{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.31978680879413723,
  "global_step": 150,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 5.681818181818182e-08,
      "loss": 1.4294,
      "step": 1
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.1363636363636364e-07,
      "loss": 1.3293,
      "step": 2
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.7045454545454545e-07,
      "loss": 1.0797,
      "step": 3
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.2727272727272729e-07,
      "loss": 0.9853,
      "step": 4
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.840909090909091e-07,
      "loss": 0.983,
      "step": 5
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.409090909090909e-07,
      "loss": 0.9868,
      "step": 6
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.9772727272727276e-07,
      "loss": 1.0148,
      "step": 7
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.5454545454545457e-07,
      "loss": 1.0126,
      "step": 8
    },
    {
      "epoch": 0.02,
      "learning_rate": 5.113636363636364e-07,
      "loss": 0.9555,
      "step": 9
    },
    {
      "epoch": 0.02,
      "learning_rate": 5.681818181818182e-07,
      "loss": 1.0721,
      "step": 10
    },
    {
      "epoch": 0.02,
      "learning_rate": 6.25e-07,
      "loss": 1.0352,
      "step": 11
    },
    {
      "epoch": 0.03,
      "learning_rate": 6.818181818181818e-07,
      "loss": 0.9382,
      "step": 12
    },
    {
      "epoch": 0.03,
      "learning_rate": 7.386363636363638e-07,
      "loss": 1.0059,
      "step": 13
    },
    {
      "epoch": 0.03,
      "learning_rate": 7.954545454545455e-07,
      "loss": 0.9997,
      "step": 14
    },
    {
      "epoch": 0.03,
      "learning_rate": 8.522727272727273e-07,
      "loss": 0.9663,
      "step": 15
    },
    {
      "epoch": 0.03,
      "learning_rate": 9.090909090909091e-07,
      "loss": 0.9601,
      "step": 16
    },
    {
      "epoch": 0.04,
      "learning_rate": 9.65909090909091e-07,
      "loss": 0.9271,
      "step": 17
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.0227272727272729e-06,
      "loss": 0.9655,
      "step": 18
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.0795454545454546e-06,
      "loss": 0.8991,
      "step": 19
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.1363636363636364e-06,
      "loss": 0.9081,
      "step": 20
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.1931818181818183e-06,
      "loss": 0.9127,
      "step": 21
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.25e-06,
      "loss": 0.9912,
      "step": 22
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.3068181818181819e-06,
      "loss": 0.9077,
      "step": 23
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.3636363636363636e-06,
      "loss": 0.9739,
      "step": 24
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.4204545454545458e-06,
      "loss": 0.9148,
      "step": 25
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.4772727272727275e-06,
      "loss": 0.8668,
      "step": 26
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.5340909090909093e-06,
      "loss": 0.8951,
      "step": 27
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.590909090909091e-06,
      "loss": 0.842,
      "step": 28
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.6477272727272728e-06,
      "loss": 0.8439,
      "step": 29
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.7045454545454546e-06,
      "loss": 0.8471,
      "step": 30
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.7613636363636365e-06,
      "loss": 0.9474,
      "step": 31
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.8181818181818183e-06,
      "loss": 0.8957,
      "step": 32
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.8750000000000003e-06,
      "loss": 0.8596,
      "step": 33
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.931818181818182e-06,
      "loss": 0.8183,
      "step": 34
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.9886363636363638e-06,
      "loss": 0.9198,
      "step": 35
    },
    {
      "epoch": 0.08,
      "learning_rate": 2.0454545454545457e-06,
      "loss": 0.8402,
      "step": 36
    },
    {
      "epoch": 0.08,
      "learning_rate": 2.1022727272727277e-06,
      "loss": 0.9209,
      "step": 37
    },
    {
      "epoch": 0.08,
      "learning_rate": 2.1590909090909092e-06,
      "loss": 0.8554,
      "step": 38
    },
    {
      "epoch": 0.08,
      "learning_rate": 2.2159090909090912e-06,
      "loss": 0.9276,
      "step": 39
    },
    {
      "epoch": 0.09,
      "learning_rate": 2.2727272727272728e-06,
      "loss": 0.8872,
      "step": 40
    },
    {
      "epoch": 0.09,
      "learning_rate": 2.3295454545454547e-06,
      "loss": 0.9047,
      "step": 41
    },
    {
      "epoch": 0.09,
      "learning_rate": 2.3863636363636367e-06,
      "loss": 0.8514,
      "step": 42
    },
    {
      "epoch": 0.09,
      "learning_rate": 2.4431818181818182e-06,
      "loss": 0.7976,
      "step": 43
    },
    {
      "epoch": 0.09,
      "learning_rate": 2.5e-06,
      "loss": 0.8126,
      "step": 44
    },
    {
      "epoch": 0.1,
      "learning_rate": 2.556818181818182e-06,
      "loss": 0.8536,
      "step": 45
    },
    {
      "epoch": 0.1,
      "learning_rate": 2.6136363636363637e-06,
      "loss": 0.8228,
      "step": 46
    },
    {
      "epoch": 0.1,
      "learning_rate": 2.6704545454545457e-06,
      "loss": 0.8439,
      "step": 47
    },
    {
      "epoch": 0.1,
      "learning_rate": 2.7272727272727272e-06,
      "loss": 0.7859,
      "step": 48
    },
    {
      "epoch": 0.1,
      "learning_rate": 2.784090909090909e-06,
      "loss": 0.8424,
      "step": 49
    },
    {
      "epoch": 0.11,
      "learning_rate": 2.8409090909090916e-06,
      "loss": 0.7674,
      "step": 50
    },
    {
      "epoch": 0.11,
      "learning_rate": 2.897727272727273e-06,
      "loss": 0.8022,
      "step": 51
    },
    {
      "epoch": 0.11,
      "learning_rate": 2.954545454545455e-06,
      "loss": 0.7343,
      "step": 52
    },
    {
      "epoch": 0.11,
      "learning_rate": 3.0113636363636366e-06,
      "loss": 0.7343,
      "step": 53
    },
    {
      "epoch": 0.12,
      "learning_rate": 3.0681818181818186e-06,
      "loss": 0.7848,
      "step": 54
    },
    {
      "epoch": 0.12,
      "learning_rate": 3.125e-06,
      "loss": 0.8391,
      "step": 55
    },
    {
      "epoch": 0.12,
      "learning_rate": 3.181818181818182e-06,
      "loss": 0.7934,
      "step": 56
    },
    {
      "epoch": 0.12,
      "learning_rate": 3.2386363636363637e-06,
      "loss": 0.774,
      "step": 57
    },
    {
      "epoch": 0.12,
      "learning_rate": 3.2954545454545456e-06,
      "loss": 0.7292,
      "step": 58
    },
    {
      "epoch": 0.13,
      "learning_rate": 3.352272727272727e-06,
      "loss": 0.7393,
      "step": 59
    },
    {
      "epoch": 0.13,
      "learning_rate": 3.409090909090909e-06,
      "loss": 0.7484,
      "step": 60
    },
    {
      "epoch": 0.13,
      "learning_rate": 3.4659090909090915e-06,
      "loss": 0.7077,
      "step": 61
    },
    {
      "epoch": 0.13,
      "learning_rate": 3.522727272727273e-06,
      "loss": 0.8003,
      "step": 62
    },
    {
      "epoch": 0.13,
      "learning_rate": 3.579545454545455e-06,
      "loss": 0.7998,
      "step": 63
    },
    {
      "epoch": 0.14,
      "learning_rate": 3.6363636363636366e-06,
      "loss": 0.7882,
      "step": 64
    },
    {
      "epoch": 0.14,
      "learning_rate": 3.6931818181818186e-06,
      "loss": 0.7326,
      "step": 65
    },
    {
      "epoch": 0.14,
      "learning_rate": 3.7500000000000005e-06,
      "loss": 0.7945,
      "step": 66
    },
    {
      "epoch": 0.14,
      "learning_rate": 3.806818181818182e-06,
      "loss": 0.7518,
      "step": 67
    },
    {
      "epoch": 0.14,
      "learning_rate": 3.863636363636364e-06,
      "loss": 0.711,
      "step": 68
    },
    {
      "epoch": 0.15,
      "learning_rate": 3.9204545454545456e-06,
      "loss": 0.7393,
      "step": 69
    },
    {
      "epoch": 0.15,
      "learning_rate": 3.9772727272727275e-06,
      "loss": 0.7285,
      "step": 70
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.0340909090909095e-06,
      "loss": 0.6976,
      "step": 71
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.0909090909090915e-06,
      "loss": 0.7845,
      "step": 72
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.1477272727272734e-06,
      "loss": 0.7104,
      "step": 73
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.204545454545455e-06,
      "loss": 0.7021,
      "step": 74
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.2613636363636365e-06,
      "loss": 0.7425,
      "step": 75
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.3181818181818185e-06,
      "loss": 0.6481,
      "step": 76
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.3750000000000005e-06,
      "loss": 0.7236,
      "step": 77
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.4318181818181824e-06,
      "loss": 0.6932,
      "step": 78
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.4886363636363636e-06,
      "loss": 0.805,
      "step": 79
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.5454545454545455e-06,
      "loss": 0.7581,
      "step": 80
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.6022727272727275e-06,
      "loss": 0.7845,
      "step": 81
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.6590909090909095e-06,
      "loss": 0.722,
      "step": 82
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.715909090909091e-06,
      "loss": 0.7098,
      "step": 83
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.772727272727273e-06,
      "loss": 0.6372,
      "step": 84
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.829545454545455e-06,
      "loss": 0.6828,
      "step": 85
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.8863636363636365e-06,
      "loss": 0.655,
      "step": 86
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.9431818181818184e-06,
      "loss": 0.7154,
      "step": 87
    },
    {
      "epoch": 0.19,
      "learning_rate": 5e-06,
      "loss": 0.7593,
      "step": 88
    },
    {
      "epoch": 0.19,
      "learning_rate": 5.056818181818182e-06,
      "loss": 0.7319,
      "step": 89
    },
    {
      "epoch": 0.19,
      "learning_rate": 5.113636363636364e-06,
      "loss": 0.7033,
      "step": 90
    },
    {
      "epoch": 0.19,
      "learning_rate": 5.170454545454546e-06,
      "loss": 0.6835,
      "step": 91
    },
    {
      "epoch": 0.2,
      "learning_rate": 5.2272727272727274e-06,
      "loss": 0.7718,
      "step": 92
    },
    {
      "epoch": 0.2,
      "learning_rate": 5.28409090909091e-06,
      "loss": 0.7964,
      "step": 93
    },
    {
      "epoch": 0.2,
      "learning_rate": 5.340909090909091e-06,
      "loss": 0.727,
      "step": 94
    },
    {
      "epoch": 0.2,
      "learning_rate": 5.397727272727273e-06,
      "loss": 0.7526,
      "step": 95
    },
    {
      "epoch": 0.2,
      "learning_rate": 5.4545454545454545e-06,
      "loss": 0.7241,
      "step": 96
    },
    {
      "epoch": 0.21,
      "learning_rate": 5.511363636363637e-06,
      "loss": 0.7246,
      "step": 97
    },
    {
      "epoch": 0.21,
      "learning_rate": 5.568181818181818e-06,
      "loss": 0.7095,
      "step": 98
    },
    {
      "epoch": 0.21,
      "learning_rate": 5.625e-06,
      "loss": 0.6755,
      "step": 99
    },
    {
      "epoch": 0.21,
      "learning_rate": 5.681818181818183e-06,
      "loss": 0.6322,
      "step": 100
    },
    {
      "epoch": 0.22,
      "learning_rate": 5.738636363636364e-06,
      "loss": 0.7094,
      "step": 101
    },
    {
      "epoch": 0.22,
      "learning_rate": 5.795454545454546e-06,
      "loss": 0.7256,
      "step": 102
    },
    {
      "epoch": 0.22,
      "learning_rate": 5.852272727272727e-06,
      "loss": 0.7096,
      "step": 103
    },
    {
      "epoch": 0.22,
      "learning_rate": 5.90909090909091e-06,
      "loss": 0.7148,
      "step": 104
    },
    {
      "epoch": 0.22,
      "learning_rate": 5.965909090909091e-06,
      "loss": 0.6763,
      "step": 105
    },
    {
      "epoch": 0.23,
      "learning_rate": 6.022727272727273e-06,
      "loss": 0.6761,
      "step": 106
    },
    {
      "epoch": 0.23,
      "learning_rate": 6.079545454545454e-06,
      "loss": 0.7454,
      "step": 107
    },
    {
      "epoch": 0.23,
      "learning_rate": 6.136363636363637e-06,
      "loss": 0.694,
      "step": 108
    },
    {
      "epoch": 0.23,
      "learning_rate": 6.193181818181818e-06,
      "loss": 0.6797,
      "step": 109
    },
    {
      "epoch": 0.23,
      "learning_rate": 6.25e-06,
      "loss": 0.6294,
      "step": 110
    },
    {
      "epoch": 0.24,
      "learning_rate": 6.306818181818183e-06,
      "loss": 0.709,
      "step": 111
    },
    {
      "epoch": 0.24,
      "learning_rate": 6.363636363636364e-06,
      "loss": 0.6536,
      "step": 112
    },
    {
      "epoch": 0.24,
      "learning_rate": 6.420454545454546e-06,
      "loss": 0.7591,
      "step": 113
    },
    {
      "epoch": 0.24,
      "learning_rate": 6.477272727272727e-06,
      "loss": 0.6292,
      "step": 114
    },
    {
      "epoch": 0.25,
      "learning_rate": 6.53409090909091e-06,
      "loss": 0.7554,
      "step": 115
    },
    {
      "epoch": 0.25,
      "learning_rate": 6.590909090909091e-06,
      "loss": 0.6496,
      "step": 116
    },
    {
      "epoch": 0.25,
      "learning_rate": 6.647727272727273e-06,
      "loss": 0.7014,
      "step": 117
    },
    {
      "epoch": 0.25,
      "learning_rate": 6.704545454545454e-06,
      "loss": 0.6783,
      "step": 118
    },
    {
      "epoch": 0.25,
      "learning_rate": 6.761363636363637e-06,
      "loss": 0.6998,
      "step": 119
    },
    {
      "epoch": 0.26,
      "learning_rate": 6.818181818181818e-06,
      "loss": 0.6159,
      "step": 120
    },
    {
      "epoch": 0.26,
      "learning_rate": 6.875e-06,
      "loss": 0.7181,
      "step": 121
    },
    {
      "epoch": 0.26,
      "learning_rate": 6.931818181818183e-06,
      "loss": 0.7008,
      "step": 122
    },
    {
      "epoch": 0.26,
      "learning_rate": 6.988636363636364e-06,
      "loss": 0.6097,
      "step": 123
    },
    {
      "epoch": 0.26,
      "learning_rate": 7.045454545454546e-06,
      "loss": 0.7091,
      "step": 124
    },
    {
      "epoch": 0.27,
      "learning_rate": 7.102272727272727e-06,
      "loss": 0.6313,
      "step": 125
    },
    {
      "epoch": 0.27,
      "learning_rate": 7.15909090909091e-06,
      "loss": 0.6355,
      "step": 126
    },
    {
      "epoch": 0.27,
      "learning_rate": 7.215909090909091e-06,
      "loss": 0.6559,
      "step": 127
    },
    {
      "epoch": 0.27,
      "learning_rate": 7.272727272727273e-06,
      "loss": 0.6066,
      "step": 128
    },
    {
      "epoch": 0.28,
      "learning_rate": 7.329545454545455e-06,
      "loss": 0.6655,
      "step": 129
    },
    {
      "epoch": 0.28,
      "learning_rate": 7.386363636363637e-06,
      "loss": 0.6623,
      "step": 130
    },
    {
      "epoch": 0.28,
      "learning_rate": 7.443181818181818e-06,
      "loss": 0.6746,
      "step": 131
    },
    {
      "epoch": 0.28,
      "learning_rate": 7.500000000000001e-06,
      "loss": 0.6795,
      "step": 132
    },
    {
      "epoch": 0.28,
      "learning_rate": 7.556818181818183e-06,
      "loss": 0.6885,
      "step": 133
    },
    {
      "epoch": 0.29,
      "learning_rate": 7.613636363636364e-06,
      "loss": 0.6868,
      "step": 134
    },
    {
      "epoch": 0.29,
      "learning_rate": 7.670454545454547e-06,
      "loss": 0.6539,
      "step": 135
    },
    {
      "epoch": 0.29,
      "learning_rate": 7.727272727272727e-06,
      "loss": 0.6463,
      "step": 136
    },
    {
      "epoch": 0.29,
      "learning_rate": 7.784090909090911e-06,
      "loss": 0.6706,
      "step": 137
    },
    {
      "epoch": 0.29,
      "learning_rate": 7.840909090909091e-06,
      "loss": 0.6806,
      "step": 138
    },
    {
      "epoch": 0.3,
      "learning_rate": 7.897727272727273e-06,
      "loss": 0.6643,
      "step": 139
    },
    {
      "epoch": 0.3,
      "learning_rate": 7.954545454545455e-06,
      "loss": 0.6554,
      "step": 140
    },
    {
      "epoch": 0.3,
      "learning_rate": 8.011363636363637e-06,
      "loss": 0.6783,
      "step": 141
    },
    {
      "epoch": 0.3,
      "learning_rate": 8.068181818181819e-06,
      "loss": 0.6641,
      "step": 142
    },
    {
      "epoch": 0.3,
      "learning_rate": 8.125000000000001e-06,
      "loss": 0.6763,
      "step": 143
    },
    {
      "epoch": 0.31,
      "learning_rate": 8.181818181818183e-06,
      "loss": 0.7325,
      "step": 144
    },
    {
      "epoch": 0.31,
      "learning_rate": 8.238636363636365e-06,
      "loss": 0.6366,
      "step": 145
    },
    {
      "epoch": 0.31,
      "learning_rate": 8.295454545454547e-06,
      "loss": 0.7373,
      "step": 146
    },
    {
      "epoch": 0.31,
      "learning_rate": 8.352272727272727e-06,
      "loss": 0.7105,
      "step": 147
    },
    {
      "epoch": 0.32,
      "learning_rate": 8.40909090909091e-06,
      "loss": 0.6865,
      "step": 148
    },
    {
      "epoch": 0.32,
      "learning_rate": 8.465909090909091e-06,
      "loss": 0.6784,
      "step": 149
    },
    {
      "epoch": 0.32,
      "learning_rate": 8.522727272727273e-06,
      "loss": 0.6769,
      "step": 150
    }
  ],
  "max_steps": 11725,
  "num_train_epochs": 25,
  "total_flos": 3.974479252291584e+17,
  "trial_name": null,
  "trial_params": null
}