{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.06,
  "eval_steps": 1000,
  "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": 6.666666666666667e-05,
      "loss": 2.3598,
      "step": 1
    },
    {
      "epoch": 0.0,
      "learning_rate": 0.00013333333333333334,
      "loss": 2.1113,
      "step": 2
    },
    {
      "epoch": 0.0,
      "learning_rate": 0.0002,
      "loss": 2.0722,
      "step": 3
    },
    {
      "epoch": 0.0,
      "learning_rate": 0.00019863945578231293,
      "loss": 2.1798,
      "step": 4
    },
    {
      "epoch": 0.0,
      "learning_rate": 0.00019727891156462587,
      "loss": 1.9333,
      "step": 5
    },
    {
      "epoch": 0.0,
      "learning_rate": 0.0001959183673469388,
      "loss": 2.4093,
      "step": 6
    },
    {
      "epoch": 0.0,
      "learning_rate": 0.0001945578231292517,
      "loss": 2.6837,
      "step": 7
    },
    {
      "epoch": 0.0,
      "learning_rate": 0.00019319727891156462,
      "loss": 1.5956,
      "step": 8
    },
    {
      "epoch": 0.0,
      "learning_rate": 0.00019183673469387756,
      "loss": 2.02,
      "step": 9
    },
    {
      "epoch": 0.0,
      "learning_rate": 0.00019047619047619048,
      "loss": 2.0534,
      "step": 10
    },
    {
      "epoch": 0.0,
      "learning_rate": 0.00018911564625850343,
      "loss": 1.804,
      "step": 11
    },
    {
      "epoch": 0.0,
      "learning_rate": 0.00018775510204081634,
      "loss": 2.0297,
      "step": 12
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00018639455782312926,
      "loss": 1.8164,
      "step": 13
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0001850340136054422,
      "loss": 1.8929,
      "step": 14
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00018367346938775512,
      "loss": 1.7967,
      "step": 15
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00018231292517006804,
      "loss": 1.5366,
      "step": 16
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00018095238095238095,
      "loss": 1.7421,
      "step": 17
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0001795918367346939,
      "loss": 2.5308,
      "step": 18
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00017823129251700681,
      "loss": 1.5277,
      "step": 19
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00017687074829931973,
      "loss": 1.9084,
      "step": 20
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00017551020408163265,
      "loss": 1.9708,
      "step": 21
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0001741496598639456,
      "loss": 1.8567,
      "step": 22
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0001727891156462585,
      "loss": 1.5694,
      "step": 23
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00017142857142857143,
      "loss": 1.8519,
      "step": 24
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00017006802721088434,
      "loss": 1.663,
      "step": 25
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00016870748299319729,
      "loss": 1.9309,
      "step": 26
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00016734693877551023,
      "loss": 1.551,
      "step": 27
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00016598639455782315,
      "loss": 1.4782,
      "step": 28
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00016462585034013606,
      "loss": 1.6643,
      "step": 29
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00016326530612244898,
      "loss": 1.5571,
      "step": 30
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00016190476190476192,
      "loss": 1.6747,
      "step": 31
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00016054421768707484,
      "loss": 1.3539,
      "step": 32
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00015918367346938776,
      "loss": 2.0579,
      "step": 33
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00015782312925170067,
      "loss": 1.5418,
      "step": 34
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00015646258503401362,
      "loss": 1.8159,
      "step": 35
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00015510204081632654,
      "loss": 1.6081,
      "step": 36
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00015374149659863945,
      "loss": 1.5529,
      "step": 37
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00015238095238095237,
      "loss": 1.5456,
      "step": 38
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0001510204081632653,
      "loss": 2.1016,
      "step": 39
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00014965986394557826,
      "loss": 1.4394,
      "step": 40
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00014829931972789117,
      "loss": 1.4033,
      "step": 41
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0001469387755102041,
      "loss": 1.5887,
      "step": 42
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.000145578231292517,
      "loss": 1.8761,
      "step": 43
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00014421768707482995,
      "loss": 1.67,
      "step": 44
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00014285714285714287,
      "loss": 1.8064,
      "step": 45
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00014149659863945578,
      "loss": 1.4283,
      "step": 46
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0001401360544217687,
      "loss": 1.8028,
      "step": 47
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00013877551020408165,
      "loss": 1.5101,
      "step": 48
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00013741496598639456,
      "loss": 1.7825,
      "step": 49
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00013605442176870748,
      "loss": 1.3797,
      "step": 50
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0001346938775510204,
      "loss": 1.7408,
      "step": 51
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00013333333333333334,
      "loss": 1.6008,
      "step": 52
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00013197278911564626,
      "loss": 1.5121,
      "step": 53
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00013061224489795917,
      "loss": 1.3599,
      "step": 54
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00012925170068027212,
      "loss": 1.4465,
      "step": 55
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00012789115646258506,
      "loss": 1.6332,
      "step": 56
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00012653061224489798,
      "loss": 1.6305,
      "step": 57
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0001251700680272109,
      "loss": 1.5599,
      "step": 58
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0001238095238095238,
      "loss": 1.8265,
      "step": 59
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00012244897959183676,
      "loss": 1.7149,
      "step": 60
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00012108843537414967,
      "loss": 1.7577,
      "step": 61
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00011972789115646259,
      "loss": 1.5519,
      "step": 62
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00011836734693877552,
      "loss": 1.4681,
      "step": 63
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00011700680272108844,
      "loss": 1.3328,
      "step": 64
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00011564625850340137,
      "loss": 1.4249,
      "step": 65
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00011428571428571428,
      "loss": 1.8161,
      "step": 66
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00011292517006802721,
      "loss": 1.4775,
      "step": 67
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00011156462585034013,
      "loss": 1.5223,
      "step": 68
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00011020408163265306,
      "loss": 1.5734,
      "step": 69
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.000108843537414966,
      "loss": 1.5901,
      "step": 70
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00010748299319727892,
      "loss": 2.1238,
      "step": 71
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00010612244897959185,
      "loss": 1.8808,
      "step": 72
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00010476190476190477,
      "loss": 1.3648,
      "step": 73
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0001034013605442177,
      "loss": 1.3667,
      "step": 74
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00010204081632653062,
      "loss": 1.3925,
      "step": 75
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00010068027210884355,
      "loss": 1.8534,
      "step": 76
    },
    {
      "epoch": 0.03,
      "learning_rate": 9.931972789115646e-05,
      "loss": 1.6686,
      "step": 77
    },
    {
      "epoch": 0.03,
      "learning_rate": 9.79591836734694e-05,
      "loss": 1.5057,
      "step": 78
    },
    {
      "epoch": 0.03,
      "learning_rate": 9.659863945578231e-05,
      "loss": 1.4539,
      "step": 79
    },
    {
      "epoch": 0.03,
      "learning_rate": 9.523809523809524e-05,
      "loss": 1.6102,
      "step": 80
    },
    {
      "epoch": 0.03,
      "learning_rate": 9.387755102040817e-05,
      "loss": 1.3404,
      "step": 81
    },
    {
      "epoch": 0.03,
      "learning_rate": 9.25170068027211e-05,
      "loss": 1.5923,
      "step": 82
    },
    {
      "epoch": 0.03,
      "learning_rate": 9.115646258503402e-05,
      "loss": 1.34,
      "step": 83
    },
    {
      "epoch": 0.03,
      "learning_rate": 8.979591836734695e-05,
      "loss": 1.9159,
      "step": 84
    },
    {
      "epoch": 0.03,
      "learning_rate": 8.843537414965987e-05,
      "loss": 1.5353,
      "step": 85
    },
    {
      "epoch": 0.03,
      "learning_rate": 8.70748299319728e-05,
      "loss": 1.5158,
      "step": 86
    },
    {
      "epoch": 0.03,
      "learning_rate": 8.571428571428571e-05,
      "loss": 1.3898,
      "step": 87
    },
    {
      "epoch": 0.04,
      "learning_rate": 8.435374149659864e-05,
      "loss": 1.6535,
      "step": 88
    },
    {
      "epoch": 0.04,
      "learning_rate": 8.299319727891157e-05,
      "loss": 1.6544,
      "step": 89
    },
    {
      "epoch": 0.04,
      "learning_rate": 8.163265306122449e-05,
      "loss": 1.8656,
      "step": 90
    },
    {
      "epoch": 0.04,
      "learning_rate": 8.027210884353742e-05,
      "loss": 1.465,
      "step": 91
    },
    {
      "epoch": 0.04,
      "learning_rate": 7.891156462585034e-05,
      "loss": 1.0163,
      "step": 92
    },
    {
      "epoch": 0.04,
      "learning_rate": 7.755102040816327e-05,
      "loss": 1.61,
      "step": 93
    },
    {
      "epoch": 0.04,
      "learning_rate": 7.619047619047618e-05,
      "loss": 2.0532,
      "step": 94
    },
    {
      "epoch": 0.04,
      "learning_rate": 7.482993197278913e-05,
      "loss": 1.4248,
      "step": 95
    },
    {
      "epoch": 0.04,
      "learning_rate": 7.346938775510205e-05,
      "loss": 1.4999,
      "step": 96
    },
    {
      "epoch": 0.04,
      "learning_rate": 7.210884353741498e-05,
      "loss": 1.6352,
      "step": 97
    },
    {
      "epoch": 0.04,
      "learning_rate": 7.074829931972789e-05,
      "loss": 1.6345,
      "step": 98
    },
    {
      "epoch": 0.04,
      "learning_rate": 6.938775510204082e-05,
      "loss": 1.6019,
      "step": 99
    },
    {
      "epoch": 0.04,
      "learning_rate": 6.802721088435374e-05,
      "loss": 1.4834,
      "step": 100
    },
    {
      "epoch": 0.04,
      "learning_rate": 6.666666666666667e-05,
      "loss": 1.4514,
      "step": 101
    },
    {
      "epoch": 0.04,
      "learning_rate": 6.530612244897959e-05,
      "loss": 1.5484,
      "step": 102
    },
    {
      "epoch": 0.04,
      "learning_rate": 6.394557823129253e-05,
      "loss": 1.6164,
      "step": 103
    },
    {
      "epoch": 0.04,
      "learning_rate": 6.258503401360545e-05,
      "loss": 1.6819,
      "step": 104
    },
    {
      "epoch": 0.04,
      "learning_rate": 6.122448979591838e-05,
      "loss": 1.8219,
      "step": 105
    },
    {
      "epoch": 0.04,
      "learning_rate": 5.9863945578231295e-05,
      "loss": 1.385,
      "step": 106
    },
    {
      "epoch": 0.04,
      "learning_rate": 5.850340136054422e-05,
      "loss": 1.6358,
      "step": 107
    },
    {
      "epoch": 0.04,
      "learning_rate": 5.714285714285714e-05,
      "loss": 1.5059,
      "step": 108
    },
    {
      "epoch": 0.04,
      "learning_rate": 5.5782312925170065e-05,
      "loss": 1.6794,
      "step": 109
    },
    {
      "epoch": 0.04,
      "learning_rate": 5.4421768707483e-05,
      "loss": 1.4888,
      "step": 110
    },
    {
      "epoch": 0.04,
      "learning_rate": 5.3061224489795926e-05,
      "loss": 1.3177,
      "step": 111
    },
    {
      "epoch": 0.04,
      "learning_rate": 5.170068027210885e-05,
      "loss": 1.5026,
      "step": 112
    },
    {
      "epoch": 0.05,
      "learning_rate": 5.034013605442177e-05,
      "loss": 1.8063,
      "step": 113
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.89795918367347e-05,
      "loss": 1.8037,
      "step": 114
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.761904761904762e-05,
      "loss": 1.6401,
      "step": 115
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.625850340136055e-05,
      "loss": 1.7781,
      "step": 116
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.4897959183673474e-05,
      "loss": 1.7268,
      "step": 117
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.35374149659864e-05,
      "loss": 1.6482,
      "step": 118
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.217687074829932e-05,
      "loss": 1.5628,
      "step": 119
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.0816326530612245e-05,
      "loss": 1.3526,
      "step": 120
    },
    {
      "epoch": 0.05,
      "learning_rate": 3.945578231292517e-05,
      "loss": 1.3096,
      "step": 121
    },
    {
      "epoch": 0.05,
      "learning_rate": 3.809523809523809e-05,
      "loss": 1.6429,
      "step": 122
    },
    {
      "epoch": 0.05,
      "learning_rate": 3.673469387755102e-05,
      "loss": 1.4119,
      "step": 123
    },
    {
      "epoch": 0.05,
      "learning_rate": 3.5374149659863946e-05,
      "loss": 1.4932,
      "step": 124
    },
    {
      "epoch": 0.05,
      "learning_rate": 3.401360544217687e-05,
      "loss": 1.8325,
      "step": 125
    },
    {
      "epoch": 0.05,
      "learning_rate": 3.265306122448979e-05,
      "loss": 1.6309,
      "step": 126
    },
    {
      "epoch": 0.05,
      "learning_rate": 3.1292517006802724e-05,
      "loss": 1.6754,
      "step": 127
    },
    {
      "epoch": 0.05,
      "learning_rate": 2.9931972789115647e-05,
      "loss": 1.1972,
      "step": 128
    },
    {
      "epoch": 0.05,
      "learning_rate": 2.857142857142857e-05,
      "loss": 1.6411,
      "step": 129
    },
    {
      "epoch": 0.05,
      "learning_rate": 2.72108843537415e-05,
      "loss": 2.031,
      "step": 130
    },
    {
      "epoch": 0.05,
      "learning_rate": 2.5850340136054425e-05,
      "loss": 1.726,
      "step": 131
    },
    {
      "epoch": 0.05,
      "learning_rate": 2.448979591836735e-05,
      "loss": 1.0537,
      "step": 132
    },
    {
      "epoch": 0.05,
      "learning_rate": 2.3129251700680275e-05,
      "loss": 1.2133,
      "step": 133
    },
    {
      "epoch": 0.05,
      "learning_rate": 2.17687074829932e-05,
      "loss": 1.4264,
      "step": 134
    },
    {
      "epoch": 0.05,
      "learning_rate": 2.0408163265306123e-05,
      "loss": 1.8015,
      "step": 135
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.9047619047619046e-05,
      "loss": 1.6847,
      "step": 136
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.7687074829931973e-05,
      "loss": 1.6395,
      "step": 137
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.6326530612244897e-05,
      "loss": 1.3069,
      "step": 138
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.4965986394557824e-05,
      "loss": 1.7292,
      "step": 139
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.360544217687075e-05,
      "loss": 1.5996,
      "step": 140
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.2244897959183674e-05,
      "loss": 1.5048,
      "step": 141
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.08843537414966e-05,
      "loss": 1.5075,
      "step": 142
    },
    {
      "epoch": 0.06,
      "learning_rate": 9.523809523809523e-06,
      "loss": 1.4126,
      "step": 143
    },
    {
      "epoch": 0.06,
      "learning_rate": 8.163265306122448e-06,
      "loss": 1.4858,
      "step": 144
    },
    {
      "epoch": 0.06,
      "learning_rate": 6.802721088435375e-06,
      "loss": 1.4445,
      "step": 145
    },
    {
      "epoch": 0.06,
      "learning_rate": 5.4421768707483e-06,
      "loss": 1.2193,
      "step": 146
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.081632653061224e-06,
      "loss": 1.5606,
      "step": 147
    },
    {
      "epoch": 0.06,
      "learning_rate": 2.72108843537415e-06,
      "loss": 1.5721,
      "step": 148
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.360544217687075e-06,
      "loss": 1.8318,
      "step": 149
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0,
      "loss": 1.4401,
      "step": 150
    }
  ],
  "logging_steps": 1,
  "max_steps": 150,
  "num_train_epochs": 1,
  "save_steps": 5,
  "total_flos": 1.863424689143808e+16,
  "trial_name": null,
  "trial_params": null
}