
      "loss": 1.3464,
      "step": 1218
    },
    {
      "epoch": 0.36,
      "learning_rate": 5.952618631969766e-05,
      "loss": 1.4277,
      "step": 1219
    },
    {
      "epoch": 0.36,
      "learning_rate": 5.9492993268031225e-05,
      "loss": 1.3417,
      "step": 1220
    },
    {
      "epoch": 0.36,
      "learning_rate": 5.9459782603284384e-05,
      "loss": 1.3843,
      "step": 1221
    },
    {
      "epoch": 0.36,
      "learning_rate": 5.9426554355464944e-05,
      "loss": 1.352,
      "step": 1222
    },
    {
      "epoch": 0.36,
      "learning_rate": 5.939330855459661e-05,
      "loss": 1.3184,
      "step": 1223
    },
    {
      "epoch": 0.36,
      "learning_rate": 5.936004523071894e-05,
      "loss": 1.3992,
      "step": 1224
    },
    {
      "epoch": 0.36,
      "learning_rate": 5.932676441388731e-05,
      "loss": 1.3183,
      "step": 1225
    },
    {
      "epoch": 0.36,
      "learning_rate": 5.929346613417296e-05,
      "loss": 1.3492,
      "step": 1226
    },
    {
      "epoch": 0.36,
      "learning_rate": 5.926015042166283e-05,
      "loss": 1.38,
      "step": 1227
    },
    {
      "epoch": 0.36,
      "learning_rate": 5.922681730645968e-05,
      "loss": 1.3634,
      "step": 1228
    },
    {
      "epoch": 0.36,
      "learning_rate": 5.919346681868195e-05,
      "loss": 1.3223,
      "step": 1229
    },
    {
      "epoch": 0.36,
      "learning_rate": 5.91600989884638e-05,
      "loss": 1.3518,
      "step": 1230
    },
    {
      "epoch": 0.36,
      "learning_rate": 5.912671384595504e-05,
      "loss": 1.3404,
      "step": 1231
    },
    {
      "epoch": 0.36,
      "learning_rate": 5.909331142132114e-05,
      "loss": 1.3751,
      "step": 1232
    },
    {
      "epoch": 0.36,
      "learning_rate": 5.905989174474319e-05,
      "loss": 1.3541,
      "step": 1233
    },
    {
      "epoch": 0.36,
      "learning_rate": 5.902645484641785e-05,
      "loss": 1.3202,
      "step": 1234
    },
    {
      "epoch": 0.36,
      "learning_rate": 5.899300075655734e-05,
      "loss": 1.3419,
      "step": 1235
    },
    {
      "epoch": 0.36,
      "learning_rate": 5.895952950538944e-05,
      "loss": 1.3577,
      "step": 1236
    },
    {
      "epoch": 0.36,
      "learning_rate": 5.892604112315741e-05,
      "loss": 1.348,
      "step": 1237
    },
    {
      "epoch": 0.36,
      "learning_rate": 5.889253564011999e-05,
      "loss": 1.3128,
      "step": 1238
    },
    {
      "epoch": 0.36,
      "learning_rate": 5.8859013086551394e-05,
      "loss": 1.3995,
      "step": 1239
    },
    {
      "epoch": 0.36,
      "learning_rate": 5.882547349274125e-05,
      "loss": 1.333,
      "step": 1240
    },
    {
      "epoch": 0.36,
      "learning_rate": 5.879191688899456e-05,
      "loss": 1.3532,
      "step": 1241
    },
    {
      "epoch": 0.36,
      "learning_rate": 5.8758343305631734e-05,
      "loss": 1.3281,
      "step": 1242
    },
    {
      "epoch": 0.36,
      "learning_rate": 5.872475277298847e-05,
      "loss": 1.3446,
      "step": 1243
    },
    {
      "epoch": 0.36,
      "learning_rate": 5.869114532141586e-05,
      "loss": 1.3548,
      "step": 1244
    },
    {
      "epoch": 0.37,
      "learning_rate": 5.8657520981280195e-05,
      "loss": 1.3182,
      "step": 1245
    },
    {
      "epoch": 0.37,
      "learning_rate": 5.86238797829631e-05,
      "loss": 1.3321,
      "step": 1246
    },
    {
      "epoch": 0.37,
      "learning_rate": 5.8590221756861395e-05,
      "loss": 1.3825,
      "step": 1247
    },
    {
      "epoch": 0.37,
      "learning_rate": 5.855654693338711e-05,
      "loss": 1.3687,
      "step": 1248
    },
    {
      "epoch": 0.37,
      "learning_rate": 5.8522855342967434e-05,
      "loss": 1.3752,
      "step": 1249
    },
    {
      "epoch": 0.37,
      "learning_rate": 5.848914701604476e-05,
      "loss": 1.343,
      "step": 1250
    },
    {
      "epoch": 0.37,
      "learning_rate": 5.845542198307654e-05,
      "loss": 1.3619,
      "step": 1251
    },
    {
      "epoch": 0.37,
      "learning_rate": 5.8421680274535374e-05,
      "loss": 1.3333,
      "step": 1252
    },
    {
      "epoch": 0.37,
      "learning_rate": 5.838792192090889e-05,
      "loss": 1.3347,
      "step": 1253
    },
    {
      "epoch": 0.37,
      "learning_rate": 5.835414695269976e-05,
      "loss": 1.3822,
      "step": 1254
    },
    {
      "epoch": 0.37,
      "learning_rate": 5.8320355400425714e-05,
      "loss": 1.3525,
      "step": 1255
    },
    {
      "epoch": 0.37,
      "learning_rate": 5.82865472946194e-05,
      "loss": 1.3225,
      "step": 1256
    },
    {
      "epoch": 0.37,
      "learning_rate": 5.825272266582847e-05,
      "loss": 1.3568,
      "step": 1257
    },
    {
      "epoch": 0.37,
      "learning_rate": 5.821888154461549e-05,
      "loss": 1.3643,
      "step": 1258
    },
    {
      "epoch": 0.37,
      "learning_rate": 5.8185023961557916e-05,
      "loss": 1.3287,
      "step": 1259
    },
    {
      "epoch": 0.37,
      "learning_rate": 5.8151149947248106e-05,
      "loss": 1.35,
      "step": 1260
    },
    {
      "epoch": 0.37,
      "learning_rate": 5.8117259532293246e-05,
      "loss": 1.347,
      "step": 1261
    },
    {
      "epoch": 0.37,
      "learning_rate": 5.8083352747315345e-05,
      "loss": 1.367,
      "step": 1262
    },
    {
      "epoch": 0.37,
      "learning_rate": 5.8049429622951194e-05,
      "loss": 1.3318,
      "step": 1263
    },
    {
      "epoch": 0.37,
      "learning_rate": 5.8015490189852365e-05,
      "loss": 1.3634,
      "step": 1264
    },
    {
      "epoch": 0.37,
      "learning_rate": 5.798153447868516e-05,
      "loss": 1.3485,
      "step": 1265
    },
    {
      "epoch": 0.37,
      "learning_rate": 5.794756252013057e-05,
      "loss": 1.3954,
      "step": 1266
    },
    {
      "epoch": 0.37,
      "learning_rate": 5.7913574344884314e-05,
      "loss": 1.3475,
      "step": 1267
    },
    {
      "epoch": 0.37,
      "learning_rate": 5.7879569983656694e-05,
      "loss": 1.3377,
      "step": 1268
    },
    {
      "epoch": 0.37,
      "learning_rate": 5.78455494671727e-05,
      "loss": 1.3657,
      "step": 1269
    },
    {
      "epoch": 0.37,
      "learning_rate": 5.7811512826171894e-05,
      "loss": 1.3459,
      "step": 1270
    },
    {
      "epoch": 0.37,
      "learning_rate": 5.77774600914084e-05,
      "loss": 1.3223,
      "step": 1271
    },
    {
      "epoch": 0.37,
      "learning_rate": 5.7743391293650895e-05,
      "loss": 1.3277,
      "step": 1272
    },
    {
      "epoch": 0.37,
      "learning_rate": 5.770930646368257e-05,
      "loss": 1.3707,
      "step": 1273
    },
    {
      "epoch": 0.37,
      "learning_rate": 5.7675205632301096e-05,
      "loss": 1.3402,
      "step": 1274
    },
    {
      "epoch": 0.37,
      "learning_rate": 5.764108883031861e-05,
      "loss": 1.336,
      "step": 1275
    },
    {
      "epoch": 0.37,
      "learning_rate": 5.7606956088561664e-05,
      "loss": 1.3107,
      "step": 1276
    },
    {
      "epoch": 0.37,
      "learning_rate": 5.757280743787125e-05,
      "loss": 1.3793,
      "step": 1277
    },
    {
      "epoch": 0.37,
      "learning_rate": 5.7538642909102654e-05,
      "loss": 1.3571,
      "step": 1278
    },
    {
      "epoch": 0.38,
      "learning_rate": 5.750446253312562e-05,
      "loss": 1.3524,
      "step": 1279
    },
    {
      "epoch": 0.38,
      "learning_rate": 5.747026634082412e-05,
      "loss": 1.3535,
      "step": 1280
    },
    {
      "epoch": 0.38,
      "learning_rate": 5.743605436309645e-05,
      "loss": 1.3684,
      "step": 1281
    },
    {
      "epoch": 0.38,
      "learning_rate": 5.740182663085519e-05,
      "loss": 1.375,
      "step": 1282
    },
    {
      "epoch": 0.38,
      "learning_rate": 5.736758317502714e-05,
      "loss": 1.3781,
      "step": 1283
    },
    {
      "epoch": 0.38,
      "learning_rate": 5.7333324026553264e-05,
      "loss": 1.3732,
      "step": 1284
    },
    {
      "epoch": 0.38,
      "learning_rate": 5.729904921638877e-05,
      "loss": 1.3345,
      "step": 1285
    },
    {
      "epoch": 0.38,
      "learning_rate": 5.7264758775502986e-05,
      "loss": 1.3591,
      "step": 1286
    },
    {
      "epoch": 0.38,
      "learning_rate": 5.723045273487937e-05,
      "loss": 1.3373,
      "step": 1287
    },
    {
      "epoch": 0.38,
      "learning_rate": 5.719613112551546e-05,
      "loss": 1.3948,
      "step": 1288
    },
    {
      "epoch": 0.38,
      "learning_rate": 5.7161793978422895e-05,
      "loss": 1.3502,
      "step": 1289
    },
    {
      "epoch": 0.38,
      "learning_rate": 5.712744132462732e-05,
      "loss": 1.364,
      "step": 1290
    },
    {
      "epoch": 0.38,
      "learning_rate": 5.709307319516841e-05,
      "loss": 1.3257,
      "step": 1291
    },
    {
      "epoch": 0.38,
      "learning_rate": 5.7058689621099803e-05,
      "loss": 1.3829,
      "step": 1292
    },
    {
      "epoch": 0.38,
      "learning_rate": 5.702429063348912e-05,
      "loss": 1.3489,
      "step": 1293
    },
    {
      "epoch": 0.38,
      "learning_rate": 5.6989876263417884e-05,
      "loss": 1.3713,
      "step": 1294
    },
    {
      "epoch": 0.38,
      "learning_rate": 5.6955446541981546e-05,
      "loss": 1.3407,
      "step": 1295
    },
    {
      "epoch": 0.38,
      "learning_rate": 5.69210015002894e-05,
      "loss": 1.3283,
      "step": 1296
    },
    {
      "epoch": 0.38,
      "learning_rate": 5.688654116946459e-05,
      "loss": 1.3342,
      "step": 1297
    },
    {
      "epoch": 0.38,
      "learning_rate": 5.685206558064407e-05,
      "loss": 1.3356,
      "step": 1298
    },
    {
      "epoch": 0.38,
      "learning_rate": 5.6817574764978594e-05,
      "loss": 1.3943,
      "step": 1299
    },
    {
      "epoch": 0.38,
      "learning_rate": 5.678306875363268e-05,
      "loss": 1.3543,
      "step": 1300
    },
    {
      "epoch": 0.38,
      "eval_loss": 0.12639901041984558,
      "eval_runtime": 24.668,
      "eval_samples_per_second": 81.077,
      "eval_steps_per_second": 0.649,
      "step": 1300
    },
    {
      "epoch": 0.38,
      "learning_rate": 5.6748547577784554e-05,
      "loss": 1.3064,
      "step": 1301
    },
    {
      "epoch": 0.38,
      "learning_rate": 5.671401126862615e-05,
      "loss": 1.3409,
      "step": 1302
    },
    {
      "epoch": 0.38,
      "learning_rate": 5.667945985736308e-05,
      "loss": 1.4306,
      "step": 1303
    },
    {
      "epoch": 0.38,
      "learning_rate": 5.664489337521463e-05,
      "loss": 1.3559,
      "step": 1304
    },
    {
      "epoch": 0.38,
      "learning_rate": 5.661031185341363e-05,
      "loss": 1.3144,
      "step": 1305
    },
    {
      "epoch": 0.38,
      "learning_rate": 5.657571532320658e-05,
      "loss": 1.375,
      "step": 1306
    },
    {
      "epoch": 0.38,
      "learning_rate": 5.6541103815853496e-05,
      "loss": 1.3723,
      "step": 1307
    },
    {
      "epoch": 0.38,
      "learning_rate": 5.6506477362627926e-05,
      "loss": 1.3533,
      "step": 1308
    },
    {
      "epoch": 0.38,
      "learning_rate": 5.647183599481694e-05,
      "loss": 1.2937,
      "step": 1309
    },
    {
      "epoch": 0.38,
      "learning_rate": 5.6437179743721075e-05,
      "loss": 1.3527,
      "step": 1310
    },
    {
      "epoch": 0.38,
      "learning_rate": 5.640250864065434e-05,
      "loss": 1.3708,
      "step": 1311
    },
    {
      "epoch": 0.38,
      "learning_rate": 5.6367822716944095e-05,
      "loss": 1.3469,
      "step": 1312
    },
    {
      "epoch": 0.39,
      "learning_rate": 5.6333122003931186e-05,
      "loss": 1.3701,
      "step": 1313
    },
    {
      "epoch": 0.39,
      "learning_rate": 5.629840653296974e-05,
      "loss": 1.3732,
      "step": 1314
    },
    {
      "epoch": 0.39,
      "learning_rate": 5.626367633542728e-05,
      "loss": 1.3294,
      "step": 1315
    },
    {
      "epoch": 0.39,
      "learning_rate": 5.622893144268458e-05,
      "loss": 1.3438,
      "step": 1316
    },
    {
      "epoch": 0.39,
      "learning_rate": 5.6194171886135754e-05,
      "loss": 1.3662,
      "step": 1317
    },
    {
      "epoch": 0.39,
      "learning_rate": 5.615939769718809e-05,
      "loss": 1.3315,
      "step": 1318
    },
    {
      "epoch": 0.39,
      "learning_rate": 5.612460890726216e-05,
      "loss": 1.352,
      "step": 1319
    },
    {
      "epoch": 0.39,
      "learning_rate": 5.608980554779171e-05,
      "loss": 1.3524,
      "step": 1320
    },
    {
      "epoch": 0.39,
      "learning_rate": 5.605498765022366e-05,
      "loss": 1.3169,
      "step": 1321
    },
    {
      "epoch": 0.39,
      "learning_rate": 5.602015524601802e-05,
      "loss": 1.3533,
      "step": 1322
    },
    {
      "epoch": 0.39,
      "learning_rate": 5.5985308366647985e-05,
      "loss": 1.3195,
      "step": 1323
    },
    {
      "epoch": 0.39,
      "learning_rate": 5.595044704359974e-05,
      "loss": 1.3552,
      "step": 1324
    },
    {
      "epoch": 0.39,
      "learning_rate": 5.59155713083726e-05,
      "loss": 1.3847,
      "step": 1325
    },
    {
      "epoch": 0.39,
      "learning_rate": 5.588068119247886e-05,
      "loss": 1.3543,
      "step": 1326
    },
    {
      "epoch": 0.39,
      "learning_rate": 5.584577672744382e-05,
      "loss": 1.33,
      "step": 1327
    },
    {
      "epoch": 0.39,
      "learning_rate": 5.5810857944805744e-05,
      "loss": 1.3055,
      "step": 1328
    },
    {
      "epoch": 0.39,
      "learning_rate": 5.577592487611581e-05,
      "loss": 1.3662,
      "step": 1329
    },
    {
      "epoch": 0.39,
      "learning_rate": 5.574097755293816e-05,
      "loss": 1.3242,
      "step": 1330
    },
    {
      "epoch": 0.39,
      "learning_rate": 5.570601600684974e-05,
      "loss": 1.3101,
      "step": 1331
    },
    {
      "epoch": 0.39,
      "learning_rate": 5.567104026944044e-05,
      "loss": 1.4008,
      "step": 1332
    },
    {
      "epoch": 0.39,
      "learning_rate": 5.5636050372312896e-05,
      "loss": 1.3579,
      "step": 1333
    },
    {
      "epoch": 0.39,
      "learning_rate": 5.560104634708254e-05,
      "loss": 1.3676,
      "step": 1334
    },
    {
      "epoch": 0.39,
      "learning_rate": 5.5566028225377645e-05,
      "loss": 1.325,
      "step": 1335
    },
    {
      "epoch": 0.39,
      "learning_rate": 5.553099603883913e-05,
      "loss": 1.3566,
      "step": 1336
    },
    {
      "epoch": 0.39,
      "learning_rate": 5.5495949819120676e-05,
      "loss": 1.3672,
      "step": 1337
    },
    {
      "epoch": 0.39,
      "learning_rate": 5.546088959788862e-05,
      "loss": 1.3204,
      "step": 1338
    },
    {
      "epoch": 0.39,
      "learning_rate": 5.5425815406821984e-05,
      "loss": 1.3678,
      "step": 1339
    },
    {
      "epoch": 0.39,
      "learning_rate": 5.539072727761238e-05,
      "loss": 1.3447,
      "step": 1340
    },
    {
      "epoch": 0.39,
      "learning_rate": 5.535562524196399e-05,
      "loss": 1.3748,
      "step": 1341
    },
    {
      "epoch": 0.39,
      "learning_rate": 5.5320509331593644e-05,
      "loss": 1.3437,
      "step": 1342
    },
    {
      "epoch": 0.39,
      "learning_rate": 5.528537957823061e-05,
      "loss": 1.3259,
      "step": 1343
    },
    {
      "epoch": 0.39,
      "learning_rate": 5.5250236013616766e-05,
      "loss": 1.3488,
      "step": 1344
    },
    {
      "epoch": 0.39,
      "learning_rate": 5.521507866950636e-05,
      "loss": 1.3266,
      "step": 1345
    },
    {
      "epoch": 0.39,
      "learning_rate": 5.517990757766619e-05,
      "loss": 1.3167,
      "step": 1346
    },
    {
      "epoch": 0.4,
      "learning_rate": 5.51447227698754e-05,
      "loss": 1.3166,
      "step": 1347
    },
    {
      "epoch": 0.4,
      "learning_rate": 5.510952427792559e-05,
      "loss": 1.4062,
      "step": 1348
    },
    {
      "epoch": 0.4,
      "learning_rate": 5.507431213362067e-05,
      "loss": 1.3962,
      "step": 1349
    },
    {
      "epoch": 0.4,
      "learning_rate": 5.5039086368776917e-05,
      "loss": 1.3542,
      "step": 1350
    },
    {
      "epoch": 0.4,
      "learning_rate": 5.500384701522291e-05,
      "loss": 1.3419,
      "step": 1351
    },
    {
      "epoch": 0.4,
      "learning_rate": 5.4968594104799515e-05,
      "loss": 1.3706,
      "step": 1352
    },
    {
      "epoch": 0.4,
      "learning_rate": 5.493332766935981e-05,
      "loss": 1.3761,
      "step": 1353
    },
    {
      "epoch": 0.4,
      "learning_rate": 5.489804774076914e-05,
      "loss": 1.3292,
      "step": 1354
    },
    {
      "epoch": 0.4,
      "learning_rate": 5.4862754350905e-05,
      "loss": 1.3335,
      "step": 1355
    },
    {
      "epoch": 0.4,
      "learning_rate": 5.48274475316571e-05,
      "loss": 1.3948,
      "step": 1356
    },
    {
      "epoch": 0.4,
      "learning_rate": 5.479212731492723e-05,
      "loss": 1.3258,
      "step": 1357
    },
    {
      "epoch": 0.4,
      "learning_rate": 5.4756793732629335e-05,
      "loss": 1.3009,
      "step": 1358
    },
    {
      "epoch": 0.4,
      "learning_rate": 5.472144681668938e-05,
      "loss": 1.3647,
      "step": 1359
    },
    {
      "epoch": 0.4,
      "learning_rate": 5.4686086599045424e-05,
      "loss": 1.3574,
      "step": 1360
    },
    {
      "epoch": 0.4,
      "learning_rate": 5.465071311164753e-05,
      "loss": 1.3108,
      "step": 1361
    },
    {
      "epoch": 0.4,
      "learning_rate": 5.4615326386457755e-05,
      "loss": 1.36,
      "step": 1362
    },
    {
      "epoch": 0.4,
      "learning_rate": 5.45799264554501e-05,
      "loss": 1.3399,
      "step": 1363
    },
    {
      "epoch": 0.4,
      "learning_rate": 5.4544513350610527e-05,
      "loss": 1.355,
      "step": 1364
    },
    {
      "epoch": 0.4,
      "learning_rate": 5.4509087103936866e-05,
      "loss": 1.3424,
      "step": 1365
    },
    {
      "epoch": 0.4,
      "learning_rate": 5.447364774743887e-05,
      "loss": 1.3227,
      "step": 1366
    },
    {
      "epoch": 0.4,
      "learning_rate": 5.443819531313807e-05,
      "loss": 1.3209,
      "step": 1367
    },
    {
      "epoch": 0.4,
      "learning_rate": 5.440272983306789e-05,
      "loss": 1.3286,
      "step": 1368
    },
    {
      "epoch": 0.4,
      "learning_rate": 5.436725133927347e-05,
      "loss": 1.3217,
      "step": 1369
    },
    {
      "epoch": 0.4,
      "learning_rate": 5.4331759863811764e-05,
      "loss": 1.2709,
      "step": 1370
    },
    {
      "epoch": 0.4,
      "learning_rate": 5.429625543875143e-05,
      "loss": 1.3699,
      "step": 1371
    },
    {
      "epoch": 0.4,
      "learning_rate": 5.426073809617284e-05,
      "loss": 1.2988,
      "step": 1372
    },
    {
      "epoch": 0.4,
      "learning_rate": 5.4225207868167994e-05,
      "loss": 1.3489,
      "step": 1373
    },
    {
      "epoch": 0.4,
      "learning_rate": 5.418966478684061e-05,
      "loss": 1.3632,
      "step": 1374
    },
    {
      "epoch": 0.4,
      "learning_rate": 5.415410888430595e-05,
      "loss": 1.3064,
      "step": 1375
    },
    {
      "epoch": 0.4,
      "learning_rate": 5.4118540192690905e-05,
      "loss": 1.3105,
      "step": 1376
    },
    {
      "epoch": 0.4,
      "learning_rate": 5.408295874413389e-05,
      "loss": 1.3146,
      "step": 1377
    },
    {
      "epoch": 0.4,
      "learning_rate": 5.404736457078489e-05,
      "loss": 1.3923,
      "step": 1378
    },
    {
      "epoch": 0.4,
      "learning_rate": 5.401175770480534e-05,
      "loss": 1.3058,
      "step": 1379
    },
    {
      "epoch": 0.4,
      "learning_rate": 5.397613817836818e-05,
      "loss": 1.3104,
      "step": 1380
    },
    {
      "epoch": 0.41,
      "learning_rate": 5.3940506023657745e-05,
      "loss": 1.3358,
      "step": 1381
    },
    {
      "epoch": 0.41,
      "learning_rate": 5.390486127286984e-05,
      "loss": 1.377,
      "step": 1382
    },
    {
      "epoch": 0.41,
      "learning_rate": 5.38692039582116e-05,
      "loss": 1.2862,
      "step": 1383
    },
    {
      "epoch": 0.41,
      "learning_rate": 5.3833534111901555e-05,
      "loss": 1.3436,
      "step": 1384
    },
    {
      "epoch": 0.41,
      "learning_rate": 5.379785176616951e-05,
      "loss": 1.3288,
      "step": 1385
    },
    {
      "epoch": 0.41,
      "learning_rate": 5.3762156953256596e-05,
      "loss": 1.376,
      "step": 1386
    },
    {
      "epoch": 0.41,
      "learning_rate": 5.3726449705415214e-05,
      "loss": 1.2981,
      "step": 1387
    },
    {
      "epoch": 0.41,
      "learning_rate": 5.3690730054908985e-05,
      "loss": 1.3244,
      "step": 1388
    },
    {
      "epoch": 0.41,
      "learning_rate": 5.365499803401273e-05,
      "loss": 1.3828,
      "step": 1389
    },
    {
      "epoch": 0.41,
      "learning_rate": 5.361925367501248e-05,
      "loss": 1.3495,
      "step": 1390
    },
    {
      "epoch": 0.41,
      "learning_rate": 5.3583497010205356e-05,
      "loss": 1.2911,
      "step": 1391
    },
    {
      "epoch": 0.41,
      "learning_rate": 5.354772807189967e-05,
      "loss": 1.3158,
      "step": 1392
    },
    {
      "epoch": 0.41,
      "learning_rate": 5.3511946892414775e-05,
      "loss": 1.3606,
      "step": 1393
    },
    {
      "epoch": 0.41,
      "learning_rate": 5.34761535040811e-05,
      "loss": 1.3096,
      "step": 1394
    },
    {
      "epoch": 0.41,
      "learning_rate": 5.3440347939240086e-05,
      "loss": 1.3594,
      "step": 1395
    },
    {
      "epoch": 0.41,
      "learning_rate": 5.340453023024422e-05,
      "loss": 1.3525,
      "step": 1396
    },
    {
      "epoch": 0.41,
      "learning_rate": 5.336870040945691e-05,
      "loss": 1.3859,
      "step": 1397
    },
    {
      "epoch": 0.41,
      "learning_rate": 5.333285850925256e-05,
      "loss": 1.3238,
      "step": 1398
    },
    {
      "epoch": 0.41,
      "learning_rate": 5.3297004562016436e-05,
      "loss": 1.3199,
      "step": 1399
    },
    {
      "epoch": 0.41,
      "learning_rate": 5.3261138600144745e-05,
      "loss": 1.3119,
      "step": 1400
    },
    {
      "epoch": 0.41,
      "eval_loss": 0.12737244367599487,
      "eval_runtime": 24.4081,
      "eval_samples_per_second": 81.94,
      "eval_steps_per_second": 0.656,
      "step": 1400
    },
    {
      "epoch": 0.41,
      "learning_rate": 5.3225260656044484e-05,
      "loss": 1.3797,
      "step": 1401
    },
    {
      "epoch": 0.41,
      "learning_rate": 5.318937076213354e-05,
      "loss": 1.2837,
      "step": 1402
    },
    {
      "epoch": 0.41,
      "learning_rate": 5.315346895084056e-05,
      "loss": 1.3252,
      "step": 1403
    },
    {
      "epoch": 0.41,
      "learning_rate": 5.3117555254604956e-05,
      "loss": 1.4006,
      "step": 1404
    },
    {
      "epoch": 0.41,
      "learning_rate": 5.3081629705876915e-05,
      "loss": 1.3279,
      "step": 1405
    },
    {
      "epoch": 0.41,
      "learning_rate": 5.3045692337117294e-05,
      "loss": 1.3905,
      "step": 1406
    },
    {
      "epoch": 0.41,
      "learning_rate": 5.300974318079766e-05,
      "loss": 1.3005,
      "step": 1407
    },
    {
      "epoch": 0.41,
      "learning_rate": 5.297378226940019e-05,
      "loss": 1.3374,
      "step": 1408
    },
    {
      "epoch": 0.41,
      "learning_rate": 5.293780963541775e-05,
      "loss": 1.302,
      "step": 1409
    },
    {
      "epoch": 0.41,
      "learning_rate": 5.290182531135372e-05,
      "loss": 1.3665,
      "step": 1410
    },
    {
      "epoch": 0.41,
      "learning_rate": 5.2865829329722093e-05,
      "loss": 1.3417,
      "step": 1411
    },
    {
      "epoch": 0.41,
      "learning_rate": 5.2829821723047386e-05,
      "loss": 1.3647,
      "step": 1412
    },
    {
      "epoch": 0.41,
      "learning_rate": 5.279380252386461e-05,
      "loss": 1.329,
      "step": 1413
    },
    {
      "epoch": 0.41,
      "learning_rate": 5.275777176471927e-05,
      "loss": 1.324,
      "step": 1414
    },
    {
      "epoch": 0.42,
      "learning_rate": 5.272172947816729e-05,
      "loss": 1.3447,
      "step": 1415
    },
    {
      "epoch": 0.42,
      "learning_rate": 5.268567569677502e-05,
      "loss": 1.3662,
      "step": 1416
    },
    {
      "epoch": 0.42,
      "learning_rate": 5.264961045311921e-05,
      "loss": 1.3524,
      "step": 1417
    },
    {
      "epoch": 0.42,
      "learning_rate": 5.2613533779786945e-05,
      "loss": 1.2874,
      "step": 1418
    },
    {
      "epoch": 0.42,
      "learning_rate": 5.2577445709375666e-05,
      "loss": 1.4179,
      "step": 1419
    },
    {
      "epoch": 0.42,
      "learning_rate": 5.2541346274493065e-05,
      "loss": 1.3462,
      "step": 1420
    },
    {
      "epoch": 0.42,
      "learning_rate": 5.250523550775716e-05,
      "loss": 1.3295,
      "step": 1421
    },
    {
      "epoch": 0.42,
      "learning_rate": 5.246911344179616e-05,
      "loss": 1.3016,
      "step": 1422
    },
    {
      "epoch": 0.42,
      "learning_rate": 5.243298010924852e-05,
      "loss": 1.3403,
      "step": 1423
    },
    {
      "epoch": 0.42,
      "learning_rate": 5.239683554276283e-05,
      "loss": 1.3318,
      "step": 1424
    },
    {
      "epoch": 0.42,
      "learning_rate": 5.23606797749979e-05,
      "loss": 1.2756,
      "step": 1425
    },
    {
      "epoch": 0.42,
      "learning_rate": 5.232451283862259e-05,
      "loss": 1.2976,
      "step": 1426
    },
    {
      "epoch": 0.42,
      "learning_rate": 5.2288334766315896e-05,
      "loss": 1.3963,
      "step": 1427
    },
    {
      "epoch": 0.42,
      "learning_rate": 5.225214559076683e-05,
      "loss": 1.3155,
      "step": 1428
    },
    {
      "epoch": 0.42,
      "learning_rate": 5.221594534467452e-05,
      "loss": 1.3423,
      "step": 1429
    },
    {
      "epoch": 0.42,
      "learning_rate": 5.217973406074798e-05,
      "loss": 1.3264,
      "step": 1430
    },
    {
      "epoch": 0.42,
      "learning_rate": 5.2143511771706324e-05,
      "loss": 1.344,
      "step": 1431
    },
    {
      "epoch": 0.42,
      "learning_rate": 5.210727851027851e-05,
      "loss": 1.3426,
      "step": 1432
    },
    {
      "epoch": 0.42,
      "learning_rate": 5.207103430920345e-05,
      "loss": 1.3934,
      "step": 1433
    },
    {
      "epoch": 0.42,
      "learning_rate": 5.203477920122996e-05,
      "loss": 1.355,
      "step": 1434
    },
    {
      "epoch": 0.42,
      "learning_rate": 5.199851321911667e-05,
      "loss": 1.3506,
      "step": 1435
    },
    {
      "epoch": 0.42,
      "learning_rate": 5.1962236395632055e-05,
      "loss": 1.3267,
      "step": 1436
    },
    {
      "epoch": 0.42,
      "learning_rate": 5.1925948763554396e-05,
      "loss": 1.2905,
      "step": 1437
    },
    {
      "epoch": 0.42,
      "learning_rate": 5.1889650355671725e-05,
      "loss": 1.3592,
      "step": 1438
    },
    {
      "epoch": 0.42,
      "learning_rate": 5.1853341204781815e-05,
      "loss": 1.3158,
      "step": 1439
    },
    {
      "epoch": 0.42,
      "learning_rate": 5.1817021343692156e-05,
      "loss": 1.29,
      "step": 1440
    },
    {
      "epoch": 0.42,
      "learning_rate": 5.178069080521989e-05,
      "loss": 1.3404,
      "step": 1441
    },
    {
      "epoch": 0.42,
      "learning_rate": 5.174434962219182e-05,
      "loss": 1.388,
      "step": 1442
    },
    {
      "epoch": 0.42,
      "learning_rate": 5.1707997827444394e-05,
      "loss": 1.3279,
      "step": 1443
    },
    {
      "epoch": 0.42,
      "learning_rate": 5.167163545382359e-05,
      "loss": 1.2866,
      "step": 1444
    },
    {
      "epoch": 0.42,
      "learning_rate": 5.163526253418501e-05,
      "loss": 1.3182,
      "step": 1445
    },
    {
      "epoch": 0.42,
      "learning_rate": 5.1598879101393734e-05,
      "loss": 1.3818,
      "step": 1446
    },
    {
      "epoch": 0.42,
      "learning_rate": 5.156248518832438e-05,
      "loss": 1.3141,
      "step": 1447
    },
    {
      "epoch": 0.42,
      "learning_rate": 5.152608082786098e-05,
      "loss": 1.3134,
      "step": 1448
    },
    {
      "epoch": 0.43,
      "learning_rate": 5.1489666052897075e-05,
      "loss": 1.4078,
      "step": 1449
    },
    {
      "epoch": 0.43,
      "learning_rate": 5.1453240896335564e-05,
      "loss": 1.333,
      "step": 1450
    },
    {
      "epoch": 0.43,
      "learning_rate": 5.1416805391088765e-05,
      "loss": 1.3424,
      "step": 1451
    },
    {
      "epoch": 0.43,
      "learning_rate": 5.13803595700783e-05,
      "loss": 1.2766,
      "step": 1452
    },
    {
      "epoch": 0.43,
      "learning_rate": 5.1343903466235174e-05,
      "loss": 1.3573,
      "step": 1453
    },
    {
      "epoch": 0.43,
      "learning_rate": 5.130743711249961e-05,
      "loss": 1.3733,
      "step": 1454
    },
    {
      "epoch": 0.43,
      "learning_rate": 5.127096054182116e-05,
      "loss": 1.372,
      "step": 1455
    },
    {
      "epoch": 0.43,
      "learning_rate": 5.123447378715856e-05,
      "loss": 1.3083,
      "step": 1456
    },
    {
      "epoch": 0.43,
      "learning_rate": 5.1197976881479786e-05,
      "loss": 1.3941,
      "step": 1457
    },
    {
      "epoch": 0.43,
      "learning_rate": 5.116146985776194e-05,
      "loss": 1.3385,
      "step": 1458
    },
    {
      "epoch": 0.43,
      "learning_rate": 5.112495274899131e-05,
      "loss": 1.3056,
      "step": 1459
    },
    {
      "epoch": 0.43,
      "learning_rate": 5.108842558816327e-05,
      "loss": 1.3337,
      "step": 1460
    },
    {
      "epoch": 0.43,
      "learning_rate": 5.10518884082823e-05,
      "loss": 1.3294,
      "step": 1461
    },
    {
      "epoch": 0.43,
      "learning_rate": 5.101534124236189e-05,
      "loss": 1.3105,
      "step": 1462
    },
    {
      "epoch": 0.43,
      "learning_rate": 5.0978784123424617e-05,
      "loss": 1.2798,
      "step": 1463
    },
    {
      "epoch": 0.43,
      "learning_rate": 5.094221708450199e-05,
      "loss": 1.345,
      "step": 1464
    },
    {
      "epoch": 0.43,
      "learning_rate": 5.090564015863452e-05,
      "loss": 1.339,
      "step": 1465
    },
    {
      "epoch": 0.43,
      "learning_rate": 5.086905337887162e-05,
      "loss": 1.2971,
      "step": 1466
    },
    {
      "epoch": 0.43,
      "learning_rate": 5.083245677827165e-05,
      "loss": 1.3337,
      "step": 1467
    },
    {
      "epoch": 0.43,
      "learning_rate": 5.0795850389901784e-05,
      "loss": 1.4053,
      "step": 1468
    },
    {
      "epoch": 0.43,
      "learning_rate": 5.075923424683811e-05,
      "loss": 1.2797,
      "step": 1469
    },
    {
      "epoch": 0.43,
      "learning_rate": 5.0722608382165473e-05,
      "loss": 1.3172,
      "step": 1470
    },
    {
      "epoch": 0.43,
      "learning_rate": 5.068597282897754e-05,
      "loss": 1.3616,
      "step": 1471
    },
    {
      "epoch": 0.43,
      "learning_rate": 5.06493276203767e-05,
      "loss": 1.3528,
      "step": 1472
    },
    {
      "epoch": 0.43,
      "learning_rate": 5.061267278947408e-05,
      "loss": 1.3289,
      "step": 1473
    },
    {
      "epoch": 0.43,
      "learning_rate": 5.05760083693895e-05,
      "loss": 1.2941,
      "step": 1474
    },
    {
      "epoch": 0.43,
      "learning_rate": 5.053933439325147e-05,
      "loss": 1.3064,
      "step": 1475
    },
    {
      "epoch": 0.43,
      "learning_rate": 5.050265089419709e-05,
      "loss": 1.3308,
      "step": 1476
    },
    {
      "epoch": 0.43,
      "learning_rate": 5.04659579053721e-05,
      "loss": 1.3564,
      "step": 1477
    },
    {
      "epoch": 0.43,
      "learning_rate": 5.042925545993079e-05,
      "loss": 1.2551,
      "step": 1478
    },
    {
      "epoch": 0.43,
      "learning_rate": 5.039254359103602e-05,
      "loss": 1.2842,
      "step": 1479
    },
    {
      "epoch": 0.43,
      "learning_rate": 5.035582233185914e-05,
      "loss": 1.3351,
      "step": 1480
    },
    {
      "epoch": 0.43,
      "learning_rate": 5.031909171557998e-05,
      "loss": 1.389,
      "step": 1481
    },
    {
      "epoch": 0.43,
      "learning_rate": 5.0282351775386866e-05,
      "loss": 1.3236,
      "step": 1482
    },
    {
      "epoch": 0.44,
      "learning_rate": 5.02456025444765e-05,
      "loss": 1.3634,
      "step": 1483
    },
    {
      "epoch": 0.44,
      "learning_rate": 5.0208844056054025e-05,
      "loss": 1.3284,
      "step": 1484
    },
    {
      "epoch": 0.44,
      "learning_rate": 5.0172076343332914e-05,
      "loss": 1.329,
      "step": 1485
    },
    {
      "epoch": 0.44,
      "learning_rate": 5.013529943953497e-05,
      "loss": 1.3316,
      "step": 1486
    },
    {
      "epoch": 0.44,
      "learning_rate": 5.0098513377890336e-05,
      "loss": 1.3781,
      "step": 1487
    },
    {
      "epoch": 0.44,
      "learning_rate": 5.0061718191637394e-05,
      "loss": 1.3008,
      "step": 1488
    },
    {
      "epoch": 0.44,
      "learning_rate": 5.0024913914022796e-05,
      "loss": 1.292,
      "step": 1489
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.99881005783014e-05,
      "loss": 1.3426,
      "step": 1490
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.995127821773625e-05,
      "loss": 1.3207,
      "step": 1491
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.991444686559853e-05,
      "loss": 1.3345,
      "step": 1492
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.987760655516757e-05,
      "loss": 1.3345,
      "step": 1493
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.984075731973079e-05,
      "loss": 1.3544,
      "step": 1494
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.980389919258364e-05,
      "loss": 1.3371,
      "step": 1495
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.976703220702967e-05,
      "loss": 1.3507,
      "step": 1496
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.973015639638037e-05,
      "loss": 1.2927,
      "step": 1497
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.9693271793955255e-05,
      "loss": 1.3525,
      "step": 1498
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.965637843308173e-05,
      "loss": 1.3101,
      "step": 1499
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.961947634709518e-05,
      "loss": 1.3267,
      "step": 1500
    },
    {
      "epoch": 0.44,
      "eval_loss": 0.1289711445569992,
      "eval_runtime": 24.6907,
      "eval_samples_per_second": 81.002,
      "eval_steps_per_second": 0.648,
      "step": 1500
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.9582565569338806e-05,
      "loss": 1.3406,
      "step": 1501
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.954564613316371e-05,
      "loss": 1.318,
      "step": 1502
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.95087180719288e-05,
      "loss": 1.3414,
      "step": 1503
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.947178141900079e-05,
      "loss": 1.3382,
      "step": 1504
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.9434836207754104e-05,
      "loss": 1.3201,
      "step": 1505
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.939788247157099e-05,
      "loss": 1.3079,
      "step": 1506
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.936092024384132e-05,
      "loss": 1.2724,
      "step": 1507
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.9323949557962684e-05,
      "loss": 1.2962,
      "step": 1508
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.928697044734029e-05,
      "loss": 1.2819,
      "step": 1509
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.924998294538697e-05,
      "loss": 1.3792,
      "step": 1510
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.9212987085523123e-05,
      "loss": 1.2954,
      "step": 1511
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.917598290117674e-05,
      "loss": 1.3168,
      "step": 1512
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.913897042578327e-05,
      "loss": 1.3593,
      "step": 1513
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.910194969278573e-05,
      "loss": 1.3198,
      "step": 1514
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.906492073563452e-05,
      "loss": 1.365,
      "step": 1515
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.902788358778753e-05,
      "loss": 1.3292,
      "step": 1516
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.8990838282710015e-05,
      "loss": 1.4016,
      "step": 1517
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.8953784853874624e-05,
      "loss": 1.3013,
      "step": 1518
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.8916723334761334e-05,
      "loss": 1.3434,
      "step": 1519
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.887965375885744e-05,
      "loss": 1.3012,
      "step": 1520
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.88425761596575e-05,
      "loss": 1.3854,
      "step": 1521
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.880549057066335e-05,
      "loss": 1.3082,
      "step": 1522
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.8768397025383996e-05,
      "loss": 1.3121,
      "step": 1523
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.873129555733571e-05,
      "loss": 1.334,
      "step": 1524
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.869418620004184e-05,
      "loss": 1.3792,
      "step": 1525
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.865706898703293e-05,
      "loss": 1.3608,
      "step": 1526
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.861994395184657e-05,
      "loss": 1.3345,
      "step": 1527
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.858281112802745e-05,
      "loss": 1.3449,
      "step": 1528
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.8545670549127295e-05,
      "loss": 1.3325,
      "step": 1529
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.8508522248704816e-05,
      "loss": 1.3026,
      "step": 1530
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.8471366260325715e-05,
      "loss": 1.3181,
      "step": 1531
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.843420261756266e-05,
      "loss": 1.3538,
      "step": 1532
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.83970313539952e-05,
      "loss": 1.3219,
      "step": 1533
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.8359852503209785e-05,
      "loss": 1.2695,
      "step": 1534
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.832266609879972e-05,
      "loss": 1.3187,
      "step": 1535
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.828547217436515e-05,
      "loss": 1.3121,
      "step": 1536
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.824827076351298e-05,
      "loss": 1.2867,
      "step": 1537
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.821106189985693e-05,
      "loss": 1.3149,
      "step": 1538
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.817384561701738e-05,
      "loss": 1.3152,
      "step": 1539
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.8136621948621525e-05,
      "loss": 1.3443,
      "step": 1540
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.809939092830309e-05,
      "loss": 1.348,
      "step": 1541
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.8062152589702573e-05,
      "loss": 1.3102,
      "step": 1542
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.8024906966467e-05,
      "loss": 1.301,
      "step": 1543
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.798765409225004e-05,
      "loss": 1.3714,
      "step": 1544
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.7950394000711844e-05,
      "loss": 1.2528,
      "step": 1545
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.7913126725519164e-05,
      "loss": 1.3302,
      "step": 1546
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.787585230034518e-05,
      "loss": 1.3359,
      "step": 1547
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.783857075886956e-05,
      "loss": 1.3627,
      "step": 1548
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.7801282134778395e-05,
      "loss": 1.3164,
      "step": 1549
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.7763986461764186e-05,
      "loss": 1.3441,
      "step": 1550
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.772668377352579e-05,
      "loss": 1.3688,
      "step": 1551
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.76893741037684e-05,
      "loss": 1.3072,
      "step": 1552
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.7652057486203525e-05,
      "loss": 1.3128,
      "step": 1553
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.7614733954548974e-05,
      "loss": 1.3613,
      "step": 1554
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.757740354252876e-05,
      "loss": 1.3199,
      "step": 1555
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.7540066283873134e-05,
      "loss": 1.3488,
      "step": 1556
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.750272221231854e-05,
      "loss": 1.3038,
      "step": 1557
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.746537136160757e-05,
      "loss": 1.348,
      "step": 1558
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.7428013765488946e-05,
      "loss": 1.3155,
      "step": 1559
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.7390649457717476e-05,
      "loss": 1.3273,
      "step": 1560
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.735327847205402e-05,
      "loss": 1.3423,
      "step": 1561
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.731590084226552e-05,
      "loss": 1.3632,
      "step": 1562
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.727851660212487e-05,
      "loss": 1.2966,
      "step": 1563
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.724112578541098e-05,
      "loss": 1.3457,
      "step": 1564
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.7203728425908644e-05,
      "loss": 1.2946,
      "step": 1565
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.716632455740865e-05,
      "loss": 1.3452,
      "step": 1566
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.712891421370757e-05,
      "loss": 1.3059,
      "step": 1567
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.709149742860792e-05,
      "loss": 1.3404,
      "step": 1568
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.705407423591797e-05,
      "loss": 1.3027,
      "step": 1569
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.7016644669451806e-05,
      "loss": 1.3268,
      "step": 1570
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.697920876302928e-05,
      "loss": 1.3423,
      "step": 1571
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.694176655047596e-05,
      "loss": 1.3156,
      "step": 1572
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.690431806562311e-05,
      "loss": 1.3373,
      "step": 1573
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.686686334230766e-05,
      "loss": 1.3333,
      "step": 1574
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.682940241437221e-05,
      "loss": 1.3011,
      "step": 1575
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.679193531566491e-05,
      "loss": 1.2923,
      "step": 1576
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.675446208003954e-05,
      "loss": 1.3618,
      "step": 1577
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.6716982741355386e-05,
      "loss": 1.3239,
      "step": 1578
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.667949733347726e-05,
      "loss": 1.2958,
      "step": 1579
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.664200589027546e-05,
      "loss": 1.3201,
      "step": 1580
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.660450844562576e-05,
      "loss": 1.2923,
      "step": 1581
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.6567005033409294e-05,
      "loss": 1.3134,
      "step": 1582
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.652949568751267e-05,
      "loss": 1.3365,
      "step": 1583
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.649198044182779e-05,
      "loss": 1.3283,
      "step": 1584
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.645445933025192e-05,
      "loss": 1.4115,
      "step": 1585
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.6416932386687624e-05,
      "loss": 1.3075,
      "step": 1586
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.637939964504274e-05,
      "loss": 1.3186,
      "step": 1587
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.63418611392303e-05,
      "loss": 1.3817,
      "step": 1588
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.630431690316861e-05,
      "loss": 1.3419,
      "step": 1589
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.6266766970781116e-05,
      "loss": 1.2899,
      "step": 1590
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.622921137599643e-05,
      "loss": 1.2792,
      "step": 1591
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.619165015274824e-05,
      "loss": 1.3751,
      "step": 1592
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.615408333497538e-05,
      "loss": 1.2905,
      "step": 1593
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.611651095662168e-05,
      "loss": 1.3497,
      "step": 1594
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.607893305163604e-05,
      "loss": 1.3325,
      "step": 1595
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.6041349653972336e-05,
      "loss": 1.3999,
      "step": 1596
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.6003760797589404e-05,
      "loss": 1.2777,
      "step": 1597
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.5966166516450985e-05,
      "loss": 1.3144,
      "step": 1598
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.59285668445258e-05,
      "loss": 1.3161,
      "step": 1599
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.589096181578734e-05,
      "loss": 1.3084,
      "step": 1600
    },
    {
      "epoch": 0.47,
      "eval_loss": 0.12573568522930145,
      "eval_runtime": 24.516,
      "eval_samples_per_second": 81.579,
      "eval_steps_per_second": 0.653,
      "step": 1600
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.585335146421404e-05,
      "loss": 1.3642,
      "step": 1601
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.581573582378906e-05,
      "loss": 1.3107,
      "step": 1602
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.577811492850039e-05,
      "loss": 1.3872,
      "step": 1603
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.574048881234073e-05,
      "loss": 1.3334,
      "step": 1604
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.5702857509307536e-05,
      "loss": 1.2477,
      "step": 1605
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.5665221053402915e-05,
      "loss": 1.3203,
      "step": 1606
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.562757947863365e-05,
      "loss": 1.3816,
      "step": 1607
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.558993281901116e-05,
      "loss": 1.2813,
      "step": 1608
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.5552281108551435e-05,
      "loss": 1.2786,
      "step": 1609
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.551462438127503e-05,
      "loss": 1.3509,
      "step": 1610
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.547696267120705e-05,
      "loss": 1.3546,
      "step": 1611
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.543929601237709e-05,
      "loss": 1.316,
      "step": 1612
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.540162443881922e-05,
      "loss": 1.3354,
      "step": 1613
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.536394798457195e-05,
      "loss": 1.2978,
      "step": 1614
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.5326266683678226e-05,
      "loss": 1.319,
      "step": 1615
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.5288580570185294e-05,
      "loss": 1.318,
      "step": 1616
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.525088967814485e-05,
      "loss": 1.3458,
      "step": 1617
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.5213194041612824e-05,
      "loss": 1.3451,
      "step": 1618
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.517549369464949e-05,
      "loss": 1.2594,
      "step": 1619
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.513778867131936e-05,
      "loss": 1.3143,
      "step": 1620
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.510007900569116e-05,
      "loss": 1.32,
      "step": 1621
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.506236473183782e-05,
      "loss": 1.3629,
      "step": 1622
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.5024645883836426e-05,
      "loss": 1.3109,
      "step": 1623
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.498692249576821e-05,
      "loss": 1.3158,
      "step": 1624
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.494919460171851e-05,
      "loss": 1.276,
      "step": 1625
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.4911462235776704e-05,
      "loss": 1.3392,
      "step": 1626
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.4873725432036255e-05,
      "loss": 1.288,
      "step": 1627
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.4835984224594586e-05,
      "loss": 1.2946,
      "step": 1628
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.479823864755316e-05,
      "loss": 1.2925,
      "step": 1629
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.476048873501733e-05,
      "loss": 1.3792,
      "step": 1630
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.472273452109639e-05,
      "loss": 1.2779,
      "step": 1631
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.468497603990352e-05,
      "loss": 1.3084,
      "step": 1632
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.464721332555577e-05,
      "loss": 1.374,
      "step": 1633
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.4609446412174e-05,
      "loss": 1.2989,
      "step": 1634
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.457167533388286e-05,
      "loss": 1.3216,
      "step": 1635
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.4533900124810756e-05,
      "loss": 1.2977,
      "step": 1636
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.449612081908987e-05,
      "loss": 1.339,
      "step": 1637
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.445833745085602e-05,
      "loss": 1.2815,
      "step": 1638
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.4420550054248756e-05,
      "loss": 1.3592,
      "step": 1639
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.438275866341123e-05,
      "loss": 1.3131,
      "step": 1640
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.4344963312490216e-05,
      "loss": 1.3288,
      "step": 1641
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.4307164035636075e-05,
      "loss": 1.3354,
      "step": 1642
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.4269360867002675e-05,
      "loss": 1.3171,
      "step": 1643
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.423155384074747e-05,
      "loss": 1.3129,
      "step": 1644
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.4193742991031325e-05,
      "loss": 1.3287,
      "step": 1645
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.415592835201862e-05,
      "loss": 1.3433,
      "step": 1646
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.4118109957877114e-05,
      "loss": 1.2809,
      "step": 1647
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.408028784277799e-05,
      "loss": 1.3841,
      "step": 1648
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.404246204089578e-05,
      "loss": 1.3282,
      "step": 1649
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.4004632586408366e-05,
      "loss": 1.2509,
      "step": 1650
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.396679951349687e-05,
      "loss": 1.2912,
      "step": 1651
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.392896285634577e-05,
      "loss": 1.3519,
      "step": 1652
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.389112264914273e-05,
      "loss": 1.3311,
      "step": 1653
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.385327892607861e-05,
      "loss": 1.3475,
      "step": 1654
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.381543172134748e-05,
      "loss": 1.3375,
      "step": 1655
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.377758106914655e-05,
      "loss": 1.351,
      "step": 1656
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.373972700367613e-05,
      "loss": 1.2965,
      "step": 1657
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.370186955913962e-05,
      "loss": 1.2933,
      "step": 1658
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.3664008769743484e-05,
      "loss": 1.2752,
      "step": 1659
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.362614466969721e-05,
      "loss": 1.3498,
      "step": 1660
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.3588277293213236e-05,
      "loss": 1.3237,
      "step": 1661
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.3550406674507026e-05,
      "loss": 1.3121,
      "step": 1662
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.351253284779692e-05,
      "loss": 1.3118,
      "step": 1663
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.3474655847304177e-05,
      "loss": 1.2915,
      "step": 1664
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.343677570725293e-05,
      "loss": 1.3193,
      "step": 1665
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.3398892461870146e-05,
      "loss": 1.3559,
      "step": 1666
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.336100614538557e-05,
      "loss": 1.3361,
      "step": 1667
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.332311679203177e-05,
      "loss": 1.3126,
      "step": 1668
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.3285224436044026e-05,
      "loss": 1.2813,
      "step": 1669
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.324732911166034e-05,
      "loss": 1.3133,
      "step": 1670
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.3209430853121395e-05,
      "loss": 1.3254,
      "step": 1671
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.317152969467053e-05,
      "loss": 1.3145,
      "step": 1672
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.313362567055367e-05,
      "loss": 1.3175,
      "step": 1673
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.3095718815019395e-05,
      "loss": 1.3076,
      "step": 1674
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.305780916231878e-05,
      "loss": 1.3127,
      "step": 1675
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.301989674670547e-05,
      "loss": 1.2822,
      "step": 1676
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.298198160243558e-05,
      "loss": 1.3067,
      "step": 1677
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.294406376376771e-05,
      "loss": 1.3409,
      "step": 1678
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.290614326496289e-05,
      "loss": 1.3005,
      "step": 1679
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.286822014028453e-05,
      "loss": 1.317,
      "step": 1680
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.283029442399844e-05,
      "loss": 1.2928,
      "step": 1681
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.2792366150372784e-05,
      "loss": 1.2939,
      "step": 1682
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.2754435353677985e-05,
      "loss": 1.3105,
      "step": 1683
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.27165020681868e-05,
      "loss": 1.3318,
      "step": 1684
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.2678566328174196e-05,
      "loss": 1.3283,
      "step": 1685
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.2640628167917387e-05,
      "loss": 1.3281,
      "step": 1686
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.260268762169575e-05,
      "loss": 1.3179,
      "step": 1687
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.2564744723790835e-05,
      "loss": 1.3296,
      "step": 1688
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.25267995084863e-05,
      "loss": 1.2747,
      "step": 1689
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.248885201006792e-05,
      "loss": 1.352,
      "step": 1690
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.245090226282352e-05,
      "loss": 1.256,
      "step": 1691
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.241295030104295e-05,
      "loss": 1.2635,
      "step": 1692
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.237499615901805e-05,
      "loss": 1.3027,
      "step": 1693
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.2337039871042684e-05,
      "loss": 1.3289,
      "step": 1694
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.2299081471412586e-05,
      "loss": 1.306,
      "step": 1695
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.2261120994425464e-05,
      "loss": 1.2563,
      "step": 1696
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.2223158474380834e-05,
      "loss": 1.303,
      "step": 1697
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.218519394558013e-05,
      "loss": 1.2909,
      "step": 1698
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.214722744232654e-05,
      "loss": 1.2942,
      "step": 1699
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.210925899892507e-05,
      "loss": 1.2985,
      "step": 1700
    },
    {
      "epoch": 0.5,
      "eval_loss": 0.1250637024641037,
      "eval_runtime": 24.3845,
      "eval_samples_per_second": 82.019,
      "eval_steps_per_second": 0.656,
      "step": 1700
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.2071288649682455e-05,
      "loss": 1.3565,
      "step": 1701
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.20333164289072e-05,
      "loss": 1.2961,
      "step": 1702
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.199534237090943e-05,
      "loss": 1.3133,
      "step": 1703
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.1957366510001e-05,
      "loss": 1.3518,
      "step": 1704
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.191938888049534e-05,
      "loss": 1.3327,
      "step": 1705
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.188140951670751e-05,
      "loss": 1.3062,
      "step": 1706
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.184342845295414e-05,
      "loss": 1.3093,
      "step": 1707
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.1805445723553346e-05,
      "loss": 1.3004,
      "step": 1708
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.1767461362824824e-05,
      "loss": 1.3283,
      "step": 1709
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.172947540508969e-05,
      "loss": 1.2971,
      "step": 1710
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.169148788467051e-05,
      "loss": 1.2979,
      "step": 1711
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.1653498835891304e-05,
      "loss": 1.3378,
      "step": 1712
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.1615508293077394e-05,
      "loss": 1.2779,
      "step": 1713
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.157751629055553e-05,
      "loss": 1.3048,
      "step": 1714
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.153952286265375e-05,
      "loss": 1.3023,
      "step": 1715
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.150152804370136e-05,
      "loss": 1.364,
      "step": 1716
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.146353186802895e-05,
      "loss": 1.3194,
      "step": 1717
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.142553436996834e-05,
      "loss": 1.2827,
      "step": 1718
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.1387535583852525e-05,
      "loss": 1.346,
      "step": 1719
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.1349535544015655e-05,
      "loss": 1.2819,
      "step": 1720
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.131153428479306e-05,
      "loss": 1.2768,
      "step": 1721
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.127353184052111e-05,
      "loss": 1.279,
      "step": 1722
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.12355282455373e-05,
      "loss": 1.3018,
      "step": 1723
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.119752353418012e-05,
      "loss": 1.3279,
      "step": 1724
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.1159517740789114e-05,
      "loss": 1.2705,
      "step": 1725
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.1121510899704766e-05,
      "loss": 1.3457,
      "step": 1726
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.1083503045268526e-05,
      "loss": 1.3252,
      "step": 1727
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.1045494211822756e-05,
      "loss": 1.3029,
      "step": 1728
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.1007484433710684e-05,
      "loss": 1.241,
      "step": 1729
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.0969473745276434e-05,
      "loss": 1.3415,
      "step": 1730
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.093146218086491e-05,
      "loss": 1.3393,
      "step": 1731
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.0893449774821814e-05,
      "loss": 1.3265,
      "step": 1732
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.085543656149365e-05,
      "loss": 1.3064,
      "step": 1733
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.081742257522759e-05,
      "loss": 1.3113,
      "step": 1734
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.077940785037156e-05,
      "loss": 1.3173,
      "step": 1735
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.074139242127411e-05,
      "loss": 1.3347,
      "step": 1736
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.070337632228445e-05,
      "loss": 1.2996,
      "step": 1737
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.0665359587752394e-05,
      "loss": 1.2962,
      "step": 1738
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.0627342252028316e-05,
      "loss": 1.2962,
      "step": 1739
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.058932434946314e-05,
      "loss": 1.3119,
      "step": 1740
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.055130591440834e-05,
      "loss": 1.3299,
      "step": 1741
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.051328698121579e-05,
      "loss": 1.3256,
      "step": 1742
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.04752675842379e-05,
      "loss": 1.2831,
      "step": 1743
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.0437247757827417e-05,
      "loss": 1.2794,
      "step": 1744
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.039922753633757e-05,
      "loss": 1.2921,
      "step": 1745
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.036120695412184e-05,
      "loss": 1.3653,
      "step": 1746
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.032318604553414e-05,
      "loss": 1.3592,
      "step": 1747
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.028516484492857e-05,
      "loss": 1.2766,
      "step": 1748
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.02471433866596e-05,
      "loss": 1.348,
      "step": 1749
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.0209121705081845e-05,
      "loss": 1.2858,
      "step": 1750
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.0171099834550185e-05,
      "loss": 1.2677,
      "step": 1751
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.013307780941963e-05,
      "loss": 1.2854,
      "step": 1752
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.009505566404535e-05,
      "loss": 1.2893,
      "step": 1753
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.005703343278262e-05,
      "loss": 1.3607,
      "step": 1754
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.0019011149986774e-05,
      "loss": 1.2593,
      "step": 1755
    },
    {
      "epoch": 0.52,
      "learning_rate": 3.998098885001323e-05,
      "loss": 1.3124,
      "step": 1756
    },
    {
      "epoch": 0.52,
      "learning_rate": 3.9942966567217396e-05,
      "loss": 1.3126,
      "step": 1757
    },
    {
      "epoch": 0.52,
      "learning_rate": 3.990494433595466e-05,
      "loss": 1.2887,
      "step": 1758
    },
    {
      "epoch": 0.52,
      "learning_rate": 3.9866922190580376e-05,
      "loss": 1.3259,
      "step": 1759
    },
    {
      "epoch": 0.52,
      "learning_rate": 3.982890016544982e-05,
      "loss": 1.315,
      "step": 1760
    },
    {
      "epoch": 0.52,
      "learning_rate": 3.979087829491816e-05,
      "loss": 1.3628,
      "step": 1761
    },
    {
      "epoch": 0.52,
      "learning_rate": 3.975285661334041e-05,
      "loss": 1.2928,
      "step": 1762
    },
    {
      "epoch": 0.52,
      "learning_rate": 3.9714835155071435e-05,
      "loss": 1.3065,
      "step": 1763
    },
    {
      "epoch": 0.52,
      "learning_rate": 3.967681395446588e-05,
      "loss": 1.3409,
      "step": 1764
    },
    {
      "epoch": 0.52,
      "learning_rate": 3.963879304587816e-05,
      "loss": 1.3625,
      "step": 1765
    },
    {
      "epoch": 0.52,
      "learning_rate": 3.960077246366244e-05,
      "loss": 1.2532,
      "step": 1766
    },
    {
      "epoch": 0.52,
      "learning_rate": 3.956275224217258e-05,
      "loss": 1.2926,
      "step": 1767
    },
    {
      "epoch": 0.52,
      "learning_rate": 3.952473241576211e-05,
      "loss": 1.2772,
      "step": 1768
    },
    {
      "epoch": 0.52,
      "learning_rate": 3.9486713018784214e-05,
      "loss": 1.3211,
      "step": 1769
    },
    {
      "epoch": 0.52,
      "learning_rate": 3.9448694085591674e-05,
      "loss": 1.3049,
      "step": 1770
    },
    {
      "epoch": 0.52,
      "learning_rate": 3.941067565053685e-05,
      "loss": 1.2892,
      "step": 1771
    },
    {
      "epoch": 0.52,
      "learning_rate": 3.937265774797169e-05,
      "loss": 1.3263,
      "step": 1772
    },
    {
      "epoch": 0.52,
      "learning_rate": 3.933464041224761e-05,
      "loss": 1.2876,
      "step": 1773
    },
    {
      "epoch": 0.52,
      "learning_rate": 3.929662367771556e-05,
      "loss": 1.2873,
      "step": 1774
    },
    {
      "epoch": 0.52,
      "learning_rate": 3.925860757872591e-05,
      "loss": 1.2619,
      "step": 1775
    },
    {
      "epoch": 0.52,
      "learning_rate": 3.922059214962846e-05,
      "loss": 1.3191,
      "step": 1776
    },
    {
      "epoch": 0.52,
      "learning_rate": 3.918257742477243e-05,
      "loss": 1.3316,
      "step": 1777
    },
    {
      "epoch": 0.52,
      "learning_rate": 3.914456343850637e-05,
      "loss": 1.2509,
      "step": 1778
    },
    {
      "epoch": 0.52,
      "learning_rate": 3.9106550225178206e-05,
      "loss": 1.2963,
      "step": 1779
    },
    {
      "epoch": 0.52,
      "learning_rate": 3.906853781913511e-05,
      "loss": 1.3364,
      "step": 1780
    },
    {
      "epoch": 0.52,
      "learning_rate": 3.903052625472359e-05,
      "loss": 1.3181,
      "step": 1781
    },
    {
      "epoch": 0.52,
      "learning_rate": 3.899251556628933e-05,
      "loss": 1.2798,
      "step": 1782
    },
    {
      "epoch": 0.52,
      "learning_rate": 3.895450578817727e-05,
      "loss": 1.2428,
      "step": 1783
    },
    {
      "epoch": 0.52,
      "learning_rate": 3.891649695473149e-05,
      "loss": 1.3046,
      "step": 1784
    },
    {
      "epoch": 0.52,
      "learning_rate": 3.887848910029525e-05,
      "loss": 1.3501,
      "step": 1785
    },
    {
      "epoch": 0.52,
      "learning_rate": 3.88404822592109e-05,
      "loss": 1.2689,
      "step": 1786
    },
    {
      "epoch": 0.52,
      "learning_rate": 3.880247646581989e-05,
      "loss": 1.3381,
      "step": 1787
    },
    {
      "epoch": 0.52,
      "learning_rate": 3.8764471754462714e-05,
      "loss": 1.2986,
      "step": 1788
    },
    {
      "epoch": 0.52,
      "learning_rate": 3.8726468159478904e-05,
      "loss": 1.2904,
      "step": 1789
    },
    {
      "epoch": 0.53,
      "learning_rate": 3.868846571520696e-05,
      "loss": 1.3375,
      "step": 1790
    },
    {
      "epoch": 0.53,
      "learning_rate": 3.865046445598435e-05,
      "loss": 1.3237,
      "step": 1791
    },
    {
      "epoch": 0.53,
      "learning_rate": 3.861246441614749e-05,
      "loss": 1.299,
      "step": 1792
    },
    {
      "epoch": 0.53,
      "learning_rate": 3.857446563003167e-05,
      "loss": 1.3175,
      "step": 1793
    },
    {
      "epoch": 0.53,
      "learning_rate": 3.853646813197106e-05,
      "loss": 1.3175,
      "step": 1794
    },
    {
      "epoch": 0.53,
      "learning_rate": 3.849847195629865e-05,
      "loss": 1.2389,
      "step": 1795
    },
    {
      "epoch": 0.53,
      "learning_rate": 3.846047713734627e-05,
      "loss": 1.2977,
      "step": 1796
    },
    {
      "epoch": 0.53,
      "learning_rate": 3.842248370944447e-05,
      "loss": 1.3138,
      "step": 1797
    },
    {
      "epoch": 0.53,
      "learning_rate": 3.838449170692262e-05,
      "loss": 1.2987,
      "step": 1798
    },
    {
      "epoch": 0.53,
      "learning_rate": 3.8346501164108716e-05,
      "loss": 1.362,
      "step": 1799
    },
    {
      "epoch": 0.53,
      "learning_rate": 3.83085121153295e-05,
      "loss": 1.2754,
      "step": 1800
    },
    {
      "epoch": 0.53,
      "eval_loss": 0.11963999271392822,
      "eval_runtime": 24.3979,
      "eval_samples_per_second": 81.974,
      "eval_steps_per_second": 0.656,
      "step": 1800
    },
    {
      "epoch": 0.53,
      "learning_rate": 3.8270524594910315e-05,
      "loss": 1.3175,
      "step": 1801
    },
    {
      "epoch": 0.53,
      "learning_rate": 3.823253863717519e-05,
      "loss": 1.3207,
      "step": 1802
    },
    {
      "epoch": 0.53,
      "learning_rate": 3.819455427644666e-05,
      "loss": 1.2735,
      "step": 1803
    },
    {
      "epoch": 0.53,
      "learning_rate": 3.8156571547045875e-05,
      "loss": 1.2781,
      "step": 1804
    },
    {
      "epoch": 0.53,
      "learning_rate": 3.811859048329249e-05,
      "loss": 1.2988,
      "step": 1805
    },
    {
      "epoch": 0.53,
      "learning_rate": 3.8080611119504666e-05,
      "loss": 1.3627,
      "step": 1806
    },
    {
      "epoch": 0.53,
      "learning_rate": 3.8042633489999e-05,
      "loss": 1.2819,
      "step": 1807
    },
    {
      "epoch": 0.53,
      "learning_rate": 3.800465762909057e-05,
      "loss": 1.3025,
      "step": 1808
    },
    {
      "epoch": 0.53,
      "learning_rate": 3.7966683571092805e-05,
      "loss": 1.2675,
      "step": 1809
    },
    {
      "epoch": 0.53,
      "learning_rate": 3.7928711350317545e-05,
      "loss": 1.3414,
      "step": 1810
    },
    {
      "epoch": 0.53,
      "learning_rate": 3.789074100107494e-05,
      "loss": 1.2875,
      "step": 1811
    },
    {
      "epoch": 0.53,
      "learning_rate": 3.7852772557673477e-05,
      "loss": 1.2895,
      "step": 1812
    },
    {
      "epoch": 0.53,
      "learning_rate": 3.781480605441989e-05,
      "loss": 1.2693,
      "step": 1813
    },
    {
      "epoch": 0.53,
      "learning_rate": 3.777684152561917e-05,
      "loss": 1.3053,
      "step": 1814
    },
    {
      "epoch": 0.53,
      "learning_rate": 3.7738879005574556e-05,
      "loss": 1.3055,
      "step": 1815
    },
    {
      "epoch": 0.53,
      "learning_rate": 3.770091852858743e-05,
      "loss": 1.2972,
      "step": 1816
    },
    {
      "epoch": 0.53,
      "learning_rate": 3.7662960128957336e-05,
      "loss": 1.3423,
      "step": 1817
    },
    {
      "epoch": 0.53,
      "learning_rate": 3.7625003840981976e-05,
      "loss": 1.2805,
      "step": 1818
    },
    {
      "epoch": 0.53,
      "learning_rate": 3.758704969895708e-05,
      "loss": 1.2707,
      "step": 1819
    },
    {
      "epoch": 0.53,
      "learning_rate": 3.75490977371765e-05,
      "loss": 1.2824,
      "step": 1820
    },
    {
      "epoch": 0.53,
      "learning_rate": 3.751114798993209e-05,
      "loss": 1.3229,
      "step": 1821
    },
    {
      "epoch": 0.53,
      "learning_rate": 3.7473200491513716e-05,
      "loss": 1.2698,
      "step": 1822
    },
    {
      "epoch": 0.53,
      "learning_rate": 3.743525527620918e-05,
      "loss": 1.2623,
      "step": 1823
    },
    {
      "epoch": 0.54,
      "learning_rate": 3.739731237830427e-05,
      "loss": 1.2989,
      "step": 1824
    },
    {
      "epoch": 0.54,
      "learning_rate": 3.735937183208263e-05,
      "loss": 1.3364,
      "step": 1825
    },
    {
      "epoch": 0.54,
      "learning_rate": 3.732143367182581e-05,
      "loss": 1.2686,
      "step": 1826
    },
    {
      "epoch": 0.54,
      "learning_rate": 3.728349793181321e-05,
      "loss": 1.2572,
      "step": 1827
    },
    {
      "epoch": 0.54,
      "learning_rate": 3.724556464632203e-05,
      "loss": 1.2923,
      "step": 1828
    },
    {
      "epoch": 0.54,
      "learning_rate": 3.720763384962723e-05,
      "loss": 1.3484,
      "step": 1829
    },
    {
      "epoch": 0.54,
      "learning_rate": 3.7169705576001565e-05,
      "loss": 1.3086,
      "step": 1830
    },
    {
      "epoch": 0.54,
      "learning_rate": 3.7131779859715475e-05,
      "loss": 1.2899,
      "step": 1831
    },
    {
      "epoch": 0.54,
      "learning_rate": 3.7093856735037126e-05,
      "loss": 1.3417,
      "step": 1832
    },
    {
      "epoch": 0.54,
      "learning_rate": 3.7055936236232296e-05,
      "loss": 1.3113,
      "step": 1833
    },
    {
      "epoch": 0.54,
      "learning_rate": 3.701801839756443e-05,
      "loss": 1.2167,
      "step": 1834
    },
    {
      "epoch": 0.54,
      "learning_rate": 3.698010325329454e-05,
      "loss": 1.2844,
      "step": 1835
    },
    {
      "epoch": 0.54,
      "learning_rate": 3.6942190837681235e-05,
      "loss": 1.267,
      "step": 1836
    },
    {
      "epoch": 0.54,
      "learning_rate": 3.690428118498062e-05,
      "loss": 1.3066,
      "step": 1837
    },
    {
      "epoch": 0.54,
      "learning_rate": 3.686637432944634e-05,
      "loss": 1.3051,
      "step": 1838
    },
    {
      "epoch": 0.54,
      "learning_rate": 3.6828470305329485e-05,
      "loss": 1.2396,
      "step": 1839
    },
    {
      "epoch": 0.54,
      "learning_rate": 3.679056914687861e-05,
      "loss": 1.3507,
      "step": 1840
    },
    {
      "epoch": 0.54,
      "learning_rate": 3.675267088833966e-05,
      "loss": 1.2689,
      "step": 1841
    },
    {
      "epoch": 0.54,
      "learning_rate": 3.6714775563955974e-05,
      "loss": 1.295,
      "step": 1842
    },
    {
      "epoch": 0.54,
      "learning_rate": 3.6676883207968226e-05,
      "loss": 1.2977,
      "step": 1843
    },
    {
      "epoch": 0.54,
      "learning_rate": 3.6638993854614434e-05,
      "loss": 1.2781,
      "step": 1844
    },
    {
      "epoch": 0.54,
      "learning_rate": 3.660110753812986e-05,
      "loss": 1.2672,
      "step": 1845
    },
    {
      "epoch": 0.54,
      "learning_rate": 3.656322429274708e-05,
      "loss": 1.3403,
      "step": 1846
    },
    {
      "epoch": 0.54,
      "learning_rate": 3.652534415269583e-05,
      "loss": 1.2922,
      "step": 1847
    },
    {
      "epoch": 0.54,
      "learning_rate": 3.648746715220309e-05,
      "loss": 1.3216,
      "step": 1848
    },
    {
      "epoch": 0.54,
      "learning_rate": 3.644959332549299e-05,
      "loss": 1.2824,
      "step": 1849
    },
    {
      "epoch": 0.54,
      "learning_rate": 3.6411722706786784e-05,
      "loss": 1.2775,
      "step": 1850
    },
    {
      "epoch": 0.54,
      "learning_rate": 3.637385533030282e-05,
      "loss": 1.3366,
      "step": 1851
    },
    {
      "epoch": 0.54,
      "learning_rate": 3.6335991230256536e-05,
      "loss": 1.2807,
      "step": 1852
    },
    {
      "epoch": 0.54,
      "learning_rate": 3.6298130440860394e-05,
      "loss": 1.2819,
      "step": 1853
    },
    {
      "epoch": 0.54,
      "learning_rate": 3.626027299632389e-05,
      "loss": 1.3051,
      "step": 1854
    },
    {
      "epoch": 0.54,
      "learning_rate": 3.6222418930853474e-05,
      "loss": 1.3106,
      "step": 1855
    },
    {
      "epoch": 0.54,
      "learning_rate": 3.618456827865254e-05,
      "loss": 1.2978,
      "step": 1856
    },
    {
      "epoch": 0.54,
      "learning_rate": 3.614672107392141e-05,
      "loss": 1.3144,
      "step": 1857
    },
    {
      "epoch": 0.55,
      "learning_rate": 3.6108877350857296e-05,
      "loss": 1.2938,
      "step": 1858
    },
    {
      "epoch": 0.55,
      "learning_rate": 3.6071037143654235e-05,
      "loss": 1.3015,
      "step": 1859
    },
    {
      "epoch": 0.55,
      "learning_rate": 3.603320048650314e-05,
      "loss": 1.257,
      "step": 1860
    },
    {
      "epoch": 0.55,
      "learning_rate": 3.599536741359165e-05,
      "loss": 1.2707,
      "step": 1861
    },
    {
      "epoch": 0.55,
      "learning_rate": 3.5957537959104226e-05,
      "loss": 1.3495,
      "step": 1862
    },
    {
      "epoch": 0.55,
      "learning_rate": 3.5919712157222014e-05,
      "loss": 1.2192,
      "step": 1863
    },
    {
      "epoch": 0.55,
      "learning_rate": 3.58818900421229e-05,
      "loss": 1.2999,
      "step": 1864
    },
    {
      "epoch": 0.55,
      "learning_rate": 3.5844071647981396e-05,
      "loss": 1.3092,
      "step": 1865
    },
    {
      "epoch": 0.55,
      "learning_rate": 3.580625700896869e-05,
      "loss": 1.3539,
      "step": 1866
    },
    {
      "epoch": 0.55,
      "learning_rate": 3.576844615925255e-05,
      "loss": 1.2931,
      "step": 1867
    },
    {
      "epoch": 0.55,
      "learning_rate": 3.573063913299733e-05,
      "loss": 1.3017,
      "step": 1868
    },
    {
      "epoch": 0.55,
      "learning_rate": 3.5692835964363945e-05,
      "loss": 1.285,
      "step": 1869
    },
    {
      "epoch": 0.55,
      "learning_rate": 3.56550366875098e-05,
      "loss": 1.2721,
      "step": 1870
    },
    {
      "epoch": 0.55,
      "learning_rate": 3.561724133658878e-05,
      "loss": 1.3359,
      "step": 1871
    },
    {
      "epoch": 0.55,
      "learning_rate": 3.557944994575125e-05,
      "loss": 1.2648,
      "step": 1872
    },
    {
      "epoch": 0.55,
      "learning_rate": 3.554166254914399e-05,
      "loss": 1.3089,
      "step": 1873
    },
    {
      "epoch": 0.55,
      "learning_rate": 3.5503879180910144e-05,
      "loss": 1.2983,
      "step": 1874
    },
    {
      "epoch": 0.55,
      "learning_rate": 3.546609987518925e-05,
      "loss": 1.2959,
      "step": 1875
    },
    {
      "epoch": 0.55,
      "learning_rate": 3.542832466611715e-05,
      "loss": 1.3089,
      "step": 1876
    },
    {
      "epoch": 0.55,
      "learning_rate": 3.539055358782601e-05,
      "loss": 1.3284,
      "step": 1877
    },
    {
      "epoch": 0.55,
      "learning_rate": 3.535278667444423e-05,
      "loss": 1.2928,
      "step": 1878
    },
    {
      "epoch": 0.55,
      "learning_rate": 3.531502396009648e-05,
      "loss": 1.2307,
      "step": 1879
    },
    {
      "epoch": 0.55,
      "learning_rate": 3.527726547890362e-05,
      "loss": 1.3063,
      "step": 1880
    },
    {
      "epoch": 0.55,
      "learning_rate": 3.523951126498268e-05,
      "loss": 1.3395,
      "step": 1881
    },
    {
      "epoch": 0.55,
      "learning_rate": 3.5201761352446844e-05,
      "loss": 1.268,
      "step": 1882
    },
    {
      "epoch": 0.55,
      "learning_rate": 3.5164015775405414e-05,
      "loss": 1.2714,
      "step": 1883
    },
    {
      "epoch": 0.55,
      "learning_rate": 3.512627456796375e-05,
      "loss": 1.2705,
      "step": 1884
    },
    {
      "epoch": 0.55,
      "learning_rate": 3.5088537764223295e-05,
      "loss": 1.3012,
      "step": 1885
    },
    {
      "epoch": 0.55,
      "learning_rate": 3.505080539828149e-05,
      "loss": 1.2715,
      "step": 1886
    },
    {
      "epoch": 0.55,
      "learning_rate": 3.50130775042318e-05,
      "loss": 1.3258,
      "step": 1887
    },
    {
      "epoch": 0.55,
      "learning_rate": 3.4975354116163594e-05,
      "loss": 1.2739,
      "step": 1888
    },
    {
      "epoch": 0.55,
      "learning_rate": 3.49376352681622e-05,
      "loss": 1.2802,
      "step": 1889
    },
    {
      "epoch": 0.55,
      "learning_rate": 3.489992099430886e-05,
      "loss": 1.3022,
      "step": 1890
    },
    {
      "epoch": 0.55,
      "learning_rate": 3.4862211328680655e-05,
      "loss": 1.306,
      "step": 1891
    },
    {
      "epoch": 0.56,
      "learning_rate": 3.482450630535052e-05,
      "loss": 1.3272,
      "step": 1892
    },
    {
      "epoch": 0.56,
      "learning_rate": 3.478680595838719e-05,
      "loss": 1.3474,
      "step": 1893
    },
    {
      "epoch": 0.56,
      "learning_rate": 3.4749110321855176e-05,
      "loss": 1.2565,
      "step": 1894
    },
    {
      "epoch": 0.56,
      "learning_rate": 3.471141942981472e-05,
      "loss": 1.2476,
      "step": 1895
    },
    {
      "epoch": 0.56,
      "learning_rate": 3.4673733316321794e-05,
      "loss": 1.27,
      "step": 1896
    },
    {
      "epoch": 0.56,
      "learning_rate": 3.4636052015428055e-05,
      "loss": 1.3043,
      "step": 1897
    },
    {
      "epoch": 0.56,
      "learning_rate": 3.4598375561180783e-05,
      "loss": 1.2715,
      "step": 1898
    },
    {
      "epoch": 0.56,
      "learning_rate": 3.4560703987622924e-05,
      "loss": 1.2638,
      "step": 1899
    },
    {
      "epoch": 0.56,
      "learning_rate": 3.452303732879296e-05,
      "loss": 1.2906,
      "step": 1900
    },
    {
      "epoch": 0.56,
      "eval_loss": 0.12069112062454224,
      "eval_runtime": 25.2088,
      "eval_samples_per_second": 79.337,
      "eval_steps_per_second": 0.635,
      "step": 1900
    },
    {
      "epoch": 0.56,
      "learning_rate": 3.4485375618724986e-05,
      "loss": 1.2865,
      "step": 1901
    },
    {
      "epoch": 0.56,
      "learning_rate": 3.444771889144858e-05,
      "loss": 1.2703,
      "step": 1902
    },
    {
      "epoch": 0.56,
      "learning_rate": 3.441006718098885e-05,
      "loss": 1.2536,
      "step": 1903
    },
    {
      "epoch": 0.56,
      "learning_rate": 3.4372420521366354e-05,
      "loss": 1.3122,
      "step": 1904
    },
    {
      "epoch": 0.56,
      "learning_rate": 3.43347789465971e-05,
      "loss": 1.273,
      "step": 1905
    },
    {
      "epoch": 0.56,
      "learning_rate": 3.429714249069248e-05,
      "loss": 1.2504,
      "step": 1906
    },
    {
      "epoch": 0.56,
      "learning_rate": 3.4259511187659285e-05,
      "loss": 1.2751,
      "step": 1907
    },
    {
      "epoch": 0.56,
      "learning_rate": 3.422188507149962e-05,
      "loss": 1.267,
      "step": 1908
    },
    {
      "epoch": 0.56,
      "learning_rate": 3.4184264176210946e-05,
      "loss": 1.2478,
      "step": 1909
    },
    {
      "epoch": 0.56,
      "learning_rate": 3.414664853578597e-05,
      "loss": 1.3067,
      "step": 1910
    },
    {
      "epoch": 0.56,
      "learning_rate": 3.410903818421266e-05,
      "loss": 1.2967,
      "step": 1911
    },
    {
      "epoch": 0.56,
      "learning_rate": 3.4071433155474216e-05,
      "loss": 1.2767,
      "step": 1912
    },
    {
      "epoch": 0.56,
      "learning_rate": 3.403383348354902e-05,
      "loss": 1.2698,
      "step": 1913
    },
    {
      "epoch": 0.56,
      "learning_rate": 3.3996239202410617e-05,
      "loss": 1.2983,
      "step": 1914
    },
    {
      "epoch": 0.56,
      "learning_rate": 3.395865034602768e-05,
      "loss": 1.3113,
      "step": 1915
    },
    {
      "epoch": 0.56,
      "learning_rate": 3.3921066948363956e-05,
      "loss": 1.3003,
      "step": 1916
    },
    {
      "epoch": 0.56,
      "learning_rate": 3.3883489043378324e-05,
      "loss": 1.2903,
      "step": 1917
    },
    {
      "epoch": 0.56,
      "learning_rate": 3.3845916665024626e-05,
      "loss": 1.274,
      "step": 1918
    },
    {
      "epoch": 0.56,
      "learning_rate": 3.380834984725176e-05,
      "loss": 1.3062,
      "step": 1919
    },
    {
      "epoch": 0.56,
      "learning_rate": 3.3770788624003574e-05,
      "loss": 1.2646,
      "step": 1920
    },
    {
      "epoch": 0.56,
      "learning_rate": 3.373323302921888e-05,
      "loss": 1.2929,
      "step": 1921
    },
    {
      "epoch": 0.56,
      "learning_rate": 3.3695683096831394e-05,
      "loss": 1.2956,
      "step": 1922
    },
    {
      "epoch": 0.56,
      "learning_rate": 3.36581388607697e-05,
      "loss": 1.2732,
      "step": 1923
    },
    {
      "epoch": 0.56,
      "learning_rate": 3.3620600354957276e-05,
      "loss": 1.2567,
      "step": 1924
    },
    {
      "epoch": 0.56,
      "learning_rate": 3.3583067613312396e-05,
      "loss": 1.2301,
      "step": 1925
    },
    {
      "epoch": 0.57,
      "learning_rate": 3.354554066974809e-05,
      "loss": 1.3254,
      "step": 1926
    },
    {
      "epoch": 0.57,
      "learning_rate": 3.350801955817223e-05,
      "loss": 1.2792,
      "step": 1927
    },
    {
      "epoch": 0.57,
      "learning_rate": 3.347050431248735e-05,
      "loss": 1.2685,
      "step": 1928
    },
    {
      "epoch": 0.57,
      "learning_rate": 3.3432994966590726e-05,
      "loss": 1.2609,
      "step": 1929
    },
    {
      "epoch": 0.57,
      "learning_rate": 3.3395491554374266e-05,
      "loss": 1.3009,
      "step": 1930
    },
    {
      "epoch": 0.57,
      "learning_rate": 3.335799410972456e-05,
      "loss": 1.257,
      "step": 1931
    },
    {
      "epoch": 0.57,
      "learning_rate": 3.332050266652276e-05,
      "loss": 1.2878,
      "step": 1932
    },
    {
      "epoch": 0.57,
      "learning_rate": 3.3283017258644634e-05,
      "loss": 1.264,
      "step": 1933
    },
    {
      "epoch": 0.57,
      "learning_rate": 3.324553791996048e-05,
      "loss": 1.3178,
      "step": 1934
    },
    {
      "epoch": 0.57,
      "learning_rate": 3.32080646843351e-05,
      "loss": 1.3082,
      "step": 1935
    },
    {
      "epoch": 0.57,
      "learning_rate": 3.31705975856278e-05,
      "loss": 1.2647,
      "step": 1936
    },
    {
      "epoch": 0.57,
      "learning_rate": 3.313313665769235e-05,
      "loss": 1.2802,
      "step": 1937
    },
    {
      "epoch": 0.57,
      "learning_rate": 3.30956819343769e-05,
      "loss": 1.2614,
      "step": 1938
    },
    {
      "epoch": 0.57,
      "learning_rate": 3.3058233449524056e-05,
      "loss": 1.3316,
      "step": 1939
    },
    {
      "epoch": 0.57,
      "learning_rate": 3.302079123697073e-05,
      "loss": 1.3094,
      "step": 1940
    },
    {
      "epoch": 0.57,
      "learning_rate": 3.298335533054821e-05,
      "loss": 1.2493,
      "step": 1941
    },
    {
      "epoch": 0.57,
      "learning_rate": 3.294592576408204e-05,
      "loss": 1.25,
      "step": 1942
    },
    {
      "epoch": 0.57,
      "learning_rate": 3.290850257139209e-05,
      "loss": 1.2832,
      "step": 1943
    },
    {
      "epoch": 0.57,
      "learning_rate": 3.2871085786292435e-05,
      "loss": 1.2659,
      "step": 1944
    },
    {
      "epoch": 0.57,
      "learning_rate": 3.2833675442591366e-05,
      "loss": 1.3033,
      "step": 1945
    },
    {
      "epoch": 0.57,
      "learning_rate": 3.279627157409136e-05,
      "loss": 1.2674,
      "step": 1946
    },
    {
      "epoch": 0.57,
      "learning_rate": 3.275887421458903e-05,
      "loss": 1.2916,
      "step": 1947
    },
    {
      "epoch": 0.57,
      "learning_rate": 3.2721483397875135e-05,
      "loss": 1.2843,
      "step": 1948
    },
    {
      "epoch": 0.57,
      "learning_rate": 3.268409915773449e-05,
      "loss": 1.3228,
      "step": 1949
    },
    {
      "epoch": 0.57,
      "learning_rate": 3.2646721527945984e-05,
      "loss": 1.2859,
      "step": 1950
    },
    {
      "epoch": 0.57,
      "learning_rate": 3.260935054228254e-05,
      "loss": 1.2659,
      "step": 1951
    },
    {
      "epoch": 0.57,
      "learning_rate": 3.257198623451107e-05,
      "loss": 1.2488,
      "step": 1952
    },
    {
      "epoch": 0.57,
      "learning_rate": 3.253462863839243e-05,
      "loss": 1.3218,
      "step": 1953
    },
    {
      "epoch": 0.57,
      "learning_rate": 3.2497277787681466e-05,
      "loss": 1.287,
      "step": 1954
    },
    {
      "epoch": 0.57,
      "learning_rate": 3.2459933716126866e-05,
      "loss": 1.2386,
      "step": 1955
    },
    {
      "epoch": 0.57,
      "learning_rate": 3.2422596457471245e-05,
      "loss": 1.3374,
      "step": 1956
    },
    {
      "epoch": 0.57,
      "learning_rate": 3.238526604545103e-05,
      "loss": 1.2671,
      "step": 1957
    },
    {
      "epoch": 0.57,
      "learning_rate": 3.2347942513796475e-05,
      "loss": 1.3083,
      "step": 1958
    },
    {
      "epoch": 0.57,
      "learning_rate": 3.2310625896231606e-05,
      "loss": 1.2443,
      "step": 1959
    },
    {
      "epoch": 0.58,
      "learning_rate": 3.227331622647422e-05,
      "loss": 1.3063,
      "step": 1960
    },
    {
      "epoch": 0.58,
      "learning_rate": 3.223601353823582e-05,
      "loss": 1.2786,
      "step": 1961
    },
    {
      "epoch": 0.58,
      "learning_rate": 3.219871786522161e-05,
      "loss": 1.255,
      "step": 1962
    },
    {
      "epoch": 0.58,
      "learning_rate": 3.216142924113046e-05,
      "loss": 1.2737,
      "step": 1963
    },
    {
      "epoch": 0.58,
      "learning_rate": 3.2124147699654834e-05,
      "loss": 1.3497,
      "step": 1964
    },
    {
      "epoch": 0.58,
      "learning_rate": 3.208687327448085e-05,
      "loss": 1.2816,
      "step": 1965
    },
    {
      "epoch": 0.58,
      "learning_rate": 3.204960599928816e-05,
      "loss": 1.2894,
      "step": 1966
    },
    {
      "epoch": 0.58,
      "learning_rate": 3.2012345907749985e-05,
      "loss": 1.3062,
      "step": 1967
    },
    {
      "epoch": 0.58,
      "learning_rate": 3.1975093033533003e-05,
      "loss": 1.3122,
      "step": 1968
    },
    {
      "epoch": 0.58,
      "learning_rate": 3.193784741029745e-05,
      "loss": 1.3216,
      "step": 1969
    },
    {
      "epoch": 0.58,
      "learning_rate": 3.190060907169692e-05,
      "loss": 1.3057,
      "step": 1970
    },
    {
      "epoch": 0.58,
      "learning_rate": 3.18633780513785e-05,
      "loss": 1.2744,
      "step": 1971
    },
    {
      "epoch": 0.58,
      "learning_rate": 3.1826154382982625e-05,
      "loss": 1.3115,
      "step": 1972
    },
    {
      "epoch": 0.58,
      "learning_rate": 3.1788938100143086e-05,
      "loss": 1.3188,
      "step": 1973
    },
    {
      "epoch": 0.58,
      "learning_rate": 3.175172923648703e-05,
      "loss": 1.3241,
      "step": 1974
    },
    {
      "epoch": 0.58,
      "learning_rate": 3.171452782563486e-05,
      "loss": 1.2751,
      "step": 1975
    },
    {
      "epoch": 0.58,
      "learning_rate": 3.167733390120029e-05,
      "loss": 1.256,
      "step": 1976
    },
    {
      "epoch": 0.58,
      "learning_rate": 3.164014749679023e-05,
      "loss": 1.2658,
      "step": 1977
    },
    {
      "epoch": 0.58,
      "learning_rate": 3.160296864600482e-05,
      "loss": 1.2596,
      "step": 1978
    },
    {
      "epoch": 0.58,
      "learning_rate": 3.156579738243735e-05,
      "loss": 1.2895,
      "step": 1979
    },
    {
      "epoch": 0.58,
      "learning_rate": 3.152863373967429e-05,
      "loss": 1.24,
      "step": 1980
    },
    {
      "epoch": 0.58,
      "learning_rate": 3.149147775129519e-05,
      "loss": 1.2583,
      "step": 1981
    },
    {
      "epoch": 0.58,
      "learning_rate": 3.1454329450872725e-05,
      "loss": 1.2948,
      "step": 1982
    },
    {
      "epoch": 0.58,
      "learning_rate": 3.1417188871972554e-05,
      "loss": 1.3102,
      "step": 1983
    },
    {
      "epoch": 0.58,
      "learning_rate": 3.1380056048153445e-05,
      "loss": 1.2686,
      "step": 1984
    },
    {
      "epoch": 0.58,
      "learning_rate": 3.1342931012967084e-05,
      "loss": 1.32,
      "step": 1985
    },
    {
      "epoch": 0.58,
      "learning_rate": 3.1305813799958166e-05,
      "loss": 1.2893,
      "step": 1986
    },
    {
      "epoch": 0.58,
      "learning_rate": 3.12687044426643e-05,
      "loss": 1.245,
      "step": 1987
    },
    {
      "epoch": 0.58,
      "learning_rate": 3.123160297461601e-05,
      "loss": 1.2896,
      "step": 1988
    },
    {
      "epoch": 0.58,
      "learning_rate": 3.1194509429336666e-05,
      "loss": 1.2341,
      "step": 1989
    },
    {
      "epoch": 0.58,
      "learning_rate": 3.1157423840342516e-05,
      "loss": 1.2995,
      "step": 1990
    },
    {
      "epoch": 0.58,
      "learning_rate": 3.112034624114257e-05,
      "loss": 1.2572,
      "step": 1991
    },
    {
      "epoch": 0.58,
      "learning_rate": 3.108327666523867e-05,
      "loss": 1.2802,
      "step": 1992
    },
    {
      "epoch": 0.58,
      "learning_rate": 3.104621514612538e-05,
      "loss": 1.273,
      "step": 1993
    },
    {
      "epoch": 0.59,
      "learning_rate": 3.100916171728999e-05,
      "loss": 1.2993,
      "step": 1994
    },
    {
      "epoch": 0.59,
      "learning_rate": 3.097211641221248e-05,
      "loss": 1.278,
      "step": 1995
    },
    {
      "epoch": 0.59,
      "learning_rate": 3.0935079264365485e-05,
      "loss": 1.2976,
      "step": 1996
    },
    {
      "epoch": 0.59,
      "learning_rate": 3.089805030721428e-05,
      "loss": 1.2468,
      "step": 1997
    },
    {
      "epoch": 0.59,
      "learning_rate": 3.086102957421672e-05,
      "loss": 1.2535,
      "step": 1998
    },
    {
      "epoch": 0.59,
      "learning_rate": 3.082401709882327e-05,
      "loss": 1.2311,
      "step": 1999
    },
    {
      "epoch": 0.59,
      "learning_rate": 3.078701291447689e-05,
      "loss": 1.2886,
      "step": 2000
    },
    {
      "epoch": 0.59,
      "eval_loss": 0.11948389559984207,
      "eval_runtime": 24.5965,
      "eval_samples_per_second": 81.312,
      "eval_steps_per_second": 0.65,
      "step": 2000
    },
    {
      "epoch": 0.59,
      "learning_rate": 3.075001705461305e-05,
      "loss": 1.335,
      "step": 2001
    },
    {
      "epoch": 0.59,
      "learning_rate": 3.071302955265973e-05,
      "loss": 1.2889,
      "step": 2002
    },
    {
      "epoch": 0.59,
      "learning_rate": 3.0676050442037336e-05,
      "loss": 1.327,
      "step": 2003
    },
    {
      "epoch": 0.59,
      "learning_rate": 3.06390797561587e-05,
      "loss": 1.2386,
      "step": 2004
    },
    {
      "epoch": 0.59,
      "learning_rate": 3.060211752842903e-05,
      "loss": 1.3083,
      "step": 2005
    },
    {
      "epoch": 0.59,
      "learning_rate": 3.0565163792245917e-05,
      "loss": 1.257,
      "step": 2006
    },
    {
      "epoch": 0.59,
      "learning_rate": 3.052821858099924e-05,
      "loss": 1.2566,
      "step": 2007
    },
    {
      "epoch": 0.59,
      "learning_rate": 3.0491281928071217e-05,
      "loss": 1.2586,
      "step": 2008
    },
    {
      "epoch": 0.59,
      "learning_rate": 3.0454353866836302e-05,
      "loss": 1.312,
      "step": 2009
    },
    {
      "epoch": 0.59,
      "learning_rate": 3.0417434430661208e-05,
      "loss": 1.2886,
      "step": 2010
    },
    {
      "epoch": 0.59,
      "learning_rate": 3.0380523652904835e-05,
      "loss": 1.2788,
      "step": 2011
    },
    {
      "epoch": 0.59,
      "learning_rate": 3.034362156691828e-05,
      "loss": 1.2422,
      "step": 2012
    },
    {
      "epoch": 0.59,
      "learning_rate": 3.030672820604476e-05,
      "loss": 1.3027,
      "step": 2013
    },
    {
      "epoch": 0.59,
      "learning_rate": 3.0269843603619643e-05,
      "loss": 1.2645,
      "step": 2014
    },
    {
      "epoch": 0.59,
      "learning_rate": 3.0232967792970344e-05,
      "loss": 1.2768,
      "step": 2015
    },
    {
      "epoch": 0.59,
      "learning_rate": 3.019610080741637e-05,
      "loss": 1.2942,
      "step": 2016
    },
    {
      "epoch": 0.59,
      "learning_rate": 3.015924268026923e-05,
      "loss": 1.3036,
      "step": 2017
    },
    {
      "epoch": 0.59,
      "learning_rate": 3.012239344483244e-05,
      "loss": 1.2268,
      "step": 2018
    },
    {
      "epoch": 0.59,
      "learning_rate": 3.0085553134401476e-05,
      "loss": 1.2829,
      "step": 2019
    },
    {
      "epoch": 0.59,
      "learning_rate": 3.004872178226376e-05,
      "loss": 1.2761,
      "step": 2020
    },
    {
      "epoch": 0.59,
      "learning_rate": 3.0011899421698613e-05,
      "loss": 1.2684,
      "step": 2021
    },
    {
      "epoch": 0.59,
      "learning_rate": 2.9975086085977207e-05,
      "loss": 1.244,
      "step": 2022
    },
    {
      "epoch": 0.59,
      "learning_rate": 2.993828180836262e-05,
      "loss": 1.2528,
      "step": 2023
    },
    {
      "epoch": 0.59,
      "learning_rate": 2.9901486622109677e-05,
      "loss": 1.2759,
      "step": 2024
    },
    {
      "epoch": 0.59,
      "learning_rate": 2.9864700560465044e-05,
      "loss": 1.2332,
      "step": 2025
    },
    {
      "epoch": 0.59,
      "learning_rate": 2.98279236566671e-05,
      "loss": 1.2944,
      "step": 2026
    },
    {
      "epoch": 0.59,
      "learning_rate": 2.9791155943945985e-05,
      "loss": 1.2872,
      "step": 2027
    },
    {
      "epoch": 0.59,
      "learning_rate": 2.9754397455523497e-05,
      "loss": 1.3043,
      "step": 2028
    },
    {
      "epoch": 0.6,
      "learning_rate": 2.9717648224613147e-05,
      "loss": 1.2353,
      "step": 2029
    },
    {
      "epoch": 0.6,
      "learning_rate": 2.9680908284420024e-05,
      "loss": 1.2279,
      "step": 2030
    },
    {
      "epoch": 0.6,
      "learning_rate": 2.9644177668140873e-05,
      "loss": 1.3347,
      "step": 2031
    },
    {
      "epoch": 0.6,
      "learning_rate": 2.9607456408963987e-05,
      "loss": 1.262,
      "step": 2032
    },
    {
      "epoch": 0.6,
      "learning_rate": 2.957074454006921e-05,
      "loss": 1.2914,
      "step": 2033
    },
    {
      "epoch": 0.6,
      "learning_rate": 2.95340420946279e-05,
      "loss": 1.3127,
      "step": 2034
    },
    {
      "epoch": 0.6,
      "learning_rate": 2.949734910580291e-05,
      "loss": 1.2757,
      "step": 2035
    },
    {
      "epoch": 0.6,
      "learning_rate": 2.946066560674853e-05,
      "loss": 1.2544,
      "step": 2036
    },
    {
      "epoch": 0.6,
      "learning_rate": 2.9423991630610503e-05,
      "loss": 1.2869,
      "step": 2037
    },
    {
      "epoch": 0.6,
      "learning_rate": 2.9387327210525942e-05,
      "loss": 1.285,
      "step": 2038
    },
    {
      "epoch": 0.6,
      "learning_rate": 2.9350672379623318e-05,
      "loss": 1.2906,
      "step": 2039
    },
    {
      "epoch": 0.6,
      "learning_rate": 2.9314027171022482e-05,
      "loss": 1.315,
      "step": 2040
    },
    {
      "epoch": 0.6,
      "learning_rate": 2.9277391617834533e-05,
      "loss": 1.2116,
      "step": 2041
    },
    {
      "epoch": 0.6,
      "learning_rate": 2.9240765753161904e-05,
      "loss": 1.3029,
      "step": 2042
    },
    {
      "epoch": 0.6,
      "learning_rate": 2.9204149610098223e-05,
      "loss": 1.2744,
      "step": 2043
    },
    {
      "epoch": 0.6,
      "learning_rate": 2.9167543221728373e-05,
      "loss": 1.2328,
      "step": 2044
    },
    {
      "epoch": 0.6,
      "learning_rate": 2.9130946621128395e-05,
      "loss": 1.2368,
      "step": 2045
    },
    {
      "epoch": 0.6,
      "learning_rate": 2.9094359841365496e-05,
      "loss": 1.2629,
      "step": 2046
    },
    {
      "epoch": 0.6,
      "learning_rate": 2.9057782915498026e-05,
      "loss": 1.308,
      "step": 2047
    },
    {
      "epoch": 0.6,
      "learning_rate": 2.902121587657539e-05,
      "loss": 1.23,
      "step": 2048
    },
    {
      "epoch": 0.6,
      "learning_rate": 2.898465875763812e-05,
      "loss": 1.2787,
      "step": 2049
    },
    {
      "epoch": 0.6,
      "learning_rate": 2.8948111591717713e-05,
      "loss": 1.3179,
      "step": 2050
    },
    {
      "epoch": 0.6,
      "learning_rate": 2.891157441183674e-05,
      "loss": 1.2561,
      "step": 2051
    },
    {
      "epoch": 0.6,
      "learning_rate": 2.8875047251008702e-05,
      "loss": 1.273,
      "step": 2052
    },
    {
      "epoch": 0.6,
      "learning_rate": 2.8838530142238076e-05,
      "loss": 1.2504,
      "step": 2053
    },
    {
      "epoch": 0.6,
      "learning_rate": 2.8802023118520227e-05,
      "loss": 1.2608,
      "step": 2054
    },
    {
      "epoch": 0.6,
      "learning_rate": 2.8765526212841445e-05,
      "loss": 1.2993,
      "step": 2055
    },
    {
      "epoch": 0.6,
      "learning_rate": 2.8729039458178846e-05,
      "loss": 1.2809,
      "step": 2056
    },
    {
      "epoch": 0.6,
      "learning_rate": 2.86925628875004e-05,
      "loss": 1.2688,
      "step": 2057
    },
    {
      "epoch": 0.6,
      "learning_rate": 2.865609653376484e-05,
      "loss": 1.2867,
      "step": 2058
    },
    {
      "epoch": 0.6,
      "learning_rate": 2.8619640429921705e-05,
      "loss": 1.2277,
      "step": 2059
    },
    {
      "epoch": 0.6,
      "learning_rate": 2.8583194608911245e-05,
      "loss": 1.2669,
      "step": 2060
    },
    {
      "epoch": 0.6,
      "learning_rate": 2.8546759103664446e-05,
      "loss": 1.3178,
      "step": 2061
    },
    {
      "epoch": 0.6,
      "learning_rate": 2.851033394710293e-05,
      "loss": 1.2871,
      "step": 2062
    },
    {
      "epoch": 0.61,
      "learning_rate": 2.8473919172139034e-05,
      "loss": 1.265,
      "step": 2063
    },
    {
      "epoch": 0.61,
      "learning_rate": 2.8437514811675636e-05,
      "loss": 1.2737,
      "step": 2064
    },
    {
      "epoch": 0.61,
      "learning_rate": 2.8401120898606272e-05,
      "loss": 1.3215,
      "step": 2065
    },
    {
      "epoch": 0.61,
      "learning_rate": 2.8364737465814993e-05,
      "loss": 1.2623,
      "step": 2066
    },
    {
      "epoch": 0.61,
      "learning_rate": 2.832836454617641e-05,
      "loss": 1.2492,
      "step": 2067
    },
    {
      "epoch": 0.61,
      "learning_rate": 2.8292002172555616e-05,
      "loss": 1.2173,
      "step": 2068
    },
    {
      "epoch": 0.61,
      "learning_rate": 2.8255650377808185e-05,
      "loss": 1.2751,
      "step": 2069
    },
    {
      "epoch": 0.61,
      "learning_rate": 2.8219309194780113e-05,
      "loss": 1.272,
      "step": 2070
    },
    {
      "epoch": 0.61,
      "learning_rate": 2.8182978656307854e-05,
      "loss": 1.2682,
      "step": 2071
    },
    {
      "epoch": 0.61,
      "learning_rate": 2.814665879521818e-05,
      "loss": 1.2393,
      "step": 2072
    },
    {
      "epoch": 0.61,
      "learning_rate": 2.8110349644328275e-05,
      "loss": 1.3219,
      "step": 2073
    },
    {
      "epoch": 0.61,
      "learning_rate": 2.8074051236445607e-05,
      "loss": 1.2346,
      "step": 2074
    },
    {
      "epoch": 0.61,
      "learning_rate": 2.8037763604367962e-05,
      "loss": 1.2936,
      "step": 2075
    },
    {
      "epoch": 0.61,
      "learning_rate": 2.800148678088335e-05,
      "loss": 1.2525,
      "step": 2076
    },
    {
      "epoch": 0.61,
      "learning_rate": 2.796522079877006e-05,
      "loss": 1.252,
      "step": 2077
    },
    {
      "epoch": 0.61,
      "learning_rate": 2.7928965690796562e-05,
      "loss": 1.2554,
      "step": 2078
    },
    {
      "epoch": 0.61,
      "learning_rate": 2.7892721489721516e-05,
      "loss": 1.2498,
      "step": 2079
    },
    {
      "epoch": 0.61,
      "learning_rate": 2.7856488228293693e-05,
      "loss": 1.2822,
      "step": 2080
    },
    {
      "epoch": 0.61,
      "learning_rate": 2.7820265939252036e-05,
      "loss": 1.2625,
      "step": 2081
    },
    {
      "epoch": 0.61,
      "learning_rate": 2.7784054655325506e-05,
      "loss": 1.2813,
      "step": 2082
    },
    {
      "epoch": 0.61,
      "learning_rate": 2.7747854409233187e-05,
      "loss": 1.2518,
      "step": 2083
    },
    {
      "epoch": 0.61,
      "learning_rate": 2.7711665233684127e-05,
      "loss": 1.3049,
      "step": 2084
    },
    {
      "epoch": 0.61,
      "learning_rate": 2.767548716137743e-05,
      "loss": 1.2807,
      "step": 2085
    },
    {
      "epoch": 0.61,
      "learning_rate": 2.7639320225002108e-05,
      "loss": 1.2758,
      "step": 2086
    },
    {
      "epoch": 0.61,
      "learning_rate": 2.7603164457237172e-05,
      "loss": 1.2759,
      "step": 2087
    },
    {
      "epoch": 0.61,
      "learning_rate": 2.7567019890751493e-05,
      "loss": 1.3017,
      "step": 2088
    },
    {
      "epoch": 0.61,
      "learning_rate": 2.753088655820385e-05,
      "loss": 1.2906,
      "step": 2089
    },
    {
      "epoch": 0.61,
      "learning_rate": 2.749476449224285e-05,
      "loss": 1.2719,
      "step": 2090
    },
    {
      "epoch": 0.61,
      "learning_rate": 2.745865372550695e-05,
      "loss": 1.2734,
      "step": 2091
    },
    {
      "epoch": 0.61,
      "learning_rate": 2.7422554290624348e-05,
      "loss": 1.2289,
      "step": 2092
    },
    {
      "epoch": 0.61,
      "learning_rate": 2.7386466220213065e-05,
      "loss": 1.3234,
      "step": 2093
    },
    {
      "epoch": 0.61,
      "learning_rate": 2.7350389546880803e-05,
      "loss": 1.2704,
      "step": 2094
    },
    {
      "epoch": 0.61,
      "learning_rate": 2.7314324303224994e-05,
      "loss": 1.2987,
      "step": 2095
    },
    {
      "epoch": 0.61,
      "learning_rate": 2.727827052183272e-05,
      "loss": 1.2206,
      "step": 2096
    },
    {
      "epoch": 0.62,
      "learning_rate": 2.7242228235280746e-05,
      "loss": 1.2969,
      "step": 2097
    },
    {
      "epoch": 0.62,
      "learning_rate": 2.7206197476135403e-05,
      "loss": 1.2652,
      "step": 2098
    },
    {
      "epoch": 0.62,
      "learning_rate": 2.7170178276952627e-05,
      "loss": 1.2619,
      "step": 2099
    },
    {
      "epoch": 0.62,
      "learning_rate": 2.7134170670277923e-05,
      "loss": 1.2497,
      "step": 2100
    },
    {
      "epoch": 0.62,
      "eval_loss": 0.11877856403589249,
      "eval_runtime": 24.594,
      "eval_samples_per_second": 81.321,
      "eval_steps_per_second": 0.651,
      "step": 2100
    },
    {
      "epoch": 0.62,
      "learning_rate": 2.7098174688646293e-05,
      "loss": 1.2859,
      "step": 2101
    },
    {
      "epoch": 0.62,
      "learning_rate": 2.7062190364582268e-05,
      "loss": 1.2402,
      "step": 2102
    },
    {
      "epoch": 0.62,
      "learning_rate": 2.7026217730599814e-05,
      "loss": 1.2968,
      "step": 2103
    },
    {
      "epoch": 0.62,
      "learning_rate": 2.6990256819202353e-05,
      "loss": 1.2992,
      "step": 2104
    },
    {
      "epoch": 0.62,
      "learning_rate": 2.695430766288271e-05,
      "loss": 1.2159,
      "step": 2105
    },
    {
      "epoch": 0.62,
      "learning_rate": 2.691837029412309e-05,
      "loss": 1.2908,
      "step": 2106
    },
    {
      "epoch": 0.62,
      "learning_rate": 2.6882444745395044e-05,
      "loss": 1.297,
      "step": 2107
    },
    {
      "epoch": 0.62,
      "learning_rate": 2.6846531049159454e-05,
      "loss": 1.2995,
      "step": 2108
    },
    {
      "epoch": 0.62,
      "learning_rate": 2.6810629237866462e-05,
      "loss": 1.2352,
      "step": 2109
    },
    {
      "epoch": 0.62,
      "learning_rate": 2.677473934395552e-05,
      "loss": 1.2636,
      "step": 2110
    },
    {
      "epoch": 0.62,
      "learning_rate": 2.673886139985526e-05,
      "loss": 1.2827,
      "step": 2111
    },
    {
      "epoch": 0.62,
      "learning_rate": 2.6702995437983557e-05,
      "loss": 1.2675,
      "step": 2112
    },
    {
      "epoch": 0.62,
      "learning_rate": 2.6667141490747454e-05,
      "loss": 1.2757,
      "step": 2113
    },
    {
      "epoch": 0.62,
      "learning_rate": 2.6631299590543095e-05,
      "loss": 1.3055,
      "step": 2114
    },
    {
      "epoch": 0.62,
      "learning_rate": 2.6595469769755805e-05,
      "loss": 1.2427,
      "step": 2115
    },
    {
      "epoch": 0.62,
      "learning_rate": 2.6559652060759928e-05,
      "loss": 1.258,
      "step": 2116
    },
    {
      "epoch": 0.62,
      "learning_rate": 2.6523846495918926e-05,
      "loss": 1.2685,
      "step": 2117
    },
    {
      "epoch": 0.62,
      "learning_rate": 2.648805310758524e-05,
      "loss": 1.2814,
      "step": 2118
    },
    {
      "epoch": 0.62,
      "learning_rate": 2.645227192810034e-05,
      "loss": 1.2521,
      "step": 2119
    },
    {
      "epoch": 0.62,
      "learning_rate": 2.6416502989794647e-05,
      "loss": 1.3092,
      "step": 2120
    },
    {
      "epoch": 0.62,
      "learning_rate": 2.638074632498754e-05,
      "loss": 1.2792,
      "step": 2121
    },
    {
      "epoch": 0.62,
      "learning_rate": 2.634500196598729e-05,
      "loss": 1.2842,
      "step": 2122
    },
    {
      "epoch": 0.62,
      "learning_rate": 2.6309269945091025e-05,
      "loss": 1.266,
      "step": 2123
    },
    {
      "epoch": 0.62,
      "learning_rate": 2.62735502945848e-05,
      "loss": 1.277,
      "step": 2124
    },
    {
      "epoch": 0.62,
      "learning_rate": 2.6237843046743407e-05,
      "loss": 1.2642,
      "step": 2125
    },
    {
      "epoch": 0.62,
      "learning_rate": 2.6202148233830505e-05,
      "loss": 1.2382,
      "step": 2126
    },
    {
      "epoch": 0.62,
      "learning_rate": 2.6166465888098458e-05,
      "loss": 1.2786,
      "step": 2127
    },
    {
      "epoch": 0.62,
      "learning_rate": 2.6130796041788404e-05,
      "loss": 1.2744,
      "step": 2128
    },
    {
      "epoch": 0.62,
      "learning_rate": 2.6095138727130166e-05,
      "loss": 1.2573,
      "step": 2129
    },
    {
      "epoch": 0.62,
      "learning_rate": 2.6059493976342262e-05,
      "loss": 1.2936,
      "step": 2130
    },
    {
      "epoch": 0.63,
      "learning_rate": 2.6023861821631835e-05,
      "loss": 1.2255,
      "step": 2131
    },
    {
      "epoch": 0.63,
      "learning_rate": 2.5988242295194666e-05,
      "loss": 1.2958,
      "step": 2132
    },
    {
      "epoch": 0.63,
      "learning_rate": 2.5952635429215117e-05,
      "loss": 1.2912,
      "step": 2133
    },
    {
      "epoch": 0.63,
      "learning_rate": 2.5917041255866117e-05,
      "loss": 1.2889,
      "step": 2134
    },
    {
      "epoch": 0.63,
      "learning_rate": 2.588145980730911e-05,
      "loss": 1.257,
      "step": 2135
    },
    {
      "epoch": 0.63,
      "learning_rate": 2.5845891115694067e-05,
      "loss": 1.3095,
      "step": 2136
    },
    {
      "epoch": 0.63,
      "learning_rate": 2.5810335213159404e-05,
      "loss": 1.2672,
      "step": 2137
    },
    {
      "epoch": 0.63,
      "learning_rate": 2.5774792131832012e-05,
      "loss": 1.2579,
      "step": 2138
    },
    {
      "epoch": 0.63,
      "learning_rate": 2.5739261903827177e-05,
      "loss": 1.2832,
      "step": 2139
    },
    {
      "epoch": 0.63,
      "learning_rate": 2.5703744561248577e-05,
      "loss": 1.304,
      "step": 2140
    },
    {
      "epoch": 0.63,
      "learning_rate": 2.566824013618824e-05,
      "loss": 1.2841,
      "step": 2141
    },
    {
      "epoch": 0.63,
      "learning_rate": 2.5632748660726537e-05,
      "loss": 1.2342,
      "step": 2142
    },
    {
      "epoch": 0.63,
      "learning_rate": 2.559727016693212e-05,
      "loss": 1.2839,
      "step": 2143
    },
    {
      "epoch": 0.63,
      "learning_rate": 2.5561804686861935e-05,
      "loss": 1.2687,
      "step": 2144
    },
    {
      "epoch": 0.63,
      "learning_rate": 2.552635225256114e-05,
      "loss": 1.2393,
      "step": 2145
    },
    {
      "epoch": 0.63,
      "learning_rate": 2.549091289606313e-05,
      "loss": 1.2649,
      "step": 2146
    },
    {
      "epoch": 0.63,
      "learning_rate": 2.5455486649389477e-05,
      "loss": 1.242,
      "step": 2147
    },
    {
      "epoch": 0.63,
      "learning_rate": 2.54200735445499e-05,
      "loss": 1.3392,
      "step": 2148
    },
    {
      "epoch": 0.63,
      "learning_rate": 2.5384673613542258e-05,
      "loss": 1.2608,
      "step": 2149
    },
    {
      "epoch": 0.63,
      "learning_rate": 2.5349286888352488e-05,
      "loss": 1.3045,
      "step": 2150
    },
    {
      "epoch": 0.63,
      "learning_rate": 2.5313913400954593e-05,
      "loss": 1.31,
      "step": 2151
    },
    {
      "epoch": 0.63,
      "learning_rate": 2.5278553183310646e-05,
      "loss": 1.2655,
      "step": 2152
    },
    {
      "epoch": 0.63,
      "learning_rate": 2.5243206267370685e-05,
      "loss": 1.231,
      "step": 2153
    },
    {
      "epoch": 0.63,
      "learning_rate": 2.5207872685072783e-05,
      "loss": 1.2513,
      "step": 2154
    },
    {
      "epoch": 0.63,
      "learning_rate": 2.5172552468342913e-05,
      "loss": 1.3289,
      "step": 2155
    },
    {
      "epoch": 0.63,
      "learning_rate": 2.5137245649095015e-05,
      "loss": 1.3057,
      "step": 2156
    },
    {
      "epoch": 0.63,
      "learning_rate": 2.510195225923088e-05,
      "loss": 1.2272,
      "step": 2157
    },
    {
      "epoch": 0.63,
      "learning_rate": 2.506667233064021e-05,
      "loss": 1.2706,
      "step": 2158
    },
    {
      "epoch": 0.63,
      "learning_rate": 2.5031405895200505e-05,
      "loss": 1.2998,
      "step": 2159
    },
    {
      "epoch": 0.63,
      "learning_rate": 2.4996152984777098e-05,
      "loss": 1.2309,
      "step": 2160
    },
    {
      "epoch": 0.63,
      "learning_rate": 2.496091363122309e-05,
      "loss": 1.2631,
      "step": 2161
    },
    {
      "epoch": 0.63,
      "learning_rate": 2.4925687866379344e-05,
      "loss": 1.2622,
      "step": 2162
    },
    {
      "epoch": 0.63,
      "learning_rate": 2.489047572207442e-05,
      "loss": 1.2395,
      "step": 2163
    },
    {
      "epoch": 0.63,
      "learning_rate": 2.485527723012461e-05,
      "loss": 1.2657,
      "step": 2164
    },
    {
      "epoch": 0.64,
      "learning_rate": 2.4820092422333827e-05,
      "loss": 1.2902,
      "step": 2165
    },
    {
      "epoch": 0.64,
      "learning_rate": 2.4784921330493646e-05,
      "loss": 1.2901,
      "step": 2166
    },
    {
      "epoch": 0.64,
      "learning_rate": 2.474976398638325e-05,
      "loss": 1.2893,
      "step": 2167
    },
    {
      "epoch": 0.64,
      "learning_rate": 2.4714620421769394e-05,
      "loss": 1.263,
      "step": 2168
    },
    {
      "epoch": 0.64,
      "learning_rate": 2.4679490668406373e-05,
      "loss": 1.2496,
      "step": 2169
    },
    {
      "epoch": 0.64,
      "learning_rate": 2.4644374758036015e-05,
      "loss": 1.3173,
      "step": 2170
    },
    {
      "epoch": 0.64,
      "learning_rate": 2.4609272722387636e-05,
      "loss": 1.2312,
      "step": 2171
    },
    {
      "epoch": 0.64,
      "learning_rate": 2.4574184593178022e-05,
      "loss": 1.2481,
      "step": 2172
    },
    {
      "epoch": 0.64,
      "learning_rate": 2.4539110402111383e-05,
      "loss": 1.2617,
      "step": 2173
    },
    {
      "epoch": 0.64,
      "learning_rate": 2.450405018087933e-05,
      "loss": 1.3095,
      "step": 2174
    },
    {
      "epoch": 0.64,
      "learning_rate": 2.4469003961160883e-05,
      "loss": 1.2213,
      "step": 2175
    },
    {
      "epoch": 0.64,
      "learning_rate": 2.4433971774622368e-05,
      "loss": 1.2374,
      "step": 2176
    },
    {
      "epoch": 0.64,
      "learning_rate": 2.4398953652917464e-05,
      "loss": 1.2518,
      "step": 2177
    },
    {
      "epoch": 0.64,
      "learning_rate": 2.436394962768712e-05,
      "loss": 1.3257,
      "step": 2178
    },
    {
      "epoch": 0.64,
      "learning_rate": 2.4328959730559573e-05,
      "loss": 1.2227,
      "step": 2179
    },
    {
      "epoch": 0.64,
      "learning_rate": 2.4293983993150256e-05,
      "loss": 1.2663,
      "step": 2180
    },
    {
      "epoch": 0.64,
      "learning_rate": 2.4259022447061858e-05,
      "loss": 1.2969,
      "step": 2181
    },
    {
      "epoch": 0.64,
      "learning_rate": 2.4224075123884195e-05,
      "loss": 1.2596,
      "step": 2182
    },
    {
      "epoch": 0.64,
      "learning_rate": 2.4189142055194276e-05,
      "loss": 1.199,
      "step": 2183
    },
    {
      "epoch": 0.64,
      "learning_rate": 2.4154223272556184e-05,
      "loss": 1.287,
      "step": 2184
    },
    {
      "epoch": 0.64,
      "learning_rate": 2.411931880752114e-05,
      "loss": 1.2886,
      "step": 2185
    },
    {
      "epoch": 0.64,
      "learning_rate": 2.40844286916274e-05,
      "loss": 1.2447,
      "step": 2186
    },
    {
      "epoch": 0.64,
      "learning_rate": 2.404955295640026e-05,
      "loss": 1.257,
      "step": 2187
    },
    {
      "epoch": 0.64,
      "learning_rate": 2.4014691633352038e-05,
      "loss": 1.2159,
      "step": 2188
    },
    {
      "epoch": 0.64,
      "learning_rate": 2.3979844753981984e-05,
      "loss": 1.2866,
      "step": 2189
    },
    {
      "epoch": 0.64,
      "learning_rate": 2.3945012349776358e-05,
      "loss": 1.2717,
      "step": 2190
    },
    {
      "epoch": 0.64,
      "learning_rate": 2.3910194452208292e-05,
      "loss": 1.2025,
      "step": 2191
    },
    {
      "epoch": 0.64,
      "learning_rate": 2.3875391092737853e-05,
      "loss": 1.2589,
      "step": 2192
    },
    {
      "epoch": 0.64,
      "learning_rate": 2.3840602302811923e-05,
      "loss": 1.2504,
      "step": 2193
    },
    {
      "epoch": 0.64,
      "learning_rate": 2.3805828113864272e-05,
      "loss": 1.2905,
      "step": 2194
    },
    {
      "epoch": 0.64,
      "learning_rate": 2.3771068557315424e-05,
      "loss": 1.2466,
      "step": 2195
    },
    {
      "epoch": 0.64,
      "learning_rate": 2.3736323664572734e-05,
      "loss": 1.232,
      "step": 2196
    },
    {
      "epoch": 0.64,
      "learning_rate": 2.3701593467030265e-05,
      "loss": 1.3036,
      "step": 2197
    },
    {
      "epoch": 0.64,
      "learning_rate": 2.3666877996068824e-05,
      "loss": 1.2539,
      "step": 2198
    },
    {
      "epoch": 0.65,
      "learning_rate": 2.3632177283055908e-05,
      "loss": 1.256,
      "step": 2199
    },
    {
      "epoch": 0.65,
      "learning_rate": 2.3597491359345675e-05,
      "loss": 1.2985,
      "step": 2200
    },
    {
      "epoch": 0.65,
      "eval_loss": 0.11855418980121613,
      "eval_runtime": 24.6365,
      "eval_samples_per_second": 81.18,
      "eval_steps_per_second": 0.649,
      "step": 2200
    },
    {
      "epoch": 0.65,
      "learning_rate": 2.356282025627893e-05,
      "loss": 1.256,
      "step": 2201
    },
    {
      "epoch": 0.65,
      "learning_rate": 2.3528164005183067e-05,
      "loss": 1.2158,
      "step": 2202
    },
    {
      "epoch": 0.65,
      "learning_rate": 2.3493522637372087e-05,
      "loss": 1.2627,
      "step": 2203
    },
    {
      "epoch": 0.65,
      "learning_rate": 2.3458896184146517e-05,
      "loss": 1.272,
      "step": 2204
    },
    {
      "epoch": 0.65,
      "learning_rate": 2.3424284676793428e-05,
      "loss": 1.3215,
      "step": 2205
    },
    {
      "epoch": 0.65,
      "learning_rate": 2.338968814658637e-05,
      "loss": 1.2625,
      "step": 2206
    },
    {
      "epoch": 0.65,
      "learning_rate": 2.3355106624785392e-05,
      "loss": 1.2564,
      "step": 2207
    },
    {
      "epoch": 0.65,
      "learning_rate": 2.332054014263692e-05,
      "loss": 1.3071,
      "step": 2208
    },
    {
      "epoch": 0.65,
      "learning_rate": 2.3285988731373852e-05,
      "loss": 1.2588,
      "step": 2209
    },
    {
      "epoch": 0.65,
      "learning_rate": 2.3251452422215466e-05,
      "loss": 1.2609,
      "step": 2210
    },
    {
      "epoch": 0.65,
      "learning_rate": 2.3216931246367335e-05,
      "loss": 1.2539,
      "step": 2211
    },
    {
      "epoch": 0.65,
      "learning_rate": 2.318242523502141e-05,
      "loss": 1.2492,
      "step": 2212
    },
    {
      "epoch": 0.65,
      "learning_rate": 2.3147934419355935e-05,
      "loss": 1.2336,
      "step": 2213
    },
    {
      "epoch": 0.65,
      "learning_rate": 2.3113458830535432e-05,
      "loss": 1.28,
      "step": 2214
    },
    {
      "epoch": 0.65,
      "learning_rate": 2.3078998499710613e-05,
      "loss": 1.2764,
      "step": 2215
    },
    {
      "epoch": 0.65,
      "learning_rate": 2.3044553458018454e-05,
      "loss": 1.2551,
      "step": 2216
    },
    {
      "epoch": 0.65,
      "learning_rate": 2.301012373658211e-05,
      "loss": 1.2251,
      "step": 2217
    },
    {
      "epoch": 0.65,
      "learning_rate": 2.2975709366510887e-05,
      "loss": 1.2493,
      "step": 2218
    },
    {
      "epoch": 0.65,
      "learning_rate": 2.2941310378900203e-05,
      "loss": 1.2662,
      "step": 2219
    },
    {
      "epoch": 0.65,
      "learning_rate": 2.2906926804831598e-05,
      "loss": 1.2551,
      "step": 2220
    },
    {
      "epoch": 0.65,
      "learning_rate": 2.2872558675372674e-05,
      "loss": 1.2538,
      "step": 2221
    },
    {
      "epoch": 0.65,
      "learning_rate": 2.2838206021577108e-05,
      "loss": 1.2619,
      "step": 2222
    },
    {
      "epoch": 0.65,
      "learning_rate": 2.2803868874484536e-05,
      "loss": 1.2853,
      "step": 2223
    },
    {
      "epoch": 0.65,
      "learning_rate": 2.276954726512063e-05,
      "loss": 1.2659,
      "step": 2224
    },
    {
      "epoch": 0.65,
      "learning_rate": 2.2735241224497027e-05,
      "loss": 1.2268,
      "step": 2225
    },
    {
      "epoch": 0.65,
      "learning_rate": 2.2700950783611252e-05,
      "loss": 1.2466,
      "step": 2226
    },
    {
      "epoch": 0.65,
      "learning_rate": 2.2666675973446763e-05,
      "loss": 1.3015,
      "step": 2227
    },
    {
      "epoch": 0.65,
      "learning_rate": 2.2632416824972886e-05,
      "loss": 1.2587,
      "step": 2228
    },
    {
      "epoch": 0.65,
      "learning_rate": 2.2598173369144815e-05,
      "loss": 1.2257,
      "step": 2229
    },
    {
      "epoch": 0.65,
      "learning_rate": 2.256394563690357e-05,
      "loss": 1.2616,
      "step": 2230
    },
    {
      "epoch": 0.65,
      "learning_rate": 2.2529733659175906e-05,
      "loss": 1.2229,
      "step": 2231
    },
    {
      "epoch": 0.65,
      "learning_rate": 2.2495537466874402e-05,
      "loss": 1.2487,
      "step": 2232
    },
    {
      "epoch": 0.66,
      "learning_rate": 2.2461357090897352e-05,
      "loss": 1.2669,
      "step": 2233
    },
    {
      "epoch": 0.66,
      "learning_rate": 2.2427192562128786e-05,
      "loss": 1.2863,
      "step": 2234
    },
    {
      "epoch": 0.66,
      "learning_rate": 2.239304391143835e-05,
      "loss": 1.2698,
      "step": 2235
    },
    {
      "epoch": 0.66,
      "learning_rate": 2.2358911169681405e-05,
      "loss": 1.259,
      "step": 2236
    },
    {
      "epoch": 0.66,
      "learning_rate": 2.2324794367698914e-05,
      "loss": 1.2678,
      "step": 2237
    },
    {
      "epoch": 0.66,
      "learning_rate": 2.2290693536317444e-05,
      "loss": 1.277,
      "step": 2238
    },
    {
      "epoch": 0.66,
      "learning_rate": 2.2256608706349118e-05,
      "loss": 1.2437,
      "step": 2239
    },
    {
      "epoch": 0.66,
      "learning_rate": 2.2222539908591608e-05,
      "loss": 1.244,
      "step": 2240
    },
    {
      "epoch": 0.66,
      "learning_rate": 2.2188487173828116e-05,
      "loss": 1.2023,
      "step": 2241
    },
    {
      "epoch": 0.66,
      "learning_rate": 2.2154450532827315e-05,
      "loss": 1.3176,
      "step": 2242
    },
    {
      "epoch": 0.66,
      "learning_rate": 2.212043001634332e-05,
      "loss": 1.2036,
      "step": 2243
    },
    {
      "epoch": 0.66,
      "learning_rate": 2.2086425655115703e-05,
      "loss": 1.2621,
      "step": 2244
    },
    {
      "epoch": 0.66,
      "learning_rate": 2.205243747986943e-05,
      "loss": 1.2734,
      "step": 2245
    },
    {
      "epoch": 0.66,
      "learning_rate": 2.2018465521314857e-05,
      "loss": 1.2831,
      "step": 2246
    },
    {
      "epoch": 0.66,
      "learning_rate": 2.1984509810147645e-05,
      "loss": 1.2566,
      "step": 2247
    },
    {
      "epoch": 0.66,
      "learning_rate": 2.1950570377048813e-05,
      "loss": 1.289,
      "step": 2248
    },
    {
      "epoch": 0.66,
      "learning_rate": 2.1916647252684678e-05,
      "loss": 1.2823,
      "step": 2249
    },
    {
      "epoch": 0.66,
      "learning_rate": 2.188274046770677e-05,
      "loss": 1.2266,
      "step": 2250
    },
    {
      "epoch": 0.66,
      "learning_rate": 2.18488500527519e-05,
      "loss": 1.2698,
      "step": 2251
    },
    {
      "epoch": 0.66,
      "learning_rate": 2.1814976038442087e-05,
      "loss": 1.2187,
      "step": 2252
    },
    {
      "epoch": 0.66,
      "learning_rate": 2.178111845538453e-05,
      "loss": 1.3168,
      "step": 2253
    },
    {
      "epoch": 0.66,
      "learning_rate": 2.1747277334171545e-05,
      "loss": 1.2259,
      "step": 2254
    },
    {
      "epoch": 0.66,
      "learning_rate": 2.1713452705380612e-05,
      "loss": 1.2816,
      "step": 2255
    },
    {
      "epoch": 0.66,
      "learning_rate": 2.1679644599574293e-05,
      "loss": 1.2566,
      "step": 2256
    },
    {
      "epoch": 0.66,
      "learning_rate": 2.1645853047300248e-05,
      "loss": 1.3025,
      "step": 2257
    },
    {
      "epoch": 0.66,
      "learning_rate": 2.1612078079091125e-05,
      "loss": 1.2821,
      "step": 2258
    },
    {
      "epoch": 0.66,
      "learning_rate": 2.157831972546463e-05,
      "loss": 1.2499,
      "step": 2259
    },
    {
      "epoch": 0.66,
      "learning_rate": 2.1544578016923454e-05,
      "loss": 1.23,
      "step": 2260
    },
    {
      "epoch": 0.66,
      "learning_rate": 2.151085298395525e-05,
      "loss": 1.2619,
      "step": 2261
    },
    {
      "epoch": 0.66,
      "learning_rate": 2.1477144657032566e-05,
      "loss": 1.2541,
      "step": 2262
    },
    {
      "epoch": 0.66,
      "learning_rate": 2.144345306661291e-05,
      "loss": 1.2347,
      "step": 2263
    },
    {
      "epoch": 0.66,
      "learning_rate": 2.1409778243138618e-05,
      "loss": 1.2679,
      "step": 2264
    },
    {
      "epoch": 0.66,
      "learning_rate": 2.1376120217036904e-05,
      "loss": 1.2237,
      "step": 2265
    },
    {
      "epoch": 0.66,
      "learning_rate": 2.1342479018719818e-05,
      "loss": 1.2391,
      "step": 2266
    },
    {
      "epoch": 0.67,
      "learning_rate": 2.1308854678584162e-05,
      "loss": 1.2798,
      "step": 2267
    },
    {
      "epoch": 0.67,
      "learning_rate": 2.1275247227011536e-05,
      "loss": 1.2453,
      "step": 2268
    },
    {
      "epoch": 0.67,
      "learning_rate": 2.124165669436828e-05,
      "loss": 1.2812,
      "step": 2269
    },
    {
      "epoch": 0.67,
      "learning_rate": 2.1208083111005458e-05,
      "loss": 1.2221,
      "step": 2270
    },
    {
      "epoch": 0.67,
      "learning_rate": 2.1174526507258762e-05,
      "loss": 1.2548,
      "step": 2271
    },
    {
      "epoch": 0.67,
      "learning_rate": 2.1140986913448606e-05,
      "loss": 1.2705,
      "step": 2272
    },
    {
      "epoch": 0.67,
      "learning_rate": 2.1107464359880022e-05,
      "loss": 1.2134,
      "step": 2273
    },
    {
      "epoch": 0.67,
      "learning_rate": 2.1073958876842605e-05,
      "loss": 1.179,
      "step": 2274
    },
    {
      "epoch": 0.67,
      "learning_rate": 2.104047049461057e-05,
      "loss": 1.2881,
      "step": 2275
    },
    {
      "epoch": 0.67,
      "learning_rate": 2.1006999243442664e-05,
      "loss": 1.2468,
      "step": 2276
    },
    {
      "epoch": 0.67,
      "learning_rate": 2.097354515358217e-05,
      "loss": 1.2831,
      "step": 2277
    },
    {
      "epoch": 0.67,
      "learning_rate": 2.0940108255256823e-05,
      "loss": 1.2548,
      "step": 2278
    },
    {
      "epoch": 0.67,
      "learning_rate": 2.090668857867886e-05,
      "loss": 1.2709,
      "step": 2279
    },
    {
      "epoch": 0.67,
      "learning_rate": 2.0873286154044967e-05,
      "loss": 1.254,
      "step": 2280
    },
    {
      "epoch": 0.67,
      "learning_rate": 2.0839901011536216e-05,
      "loss": 1.2239,
      "step": 2281
    },
    {
      "epoch": 0.67,
      "learning_rate": 2.0806533181318063e-05,
      "loss": 1.2516,
      "step": 2282
    },
    {
      "epoch": 0.67,
      "learning_rate": 2.0773182693540325e-05,
      "loss": 1.2853,
      "step": 2283
    },
    {
      "epoch": 0.67,
      "learning_rate": 2.0739849578337167e-05,
      "loss": 1.2284,
      "step": 2284
    },
    {
      "epoch": 0.67,
      "learning_rate": 2.070653386582706e-05,
      "loss": 1.228,
      "step": 2285
    },
    {
      "epoch": 0.67,
      "learning_rate": 2.0673235586112698e-05,
      "loss": 1.2463,
      "step": 2286
    },
    {
      "epoch": 0.67,
      "learning_rate": 2.0639954769281082e-05,
      "loss": 1.2998,
      "step": 2287
    },
    {
      "epoch": 0.67,
      "learning_rate": 2.0606691445403398e-05,
      "loss": 1.277,
      "step": 2288
    },
    {
      "epoch": 0.67,
      "learning_rate": 2.057344564453507e-05,
      "loss": 1.2121,
      "step": 2289
    },
    {
      "epoch": 0.67,
      "learning_rate": 2.054021739671562e-05,
      "loss": 1.2894,
      "step": 2290
    },
    {
      "epoch": 0.67,
      "learning_rate": 2.0507006731968775e-05,
      "loss": 1.2377,
      "step": 2291
    },
    {
      "epoch": 0.67,
      "learning_rate": 2.0473813680302337e-05,
      "loss": 1.2422,
      "step": 2292
    },
    {
      "epoch": 0.67,
      "learning_rate": 2.0440638271708226e-05,
      "loss": 1.2078,
      "step": 2293
    },
    {
      "epoch": 0.67,
      "learning_rate": 2.0407480536162368e-05,
      "loss": 1.3223,
      "step": 2294
    },
    {
      "epoch": 0.67,
      "learning_rate": 2.037434050362477e-05,
      "loss": 1.2316,
      "step": 2295
    },
    {
      "epoch": 0.67,
      "learning_rate": 2.0341218204039416e-05,
      "loss": 1.246,
      "step": 2296
    },
    {
      "epoch": 0.67,
      "learning_rate": 2.0308113667334303e-05,
      "loss": 1.2729,
      "step": 2297
    },
    {
      "epoch": 0.67,
      "learning_rate": 2.0275026923421315e-05,
      "loss": 1.2566,
      "step": 2298
    },
    {
      "epoch": 0.67,
      "learning_rate": 2.0241958002196308e-05,
      "loss": 1.2499,
      "step": 2299
    },
    {
      "epoch": 0.67,
      "learning_rate": 2.020890693353905e-05,
      "loss": 1.2566,
      "step": 2300
    },
    {
      "epoch": 0.67,
      "eval_loss": 0.11811155080795288,
      "eval_runtime": 24.448,
      "eval_samples_per_second": 81.806,
      "eval_steps_per_second": 0.654,
      "step": 2300
    },
    {
      "epoch": 0.68,
      "learning_rate": 2.0175873747313118e-05,
      "loss": 1.2615,
      "step": 2301
    },
    {
      "epoch": 0.68,
      "learning_rate": 2.0142858473365957e-05,
      "loss": 1.2863,
      "step": 2302
    },
    {
      "epoch": 0.68,
      "learning_rate": 2.0109861141528853e-05,
      "loss": 1.2716,
      "step": 2303
    },
    {
      "epoch": 0.68,
      "learning_rate": 2.0076881781616856e-05,
      "loss": 1.2606,
      "step": 2304
    },
    {
      "epoch": 0.68,
      "learning_rate": 2.004392042342879e-05,
      "loss": 1.2657,
      "step": 2305
    },
    {
      "epoch": 0.68,
      "learning_rate": 2.0010977096747184e-05,
      "loss": 1.2354,
      "step": 2306
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.997805183133831e-05,
      "loss": 1.2393,
      "step": 2307
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.9945144656952103e-05,
      "loss": 1.2591,
      "step": 2308
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.9912255603322174e-05,
      "loss": 1.2773,
      "step": 2309
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.9879384700165706e-05,
      "loss": 1.211,
      "step": 2310
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.9846531977183536e-05,
      "loss": 1.2424,
      "step": 2311
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.9813697464060056e-05,
      "loss": 1.2258,
      "step": 2312
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.9780881190463222e-05,
      "loss": 1.2652,
      "step": 2313
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.9748083186044456e-05,
      "loss": 1.2576,
      "step": 2314
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.971530348043873e-05,
      "loss": 1.2408,
      "step": 2315
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.968254210326445e-05,
      "loss": 1.2493,
      "step": 2316
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.9649799084123504e-05,
      "loss": 1.3321,
      "step": 2317
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.9617074452601113e-05,
      "loss": 1.2754,
      "step": 2318
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.958436823826595e-05,
      "loss": 1.1769,
      "step": 2319
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.9551680470670035e-05,
      "loss": 1.2696,
      "step": 2320
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.9519011179348717e-05,
      "loss": 1.2095,
      "step": 2321
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.9486360393820628e-05,
      "loss": 1.2401,
      "step": 2322
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.9453728143587697e-05,
      "loss": 1.2474,
      "step": 2323
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.9421114458135138e-05,
      "loss": 1.2745,
      "step": 2324
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.9388519366931316e-05,
      "loss": 1.2282,
      "step": 2325
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.935594289942786e-05,
      "loss": 1.2392,
      "step": 2326
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.9323385085059544e-05,
      "loss": 1.2246,
      "step": 2327
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.9290845953244307e-05,
      "loss": 1.278,
      "step": 2328
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.9258325533383166e-05,
      "loss": 1.2167,
      "step": 2329
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.922582385486027e-05,
      "loss": 1.242,
      "step": 2330
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.919334094704282e-05,
      "loss": 1.2719,
      "step": 2331
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.9160876839281077e-05,
      "loss": 1.2855,
      "step": 2332
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.9128431560908263e-05,
      "loss": 1.2537,
      "step": 2333
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.909600514124063e-05,
      "loss": 1.2326,
      "step": 2334
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.9063597609577388e-05,
      "loss": 1.291,
      "step": 2335
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.9031208995200687e-05,
      "loss": 1.2172,
      "step": 2336
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.8998839327375532e-05,
      "loss": 1.2588,
      "step": 2337
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.896648863534988e-05,
      "loss": 1.2368,
      "step": 2338
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.8934156948354475e-05,
      "loss": 1.2894,
      "step": 2339
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.890184429560294e-05,
      "loss": 1.2531,
      "step": 2340
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.88695507062917e-05,
      "loss": 1.2341,
      "step": 2341
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.8837276209599896e-05,
      "loss": 1.2534,
      "step": 2342
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.8805020834689487e-05,
      "loss": 1.272,
      "step": 2343
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.8772784610705107e-05,
      "loss": 1.2104,
      "step": 2344
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.8740567566774137e-05,
      "loss": 1.2886,
      "step": 2345
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.8708369732006555e-05,
      "loss": 1.2731,
      "step": 2346
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.867619113549504e-05,
      "loss": 1.2391,
      "step": 2347
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.864403180631489e-05,
      "loss": 1.2516,
      "step": 2348
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.8611891773523934e-05,
      "loss": 1.2218,
      "step": 2349
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.8579771066162626e-05,
      "loss": 1.2865,
      "step": 2350
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.8547669713253943e-05,
      "loss": 1.2438,
      "step": 2351
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.8515587743803376e-05,
      "loss": 1.262,
      "step": 2352
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.8483525186798875e-05,
      "loss": 1.2495,
      "step": 2353
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.8451482071210875e-05,
      "loss": 1.2554,
      "step": 2354
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.8419458425992243e-05,
      "loss": 1.2481,
      "step": 2355
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.8387454280078268e-05,
      "loss": 1.2407,
      "step": 2356
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.835546966238657e-05,
      "loss": 1.2507,
      "step": 2357
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.832350460181717e-05,
      "loss": 1.2699,
      "step": 2358
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.82915591272524e-05,
      "loss": 1.1869,
      "step": 2359
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.8259633267556923e-05,
      "loss": 1.1662,
      "step": 2360
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.822772705157762e-05,
      "loss": 1.294,
      "step": 2361
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.8195840508143673e-05,
      "loss": 1.2864,
      "step": 2362
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.8163973666066473e-05,
      "loss": 1.2462,
      "step": 2363
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.813212655413962e-05,
      "loss": 1.2157,
      "step": 2364
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.8100299201138857e-05,
      "loss": 1.2405,
      "step": 2365
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.8068491635822097e-05,
      "loss": 1.24,
      "step": 2366
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.803670388692938e-05,
      "loss": 1.2725,
      "step": 2367
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.8004935983182835e-05,
      "loss": 1.2409,
      "step": 2368
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.7973187953286634e-05,
      "loss": 1.288,
      "step": 2369
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.7941459825927023e-05,
      "loss": 1.2169,
      "step": 2370
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.7909751629772258e-05,
      "loss": 1.1776,
      "step": 2371
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.787806339347259e-05,
      "loss": 1.2889,
      "step": 2372
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.784639514566021e-05,
      "loss": 1.2779,
      "step": 2373
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.781474691494927e-05,
      "loss": 1.2556,
      "step": 2374
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.7783118729935838e-05,
      "loss": 1.2782,
      "step": 2375
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.775151061919789e-05,
      "loss": 1.2685,
      "step": 2376
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.7719922611295177e-05,
      "loss": 1.2556,
      "step": 2377
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.7688354734769378e-05,
      "loss": 1.2534,
      "step": 2378
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.765680701814395e-05,
      "loss": 1.2165,
      "step": 2379
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.762527948992414e-05,
      "loss": 1.2971,
      "step": 2380
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.759377217859693e-05,
      "loss": 1.1959,
      "step": 2381
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.756228511263106e-05,
      "loss": 1.2311,
      "step": 2382
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.753081832047697e-05,
      "loss": 1.2669,
      "step": 2383
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.7499371830566794e-05,
      "loss": 1.2795,
      "step": 2384
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.7467945671314284e-05,
      "loss": 1.2406,
      "step": 2385
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.743653987111486e-05,
      "loss": 1.1918,
      "step": 2386
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.740515445834553e-05,
      "loss": 1.2611,
      "step": 2387
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.7373789461364904e-05,
      "loss": 1.2804,
      "step": 2388
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.734244490851309e-05,
      "loss": 1.2379,
      "step": 2389
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.7311120828111772e-05,
      "loss": 1.2276,
      "step": 2390
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.7279817248464132e-05,
      "loss": 1.226,
      "step": 2391
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.724853419785482e-05,
      "loss": 1.2891,
      "step": 2392
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.7217271704549913e-05,
      "loss": 1.2269,
      "step": 2393
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.718602979679695e-05,
      "loss": 1.2288,
      "step": 2394
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.715480850282486e-05,
      "loss": 1.2531,
      "step": 2395
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.7123607850843947e-05,
      "loss": 1.2267,
      "step": 2396
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.7092427869045836e-05,
      "loss": 1.2184,
      "step": 2397
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.7061268585603507e-05,
      "loss": 1.2751,
      "step": 2398
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.7030130028671235e-05,
      "loss": 1.3224,
      "step": 2399
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.6999012226384572e-05,
      "loss": 1.2363,
      "step": 2400
    },
    {
      "epoch": 0.7,
      "eval_loss": 0.12007694691419601,
      "eval_runtime": 24.699,
      "eval_samples_per_second": 80.975,
      "eval_steps_per_second": 0.648,
      "step": 2400
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.6967915206860274e-05,
      "loss": 1.2254,
      "step": 2401
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.693683899819637e-05,
      "loss": 1.2336,
      "step": 2402
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.6905783628472074e-05,
      "loss": 1.2907,
      "step": 2403
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.6874749125747737e-05,
      "loss": 1.2646,
      "step": 2404
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.6843735518064895e-05,
      "loss": 1.221,
      "step": 2405
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.6812742833446187e-05,
      "loss": 1.253,
      "step": 2406
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.678177109989537e-05,
      "loss": 1.2584,
      "step": 2407
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.6750820345397217e-05,
      "loss": 1.2554,
      "step": 2408
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.6719890597917588e-05,
      "loss": 1.2153,
      "step": 2409
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.668898188540336e-05,
      "loss": 1.2268,
      "step": 2410
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.6658094235782405e-05,
      "loss": 1.2531,
      "step": 2411
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.662722767696353e-05,
      "loss": 1.2358,
      "step": 2412
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.659638223683653e-05,
      "loss": 1.2201,
      "step": 2413
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.6565557943272077e-05,
      "loss": 1.3041,
      "step": 2414
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.6534754824121767e-05,
      "loss": 1.1871,
      "step": 2415
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.6503972907218067e-05,
      "loss": 1.2208,
      "step": 2416
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.6473212220374245e-05,
      "loss": 1.2669,
      "step": 2417
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.644247279138442e-05,
      "loss": 1.2488,
      "step": 2418
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.641175464802351e-05,
      "loss": 1.2402,
      "step": 2419
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.6381057818047196e-05,
      "loss": 1.1869,
      "step": 2420
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.635038232919186e-05,
      "loss": 1.2312,
      "step": 2421
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.631972820917465e-05,
      "loss": 1.2659,
      "step": 2422
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.628909548569339e-05,
      "loss": 1.1862,
      "step": 2423
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.6258484186426585e-05,
      "loss": 1.2562,
      "step": 2424
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.6227894339033334e-05,
      "loss": 1.2712,
      "step": 2425
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.6197325971153408e-05,
      "loss": 1.2487,
      "step": 2426
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.6166779110407155e-05,
      "loss": 1.2378,
      "step": 2427
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.6136253784395455e-05,
      "loss": 1.2381,
      "step": 2428
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.610575002069978e-05,
      "loss": 1.2484,
      "step": 2429
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.607526784688209e-05,
      "loss": 1.2071,
      "step": 2430
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.604480729048486e-05,
      "loss": 1.2194,
      "step": 2431
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.6014368379030987e-05,
      "loss": 1.2515,
      "step": 2432
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.5983951140023864e-05,
      "loss": 1.2542,
      "step": 2433
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.5953555600947263e-05,
      "loss": 1.2154,
      "step": 2434
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.5923181789265393e-05,
      "loss": 1.2378,
      "step": 2435
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.589282973242276e-05,
      "loss": 1.277,
      "step": 2436
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.5862499457844283e-05,
      "loss": 1.3002,
      "step": 2437
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.583219099293516e-05,
      "loss": 1.2509,
      "step": 2438
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.580190436508091e-05,
      "loss": 1.2876,
      "step": 2439
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.5771639601647286e-05,
      "loss": 1.2338,
      "step": 2440
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.574139672998031e-05,
      "loss": 1.257,
      "step": 2441
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.5711175777406217e-05,
      "loss": 1.2704,
      "step": 2442
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.568097677123146e-05,
      "loss": 1.249,
      "step": 2443
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.5650799738742604e-05,
      "loss": 1.2386,
      "step": 2444
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.5620644707206406e-05,
      "loss": 1.2212,
      "step": 2445
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.5590511703869745e-05,
      "loss": 1.2294,
      "step": 2446
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.556040075595959e-05,
      "loss": 1.2305,
      "step": 2447
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.5530311890682946e-05,
      "loss": 1.2587,
      "step": 2448
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.5500245135226917e-05,
      "loss": 1.2768,
      "step": 2449
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.5470200516758606e-05,
      "loss": 1.1653,
      "step": 2450
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.544017806242513e-05,
      "loss": 1.2504,
      "step": 2451
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.5410177799353524e-05,
      "loss": 1.2825,
      "step": 2452
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.5380199754650838e-05,
      "loss": 1.2358,
      "step": 2453
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.5350243955404015e-05,
      "loss": 1.2393,
      "step": 2454
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.532031042867992e-05,
      "loss": 1.1987,
      "step": 2455
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.529039920152525e-05,
      "loss": 1.2591,
      "step": 2456
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.5260510300966593e-05,
      "loss": 1.2111,
      "step": 2457
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.5230643754010355e-05,
      "loss": 1.2552,
      "step": 2458
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.5200799587642747e-05,
      "loss": 1.2698,
      "step": 2459
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.517097782882973e-05,
      "loss": 1.2331,
      "step": 2460
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.5141178504517048e-05,
      "loss": 1.2203,
      "step": 2461
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.5111401641630168e-05,
      "loss": 1.207,
      "step": 2462
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.508164726707427e-05,
      "loss": 1.2449,
      "step": 2463
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.5051915407734181e-05,
      "loss": 1.2555,
      "step": 2464
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.5022206090474418e-05,
      "loss": 1.2263,
      "step": 2465
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.4992519342139118e-05,
      "loss": 1.234,
      "step": 2466
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.496285518955204e-05,
      "loss": 1.2564,
      "step": 2467
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.4933213659516485e-05,
      "loss": 1.2294,
      "step": 2468
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.4903594778815361e-05,
      "loss": 1.2796,
      "step": 2469
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.4873998574211075e-05,
      "loss": 1.246,
      "step": 2470
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.484442507244559e-05,
      "loss": 1.2505,
      "step": 2471
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.4814874300240295e-05,
      "loss": 1.2326,
      "step": 2472
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.4785346284296078e-05,
      "loss": 1.247,
      "step": 2473
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.4755841051293272e-05,
      "loss": 1.2431,
      "step": 2474
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.472635862789162e-05,
      "loss": 1.2322,
      "step": 2475
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.4696899040730217e-05,
      "loss": 1.2192,
      "step": 2476
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.4667462316427576e-05,
      "loss": 1.2225,
      "step": 2477
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.4638048481581537e-05,
      "loss": 1.2898,
      "step": 2478
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.4608657562769231e-05,
      "loss": 1.2722,
      "step": 2479
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.4579289586547112e-05,
      "loss": 1.2377,
      "step": 2480
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.4549944579450897e-05,
      "loss": 1.2683,
      "step": 2481
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.452062256799556e-05,
      "loss": 1.2552,
      "step": 2482
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.4491323578675265e-05,
      "loss": 1.1738,
      "step": 2483
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.44620476379634e-05,
      "loss": 1.244,
      "step": 2484
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.4432794772312519e-05,
      "loss": 1.224,
      "step": 2485
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.440356500815435e-05,
      "loss": 1.2493,
      "step": 2486
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.4374358371899698e-05,
      "loss": 1.2124,
      "step": 2487
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.4345174889938514e-05,
      "loss": 1.231,
      "step": 2488
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.43160145886398e-05,
      "loss": 1.2927,
      "step": 2489
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.4286877494351626e-05,
      "loss": 1.2223,
      "step": 2490
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.4257763633401114e-05,
      "loss": 1.2237,
      "step": 2491
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.4228673032094342e-05,
      "loss": 1.2503,
      "step": 2492
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.4199605716716414e-05,
      "loss": 1.2374,
      "step": 2493
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.4170561713531381e-05,
      "loss": 1.2554,
      "step": 2494
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.4141541048782244e-05,
      "loss": 1.2518,
      "step": 2495
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.4112543748690874e-05,
      "loss": 1.2048,
      "step": 2496
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.4083569839458071e-05,
      "loss": 1.2436,
      "step": 2497
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.4054619347263487e-05,
      "loss": 1.2357,
      "step": 2498
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.4025692298265634e-05,
      "loss": 1.2359,
      "step": 2499
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.3996788718601795e-05,
      "loss": 1.2,
      "step": 2500
    },
    {
      "epoch": 0.73,
      "eval_loss": 0.1154995933175087,
      "eval_runtime": 25.0354,
      "eval_samples_per_second": 79.887,
      "eval_steps_per_second": 0.639,
      "step": 2500
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.3967908634388088e-05,
      "loss": 1.2448,
      "step": 2501
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.3939052071719398e-05,
      "loss": 1.1813,
      "step": 2502
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.3910219056669335e-05,
      "loss": 1.2292,
      "step": 2503
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.3881409615290244e-05,
      "loss": 1.2634,
      "step": 2504
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.3852623773613178e-05,
      "loss": 1.2264,
      "step": 2505
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.382386155764787e-05,
      "loss": 1.2159,
      "step": 2506
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.3795122993382673e-05,
      "loss": 1.2163,
      "step": 2507
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.3766408106784601e-05,
      "loss": 1.271,
      "step": 2508
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.3737716923799265e-05,
      "loss": 1.1833,
      "step": 2509
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.3709049470350873e-05,
      "loss": 1.2695,
      "step": 2510
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.3680405772342145e-05,
      "loss": 1.2319,
      "step": 2511
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.3651785855654387e-05,
      "loss": 1.2132,
      "step": 2512
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.3623189746147398e-05,
      "loss": 1.2702,
      "step": 2513
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.3594617469659474e-05,
      "loss": 1.2386,
      "step": 2514
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.3566069052007351e-05,
      "loss": 1.1893,
      "step": 2515
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.3537544518986225e-05,
      "loss": 1.2318,
      "step": 2516
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.3509043896369725e-05,
      "loss": 1.2074,
      "step": 2517
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.3480567209909863e-05,
      "loss": 1.2181,
      "step": 2518
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.3452114485337e-05,
      "loss": 1.2923,
      "step": 2519
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.3423685748359883e-05,
      "loss": 1.2487,
      "step": 2520
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.3395281024665559e-05,
      "loss": 1.2368,
      "step": 2521
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.3366900339919413e-05,
      "loss": 1.2459,
      "step": 2522
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.333854371976505e-05,
      "loss": 1.2546,
      "step": 2523
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.331021118982438e-05,
      "loss": 1.1991,
      "step": 2524
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.3281902775697533e-05,
      "loss": 1.243,
      "step": 2525
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.3253618502962872e-05,
      "loss": 1.2207,
      "step": 2526
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.3225358397176868e-05,
      "loss": 1.2564,
      "step": 2527
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.319712248387424e-05,
      "loss": 1.2266,
      "step": 2528
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.3168910788567812e-05,
      "loss": 1.1931,
      "step": 2529
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.3140723336748554e-05,
      "loss": 1.2316,
      "step": 2530
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.3112560153885467e-05,
      "loss": 1.2273,
      "step": 2531
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.3084421265425689e-05,
      "loss": 1.2325,
      "step": 2532
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.3056306696794377e-05,
      "loss": 1.2264,
      "step": 2533
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.3028216473394731e-05,
      "loss": 1.2132,
      "step": 2534
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.3000150620607914e-05,
      "loss": 1.2362,
      "step": 2535
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.297210916379311e-05,
      "loss": 1.2117,
      "step": 2536
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.2944092128287448e-05,
      "loss": 1.2321,
      "step": 2537
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.2916099539405997e-05,
      "loss": 1.2319,
      "step": 2538
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.2888131422441705e-05,
      "loss": 1.2277,
      "step": 2539
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.2860187802665443e-05,
      "loss": 1.2549,
      "step": 2540
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.2832268705325936e-05,
      "loss": 1.2441,
      "step": 2541
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.2804374155649768e-05,
      "loss": 1.2355,
      "step": 2542
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.2776504178841291e-05,
      "loss": 1.2515,
      "step": 2543
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.274865880008271e-05,
      "loss": 1.2325,
      "step": 2544
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.2720838044533981e-05,
      "loss": 1.235,
      "step": 2545
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.2693041937332828e-05,
      "loss": 1.2889,
      "step": 2546
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.266527050359466e-05,
      "loss": 1.1998,
      "step": 2547
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.2637523768412633e-05,
      "loss": 1.2619,
      "step": 2548
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.2609801756857584e-05,
      "loss": 1.2925,
      "step": 2549
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.2582104493978004e-05,
      "loss": 1.2543,
      "step": 2550
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.2554432004800004e-05,
      "loss": 1.2287,
      "step": 2551
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.2526784314327326e-05,
      "loss": 1.2638,
      "step": 2552
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.249916144754133e-05,
      "loss": 1.225,
      "step": 2553
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.2471563429400888e-05,
      "loss": 1.2406,
      "step": 2554
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.2443990284842471e-05,
      "loss": 1.248,
      "step": 2555
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.2416442038780052e-05,
      "loss": 1.236,
      "step": 2556
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.2388918716105134e-05,
      "loss": 1.2519,
      "step": 2557
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.2361420341686641e-05,
      "loss": 1.1818,
      "step": 2558
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.2333946940371013e-05,
      "loss": 1.227,
      "step": 2559
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.2306498536982101e-05,
      "loss": 1.2499,
      "step": 2560
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.2279075156321185e-05,
      "loss": 1.2406,
      "step": 2561
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.2251676823166902e-05,
      "loss": 1.2111,
      "step": 2562
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.22243035622753e-05,
      "loss": 1.2227,
      "step": 2563
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.219695539837972e-05,
      "loss": 1.2179,
      "step": 2564
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.2169632356190877e-05,
      "loss": 1.2715,
      "step": 2565
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.2142334460396775e-05,
      "loss": 1.2443,
      "step": 2566
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.2115061735662667e-05,
      "loss": 1.2308,
      "step": 2567
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.2087814206631095e-05,
      "loss": 1.2608,
      "step": 2568
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.2060591897921828e-05,
      "loss": 1.2301,
      "step": 2569
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.2033394834131849e-05,
      "loss": 1.2317,
      "step": 2570
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.20062230398353e-05,
      "loss": 1.2301,
      "step": 2571
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.1979076539583536e-05,
      "loss": 1.3083,
      "step": 2572
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.1951955357905033e-05,
      "loss": 1.2181,
      "step": 2573
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.1924859519305398e-05,
      "loss": 1.2385,
      "step": 2574
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.1897789048267324e-05,
      "loss": 1.2355,
      "step": 2575
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.18707439692506e-05,
      "loss": 1.2356,
      "step": 2576
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.1843724306692077e-05,
      "loss": 1.1748,
      "step": 2577
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.1816730085005603e-05,
      "loss": 1.2271,
      "step": 2578
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.1789761328582082e-05,
      "loss": 1.2511,
      "step": 2579
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.1762818061789391e-05,
      "loss": 1.2503,
      "step": 2580
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.1735900308972385e-05,
      "loss": 1.235,
      "step": 2581
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.1709008094452838e-05,
      "loss": 1.2292,
      "step": 2582
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.1682141442529473e-05,
      "loss": 1.2627,
      "step": 2583
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.1655300377477915e-05,
      "loss": 1.2375,
      "step": 2584
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.1628484923550673e-05,
      "loss": 1.2418,
      "step": 2585
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.1601695104977088e-05,
      "loss": 1.2385,
      "step": 2586
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.1574930945963367e-05,
      "loss": 1.2818,
      "step": 2587
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.1548192470692516e-05,
      "loss": 1.24,
      "step": 2588
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.152147970332436e-05,
      "loss": 1.2065,
      "step": 2589
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.1494792667995448e-05,
      "loss": 1.2575,
      "step": 2590
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.1468131388819117e-05,
      "loss": 1.2718,
      "step": 2591
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.1441495889885422e-05,
      "loss": 1.2654,
      "step": 2592
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.1414886195261135e-05,
      "loss": 1.1758,
      "step": 2593
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.1388302328989673e-05,
      "loss": 1.2959,
      "step": 2594
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.1361744315091156e-05,
      "loss": 1.1677,
      "step": 2595
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.1335212177562337e-05,
      "loss": 1.2087,
      "step": 2596
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.130870594037658e-05,
      "loss": 1.183,
      "step": 2597
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.128222562748384e-05,
      "loss": 1.2485,
      "step": 2598
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.1255771262810665e-05,
      "loss": 1.2639,
      "step": 2599
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.1229342870260145e-05,
      "loss": 1.1955,
      "step": 2600
    },
    {
      "epoch": 0.76,
      "eval_loss": 0.11455909162759781,
      "eval_runtime": 24.6076,
      "eval_samples_per_second": 81.276,
      "eval_steps_per_second": 0.65,
      "step": 2600
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.1202940473711936e-05,
      "loss": 1.2584,
      "step": 2601
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.117656409702212e-05,
      "loss": 1.2378,
      "step": 2602
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.1150213764023357e-05,
      "loss": 1.2373,
      "step": 2603
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.1123889498524738e-05,
      "loss": 1.1769,
      "step": 2604
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.1097591324311816e-05,
      "loss": 1.2034,
      "step": 2605
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.1071319265146534e-05,
      "loss": 1.2334,
      "step": 2606
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.1045073344767277e-05,
      "loss": 1.2454,
      "step": 2607
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.1018853586888794e-05,
      "loss": 1.2275,
      "step": 2608
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.0992660015202215e-05,
      "loss": 1.2286,
      "step": 2609
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.0966492653374958e-05,
      "loss": 1.2538,
      "step": 2610
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.0940351525050815e-05,
      "loss": 1.2119,
      "step": 2611
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.0914236653849848e-05,
      "loss": 1.2354,
      "step": 2612
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.0888148063368411e-05,
      "loss": 1.2354,
      "step": 2613
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.0862085777179078e-05,
      "loss": 1.2388,
      "step": 2614
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.0836049818830685e-05,
      "loss": 1.2042,
      "step": 2615
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.0810040211848269e-05,
      "loss": 1.147,
      "step": 2616
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.0784056979733078e-05,
      "loss": 1.2882,
      "step": 2617
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.075810014596248e-05,
      "loss": 1.2108,
      "step": 2618
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.0732169733990036e-05,
      "loss": 1.2464,
      "step": 2619
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.0706265767245419e-05,
      "loss": 1.2367,
      "step": 2620
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.068038826913441e-05,
      "loss": 1.2634,
      "step": 2621
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.0654537263038854e-05,
      "loss": 1.2037,
      "step": 2622
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.0628712772316682e-05,
      "loss": 1.2043,
      "step": 2623
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.060291482030186e-05,
      "loss": 1.2435,
      "step": 2624
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.0577143430304386e-05,
      "loss": 1.1848,
      "step": 2625
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.0551398625610227e-05,
      "loss": 1.2369,
      "step": 2626
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.0525680429481353e-05,
      "loss": 1.2523,
      "step": 2627
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.04999888651557e-05,
      "loss": 1.278,
      "step": 2628
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.0474323955847101e-05,
      "loss": 1.2587,
      "step": 2629
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.0448685724745347e-05,
      "loss": 1.1863,
      "step": 2630
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.0423074195016101e-05,
      "loss": 1.2123,
      "step": 2631
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.0397489389800918e-05,
      "loss": 1.244,
      "step": 2632
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.0371931332217175e-05,
      "loss": 1.2313,
      "step": 2633
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.034640004535811e-05,
      "loss": 1.1999,
      "step": 2634
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.0320895552292755e-05,
      "loss": 1.2186,
      "step": 2635
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.0295417876065957e-05,
      "loss": 1.2526,
      "step": 2636
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.0269967039698288e-05,
      "loss": 1.2198,
      "step": 2637
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.0244543066186102e-05,
      "loss": 1.1616,
      "step": 2638
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.0219145978501492e-05,
      "loss": 1.244,
      "step": 2639
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.0193775799592211e-05,
      "loss": 1.2348,
      "step": 2640
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.0168432552381757e-05,
      "loss": 1.2204,
      "step": 2641
    },
    {
      "epoch": 0.78,
      "learning_rate": 1.014311625976923e-05,
      "loss": 1.2469,
      "step": 2642
    },
    {
      "epoch": 0.78,
      "learning_rate": 1.0117826944629425e-05,
      "loss": 1.2617,
      "step": 2643
    },
    {
      "epoch": 0.78,
      "learning_rate": 1.0092564629812744e-05,
      "loss": 1.2143,
      "step": 2644
    },
    {
      "epoch": 0.78,
      "learning_rate": 1.0067329338145209e-05,
      "loss": 1.2124,
      "step": 2645
    },
    {
      "epoch": 0.78,
      "learning_rate": 1.0042121092428378e-05,
      "loss": 1.2113,
      "step": 2646
    },
    {
      "epoch": 0.78,
      "learning_rate": 1.001693991543942e-05,
      "loss": 1.3103,
      "step": 2647
    },
    {
      "epoch": 0.78,
      "learning_rate": 9.991785829931024e-06,
      "loss": 1.2004,
      "step": 2648
    },
    {
      "epoch": 0.78,
      "learning_rate": 9.966658858631421e-06,
      "loss": 1.2091,
      "step": 2649
    },
    {
      "epoch": 0.78,
      "learning_rate": 9.941559024244296e-06,
      "loss": 1.2327,
      "step": 2650
    },
    {
      "epoch": 0.78,
      "learning_rate": 9.916486349448862e-06,
      "loss": 1.248,
      "step": 2651
    },
    {
      "epoch": 0.78,
      "learning_rate": 9.891440856899783e-06,
      "loss": 1.1986,
      "step": 2652
    },
    {
      "epoch": 0.78,
      "learning_rate": 9.866422569227133e-06,
      "loss": 1.2516,
      "step": 2653
    },
    {
      "epoch": 0.78,
      "learning_rate": 9.84143150903643e-06,
      "loss": 1.2427,
      "step": 2654
    },
    {
      "epoch": 0.78,
      "learning_rate": 9.816467698908592e-06,
      "loss": 1.2454,
      "step": 2655
    },
    {
      "epoch": 0.78,
      "learning_rate": 9.791531161399911e-06,
      "loss": 1.2508,
      "step": 2656
    },
    {
      "epoch": 0.78,
      "learning_rate": 9.766621919042025e-06,
      "loss": 1.2134,
      "step": 2657
    },
    {
      "epoch": 0.78,
      "learning_rate": 9.74173999434192e-06,
      "loss": 1.2839,
      "step": 2658
    },
    {
      "epoch": 0.78,
      "learning_rate": 9.716885409781893e-06,
      "loss": 1.2116,
      "step": 2659
    },
    {
      "epoch": 0.78,
      "learning_rate": 9.692058187819561e-06,
      "loss": 1.1961,
      "step": 2660
    },
    {
      "epoch": 0.78,
      "learning_rate": 9.667258350887776e-06,
      "loss": 1.2083,
      "step": 2661
    },
    {
      "epoch": 0.78,
      "learning_rate": 9.642485921394673e-06,
      "loss": 1.2718,
      "step": 2662
    },
    {
      "epoch": 0.78,
      "learning_rate": 9.61774092172362e-06,
      "loss": 1.2387,
      "step": 2663
    },
    {
      "epoch": 0.78,
      "learning_rate": 9.593023374233206e-06,
      "loss": 1.2312,
      "step": 2664
    },
    {
      "epoch": 0.78,
      "learning_rate": 9.568333301257193e-06,
      "loss": 1.184,
      "step": 2665
    },
    {
      "epoch": 0.78,
      "learning_rate": 9.543670725104533e-06,
      "loss": 1.2414,
      "step": 2666
    },
    {
      "epoch": 0.78,
      "learning_rate": 9.519035668059345e-06,
      "loss": 1.243,
      "step": 2667
    },
    {
      "epoch": 0.78,
      "learning_rate": 9.494428152380868e-06,
      "loss": 1.2126,
      "step": 2668
    },
    {
      "epoch": 0.78,
      "learning_rate": 9.469848200303446e-06,
      "loss": 1.265,
      "step": 2669
    },
    {
      "epoch": 0.78,
      "learning_rate": 9.445295834036537e-06,
      "loss": 1.1648,
      "step": 2670
    },
    {
      "epoch": 0.78,
      "learning_rate": 9.420771075764663e-06,
      "loss": 1.2899,
      "step": 2671
    },
    {
      "epoch": 0.78,
      "learning_rate": 9.39627394764742e-06,
      "loss": 1.2146,
      "step": 2672
    },
    {
      "epoch": 0.78,
      "learning_rate": 9.371804471819401e-06,
      "loss": 1.2437,
      "step": 2673
    },
    {
      "epoch": 0.78,
      "learning_rate": 9.347362670390242e-06,
      "loss": 1.2059,
      "step": 2674
    },
    {
      "epoch": 0.78,
      "learning_rate": 9.322948565444565e-06,
      "loss": 1.2322,
      "step": 2675
    },
    {
      "epoch": 0.79,
      "learning_rate": 9.298562179041996e-06,
      "loss": 1.1642,
      "step": 2676
    },
    {
      "epoch": 0.79,
      "learning_rate": 9.27420353321704e-06,
      "loss": 1.2458,
      "step": 2677
    },
    {
      "epoch": 0.79,
      "learning_rate": 9.249872649979203e-06,
      "loss": 1.2401,
      "step": 2678
    },
    {
      "epoch": 0.79,
      "learning_rate": 9.225569551312894e-06,
      "loss": 1.1887,
      "step": 2679
    },
    {
      "epoch": 0.79,
      "learning_rate": 9.201294259177413e-06,
      "loss": 1.225,
      "step": 2680
    },
    {
      "epoch": 0.79,
      "learning_rate": 9.177046795506918e-06,
      "loss": 1.2292,
      "step": 2681
    },
    {
      "epoch": 0.79,
      "learning_rate": 9.152827182210445e-06,
      "loss": 1.2199,
      "step": 2682
    },
    {
      "epoch": 0.79,
      "learning_rate": 9.128635441171854e-06,
      "loss": 1.1984,
      "step": 2683
    },
    {
      "epoch": 0.79,
      "learning_rate": 9.10447159424984e-06,
      "loss": 1.2554,
      "step": 2684
    },
    {
      "epoch": 0.79,
      "learning_rate": 9.080335663277853e-06,
      "loss": 1.2628,
      "step": 2685
    },
    {
      "epoch": 0.79,
      "learning_rate": 9.056227670064155e-06,
      "loss": 1.2014,
      "step": 2686
    },
    {
      "epoch": 0.79,
      "learning_rate": 9.032147636391752e-06,
      "loss": 1.2398,
      "step": 2687
    },
    {
      "epoch": 0.79,
      "learning_rate": 9.008095584018406e-06,
      "loss": 1.279,
      "step": 2688
    },
    {
      "epoch": 0.79,
      "learning_rate": 8.984071534676553e-06,
      "loss": 1.2308,
      "step": 2689
    },
    {
      "epoch": 0.79,
      "learning_rate": 8.960075510073358e-06,
      "loss": 1.2382,
      "step": 2690
    },
    {
      "epoch": 0.79,
      "learning_rate": 8.936107531890665e-06,
      "loss": 1.2046,
      "step": 2691
    },
    {
      "epoch": 0.79,
      "learning_rate": 8.912167621784981e-06,
      "loss": 1.2601,
      "step": 2692
    },
    {
      "epoch": 0.79,
      "learning_rate": 8.888255801387417e-06,
      "loss": 1.212,
      "step": 2693
    },
    {
      "epoch": 0.79,
      "learning_rate": 8.86437209230374e-06,
      "loss": 1.2477,
      "step": 2694
    },
    {
      "epoch": 0.79,
      "learning_rate": 8.840516516114297e-06,
      "loss": 1.2104,
      "step": 2695
    },
    {
      "epoch": 0.79,
      "learning_rate": 8.81668909437404e-06,
      "loss": 1.264,
      "step": 2696
    },
    {
      "epoch": 0.79,
      "learning_rate": 8.79288984861244e-06,
      "loss": 1.2217,
      "step": 2697
    },
    {
      "epoch": 0.79,
      "learning_rate": 8.76911880033354e-06,
      "loss": 1.1972,
      "step": 2698
    },
    {
      "epoch": 0.79,
      "learning_rate": 8.745375971015905e-06,
      "loss": 1.2174,
      "step": 2699
    },
    {
      "epoch": 0.79,
      "learning_rate": 8.721661382112598e-06,
      "loss": 1.2837,
      "step": 2700
    },
    {
      "epoch": 0.79,
      "eval_loss": 0.11844494938850403,
      "eval_runtime": 24.3874,
      "eval_samples_per_second": 82.01,
      "eval_steps_per_second": 0.656,
      "step": 2700
    },
    {
      "epoch": 0.79,
      "learning_rate": 8.697975055051148e-06,
      "loss": 1.2068,
      "step": 2701
    },
    {
      "epoch": 0.79,
      "learning_rate": 8.67431701123357e-06,
      "loss": 1.2612,
      "step": 2702
    },
    {
      "epoch": 0.79,
      "learning_rate": 8.65068727203633e-06,
      "loss": 1.2669,
      "step": 2703
    },
    {
      "epoch": 0.79,
      "learning_rate": 8.627085858810273e-06,
      "loss": 1.234,
      "step": 2704
    },
    {
      "epoch": 0.79,
      "learning_rate": 8.603512792880706e-06,
      "loss": 1.1902,
      "step": 2705
    },
    {
      "epoch": 0.79,
      "learning_rate": 8.579968095547287e-06,
      "loss": 1.1747,
      "step": 2706
    },
    {
      "epoch": 0.79,
      "learning_rate": 8.556451788084068e-06,
      "loss": 1.2325,
      "step": 2707
    },
    {
      "epoch": 0.79,
      "learning_rate": 8.53296389173941e-06,
      "loss": 1.1815,
      "step": 2708
    },
    {
      "epoch": 0.79,
      "learning_rate": 8.509504427736042e-06,
      "loss": 1.2421,
      "step": 2709
    },
    {
      "epoch": 0.8,
      "learning_rate": 8.486073417270977e-06,
      "loss": 1.2172,
      "step": 2710
    },
    {
      "epoch": 0.8,
      "learning_rate": 8.462670881515551e-06,
      "loss": 1.2663,
      "step": 2711
    },
    {
      "epoch": 0.8,
      "learning_rate": 8.43929684161532e-06,
      "loss": 1.2126,
      "step": 2712
    },
    {
      "epoch": 0.8,
      "learning_rate": 8.415951318690134e-06,
      "loss": 1.2192,
      "step": 2713
    },
    {
      "epoch": 0.8,
      "learning_rate": 8.392634333834073e-06,
      "loss": 1.2315,
      "step": 2714
    },
    {
      "epoch": 0.8,
      "learning_rate": 8.369345908115405e-06,
      "loss": 1.2324,
      "step": 2715
    },
    {
      "epoch": 0.8,
      "learning_rate": 8.346086062576625e-06,
      "loss": 1.2132,
      "step": 2716
    },
    {
      "epoch": 0.8,
      "learning_rate": 8.32285481823437e-06,
      "loss": 1.1859,
      "step": 2717
    },
    {
      "epoch": 0.8,
      "learning_rate": 8.299652196079462e-06,
      "loss": 1.246,
      "step": 2718
    },
    {
      "epoch": 0.8,
      "learning_rate": 8.276478217076853e-06,
      "loss": 1.2097,
      "step": 2719
    },
    {
      "epoch": 0.8,
      "learning_rate": 8.253332902165626e-06,
      "loss": 1.2351,
      "step": 2720
    },
    {
      "epoch": 0.8,
      "learning_rate": 8.230216272258924e-06,
      "loss": 1.256,
      "step": 2721
    },
    {
      "epoch": 0.8,
      "learning_rate": 8.20712834824402e-06,
      "loss": 1.2387,
      "step": 2722
    },
    {
      "epoch": 0.8,
      "learning_rate": 8.184069150982217e-06,
      "loss": 1.2777,
      "step": 2723
    },
    {
      "epoch": 0.8,
      "learning_rate": 8.161038701308892e-06,
      "loss": 1.2382,
      "step": 2724
    },
    {
      "epoch": 0.8,
      "learning_rate": 8.138037020033405e-06,
      "loss": 1.2049,
      "step": 2725
    },
    {
      "epoch": 0.8,
      "learning_rate": 8.115064127939152e-06,
      "loss": 1.2584,
      "step": 2726
    },
    {
      "epoch": 0.8,
      "learning_rate": 8.092120045783525e-06,
      "loss": 1.1788,
      "step": 2727
    },
    {
      "epoch": 0.8,
      "learning_rate": 8.069204794297843e-06,
      "loss": 1.2454,
      "step": 2728
    },
    {
      "epoch": 0.8,
      "learning_rate": 8.046318394187413e-06,
      "loss": 1.2364,
      "step": 2729
    },
    {
      "epoch": 0.8,
      "learning_rate": 8.02346086613146e-06,
      "loss": 1.2618,
      "step": 2730
    },
    {
      "epoch": 0.8,
      "learning_rate": 8.000632230783139e-06,
      "loss": 1.2176,
      "step": 2731
    },
    {
      "epoch": 0.8,
      "learning_rate": 7.97783250876945e-06,
      "loss": 1.2166,
      "step": 2732
    },
    {
      "epoch": 0.8,
      "learning_rate": 7.955061720691315e-06,
      "loss": 1.2745,
      "step": 2733
    },
    {
      "epoch": 0.8,
      "learning_rate": 7.932319887123498e-06,
      "loss": 1.2356,
      "step": 2734
    },
    {
      "epoch": 0.8,
      "learning_rate": 7.909607028614608e-06,
      "loss": 1.2045,
      "step": 2735
    },
    {
      "epoch": 0.8,
      "learning_rate": 7.886923165687048e-06,
      "loss": 1.2222,
      "step": 2736
    },
    {
      "epoch": 0.8,
      "learning_rate": 7.864268318837051e-06,
      "loss": 1.2042,
      "step": 2737
    },
    {
      "epoch": 0.8,
      "learning_rate": 7.841642508534617e-06,
      "loss": 1.2726,
      "step": 2738
    },
    {
      "epoch": 0.8,
      "learning_rate": 7.81904575522352e-06,
      "loss": 1.2902,
      "step": 2739
    },
    {
      "epoch": 0.8,
      "learning_rate": 7.79647807932126e-06,
      "loss": 1.2236,
      "step": 2740
    },
    {
      "epoch": 0.8,
      "learning_rate": 7.773939501219079e-06,
      "loss": 1.2082,
      "step": 2741
    },
    {
      "epoch": 0.8,
      "learning_rate": 7.751430041281933e-06,
      "loss": 1.2309,
      "step": 2742
    },
    {
      "epoch": 0.8,
      "learning_rate": 7.728949719848464e-06,
      "loss": 1.22,
      "step": 2743
    },
    {
      "epoch": 0.81,
      "learning_rate": 7.706498557230957e-06,
      "loss": 1.2023,
      "step": 2744
    },
    {
      "epoch": 0.81,
      "learning_rate": 7.684076573715398e-06,
      "loss": 1.2432,
      "step": 2745
    },
    {
      "epoch": 0.81,
      "learning_rate": 7.661683789561372e-06,
      "loss": 1.2366,
      "step": 2746
    },
    {
      "epoch": 0.81,
      "learning_rate": 7.639320225002106e-06,
      "loss": 1.2486,
      "step": 2747
    },
    {
      "epoch": 0.81,
      "learning_rate": 7.6169859002443954e-06,
      "loss": 1.2309,
      "step": 2748
    },
    {
      "epoch": 0.81,
      "learning_rate": 7.594680835468642e-06,
      "loss": 1.207,
      "step": 2749
    },
    {
      "epoch": 0.81,
      "learning_rate": 7.5724050508288e-06,
      "loss": 1.201,
      "step": 2750
    },
    {
      "epoch": 0.81,
      "learning_rate": 7.55015856645239e-06,
      "loss": 1.1928,
      "step": 2751
    },
    {
      "epoch": 0.81,
      "learning_rate": 7.527941402440388e-06,
      "loss": 1.208,
      "step": 2752
    },
    {
      "epoch": 0.81,
      "learning_rate": 7.505753578867354e-06,
      "loss": 1.2131,
      "step": 2753
    },
    {
      "epoch": 0.81,
      "learning_rate": 7.483595115781304e-06,
      "loss": 1.2215,
      "step": 2754
    },
    {
      "epoch": 0.81,
      "learning_rate": 7.461466033203737e-06,
      "loss": 1.2117,
      "step": 2755
    },
    {
      "epoch": 0.81,
      "learning_rate": 7.439366351129584e-06,
      "loss": 1.2395,
      "step": 2756
    },
    {
      "epoch": 0.81,
      "learning_rate": 7.417296089527224e-06,
      "loss": 1.1907,
      "step": 2757
    },
    {
      "epoch": 0.81,
      "learning_rate": 7.395255268338459e-06,
      "loss": 1.2015,
      "step": 2758
    },
    {
      "epoch": 0.81,
      "learning_rate": 7.3732439074784975e-06,
      "loss": 1.1883,
      "step": 2759
    },
    {
      "epoch": 0.81,
      "learning_rate": 7.351262026835888e-06,
      "loss": 1.2204,
      "step": 2760
    },
    {
      "epoch": 0.81,
      "learning_rate": 7.329309646272578e-06,
      "loss": 1.2139,
      "step": 2761
    },
    {
      "epoch": 0.81,
      "learning_rate": 7.3073867856238605e-06,
      "loss": 1.17,
      "step": 2762
    },
    {
      "epoch": 0.81,
      "learning_rate": 7.285493464698343e-06,
      "loss": 1.2624,
      "step": 2763
    },
    {
      "epoch": 0.81,
      "learning_rate": 7.263629703277937e-06,
      "loss": 1.1784,
      "step": 2764
    },
    {
      "epoch": 0.81,
      "learning_rate": 7.241795521117851e-06,
      "loss": 1.2237,
      "step": 2765
    },
    {
      "epoch": 0.81,
      "learning_rate": 7.219990937946581e-06,
      "loss": 1.2276,
      "step": 2766
    },
    {
      "epoch": 0.81,
      "learning_rate": 7.19821597346587e-06,
      "loss": 1.2761,
      "step": 2767
    },
    {
      "epoch": 0.81,
      "learning_rate": 7.176470647350675e-06,
      "loss": 1.211,
      "step": 2768
    },
    {
      "epoch": 0.81,
      "learning_rate": 7.1547549792492055e-06,
      "loss": 1.2301,
      "step": 2769
    },
    {
      "epoch": 0.81,
      "learning_rate": 7.1330689887828584e-06,
      "loss": 1.1788,
      "step": 2770
    },
    {
      "epoch": 0.81,
      "learning_rate": 7.111412695546236e-06,
      "loss": 1.2365,
      "step": 2771
    },
    {
      "epoch": 0.81,
      "learning_rate": 7.0897861191070535e-06,
      "loss": 1.2359,
      "step": 2772
    },
    {
      "epoch": 0.81,
      "learning_rate": 7.068189279006237e-06,
      "loss": 1.2164,
      "step": 2773
    },
    {
      "epoch": 0.81,
      "learning_rate": 7.046622194757806e-06,
      "loss": 1.2178,
      "step": 2774
    },
    {
      "epoch": 0.81,
      "learning_rate": 7.025084885848925e-06,
      "loss": 1.2217,
      "step": 2775
    },
    {
      "epoch": 0.81,
      "learning_rate": 7.003577371739809e-06,
      "loss": 1.2481,
      "step": 2776
    },
    {
      "epoch": 0.81,
      "learning_rate": 6.982099671863789e-06,
      "loss": 1.2377,
      "step": 2777
    },
    {
      "epoch": 0.82,
      "learning_rate": 6.960651805627248e-06,
      "loss": 1.2646,
      "step": 2778
    },
    {
      "epoch": 0.82,
      "learning_rate": 6.939233792409617e-06,
      "loss": 1.248,
      "step": 2779
    },
    {
      "epoch": 0.82,
      "learning_rate": 6.917845651563327e-06,
      "loss": 1.2326,
      "step": 2780
    },
    {
      "epoch": 0.82,
      "learning_rate": 6.896487402413852e-06,
      "loss": 1.2271,
      "step": 2781
    },
    {
      "epoch": 0.82,
      "learning_rate": 6.875159064259649e-06,
      "loss": 1.2621,
      "step": 2782
    },
    {
      "epoch": 0.82,
      "learning_rate": 6.853860656372125e-06,
      "loss": 1.2149,
      "step": 2783
    },
    {
      "epoch": 0.82,
      "learning_rate": 6.832592197995662e-06,
      "loss": 1.2139,
      "step": 2784
    },
    {
      "epoch": 0.82,
      "learning_rate": 6.8113537083475875e-06,
      "loss": 1.215,
      "step": 2785
    },
    {
      "epoch": 0.82,
      "learning_rate": 6.790145206618146e-06,
      "loss": 1.2323,
      "step": 2786
    },
    {
      "epoch": 0.82,
      "learning_rate": 6.768966711970466e-06,
      "loss": 1.1941,
      "step": 2787
    },
    {
      "epoch": 0.82,
      "learning_rate": 6.747818243540587e-06,
      "loss": 1.2535,
      "step": 2788
    },
    {
      "epoch": 0.82,
      "learning_rate": 6.726699820437419e-06,
      "loss": 1.2215,
      "step": 2789
    },
    {
      "epoch": 0.82,
      "learning_rate": 6.7056114617426935e-06,
      "loss": 1.227,
      "step": 2790
    },
    {
      "epoch": 0.82,
      "learning_rate": 6.684553186511022e-06,
      "loss": 1.2417,
      "step": 2791
    },
    {
      "epoch": 0.82,
      "learning_rate": 6.6635250137697935e-06,
      "loss": 1.2695,
      "step": 2792
    },
    {
      "epoch": 0.82,
      "learning_rate": 6.642526962519218e-06,
      "loss": 1.2686,
      "step": 2793
    },
    {
      "epoch": 0.82,
      "learning_rate": 6.621559051732287e-06,
      "loss": 1.2237,
      "step": 2794
    },
    {
      "epoch": 0.82,
      "learning_rate": 6.60062130035477e-06,
      "loss": 1.2168,
      "step": 2795
    },
    {
      "epoch": 0.82,
      "learning_rate": 6.579713727305148e-06,
      "loss": 1.2229,
      "step": 2796
    },
    {
      "epoch": 0.82,
      "learning_rate": 6.558836351474673e-06,
      "loss": 1.2629,
      "step": 2797
    },
    {
      "epoch": 0.82,
      "learning_rate": 6.537989191727292e-06,
      "loss": 1.2359,
      "step": 2798
    },
    {
      "epoch": 0.82,
      "learning_rate": 6.517172266899669e-06,
      "loss": 1.2424,
      "step": 2799
    },
    {
      "epoch": 0.82,
      "learning_rate": 6.496385595801116e-06,
      "loss": 1.2364,
      "step": 2800
    },
    {
      "epoch": 0.82,
      "eval_loss": 0.11666639894247055,
      "eval_runtime": 24.6904,
      "eval_samples_per_second": 81.003,
      "eval_steps_per_second": 0.648,
      "step": 2800
    },
    {
      "epoch": 0.82,
      "learning_rate": 6.475629197213637e-06,
      "loss": 1.2704,
      "step": 2801
    },
    {
      "epoch": 0.82,
      "learning_rate": 6.454903089891878e-06,
      "loss": 1.2297,
      "step": 2802
    },
    {
      "epoch": 0.82,
      "learning_rate": 6.4342072925631125e-06,
      "loss": 1.2271,
      "step": 2803
    },
    {
      "epoch": 0.82,
      "learning_rate": 6.413541823927216e-06,
      "loss": 1.2392,
      "step": 2804
    },
    {
      "epoch": 0.82,
      "learning_rate": 6.3929067026566735e-06,
      "loss": 1.2415,
      "step": 2805
    },
    {
      "epoch": 0.82,
      "learning_rate": 6.3723019473965575e-06,
      "loss": 1.2173,
      "step": 2806
    },
    {
      "epoch": 0.82,
      "learning_rate": 6.351727576764477e-06,
      "loss": 1.183,
      "step": 2807
    },
    {
      "epoch": 0.82,
      "learning_rate": 6.331183609350602e-06,
      "loss": 1.2815,
      "step": 2808
    },
    {
      "epoch": 0.82,
      "learning_rate": 6.310670063717638e-06,
      "loss": 1.2043,
      "step": 2809
    },
    {
      "epoch": 0.82,
      "learning_rate": 6.2901869584008014e-06,
      "loss": 1.2583,
      "step": 2810
    },
    {
      "epoch": 0.82,
      "learning_rate": 6.269734311907782e-06,
      "loss": 1.204,
      "step": 2811
    },
    {
      "epoch": 0.82,
      "learning_rate": 6.249312142718768e-06,
      "loss": 1.247,
      "step": 2812
    },
    {
      "epoch": 0.83,
      "learning_rate": 6.228920469286408e-06,
      "loss": 1.1945,
      "step": 2813
    },
    {
      "epoch": 0.83,
      "learning_rate": 6.208559310035807e-06,
      "loss": 1.1955,
      "step": 2814
    },
    {
      "epoch": 0.83,
      "learning_rate": 6.188228683364461e-06,
      "loss": 1.2397,
      "step": 2815
    },
    {
      "epoch": 0.83,
      "learning_rate": 6.167928607642317e-06,
      "loss": 1.2503,
      "step": 2816
    },
    {
      "epoch": 0.83,
      "learning_rate": 6.147659101211694e-06,
      "loss": 1.2377,
      "step": 2817
    },
    {
      "epoch": 0.83,
      "learning_rate": 6.127420182387314e-06,
      "loss": 1.2282,
      "step": 2818
    },
    {
      "epoch": 0.83,
      "learning_rate": 6.107211869456228e-06,
      "loss": 1.2269,
      "step": 2819
    },
    {
      "epoch": 0.83,
      "learning_rate": 6.087034180677855e-06,
      "loss": 1.2014,
      "step": 2820
    },
    {
      "epoch": 0.83,
      "learning_rate": 6.066887134283939e-06,
      "loss": 1.1965,
      "step": 2821
    },
    {
      "epoch": 0.83,
      "learning_rate": 6.046770748478552e-06,
      "loss": 1.2002,
      "step": 2822
    },
    {
      "epoch": 0.83,
      "learning_rate": 6.026685041438018e-06,
      "loss": 1.2899,
      "step": 2823
    },
    {
      "epoch": 0.83,
      "learning_rate": 6.006630031310989e-06,
      "loss": 1.2434,
      "step": 2824
    },
    {
      "epoch": 0.83,
      "learning_rate": 5.986605736218352e-06,
      "loss": 1.2274,
      "step": 2825
    },
    {
      "epoch": 0.83,
      "learning_rate": 5.9666121742532765e-06,
      "loss": 1.2363,
      "step": 2826
    },
    {
      "epoch": 0.83,
      "learning_rate": 5.9466493634810965e-06,
      "loss": 1.2649,
      "step": 2827
    },
    {
      "epoch": 0.83,
      "learning_rate": 5.926717321939417e-06,
      "loss": 1.1863,
      "step": 2828
    },
    {
      "epoch": 0.83,
      "learning_rate": 5.906816067638028e-06,
      "loss": 1.2161,
      "step": 2829
    },
    {
      "epoch": 0.83,
      "learning_rate": 5.886945618558906e-06,
      "loss": 1.2488,
      "step": 2830
    },
    {
      "epoch": 0.83,
      "learning_rate": 5.867105992656163e-06,
      "loss": 1.252,
      "step": 2831
    },
    {
      "epoch": 0.83,
      "learning_rate": 5.8472972078560975e-06,
      "loss": 1.209,
      "step": 2832
    },
    {
      "epoch": 0.83,
      "learning_rate": 5.827519282057128e-06,
      "loss": 1.2218,
      "step": 2833
    },
    {
      "epoch": 0.83,
      "learning_rate": 5.807772233129792e-06,
      "loss": 1.2406,
      "step": 2834
    },
    {
      "epoch": 0.83,
      "learning_rate": 5.788056078916704e-06,
      "loss": 1.2355,
      "step": 2835
    },
    {
      "epoch": 0.83,
      "learning_rate": 5.7683708372325975e-06,
      "loss": 1.2175,
      "step": 2836
    },
    {
      "epoch": 0.83,
      "learning_rate": 5.748716525864262e-06,
      "loss": 1.2035,
      "step": 2837
    },
    {
      "epoch": 0.83,
      "learning_rate": 5.729093162570545e-06,
      "loss": 1.2608,
      "step": 2838
    },
    {
      "epoch": 0.83,
      "learning_rate": 5.709500765082307e-06,
      "loss": 1.2076,
      "step": 2839
    },
    {
      "epoch": 0.83,
      "learning_rate": 5.689939351102456e-06,
      "loss": 1.2484,
      "step": 2840
    },
    {
      "epoch": 0.83,
      "learning_rate": 5.6704089383058915e-06,
      "loss": 1.2032,
      "step": 2841
    },
    {
      "epoch": 0.83,
      "learning_rate": 5.650909544339511e-06,
      "loss": 1.2321,
      "step": 2842
    },
    {
      "epoch": 0.83,
      "learning_rate": 5.631441186822168e-06,
      "loss": 1.1893,
      "step": 2843
    },
    {
      "epoch": 0.83,
      "learning_rate": 5.61200388334469e-06,
      "loss": 1.2397,
      "step": 2844
    },
    {
      "epoch": 0.83,
      "learning_rate": 5.592597651469835e-06,
      "loss": 1.2058,
      "step": 2845
    },
    {
      "epoch": 0.83,
      "learning_rate": 5.573222508732303e-06,
      "loss": 1.2172,
      "step": 2846
    },
    {
      "epoch": 0.84,
      "learning_rate": 5.553878472638667e-06,
      "loss": 1.2746,
      "step": 2847
    },
    {
      "epoch": 0.84,
      "learning_rate": 5.534565560667426e-06,
      "loss": 1.1764,
      "step": 2848
    },
    {
      "epoch": 0.84,
      "learning_rate": 5.515283790268946e-06,
      "loss": 1.2196,
      "step": 2849
    },
    {
      "epoch": 0.84,
      "learning_rate": 5.496033178865463e-06,
      "loss": 1.2319,
      "step": 2850
    },
    {
      "epoch": 0.84,
      "learning_rate": 5.476813743851037e-06,
      "loss": 1.1739,
      "step": 2851
    },
    {
      "epoch": 0.84,
      "learning_rate": 5.457625502591582e-06,
      "loss": 1.2115,
      "step": 2852
    },
    {
      "epoch": 0.84,
      "learning_rate": 5.4384684724248096e-06,
      "loss": 1.2597,
      "step": 2853
    },
    {
      "epoch": 0.84,
      "learning_rate": 5.419342670660248e-06,
      "loss": 1.2105,
      "step": 2854
    },
    {
      "epoch": 0.84,
      "learning_rate": 5.4002481145791895e-06,
      "loss": 1.2293,
      "step": 2855
    },
    {
      "epoch": 0.84,
      "learning_rate": 5.381184821434699e-06,
      "loss": 1.2067,
      "step": 2856
    },
    {
      "epoch": 0.84,
      "learning_rate": 5.362152808451618e-06,
      "loss": 1.2765,
      "step": 2857
    },
    {
      "epoch": 0.84,
      "learning_rate": 5.343152092826476e-06,
      "loss": 1.2315,
      "step": 2858
    },
    {
      "epoch": 0.84,
      "learning_rate": 5.324182691727564e-06,
      "loss": 1.2039,
      "step": 2859
    },
    {
      "epoch": 0.84,
      "learning_rate": 5.30524462229487e-06,
      "loss": 1.2252,
      "step": 2860
    },
    {
      "epoch": 0.84,
      "learning_rate": 5.286337901640069e-06,
      "loss": 1.2742,
      "step": 2861
    },
    {
      "epoch": 0.84,
      "learning_rate": 5.267462546846495e-06,
      "loss": 1.1883,
      "step": 2862
    },
    {
      "epoch": 0.84,
      "learning_rate": 5.2486185749691665e-06,
      "loss": 1.2121,
      "step": 2863
    },
    {
      "epoch": 0.84,
      "learning_rate": 5.229806003034741e-06,
      "loss": 1.2229,
      "step": 2864
    },
    {
      "epoch": 0.84,
      "learning_rate": 5.211024848041479e-06,
      "loss": 1.2716,
      "step": 2865
    },
    {
      "epoch": 0.84,
      "learning_rate": 5.192275126959292e-06,
      "loss": 1.2361,
      "step": 2866
    },
    {
      "epoch": 0.84,
      "learning_rate": 5.173556856729649e-06,
      "loss": 1.1855,
      "step": 2867
    },
    {
      "epoch": 0.84,
      "learning_rate": 5.154870054265635e-06,
      "loss": 1.2309,
      "step": 2868
    },
    {
      "epoch": 0.84,
      "learning_rate": 5.1362147364518854e-06,
      "loss": 1.2629,
      "step": 2869
    },
    {
      "epoch": 0.84,
      "learning_rate": 5.117590920144606e-06,
      "loss": 1.2222,
      "step": 2870
    },
    {
      "epoch": 0.84,
      "learning_rate": 5.0989986221715006e-06,
      "loss": 1.2088,
      "step": 2871
    },
    {
      "epoch": 0.84,
      "learning_rate": 5.080437859331823e-06,
      "loss": 1.231,
      "step": 2872
    },
    {
      "epoch": 0.84,
      "learning_rate": 5.061908648396338e-06,
      "loss": 1.2033,
      "step": 2873
    },
    {
      "epoch": 0.84,
      "learning_rate": 5.043411006107293e-06,
      "loss": 1.2219,
      "step": 2874
    },
    {
      "epoch": 0.84,
      "learning_rate": 5.024944949178392e-06,
      "loss": 1.2036,
      "step": 2875
    },
    {
      "epoch": 0.84,
      "learning_rate": 5.006510494294831e-06,
      "loss": 1.2974,
      "step": 2876
    },
    {
      "epoch": 0.84,
      "learning_rate": 4.988107658113235e-06,
      "loss": 1.1788,
      "step": 2877
    },
    {
      "epoch": 0.84,
      "learning_rate": 4.969736457261669e-06,
      "loss": 1.2012,
      "step": 2878
    },
    {
      "epoch": 0.84,
      "learning_rate": 4.951396908339594e-06,
      "loss": 1.2058,
      "step": 2879
    },
    {
      "epoch": 0.84,
      "learning_rate": 4.933089027917892e-06,
      "loss": 1.2347,
      "step": 2880
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.914812832538829e-06,
      "loss": 1.2341,
      "step": 2881
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.8965683387160215e-06,
      "loss": 1.2282,
      "step": 2882
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.878355562934465e-06,
      "loss": 1.2579,
      "step": 2883
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.860174521650485e-06,
      "loss": 1.2452,
      "step": 2884
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.8420252312917445e-06,
      "loss": 1.2058,
      "step": 2885
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.823907708257198e-06,
      "loss": 1.2358,
      "step": 2886
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.805821968917106e-06,
      "loss": 1.2515,
      "step": 2887
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.787768029613014e-06,
      "loss": 1.2075,
      "step": 2888
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.769745906657739e-06,
      "loss": 1.2191,
      "step": 2889
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.7517556163353275e-06,
      "loss": 1.1623,
      "step": 2890
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.733797174901087e-06,
      "loss": 1.2402,
      "step": 2891
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.71587059858154e-06,
      "loss": 1.2316,
      "step": 2892
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.697975903574419e-06,
      "loss": 1.1947,
      "step": 2893
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.6801131060486334e-06,
      "loss": 1.1914,
      "step": 2894
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.662282222144297e-06,
      "loss": 1.2065,
      "step": 2895
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.6444832679726705e-06,
      "loss": 1.203,
      "step": 2896
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.626716259616175e-06,
      "loss": 1.2354,
      "step": 2897
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.608981213128347e-06,
      "loss": 1.2328,
      "step": 2898
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.59127814453387e-06,
      "loss": 1.2053,
      "step": 2899
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.573607069828514e-06,
      "loss": 1.2304,
      "step": 2900
    },
    {
      "epoch": 0.85,
      "eval_loss": 0.1184241771697998,
      "eval_runtime": 24.4832,
      "eval_samples_per_second": 81.689,
      "eval_steps_per_second": 0.654,
      "step": 2900
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.555968004979163e-06,
      "loss": 1.2277,
      "step": 2901
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.538360965923745e-06,
      "loss": 1.2671,
      "step": 2902
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.520785968571266e-06,
      "loss": 1.1668,
      "step": 2903
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.503243028801794e-06,
      "loss": 1.2114,
      "step": 2904
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.485732162466421e-06,
      "loss": 1.204,
      "step": 2905
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.468253385387246e-06,
      "loss": 1.2262,
      "step": 2906
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.450806713357394e-06,
      "loss": 1.2251,
      "step": 2907
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.433392162140968e-06,
      "loss": 1.2646,
      "step": 2908
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.4160097474730624e-06,
      "loss": 1.147,
      "step": 2909
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.3986594850597084e-06,
      "loss": 1.2801,
      "step": 2910
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.381341390577909e-06,
      "loss": 1.2174,
      "step": 2911
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.364055479675595e-06,
      "loss": 1.2117,
      "step": 2912
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.3468017679716245e-06,
      "loss": 1.2398,
      "step": 2913
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.32958027105574e-06,
      "loss": 1.207,
      "step": 2914
    },
    {
      "epoch": 0.86,
      "learning_rate": 4.312391004488592e-06,
      "loss": 1.2114,
      "step": 2915
    },
    {
      "epoch": 0.86,
      "learning_rate": 4.2952339838017075e-06,
      "loss": 1.2373,
      "step": 2916
    },
    {
      "epoch": 0.86,
      "learning_rate": 4.278109224497486e-06,
      "loss": 1.2899,
      "step": 2917
    },
    {
      "epoch": 0.86,
      "learning_rate": 4.261016742049155e-06,
      "loss": 1.2266,
      "step": 2918
    },
    {
      "epoch": 0.86,
      "learning_rate": 4.2439565519007916e-06,
      "loss": 1.1913,
      "step": 2919
    },
    {
      "epoch": 0.86,
      "learning_rate": 4.2269286694673e-06,
      "loss": 1.2056,
      "step": 2920
    },
    {
      "epoch": 0.86,
      "learning_rate": 4.209933110134392e-06,
      "loss": 1.2724,
      "step": 2921
    },
    {
      "epoch": 0.86,
      "learning_rate": 4.192969889258552e-06,
      "loss": 1.1911,
      "step": 2922
    },
    {
      "epoch": 0.86,
      "learning_rate": 4.176039022167064e-06,
      "loss": 1.1573,
      "step": 2923
    },
    {
      "epoch": 0.86,
      "learning_rate": 4.159140524157983e-06,
      "loss": 1.261,
      "step": 2924
    },
    {
      "epoch": 0.86,
      "learning_rate": 4.142274410500111e-06,
      "loss": 1.244,
      "step": 2925
    },
    {
      "epoch": 0.86,
      "learning_rate": 4.12544069643297e-06,
      "loss": 1.2519,
      "step": 2926
    },
    {
      "epoch": 0.86,
      "learning_rate": 4.108639397166831e-06,
      "loss": 1.2032,
      "step": 2927
    },
    {
      "epoch": 0.86,
      "learning_rate": 4.0918705278826685e-06,
      "loss": 1.2047,
      "step": 2928
    },
    {
      "epoch": 0.86,
      "learning_rate": 4.075134103732166e-06,
      "loss": 1.2571,
      "step": 2929
    },
    {
      "epoch": 0.86,
      "learning_rate": 4.0584301398376615e-06,
      "loss": 1.2438,
      "step": 2930
    },
    {
      "epoch": 0.86,
      "learning_rate": 4.041758651292193e-06,
      "loss": 1.223,
      "step": 2931
    },
    {
      "epoch": 0.86,
      "learning_rate": 4.025119653159442e-06,
      "loss": 1.218,
      "step": 2932
    },
    {
      "epoch": 0.86,
      "learning_rate": 4.008513160473726e-06,
      "loss": 1.2348,
      "step": 2933
    },
    {
      "epoch": 0.86,
      "learning_rate": 3.991939188240009e-06,
      "loss": 1.2058,
      "step": 2934
    },
    {
      "epoch": 0.86,
      "learning_rate": 3.975397751433856e-06,
      "loss": 1.2057,
      "step": 2935
    },
    {
      "epoch": 0.86,
      "learning_rate": 3.958888865001451e-06,
      "loss": 1.2426,
      "step": 2936
    },
    {
      "epoch": 0.86,
      "learning_rate": 3.942412543859546e-06,
      "loss": 1.1983,
      "step": 2937
    },
    {
      "epoch": 0.86,
      "learning_rate": 3.925968802895477e-06,
      "loss": 1.1782,
      "step": 2938
    },
    {
      "epoch": 0.86,
      "learning_rate": 3.909557656967154e-06,
      "loss": 1.1886,
      "step": 2939
    },
    {
      "epoch": 0.86,
      "learning_rate": 3.893179120903012e-06,
      "loss": 1.2433,
      "step": 2940
    },
    {
      "epoch": 0.86,
      "learning_rate": 3.876833209502051e-06,
      "loss": 1.1762,
      "step": 2941
    },
    {
      "epoch": 0.86,
      "learning_rate": 3.860519937533753e-06,
      "loss": 1.2309,
      "step": 2942
    },
    {
      "epoch": 0.86,
      "learning_rate": 3.8442393197381456e-06,
      "loss": 1.1942,
      "step": 2943
    },
    {
      "epoch": 0.86,
      "learning_rate": 3.827991370825732e-06,
      "loss": 1.251,
      "step": 2944
    },
    {
      "epoch": 0.86,
      "learning_rate": 3.8117761054775116e-06,
      "loss": 1.1705,
      "step": 2945
    },
    {
      "epoch": 0.86,
      "learning_rate": 3.795593538344928e-06,
      "loss": 1.2067,
      "step": 2946
    },
    {
      "epoch": 0.86,
      "learning_rate": 3.7794436840498947e-06,
      "loss": 1.2723,
      "step": 2947
    },
    {
      "epoch": 0.86,
      "learning_rate": 3.7633265571847787e-06,
      "loss": 1.2135,
      "step": 2948
    },
    {
      "epoch": 0.87,
      "learning_rate": 3.7472421723123662e-06,
      "loss": 1.2081,
      "step": 2949
    },
    {
      "epoch": 0.87,
      "learning_rate": 3.731190543965837e-06,
      "loss": 1.2205,
      "step": 2950
    },
    {
      "epoch": 0.87,
      "learning_rate": 3.7151716866488065e-06,
      "loss": 1.2077,
      "step": 2951
    },
    {
      "epoch": 0.87,
      "learning_rate": 3.69918561483527e-06,
      "loss": 1.2172,
      "step": 2952
    },
    {
      "epoch": 0.87,
      "learning_rate": 3.683232342969594e-06,
      "loss": 1.2355,
      "step": 2953
    },
    {
      "epoch": 0.87,
      "learning_rate": 3.6673118854664956e-06,
      "loss": 1.1658,
      "step": 2954
    },
    {
      "epoch": 0.87,
      "learning_rate": 3.6514242567110738e-06,
      "loss": 1.2889,
      "step": 2955
    },
    {
      "epoch": 0.87,
      "learning_rate": 3.6355694710587463e-06,
      "loss": 1.1549,
      "step": 2956
    },
    {
      "epoch": 0.87,
      "learning_rate": 3.6197475428352415e-06,
      "loss": 1.2331,
      "step": 2957
    },
    {
      "epoch": 0.87,
      "learning_rate": 3.603958486336625e-06,
      "loss": 1.1841,
      "step": 2958
    },
    {
      "epoch": 0.87,
      "learning_rate": 3.5882023158292455e-06,
      "loss": 1.2429,
      "step": 2959
    },
    {
      "epoch": 0.87,
      "learning_rate": 3.5724790455497504e-06,
      "loss": 1.2262,
      "step": 2960
    },
    {
      "epoch": 0.87,
      "learning_rate": 3.5567886897050418e-06,
      "loss": 1.1866,
      "step": 2961
    },
    {
      "epoch": 0.87,
      "learning_rate": 3.5411312624722903e-06,
      "loss": 1.2479,
      "step": 2962
    },
    {
      "epoch": 0.87,
      "learning_rate": 3.5255067779989173e-06,
      "loss": 1.1975,
      "step": 2963
    },
    {
      "epoch": 0.87,
      "learning_rate": 3.50991525040258e-06,
      "loss": 1.2536,
      "step": 2964
    },
    {
      "epoch": 0.87,
      "learning_rate": 3.4943566937711438e-06,
      "loss": 1.1653,
      "step": 2965
    },
    {
      "epoch": 0.87,
      "learning_rate": 3.4788311221626914e-06,
      "loss": 1.2561,
      "step": 2966
    },
    {
      "epoch": 0.87,
      "learning_rate": 3.463338549605508e-06,
      "loss": 1.1945,
      "step": 2967
    },
    {
      "epoch": 0.87,
      "learning_rate": 3.4478789900980545e-06,
      "loss": 1.1836,
      "step": 2968
    },
    {
      "epoch": 0.87,
      "learning_rate": 3.432452457608957e-06,
      "loss": 1.23,
      "step": 2969
    },
    {
      "epoch": 0.87,
      "learning_rate": 3.4170589660770117e-06,
      "loss": 1.2527,
      "step": 2970
    },
    {
      "epoch": 0.87,
      "learning_rate": 3.4016985294111503e-06,
      "loss": 1.2219,
      "step": 2971
    },
    {
      "epoch": 0.87,
      "learning_rate": 3.386371161490458e-06,
      "loss": 1.2407,
      "step": 2972
    },
    {
      "epoch": 0.87,
      "learning_rate": 3.371076876164101e-06,
      "loss": 1.2047,
      "step": 2973
    },
    {
      "epoch": 0.87,
      "learning_rate": 3.355815687251389e-06,
      "loss": 1.2244,
      "step": 2974
    },
    {
      "epoch": 0.87,
      "learning_rate": 3.340587608541719e-06,
      "loss": 1.1801,
      "step": 2975
    },
    {
      "epoch": 0.87,
      "learning_rate": 3.3253926537945634e-06,
      "loss": 1.2002,
      "step": 2976
    },
    {
      "epoch": 0.87,
      "learning_rate": 3.3102308367394743e-06,
      "loss": 1.2499,
      "step": 2977
    },
    {
      "epoch": 0.87,
      "learning_rate": 3.295102171076039e-06,
      "loss": 1.1515,
      "step": 2978
    },
    {
      "epoch": 0.87,
      "learning_rate": 3.2800066704739185e-06,
      "loss": 1.2028,
      "step": 2979
    },
    {
      "epoch": 0.87,
      "learning_rate": 3.264944348572807e-06,
      "loss": 1.2098,
      "step": 2980
    },
    {
      "epoch": 0.87,
      "learning_rate": 3.2499152189823913e-06,
      "loss": 1.2218,
      "step": 2981
    },
    {
      "epoch": 0.87,
      "learning_rate": 3.234919295282395e-06,
      "loss": 1.1926,
      "step": 2982
    },
    {
      "epoch": 0.88,
      "learning_rate": 3.2199565910225305e-06,
      "loss": 1.2407,
      "step": 2983
    },
    {
      "epoch": 0.88,
      "learning_rate": 3.2050271197224903e-06,
      "loss": 1.1564,
      "step": 2984
    },
    {
      "epoch": 0.88,
      "learning_rate": 3.1901308948719368e-06,
      "loss": 1.2477,
      "step": 2985
    },
    {
      "epoch": 0.88,
      "learning_rate": 3.1752679299304944e-06,
      "loss": 1.2311,
      "step": 2986
    },
    {
      "epoch": 0.88,
      "learning_rate": 3.1604382383277453e-06,
      "loss": 1.2041,
      "step": 2987
    },
    {
      "epoch": 0.88,
      "learning_rate": 3.145641833463198e-06,
      "loss": 1.2169,
      "step": 2988
    },
    {
      "epoch": 0.88,
      "learning_rate": 3.1308787287062814e-06,
      "loss": 1.2583,
      "step": 2989
    },
    {
      "epoch": 0.88,
      "learning_rate": 3.1161489373963395e-06,
      "loss": 1.2083,
      "step": 2990
    },
    {
      "epoch": 0.88,
      "learning_rate": 3.101452472842619e-06,
      "loss": 1.2041,
      "step": 2991
    },
    {
      "epoch": 0.88,
      "learning_rate": 3.086789348324253e-06,
      "loss": 1.2512,
      "step": 2992
    },
    {
      "epoch": 0.88,
      "learning_rate": 3.0721595770902435e-06,
      "loss": 1.204,
      "step": 2993
    },
    {
      "epoch": 0.88,
      "learning_rate": 3.0575631723594613e-06,
      "loss": 1.2294,
      "step": 2994
    },
    {
      "epoch": 0.88,
      "learning_rate": 3.043000147320623e-06,
      "loss": 1.1997,
      "step": 2995
    },
    {
      "epoch": 0.88,
      "learning_rate": 3.0284705151323046e-06,
      "loss": 1.2148,
      "step": 2996
    },
    {
      "epoch": 0.88,
      "learning_rate": 3.013974288922876e-06,
      "loss": 1.2245,
      "step": 2997
    },
    {
      "epoch": 0.88,
      "learning_rate": 2.9995114817905493e-06,
      "loss": 1.2108,
      "step": 2998
    },
    {
      "epoch": 0.88,
      "learning_rate": 2.9850821068033274e-06,
      "loss": 1.1822,
      "step": 2999
    },
    {
      "epoch": 0.88,
      "learning_rate": 2.9706861769990225e-06,
      "loss": 1.24,
      "step": 3000
    },
    {
      "epoch": 0.88,
      "eval_loss": 0.11790976673364639,
      "eval_runtime": 24.5107,
      "eval_samples_per_second": 81.597,
      "eval_steps_per_second": 0.653,
      "step": 3000
    },
    {
      "epoch": 0.88,
      "learning_rate": 2.956323705385198e-06,
      "loss": 1.2293,
      "step": 3001
    },
    {
      "epoch": 0.88,
      "learning_rate": 2.941994704939215e-06,
      "loss": 1.2306,
      "step": 3002
    },
    {
      "epoch": 0.88,
      "learning_rate": 2.927699188608171e-06,
      "loss": 1.2018,
      "step": 3003
    },
    {
      "epoch": 0.88,
      "learning_rate": 2.9134371693089235e-06,
      "loss": 1.2241,
      "step": 3004
    },
    {
      "epoch": 0.88,
      "learning_rate": 2.89920865992805e-06,
      "loss": 1.2112,
      "step": 3005
    },
    {
      "epoch": 0.88,
      "learning_rate": 2.885013673321857e-06,
      "loss": 1.2146,
      "step": 3006
    },
    {
      "epoch": 0.88,
      "learning_rate": 2.8708522223163694e-06,
      "loss": 1.2694,
      "step": 3007
    },
    {
      "epoch": 0.88,
      "learning_rate": 2.856724319707289e-06,
      "loss": 1.2034,
      "step": 3008
    },
    {
      "epoch": 0.88,
      "learning_rate": 2.8426299782600186e-06,
      "loss": 1.1902,
      "step": 3009
    },
    {
      "epoch": 0.88,
      "learning_rate": 2.828569210709642e-06,
      "loss": 1.2194,
      "step": 3010
    },
    {
      "epoch": 0.88,
      "learning_rate": 2.8145420297609026e-06,
      "loss": 1.25,
      "step": 3011
    },
    {
      "epoch": 0.88,
      "learning_rate": 2.800548448088183e-06,
      "loss": 1.2186,
      "step": 3012
    },
    {
      "epoch": 0.88,
      "learning_rate": 2.7865884783355234e-06,
      "loss": 1.1936,
      "step": 3013
    },
    {
      "epoch": 0.88,
      "learning_rate": 2.77266213311659e-06,
      "loss": 1.1917,
      "step": 3014
    },
    {
      "epoch": 0.88,
      "learning_rate": 2.75876942501466e-06,
      "loss": 1.2397,
      "step": 3015
    },
    {
      "epoch": 0.88,
      "learning_rate": 2.7449103665826336e-06,
      "loss": 1.2389,
      "step": 3016
    },
    {
      "epoch": 0.89,
      "learning_rate": 2.7310849703429788e-06,
      "loss": 1.2235,
      "step": 3017
    },
    {
      "epoch": 0.89,
      "learning_rate": 2.717293248787769e-06,
      "loss": 1.1986,
      "step": 3018
    },
    {
      "epoch": 0.89,
      "learning_rate": 2.7035352143786543e-06,
      "loss": 1.2184,
      "step": 3019
    },
    {
      "epoch": 0.89,
      "learning_rate": 2.6898108795468413e-06,
      "loss": 1.2179,
      "step": 3020
    },
    {
      "epoch": 0.89,
      "learning_rate": 2.676120256693069e-06,
      "loss": 1.2383,
      "step": 3021
    },
    {
      "epoch": 0.89,
      "learning_rate": 2.662463358187637e-06,
      "loss": 1.2561,
      "step": 3022
    },
    {
      "epoch": 0.89,
      "learning_rate": 2.648840196370368e-06,
      "loss": 1.2288,
      "step": 3023
    },
    {
      "epoch": 0.89,
      "learning_rate": 2.635250783550607e-06,
      "loss": 1.2262,
      "step": 3024
    },
    {
      "epoch": 0.89,
      "learning_rate": 2.621695132007185e-06,
      "loss": 1.1919,
      "step": 3025
    },
    {
      "epoch": 0.89,
      "learning_rate": 2.6081732539884465e-06,
      "loss": 1.2442,
      "step": 3026
    },
    {
      "epoch": 0.89,
      "learning_rate": 2.594685161712214e-06,
      "loss": 1.1955,
      "step": 3027
    },
    {
      "epoch": 0.89,
      "learning_rate": 2.5812308673657825e-06,
      "loss": 1.2391,
      "step": 3028
    },
    {
      "epoch": 0.89,
      "learning_rate": 2.5678103831059044e-06,
      "loss": 1.1847,
      "step": 3029
    },
    {
      "epoch": 0.89,
      "learning_rate": 2.5544237210587896e-06,
      "loss": 1.2489,
      "step": 3030
    },
    {
      "epoch": 0.89,
      "learning_rate": 2.5410708933200924e-06,
      "loss": 1.2014,
      "step": 3031
    },
    {
      "epoch": 0.89,
      "learning_rate": 2.527751911954872e-06,
      "loss": 1.2042,
      "step": 3032
    },
    {
      "epoch": 0.89,
      "learning_rate": 2.5144667889976316e-06,
      "loss": 1.2115,
      "step": 3033
    },
    {
      "epoch": 0.89,
      "learning_rate": 2.501215536452266e-06,
      "loss": 1.252,
      "step": 3034
    },
    {
      "epoch": 0.89,
      "learning_rate": 2.487998166292083e-06,
      "loss": 1.2272,
      "step": 3035
    },
    {
      "epoch": 0.89,
      "learning_rate": 2.474814690459746e-06,
      "loss": 1.2268,
      "step": 3036
    },
    {
      "epoch": 0.89,
      "learning_rate": 2.4616651208673183e-06,
      "loss": 1.2386,
      "step": 3037
    },
    {
      "epoch": 0.89,
      "learning_rate": 2.4485494693962197e-06,
      "loss": 1.165,
      "step": 3038
    },
    {
      "epoch": 0.89,
      "learning_rate": 2.4354677478972334e-06,
      "loss": 1.2183,
      "step": 3039
    },
    {
      "epoch": 0.89,
      "learning_rate": 2.422419968190455e-06,
      "loss": 1.2144,
      "step": 3040
    },
    {
      "epoch": 0.89,
      "learning_rate": 2.4094061420653382e-06,
      "loss": 1.2537,
      "step": 3041
    },
    {
      "epoch": 0.89,
      "learning_rate": 2.3964262812806504e-06,
      "loss": 1.166,
      "step": 3042
    },
    {
      "epoch": 0.89,
      "learning_rate": 2.3834803975644772e-06,
      "loss": 1.2067,
      "step": 3043
    },
    {
      "epoch": 0.89,
      "learning_rate": 2.370568502614181e-06,
      "loss": 1.2219,
      "step": 3044
    },
    {
      "epoch": 0.89,
      "learning_rate": 2.3576906080964344e-06,
      "loss": 1.2361,
      "step": 3045
    },
    {
      "epoch": 0.89,
      "learning_rate": 2.344846725647183e-06,
      "loss": 1.2276,
      "step": 3046
    },
    {
      "epoch": 0.89,
      "learning_rate": 2.332036866871645e-06,
      "loss": 1.2302,
      "step": 3047
    },
    {
      "epoch": 0.89,
      "learning_rate": 2.3192610433442784e-06,
      "loss": 1.1819,
      "step": 3048
    },
    {
      "epoch": 0.89,
      "learning_rate": 2.306519266608809e-06,
      "loss": 1.266,
      "step": 3049
    },
    {
      "epoch": 0.89,
      "learning_rate": 2.2938115481781864e-06,
      "loss": 1.1758,
      "step": 3050
    },
    {
      "epoch": 0.9,
      "learning_rate": 2.2811378995346e-06,
      "loss": 1.2106,
      "step": 3051
    },
    {
      "epoch": 0.9,
      "learning_rate": 2.268498332129441e-06,
      "loss": 1.2416,
      "step": 3052
    },
    {
      "epoch": 0.9,
      "learning_rate": 2.2558928573833016e-06,
      "loss": 1.2043,
      "step": 3053
    },
    {
      "epoch": 0.9,
      "learning_rate": 2.2433214866859874e-06,
      "loss": 1.2419,
      "step": 3054
    },
    {
      "epoch": 0.9,
      "learning_rate": 2.2307842313964876e-06,
      "loss": 1.2153,
      "step": 3055
    },
    {
      "epoch": 0.9,
      "learning_rate": 2.2182811028429497e-06,
      "loss": 1.2821,
      "step": 3056
    },
    {
      "epoch": 0.9,
      "learning_rate": 2.2058121123226916e-06,
      "loss": 1.2103,
      "step": 3057
    },
    {
      "epoch": 0.9,
      "learning_rate": 2.1933772711021995e-06,
      "loss": 1.2128,
      "step": 3058
    },
    {
      "epoch": 0.9,
      "learning_rate": 2.1809765904170942e-06,
      "loss": 1.182,
      "step": 3059
    },
    {
      "epoch": 0.9,
      "learning_rate": 2.1686100814721157e-06,
      "loss": 1.254,
      "step": 3060
    },
    {
      "epoch": 0.9,
      "learning_rate": 2.1562777554411474e-06,
      "loss": 1.2182,
      "step": 3061
    },
    {
      "epoch": 0.9,
      "learning_rate": 2.1439796234671874e-06,
      "loss": 1.2066,
      "step": 3062
    },
    {
      "epoch": 0.9,
      "learning_rate": 2.1317156966623243e-06,
      "loss": 1.1914,
      "step": 3063
    },
    {
      "epoch": 0.9,
      "learning_rate": 2.1194859861077477e-06,
      "loss": 1.2006,
      "step": 3064
    },
    {
      "epoch": 0.9,
      "learning_rate": 2.107290502853725e-06,
      "loss": 1.2594,
      "step": 3065
    },
    {
      "epoch": 0.9,
      "learning_rate": 2.0951292579196057e-06,
      "loss": 1.2084,
      "step": 3066
    },
    {
      "epoch": 0.9,
      "learning_rate": 2.083002262293805e-06,
      "loss": 1.2186,
      "step": 3067
    },
    {
      "epoch": 0.9,
      "learning_rate": 2.0709095269337755e-06,
      "loss": 1.2215,
      "step": 3068
    },
    {
      "epoch": 0.9,
      "learning_rate": 2.0588510627660253e-06,
      "loss": 1.214,
      "step": 3069
    },
    {
      "epoch": 0.9,
      "learning_rate": 2.0468268806861015e-06,
      "loss": 1.1584,
      "step": 3070
    },
    {
      "epoch": 0.9,
      "learning_rate": 2.0348369915585666e-06,
      "loss": 1.2639,
      "step": 3071
    },
    {
      "epoch": 0.9,
      "learning_rate": 2.0228814062169944e-06,
      "loss": 1.1341,
      "step": 3072
    },
    {
      "epoch": 0.9,
      "learning_rate": 2.0109601354639706e-06,
      "loss": 1.2081,
      "step": 3073
    },
    {
      "epoch": 0.9,
      "learning_rate": 1.9990731900710745e-06,
      "loss": 1.2038,
      "step": 3074
    },
    {
      "epoch": 0.9,
      "learning_rate": 1.9872205807788747e-06,
      "loss": 1.2376,
      "step": 3075
    },
    {
      "epoch": 0.9,
      "learning_rate": 1.975402318296902e-06,
      "loss": 1.1899,
      "step": 3076
    },
    {
      "epoch": 0.9,
      "learning_rate": 1.9636184133036583e-06,
      "loss": 1.1916,
      "step": 3077
    },
    {
      "epoch": 0.9,
      "learning_rate": 1.9518688764466096e-06,
      "loss": 1.2418,
      "step": 3078
    },
    {
      "epoch": 0.9,
      "learning_rate": 1.940153718342166e-06,
      "loss": 1.2071,
      "step": 3079
    },
    {
      "epoch": 0.9,
      "learning_rate": 1.9284729495756595e-06,
      "loss": 1.2302,
      "step": 3080
    },
    {
      "epoch": 0.9,
      "learning_rate": 1.916826580701363e-06,
      "loss": 1.2173,
      "step": 3081
    },
    {
      "epoch": 0.9,
      "learning_rate": 1.9052146222424772e-06,
      "loss": 1.2947,
      "step": 3082
    },
    {
      "epoch": 0.9,
      "learning_rate": 1.8936370846910845e-06,
      "loss": 1.2064,
      "step": 3083
    },
    {
      "epoch": 0.9,
      "learning_rate": 1.8820939785081771e-06,
      "loss": 1.2323,
      "step": 3084
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.8705853141236518e-06,
      "loss": 1.2466,
      "step": 3085
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.8591111019362662e-06,
      "loss": 1.2662,
      "step": 3086
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.8476713523136514e-06,
      "loss": 1.2127,
      "step": 3087
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.8362660755923079e-06,
      "loss": 1.2152,
      "step": 3088
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.824895282077579e-06,
      "loss": 1.1888,
      "step": 3089
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.813558982043655e-06,
      "loss": 1.2449,
      "step": 3090
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.8022571857335603e-06,
      "loss": 1.1833,
      "step": 3091
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.790989903359135e-06,
      "loss": 1.1922,
      "step": 3092
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.7797571451010442e-06,
      "loss": 1.2021,
      "step": 3093
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.7685589211087517e-06,
      "loss": 1.2674,
      "step": 3094
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.7573952415005235e-06,
      "loss": 1.1923,
      "step": 3095
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.7462661163634064e-06,
      "loss": 1.1775,
      "step": 3096
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.7351715557532234e-06,
      "loss": 1.2562,
      "step": 3097
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.7241115696945821e-06,
      "loss": 1.2338,
      "step": 3098
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.713086168180831e-06,
      "loss": 1.2058,
      "step": 3099
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.7020953611740765e-06,
      "loss": 1.2243,
      "step": 3100
    },
    {
      "epoch": 0.91,
      "eval_loss": 0.11909307539463043,
      "eval_runtime": 25.2086,
      "eval_samples_per_second": 79.338,
      "eval_steps_per_second": 0.635,
      "step": 3100
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.6911391586051707e-06,
      "loss": 1.2174,
      "step": 3101
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.6802175703736923e-06,
      "loss": 1.1877,
      "step": 3102
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.669330606347952e-06,
      "loss": 1.159,
      "step": 3103
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.658478276364961e-06,
      "loss": 1.2408,
      "step": 3104
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.6476605902304532e-06,
      "loss": 1.2698,
      "step": 3105
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.6368775577188546e-06,
      "loss": 1.2128,
      "step": 3106
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.6261291885732645e-06,
      "loss": 1.2157,
      "step": 3107
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.6154154925054878e-06,
      "loss": 1.2102,
      "step": 3108
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.6047364791959763e-06,
      "loss": 1.2071,
      "step": 3109
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.5940921582938694e-06,
      "loss": 1.1873,
      "step": 3110
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.5834825394169317e-06,
      "loss": 1.2295,
      "step": 3111
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.572907632151588e-06,
      "loss": 1.2481,
      "step": 3112
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.5623674460529014e-06,
      "loss": 1.2163,
      "step": 3113
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.5518619906445564e-06,
      "loss": 1.2,
      "step": 3114
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.5413912754188531e-06,
      "loss": 1.2084,
      "step": 3115
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.5309553098367125e-06,
      "loss": 1.2058,
      "step": 3116
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.520554103327645e-06,
      "loss": 1.2264,
      "step": 3117
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.510187665289773e-06,
      "loss": 1.2161,
      "step": 3118
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.4998560050897726e-06,
      "loss": 1.2033,
      "step": 3119
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.4895591320629232e-06,
      "loss": 1.2176,
      "step": 3120
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.4792970555130627e-06,
      "loss": 1.2372,
      "step": 3121
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.4690697847125956e-06,
      "loss": 1.1984,
      "step": 3122
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.458877328902455e-06,
      "loss": 1.1808,
      "step": 3123
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.4487196972921402e-06,
      "loss": 1.2559,
      "step": 3124
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.4385968990596743e-06,
      "loss": 1.1869,
      "step": 3125
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.4285089433516164e-06,
      "loss": 1.2198,
      "step": 3126
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.4184558392830306e-06,
      "loss": 1.2406,
      "step": 3127
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.4084375959374864e-06,
      "loss": 1.2032,
      "step": 3128
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.3984542223670717e-06,
      "loss": 1.2094,
      "step": 3129
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.388505727592362e-06,
      "loss": 1.2095,
      "step": 3130
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.3785921206024022e-06,
      "loss": 1.262,
      "step": 3131
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.3687134103547383e-06,
      "loss": 1.2094,
      "step": 3132
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.3588696057753725e-06,
      "loss": 1.2179,
      "step": 3133
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.3490607157587677e-06,
      "loss": 1.1915,
      "step": 3134
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.339286749167834e-06,
      "loss": 1.286,
      "step": 3135
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.3295477148339387e-06,
      "loss": 1.211,
      "step": 3136
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.3198436215568778e-06,
      "loss": 1.2351,
      "step": 3137
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.3101744781048774e-06,
      "loss": 1.2168,
      "step": 3138
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.300540293214585e-06,
      "loss": 1.2357,
      "step": 3139
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.290941075591059e-06,
      "loss": 1.215,
      "step": 3140
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.281376833907766e-06,
      "loss": 1.185,
      "step": 3141
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.2718475768065709e-06,
      "loss": 1.265,
      "step": 3142
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.262353312897715e-06,
      "loss": 1.1601,
      "step": 3143
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.2528940507598387e-06,
      "loss": 1.1799,
      "step": 3144
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.243469798939949e-06,
      "loss": 1.1758,
      "step": 3145
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.2340805659534216e-06,
      "loss": 1.232,
      "step": 3146
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.224726360283981e-06,
      "loss": 1.2176,
      "step": 3147
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.2154071903837107e-06,
      "loss": 1.2218,
      "step": 3148
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.2061230646730348e-06,
      "loss": 1.197,
      "step": 3149
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.196873991540719e-06,
      "loss": 1.2303,
      "step": 3150
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.187659979343847e-06,
      "loss": 1.2106,
      "step": 3151
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.1784810364078303e-06,
      "loss": 1.1829,
      "step": 3152
    },
    {
      "epoch": 0.93,
      "learning_rate": 1.1693371710263857e-06,
      "loss": 1.1985,
      "step": 3153
    },
    {
      "epoch": 0.93,
      "learning_rate": 1.1602283914615486e-06,
      "loss": 1.2391,
      "step": 3154
    },
    {
      "epoch": 0.93,
      "learning_rate": 1.151154705943638e-06,
      "loss": 1.2085,
      "step": 3155
    },
    {
      "epoch": 0.93,
      "learning_rate": 1.1421161226712684e-06,
      "loss": 1.1942,
      "step": 3156
    },
    {
      "epoch": 0.93,
      "learning_rate": 1.1331126498113433e-06,
      "loss": 1.261,
      "step": 3157
    },
    {
      "epoch": 0.93,
      "learning_rate": 1.1241442954990301e-06,
      "loss": 1.1779,
      "step": 3158
    },
    {
      "epoch": 0.93,
      "learning_rate": 1.1152110678377759e-06,
      "loss": 1.2233,
      "step": 3159
    },
    {
      "epoch": 0.93,
      "learning_rate": 1.1063129748992796e-06,
      "loss": 1.2103,
      "step": 3160
    },
    {
      "epoch": 0.93,
      "learning_rate": 1.0974500247235053e-06,
      "loss": 1.2403,
      "step": 3161
    },
    {
      "epoch": 0.93,
      "learning_rate": 1.0886222253186473e-06,
      "loss": 1.2104,
      "step": 3162
    },
    {
      "epoch": 0.93,
      "learning_rate": 1.0798295846611562e-06,
      "loss": 1.1962,
      "step": 3163
    },
    {
      "epoch": 0.93,
      "learning_rate": 1.0710721106957034e-06,
      "loss": 1.2059,
      "step": 3164
    },
    {
      "epoch": 0.93,
      "learning_rate": 1.062349811335186e-06,
      "loss": 1.2184,
      "step": 3165
    },
    {
      "epoch": 0.93,
      "learning_rate": 1.0536626944607264e-06,
      "loss": 1.2092,
      "step": 3166
    },
    {
      "epoch": 0.93,
      "learning_rate": 1.045010767921646e-06,
      "loss": 1.1916,
      "step": 3167
    },
    {
      "epoch": 0.93,
      "learning_rate": 1.0363940395354777e-06,
      "loss": 1.2092,
      "step": 3168
    },
    {
      "epoch": 0.93,
      "learning_rate": 1.0278125170879495e-06,
      "loss": 1.233,
      "step": 3169
    },
    {
      "epoch": 0.93,
      "learning_rate": 1.019266208332983e-06,
      "loss": 1.2049,
      "step": 3170
    },
    {
      "epoch": 0.93,
      "learning_rate": 1.0107551209926635e-06,
      "loss": 1.2276,
      "step": 3171
    },
    {
      "epoch": 0.93,
      "learning_rate": 1.0022792627572797e-06,
      "loss": 1.2149,
      "step": 3172
    },
    {
      "epoch": 0.93,
      "learning_rate": 9.938386412852652e-07,
      "loss": 1.2075,
      "step": 3173
    },
    {
      "epoch": 0.93,
      "learning_rate": 9.854332642032348e-07,
      "loss": 1.1898,
      "step": 3174
    },
    {
      "epoch": 0.93,
      "learning_rate": 9.770631391059316e-07,
      "loss": 1.2432,
      "step": 3175
    },
    {
      "epoch": 0.93,
      "learning_rate": 9.687282735562742e-07,
      "loss": 1.2802,
      "step": 3176
    },
    {
      "epoch": 0.93,
      "learning_rate": 9.604286750853054e-07,
      "loss": 1.1746,
      "step": 3177
    },
    {
      "epoch": 0.93,
      "learning_rate": 9.52164351192213e-07,
      "loss": 1.2301,
      "step": 3178
    },
    {
      "epoch": 0.93,
      "learning_rate": 9.439353093442949e-07,
      "loss": 1.1661,
      "step": 3179
    },
    {
      "epoch": 0.93,
      "learning_rate": 9.357415569769901e-07,
      "loss": 1.2088,
      "step": 3180
    },
    {
      "epoch": 0.93,
      "learning_rate": 9.275831014938385e-07,
      "loss": 1.184,
      "step": 3181
    },
    {
      "epoch": 0.93,
      "learning_rate": 9.194599502664858e-07,
      "loss": 1.23,
      "step": 3182
    },
    {
      "epoch": 0.93,
      "learning_rate": 9.113721106346918e-07,
      "loss": 1.1952,
      "step": 3183
    },
    {
      "epoch": 0.93,
      "learning_rate": 9.033195899062907e-07,
      "loss": 1.2606,
      "step": 3184
    },
    {
      "epoch": 0.93,
      "learning_rate": 8.953023953572315e-07,
      "loss": 1.2358,
      "step": 3185
    },
    {
      "epoch": 0.93,
      "learning_rate": 8.873205342315061e-07,
      "loss": 1.2365,
      "step": 3186
    },
    {
      "epoch": 0.94,
      "learning_rate": 8.793740137412166e-07,
      "loss": 1.1951,
      "step": 3187
    },
    {
      "epoch": 0.94,
      "learning_rate": 8.714628410665082e-07,
      "loss": 1.2474,
      "step": 3188
    },
    {
      "epoch": 0.94,
      "learning_rate": 8.635870233556054e-07,
      "loss": 1.1712,
      "step": 3189
    },
    {
      "epoch": 0.94,
      "learning_rate": 8.557465677247756e-07,
      "loss": 1.2063,
      "step": 3190
    },
    {
      "epoch": 0.94,
      "learning_rate": 8.479414812583298e-07,
      "loss": 1.2672,
      "step": 3191
    },
    {
      "epoch": 0.94,
      "learning_rate": 8.401717710086354e-07,
      "loss": 1.2227,
      "step": 3192
    },
    {
      "epoch": 0.94,
      "learning_rate": 8.324374439960947e-07,
      "loss": 1.1378,
      "step": 3193
    },
    {
      "epoch": 0.94,
      "learning_rate": 8.247385072091219e-07,
      "loss": 1.2223,
      "step": 3194
    },
    {
      "epoch": 0.94,
      "learning_rate": 8.170749676041656e-07,
      "loss": 1.2112,
      "step": 3195
    },
    {
      "epoch": 0.94,
      "learning_rate": 8.094468321056958e-07,
      "loss": 1.1939,
      "step": 3196
    },
    {
      "epoch": 0.94,
      "learning_rate": 8.018541076061859e-07,
      "loss": 1.1832,
      "step": 3197
    },
    {
      "epoch": 0.94,
      "learning_rate": 7.942968009661123e-07,
      "loss": 1.2064,
      "step": 3198
    },
    {
      "epoch": 0.94,
      "learning_rate": 7.867749190139506e-07,
      "loss": 1.2711,
      "step": 3199
    },
    {
      "epoch": 0.94,
      "learning_rate": 7.792884685461754e-07,
      "loss": 1.2301,
      "step": 3200
    },
    {
      "epoch": 0.94,
      "eval_loss": 0.11856154352426529,
      "eval_runtime": 24.8968,
      "eval_samples_per_second": 80.332,
      "eval_steps_per_second": 0.643,
      "step": 3200
    },
    {
      "epoch": 0.94,
      "learning_rate": 7.718374563272379e-07,
      "loss": 1.2119,
      "step": 3201
    },
    {
      "epoch": 0.94,
      "learning_rate": 7.644218890895705e-07,
      "loss": 1.2066,
      "step": 3202
    },
    {
      "epoch": 0.94,
      "learning_rate": 7.570417735335733e-07,
      "loss": 1.2263,
      "step": 3203
    },
    {
      "epoch": 0.94,
      "learning_rate": 7.496971163276324e-07,
      "loss": 1.2003,
      "step": 3204
    },
    {
      "epoch": 0.94,
      "learning_rate": 7.423879241080745e-07,
      "loss": 1.2088,
      "step": 3205
    },
    {
      "epoch": 0.94,
      "learning_rate": 7.351142034791947e-07,
      "loss": 1.2077,
      "step": 3206
    },
    {
      "epoch": 0.94,
      "learning_rate": 7.278759610132291e-07,
      "loss": 1.2038,
      "step": 3207
    },
    {
      "epoch": 0.94,
      "learning_rate": 7.206732032503638e-07,
      "loss": 1.1972,
      "step": 3208
    },
    {
      "epoch": 0.94,
      "learning_rate": 7.135059366987173e-07,
      "loss": 1.1978,
      "step": 3209
    },
    {
      "epoch": 0.94,
      "learning_rate": 7.063741678343449e-07,
      "loss": 1.3071,
      "step": 3210
    },
    {
      "epoch": 0.94,
      "learning_rate": 6.992779031012209e-07,
      "loss": 1.2284,
      "step": 3211
    },
    {
      "epoch": 0.94,
      "learning_rate": 6.922171489112428e-07,
      "loss": 1.2087,
      "step": 3212
    },
    {
      "epoch": 0.94,
      "learning_rate": 6.851919116442274e-07,
      "loss": 1.2022,
      "step": 3213
    },
    {
      "epoch": 0.94,
      "learning_rate": 6.782021976478925e-07,
      "loss": 1.2347,
      "step": 3214
    },
    {
      "epoch": 0.94,
      "learning_rate": 6.712480132378574e-07,
      "loss": 1.2221,
      "step": 3215
    },
    {
      "epoch": 0.94,
      "learning_rate": 6.643293646976468e-07,
      "loss": 1.1756,
      "step": 3216
    },
    {
      "epoch": 0.94,
      "learning_rate": 6.574462582786733e-07,
      "loss": 1.2267,
      "step": 3217
    },
    {
      "epoch": 0.94,
      "learning_rate": 6.505987002002245e-07,
      "loss": 1.2208,
      "step": 3218
    },
    {
      "epoch": 0.94,
      "learning_rate": 6.437866966494843e-07,
      "loss": 1.2152,
      "step": 3219
    },
    {
      "epoch": 0.94,
      "learning_rate": 6.370102537815026e-07,
      "loss": 1.1821,
      "step": 3220
    },
    {
      "epoch": 0.94,
      "learning_rate": 6.302693777191993e-07,
      "loss": 1.2259,
      "step": 3221
    },
    {
      "epoch": 0.95,
      "learning_rate": 6.235640745533556e-07,
      "loss": 1.2004,
      "step": 3222
    },
    {
      "epoch": 0.95,
      "learning_rate": 6.168943503426139e-07,
      "loss": 1.2232,
      "step": 3223
    },
    {
      "epoch": 0.95,
      "learning_rate": 6.102602111134648e-07,
      "loss": 1.2077,
      "step": 3224
    },
    {
      "epoch": 0.95,
      "learning_rate": 6.036616628602554e-07,
      "loss": 1.2323,
      "step": 3225
    },
    {
      "epoch": 0.95,
      "learning_rate": 5.970987115451587e-07,
      "loss": 1.2547,
      "step": 3226
    },
    {
      "epoch": 0.95,
      "learning_rate": 5.905713630981913e-07,
      "loss": 1.2138,
      "step": 3227
    },
    {
      "epoch": 0.95,
      "learning_rate": 5.840796234172085e-07,
      "loss": 1.1924,
      "step": 3228
    },
    {
      "epoch": 0.95,
      "learning_rate": 5.77623498367883e-07,
      "loss": 1.2618,
      "step": 3229
    },
    {
      "epoch": 0.95,
      "learning_rate": 5.712029937837038e-07,
      "loss": 1.2606,
      "step": 3230
    },
    {
      "epoch": 0.95,
      "learning_rate": 5.648181154659771e-07,
      "loss": 1.1964,
      "step": 3231
    },
    {
      "epoch": 0.95,
      "learning_rate": 5.584688691838258e-07,
      "loss": 1.1864,
      "step": 3232
    },
    {
      "epoch": 0.95,
      "learning_rate": 5.521552606741765e-07,
      "loss": 1.2483,
      "step": 3233
    },
    {
      "epoch": 0.95,
      "learning_rate": 5.458772956417413e-07,
      "loss": 1.2191,
      "step": 3234
    },
    {
      "epoch": 0.95,
      "learning_rate": 5.396349797590405e-07,
      "loss": 1.2066,
      "step": 3235
    },
    {
      "epoch": 0.95,
      "learning_rate": 5.334283186663758e-07,
      "loss": 1.2256,
      "step": 3236
    },
    {
      "epoch": 0.95,
      "learning_rate": 5.272573179718388e-07,
      "loss": 1.1937,
      "step": 3237
    },
    {
      "epoch": 0.95,
      "learning_rate": 5.211219832512893e-07,
      "loss": 1.1856,
      "step": 3238
    },
    {
      "epoch": 0.95,
      "learning_rate": 5.150223200483772e-07,
      "loss": 1.2324,
      "step": 3239
    },
    {
      "epoch": 0.95,
      "learning_rate": 5.08958333874503e-07,
      "loss": 1.1985,
      "step": 3240
    },
    {
      "epoch": 0.95,
      "learning_rate": 5.029300302088436e-07,
      "loss": 1.1575,
      "step": 3241
    },
    {
      "epoch": 0.95,
      "learning_rate": 4.969374144983308e-07,
      "loss": 1.2206,
      "step": 3242
    },
    {
      "epoch": 0.95,
      "learning_rate": 4.909804921576466e-07,
      "loss": 1.2243,
      "step": 3243
    },
    {
      "epoch": 0.95,
      "learning_rate": 4.850592685692279e-07,
      "loss": 1.2707,
      "step": 3244
    },
    {
      "epoch": 0.95,
      "learning_rate": 4.791737490832526e-07,
      "loss": 1.2191,
      "step": 3245
    },
    {
      "epoch": 0.95,
      "learning_rate": 4.7332393901763586e-07,
      "loss": 1.2254,
      "step": 3246
    },
    {
      "epoch": 0.95,
      "learning_rate": 4.6750984365803385e-07,
      "loss": 1.2153,
      "step": 3247
    },
    {
      "epoch": 0.95,
      "learning_rate": 4.6173146825782224e-07,
      "loss": 1.2357,
      "step": 3248
    },
    {
      "epoch": 0.95,
      "learning_rate": 4.5598881803811336e-07,
      "loss": 1.2064,
      "step": 3249
    },
    {
      "epoch": 0.95,
      "learning_rate": 4.5028189818772996e-07,
      "loss": 1.1862,
      "step": 3250
    },
    {
      "epoch": 0.95,
      "learning_rate": 4.4461071386321387e-07,
      "loss": 1.26,
      "step": 3251
    },
    {
      "epoch": 0.95,
      "learning_rate": 4.389752701888173e-07,
      "loss": 1.2199,
      "step": 3252
    },
    {
      "epoch": 0.95,
      "learning_rate": 4.3337557225650695e-07,
      "loss": 1.1857,
      "step": 3253
    },
    {
      "epoch": 0.95,
      "learning_rate": 4.27811625125929e-07,
      "loss": 1.1982,
      "step": 3254
    },
    {
      "epoch": 0.95,
      "learning_rate": 4.2228343382445746e-07,
      "loss": 1.2319,
      "step": 3255
    },
    {
      "epoch": 0.96,
      "learning_rate": 4.1679100334712787e-07,
      "loss": 1.2478,
      "step": 3256
    },
    {
      "epoch": 0.96,
      "learning_rate": 4.113343386566904e-07,
      "loss": 1.2289,
      "step": 3257
    },
    {
      "epoch": 0.96,
      "learning_rate": 4.0591344468355666e-07,
      "loss": 1.2184,
      "step": 3258
    },
    {
      "epoch": 0.96,
      "learning_rate": 4.005283263258353e-07,
      "loss": 1.2183,
      "step": 3259
    },
    {
      "epoch": 0.96,
      "learning_rate": 3.9517898844929624e-07,
      "loss": 1.2018,
      "step": 3260
    },
    {
      "epoch": 0.96,
      "learning_rate": 3.8986543588738437e-07,
      "loss": 1.2008,
      "step": 3261
    },
    {
      "epoch": 0.96,
      "learning_rate": 3.845876734412146e-07,
      "loss": 1.1776,
      "step": 3262
    },
    {
      "epoch": 0.96,
      "learning_rate": 3.793457058795591e-07,
      "loss": 1.2697,
      "step": 3263
    },
    {
      "epoch": 0.96,
      "learning_rate": 3.741395379388468e-07,
      "loss": 1.1833,
      "step": 3264
    },
    {
      "epoch": 0.96,
      "learning_rate": 3.689691743231594e-07,
      "loss": 1.1906,
      "step": 3265
    },
    {
      "epoch": 0.96,
      "learning_rate": 3.638346197042264e-07,
      "loss": 1.1974,
      "step": 3266
    },
    {
      "epoch": 0.96,
      "learning_rate": 3.587358787214301e-07,
      "loss": 1.2361,
      "step": 3267
    },
    {
      "epoch": 0.96,
      "learning_rate": 3.5367295598178307e-07,
      "loss": 1.2688,
      "step": 3268
    },
    {
      "epoch": 0.96,
      "learning_rate": 3.4864585605993705e-07,
      "loss": 1.1855,
      "step": 3269
    },
    {
      "epoch": 0.96,
      "learning_rate": 3.4365458349817415e-07,
      "loss": 1.2575,
      "step": 3270
    },
    {
      "epoch": 0.96,
      "learning_rate": 3.3869914280641125e-07,
      "loss": 1.1622,
      "step": 3271
    },
    {
      "epoch": 0.96,
      "learning_rate": 3.337795384621867e-07,
      "loss": 1.1752,
      "step": 3272
    },
    {
      "epoch": 0.96,
      "learning_rate": 3.2889577491064693e-07,
      "loss": 1.2039,
      "step": 3273
    },
    {
      "epoch": 0.96,
      "learning_rate": 3.240478565645644e-07,
      "loss": 1.2238,
      "step": 3274
    },
    {
      "epoch": 0.96,
      "learning_rate": 3.192357878043284e-07,
      "loss": 1.181,
      "step": 3275
    },
    {
      "epoch": 0.96,
      "learning_rate": 3.144595729779276e-07,
      "loss": 1.2328,
      "step": 3276
    },
    {
      "epoch": 0.96,
      "learning_rate": 3.0971921640095883e-07,
      "loss": 1.207,
      "step": 3277
    },
    {
      "epoch": 0.96,
      "learning_rate": 3.050147223566091e-07,
      "loss": 1.2321,
      "step": 3278
    },
    {
      "epoch": 0.96,
      "learning_rate": 3.003460950956694e-07,
      "loss": 1.1912,
      "step": 3279
    },
    {
      "epoch": 0.96,
      "learning_rate": 2.9571333883652966e-07,
      "loss": 1.2046,
      "step": 3280
    },
    {
      "epoch": 0.96,
      "learning_rate": 2.911164577651615e-07,
      "loss": 1.261,
      "step": 3281
    },
    {
      "epoch": 0.96,
      "learning_rate": 2.8655545603511357e-07,
      "loss": 1.1855,
      "step": 3282
    },
    {
      "epoch": 0.96,
      "learning_rate": 2.8203033776752487e-07,
      "loss": 1.2228,
      "step": 3283
    },
    {
      "epoch": 0.96,
      "learning_rate": 2.7754110705111137e-07,
      "loss": 1.2589,
      "step": 3284
    },
    {
      "epoch": 0.96,
      "learning_rate": 2.730877679421573e-07,
      "loss": 1.2224,
      "step": 3285
    },
    {
      "epoch": 0.96,
      "learning_rate": 2.686703244645239e-07,
      "loss": 1.1829,
      "step": 3286
    },
    {
      "epoch": 0.96,
      "learning_rate": 2.642887806096317e-07,
      "loss": 1.2241,
      "step": 3287
    },
    {
      "epoch": 0.96,
      "learning_rate": 2.599431403364694e-07,
      "loss": 1.2519,
      "step": 3288
    },
    {
      "epoch": 0.96,
      "learning_rate": 2.556334075715805e-07,
      "loss": 1.2351,
      "step": 3289
    },
    {
      "epoch": 0.97,
      "learning_rate": 2.51359586209059e-07,
      "loss": 1.2314,
      "step": 3290
    },
    {
      "epoch": 0.97,
      "learning_rate": 2.4712168011056693e-07,
      "loss": 1.1899,
      "step": 3291
    },
    {
      "epoch": 0.97,
      "learning_rate": 2.4291969310529907e-07,
      "loss": 1.1722,
      "step": 3292
    },
    {
      "epoch": 0.97,
      "learning_rate": 2.3875362899000054e-07,
      "loss": 1.2295,
      "step": 3293
    },
    {
      "epoch": 0.97,
      "learning_rate": 2.3462349152896247e-07,
      "loss": 1.197,
      "step": 3294
    },
    {
      "epoch": 0.97,
      "learning_rate": 2.3052928445400858e-07,
      "loss": 1.1559,
      "step": 3295
    },
    {
      "epoch": 0.97,
      "learning_rate": 2.2647101146449524e-07,
      "loss": 1.2152,
      "step": 3296
    },
    {
      "epoch": 0.97,
      "learning_rate": 2.2244867622731593e-07,
      "loss": 1.2135,
      "step": 3297
    },
    {
      "epoch": 0.97,
      "learning_rate": 2.1846228237689226e-07,
      "loss": 1.2398,
      "step": 3298
    },
    {
      "epoch": 0.97,
      "learning_rate": 2.1451183351516969e-07,
      "loss": 1.2219,
      "step": 3299
    },
    {
      "epoch": 0.97,
      "learning_rate": 2.1059733321160848e-07,
      "loss": 1.2838,
      "step": 3300
    },
    {
      "epoch": 0.97,
      "eval_loss": 0.11866208910942078,
      "eval_runtime": 24.6429,
      "eval_samples_per_second": 81.159,
      "eval_steps_per_second": 0.649,
      "step": 3300
    },
    {
      "epoch": 0.97,
      "learning_rate": 2.0671878500319707e-07,
      "loss": 1.2131,
      "step": 3301
    },
    {
      "epoch": 0.97,
      "learning_rate": 2.0287619239443003e-07,
      "loss": 1.2107,
      "step": 3302
    },
    {
      "epoch": 0.97,
      "learning_rate": 1.9906955885732104e-07,
      "loss": 1.2182,
      "step": 3303
    },
    {
      "epoch": 0.97,
      "learning_rate": 1.9529888783139883e-07,
      "loss": 1.2384,
      "step": 3304
    },
    {
      "epoch": 0.97,
      "learning_rate": 1.9156418272367584e-07,
      "loss": 1.248,
      "step": 3305
    },
    {
      "epoch": 0.97,
      "learning_rate": 1.878654469086838e-07,
      "loss": 1.1917,
      "step": 3306
    },
    {
      "epoch": 0.97,
      "learning_rate": 1.8420268372845608e-07,
      "loss": 1.1746,
      "step": 3307
    },
    {
      "epoch": 0.97,
      "learning_rate": 1.8057589649251862e-07,
      "loss": 1.2346,
      "step": 3308
    },
    {
      "epoch": 0.97,
      "learning_rate": 1.7698508847788121e-07,
      "loss": 1.2001,
      "step": 3309
    },
    {
      "epoch": 0.97,
      "learning_rate": 1.7343026292905518e-07,
      "loss": 1.2314,
      "step": 3310
    },
    {
      "epoch": 0.97,
      "learning_rate": 1.6991142305804454e-07,
      "loss": 1.1919,
      "step": 3311
    },
    {
      "epoch": 0.97,
      "learning_rate": 1.6642857204431928e-07,
      "loss": 1.1796,
      "step": 3312
    },
    {
      "epoch": 0.97,
      "learning_rate": 1.62981713034851e-07,
      "loss": 1.228,
      "step": 3313
    },
    {
      "epoch": 0.97,
      "learning_rate": 1.595708491440817e-07,
      "loss": 1.1774,
      "step": 3314
    },
    {
      "epoch": 0.97,
      "learning_rate": 1.561959834539284e-07,
      "loss": 1.2435,
      "step": 3315
    },
    {
      "epoch": 0.97,
      "learning_rate": 1.5285711901378285e-07,
      "loss": 1.1853,
      "step": 3316
    },
    {
      "epoch": 0.97,
      "learning_rate": 1.4955425884050746e-07,
      "loss": 1.2228,
      "step": 3317
    },
    {
      "epoch": 0.97,
      "learning_rate": 1.462874059184305e-07,
      "loss": 1.2004,
      "step": 3318
    },
    {
      "epoch": 0.97,
      "learning_rate": 1.4305656319935522e-07,
      "loss": 1.2149,
      "step": 3319
    },
    {
      "epoch": 0.97,
      "learning_rate": 1.3986173360253762e-07,
      "loss": 1.2187,
      "step": 3320
    },
    {
      "epoch": 0.97,
      "learning_rate": 1.367029200146952e-07,
      "loss": 1.2098,
      "step": 3321
    },
    {
      "epoch": 0.97,
      "learning_rate": 1.3358012529000263e-07,
      "loss": 1.2093,
      "step": 3322
    },
    {
      "epoch": 0.97,
      "learning_rate": 1.3049335225009175e-07,
      "loss": 1.2551,
      "step": 3323
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.2744260368404703e-07,
      "loss": 1.217,
      "step": 3324
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.2442788234840575e-07,
      "loss": 1.2462,
      "step": 3325
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.2144919096714004e-07,
      "loss": 1.2635,
      "step": 3326
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.1850653223167918e-07,
      "loss": 1.1902,
      "step": 3327
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.1559990880089189e-07,
      "loss": 1.1875,
      "step": 3328
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.1272932330108178e-07,
      "loss": 1.2146,
      "step": 3329
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.0989477832599627e-07,
      "loss": 1.2177,
      "step": 3330
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.070962764368133e-07,
      "loss": 1.2562,
      "step": 3331
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.0433382016215466e-07,
      "loss": 1.2039,
      "step": 3332
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.0160741199805036e-07,
      "loss": 1.2125,
      "step": 3333
    },
    {
      "epoch": 0.98,
      "learning_rate": 9.891705440798316e-08,
      "loss": 1.221,
      "step": 3334
    },
    {
      "epoch": 0.98,
      "learning_rate": 9.626274982284855e-08,
      "loss": 1.205,
      "step": 3335
    },
    {
      "epoch": 0.98,
      "learning_rate": 9.364450064096364e-08,
      "loss": 1.197,
      "step": 3336
    },
    {
      "epoch": 0.98,
      "learning_rate": 9.106230922808046e-08,
      "loss": 1.212,
      "step": 3337
    },
    {
      "epoch": 0.98,
      "learning_rate": 8.851617791735933e-08,
      "loss": 1.2638,
      "step": 3338
    },
    {
      "epoch": 0.98,
      "learning_rate": 8.600610900938222e-08,
      "loss": 1.2388,
      "step": 3339
    },
    {
      "epoch": 0.98,
      "learning_rate": 8.353210477214379e-08,
      "loss": 1.2256,
      "step": 3340
    },
    {
      "epoch": 0.98,
      "learning_rate": 8.109416744105148e-08,
      "loss": 1.2323,
      "step": 3341
    },
    {
      "epoch": 0.98,
      "learning_rate": 7.869229921893429e-08,
      "loss": 1.2112,
      "step": 3342
    },
    {
      "epoch": 0.98,
      "learning_rate": 7.632650227602511e-08,
      "loss": 1.1989,
      "step": 3343
    },
    {
      "epoch": 0.98,
      "learning_rate": 7.399677874995181e-08,
      "loss": 1.2037,
      "step": 3344
    },
    {
      "epoch": 0.98,
      "learning_rate": 7.170313074576384e-08,
      "loss": 1.2638,
      "step": 3345
    },
    {
      "epoch": 0.98,
      "learning_rate": 6.944556033590566e-08,
      "loss": 1.2455,
      "step": 3346
    },
    {
      "epoch": 0.98,
      "learning_rate": 6.722406956023441e-08,
      "loss": 1.1469,
      "step": 3347
    },
    {
      "epoch": 0.98,
      "learning_rate": 6.503866042599338e-08,
      "loss": 1.1828,
      "step": 3348
    },
    {
      "epoch": 0.98,
      "learning_rate": 6.288933490782967e-08,
      "loss": 1.2341,
      "step": 3349
    },
    {
      "epoch": 0.98,
      "learning_rate": 6.07760949477898e-08,
      "loss": 1.2226,
      "step": 3350
    },
    {
      "epoch": 0.98,
      "learning_rate": 5.8698942455306384e-08,
      "loss": 1.2107,
      "step": 3351
    },
    {
      "epoch": 0.98,
      "learning_rate": 5.6657879307220334e-08,
      "loss": 1.1971,
      "step": 3352
    },
    {
      "epoch": 0.98,
      "learning_rate": 5.4652907347745307e-08,
      "loss": 1.2389,
      "step": 3353
    },
    {
      "epoch": 0.98,
      "learning_rate": 5.26840283884944e-08,
      "loss": 1.2183,
      "step": 3354
    },
    {
      "epoch": 0.98,
      "learning_rate": 5.075124420846678e-08,
      "loss": 1.176,
      "step": 3355
    },
    {
      "epoch": 0.98,
      "learning_rate": 4.885455655405213e-08,
      "loss": 1.224,
      "step": 3356
    },
    {
      "epoch": 0.98,
      "learning_rate": 4.699396713901294e-08,
      "loss": 1.1713,
      "step": 3357
    },
    {
      "epoch": 0.99,
      "learning_rate": 4.516947764451107e-08,
      "loss": 1.1988,
      "step": 3358
    },
    {
      "epoch": 0.99,
      "learning_rate": 4.338108971908117e-08,
      "loss": 1.2227,
      "step": 3359
    },
    {
      "epoch": 0.99,
      "learning_rate": 4.1628804978630645e-08,
      "loss": 1.2372,
      "step": 3360
    },
    {
      "epoch": 0.99,
      "learning_rate": 3.9912625006457426e-08,
      "loss": 1.1985,
      "step": 3361
    },
    {
      "epoch": 0.99,
      "learning_rate": 3.823255135323223e-08,
      "loss": 1.2346,
      "step": 3362
    },
    {
      "epoch": 0.99,
      "learning_rate": 3.6588585537002955e-08,
      "loss": 1.2013,
      "step": 3363
    },
    {
      "epoch": 0.99,
      "learning_rate": 3.498072904319027e-08,
      "loss": 1.2219,
      "step": 3364
    },
    {
      "epoch": 0.99,
      "learning_rate": 3.34089833245832e-08,
      "loss": 1.2462,
      "step": 3365
    },
    {
      "epoch": 0.99,
      "learning_rate": 3.1873349801356814e-08,
      "loss": 1.1915,
      "step": 3366
    },
    {
      "epoch": 0.99,
      "learning_rate": 3.0373829861041204e-08,
      "loss": 1.1813,
      "step": 3367
    },
    {
      "epoch": 0.99,
      "learning_rate": 2.8910424858543673e-08,
      "loss": 1.264,
      "step": 3368
    },
    {
      "epoch": 0.99,
      "learning_rate": 2.74831361161354e-08,
      "loss": 1.1941,
      "step": 3369
    },
    {
      "epoch": 0.99,
      "learning_rate": 2.6091964923455894e-08,
      "loss": 1.2501,
      "step": 3370
    },
    {
      "epoch": 0.99,
      "learning_rate": 2.473691253752186e-08,
      "loss": 1.2374,
      "step": 3371
    },
    {
      "epoch": 0.99,
      "learning_rate": 2.341798018268726e-08,
      "loss": 1.198,
      "step": 3372
    },
    {
      "epoch": 0.99,
      "learning_rate": 2.213516905070101e-08,
      "loss": 1.2375,
      "step": 3373
    },
    {
      "epoch": 0.99,
      "learning_rate": 2.088848030064927e-08,
      "loss": 1.2139,
      "step": 3374
    },
    {
      "epoch": 0.99,
      "learning_rate": 1.9677915058999854e-08,
      "loss": 1.2488,
      "step": 3375
    },
    {
      "epoch": 0.99,
      "learning_rate": 1.8503474419557798e-08,
      "loss": 1.1891,
      "step": 3376
    },
    {
      "epoch": 0.99,
      "learning_rate": 1.7365159443509804e-08,
      "loss": 1.1486,
      "step": 3377
    },
    {
      "epoch": 0.99,
      "learning_rate": 1.6262971159384243e-08,
      "loss": 1.2394,
      "step": 3378
    },
    {
      "epoch": 0.99,
      "learning_rate": 1.5196910563082256e-08,
      "loss": 1.2217,
      "step": 3379
    },
    {
      "epoch": 0.99,
      "learning_rate": 1.416697861784222e-08,
      "loss": 1.1723,
      "step": 3380
    },
    {
      "epoch": 0.99,
      "learning_rate": 1.3173176254279717e-08,
      "loss": 1.2329,
      "step": 3381
    },
    {
      "epoch": 0.99,
      "learning_rate": 1.2215504370352016e-08,
      "loss": 1.1743,
      "step": 3382
    },
    {
      "epoch": 0.99,
      "learning_rate": 1.1293963831366939e-08,
      "loss": 1.2438,
      "step": 3383
    },
    {
      "epoch": 0.99,
      "learning_rate": 1.0408555469996195e-08,
      "loss": 1.2094,
      "step": 3384
    },
    {
      "epoch": 0.99,
      "learning_rate": 9.559280086253176e-09,
      "loss": 1.215,
      "step": 3385
    },
    {
      "epoch": 0.99,
      "learning_rate": 8.746138447519593e-09,
      "loss": 1.2585,
      "step": 3386
    },
    {
      "epoch": 0.99,
      "learning_rate": 7.969131288509957e-09,
      "loss": 1.1919,
      "step": 3387
    },
    {
      "epoch": 0.99,
      "learning_rate": 7.228259311293784e-09,
      "loss": 1.1525,
      "step": 3388
    },
    {
      "epoch": 0.99,
      "learning_rate": 6.523523185300029e-09,
      "loss": 1.1838,
      "step": 3389
    },
    {
      "epoch": 0.99,
      "learning_rate": 5.854923547294888e-09,
      "loss": 1.2912,
      "step": 3390
    },
    {
      "epoch": 0.99,
      "learning_rate": 5.222461001399559e-09,
      "loss": 1.2181,
      "step": 3391
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.626136119081359e-09,
      "loss": 1.2081,
      "step": 3392
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.065949439158168e-09,
      "loss": 1.1659,
      "step": 3393
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.541901467785103e-09,
      "loss": 1.1937,
      "step": 3394
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.053992678476725e-09,
      "loss": 1.2296,
      "step": 3395
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.602223512089275e-09,
      "loss": 1.1585,
      "step": 3396
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.186594376816231e-09,
      "loss": 1.1891,
      "step": 3397
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.8071056482105166e-09,
      "loss": 1.2397,
      "step": 3398
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.4637576691534095e-09,
      "loss": 1.2282,
      "step": 3399
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.156550749890073e-09,
      "loss": 1.1682,
      "step": 3400
    },
    {
      "epoch": 1.0,
      "eval_loss": 0.11866223812103271,
      "eval_runtime": 24.3505,
      "eval_samples_per_second": 82.134,
      "eval_steps_per_second": 0.657,
      "step": 3400
    },
    {
      "epoch": 1.0,
      "learning_rate": 8.854851679984678e-10,
      "loss": 1.2339,
      "step": 3401
    },
    {
      "epoch": 1.0,
      "learning_rate": 6.505611683937929e-10,
      "loss": 1.2304,
      "step": 3402
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.517789633551317e-10,
      "loss": 1.2208,
      "step": 3403
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.891387324899242e-10,
      "loss": 1.2399,
      "step": 3404
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.6264062274728986e-10,
      "loss": 1.271,
      "step": 3405
    },
    {
      "epoch": 1.0,
      "learning_rate": 7.228474843579136e-11,
      "loss": 1.1947,
      "step": 3406
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.8071191192348125e-11,
      "loss": 1.2193,
      "step": 3407
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.0,
      "loss": 1.2438,
      "step": 3408
    },
    {
      "epoch": 1.0,
      "step": 3408,
      "total_flos": 0.0,
      "train_loss": 0.14557739074381304,
      "train_runtime": 9584.0536,
      "train_samples_per_second": 273.278,
      "train_steps_per_second": 0.356
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 3408,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 100,
  "total_flos": 0.0,
  "train_batch_size": 12,
  "trial_name": null,
  "trial_params": null
}