{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.8213592233009708,
  "global_step": 164,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.02,
      "learning_rate": 2.631578947368421e-06,
      "loss": 1.9044,
      "step": 1
    },
    {
      "epoch": 0.05,
      "learning_rate": 5.263157894736842e-06,
      "loss": 2.018,
      "step": 2
    },
    {
      "epoch": 0.07,
      "learning_rate": 7.894736842105263e-06,
      "loss": 2.0221,
      "step": 3
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.0526315789473684e-05,
      "loss": 1.8884,
      "step": 4
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.3157894736842106e-05,
      "loss": 1.9326,
      "step": 5
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.5789473684210526e-05,
      "loss": 2.1193,
      "step": 6
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.8421052631578947e-05,
      "loss": 2.0827,
      "step": 7
    },
    {
      "epoch": 0.19,
      "learning_rate": 2.105263157894737e-05,
      "loss": 1.8846,
      "step": 8
    },
    {
      "epoch": 0.21,
      "learning_rate": 2.368421052631579e-05,
      "loss": 2.0943,
      "step": 9
    },
    {
      "epoch": 0.23,
      "learning_rate": 2.6315789473684212e-05,
      "loss": 2.0881,
      "step": 10
    },
    {
      "epoch": 0.26,
      "learning_rate": 2.8947368421052634e-05,
      "loss": 1.9275,
      "step": 11
    },
    {
      "epoch": 0.28,
      "learning_rate": 3.157894736842105e-05,
      "loss": 1.6874,
      "step": 12
    },
    {
      "epoch": 0.3,
      "learning_rate": 3.421052631578947e-05,
      "loss": 1.9874,
      "step": 13
    },
    {
      "epoch": 0.33,
      "learning_rate": 3.6842105263157895e-05,
      "loss": 1.8901,
      "step": 14
    },
    {
      "epoch": 0.35,
      "learning_rate": 3.9473684210526316e-05,
      "loss": 1.7683,
      "step": 15
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.210526315789474e-05,
      "loss": 2.0108,
      "step": 16
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.473684210526316e-05,
      "loss": 1.6579,
      "step": 17
    },
    {
      "epoch": 0.42,
      "learning_rate": 4.736842105263158e-05,
      "loss": 1.9758,
      "step": 18
    },
    {
      "epoch": 0.44,
      "learning_rate": 5e-05,
      "loss": 1.9118,
      "step": 19
    },
    {
      "epoch": 0.47,
      "learning_rate": 5.2631578947368424e-05,
      "loss": 1.8964,
      "step": 20
    },
    {
      "epoch": 0.49,
      "learning_rate": 5.526315789473685e-05,
      "loss": 1.9288,
      "step": 21
    },
    {
      "epoch": 0.51,
      "learning_rate": 5.789473684210527e-05,
      "loss": 1.8706,
      "step": 22
    },
    {
      "epoch": 0.54,
      "learning_rate": 6.052631578947369e-05,
      "loss": 1.8604,
      "step": 23
    },
    {
      "epoch": 0.56,
      "learning_rate": 6.31578947368421e-05,
      "loss": 1.9001,
      "step": 24
    },
    {
      "epoch": 0.58,
      "learning_rate": 6.578947368421054e-05,
      "loss": 1.8254,
      "step": 25
    },
    {
      "epoch": 0.61,
      "learning_rate": 6.842105263157895e-05,
      "loss": 1.8111,
      "step": 26
    },
    {
      "epoch": 0.63,
      "learning_rate": 7.105263157894737e-05,
      "loss": 1.7496,
      "step": 27
    },
    {
      "epoch": 0.65,
      "learning_rate": 7.368421052631579e-05,
      "loss": 1.7604,
      "step": 28
    },
    {
      "epoch": 0.68,
      "learning_rate": 7.631578947368422e-05,
      "loss": 1.6285,
      "step": 29
    },
    {
      "epoch": 0.7,
      "learning_rate": 7.894736842105263e-05,
      "loss": 1.7226,
      "step": 30
    },
    {
      "epoch": 0.72,
      "learning_rate": 8.157894736842105e-05,
      "loss": 1.9689,
      "step": 31
    },
    {
      "epoch": 0.75,
      "learning_rate": 8.421052631578948e-05,
      "loss": 1.7737,
      "step": 32
    },
    {
      "epoch": 0.77,
      "learning_rate": 8.68421052631579e-05,
      "loss": 1.8093,
      "step": 33
    },
    {
      "epoch": 0.79,
      "learning_rate": 8.947368421052632e-05,
      "loss": 1.6292,
      "step": 34
    },
    {
      "epoch": 0.82,
      "learning_rate": 9.210526315789474e-05,
      "loss": 1.8351,
      "step": 35
    },
    {
      "epoch": 0.84,
      "learning_rate": 9.473684210526316e-05,
      "loss": 1.918,
      "step": 36
    },
    {
      "epoch": 0.86,
      "learning_rate": 9.736842105263158e-05,
      "loss": 1.8161,
      "step": 37
    },
    {
      "epoch": 0.89,
      "learning_rate": 0.0001,
      "loss": 1.8184,
      "step": 38
    },
    {
      "epoch": 0.91,
      "learning_rate": 9.983108108108109e-05,
      "loss": 1.5839,
      "step": 39
    },
    {
      "epoch": 0.93,
      "learning_rate": 9.966216216216217e-05,
      "loss": 1.8736,
      "step": 40
    },
    {
      "epoch": 0.96,
      "learning_rate": 9.949324324324325e-05,
      "loss": 1.77,
      "step": 41
    },
    {
      "epoch": 0.98,
      "learning_rate": 9.932432432432433e-05,
      "loss": 1.7912,
      "step": 42
    },
    {
      "epoch": 1.0,
      "learning_rate": 9.915540540540541e-05,
      "loss": 1.6989,
      "step": 43
    },
    {
      "epoch": 1.03,
      "learning_rate": 9.89864864864865e-05,
      "loss": 1.7327,
      "step": 44
    },
    {
      "epoch": 1.05,
      "learning_rate": 9.881756756756757e-05,
      "loss": 1.7799,
      "step": 45
    },
    {
      "epoch": 1.07,
      "learning_rate": 9.864864864864865e-05,
      "loss": 1.8298,
      "step": 46
    },
    {
      "epoch": 1.1,
      "learning_rate": 9.847972972972973e-05,
      "loss": 1.8616,
      "step": 47
    },
    {
      "epoch": 1.12,
      "learning_rate": 9.831081081081081e-05,
      "loss": 1.7984,
      "step": 48
    },
    {
      "epoch": 1.14,
      "learning_rate": 9.81418918918919e-05,
      "loss": 1.7626,
      "step": 49
    },
    {
      "epoch": 1.17,
      "learning_rate": 9.797297297297297e-05,
      "loss": 1.8424,
      "step": 50
    },
    {
      "epoch": 1.19,
      "learning_rate": 9.780405405405407e-05,
      "loss": 1.6436,
      "step": 51
    },
    {
      "epoch": 1.21,
      "learning_rate": 9.763513513513513e-05,
      "loss": 1.6303,
      "step": 52
    },
    {
      "epoch": 1.23,
      "learning_rate": 9.746621621621623e-05,
      "loss": 1.7199,
      "step": 53
    },
    {
      "epoch": 1.26,
      "learning_rate": 9.729729729729731e-05,
      "loss": 1.6789,
      "step": 54
    },
    {
      "epoch": 1.28,
      "learning_rate": 9.712837837837838e-05,
      "loss": 1.8519,
      "step": 55
    },
    {
      "epoch": 1.3,
      "learning_rate": 9.695945945945947e-05,
      "loss": 1.7961,
      "step": 56
    },
    {
      "epoch": 1.33,
      "learning_rate": 9.679054054054054e-05,
      "loss": 1.6725,
      "step": 57
    },
    {
      "epoch": 1.35,
      "learning_rate": 9.662162162162163e-05,
      "loss": 1.7249,
      "step": 58
    },
    {
      "epoch": 1.37,
      "learning_rate": 9.64527027027027e-05,
      "loss": 1.5996,
      "step": 59
    },
    {
      "epoch": 1.4,
      "learning_rate": 9.628378378378379e-05,
      "loss": 1.7033,
      "step": 60
    },
    {
      "epoch": 1.42,
      "learning_rate": 9.611486486486487e-05,
      "loss": 1.7387,
      "step": 61
    },
    {
      "epoch": 1.44,
      "learning_rate": 9.594594594594595e-05,
      "loss": 1.7804,
      "step": 62
    },
    {
      "epoch": 1.47,
      "learning_rate": 9.577702702702703e-05,
      "loss": 1.7484,
      "step": 63
    },
    {
      "epoch": 1.49,
      "learning_rate": 9.56081081081081e-05,
      "loss": 1.7526,
      "step": 64
    },
    {
      "epoch": 1.51,
      "learning_rate": 9.54391891891892e-05,
      "loss": 1.7247,
      "step": 65
    },
    {
      "epoch": 1.54,
      "learning_rate": 9.527027027027028e-05,
      "loss": 1.6276,
      "step": 66
    },
    {
      "epoch": 1.56,
      "learning_rate": 9.510135135135136e-05,
      "loss": 1.8279,
      "step": 67
    },
    {
      "epoch": 1.58,
      "learning_rate": 9.493243243243244e-05,
      "loss": 1.7067,
      "step": 68
    },
    {
      "epoch": 1.61,
      "learning_rate": 9.476351351351352e-05,
      "loss": 1.7865,
      "step": 69
    },
    {
      "epoch": 1.63,
      "learning_rate": 9.45945945945946e-05,
      "loss": 1.6693,
      "step": 70
    },
    {
      "epoch": 1.65,
      "learning_rate": 9.442567567567568e-05,
      "loss": 1.6939,
      "step": 71
    },
    {
      "epoch": 1.68,
      "learning_rate": 9.425675675675676e-05,
      "loss": 1.8667,
      "step": 72
    },
    {
      "epoch": 1.7,
      "learning_rate": 9.408783783783784e-05,
      "loss": 1.746,
      "step": 73
    },
    {
      "epoch": 1.72,
      "learning_rate": 9.391891891891892e-05,
      "loss": 1.6951,
      "step": 74
    },
    {
      "epoch": 1.75,
      "learning_rate": 9.375e-05,
      "loss": 1.6749,
      "step": 75
    },
    {
      "epoch": 1.77,
      "learning_rate": 9.35810810810811e-05,
      "loss": 1.8773,
      "step": 76
    },
    {
      "epoch": 1.79,
      "learning_rate": 9.341216216216216e-05,
      "loss": 1.8064,
      "step": 77
    },
    {
      "epoch": 1.82,
      "learning_rate": 9.324324324324324e-05,
      "loss": 1.6596,
      "step": 78
    },
    {
      "epoch": 1.84,
      "learning_rate": 9.307432432432432e-05,
      "loss": 1.7321,
      "step": 79
    },
    {
      "epoch": 1.86,
      "learning_rate": 9.29054054054054e-05,
      "loss": 1.7029,
      "step": 80
    },
    {
      "epoch": 1.89,
      "learning_rate": 9.27364864864865e-05,
      "loss": 1.7392,
      "step": 81
    },
    {
      "epoch": 1.91,
      "learning_rate": 9.256756756756757e-05,
      "loss": 1.7904,
      "step": 82
    },
    {
      "epoch": 1.93,
      "learning_rate": 9.239864864864866e-05,
      "loss": 1.6794,
      "step": 83
    },
    {
      "epoch": 1.96,
      "learning_rate": 9.222972972972973e-05,
      "loss": 1.8827,
      "step": 84
    },
    {
      "epoch": 1.98,
      "learning_rate": 9.206081081081082e-05,
      "loss": 1.7051,
      "step": 85
    },
    {
      "epoch": 2.0,
      "learning_rate": 9.18918918918919e-05,
      "loss": 1.7069,
      "step": 86
    },
    {
      "epoch": 2.03,
      "learning_rate": 9.172297297297297e-05,
      "loss": 1.8051,
      "step": 87
    },
    {
      "epoch": 2.05,
      "learning_rate": 9.155405405405406e-05,
      "loss": 1.6665,
      "step": 88
    },
    {
      "epoch": 2.07,
      "learning_rate": 9.138513513513513e-05,
      "loss": 1.6218,
      "step": 89
    },
    {
      "epoch": 2.1,
      "learning_rate": 9.121621621621623e-05,
      "loss": 1.5848,
      "step": 90
    },
    {
      "epoch": 2.12,
      "learning_rate": 9.10472972972973e-05,
      "loss": 1.7544,
      "step": 91
    },
    {
      "epoch": 2.14,
      "learning_rate": 9.087837837837839e-05,
      "loss": 1.6454,
      "step": 92
    },
    {
      "epoch": 2.17,
      "learning_rate": 9.070945945945947e-05,
      "loss": 1.8591,
      "step": 93
    },
    {
      "epoch": 2.19,
      "learning_rate": 9.054054054054055e-05,
      "loss": 1.7025,
      "step": 94
    },
    {
      "epoch": 2.21,
      "learning_rate": 9.037162162162163e-05,
      "loss": 1.7824,
      "step": 95
    },
    {
      "epoch": 2.24,
      "learning_rate": 9.02027027027027e-05,
      "loss": 1.6049,
      "step": 96
    },
    {
      "epoch": 2.26,
      "learning_rate": 9.003378378378379e-05,
      "loss": 1.6961,
      "step": 97
    },
    {
      "epoch": 2.28,
      "learning_rate": 8.986486486486487e-05,
      "loss": 1.8705,
      "step": 98
    },
    {
      "epoch": 2.31,
      "learning_rate": 8.969594594594595e-05,
      "loss": 1.4498,
      "step": 99
    },
    {
      "epoch": 2.33,
      "learning_rate": 8.952702702702703e-05,
      "loss": 1.6967,
      "step": 100
    },
    {
      "epoch": 2.35,
      "learning_rate": 8.935810810810811e-05,
      "loss": 1.6345,
      "step": 101
    },
    {
      "epoch": 2.38,
      "learning_rate": 8.918918918918919e-05,
      "loss": 1.8037,
      "step": 102
    },
    {
      "epoch": 2.4,
      "learning_rate": 8.902027027027027e-05,
      "loss": 1.7896,
      "step": 103
    },
    {
      "epoch": 2.42,
      "learning_rate": 8.885135135135135e-05,
      "loss": 1.751,
      "step": 104
    },
    {
      "epoch": 2.45,
      "learning_rate": 8.868243243243243e-05,
      "loss": 1.482,
      "step": 105
    },
    {
      "epoch": 2.47,
      "learning_rate": 8.851351351351352e-05,
      "loss": 1.6049,
      "step": 106
    },
    {
      "epoch": 2.49,
      "learning_rate": 8.83445945945946e-05,
      "loss": 1.6029,
      "step": 107
    },
    {
      "epoch": 2.52,
      "learning_rate": 8.817567567567569e-05,
      "loss": 1.875,
      "step": 108
    },
    {
      "epoch": 2.54,
      "learning_rate": 8.800675675675676e-05,
      "loss": 1.5798,
      "step": 109
    },
    {
      "epoch": 2.56,
      "learning_rate": 8.783783783783784e-05,
      "loss": 1.4883,
      "step": 110
    },
    {
      "epoch": 2.59,
      "learning_rate": 8.766891891891892e-05,
      "loss": 1.7249,
      "step": 111
    },
    {
      "epoch": 2.61,
      "learning_rate": 8.75e-05,
      "loss": 1.5086,
      "step": 112
    },
    {
      "epoch": 2.63,
      "learning_rate": 8.73310810810811e-05,
      "loss": 1.39,
      "step": 113
    },
    {
      "epoch": 2.66,
      "learning_rate": 8.716216216216216e-05,
      "loss": 1.7481,
      "step": 114
    },
    {
      "epoch": 2.68,
      "learning_rate": 8.699324324324325e-05,
      "loss": 1.6514,
      "step": 115
    },
    {
      "epoch": 2.7,
      "learning_rate": 8.682432432432432e-05,
      "loss": 1.8731,
      "step": 116
    },
    {
      "epoch": 2.73,
      "learning_rate": 8.665540540540542e-05,
      "loss": 1.6415,
      "step": 117
    },
    {
      "epoch": 2.75,
      "learning_rate": 8.64864864864865e-05,
      "loss": 1.7659,
      "step": 118
    },
    {
      "epoch": 2.77,
      "learning_rate": 8.631756756756756e-05,
      "loss": 1.5691,
      "step": 119
    },
    {
      "epoch": 2.8,
      "learning_rate": 8.614864864864866e-05,
      "loss": 1.6315,
      "step": 120
    },
    {
      "epoch": 2.82,
      "learning_rate": 8.597972972972972e-05,
      "loss": 1.367,
      "step": 121
    },
    {
      "epoch": 2.84,
      "learning_rate": 8.581081081081082e-05,
      "loss": 1.6088,
      "step": 122
    },
    {
      "epoch": 2.87,
      "learning_rate": 8.56418918918919e-05,
      "loss": 1.7598,
      "step": 123
    },
    {
      "epoch": 2.89,
      "learning_rate": 8.547297297297298e-05,
      "loss": 1.6255,
      "step": 124
    },
    {
      "epoch": 2.91,
      "learning_rate": 8.530405405405406e-05,
      "loss": 1.5739,
      "step": 125
    },
    {
      "epoch": 2.94,
      "learning_rate": 8.513513513513514e-05,
      "loss": 1.5542,
      "step": 126
    },
    {
      "epoch": 2.96,
      "learning_rate": 8.496621621621622e-05,
      "loss": 1.5758,
      "step": 127
    },
    {
      "epoch": 2.98,
      "learning_rate": 8.47972972972973e-05,
      "loss": 1.668,
      "step": 128
    },
    {
      "epoch": 3.01,
      "learning_rate": 8.462837837837838e-05,
      "loss": 1.6283,
      "step": 129
    },
    {
      "epoch": 3.03,
      "learning_rate": 8.445945945945946e-05,
      "loss": 1.6111,
      "step": 130
    },
    {
      "epoch": 3.05,
      "learning_rate": 8.429054054054054e-05,
      "loss": 1.8061,
      "step": 131
    },
    {
      "epoch": 3.08,
      "learning_rate": 8.412162162162163e-05,
      "loss": 1.4882,
      "step": 132
    },
    {
      "epoch": 3.1,
      "learning_rate": 8.39527027027027e-05,
      "loss": 1.592,
      "step": 133
    },
    {
      "epoch": 3.12,
      "learning_rate": 8.378378378378379e-05,
      "loss": 1.638,
      "step": 134
    },
    {
      "epoch": 3.15,
      "learning_rate": 8.361486486486487e-05,
      "loss": 1.5036,
      "step": 135
    },
    {
      "epoch": 3.17,
      "learning_rate": 8.344594594594595e-05,
      "loss": 1.6721,
      "step": 136
    },
    {
      "epoch": 3.19,
      "learning_rate": 8.327702702702703e-05,
      "loss": 1.7742,
      "step": 137
    },
    {
      "epoch": 3.22,
      "learning_rate": 8.310810810810811e-05,
      "loss": 1.484,
      "step": 138
    },
    {
      "epoch": 3.24,
      "learning_rate": 8.293918918918919e-05,
      "loss": 1.3698,
      "step": 139
    },
    {
      "epoch": 3.26,
      "learning_rate": 8.277027027027028e-05,
      "loss": 1.5644,
      "step": 140
    },
    {
      "epoch": 3.29,
      "learning_rate": 8.260135135135135e-05,
      "loss": 1.6195,
      "step": 141
    },
    {
      "epoch": 3.31,
      "learning_rate": 8.243243243243243e-05,
      "loss": 1.5031,
      "step": 142
    },
    {
      "epoch": 3.33,
      "learning_rate": 8.226351351351351e-05,
      "loss": 1.5403,
      "step": 143
    },
    {
      "epoch": 3.36,
      "learning_rate": 8.209459459459459e-05,
      "loss": 1.57,
      "step": 144
    },
    {
      "epoch": 3.38,
      "learning_rate": 8.192567567567569e-05,
      "loss": 1.6454,
      "step": 145
    },
    {
      "epoch": 3.4,
      "learning_rate": 8.175675675675675e-05,
      "loss": 1.5069,
      "step": 146
    },
    {
      "epoch": 3.43,
      "learning_rate": 8.158783783783785e-05,
      "loss": 1.3782,
      "step": 147
    },
    {
      "epoch": 3.45,
      "learning_rate": 8.141891891891892e-05,
      "loss": 1.4296,
      "step": 148
    },
    {
      "epoch": 3.47,
      "learning_rate": 8.125000000000001e-05,
      "loss": 1.4303,
      "step": 149
    },
    {
      "epoch": 3.5,
      "learning_rate": 8.108108108108109e-05,
      "loss": 1.6368,
      "step": 150
    },
    {
      "epoch": 3.52,
      "learning_rate": 8.091216216216216e-05,
      "loss": 1.5951,
      "step": 151
    },
    {
      "epoch": 3.54,
      "learning_rate": 8.074324324324325e-05,
      "loss": 1.5769,
      "step": 152
    },
    {
      "epoch": 3.57,
      "learning_rate": 8.057432432432432e-05,
      "loss": 1.482,
      "step": 153
    },
    {
      "epoch": 3.59,
      "learning_rate": 8.040540540540541e-05,
      "loss": 1.5415,
      "step": 154
    },
    {
      "epoch": 3.61,
      "learning_rate": 8.02364864864865e-05,
      "loss": 1.6084,
      "step": 155
    },
    {
      "epoch": 3.63,
      "learning_rate": 8.006756756756757e-05,
      "loss": 1.5483,
      "step": 156
    },
    {
      "epoch": 3.66,
      "learning_rate": 7.989864864864865e-05,
      "loss": 1.4329,
      "step": 157
    },
    {
      "epoch": 3.68,
      "learning_rate": 7.972972972972974e-05,
      "loss": 1.6068,
      "step": 158
    },
    {
      "epoch": 3.7,
      "learning_rate": 7.956081081081082e-05,
      "loss": 1.5519,
      "step": 159
    },
    {
      "epoch": 3.73,
      "learning_rate": 7.93918918918919e-05,
      "loss": 1.7387,
      "step": 160
    },
    {
      "epoch": 3.75,
      "learning_rate": 7.922297297297298e-05,
      "loss": 1.5583,
      "step": 161
    },
    {
      "epoch": 3.77,
      "learning_rate": 7.905405405405406e-05,
      "loss": 1.5003,
      "step": 162
    },
    {
      "epoch": 3.8,
      "learning_rate": 7.888513513513514e-05,
      "loss": 1.5025,
      "step": 163
    },
    {
      "epoch": 3.82,
      "learning_rate": 7.871621621621622e-05,
      "loss": 1.5322,
      "step": 164
    }
  ],
  "max_steps": 630,
  "num_train_epochs": 15,
  "total_flos": 5.937698934423552e+16,
  "trial_name": null,
  "trial_params": null
}