{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.7384331371870313,
  "eval_steps": 500,
  "global_step": 800,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 1.020408163265306e-06,
      "loss": 0.9062,
      "step": 1
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.040816326530612e-06,
      "loss": 0.8944,
      "step": 2
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.0612244897959185e-06,
      "loss": 0.9249,
      "step": 3
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.081632653061224e-06,
      "loss": 0.9608,
      "step": 4
    },
    {
      "epoch": 0.0,
      "learning_rate": 5.102040816326531e-06,
      "loss": 0.9593,
      "step": 5
    },
    {
      "epoch": 0.01,
      "learning_rate": 6.122448979591837e-06,
      "loss": 0.8724,
      "step": 6
    },
    {
      "epoch": 0.01,
      "learning_rate": 7.142857142857143e-06,
      "loss": 0.897,
      "step": 7
    },
    {
      "epoch": 0.01,
      "learning_rate": 8.163265306122448e-06,
      "loss": 0.9658,
      "step": 8
    },
    {
      "epoch": 0.01,
      "learning_rate": 9.183673469387756e-06,
      "loss": 0.9833,
      "step": 9
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.0204081632653061e-05,
      "loss": 1.0093,
      "step": 10
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.1224489795918369e-05,
      "loss": 0.9486,
      "step": 11
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.2244897959183674e-05,
      "loss": 1.043,
      "step": 12
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.3265306122448982e-05,
      "loss": 0.894,
      "step": 13
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.4285714285714285e-05,
      "loss": 0.8912,
      "step": 14
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.5306122448979594e-05,
      "loss": 0.9604,
      "step": 15
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.6326530612244897e-05,
      "loss": 0.9794,
      "step": 16
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.7346938775510206e-05,
      "loss": 0.8549,
      "step": 17
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.836734693877551e-05,
      "loss": 0.8693,
      "step": 18
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.9387755102040817e-05,
      "loss": 0.9678,
      "step": 19
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.0408163265306123e-05,
      "loss": 0.8967,
      "step": 20
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.1428571428571428e-05,
      "loss": 0.9997,
      "step": 21
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.2448979591836737e-05,
      "loss": 0.9781,
      "step": 22
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.3469387755102043e-05,
      "loss": 0.961,
      "step": 23
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.448979591836735e-05,
      "loss": 0.939,
      "step": 24
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.5510204081632654e-05,
      "loss": 0.924,
      "step": 25
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.6530612244897963e-05,
      "loss": 0.9584,
      "step": 26
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.7551020408163265e-05,
      "loss": 1.012,
      "step": 27
    },
    {
      "epoch": 0.03,
      "learning_rate": 2.857142857142857e-05,
      "loss": 0.9738,
      "step": 28
    },
    {
      "epoch": 0.03,
      "learning_rate": 2.959183673469388e-05,
      "loss": 0.9314,
      "step": 29
    },
    {
      "epoch": 0.03,
      "learning_rate": 3.061224489795919e-05,
      "loss": 0.911,
      "step": 30
    },
    {
      "epoch": 0.03,
      "learning_rate": 3.1632653061224494e-05,
      "loss": 0.8701,
      "step": 31
    },
    {
      "epoch": 0.03,
      "learning_rate": 3.265306122448979e-05,
      "loss": 1.0048,
      "step": 32
    },
    {
      "epoch": 0.03,
      "learning_rate": 3.36734693877551e-05,
      "loss": 0.9139,
      "step": 33
    },
    {
      "epoch": 0.03,
      "learning_rate": 3.469387755102041e-05,
      "loss": 0.9547,
      "step": 34
    },
    {
      "epoch": 0.03,
      "learning_rate": 3.571428571428572e-05,
      "loss": 0.8821,
      "step": 35
    },
    {
      "epoch": 0.03,
      "learning_rate": 3.673469387755102e-05,
      "loss": 0.9439,
      "step": 36
    },
    {
      "epoch": 0.03,
      "learning_rate": 3.775510204081633e-05,
      "loss": 0.96,
      "step": 37
    },
    {
      "epoch": 0.04,
      "learning_rate": 3.8775510204081634e-05,
      "loss": 0.7702,
      "step": 38
    },
    {
      "epoch": 0.04,
      "learning_rate": 3.979591836734694e-05,
      "loss": 0.8078,
      "step": 39
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.0816326530612245e-05,
      "loss": 0.7712,
      "step": 40
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.183673469387756e-05,
      "loss": 0.9411,
      "step": 41
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.2857142857142856e-05,
      "loss": 0.9454,
      "step": 42
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.387755102040816e-05,
      "loss": 0.8914,
      "step": 43
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.4897959183673474e-05,
      "loss": 0.8872,
      "step": 44
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.591836734693878e-05,
      "loss": 0.8336,
      "step": 45
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.6938775510204086e-05,
      "loss": 0.9148,
      "step": 46
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.795918367346939e-05,
      "loss": 1.0001,
      "step": 47
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.89795918367347e-05,
      "loss": 0.9001,
      "step": 48
    },
    {
      "epoch": 0.05,
      "learning_rate": 5e-05,
      "loss": 0.7953,
      "step": 49
    },
    {
      "epoch": 0.05,
      "learning_rate": 5.102040816326531e-05,
      "loss": 0.747,
      "step": 50
    },
    {
      "epoch": 0.05,
      "learning_rate": 5.2040816326530614e-05,
      "loss": 0.8351,
      "step": 51
    },
    {
      "epoch": 0.05,
      "learning_rate": 5.3061224489795926e-05,
      "loss": 0.8445,
      "step": 52
    },
    {
      "epoch": 0.05,
      "learning_rate": 5.408163265306123e-05,
      "loss": 0.9047,
      "step": 53
    },
    {
      "epoch": 0.05,
      "learning_rate": 5.510204081632653e-05,
      "loss": 0.8999,
      "step": 54
    },
    {
      "epoch": 0.05,
      "learning_rate": 5.6122448979591836e-05,
      "loss": 0.8486,
      "step": 55
    },
    {
      "epoch": 0.05,
      "learning_rate": 5.714285714285714e-05,
      "loss": 0.7934,
      "step": 56
    },
    {
      "epoch": 0.05,
      "learning_rate": 5.816326530612245e-05,
      "loss": 0.7969,
      "step": 57
    },
    {
      "epoch": 0.05,
      "learning_rate": 5.918367346938776e-05,
      "loss": 0.8063,
      "step": 58
    },
    {
      "epoch": 0.05,
      "learning_rate": 6.0204081632653065e-05,
      "loss": 0.8719,
      "step": 59
    },
    {
      "epoch": 0.06,
      "learning_rate": 6.122448979591838e-05,
      "loss": 0.8005,
      "step": 60
    },
    {
      "epoch": 0.06,
      "learning_rate": 6.224489795918368e-05,
      "loss": 0.8339,
      "step": 61
    },
    {
      "epoch": 0.06,
      "learning_rate": 6.326530612244899e-05,
      "loss": 0.7684,
      "step": 62
    },
    {
      "epoch": 0.06,
      "learning_rate": 6.428571428571429e-05,
      "loss": 0.9206,
      "step": 63
    },
    {
      "epoch": 0.06,
      "learning_rate": 6.530612244897959e-05,
      "loss": 0.7635,
      "step": 64
    },
    {
      "epoch": 0.06,
      "learning_rate": 6.63265306122449e-05,
      "loss": 0.7626,
      "step": 65
    },
    {
      "epoch": 0.06,
      "learning_rate": 6.73469387755102e-05,
      "loss": 0.8199,
      "step": 66
    },
    {
      "epoch": 0.06,
      "learning_rate": 6.836734693877551e-05,
      "loss": 0.9005,
      "step": 67
    },
    {
      "epoch": 0.06,
      "learning_rate": 6.938775510204082e-05,
      "loss": 0.8558,
      "step": 68
    },
    {
      "epoch": 0.06,
      "learning_rate": 7.040816326530612e-05,
      "loss": 0.8113,
      "step": 69
    },
    {
      "epoch": 0.06,
      "learning_rate": 7.142857142857143e-05,
      "loss": 0.8554,
      "step": 70
    },
    {
      "epoch": 0.07,
      "learning_rate": 7.244897959183675e-05,
      "loss": 0.7537,
      "step": 71
    },
    {
      "epoch": 0.07,
      "learning_rate": 7.346938775510205e-05,
      "loss": 0.7017,
      "step": 72
    },
    {
      "epoch": 0.07,
      "learning_rate": 7.448979591836736e-05,
      "loss": 0.799,
      "step": 73
    },
    {
      "epoch": 0.07,
      "learning_rate": 7.551020408163266e-05,
      "loss": 0.8601,
      "step": 74
    },
    {
      "epoch": 0.07,
      "learning_rate": 7.653061224489796e-05,
      "loss": 0.7961,
      "step": 75
    },
    {
      "epoch": 0.07,
      "learning_rate": 7.755102040816327e-05,
      "loss": 0.8542,
      "step": 76
    },
    {
      "epoch": 0.07,
      "learning_rate": 7.857142857142858e-05,
      "loss": 0.6854,
      "step": 77
    },
    {
      "epoch": 0.07,
      "learning_rate": 7.959183673469388e-05,
      "loss": 0.8019,
      "step": 78
    },
    {
      "epoch": 0.07,
      "learning_rate": 8.061224489795919e-05,
      "loss": 0.7721,
      "step": 79
    },
    {
      "epoch": 0.07,
      "learning_rate": 8.163265306122449e-05,
      "loss": 0.793,
      "step": 80
    },
    {
      "epoch": 0.07,
      "learning_rate": 8.26530612244898e-05,
      "loss": 0.818,
      "step": 81
    },
    {
      "epoch": 0.08,
      "learning_rate": 8.367346938775511e-05,
      "loss": 0.808,
      "step": 82
    },
    {
      "epoch": 0.08,
      "learning_rate": 8.469387755102041e-05,
      "loss": 0.7093,
      "step": 83
    },
    {
      "epoch": 0.08,
      "learning_rate": 8.571428571428571e-05,
      "loss": 0.8447,
      "step": 84
    },
    {
      "epoch": 0.08,
      "learning_rate": 8.673469387755102e-05,
      "loss": 0.7826,
      "step": 85
    },
    {
      "epoch": 0.08,
      "learning_rate": 8.775510204081632e-05,
      "loss": 0.7929,
      "step": 86
    },
    {
      "epoch": 0.08,
      "learning_rate": 8.877551020408164e-05,
      "loss": 0.8044,
      "step": 87
    },
    {
      "epoch": 0.08,
      "learning_rate": 8.979591836734695e-05,
      "loss": 0.8364,
      "step": 88
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.081632653061225e-05,
      "loss": 0.8111,
      "step": 89
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.183673469387756e-05,
      "loss": 0.8399,
      "step": 90
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.285714285714286e-05,
      "loss": 0.8361,
      "step": 91
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.387755102040817e-05,
      "loss": 0.6948,
      "step": 92
    },
    {
      "epoch": 0.09,
      "learning_rate": 9.489795918367348e-05,
      "loss": 0.8313,
      "step": 93
    },
    {
      "epoch": 0.09,
      "learning_rate": 9.591836734693878e-05,
      "loss": 0.7753,
      "step": 94
    },
    {
      "epoch": 0.09,
      "learning_rate": 9.693877551020408e-05,
      "loss": 0.7655,
      "step": 95
    },
    {
      "epoch": 0.09,
      "learning_rate": 9.79591836734694e-05,
      "loss": 0.7128,
      "step": 96
    },
    {
      "epoch": 0.09,
      "learning_rate": 9.897959183673469e-05,
      "loss": 0.7241,
      "step": 97
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0001,
      "loss": 0.7168,
      "step": 98
    },
    {
      "epoch": 0.09,
      "learning_rate": 9.999997514905481e-05,
      "loss": 0.7903,
      "step": 99
    },
    {
      "epoch": 0.09,
      "learning_rate": 9.999990059624389e-05,
      "loss": 0.7593,
      "step": 100
    },
    {
      "epoch": 0.09,
      "learning_rate": 9.99997763416414e-05,
      "loss": 0.7333,
      "step": 101
    },
    {
      "epoch": 0.09,
      "learning_rate": 9.999960238537081e-05,
      "loss": 0.8234,
      "step": 102
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.999937872760505e-05,
      "loss": 0.7931,
      "step": 103
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.999910536856647e-05,
      "loss": 0.7901,
      "step": 104
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.999878230852677e-05,
      "loss": 0.6732,
      "step": 105
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.99984095478071e-05,
      "loss": 0.6699,
      "step": 106
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.9997987086778e-05,
      "loss": 0.8572,
      "step": 107
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.999751492585938e-05,
      "loss": 0.7167,
      "step": 108
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.999699306552061e-05,
      "loss": 0.8081,
      "step": 109
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.999642150628046e-05,
      "loss": 0.7586,
      "step": 110
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.999580024870706e-05,
      "loss": 0.8732,
      "step": 111
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.999512929341795e-05,
      "loss": 0.8539,
      "step": 112
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.999440864108011e-05,
      "loss": 0.6789,
      "step": 113
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.999363829240988e-05,
      "loss": 0.6847,
      "step": 114
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.999281824817304e-05,
      "loss": 0.76,
      "step": 115
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.99919485091847e-05,
      "loss": 0.8505,
      "step": 116
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.999102907630945e-05,
      "loss": 0.7612,
      "step": 117
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.999005995046122e-05,
      "loss": 0.8135,
      "step": 118
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.99890411326034e-05,
      "loss": 0.8179,
      "step": 119
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.998797262374866e-05,
      "loss": 0.7954,
      "step": 120
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.998685442495921e-05,
      "loss": 0.6992,
      "step": 121
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.998568653734653e-05,
      "loss": 0.7258,
      "step": 122
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.998446896207158e-05,
      "loss": 0.7217,
      "step": 123
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.998320170034465e-05,
      "loss": 0.7615,
      "step": 124
    },
    {
      "epoch": 0.12,
      "learning_rate": 9.998188475342547e-05,
      "loss": 0.8069,
      "step": 125
    },
    {
      "epoch": 0.12,
      "learning_rate": 9.998051812262311e-05,
      "loss": 0.7971,
      "step": 126
    },
    {
      "epoch": 0.12,
      "learning_rate": 9.997910180929609e-05,
      "loss": 0.8476,
      "step": 127
    },
    {
      "epoch": 0.12,
      "learning_rate": 9.997763581485222e-05,
      "loss": 0.85,
      "step": 128
    },
    {
      "epoch": 0.12,
      "learning_rate": 9.997612014074882e-05,
      "loss": 0.7718,
      "step": 129
    },
    {
      "epoch": 0.12,
      "learning_rate": 9.997455478849246e-05,
      "loss": 0.7807,
      "step": 130
    },
    {
      "epoch": 0.12,
      "learning_rate": 9.997293975963923e-05,
      "loss": 0.7708,
      "step": 131
    },
    {
      "epoch": 0.12,
      "learning_rate": 9.997127505579449e-05,
      "loss": 0.8352,
      "step": 132
    },
    {
      "epoch": 0.12,
      "learning_rate": 9.996956067861301e-05,
      "loss": 0.7092,
      "step": 133
    },
    {
      "epoch": 0.12,
      "learning_rate": 9.996779662979897e-05,
      "loss": 0.793,
      "step": 134
    },
    {
      "epoch": 0.12,
      "learning_rate": 9.996598291110588e-05,
      "loss": 0.719,
      "step": 135
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.996411952433667e-05,
      "loss": 0.8185,
      "step": 136
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.996220647134358e-05,
      "loss": 0.7086,
      "step": 137
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.99602437540283e-05,
      "loss": 0.8095,
      "step": 138
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.99582313743418e-05,
      "loss": 0.7667,
      "step": 139
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.995616933428451e-05,
      "loss": 0.7714,
      "step": 140
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.995405763590615e-05,
      "loss": 0.7935,
      "step": 141
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.995189628130581e-05,
      "loss": 0.7901,
      "step": 142
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.9949685272632e-05,
      "loss": 0.7186,
      "step": 143
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.994742461208251e-05,
      "loss": 0.7924,
      "step": 144
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.994511430190454e-05,
      "loss": 0.8249,
      "step": 145
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.994275434439463e-05,
      "loss": 0.7549,
      "step": 146
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.994034474189865e-05,
      "loss": 0.6845,
      "step": 147
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.993788549681185e-05,
      "loss": 0.8407,
      "step": 148
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.993537661157881e-05,
      "loss": 0.7134,
      "step": 149
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.993281808869345e-05,
      "loss": 0.7939,
      "step": 150
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.993020993069903e-05,
      "loss": 0.8207,
      "step": 151
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.992755214018819e-05,
      "loss": 0.6635,
      "step": 152
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.992484471980284e-05,
      "loss": 0.7426,
      "step": 153
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.992208767223428e-05,
      "loss": 0.8217,
      "step": 154
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.991928100022312e-05,
      "loss": 0.9371,
      "step": 155
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.991642470655929e-05,
      "loss": 0.7407,
      "step": 156
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.991351879408204e-05,
      "loss": 0.6728,
      "step": 157
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.991056326567999e-05,
      "loss": 0.883,
      "step": 158
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.9907558124291e-05,
      "loss": 0.8625,
      "step": 159
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.990450337290233e-05,
      "loss": 0.7297,
      "step": 160
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.990139901455053e-05,
      "loss": 0.7908,
      "step": 161
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.989824505232141e-05,
      "loss": 0.9057,
      "step": 162
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.989504148935017e-05,
      "loss": 0.7044,
      "step": 163
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.989178832882124e-05,
      "loss": 0.7775,
      "step": 164
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.98884855739684e-05,
      "loss": 0.8189,
      "step": 165
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.988513322807469e-05,
      "loss": 0.7743,
      "step": 166
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.988173129447252e-05,
      "loss": 0.7529,
      "step": 167
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.987827977654347e-05,
      "loss": 0.7854,
      "step": 168
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.987477867771854e-05,
      "loss": 0.7745,
      "step": 169
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.987122800147792e-05,
      "loss": 0.7373,
      "step": 170
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.986762775135114e-05,
      "loss": 0.7396,
      "step": 171
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.986397793091697e-05,
      "loss": 0.7805,
      "step": 172
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.986027854380347e-05,
      "loss": 0.7949,
      "step": 173
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.985652959368797e-05,
      "loss": 0.7751,
      "step": 174
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.985273108429707e-05,
      "loss": 0.7274,
      "step": 175
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.984888301940664e-05,
      "loss": 0.8411,
      "step": 176
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.984498540284178e-05,
      "loss": 0.7724,
      "step": 177
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.98410382384769e-05,
      "loss": 0.767,
      "step": 178
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.98370415302356e-05,
      "loss": 0.7993,
      "step": 179
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.98329952820908e-05,
      "loss": 0.8037,
      "step": 180
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.982889949806457e-05,
      "loss": 0.7844,
      "step": 181
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.982475418222829e-05,
      "loss": 0.6842,
      "step": 182
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.98205593387026e-05,
      "loss": 0.7711,
      "step": 183
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.98163149716573e-05,
      "loss": 0.7052,
      "step": 184
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.981202108531145e-05,
      "loss": 0.8863,
      "step": 185
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.980767768393335e-05,
      "loss": 0.758,
      "step": 186
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.980328477184047e-05,
      "loss": 0.806,
      "step": 187
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.979884235339958e-05,
      "loss": 0.7591,
      "step": 188
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.979435043302658e-05,
      "loss": 0.7416,
      "step": 189
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.978980901518663e-05,
      "loss": 0.769,
      "step": 190
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.978521810439405e-05,
      "loss": 0.8024,
      "step": 191
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.978057770521239e-05,
      "loss": 0.7013,
      "step": 192
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.977588782225438e-05,
      "loss": 0.8736,
      "step": 193
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.977114846018194e-05,
      "loss": 0.7576,
      "step": 194
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.976635962370617e-05,
      "loss": 0.733,
      "step": 195
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.976152131758736e-05,
      "loss": 0.7636,
      "step": 196
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.975663354663498e-05,
      "loss": 0.6974,
      "step": 197
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.975169631570765e-05,
      "loss": 0.6674,
      "step": 198
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.974670962971315e-05,
      "loss": 0.7825,
      "step": 199
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.974167349360845e-05,
      "loss": 0.712,
      "step": 200
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.973658791239965e-05,
      "loss": 0.7248,
      "step": 201
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.973145289114203e-05,
      "loss": 0.7855,
      "step": 202
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.972626843493998e-05,
      "loss": 0.6856,
      "step": 203
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.972103454894705e-05,
      "loss": 0.7341,
      "step": 204
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.971575123836591e-05,
      "loss": 0.812,
      "step": 205
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.97104185084484e-05,
      "loss": 0.8348,
      "step": 206
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.970503636449542e-05,
      "loss": 0.7607,
      "step": 207
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.969960481185703e-05,
      "loss": 0.7174,
      "step": 208
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.969412385593243e-05,
      "loss": 0.7923,
      "step": 209
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.968859350216985e-05,
      "loss": 0.7405,
      "step": 210
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.968301375606671e-05,
      "loss": 0.7062,
      "step": 211
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.967738462316947e-05,
      "loss": 0.7513,
      "step": 212
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.967170610907369e-05,
      "loss": 0.7079,
      "step": 213
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.966597821942406e-05,
      "loss": 0.7921,
      "step": 214
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.96602009599143e-05,
      "loss": 0.8295,
      "step": 215
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.965437433628722e-05,
      "loss": 0.8146,
      "step": 216
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.964849835433472e-05,
      "loss": 0.7746,
      "step": 217
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.964257301989772e-05,
      "loss": 0.7192,
      "step": 218
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.963659833886627e-05,
      "loss": 0.7096,
      "step": 219
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.963057431717939e-05,
      "loss": 0.7855,
      "step": 220
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.96245009608252e-05,
      "loss": 0.8198,
      "step": 221
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.961837827584083e-05,
      "loss": 0.7743,
      "step": 222
    },
    {
      "epoch": 0.21,
      "learning_rate": 9.96122062683125e-05,
      "loss": 0.8346,
      "step": 223
    },
    {
      "epoch": 0.21,
      "learning_rate": 9.960598494437539e-05,
      "loss": 0.8074,
      "step": 224
    },
    {
      "epoch": 0.21,
      "learning_rate": 9.959971431021373e-05,
      "loss": 0.8261,
      "step": 225
    },
    {
      "epoch": 0.21,
      "learning_rate": 9.959339437206077e-05,
      "loss": 0.7818,
      "step": 226
    },
    {
      "epoch": 0.21,
      "learning_rate": 9.958702513619875e-05,
      "loss": 0.8279,
      "step": 227
    },
    {
      "epoch": 0.21,
      "learning_rate": 9.958060660895898e-05,
      "loss": 0.8774,
      "step": 228
    },
    {
      "epoch": 0.21,
      "learning_rate": 9.957413879672168e-05,
      "loss": 0.9042,
      "step": 229
    },
    {
      "epoch": 0.21,
      "learning_rate": 9.956762170591611e-05,
      "loss": 0.8075,
      "step": 230
    },
    {
      "epoch": 0.21,
      "learning_rate": 9.956105534302052e-05,
      "loss": 0.8327,
      "step": 231
    },
    {
      "epoch": 0.21,
      "learning_rate": 9.955443971456209e-05,
      "loss": 0.8518,
      "step": 232
    },
    {
      "epoch": 0.22,
      "learning_rate": 9.954777482711702e-05,
      "loss": 0.7535,
      "step": 233
    },
    {
      "epoch": 0.22,
      "learning_rate": 9.954106068731047e-05,
      "loss": 0.7773,
      "step": 234
    },
    {
      "epoch": 0.22,
      "learning_rate": 9.953429730181653e-05,
      "loss": 0.7843,
      "step": 235
    },
    {
      "epoch": 0.22,
      "learning_rate": 9.952748467735829e-05,
      "loss": 0.7758,
      "step": 236
    },
    {
      "epoch": 0.22,
      "learning_rate": 9.952062282070774e-05,
      "loss": 0.6859,
      "step": 237
    },
    {
      "epoch": 0.22,
      "learning_rate": 9.95137117386858e-05,
      "loss": 0.8009,
      "step": 238
    },
    {
      "epoch": 0.22,
      "learning_rate": 9.950675143816239e-05,
      "loss": 0.8007,
      "step": 239
    },
    {
      "epoch": 0.22,
      "learning_rate": 9.94997419260563e-05,
      "loss": 0.7254,
      "step": 240
    },
    {
      "epoch": 0.22,
      "learning_rate": 9.949268320933522e-05,
      "loss": 0.7972,
      "step": 241
    },
    {
      "epoch": 0.22,
      "learning_rate": 9.948557529501581e-05,
      "loss": 0.7696,
      "step": 242
    },
    {
      "epoch": 0.22,
      "learning_rate": 9.947841819016361e-05,
      "loss": 0.8176,
      "step": 243
    },
    {
      "epoch": 0.23,
      "learning_rate": 9.947121190189303e-05,
      "loss": 0.7332,
      "step": 244
    },
    {
      "epoch": 0.23,
      "learning_rate": 9.946395643736741e-05,
      "loss": 0.7843,
      "step": 245
    },
    {
      "epoch": 0.23,
      "learning_rate": 9.945665180379895e-05,
      "loss": 0.7998,
      "step": 246
    },
    {
      "epoch": 0.23,
      "learning_rate": 9.944929800844873e-05,
      "loss": 0.7614,
      "step": 247
    },
    {
      "epoch": 0.23,
      "learning_rate": 9.944189505862672e-05,
      "loss": 0.7185,
      "step": 248
    },
    {
      "epoch": 0.23,
      "learning_rate": 9.943444296169168e-05,
      "loss": 0.7878,
      "step": 249
    },
    {
      "epoch": 0.23,
      "learning_rate": 9.942694172505134e-05,
      "loss": 0.7428,
      "step": 250
    },
    {
      "epoch": 0.23,
      "learning_rate": 9.941939135616219e-05,
      "loss": 0.7454,
      "step": 251
    },
    {
      "epoch": 0.23,
      "learning_rate": 9.941179186252956e-05,
      "loss": 0.7749,
      "step": 252
    },
    {
      "epoch": 0.23,
      "learning_rate": 9.940414325170764e-05,
      "loss": 0.7471,
      "step": 253
    },
    {
      "epoch": 0.23,
      "learning_rate": 9.939644553129947e-05,
      "loss": 0.7193,
      "step": 254
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.938869870895684e-05,
      "loss": 0.7657,
      "step": 255
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.938090279238042e-05,
      "loss": 0.8586,
      "step": 256
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.93730577893196e-05,
      "loss": 0.7406,
      "step": 257
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.936516370757265e-05,
      "loss": 0.7678,
      "step": 258
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.935722055498655e-05,
      "loss": 0.7676,
      "step": 259
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.934922833945714e-05,
      "loss": 0.7335,
      "step": 260
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.934118706892895e-05,
      "loss": 0.7691,
      "step": 261
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.93330967513953e-05,
      "loss": 0.7785,
      "step": 262
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.932495739489829e-05,
      "loss": 0.7941,
      "step": 263
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.931676900752875e-05,
      "loss": 0.7973,
      "step": 264
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.930853159742624e-05,
      "loss": 0.8148,
      "step": 265
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.930024517277907e-05,
      "loss": 0.7977,
      "step": 266
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.929190974182423e-05,
      "loss": 0.8086,
      "step": 267
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.928352531284748e-05,
      "loss": 0.7577,
      "step": 268
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.927509189418325e-05,
      "loss": 0.8167,
      "step": 269
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.926660949421467e-05,
      "loss": 0.694,
      "step": 270
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.925807812137358e-05,
      "loss": 0.8129,
      "step": 271
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.924949778414047e-05,
      "loss": 0.7939,
      "step": 272
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.924086849104455e-05,
      "loss": 0.8259,
      "step": 273
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.923219025066364e-05,
      "loss": 0.7405,
      "step": 274
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.922346307162424e-05,
      "loss": 0.783,
      "step": 275
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.92146869626015e-05,
      "loss": 0.7164,
      "step": 276
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.92058619323192e-05,
      "loss": 0.8625,
      "step": 277
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.919698798954975e-05,
      "loss": 0.7878,
      "step": 278
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.918806514311421e-05,
      "loss": 0.7381,
      "step": 279
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.917909340188218e-05,
      "loss": 0.7625,
      "step": 280
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.917007277477197e-05,
      "loss": 0.6736,
      "step": 281
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.916100327075038e-05,
      "loss": 0.7772,
      "step": 282
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.915188489883285e-05,
      "loss": 0.7715,
      "step": 283
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.914271766808339e-05,
      "loss": 0.8097,
      "step": 284
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.913350158761456e-05,
      "loss": 0.7812,
      "step": 285
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.912423666658753e-05,
      "loss": 0.7651,
      "step": 286
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.911492291421193e-05,
      "loss": 0.8206,
      "step": 287
    },
    {
      "epoch": 0.27,
      "learning_rate": 9.910556033974605e-05,
      "loss": 0.8193,
      "step": 288
    },
    {
      "epoch": 0.27,
      "learning_rate": 9.909614895249656e-05,
      "loss": 0.7724,
      "step": 289
    },
    {
      "epoch": 0.27,
      "learning_rate": 9.90866887618188e-05,
      "loss": 0.7614,
      "step": 290
    },
    {
      "epoch": 0.27,
      "learning_rate": 9.907717977711652e-05,
      "loss": 0.7192,
      "step": 291
    },
    {
      "epoch": 0.27,
      "learning_rate": 9.906762200784205e-05,
      "loss": 0.7833,
      "step": 292
    },
    {
      "epoch": 0.27,
      "learning_rate": 9.905801546349614e-05,
      "loss": 0.7243,
      "step": 293
    },
    {
      "epoch": 0.27,
      "learning_rate": 9.904836015362804e-05,
      "loss": 0.7702,
      "step": 294
    },
    {
      "epoch": 0.27,
      "learning_rate": 9.903865608783555e-05,
      "loss": 0.822,
      "step": 295
    },
    {
      "epoch": 0.27,
      "learning_rate": 9.902890327576483e-05,
      "loss": 0.826,
      "step": 296
    },
    {
      "epoch": 0.27,
      "learning_rate": 9.901910172711058e-05,
      "loss": 0.7978,
      "step": 297
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.900925145161586e-05,
      "loss": 0.773,
      "step": 298
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.899935245907228e-05,
      "loss": 0.8328,
      "step": 299
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.898940475931977e-05,
      "loss": 0.8035,
      "step": 300
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.897940836224673e-05,
      "loss": 0.7645,
      "step": 301
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.896936327778996e-05,
      "loss": 0.776,
      "step": 302
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.895926951593465e-05,
      "loss": 0.7883,
      "step": 303
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.894912708671438e-05,
      "loss": 0.8139,
      "step": 304
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.893893600021112e-05,
      "loss": 0.675,
      "step": 305
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.892869626655517e-05,
      "loss": 0.8393,
      "step": 306
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.891840789592524e-05,
      "loss": 0.7788,
      "step": 307
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.890807089854834e-05,
      "loss": 0.7463,
      "step": 308
    },
    {
      "epoch": 0.29,
      "learning_rate": 9.889768528469985e-05,
      "loss": 0.7821,
      "step": 309
    },
    {
      "epoch": 0.29,
      "learning_rate": 9.888725106470344e-05,
      "loss": 0.7797,
      "step": 310
    },
    {
      "epoch": 0.29,
      "learning_rate": 9.887676824893117e-05,
      "loss": 0.7933,
      "step": 311
    },
    {
      "epoch": 0.29,
      "learning_rate": 9.886623684780329e-05,
      "loss": 0.7835,
      "step": 312
    },
    {
      "epoch": 0.29,
      "learning_rate": 9.885565687178845e-05,
      "loss": 0.7426,
      "step": 313
    },
    {
      "epoch": 0.29,
      "learning_rate": 9.884502833140354e-05,
      "loss": 0.7285,
      "step": 314
    },
    {
      "epoch": 0.29,
      "learning_rate": 9.883435123721373e-05,
      "loss": 0.8611,
      "step": 315
    },
    {
      "epoch": 0.29,
      "learning_rate": 9.882362559983245e-05,
      "loss": 0.7882,
      "step": 316
    },
    {
      "epoch": 0.29,
      "learning_rate": 9.881285142992139e-05,
      "loss": 0.7597,
      "step": 317
    },
    {
      "epoch": 0.29,
      "learning_rate": 9.880202873819049e-05,
      "loss": 0.716,
      "step": 318
    },
    {
      "epoch": 0.29,
      "learning_rate": 9.879115753539792e-05,
      "loss": 0.7509,
      "step": 319
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.878023783235005e-05,
      "loss": 0.8133,
      "step": 320
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.876926963990147e-05,
      "loss": 0.8162,
      "step": 321
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.8758252968955e-05,
      "loss": 0.8104,
      "step": 322
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.874718783046163e-05,
      "loss": 0.7891,
      "step": 323
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.87360742354205e-05,
      "loss": 0.8742,
      "step": 324
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.872491219487898e-05,
      "loss": 0.8054,
      "step": 325
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.871370171993253e-05,
      "loss": 0.7886,
      "step": 326
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.87024428217248e-05,
      "loss": 0.745,
      "step": 327
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.869113551144754e-05,
      "loss": 0.8503,
      "step": 328
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.867977980034068e-05,
      "loss": 0.8197,
      "step": 329
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.866837569969218e-05,
      "loss": 0.7122,
      "step": 330
    },
    {
      "epoch": 0.31,
      "learning_rate": 9.865692322083821e-05,
      "loss": 0.7263,
      "step": 331
    },
    {
      "epoch": 0.31,
      "learning_rate": 9.86454223751629e-05,
      "loss": 0.7863,
      "step": 332
    },
    {
      "epoch": 0.31,
      "learning_rate": 9.863387317409858e-05,
      "loss": 0.7881,
      "step": 333
    },
    {
      "epoch": 0.31,
      "learning_rate": 9.862227562912555e-05,
      "loss": 0.7865,
      "step": 334
    },
    {
      "epoch": 0.31,
      "learning_rate": 9.861062975177223e-05,
      "loss": 0.7862,
      "step": 335
    },
    {
      "epoch": 0.31,
      "learning_rate": 9.859893555361505e-05,
      "loss": 0.7876,
      "step": 336
    },
    {
      "epoch": 0.31,
      "learning_rate": 9.858719304627853e-05,
      "loss": 0.6853,
      "step": 337
    },
    {
      "epoch": 0.31,
      "learning_rate": 9.85754022414351e-05,
      "loss": 0.6911,
      "step": 338
    },
    {
      "epoch": 0.31,
      "learning_rate": 9.856356315080531e-05,
      "loss": 0.7738,
      "step": 339
    },
    {
      "epoch": 0.31,
      "learning_rate": 9.855167578615762e-05,
      "loss": 0.787,
      "step": 340
    },
    {
      "epoch": 0.31,
      "learning_rate": 9.853974015930859e-05,
      "loss": 0.7608,
      "step": 341
    },
    {
      "epoch": 0.32,
      "learning_rate": 9.852775628212263e-05,
      "loss": 0.7859,
      "step": 342
    },
    {
      "epoch": 0.32,
      "learning_rate": 9.851572416651218e-05,
      "loss": 0.794,
      "step": 343
    },
    {
      "epoch": 0.32,
      "learning_rate": 9.850364382443761e-05,
      "loss": 0.7768,
      "step": 344
    },
    {
      "epoch": 0.32,
      "learning_rate": 9.849151526790726e-05,
      "loss": 0.7426,
      "step": 345
    },
    {
      "epoch": 0.32,
      "learning_rate": 9.847933850897735e-05,
      "loss": 0.8163,
      "step": 346
    },
    {
      "epoch": 0.32,
      "learning_rate": 9.846711355975205e-05,
      "loss": 0.7864,
      "step": 347
    },
    {
      "epoch": 0.32,
      "learning_rate": 9.845484043238343e-05,
      "loss": 0.8279,
      "step": 348
    },
    {
      "epoch": 0.32,
      "learning_rate": 9.844251913907142e-05,
      "loss": 0.6468,
      "step": 349
    },
    {
      "epoch": 0.32,
      "learning_rate": 9.843014969206386e-05,
      "loss": 0.7872,
      "step": 350
    },
    {
      "epoch": 0.32,
      "learning_rate": 9.841773210365646e-05,
      "loss": 0.7347,
      "step": 351
    },
    {
      "epoch": 0.32,
      "learning_rate": 9.840526638619276e-05,
      "loss": 0.6754,
      "step": 352
    },
    {
      "epoch": 0.33,
      "learning_rate": 9.839275255206417e-05,
      "loss": 0.764,
      "step": 353
    },
    {
      "epoch": 0.33,
      "learning_rate": 9.838019061370989e-05,
      "loss": 0.7955,
      "step": 354
    },
    {
      "epoch": 0.33,
      "learning_rate": 9.836758058361699e-05,
      "loss": 0.7462,
      "step": 355
    },
    {
      "epoch": 0.33,
      "learning_rate": 9.835492247432029e-05,
      "loss": 0.8099,
      "step": 356
    },
    {
      "epoch": 0.33,
      "learning_rate": 9.834221629840245e-05,
      "loss": 0.8193,
      "step": 357
    },
    {
      "epoch": 0.33,
      "learning_rate": 9.832946206849387e-05,
      "loss": 0.7293,
      "step": 358
    },
    {
      "epoch": 0.33,
      "learning_rate": 9.831665979727275e-05,
      "loss": 0.7788,
      "step": 359
    },
    {
      "epoch": 0.33,
      "learning_rate": 9.830380949746503e-05,
      "loss": 0.7238,
      "step": 360
    },
    {
      "epoch": 0.33,
      "learning_rate": 9.82909111818444e-05,
      "loss": 0.7332,
      "step": 361
    },
    {
      "epoch": 0.33,
      "learning_rate": 9.827796486323225e-05,
      "loss": 0.72,
      "step": 362
    },
    {
      "epoch": 0.34,
      "learning_rate": 9.826497055449772e-05,
      "loss": 0.7677,
      "step": 363
    },
    {
      "epoch": 0.34,
      "learning_rate": 9.825192826855767e-05,
      "loss": 0.7892,
      "step": 364
    },
    {
      "epoch": 0.34,
      "learning_rate": 9.823883801837659e-05,
      "loss": 0.7914,
      "step": 365
    },
    {
      "epoch": 0.34,
      "learning_rate": 9.822569981696671e-05,
      "loss": 0.7581,
      "step": 366
    },
    {
      "epoch": 0.34,
      "learning_rate": 9.821251367738789e-05,
      "loss": 0.7255,
      "step": 367
    },
    {
      "epoch": 0.34,
      "learning_rate": 9.819927961274764e-05,
      "loss": 0.8007,
      "step": 368
    },
    {
      "epoch": 0.34,
      "learning_rate": 9.818599763620113e-05,
      "loss": 0.7703,
      "step": 369
    },
    {
      "epoch": 0.34,
      "learning_rate": 9.817266776095116e-05,
      "loss": 0.7654,
      "step": 370
    },
    {
      "epoch": 0.34,
      "learning_rate": 9.81592900002481e-05,
      "loss": 0.7331,
      "step": 371
    },
    {
      "epoch": 0.34,
      "learning_rate": 9.814586436738998e-05,
      "loss": 0.8153,
      "step": 372
    },
    {
      "epoch": 0.34,
      "learning_rate": 9.813239087572236e-05,
      "loss": 0.7156,
      "step": 373
    },
    {
      "epoch": 0.35,
      "learning_rate": 9.811886953863841e-05,
      "loss": 0.7617,
      "step": 374
    },
    {
      "epoch": 0.35,
      "learning_rate": 9.810530036957887e-05,
      "loss": 0.8124,
      "step": 375
    },
    {
      "epoch": 0.35,
      "learning_rate": 9.809168338203199e-05,
      "loss": 0.7322,
      "step": 376
    },
    {
      "epoch": 0.35,
      "learning_rate": 9.807801858953356e-05,
      "loss": 0.7427,
      "step": 377
    },
    {
      "epoch": 0.35,
      "learning_rate": 9.80643060056669e-05,
      "loss": 0.7658,
      "step": 378
    },
    {
      "epoch": 0.35,
      "learning_rate": 9.805054564406286e-05,
      "loss": 0.7909,
      "step": 379
    },
    {
      "epoch": 0.35,
      "learning_rate": 9.803673751839977e-05,
      "loss": 0.8076,
      "step": 380
    },
    {
      "epoch": 0.35,
      "learning_rate": 9.802288164240339e-05,
      "loss": 0.8157,
      "step": 381
    },
    {
      "epoch": 0.35,
      "learning_rate": 9.800897802984698e-05,
      "loss": 0.7728,
      "step": 382
    },
    {
      "epoch": 0.35,
      "learning_rate": 9.79950266945513e-05,
      "loss": 0.7456,
      "step": 383
    },
    {
      "epoch": 0.35,
      "learning_rate": 9.798102765038449e-05,
      "loss": 0.7587,
      "step": 384
    },
    {
      "epoch": 0.36,
      "learning_rate": 9.79669809112621e-05,
      "loss": 0.8644,
      "step": 385
    },
    {
      "epoch": 0.36,
      "learning_rate": 9.795288649114717e-05,
      "loss": 0.7514,
      "step": 386
    },
    {
      "epoch": 0.36,
      "learning_rate": 9.793874440405005e-05,
      "loss": 0.8003,
      "step": 387
    },
    {
      "epoch": 0.36,
      "learning_rate": 9.79245546640285e-05,
      "loss": 0.7178,
      "step": 388
    },
    {
      "epoch": 0.36,
      "learning_rate": 9.791031728518768e-05,
      "loss": 0.7331,
      "step": 389
    },
    {
      "epoch": 0.36,
      "learning_rate": 9.789603228168008e-05,
      "loss": 0.7609,
      "step": 390
    },
    {
      "epoch": 0.36,
      "learning_rate": 9.788169966770553e-05,
      "loss": 0.8373,
      "step": 391
    },
    {
      "epoch": 0.36,
      "learning_rate": 9.78673194575112e-05,
      "loss": 0.8086,
      "step": 392
    },
    {
      "epoch": 0.36,
      "learning_rate": 9.785289166539156e-05,
      "loss": 0.8103,
      "step": 393
    },
    {
      "epoch": 0.36,
      "learning_rate": 9.783841630568838e-05,
      "loss": 0.752,
      "step": 394
    },
    {
      "epoch": 0.36,
      "learning_rate": 9.782389339279068e-05,
      "loss": 0.7923,
      "step": 395
    },
    {
      "epoch": 0.37,
      "learning_rate": 9.780932294113485e-05,
      "loss": 0.7154,
      "step": 396
    },
    {
      "epoch": 0.37,
      "learning_rate": 9.779470496520442e-05,
      "loss": 0.716,
      "step": 397
    },
    {
      "epoch": 0.37,
      "learning_rate": 9.778003947953022e-05,
      "loss": 0.8073,
      "step": 398
    },
    {
      "epoch": 0.37,
      "learning_rate": 9.77653264986903e-05,
      "loss": 0.8262,
      "step": 399
    },
    {
      "epoch": 0.37,
      "learning_rate": 9.775056603730994e-05,
      "loss": 0.7933,
      "step": 400
    },
    {
      "epoch": 0.37,
      "learning_rate": 9.773575811006156e-05,
      "loss": 0.7491,
      "step": 401
    },
    {
      "epoch": 0.37,
      "learning_rate": 9.772090273166483e-05,
      "loss": 0.7564,
      "step": 402
    },
    {
      "epoch": 0.37,
      "learning_rate": 9.770599991688654e-05,
      "loss": 0.7997,
      "step": 403
    },
    {
      "epoch": 0.37,
      "learning_rate": 9.769104968054066e-05,
      "loss": 0.7421,
      "step": 404
    },
    {
      "epoch": 0.37,
      "learning_rate": 9.767605203748828e-05,
      "loss": 0.7948,
      "step": 405
    },
    {
      "epoch": 0.37,
      "learning_rate": 9.766100700263765e-05,
      "loss": 0.7489,
      "step": 406
    },
    {
      "epoch": 0.38,
      "learning_rate": 9.764591459094407e-05,
      "loss": 0.8029,
      "step": 407
    },
    {
      "epoch": 0.38,
      "learning_rate": 9.763077481740999e-05,
      "loss": 0.7862,
      "step": 408
    },
    {
      "epoch": 0.38,
      "learning_rate": 9.761558769708489e-05,
      "loss": 0.7895,
      "step": 409
    },
    {
      "epoch": 0.38,
      "learning_rate": 9.760035324506539e-05,
      "loss": 0.76,
      "step": 410
    },
    {
      "epoch": 0.38,
      "learning_rate": 9.758507147649507e-05,
      "loss": 0.7652,
      "step": 411
    },
    {
      "epoch": 0.38,
      "learning_rate": 9.75697424065646e-05,
      "loss": 0.8068,
      "step": 412
    },
    {
      "epoch": 0.38,
      "learning_rate": 9.755436605051166e-05,
      "loss": 0.7812,
      "step": 413
    },
    {
      "epoch": 0.38,
      "learning_rate": 9.753894242362091e-05,
      "loss": 0.7383,
      "step": 414
    },
    {
      "epoch": 0.38,
      "learning_rate": 9.752347154122406e-05,
      "loss": 0.707,
      "step": 415
    },
    {
      "epoch": 0.38,
      "learning_rate": 9.750795341869969e-05,
      "loss": 0.7741,
      "step": 416
    },
    {
      "epoch": 0.38,
      "learning_rate": 9.749238807147346e-05,
      "loss": 0.7334,
      "step": 417
    },
    {
      "epoch": 0.39,
      "learning_rate": 9.747677551501789e-05,
      "loss": 0.7737,
      "step": 418
    },
    {
      "epoch": 0.39,
      "learning_rate": 9.746111576485242e-05,
      "loss": 0.7499,
      "step": 419
    },
    {
      "epoch": 0.39,
      "learning_rate": 9.744540883654349e-05,
      "loss": 0.7069,
      "step": 420
    },
    {
      "epoch": 0.39,
      "learning_rate": 9.742965474570434e-05,
      "loss": 0.7469,
      "step": 421
    },
    {
      "epoch": 0.39,
      "learning_rate": 9.741385350799514e-05,
      "loss": 0.7255,
      "step": 422
    },
    {
      "epoch": 0.39,
      "learning_rate": 9.739800513912294e-05,
      "loss": 0.7632,
      "step": 423
    },
    {
      "epoch": 0.39,
      "learning_rate": 9.738210965484159e-05,
      "loss": 0.7615,
      "step": 424
    },
    {
      "epoch": 0.39,
      "learning_rate": 9.736616707095182e-05,
      "loss": 0.7707,
      "step": 425
    },
    {
      "epoch": 0.39,
      "learning_rate": 9.735017740330115e-05,
      "loss": 0.7767,
      "step": 426
    },
    {
      "epoch": 0.39,
      "learning_rate": 9.733414066778393e-05,
      "loss": 0.8117,
      "step": 427
    },
    {
      "epoch": 0.4,
      "learning_rate": 9.731805688034126e-05,
      "loss": 0.8012,
      "step": 428
    },
    {
      "epoch": 0.4,
      "learning_rate": 9.730192605696105e-05,
      "loss": 0.7527,
      "step": 429
    },
    {
      "epoch": 0.4,
      "learning_rate": 9.728574821367793e-05,
      "loss": 0.8018,
      "step": 430
    },
    {
      "epoch": 0.4,
      "learning_rate": 9.72695233665733e-05,
      "loss": 0.7729,
      "step": 431
    },
    {
      "epoch": 0.4,
      "learning_rate": 9.725325153177527e-05,
      "loss": 0.708,
      "step": 432
    },
    {
      "epoch": 0.4,
      "learning_rate": 9.723693272545868e-05,
      "loss": 0.8237,
      "step": 433
    },
    {
      "epoch": 0.4,
      "learning_rate": 9.7220566963845e-05,
      "loss": 0.697,
      "step": 434
    },
    {
      "epoch": 0.4,
      "learning_rate": 9.720415426320245e-05,
      "loss": 0.7881,
      "step": 435
    },
    {
      "epoch": 0.4,
      "learning_rate": 9.718769463984585e-05,
      "loss": 0.8093,
      "step": 436
    },
    {
      "epoch": 0.4,
      "learning_rate": 9.717118811013669e-05,
      "loss": 0.7302,
      "step": 437
    },
    {
      "epoch": 0.4,
      "learning_rate": 9.715463469048311e-05,
      "loss": 0.8122,
      "step": 438
    },
    {
      "epoch": 0.41,
      "learning_rate": 9.713803439733981e-05,
      "loss": 0.8162,
      "step": 439
    },
    {
      "epoch": 0.41,
      "learning_rate": 9.712138724720812e-05,
      "loss": 0.7271,
      "step": 440
    },
    {
      "epoch": 0.41,
      "learning_rate": 9.710469325663593e-05,
      "loss": 0.6953,
      "step": 441
    },
    {
      "epoch": 0.41,
      "learning_rate": 9.70879524422177e-05,
      "loss": 0.8185,
      "step": 442
    },
    {
      "epoch": 0.41,
      "learning_rate": 9.707116482059446e-05,
      "loss": 0.7468,
      "step": 443
    },
    {
      "epoch": 0.41,
      "learning_rate": 9.705433040845369e-05,
      "loss": 0.7621,
      "step": 444
    },
    {
      "epoch": 0.41,
      "learning_rate": 9.703744922252947e-05,
      "loss": 0.7747,
      "step": 445
    },
    {
      "epoch": 0.41,
      "learning_rate": 9.702052127960232e-05,
      "loss": 0.7619,
      "step": 446
    },
    {
      "epoch": 0.41,
      "learning_rate": 9.700354659649926e-05,
      "loss": 0.7599,
      "step": 447
    },
    {
      "epoch": 0.41,
      "learning_rate": 9.698652519009376e-05,
      "loss": 0.6768,
      "step": 448
    },
    {
      "epoch": 0.41,
      "learning_rate": 9.696945707730575e-05,
      "loss": 0.8058,
      "step": 449
    },
    {
      "epoch": 0.42,
      "learning_rate": 9.695234227510157e-05,
      "loss": 0.7743,
      "step": 450
    },
    {
      "epoch": 0.42,
      "learning_rate": 9.6935180800494e-05,
      "loss": 0.7604,
      "step": 451
    },
    {
      "epoch": 0.42,
      "learning_rate": 9.691797267054218e-05,
      "loss": 0.7516,
      "step": 452
    },
    {
      "epoch": 0.42,
      "learning_rate": 9.690071790235164e-05,
      "loss": 0.7796,
      "step": 453
    },
    {
      "epoch": 0.42,
      "learning_rate": 9.688341651307429e-05,
      "loss": 0.7869,
      "step": 454
    },
    {
      "epoch": 0.42,
      "learning_rate": 9.686606851990834e-05,
      "loss": 0.7175,
      "step": 455
    },
    {
      "epoch": 0.42,
      "learning_rate": 9.684867394009835e-05,
      "loss": 0.8029,
      "step": 456
    },
    {
      "epoch": 0.42,
      "learning_rate": 9.683123279093522e-05,
      "loss": 0.7497,
      "step": 457
    },
    {
      "epoch": 0.42,
      "learning_rate": 9.68137450897561e-05,
      "loss": 0.8125,
      "step": 458
    },
    {
      "epoch": 0.42,
      "learning_rate": 9.67962108539444e-05,
      "loss": 0.7931,
      "step": 459
    },
    {
      "epoch": 0.42,
      "learning_rate": 9.677863010092983e-05,
      "loss": 0.7585,
      "step": 460
    },
    {
      "epoch": 0.43,
      "learning_rate": 9.676100284818834e-05,
      "loss": 0.6514,
      "step": 461
    },
    {
      "epoch": 0.43,
      "learning_rate": 9.674332911324208e-05,
      "loss": 0.7607,
      "step": 462
    },
    {
      "epoch": 0.43,
      "learning_rate": 9.672560891365938e-05,
      "loss": 0.7595,
      "step": 463
    },
    {
      "epoch": 0.43,
      "learning_rate": 9.670784226705483e-05,
      "loss": 0.7217,
      "step": 464
    },
    {
      "epoch": 0.43,
      "learning_rate": 9.669002919108913e-05,
      "loss": 0.8231,
      "step": 465
    },
    {
      "epoch": 0.43,
      "learning_rate": 9.667216970346914e-05,
      "loss": 0.7936,
      "step": 466
    },
    {
      "epoch": 0.43,
      "learning_rate": 9.665426382194789e-05,
      "loss": 0.7324,
      "step": 467
    },
    {
      "epoch": 0.43,
      "learning_rate": 9.663631156432447e-05,
      "loss": 0.7812,
      "step": 468
    },
    {
      "epoch": 0.43,
      "learning_rate": 9.661831294844414e-05,
      "loss": 0.8182,
      "step": 469
    },
    {
      "epoch": 0.43,
      "learning_rate": 9.660026799219819e-05,
      "loss": 0.7856,
      "step": 470
    },
    {
      "epoch": 0.43,
      "learning_rate": 9.658217671352396e-05,
      "loss": 0.7669,
      "step": 471
    },
    {
      "epoch": 0.44,
      "learning_rate": 9.656403913040492e-05,
      "loss": 0.7184,
      "step": 472
    },
    {
      "epoch": 0.44,
      "learning_rate": 9.654585526087047e-05,
      "loss": 0.7796,
      "step": 473
    },
    {
      "epoch": 0.44,
      "learning_rate": 9.652762512299608e-05,
      "loss": 0.8011,
      "step": 474
    },
    {
      "epoch": 0.44,
      "learning_rate": 9.65093487349032e-05,
      "loss": 0.8498,
      "step": 475
    },
    {
      "epoch": 0.44,
      "learning_rate": 9.649102611475925e-05,
      "loss": 0.787,
      "step": 476
    },
    {
      "epoch": 0.44,
      "learning_rate": 9.64726572807776e-05,
      "loss": 0.8579,
      "step": 477
    },
    {
      "epoch": 0.44,
      "learning_rate": 9.645424225121755e-05,
      "loss": 0.8518,
      "step": 478
    },
    {
      "epoch": 0.44,
      "learning_rate": 9.643578104438438e-05,
      "loss": 0.6378,
      "step": 479
    },
    {
      "epoch": 0.44,
      "learning_rate": 9.641727367862918e-05,
      "loss": 0.7723,
      "step": 480
    },
    {
      "epoch": 0.44,
      "learning_rate": 9.6398720172349e-05,
      "loss": 0.7584,
      "step": 481
    },
    {
      "epoch": 0.44,
      "learning_rate": 9.638012054398672e-05,
      "loss": 0.7664,
      "step": 482
    },
    {
      "epoch": 0.45,
      "learning_rate": 9.636147481203108e-05,
      "loss": 0.8043,
      "step": 483
    },
    {
      "epoch": 0.45,
      "learning_rate": 9.634278299501662e-05,
      "loss": 0.81,
      "step": 484
    },
    {
      "epoch": 0.45,
      "learning_rate": 9.632404511152374e-05,
      "loss": 0.7361,
      "step": 485
    },
    {
      "epoch": 0.45,
      "learning_rate": 9.630526118017858e-05,
      "loss": 0.7118,
      "step": 486
    },
    {
      "epoch": 0.45,
      "learning_rate": 9.62864312196531e-05,
      "loss": 0.8345,
      "step": 487
    },
    {
      "epoch": 0.45,
      "learning_rate": 9.626755524866499e-05,
      "loss": 0.8489,
      "step": 488
    },
    {
      "epoch": 0.45,
      "learning_rate": 9.624863328597766e-05,
      "loss": 0.7286,
      "step": 489
    },
    {
      "epoch": 0.45,
      "learning_rate": 9.622966535040028e-05,
      "loss": 0.7772,
      "step": 490
    },
    {
      "epoch": 0.45,
      "learning_rate": 9.621065146078769e-05,
      "loss": 0.7915,
      "step": 491
    },
    {
      "epoch": 0.45,
      "learning_rate": 9.619159163604039e-05,
      "loss": 0.7121,
      "step": 492
    },
    {
      "epoch": 0.46,
      "learning_rate": 9.617248589510459e-05,
      "loss": 0.7406,
      "step": 493
    },
    {
      "epoch": 0.46,
      "learning_rate": 9.615333425697213e-05,
      "loss": 0.771,
      "step": 494
    },
    {
      "epoch": 0.46,
      "learning_rate": 9.613413674068042e-05,
      "loss": 0.7503,
      "step": 495
    },
    {
      "epoch": 0.46,
      "learning_rate": 9.611489336531256e-05,
      "loss": 0.817,
      "step": 496
    },
    {
      "epoch": 0.46,
      "learning_rate": 9.609560414999718e-05,
      "loss": 0.8607,
      "step": 497
    },
    {
      "epoch": 0.46,
      "learning_rate": 9.607626911390848e-05,
      "loss": 0.7837,
      "step": 498
    },
    {
      "epoch": 0.46,
      "learning_rate": 9.605688827626623e-05,
      "loss": 0.7698,
      "step": 499
    },
    {
      "epoch": 0.46,
      "learning_rate": 9.60374616563357e-05,
      "loss": 0.7799,
      "step": 500
    },
    {
      "epoch": 0.46,
      "learning_rate": 9.601798927342769e-05,
      "loss": 0.7782,
      "step": 501
    },
    {
      "epoch": 0.46,
      "learning_rate": 9.599847114689848e-05,
      "loss": 0.7615,
      "step": 502
    },
    {
      "epoch": 0.46,
      "learning_rate": 9.597890729614985e-05,
      "loss": 0.8254,
      "step": 503
    },
    {
      "epoch": 0.47,
      "learning_rate": 9.5959297740629e-05,
      "loss": 0.7277,
      "step": 504
    },
    {
      "epoch": 0.47,
      "learning_rate": 9.593964249982855e-05,
      "loss": 0.742,
      "step": 505
    },
    {
      "epoch": 0.47,
      "learning_rate": 9.591994159328655e-05,
      "loss": 0.7961,
      "step": 506
    },
    {
      "epoch": 0.47,
      "learning_rate": 9.590019504058646e-05,
      "loss": 0.7536,
      "step": 507
    },
    {
      "epoch": 0.47,
      "learning_rate": 9.58804028613571e-05,
      "loss": 0.6764,
      "step": 508
    },
    {
      "epoch": 0.47,
      "learning_rate": 9.586056507527266e-05,
      "loss": 0.7498,
      "step": 509
    },
    {
      "epoch": 0.47,
      "learning_rate": 9.584068170205261e-05,
      "loss": 0.7324,
      "step": 510
    },
    {
      "epoch": 0.47,
      "learning_rate": 9.582075276146181e-05,
      "loss": 0.7647,
      "step": 511
    },
    {
      "epoch": 0.47,
      "learning_rate": 9.580077827331038e-05,
      "loss": 0.7503,
      "step": 512
    },
    {
      "epoch": 0.47,
      "learning_rate": 9.578075825745367e-05,
      "loss": 0.7379,
      "step": 513
    },
    {
      "epoch": 0.47,
      "learning_rate": 9.57606927337924e-05,
      "loss": 0.7972,
      "step": 514
    },
    {
      "epoch": 0.48,
      "learning_rate": 9.57405817222724e-05,
      "loss": 0.8258,
      "step": 515
    },
    {
      "epoch": 0.48,
      "learning_rate": 9.572042524288482e-05,
      "loss": 0.7676,
      "step": 516
    },
    {
      "epoch": 0.48,
      "learning_rate": 9.570022331566594e-05,
      "loss": 0.701,
      "step": 517
    },
    {
      "epoch": 0.48,
      "learning_rate": 9.567997596069724e-05,
      "loss": 0.8044,
      "step": 518
    },
    {
      "epoch": 0.48,
      "learning_rate": 9.565968319810536e-05,
      "loss": 0.7369,
      "step": 519
    },
    {
      "epoch": 0.48,
      "learning_rate": 9.563934504806207e-05,
      "loss": 0.7104,
      "step": 520
    },
    {
      "epoch": 0.48,
      "learning_rate": 9.561896153078427e-05,
      "loss": 0.7893,
      "step": 521
    },
    {
      "epoch": 0.48,
      "learning_rate": 9.559853266653393e-05,
      "loss": 0.6855,
      "step": 522
    },
    {
      "epoch": 0.48,
      "learning_rate": 9.557805847561813e-05,
      "loss": 0.7471,
      "step": 523
    },
    {
      "epoch": 0.48,
      "learning_rate": 9.5557538978389e-05,
      "loss": 0.7618,
      "step": 524
    },
    {
      "epoch": 0.48,
      "learning_rate": 9.553697419524365e-05,
      "loss": 0.768,
      "step": 525
    },
    {
      "epoch": 0.49,
      "learning_rate": 9.551636414662431e-05,
      "loss": 0.7314,
      "step": 526
    },
    {
      "epoch": 0.49,
      "learning_rate": 9.54957088530181e-05,
      "loss": 0.8571,
      "step": 527
    },
    {
      "epoch": 0.49,
      "learning_rate": 9.54750083349572e-05,
      "loss": 0.7992,
      "step": 528
    },
    {
      "epoch": 0.49,
      "learning_rate": 9.545426261301866e-05,
      "loss": 0.6739,
      "step": 529
    },
    {
      "epoch": 0.49,
      "learning_rate": 9.543347170782457e-05,
      "loss": 0.8096,
      "step": 530
    },
    {
      "epoch": 0.49,
      "learning_rate": 9.541263564004182e-05,
      "loss": 0.7678,
      "step": 531
    },
    {
      "epoch": 0.49,
      "learning_rate": 9.539175443038231e-05,
      "loss": 0.7127,
      "step": 532
    },
    {
      "epoch": 0.49,
      "learning_rate": 9.537082809960269e-05,
      "loss": 0.7429,
      "step": 533
    },
    {
      "epoch": 0.49,
      "learning_rate": 9.534985666850455e-05,
      "loss": 0.7314,
      "step": 534
    },
    {
      "epoch": 0.49,
      "learning_rate": 9.53288401579343e-05,
      "loss": 0.7705,
      "step": 535
    },
    {
      "epoch": 0.49,
      "learning_rate": 9.530777858878314e-05,
      "loss": 0.8897,
      "step": 536
    },
    {
      "epoch": 0.5,
      "learning_rate": 9.528667198198706e-05,
      "loss": 0.7304,
      "step": 537
    },
    {
      "epoch": 0.5,
      "learning_rate": 9.526552035852682e-05,
      "loss": 0.7143,
      "step": 538
    },
    {
      "epoch": 0.5,
      "learning_rate": 9.524432373942794e-05,
      "loss": 0.6726,
      "step": 539
    },
    {
      "epoch": 0.5,
      "learning_rate": 9.522308214576066e-05,
      "loss": 0.6989,
      "step": 540
    },
    {
      "epoch": 0.5,
      "learning_rate": 9.52017955986399e-05,
      "loss": 0.7255,
      "step": 541
    },
    {
      "epoch": 0.5,
      "learning_rate": 9.518046411922536e-05,
      "loss": 0.7899,
      "step": 542
    },
    {
      "epoch": 0.5,
      "learning_rate": 9.515908772872126e-05,
      "loss": 0.854,
      "step": 543
    },
    {
      "epoch": 0.5,
      "learning_rate": 9.513766644837661e-05,
      "loss": 0.7716,
      "step": 544
    },
    {
      "epoch": 0.5,
      "learning_rate": 9.511620029948491e-05,
      "loss": 0.7424,
      "step": 545
    },
    {
      "epoch": 0.5,
      "learning_rate": 9.509468930338437e-05,
      "loss": 0.7039,
      "step": 546
    },
    {
      "epoch": 0.5,
      "learning_rate": 9.507313348145771e-05,
      "loss": 0.6786,
      "step": 547
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.505153285513225e-05,
      "loss": 0.7283,
      "step": 548
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.502988744587979e-05,
      "loss": 0.749,
      "step": 549
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.500819727521673e-05,
      "loss": 0.8345,
      "step": 550
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.498646236470391e-05,
      "loss": 0.8253,
      "step": 551
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.496468273594663e-05,
      "loss": 0.8101,
      "step": 552
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.49428584105947e-05,
      "loss": 0.811,
      "step": 553
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.492098941034228e-05,
      "loss": 0.7192,
      "step": 554
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.489907575692801e-05,
      "loss": 0.7218,
      "step": 555
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.487711747213489e-05,
      "loss": 0.7933,
      "step": 556
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.485511457779029e-05,
      "loss": 0.7774,
      "step": 557
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.48330670957659e-05,
      "loss": 0.7631,
      "step": 558
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.481097504797778e-05,
      "loss": 0.8008,
      "step": 559
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.478883845638622e-05,
      "loss": 0.7514,
      "step": 560
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.476665734299587e-05,
      "loss": 0.7913,
      "step": 561
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.474443172985556e-05,
      "loss": 0.7367,
      "step": 562
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.472216163905843e-05,
      "loss": 0.9209,
      "step": 563
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.469984709274175e-05,
      "loss": 0.765,
      "step": 564
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.467748811308703e-05,
      "loss": 0.686,
      "step": 565
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.465508472231999e-05,
      "loss": 0.7763,
      "step": 566
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.463263694271038e-05,
      "loss": 0.7893,
      "step": 567
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.461014479657217e-05,
      "loss": 0.76,
      "step": 568
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.458760830626343e-05,
      "loss": 0.8255,
      "step": 569
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.456502749418625e-05,
      "loss": 0.8304,
      "step": 570
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.45424023827868e-05,
      "loss": 0.6895,
      "step": 571
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.451973299455535e-05,
      "loss": 0.8602,
      "step": 572
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.449701935202606e-05,
      "loss": 0.7552,
      "step": 573
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.44742614777772e-05,
      "loss": 0.8465,
      "step": 574
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.445145939443094e-05,
      "loss": 0.7853,
      "step": 575
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.442861312465342e-05,
      "loss": 0.6953,
      "step": 576
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.440572269115468e-05,
      "loss": 0.7563,
      "step": 577
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.43827881166887e-05,
      "loss": 0.8271,
      "step": 578
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.43598094240533e-05,
      "loss": 0.8189,
      "step": 579
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.433678663609017e-05,
      "loss": 0.7462,
      "step": 580
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.431371977568483e-05,
      "loss": 0.6822,
      "step": 581
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.429060886576661e-05,
      "loss": 0.722,
      "step": 582
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.426745392930862e-05,
      "loss": 0.77,
      "step": 583
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.424425498932779e-05,
      "loss": 0.7958,
      "step": 584
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.422101206888467e-05,
      "loss": 0.7781,
      "step": 585
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.419772519108366e-05,
      "loss": 0.7128,
      "step": 586
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.417439437907277e-05,
      "loss": 0.7529,
      "step": 587
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.415101965604372e-05,
      "loss": 0.7705,
      "step": 588
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.412760104523186e-05,
      "loss": 0.7387,
      "step": 589
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.410413856991618e-05,
      "loss": 0.7812,
      "step": 590
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.408063225341925e-05,
      "loss": 0.7532,
      "step": 591
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.405708211910729e-05,
      "loss": 0.8634,
      "step": 592
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.403348819038995e-05,
      "loss": 0.7424,
      "step": 593
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.400985049072055e-05,
      "loss": 0.8508,
      "step": 594
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.39861690435958e-05,
      "loss": 0.7567,
      "step": 595
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.396244387255601e-05,
      "loss": 0.8006,
      "step": 596
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.393867500118488e-05,
      "loss": 0.7491,
      "step": 597
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.391486245310955e-05,
      "loss": 0.7294,
      "step": 598
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.38910062520006e-05,
      "loss": 0.7661,
      "step": 599
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.3867106421572e-05,
      "loss": 0.7967,
      "step": 600
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.384316298558106e-05,
      "loss": 0.7076,
      "step": 601
    },
    {
      "epoch": 0.56,
      "learning_rate": 9.381917596782852e-05,
      "loss": 0.7469,
      "step": 602
    },
    {
      "epoch": 0.56,
      "learning_rate": 9.379514539215832e-05,
      "loss": 0.7391,
      "step": 603
    },
    {
      "epoch": 0.56,
      "learning_rate": 9.377107128245781e-05,
      "loss": 0.7592,
      "step": 604
    },
    {
      "epoch": 0.56,
      "learning_rate": 9.374695366265753e-05,
      "loss": 0.7834,
      "step": 605
    },
    {
      "epoch": 0.56,
      "learning_rate": 9.372279255673131e-05,
      "loss": 0.7512,
      "step": 606
    },
    {
      "epoch": 0.56,
      "learning_rate": 9.369858798869624e-05,
      "loss": 0.8458,
      "step": 607
    },
    {
      "epoch": 0.56,
      "learning_rate": 9.367433998261251e-05,
      "loss": 0.7464,
      "step": 608
    },
    {
      "epoch": 0.56,
      "learning_rate": 9.365004856258361e-05,
      "loss": 0.8004,
      "step": 609
    },
    {
      "epoch": 0.56,
      "learning_rate": 9.36257137527561e-05,
      "loss": 0.8022,
      "step": 610
    },
    {
      "epoch": 0.56,
      "learning_rate": 9.360133557731974e-05,
      "loss": 0.7929,
      "step": 611
    },
    {
      "epoch": 0.56,
      "learning_rate": 9.35769140605073e-05,
      "loss": 0.7267,
      "step": 612
    },
    {
      "epoch": 0.57,
      "learning_rate": 9.355244922659474e-05,
      "loss": 0.8429,
      "step": 613
    },
    {
      "epoch": 0.57,
      "learning_rate": 9.352794109990101e-05,
      "loss": 0.7628,
      "step": 614
    },
    {
      "epoch": 0.57,
      "learning_rate": 9.350338970478811e-05,
      "loss": 0.8081,
      "step": 615
    },
    {
      "epoch": 0.57,
      "learning_rate": 9.347879506566107e-05,
      "loss": 0.6582,
      "step": 616
    },
    {
      "epoch": 0.57,
      "learning_rate": 9.345415720696787e-05,
      "loss": 0.7136,
      "step": 617
    },
    {
      "epoch": 0.57,
      "learning_rate": 9.342947615319949e-05,
      "loss": 0.8122,
      "step": 618
    },
    {
      "epoch": 0.57,
      "learning_rate": 9.340475192888983e-05,
      "loss": 0.7689,
      "step": 619
    },
    {
      "epoch": 0.57,
      "learning_rate": 9.33799845586157e-05,
      "loss": 0.7049,
      "step": 620
    },
    {
      "epoch": 0.57,
      "learning_rate": 9.335517406699678e-05,
      "loss": 0.7175,
      "step": 621
    },
    {
      "epoch": 0.57,
      "learning_rate": 9.33303204786957e-05,
      "loss": 0.7389,
      "step": 622
    },
    {
      "epoch": 0.58,
      "learning_rate": 9.33054238184178e-05,
      "loss": 0.7718,
      "step": 623
    },
    {
      "epoch": 0.58,
      "learning_rate": 9.328048411091133e-05,
      "loss": 0.7743,
      "step": 624
    },
    {
      "epoch": 0.58,
      "learning_rate": 9.325550138096728e-05,
      "loss": 0.7641,
      "step": 625
    },
    {
      "epoch": 0.58,
      "learning_rate": 9.323047565341947e-05,
      "loss": 0.7585,
      "step": 626
    },
    {
      "epoch": 0.58,
      "learning_rate": 9.320540695314439e-05,
      "loss": 0.7197,
      "step": 627
    },
    {
      "epoch": 0.58,
      "learning_rate": 9.31802953050613e-05,
      "loss": 0.7304,
      "step": 628
    },
    {
      "epoch": 0.58,
      "learning_rate": 9.315514073413208e-05,
      "loss": 0.7802,
      "step": 629
    },
    {
      "epoch": 0.58,
      "learning_rate": 9.312994326536138e-05,
      "loss": 0.804,
      "step": 630
    },
    {
      "epoch": 0.58,
      "learning_rate": 9.31047029237964e-05,
      "loss": 0.7454,
      "step": 631
    },
    {
      "epoch": 0.58,
      "learning_rate": 9.3079419734527e-05,
      "loss": 0.7936,
      "step": 632
    },
    {
      "epoch": 0.58,
      "learning_rate": 9.305409372268563e-05,
      "loss": 0.707,
      "step": 633
    },
    {
      "epoch": 0.59,
      "learning_rate": 9.302872491344732e-05,
      "loss": 0.7421,
      "step": 634
    },
    {
      "epoch": 0.59,
      "learning_rate": 9.30033133320296e-05,
      "loss": 0.7916,
      "step": 635
    },
    {
      "epoch": 0.59,
      "learning_rate": 9.297785900369255e-05,
      "loss": 0.7716,
      "step": 636
    },
    {
      "epoch": 0.59,
      "learning_rate": 9.295236195373874e-05,
      "loss": 0.7937,
      "step": 637
    },
    {
      "epoch": 0.59,
      "learning_rate": 9.29268222075132e-05,
      "loss": 0.7311,
      "step": 638
    },
    {
      "epoch": 0.59,
      "learning_rate": 9.290123979040339e-05,
      "loss": 0.7741,
      "step": 639
    },
    {
      "epoch": 0.59,
      "learning_rate": 9.287561472783924e-05,
      "loss": 0.7814,
      "step": 640
    },
    {
      "epoch": 0.59,
      "learning_rate": 9.284994704529297e-05,
      "loss": 0.7485,
      "step": 641
    },
    {
      "epoch": 0.59,
      "learning_rate": 9.282423676827928e-05,
      "loss": 0.729,
      "step": 642
    },
    {
      "epoch": 0.59,
      "learning_rate": 9.279848392235512e-05,
      "loss": 0.7979,
      "step": 643
    },
    {
      "epoch": 0.59,
      "learning_rate": 9.277268853311984e-05,
      "loss": 0.7438,
      "step": 644
    },
    {
      "epoch": 0.6,
      "learning_rate": 9.274685062621498e-05,
      "loss": 0.7761,
      "step": 645
    },
    {
      "epoch": 0.6,
      "learning_rate": 9.272097022732443e-05,
      "loss": 0.7718,
      "step": 646
    },
    {
      "epoch": 0.6,
      "learning_rate": 9.269504736217427e-05,
      "loss": 0.8207,
      "step": 647
    },
    {
      "epoch": 0.6,
      "learning_rate": 9.266908205653281e-05,
      "loss": 0.8594,
      "step": 648
    },
    {
      "epoch": 0.6,
      "learning_rate": 9.264307433621054e-05,
      "loss": 0.8015,
      "step": 649
    },
    {
      "epoch": 0.6,
      "learning_rate": 9.261702422706013e-05,
      "loss": 0.785,
      "step": 650
    },
    {
      "epoch": 0.6,
      "learning_rate": 9.259093175497638e-05,
      "loss": 0.7703,
      "step": 651
    },
    {
      "epoch": 0.6,
      "learning_rate": 9.256479694589617e-05,
      "loss": 0.6738,
      "step": 652
    },
    {
      "epoch": 0.6,
      "learning_rate": 9.25386198257985e-05,
      "loss": 0.7179,
      "step": 653
    },
    {
      "epoch": 0.6,
      "learning_rate": 9.251240042070442e-05,
      "loss": 0.784,
      "step": 654
    },
    {
      "epoch": 0.6,
      "learning_rate": 9.2486138756677e-05,
      "loss": 0.7395,
      "step": 655
    },
    {
      "epoch": 0.61,
      "learning_rate": 9.245983485982133e-05,
      "loss": 0.7926,
      "step": 656
    },
    {
      "epoch": 0.61,
      "learning_rate": 9.243348875628451e-05,
      "loss": 0.7259,
      "step": 657
    },
    {
      "epoch": 0.61,
      "learning_rate": 9.24071004722555e-05,
      "loss": 0.7052,
      "step": 658
    },
    {
      "epoch": 0.61,
      "learning_rate": 9.238067003396532e-05,
      "loss": 0.7086,
      "step": 659
    },
    {
      "epoch": 0.61,
      "learning_rate": 9.235419746768676e-05,
      "loss": 0.7612,
      "step": 660
    },
    {
      "epoch": 0.61,
      "learning_rate": 9.232768279973461e-05,
      "loss": 0.6866,
      "step": 661
    },
    {
      "epoch": 0.61,
      "learning_rate": 9.230112605646539e-05,
      "loss": 0.7447,
      "step": 662
    },
    {
      "epoch": 0.61,
      "learning_rate": 9.227452726427757e-05,
      "loss": 0.7187,
      "step": 663
    },
    {
      "epoch": 0.61,
      "learning_rate": 9.224788644961133e-05,
      "loss": 0.7991,
      "step": 664
    },
    {
      "epoch": 0.61,
      "learning_rate": 9.222120363894864e-05,
      "loss": 0.7863,
      "step": 665
    },
    {
      "epoch": 0.61,
      "learning_rate": 9.219447885881323e-05,
      "loss": 0.7029,
      "step": 666
    },
    {
      "epoch": 0.62,
      "learning_rate": 9.216771213577053e-05,
      "loss": 0.7244,
      "step": 667
    },
    {
      "epoch": 0.62,
      "learning_rate": 9.214090349642769e-05,
      "loss": 0.6811,
      "step": 668
    },
    {
      "epoch": 0.62,
      "learning_rate": 9.21140529674335e-05,
      "loss": 0.722,
      "step": 669
    },
    {
      "epoch": 0.62,
      "learning_rate": 9.208716057547841e-05,
      "loss": 0.6527,
      "step": 670
    },
    {
      "epoch": 0.62,
      "learning_rate": 9.206022634729447e-05,
      "loss": 0.665,
      "step": 671
    },
    {
      "epoch": 0.62,
      "learning_rate": 9.203325030965533e-05,
      "loss": 0.6965,
      "step": 672
    },
    {
      "epoch": 0.62,
      "learning_rate": 9.20062324893762e-05,
      "loss": 0.7179,
      "step": 673
    },
    {
      "epoch": 0.62,
      "learning_rate": 9.197917291331375e-05,
      "loss": 0.7414,
      "step": 674
    },
    {
      "epoch": 0.62,
      "learning_rate": 9.19520716083663e-05,
      "loss": 0.7441,
      "step": 675
    },
    {
      "epoch": 0.62,
      "learning_rate": 9.192492860147353e-05,
      "loss": 0.7812,
      "step": 676
    },
    {
      "epoch": 0.62,
      "learning_rate": 9.189774391961663e-05,
      "loss": 0.7208,
      "step": 677
    },
    {
      "epoch": 0.63,
      "learning_rate": 9.18705175898182e-05,
      "loss": 0.7794,
      "step": 678
    },
    {
      "epoch": 0.63,
      "learning_rate": 9.184324963914225e-05,
      "loss": 0.7924,
      "step": 679
    },
    {
      "epoch": 0.63,
      "learning_rate": 9.181594009469414e-05,
      "loss": 0.8082,
      "step": 680
    },
    {
      "epoch": 0.63,
      "learning_rate": 9.178858898362057e-05,
      "loss": 0.8029,
      "step": 681
    },
    {
      "epoch": 0.63,
      "learning_rate": 9.176119633310962e-05,
      "loss": 0.7576,
      "step": 682
    },
    {
      "epoch": 0.63,
      "learning_rate": 9.17337621703906e-05,
      "loss": 0.786,
      "step": 683
    },
    {
      "epoch": 0.63,
      "learning_rate": 9.170628652273409e-05,
      "loss": 0.7728,
      "step": 684
    },
    {
      "epoch": 0.63,
      "learning_rate": 9.167876941745196e-05,
      "loss": 0.7733,
      "step": 685
    },
    {
      "epoch": 0.63,
      "learning_rate": 9.165121088189722e-05,
      "loss": 0.7388,
      "step": 686
    },
    {
      "epoch": 0.63,
      "learning_rate": 9.162361094346413e-05,
      "loss": 0.8212,
      "step": 687
    },
    {
      "epoch": 0.64,
      "learning_rate": 9.159596962958803e-05,
      "loss": 0.8539,
      "step": 688
    },
    {
      "epoch": 0.64,
      "learning_rate": 9.156828696774547e-05,
      "loss": 0.7832,
      "step": 689
    },
    {
      "epoch": 0.64,
      "learning_rate": 9.154056298545403e-05,
      "loss": 0.7833,
      "step": 690
    },
    {
      "epoch": 0.64,
      "learning_rate": 9.151279771027242e-05,
      "loss": 0.7392,
      "step": 691
    },
    {
      "epoch": 0.64,
      "learning_rate": 9.148499116980034e-05,
      "loss": 0.8068,
      "step": 692
    },
    {
      "epoch": 0.64,
      "learning_rate": 9.145714339167859e-05,
      "loss": 0.7761,
      "step": 693
    },
    {
      "epoch": 0.64,
      "learning_rate": 9.142925440358889e-05,
      "loss": 0.8031,
      "step": 694
    },
    {
      "epoch": 0.64,
      "learning_rate": 9.140132423325393e-05,
      "loss": 0.7785,
      "step": 695
    },
    {
      "epoch": 0.64,
      "learning_rate": 9.13733529084374e-05,
      "loss": 0.7673,
      "step": 696
    },
    {
      "epoch": 0.64,
      "learning_rate": 9.134534045694382e-05,
      "loss": 0.7359,
      "step": 697
    },
    {
      "epoch": 0.64,
      "learning_rate": 9.131728690661863e-05,
      "loss": 0.7609,
      "step": 698
    },
    {
      "epoch": 0.65,
      "learning_rate": 9.128919228534814e-05,
      "loss": 0.7703,
      "step": 699
    },
    {
      "epoch": 0.65,
      "learning_rate": 9.126105662105942e-05,
      "loss": 0.7415,
      "step": 700
    },
    {
      "epoch": 0.65,
      "learning_rate": 9.123287994172045e-05,
      "loss": 0.7497,
      "step": 701
    },
    {
      "epoch": 0.65,
      "learning_rate": 9.120466227533986e-05,
      "loss": 0.7732,
      "step": 702
    },
    {
      "epoch": 0.65,
      "learning_rate": 9.117640364996708e-05,
      "loss": 0.7813,
      "step": 703
    },
    {
      "epoch": 0.65,
      "learning_rate": 9.11481040936923e-05,
      "loss": 0.7253,
      "step": 704
    },
    {
      "epoch": 0.65,
      "learning_rate": 9.111976363464631e-05,
      "loss": 0.8028,
      "step": 705
    },
    {
      "epoch": 0.65,
      "learning_rate": 9.10913823010006e-05,
      "loss": 0.7449,
      "step": 706
    },
    {
      "epoch": 0.65,
      "learning_rate": 9.10629601209673e-05,
      "loss": 0.7025,
      "step": 707
    },
    {
      "epoch": 0.65,
      "learning_rate": 9.103449712279913e-05,
      "loss": 0.8306,
      "step": 708
    },
    {
      "epoch": 0.65,
      "learning_rate": 9.100599333478937e-05,
      "loss": 0.6522,
      "step": 709
    },
    {
      "epoch": 0.66,
      "learning_rate": 9.097744878527186e-05,
      "loss": 0.7668,
      "step": 710
    },
    {
      "epoch": 0.66,
      "learning_rate": 9.0948863502621e-05,
      "loss": 0.7307,
      "step": 711
    },
    {
      "epoch": 0.66,
      "learning_rate": 9.092023751525161e-05,
      "loss": 0.7509,
      "step": 712
    },
    {
      "epoch": 0.66,
      "learning_rate": 9.089157085161902e-05,
      "loss": 0.6773,
      "step": 713
    },
    {
      "epoch": 0.66,
      "learning_rate": 9.086286354021896e-05,
      "loss": 0.6925,
      "step": 714
    },
    {
      "epoch": 0.66,
      "learning_rate": 9.08341156095876e-05,
      "loss": 0.7658,
      "step": 715
    },
    {
      "epoch": 0.66,
      "learning_rate": 9.080532708830145e-05,
      "loss": 0.759,
      "step": 716
    },
    {
      "epoch": 0.66,
      "learning_rate": 9.077649800497741e-05,
      "loss": 0.7774,
      "step": 717
    },
    {
      "epoch": 0.66,
      "learning_rate": 9.074762838827267e-05,
      "loss": 0.7763,
      "step": 718
    },
    {
      "epoch": 0.66,
      "learning_rate": 9.071871826688471e-05,
      "loss": 0.7842,
      "step": 719
    },
    {
      "epoch": 0.66,
      "learning_rate": 9.068976766955131e-05,
      "loss": 0.8055,
      "step": 720
    },
    {
      "epoch": 0.67,
      "learning_rate": 9.066077662505043e-05,
      "loss": 0.7032,
      "step": 721
    },
    {
      "epoch": 0.67,
      "learning_rate": 9.063174516220028e-05,
      "loss": 0.7473,
      "step": 722
    },
    {
      "epoch": 0.67,
      "learning_rate": 9.060267330985923e-05,
      "loss": 0.659,
      "step": 723
    },
    {
      "epoch": 0.67,
      "learning_rate": 9.057356109692579e-05,
      "loss": 0.8438,
      "step": 724
    },
    {
      "epoch": 0.67,
      "learning_rate": 9.054440855233861e-05,
      "loss": 0.7924,
      "step": 725
    },
    {
      "epoch": 0.67,
      "learning_rate": 9.051521570507642e-05,
      "loss": 0.7893,
      "step": 726
    },
    {
      "epoch": 0.67,
      "learning_rate": 9.048598258415801e-05,
      "loss": 0.8331,
      "step": 727
    },
    {
      "epoch": 0.67,
      "learning_rate": 9.045670921864222e-05,
      "loss": 0.7238,
      "step": 728
    },
    {
      "epoch": 0.67,
      "learning_rate": 9.042739563762788e-05,
      "loss": 0.7218,
      "step": 729
    },
    {
      "epoch": 0.67,
      "learning_rate": 9.039804187025378e-05,
      "loss": 0.7101,
      "step": 730
    },
    {
      "epoch": 0.67,
      "learning_rate": 9.036864794569871e-05,
      "loss": 0.7545,
      "step": 731
    },
    {
      "epoch": 0.68,
      "learning_rate": 9.033921389318129e-05,
      "loss": 0.712,
      "step": 732
    },
    {
      "epoch": 0.68,
      "learning_rate": 9.03097397419601e-05,
      "loss": 0.7353,
      "step": 733
    },
    {
      "epoch": 0.68,
      "learning_rate": 9.028022552133359e-05,
      "loss": 0.642,
      "step": 734
    },
    {
      "epoch": 0.68,
      "learning_rate": 9.025067126064e-05,
      "loss": 0.7361,
      "step": 735
    },
    {
      "epoch": 0.68,
      "learning_rate": 9.022107698925735e-05,
      "loss": 0.774,
      "step": 736
    },
    {
      "epoch": 0.68,
      "learning_rate": 9.019144273660351e-05,
      "loss": 0.7763,
      "step": 737
    },
    {
      "epoch": 0.68,
      "learning_rate": 9.016176853213602e-05,
      "loss": 0.7466,
      "step": 738
    },
    {
      "epoch": 0.68,
      "learning_rate": 9.013205440535217e-05,
      "loss": 0.7748,
      "step": 739
    },
    {
      "epoch": 0.68,
      "learning_rate": 9.01023003857889e-05,
      "loss": 0.736,
      "step": 740
    },
    {
      "epoch": 0.68,
      "learning_rate": 9.007250650302287e-05,
      "loss": 0.784,
      "step": 741
    },
    {
      "epoch": 0.68,
      "learning_rate": 9.004267278667032e-05,
      "loss": 0.6835,
      "step": 742
    },
    {
      "epoch": 0.69,
      "learning_rate": 9.001279926638707e-05,
      "loss": 0.7465,
      "step": 743
    },
    {
      "epoch": 0.69,
      "learning_rate": 8.998288597186855e-05,
      "loss": 0.7459,
      "step": 744
    },
    {
      "epoch": 0.69,
      "learning_rate": 8.995293293284968e-05,
      "loss": 0.7883,
      "step": 745
    },
    {
      "epoch": 0.69,
      "learning_rate": 8.992294017910494e-05,
      "loss": 0.7669,
      "step": 746
    },
    {
      "epoch": 0.69,
      "learning_rate": 8.989290774044822e-05,
      "loss": 0.9103,
      "step": 747
    },
    {
      "epoch": 0.69,
      "learning_rate": 8.986283564673296e-05,
      "loss": 0.7368,
      "step": 748
    },
    {
      "epoch": 0.69,
      "learning_rate": 8.983272392785194e-05,
      "loss": 0.789,
      "step": 749
    },
    {
      "epoch": 0.69,
      "learning_rate": 8.98025726137373e-05,
      "loss": 0.7775,
      "step": 750
    },
    {
      "epoch": 0.69,
      "learning_rate": 8.977238173436064e-05,
      "loss": 0.7532,
      "step": 751
    },
    {
      "epoch": 0.69,
      "learning_rate": 8.974215131973281e-05,
      "loss": 0.7509,
      "step": 752
    },
    {
      "epoch": 0.7,
      "learning_rate": 8.971188139990399e-05,
      "loss": 0.7278,
      "step": 753
    },
    {
      "epoch": 0.7,
      "learning_rate": 8.968157200496363e-05,
      "loss": 0.7677,
      "step": 754
    },
    {
      "epoch": 0.7,
      "learning_rate": 8.965122316504042e-05,
      "loss": 0.7581,
      "step": 755
    },
    {
      "epoch": 0.7,
      "learning_rate": 8.962083491030222e-05,
      "loss": 0.7784,
      "step": 756
    },
    {
      "epoch": 0.7,
      "learning_rate": 8.959040727095615e-05,
      "loss": 0.723,
      "step": 757
    },
    {
      "epoch": 0.7,
      "learning_rate": 8.95599402772484e-05,
      "loss": 0.7121,
      "step": 758
    },
    {
      "epoch": 0.7,
      "learning_rate": 8.952943395946434e-05,
      "loss": 0.7357,
      "step": 759
    },
    {
      "epoch": 0.7,
      "learning_rate": 8.949888834792838e-05,
      "loss": 0.7926,
      "step": 760
    },
    {
      "epoch": 0.7,
      "learning_rate": 8.946830347300403e-05,
      "loss": 0.6883,
      "step": 761
    },
    {
      "epoch": 0.7,
      "learning_rate": 8.94376793650938e-05,
      "loss": 0.8138,
      "step": 762
    },
    {
      "epoch": 0.7,
      "learning_rate": 8.94070160546392e-05,
      "loss": 0.7476,
      "step": 763
    },
    {
      "epoch": 0.71,
      "learning_rate": 8.937631357212075e-05,
      "loss": 0.7372,
      "step": 764
    },
    {
      "epoch": 0.71,
      "learning_rate": 8.934557194805788e-05,
      "loss": 0.7099,
      "step": 765
    },
    {
      "epoch": 0.71,
      "learning_rate": 8.931479121300888e-05,
      "loss": 0.7471,
      "step": 766
    },
    {
      "epoch": 0.71,
      "learning_rate": 8.9283971397571e-05,
      "loss": 0.8266,
      "step": 767
    },
    {
      "epoch": 0.71,
      "learning_rate": 8.925311253238033e-05,
      "loss": 0.7141,
      "step": 768
    },
    {
      "epoch": 0.71,
      "learning_rate": 8.922221464811168e-05,
      "loss": 0.7603,
      "step": 769
    },
    {
      "epoch": 0.71,
      "learning_rate": 8.919127777547876e-05,
      "loss": 0.8253,
      "step": 770
    },
    {
      "epoch": 0.71,
      "learning_rate": 8.916030194523399e-05,
      "loss": 0.7468,
      "step": 771
    },
    {
      "epoch": 0.71,
      "learning_rate": 8.91292871881685e-05,
      "loss": 0.748,
      "step": 772
    },
    {
      "epoch": 0.71,
      "learning_rate": 8.909823353511215e-05,
      "loss": 0.7932,
      "step": 773
    },
    {
      "epoch": 0.71,
      "learning_rate": 8.906714101693343e-05,
      "loss": 0.7479,
      "step": 774
    },
    {
      "epoch": 0.72,
      "learning_rate": 8.903600966453948e-05,
      "loss": 0.7863,
      "step": 775
    },
    {
      "epoch": 0.72,
      "learning_rate": 8.900483950887603e-05,
      "loss": 0.7869,
      "step": 776
    },
    {
      "epoch": 0.72,
      "learning_rate": 8.897363058092743e-05,
      "loss": 0.7654,
      "step": 777
    },
    {
      "epoch": 0.72,
      "learning_rate": 8.89423829117165e-05,
      "loss": 0.7471,
      "step": 778
    },
    {
      "epoch": 0.72,
      "learning_rate": 8.891109653230463e-05,
      "loss": 0.8054,
      "step": 779
    },
    {
      "epoch": 0.72,
      "learning_rate": 8.887977147379163e-05,
      "loss": 0.7826,
      "step": 780
    },
    {
      "epoch": 0.72,
      "learning_rate": 8.884840776731583e-05,
      "loss": 0.792,
      "step": 781
    },
    {
      "epoch": 0.72,
      "learning_rate": 8.881700544405391e-05,
      "loss": 0.7965,
      "step": 782
    },
    {
      "epoch": 0.72,
      "learning_rate": 8.8785564535221e-05,
      "loss": 0.7135,
      "step": 783
    },
    {
      "epoch": 0.72,
      "learning_rate": 8.875408507207052e-05,
      "loss": 0.7067,
      "step": 784
    },
    {
      "epoch": 0.72,
      "learning_rate": 8.872256708589426e-05,
      "loss": 0.7372,
      "step": 785
    },
    {
      "epoch": 0.73,
      "learning_rate": 8.869101060802227e-05,
      "loss": 0.7643,
      "step": 786
    },
    {
      "epoch": 0.73,
      "learning_rate": 8.865941566982292e-05,
      "loss": 0.736,
      "step": 787
    },
    {
      "epoch": 0.73,
      "learning_rate": 8.862778230270275e-05,
      "loss": 0.8422,
      "step": 788
    },
    {
      "epoch": 0.73,
      "learning_rate": 8.859611053810653e-05,
      "loss": 0.662,
      "step": 789
    },
    {
      "epoch": 0.73,
      "learning_rate": 8.856440040751719e-05,
      "loss": 0.8046,
      "step": 790
    },
    {
      "epoch": 0.73,
      "learning_rate": 8.853265194245579e-05,
      "loss": 0.7361,
      "step": 791
    },
    {
      "epoch": 0.73,
      "learning_rate": 8.85008651744815e-05,
      "loss": 0.8587,
      "step": 792
    },
    {
      "epoch": 0.73,
      "learning_rate": 8.84690401351916e-05,
      "loss": 0.6971,
      "step": 793
    },
    {
      "epoch": 0.73,
      "learning_rate": 8.843717685622135e-05,
      "loss": 0.8031,
      "step": 794
    },
    {
      "epoch": 0.73,
      "learning_rate": 8.840527536924408e-05,
      "loss": 0.772,
      "step": 795
    },
    {
      "epoch": 0.73,
      "learning_rate": 8.837333570597104e-05,
      "loss": 0.7309,
      "step": 796
    },
    {
      "epoch": 0.74,
      "learning_rate": 8.83413578981515e-05,
      "loss": 0.7328,
      "step": 797
    },
    {
      "epoch": 0.74,
      "learning_rate": 8.830934197757259e-05,
      "loss": 0.788,
      "step": 798
    },
    {
      "epoch": 0.74,
      "learning_rate": 8.827728797605934e-05,
      "loss": 0.7834,
      "step": 799
    },
    {
      "epoch": 0.74,
      "learning_rate": 8.824519592547466e-05,
      "loss": 0.752,
      "step": 800
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 3249,
  "num_train_epochs": 3,
  "save_steps": 100,
  "total_flos": 4.0704187821981696e+18,
  "trial_name": null,
  "trial_params": null
}