wav2vec2-2-gpt2-grid-search / trainer_state.json
sanchit-gandhi's picture
End of training
e1ca89d
raw
history blame
No virus
50.3 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 1.0,
"global_step": 446,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 2.0000000000000002e-07,
"loss": 4.7719,
"step": 1
},
{
"epoch": 0.0,
"learning_rate": 4.0000000000000003e-07,
"loss": 4.9065,
"step": 2
},
{
"epoch": 0.01,
"learning_rate": 4.0000000000000003e-07,
"loss": 4.885,
"step": 3
},
{
"epoch": 0.01,
"learning_rate": 6.000000000000001e-07,
"loss": 4.8861,
"step": 4
},
{
"epoch": 0.01,
"learning_rate": 8.000000000000001e-07,
"loss": 4.8113,
"step": 5
},
{
"epoch": 0.01,
"learning_rate": 1.0000000000000002e-06,
"loss": 4.8648,
"step": 6
},
{
"epoch": 0.02,
"learning_rate": 1.2000000000000002e-06,
"loss": 4.8398,
"step": 7
},
{
"epoch": 0.02,
"learning_rate": 1.4000000000000001e-06,
"loss": 4.8766,
"step": 8
},
{
"epoch": 0.02,
"learning_rate": 1.6000000000000001e-06,
"loss": 4.8603,
"step": 9
},
{
"epoch": 0.02,
"learning_rate": 1.8e-06,
"loss": 4.7241,
"step": 10
},
{
"epoch": 0.02,
"learning_rate": 2.0000000000000003e-06,
"loss": 4.8192,
"step": 11
},
{
"epoch": 0.03,
"learning_rate": 2.2e-06,
"loss": 4.7752,
"step": 12
},
{
"epoch": 0.03,
"learning_rate": 2.4000000000000003e-06,
"loss": 4.7171,
"step": 13
},
{
"epoch": 0.03,
"learning_rate": 2.6e-06,
"loss": 4.707,
"step": 14
},
{
"epoch": 0.03,
"learning_rate": 2.8000000000000003e-06,
"loss": 4.5768,
"step": 15
},
{
"epoch": 0.04,
"learning_rate": 3e-06,
"loss": 4.7214,
"step": 16
},
{
"epoch": 0.04,
"learning_rate": 3.2000000000000003e-06,
"loss": 4.6155,
"step": 17
},
{
"epoch": 0.04,
"learning_rate": 3.4000000000000005e-06,
"loss": 4.6982,
"step": 18
},
{
"epoch": 0.04,
"learning_rate": 3.6e-06,
"loss": 4.5043,
"step": 19
},
{
"epoch": 0.04,
"learning_rate": 3.8e-06,
"loss": 4.6563,
"step": 20
},
{
"epoch": 0.05,
"learning_rate": 4.000000000000001e-06,
"loss": 4.6315,
"step": 21
},
{
"epoch": 0.05,
"learning_rate": 4.2000000000000004e-06,
"loss": 4.5689,
"step": 22
},
{
"epoch": 0.05,
"learning_rate": 4.4e-06,
"loss": 4.6517,
"step": 23
},
{
"epoch": 0.05,
"learning_rate": 4.6e-06,
"loss": 4.626,
"step": 24
},
{
"epoch": 0.06,
"learning_rate": 4.800000000000001e-06,
"loss": 4.4167,
"step": 25
},
{
"epoch": 0.06,
"learning_rate": 5e-06,
"loss": 4.6779,
"step": 26
},
{
"epoch": 0.06,
"learning_rate": 5.2e-06,
"loss": 4.5106,
"step": 27
},
{
"epoch": 0.06,
"learning_rate": 5.4e-06,
"loss": 4.3687,
"step": 28
},
{
"epoch": 0.07,
"learning_rate": 5.600000000000001e-06,
"loss": 4.3321,
"step": 29
},
{
"epoch": 0.07,
"learning_rate": 5.8e-06,
"loss": 4.5401,
"step": 30
},
{
"epoch": 0.07,
"learning_rate": 6e-06,
"loss": 4.5029,
"step": 31
},
{
"epoch": 0.07,
"learning_rate": 6.2e-06,
"loss": 4.3607,
"step": 32
},
{
"epoch": 0.07,
"learning_rate": 6.4000000000000006e-06,
"loss": 4.4283,
"step": 33
},
{
"epoch": 0.08,
"learning_rate": 6.6e-06,
"loss": 4.3708,
"step": 34
},
{
"epoch": 0.08,
"learning_rate": 6.800000000000001e-06,
"loss": 4.4411,
"step": 35
},
{
"epoch": 0.08,
"learning_rate": 7.000000000000001e-06,
"loss": 4.408,
"step": 36
},
{
"epoch": 0.08,
"learning_rate": 7.2e-06,
"loss": 4.4535,
"step": 37
},
{
"epoch": 0.09,
"learning_rate": 7.4e-06,
"loss": 4.3956,
"step": 38
},
{
"epoch": 0.09,
"learning_rate": 7.6e-06,
"loss": 4.5053,
"step": 39
},
{
"epoch": 0.09,
"learning_rate": 7.8e-06,
"loss": 4.3493,
"step": 40
},
{
"epoch": 0.09,
"learning_rate": 8.000000000000001e-06,
"loss": 4.3782,
"step": 41
},
{
"epoch": 0.09,
"learning_rate": 8.200000000000001e-06,
"loss": 4.4784,
"step": 42
},
{
"epoch": 0.1,
"learning_rate": 8.400000000000001e-06,
"loss": 4.2425,
"step": 43
},
{
"epoch": 0.1,
"learning_rate": 8.599999999999999e-06,
"loss": 4.4174,
"step": 44
},
{
"epoch": 0.1,
"learning_rate": 8.8e-06,
"loss": 4.5094,
"step": 45
},
{
"epoch": 0.1,
"learning_rate": 9e-06,
"loss": 4.5613,
"step": 46
},
{
"epoch": 0.11,
"learning_rate": 9.2e-06,
"loss": 4.4531,
"step": 47
},
{
"epoch": 0.11,
"learning_rate": 9.4e-06,
"loss": 4.7066,
"step": 48
},
{
"epoch": 0.11,
"learning_rate": 9.600000000000001e-06,
"loss": 4.8332,
"step": 49
},
{
"epoch": 0.11,
"learning_rate": 9.600000000000001e-06,
"loss": 4.5146,
"step": 50
},
{
"epoch": 0.11,
"learning_rate": 9.800000000000001e-06,
"loss": 4.3416,
"step": 51
},
{
"epoch": 0.12,
"learning_rate": 1e-05,
"loss": 4.214,
"step": 52
},
{
"epoch": 0.12,
"learning_rate": 1.02e-05,
"loss": 4.3766,
"step": 53
},
{
"epoch": 0.12,
"learning_rate": 1.04e-05,
"loss": 4.2895,
"step": 54
},
{
"epoch": 0.12,
"learning_rate": 1.06e-05,
"loss": 4.1125,
"step": 55
},
{
"epoch": 0.13,
"learning_rate": 1.08e-05,
"loss": 4.3304,
"step": 56
},
{
"epoch": 0.13,
"learning_rate": 1.1000000000000001e-05,
"loss": 4.3424,
"step": 57
},
{
"epoch": 0.13,
"learning_rate": 1.1200000000000001e-05,
"loss": 4.4227,
"step": 58
},
{
"epoch": 0.13,
"learning_rate": 1.1400000000000001e-05,
"loss": 4.3562,
"step": 59
},
{
"epoch": 0.13,
"learning_rate": 1.16e-05,
"loss": 4.217,
"step": 60
},
{
"epoch": 0.14,
"learning_rate": 1.18e-05,
"loss": 4.2907,
"step": 61
},
{
"epoch": 0.14,
"learning_rate": 1.2e-05,
"loss": 4.246,
"step": 62
},
{
"epoch": 0.14,
"learning_rate": 1.22e-05,
"loss": 4.2075,
"step": 63
},
{
"epoch": 0.14,
"learning_rate": 1.24e-05,
"loss": 4.3065,
"step": 64
},
{
"epoch": 0.15,
"learning_rate": 1.2600000000000001e-05,
"loss": 4.285,
"step": 65
},
{
"epoch": 0.15,
"learning_rate": 1.2800000000000001e-05,
"loss": 4.1584,
"step": 66
},
{
"epoch": 0.15,
"learning_rate": 1.3000000000000001e-05,
"loss": 4.2986,
"step": 67
},
{
"epoch": 0.15,
"learning_rate": 1.32e-05,
"loss": 4.1812,
"step": 68
},
{
"epoch": 0.15,
"learning_rate": 1.3400000000000002e-05,
"loss": 4.229,
"step": 69
},
{
"epoch": 0.16,
"learning_rate": 1.3600000000000002e-05,
"loss": 4.1952,
"step": 70
},
{
"epoch": 0.16,
"learning_rate": 1.3800000000000002e-05,
"loss": 4.3216,
"step": 71
},
{
"epoch": 0.16,
"learning_rate": 1.4000000000000001e-05,
"loss": 4.3481,
"step": 72
},
{
"epoch": 0.16,
"learning_rate": 1.42e-05,
"loss": 4.1716,
"step": 73
},
{
"epoch": 0.17,
"learning_rate": 1.44e-05,
"loss": 4.2743,
"step": 74
},
{
"epoch": 0.17,
"learning_rate": 1.4599999999999999e-05,
"loss": 4.2567,
"step": 75
},
{
"epoch": 0.17,
"learning_rate": 1.48e-05,
"loss": 4.2005,
"step": 76
},
{
"epoch": 0.17,
"learning_rate": 1.5e-05,
"loss": 4.2588,
"step": 77
},
{
"epoch": 0.17,
"learning_rate": 1.52e-05,
"loss": 4.2041,
"step": 78
},
{
"epoch": 0.18,
"learning_rate": 1.54e-05,
"loss": 4.1828,
"step": 79
},
{
"epoch": 0.18,
"learning_rate": 1.56e-05,
"loss": 4.3437,
"step": 80
},
{
"epoch": 0.18,
"learning_rate": 1.58e-05,
"loss": 4.2745,
"step": 81
},
{
"epoch": 0.18,
"learning_rate": 1.6000000000000003e-05,
"loss": 4.2209,
"step": 82
},
{
"epoch": 0.19,
"learning_rate": 1.62e-05,
"loss": 4.339,
"step": 83
},
{
"epoch": 0.19,
"learning_rate": 1.6400000000000002e-05,
"loss": 4.2074,
"step": 84
},
{
"epoch": 0.19,
"learning_rate": 1.66e-05,
"loss": 4.287,
"step": 85
},
{
"epoch": 0.19,
"learning_rate": 1.6800000000000002e-05,
"loss": 4.2516,
"step": 86
},
{
"epoch": 0.2,
"learning_rate": 1.7000000000000003e-05,
"loss": 4.3282,
"step": 87
},
{
"epoch": 0.2,
"learning_rate": 1.7199999999999998e-05,
"loss": 4.2117,
"step": 88
},
{
"epoch": 0.2,
"learning_rate": 1.74e-05,
"loss": 4.3481,
"step": 89
},
{
"epoch": 0.2,
"learning_rate": 1.76e-05,
"loss": 4.3362,
"step": 90
},
{
"epoch": 0.2,
"learning_rate": 1.78e-05,
"loss": 4.2163,
"step": 91
},
{
"epoch": 0.21,
"learning_rate": 1.8e-05,
"loss": 4.4685,
"step": 92
},
{
"epoch": 0.21,
"learning_rate": 1.8200000000000002e-05,
"loss": 4.3722,
"step": 93
},
{
"epoch": 0.21,
"learning_rate": 1.84e-05,
"loss": 4.3702,
"step": 94
},
{
"epoch": 0.21,
"learning_rate": 1.86e-05,
"loss": 4.2807,
"step": 95
},
{
"epoch": 0.22,
"learning_rate": 1.88e-05,
"loss": 4.5093,
"step": 96
},
{
"epoch": 0.22,
"learning_rate": 1.9e-05,
"loss": 4.4255,
"step": 97
},
{
"epoch": 0.22,
"learning_rate": 1.9200000000000003e-05,
"loss": 4.2955,
"step": 98
},
{
"epoch": 0.22,
"learning_rate": 1.94e-05,
"loss": 4.631,
"step": 99
},
{
"epoch": 0.22,
"learning_rate": 1.9600000000000002e-05,
"loss": 4.4001,
"step": 100
},
{
"epoch": 0.23,
"learning_rate": 1.9800000000000004e-05,
"loss": 4.2505,
"step": 101
},
{
"epoch": 0.23,
"learning_rate": 2e-05,
"loss": 4.1911,
"step": 102
},
{
"epoch": 0.23,
"learning_rate": 2.0200000000000003e-05,
"loss": 4.0639,
"step": 103
},
{
"epoch": 0.23,
"learning_rate": 2.04e-05,
"loss": 4.1361,
"step": 104
},
{
"epoch": 0.24,
"learning_rate": 2.06e-05,
"loss": 4.0851,
"step": 105
},
{
"epoch": 0.24,
"learning_rate": 2.08e-05,
"loss": 4.1384,
"step": 106
},
{
"epoch": 0.24,
"learning_rate": 2.1e-05,
"loss": 4.2131,
"step": 107
},
{
"epoch": 0.24,
"learning_rate": 2.12e-05,
"loss": 4.1188,
"step": 108
},
{
"epoch": 0.24,
"learning_rate": 2.1400000000000002e-05,
"loss": 4.1356,
"step": 109
},
{
"epoch": 0.25,
"learning_rate": 2.16e-05,
"loss": 4.1708,
"step": 110
},
{
"epoch": 0.25,
"learning_rate": 2.18e-05,
"loss": 4.2002,
"step": 111
},
{
"epoch": 0.25,
"learning_rate": 2.2000000000000003e-05,
"loss": 4.1928,
"step": 112
},
{
"epoch": 0.25,
"learning_rate": 2.22e-05,
"loss": 4.1811,
"step": 113
},
{
"epoch": 0.26,
"learning_rate": 2.2400000000000002e-05,
"loss": 4.1331,
"step": 114
},
{
"epoch": 0.26,
"learning_rate": 2.26e-05,
"loss": 4.184,
"step": 115
},
{
"epoch": 0.26,
"learning_rate": 2.2800000000000002e-05,
"loss": 4.2902,
"step": 116
},
{
"epoch": 0.26,
"learning_rate": 2.3000000000000003e-05,
"loss": 4.1586,
"step": 117
},
{
"epoch": 0.26,
"learning_rate": 2.32e-05,
"loss": 4.3134,
"step": 118
},
{
"epoch": 0.27,
"learning_rate": 2.3400000000000003e-05,
"loss": 4.2146,
"step": 119
},
{
"epoch": 0.27,
"learning_rate": 2.36e-05,
"loss": 4.239,
"step": 120
},
{
"epoch": 0.27,
"learning_rate": 2.38e-05,
"loss": 4.0619,
"step": 121
},
{
"epoch": 0.27,
"learning_rate": 2.4e-05,
"loss": 4.1774,
"step": 122
},
{
"epoch": 0.28,
"learning_rate": 2.4200000000000002e-05,
"loss": 4.1803,
"step": 123
},
{
"epoch": 0.28,
"learning_rate": 2.44e-05,
"loss": 4.1746,
"step": 124
},
{
"epoch": 0.28,
"learning_rate": 2.46e-05,
"loss": 4.2787,
"step": 125
},
{
"epoch": 0.28,
"learning_rate": 2.48e-05,
"loss": 4.2883,
"step": 126
},
{
"epoch": 0.28,
"learning_rate": 2.5e-05,
"loss": 4.2007,
"step": 127
},
{
"epoch": 0.29,
"learning_rate": 2.5200000000000003e-05,
"loss": 4.1726,
"step": 128
},
{
"epoch": 0.29,
"learning_rate": 2.54e-05,
"loss": 4.0857,
"step": 129
},
{
"epoch": 0.29,
"learning_rate": 2.5600000000000002e-05,
"loss": 4.236,
"step": 130
},
{
"epoch": 0.29,
"learning_rate": 2.58e-05,
"loss": 4.1492,
"step": 131
},
{
"epoch": 0.3,
"learning_rate": 2.6000000000000002e-05,
"loss": 4.2103,
"step": 132
},
{
"epoch": 0.3,
"learning_rate": 2.6200000000000003e-05,
"loss": 4.1179,
"step": 133
},
{
"epoch": 0.3,
"learning_rate": 2.64e-05,
"loss": 4.1823,
"step": 134
},
{
"epoch": 0.3,
"learning_rate": 2.6600000000000003e-05,
"loss": 4.3251,
"step": 135
},
{
"epoch": 0.3,
"learning_rate": 2.6800000000000004e-05,
"loss": 4.1815,
"step": 136
},
{
"epoch": 0.31,
"learning_rate": 2.7000000000000002e-05,
"loss": 4.2615,
"step": 137
},
{
"epoch": 0.31,
"learning_rate": 2.7200000000000004e-05,
"loss": 4.3035,
"step": 138
},
{
"epoch": 0.31,
"learning_rate": 2.7400000000000002e-05,
"loss": 4.3086,
"step": 139
},
{
"epoch": 0.31,
"learning_rate": 2.7600000000000003e-05,
"loss": 4.2835,
"step": 140
},
{
"epoch": 0.32,
"learning_rate": 2.7800000000000005e-05,
"loss": 4.3894,
"step": 141
},
{
"epoch": 0.32,
"learning_rate": 2.8000000000000003e-05,
"loss": 4.3306,
"step": 142
},
{
"epoch": 0.32,
"learning_rate": 2.8199999999999998e-05,
"loss": 4.2895,
"step": 143
},
{
"epoch": 0.32,
"learning_rate": 2.84e-05,
"loss": 4.3263,
"step": 144
},
{
"epoch": 0.33,
"learning_rate": 2.86e-05,
"loss": 4.2518,
"step": 145
},
{
"epoch": 0.33,
"learning_rate": 2.88e-05,
"loss": 4.3215,
"step": 146
},
{
"epoch": 0.33,
"learning_rate": 2.9e-05,
"loss": 4.598,
"step": 147
},
{
"epoch": 0.33,
"learning_rate": 2.9199999999999998e-05,
"loss": 4.7036,
"step": 148
},
{
"epoch": 0.33,
"learning_rate": 2.94e-05,
"loss": 4.4583,
"step": 149
},
{
"epoch": 0.34,
"learning_rate": 2.96e-05,
"loss": 4.5292,
"step": 150
},
{
"epoch": 0.34,
"learning_rate": 2.98e-05,
"loss": 4.1027,
"step": 151
},
{
"epoch": 0.34,
"learning_rate": 3e-05,
"loss": 4.1962,
"step": 152
},
{
"epoch": 0.34,
"learning_rate": 3.02e-05,
"loss": 4.2274,
"step": 153
},
{
"epoch": 0.35,
"learning_rate": 3.04e-05,
"loss": 4.0786,
"step": 154
},
{
"epoch": 0.35,
"learning_rate": 3.06e-05,
"loss": 4.2101,
"step": 155
},
{
"epoch": 0.35,
"learning_rate": 3.08e-05,
"loss": 4.0632,
"step": 156
},
{
"epoch": 0.35,
"learning_rate": 3.1e-05,
"loss": 4.0863,
"step": 157
},
{
"epoch": 0.35,
"learning_rate": 3.12e-05,
"loss": 4.184,
"step": 158
},
{
"epoch": 0.36,
"learning_rate": 3.1400000000000004e-05,
"loss": 4.1929,
"step": 159
},
{
"epoch": 0.36,
"learning_rate": 3.16e-05,
"loss": 4.2031,
"step": 160
},
{
"epoch": 0.36,
"learning_rate": 3.18e-05,
"loss": 4.2709,
"step": 161
},
{
"epoch": 0.36,
"learning_rate": 3.2000000000000005e-05,
"loss": 4.212,
"step": 162
},
{
"epoch": 0.37,
"learning_rate": 3.2200000000000003e-05,
"loss": 4.1633,
"step": 163
},
{
"epoch": 0.37,
"learning_rate": 3.24e-05,
"loss": 4.2215,
"step": 164
},
{
"epoch": 0.37,
"learning_rate": 3.26e-05,
"loss": 4.0926,
"step": 165
},
{
"epoch": 0.37,
"learning_rate": 3.2800000000000004e-05,
"loss": 4.1606,
"step": 166
},
{
"epoch": 0.37,
"learning_rate": 3.3e-05,
"loss": 4.0721,
"step": 167
},
{
"epoch": 0.38,
"learning_rate": 3.32e-05,
"loss": 4.1272,
"step": 168
},
{
"epoch": 0.38,
"learning_rate": 3.3400000000000005e-05,
"loss": 4.2054,
"step": 169
},
{
"epoch": 0.38,
"learning_rate": 3.3600000000000004e-05,
"loss": 4.2183,
"step": 170
},
{
"epoch": 0.38,
"learning_rate": 3.38e-05,
"loss": 4.2042,
"step": 171
},
{
"epoch": 0.39,
"learning_rate": 3.4000000000000007e-05,
"loss": 4.275,
"step": 172
},
{
"epoch": 0.39,
"learning_rate": 3.4200000000000005e-05,
"loss": 4.0782,
"step": 173
},
{
"epoch": 0.39,
"learning_rate": 3.4399999999999996e-05,
"loss": 4.2265,
"step": 174
},
{
"epoch": 0.39,
"learning_rate": 3.46e-05,
"loss": 4.2685,
"step": 175
},
{
"epoch": 0.39,
"learning_rate": 3.48e-05,
"loss": 4.1473,
"step": 176
},
{
"epoch": 0.4,
"learning_rate": 3.5e-05,
"loss": 4.0942,
"step": 177
},
{
"epoch": 0.4,
"learning_rate": 3.52e-05,
"loss": 4.2087,
"step": 178
},
{
"epoch": 0.4,
"learning_rate": 3.54e-05,
"loss": 4.2813,
"step": 179
},
{
"epoch": 0.4,
"learning_rate": 3.56e-05,
"loss": 4.2491,
"step": 180
},
{
"epoch": 0.41,
"learning_rate": 3.58e-05,
"loss": 4.1937,
"step": 181
},
{
"epoch": 0.41,
"learning_rate": 3.6e-05,
"loss": 4.2598,
"step": 182
},
{
"epoch": 0.41,
"learning_rate": 3.62e-05,
"loss": 4.2181,
"step": 183
},
{
"epoch": 0.41,
"learning_rate": 3.6400000000000004e-05,
"loss": 4.236,
"step": 184
},
{
"epoch": 0.41,
"learning_rate": 3.66e-05,
"loss": 4.2193,
"step": 185
},
{
"epoch": 0.42,
"learning_rate": 3.68e-05,
"loss": 4.0767,
"step": 186
},
{
"epoch": 0.42,
"learning_rate": 3.7e-05,
"loss": 4.112,
"step": 187
},
{
"epoch": 0.42,
"learning_rate": 3.72e-05,
"loss": 4.0809,
"step": 188
},
{
"epoch": 0.42,
"learning_rate": 3.74e-05,
"loss": 4.2022,
"step": 189
},
{
"epoch": 0.43,
"learning_rate": 3.76e-05,
"loss": 4.3143,
"step": 190
},
{
"epoch": 0.43,
"learning_rate": 3.7800000000000004e-05,
"loss": 4.2295,
"step": 191
},
{
"epoch": 0.43,
"learning_rate": 3.8e-05,
"loss": 4.1771,
"step": 192
},
{
"epoch": 0.43,
"learning_rate": 3.82e-05,
"loss": 4.157,
"step": 193
},
{
"epoch": 0.43,
"learning_rate": 3.8400000000000005e-05,
"loss": 4.2961,
"step": 194
},
{
"epoch": 0.44,
"learning_rate": 3.86e-05,
"loss": 4.42,
"step": 195
},
{
"epoch": 0.44,
"learning_rate": 3.88e-05,
"loss": 4.3345,
"step": 196
},
{
"epoch": 0.44,
"learning_rate": 3.9000000000000006e-05,
"loss": 4.41,
"step": 197
},
{
"epoch": 0.44,
"learning_rate": 3.9200000000000004e-05,
"loss": 4.5266,
"step": 198
},
{
"epoch": 0.45,
"learning_rate": 3.94e-05,
"loss": 4.4871,
"step": 199
},
{
"epoch": 0.45,
"learning_rate": 3.960000000000001e-05,
"loss": 4.4572,
"step": 200
},
{
"epoch": 0.45,
"learning_rate": 3.9800000000000005e-05,
"loss": 4.2354,
"step": 201
},
{
"epoch": 0.45,
"learning_rate": 4e-05,
"loss": 4.1817,
"step": 202
},
{
"epoch": 0.46,
"learning_rate": 4.02e-05,
"loss": 4.214,
"step": 203
},
{
"epoch": 0.46,
"learning_rate": 4.0400000000000006e-05,
"loss": 4.1809,
"step": 204
},
{
"epoch": 0.46,
"learning_rate": 4.0600000000000004e-05,
"loss": 4.2551,
"step": 205
},
{
"epoch": 0.46,
"learning_rate": 4.08e-05,
"loss": 4.2098,
"step": 206
},
{
"epoch": 0.46,
"learning_rate": 4.1e-05,
"loss": 4.0845,
"step": 207
},
{
"epoch": 0.47,
"learning_rate": 4.12e-05,
"loss": 4.2135,
"step": 208
},
{
"epoch": 0.47,
"learning_rate": 4.14e-05,
"loss": 4.2722,
"step": 209
},
{
"epoch": 0.47,
"learning_rate": 4.16e-05,
"loss": 4.2086,
"step": 210
},
{
"epoch": 0.47,
"learning_rate": 4.18e-05,
"loss": 4.0768,
"step": 211
},
{
"epoch": 0.48,
"learning_rate": 4.2e-05,
"loss": 4.1328,
"step": 212
},
{
"epoch": 0.48,
"learning_rate": 4.22e-05,
"loss": 4.0752,
"step": 213
},
{
"epoch": 0.48,
"learning_rate": 4.24e-05,
"loss": 4.1268,
"step": 214
},
{
"epoch": 0.48,
"learning_rate": 4.26e-05,
"loss": 4.1781,
"step": 215
},
{
"epoch": 0.48,
"learning_rate": 4.2800000000000004e-05,
"loss": 4.163,
"step": 216
},
{
"epoch": 0.49,
"learning_rate": 4.3e-05,
"loss": 4.193,
"step": 217
},
{
"epoch": 0.49,
"learning_rate": 4.32e-05,
"loss": 4.1417,
"step": 218
},
{
"epoch": 0.49,
"learning_rate": 4.3400000000000005e-05,
"loss": 4.1006,
"step": 219
},
{
"epoch": 0.49,
"learning_rate": 4.36e-05,
"loss": 4.183,
"step": 220
},
{
"epoch": 0.5,
"learning_rate": 4.38e-05,
"loss": 4.1586,
"step": 221
},
{
"epoch": 0.5,
"learning_rate": 4.4000000000000006e-05,
"loss": 4.1583,
"step": 222
},
{
"epoch": 0.5,
"learning_rate": 4.4200000000000004e-05,
"loss": 4.1894,
"step": 223
},
{
"epoch": 0.5,
"learning_rate": 4.44e-05,
"loss": 4.1255,
"step": 224
},
{
"epoch": 0.5,
"learning_rate": 4.46e-05,
"loss": 4.2794,
"step": 225
},
{
"epoch": 0.51,
"learning_rate": 4.4800000000000005e-05,
"loss": 4.1867,
"step": 226
},
{
"epoch": 0.51,
"learning_rate": 4.5e-05,
"loss": 4.2804,
"step": 227
},
{
"epoch": 0.51,
"learning_rate": 4.52e-05,
"loss": 4.2764,
"step": 228
},
{
"epoch": 0.51,
"learning_rate": 4.5400000000000006e-05,
"loss": 4.1211,
"step": 229
},
{
"epoch": 0.52,
"learning_rate": 4.5600000000000004e-05,
"loss": 4.2534,
"step": 230
},
{
"epoch": 0.52,
"learning_rate": 4.58e-05,
"loss": 4.2681,
"step": 231
},
{
"epoch": 0.52,
"learning_rate": 4.600000000000001e-05,
"loss": 4.0795,
"step": 232
},
{
"epoch": 0.52,
"learning_rate": 4.6200000000000005e-05,
"loss": 4.1336,
"step": 233
},
{
"epoch": 0.52,
"learning_rate": 4.64e-05,
"loss": 4.1979,
"step": 234
},
{
"epoch": 0.53,
"learning_rate": 4.660000000000001e-05,
"loss": 4.2561,
"step": 235
},
{
"epoch": 0.53,
"learning_rate": 4.6800000000000006e-05,
"loss": 4.2117,
"step": 236
},
{
"epoch": 0.53,
"learning_rate": 4.7e-05,
"loss": 4.192,
"step": 237
},
{
"epoch": 0.53,
"learning_rate": 4.72e-05,
"loss": 4.1534,
"step": 238
},
{
"epoch": 0.54,
"learning_rate": 4.74e-05,
"loss": 4.1088,
"step": 239
},
{
"epoch": 0.54,
"learning_rate": 4.76e-05,
"loss": 4.1981,
"step": 240
},
{
"epoch": 0.54,
"learning_rate": 4.78e-05,
"loss": 4.2105,
"step": 241
},
{
"epoch": 0.54,
"learning_rate": 4.8e-05,
"loss": 4.2749,
"step": 242
},
{
"epoch": 0.54,
"learning_rate": 4.82e-05,
"loss": 4.4264,
"step": 243
},
{
"epoch": 0.55,
"learning_rate": 4.8400000000000004e-05,
"loss": 4.2585,
"step": 244
},
{
"epoch": 0.55,
"learning_rate": 4.86e-05,
"loss": 4.2482,
"step": 245
},
{
"epoch": 0.55,
"learning_rate": 4.88e-05,
"loss": 4.1165,
"step": 246
},
{
"epoch": 0.55,
"learning_rate": 4.9e-05,
"loss": 4.0864,
"step": 247
},
{
"epoch": 0.56,
"learning_rate": 4.92e-05,
"loss": 4.5551,
"step": 248
},
{
"epoch": 0.56,
"learning_rate": 4.94e-05,
"loss": 4.5483,
"step": 249
},
{
"epoch": 0.56,
"learning_rate": 4.96e-05,
"loss": 4.5265,
"step": 250
},
{
"epoch": 0.56,
"learning_rate": 4.9800000000000004e-05,
"loss": 4.2203,
"step": 251
},
{
"epoch": 0.57,
"learning_rate": 5e-05,
"loss": 4.2102,
"step": 252
},
{
"epoch": 0.57,
"learning_rate": 5.02e-05,
"loss": 4.2521,
"step": 253
},
{
"epoch": 0.57,
"learning_rate": 5.0400000000000005e-05,
"loss": 4.1037,
"step": 254
},
{
"epoch": 0.57,
"learning_rate": 5.0600000000000003e-05,
"loss": 4.1975,
"step": 255
},
{
"epoch": 0.57,
"learning_rate": 5.08e-05,
"loss": 4.1548,
"step": 256
},
{
"epoch": 0.58,
"learning_rate": 5.1000000000000006e-05,
"loss": 4.1562,
"step": 257
},
{
"epoch": 0.58,
"learning_rate": 5.1200000000000004e-05,
"loss": 4.1191,
"step": 258
},
{
"epoch": 0.58,
"learning_rate": 5.14e-05,
"loss": 4.1629,
"step": 259
},
{
"epoch": 0.58,
"learning_rate": 5.16e-05,
"loss": 4.1329,
"step": 260
},
{
"epoch": 0.59,
"learning_rate": 5.1800000000000005e-05,
"loss": 4.2366,
"step": 261
},
{
"epoch": 0.59,
"learning_rate": 5.2000000000000004e-05,
"loss": 4.1222,
"step": 262
},
{
"epoch": 0.59,
"learning_rate": 5.22e-05,
"loss": 4.2949,
"step": 263
},
{
"epoch": 0.59,
"learning_rate": 5.2400000000000007e-05,
"loss": 4.1441,
"step": 264
},
{
"epoch": 0.59,
"learning_rate": 5.2600000000000005e-05,
"loss": 4.2206,
"step": 265
},
{
"epoch": 0.6,
"learning_rate": 5.28e-05,
"loss": 4.1359,
"step": 266
},
{
"epoch": 0.6,
"learning_rate": 5.300000000000001e-05,
"loss": 4.18,
"step": 267
},
{
"epoch": 0.6,
"learning_rate": 5.3200000000000006e-05,
"loss": 4.0921,
"step": 268
},
{
"epoch": 0.6,
"learning_rate": 5.3400000000000004e-05,
"loss": 4.2223,
"step": 269
},
{
"epoch": 0.61,
"learning_rate": 5.360000000000001e-05,
"loss": 4.1064,
"step": 270
},
{
"epoch": 0.61,
"learning_rate": 5.380000000000001e-05,
"loss": 4.0658,
"step": 271
},
{
"epoch": 0.61,
"learning_rate": 5.4000000000000005e-05,
"loss": 4.2146,
"step": 272
},
{
"epoch": 0.61,
"learning_rate": 5.420000000000001e-05,
"loss": 4.0509,
"step": 273
},
{
"epoch": 0.61,
"learning_rate": 5.440000000000001e-05,
"loss": 4.1473,
"step": 274
},
{
"epoch": 0.62,
"learning_rate": 5.4600000000000006e-05,
"loss": 4.2196,
"step": 275
},
{
"epoch": 0.62,
"learning_rate": 5.4800000000000004e-05,
"loss": 4.1378,
"step": 276
},
{
"epoch": 0.62,
"learning_rate": 5.500000000000001e-05,
"loss": 4.0931,
"step": 277
},
{
"epoch": 0.62,
"learning_rate": 5.520000000000001e-05,
"loss": 4.1675,
"step": 278
},
{
"epoch": 0.63,
"learning_rate": 5.5400000000000005e-05,
"loss": 4.1488,
"step": 279
},
{
"epoch": 0.63,
"learning_rate": 5.560000000000001e-05,
"loss": 4.2942,
"step": 280
},
{
"epoch": 0.63,
"learning_rate": 5.580000000000001e-05,
"loss": 4.1823,
"step": 281
},
{
"epoch": 0.63,
"learning_rate": 5.6000000000000006e-05,
"loss": 4.0571,
"step": 282
},
{
"epoch": 0.63,
"learning_rate": 5.620000000000001e-05,
"loss": 4.0695,
"step": 283
},
{
"epoch": 0.64,
"learning_rate": 5.6399999999999995e-05,
"loss": 4.2051,
"step": 284
},
{
"epoch": 0.64,
"learning_rate": 5.66e-05,
"loss": 4.2795,
"step": 285
},
{
"epoch": 0.64,
"learning_rate": 5.68e-05,
"loss": 4.1569,
"step": 286
},
{
"epoch": 0.64,
"learning_rate": 5.6999999999999996e-05,
"loss": 4.2607,
"step": 287
},
{
"epoch": 0.65,
"learning_rate": 5.72e-05,
"loss": 4.2057,
"step": 288
},
{
"epoch": 0.65,
"learning_rate": 5.74e-05,
"loss": 4.1146,
"step": 289
},
{
"epoch": 0.65,
"learning_rate": 5.76e-05,
"loss": 4.1695,
"step": 290
},
{
"epoch": 0.65,
"learning_rate": 5.7799999999999995e-05,
"loss": 4.4908,
"step": 291
},
{
"epoch": 0.65,
"learning_rate": 5.8e-05,
"loss": 4.2949,
"step": 292
},
{
"epoch": 0.66,
"learning_rate": 5.82e-05,
"loss": 4.3671,
"step": 293
},
{
"epoch": 0.66,
"learning_rate": 5.8399999999999997e-05,
"loss": 4.307,
"step": 294
},
{
"epoch": 0.66,
"learning_rate": 5.86e-05,
"loss": 4.3258,
"step": 295
},
{
"epoch": 0.66,
"learning_rate": 5.88e-05,
"loss": 4.3597,
"step": 296
},
{
"epoch": 0.67,
"learning_rate": 5.9e-05,
"loss": 4.2197,
"step": 297
},
{
"epoch": 0.67,
"learning_rate": 5.92e-05,
"loss": 4.6226,
"step": 298
},
{
"epoch": 0.67,
"learning_rate": 5.94e-05,
"loss": 4.2705,
"step": 299
},
{
"epoch": 0.67,
"learning_rate": 5.96e-05,
"loss": 4.714,
"step": 300
},
{
"epoch": 0.67,
"learning_rate": 5.9800000000000003e-05,
"loss": 4.2312,
"step": 301
},
{
"epoch": 0.68,
"learning_rate": 6e-05,
"loss": 4.2223,
"step": 302
},
{
"epoch": 0.68,
"learning_rate": 6.02e-05,
"loss": 4.1857,
"step": 303
},
{
"epoch": 0.68,
"learning_rate": 6.04e-05,
"loss": 4.215,
"step": 304
},
{
"epoch": 0.68,
"learning_rate": 6.06e-05,
"loss": 4.1659,
"step": 305
},
{
"epoch": 0.69,
"learning_rate": 6.08e-05,
"loss": 4.1269,
"step": 306
},
{
"epoch": 0.69,
"learning_rate": 6.1e-05,
"loss": 4.1849,
"step": 307
},
{
"epoch": 0.69,
"learning_rate": 6.12e-05,
"loss": 4.1578,
"step": 308
},
{
"epoch": 0.69,
"learning_rate": 6.14e-05,
"loss": 4.1545,
"step": 309
},
{
"epoch": 0.7,
"learning_rate": 6.16e-05,
"loss": 4.2542,
"step": 310
},
{
"epoch": 0.7,
"learning_rate": 6.18e-05,
"loss": 3.9987,
"step": 311
},
{
"epoch": 0.7,
"learning_rate": 6.2e-05,
"loss": 4.2024,
"step": 312
},
{
"epoch": 0.7,
"learning_rate": 6.220000000000001e-05,
"loss": 4.2768,
"step": 313
},
{
"epoch": 0.7,
"learning_rate": 6.24e-05,
"loss": 4.128,
"step": 314
},
{
"epoch": 0.71,
"learning_rate": 6.26e-05,
"loss": 4.1512,
"step": 315
},
{
"epoch": 0.71,
"learning_rate": 6.280000000000001e-05,
"loss": 4.1232,
"step": 316
},
{
"epoch": 0.71,
"learning_rate": 6.3e-05,
"loss": 4.1407,
"step": 317
},
{
"epoch": 0.71,
"learning_rate": 6.32e-05,
"loss": 4.1325,
"step": 318
},
{
"epoch": 0.72,
"learning_rate": 6.340000000000001e-05,
"loss": 4.1598,
"step": 319
},
{
"epoch": 0.72,
"learning_rate": 6.36e-05,
"loss": 4.2247,
"step": 320
},
{
"epoch": 0.72,
"learning_rate": 6.38e-05,
"loss": 4.1314,
"step": 321
},
{
"epoch": 0.72,
"learning_rate": 6.400000000000001e-05,
"loss": 3.981,
"step": 322
},
{
"epoch": 0.72,
"learning_rate": 6.42e-05,
"loss": 4.1958,
"step": 323
},
{
"epoch": 0.73,
"learning_rate": 6.440000000000001e-05,
"loss": 4.0383,
"step": 324
},
{
"epoch": 0.73,
"learning_rate": 6.460000000000001e-05,
"loss": 4.2174,
"step": 325
},
{
"epoch": 0.73,
"learning_rate": 6.48e-05,
"loss": 4.1948,
"step": 326
},
{
"epoch": 0.73,
"learning_rate": 6.500000000000001e-05,
"loss": 4.1112,
"step": 327
},
{
"epoch": 0.74,
"learning_rate": 6.52e-05,
"loss": 4.0492,
"step": 328
},
{
"epoch": 0.74,
"learning_rate": 6.54e-05,
"loss": 4.079,
"step": 329
},
{
"epoch": 0.74,
"learning_rate": 6.560000000000001e-05,
"loss": 4.2257,
"step": 330
},
{
"epoch": 0.74,
"learning_rate": 6.58e-05,
"loss": 4.2159,
"step": 331
},
{
"epoch": 0.74,
"learning_rate": 6.6e-05,
"loss": 4.0892,
"step": 332
},
{
"epoch": 0.75,
"learning_rate": 6.620000000000001e-05,
"loss": 4.2605,
"step": 333
},
{
"epoch": 0.75,
"learning_rate": 6.64e-05,
"loss": 4.2019,
"step": 334
},
{
"epoch": 0.75,
"learning_rate": 6.66e-05,
"loss": 4.3192,
"step": 335
},
{
"epoch": 0.75,
"learning_rate": 6.680000000000001e-05,
"loss": 4.2142,
"step": 336
},
{
"epoch": 0.76,
"learning_rate": 6.7e-05,
"loss": 4.1008,
"step": 337
},
{
"epoch": 0.76,
"learning_rate": 6.720000000000001e-05,
"loss": 4.1991,
"step": 338
},
{
"epoch": 0.76,
"learning_rate": 6.740000000000001e-05,
"loss": 4.2477,
"step": 339
},
{
"epoch": 0.76,
"learning_rate": 6.76e-05,
"loss": 4.277,
"step": 340
},
{
"epoch": 0.76,
"learning_rate": 6.780000000000001e-05,
"loss": 4.1477,
"step": 341
},
{
"epoch": 0.77,
"learning_rate": 6.800000000000001e-05,
"loss": 4.3392,
"step": 342
},
{
"epoch": 0.77,
"learning_rate": 6.82e-05,
"loss": 4.2372,
"step": 343
},
{
"epoch": 0.77,
"learning_rate": 6.840000000000001e-05,
"loss": 4.355,
"step": 344
},
{
"epoch": 0.77,
"learning_rate": 6.860000000000001e-05,
"loss": 4.0971,
"step": 345
},
{
"epoch": 0.78,
"learning_rate": 6.879999999999999e-05,
"loss": 4.2799,
"step": 346
},
{
"epoch": 0.78,
"learning_rate": 6.9e-05,
"loss": 4.3116,
"step": 347
},
{
"epoch": 0.78,
"learning_rate": 6.92e-05,
"loss": 4.2765,
"step": 348
},
{
"epoch": 0.78,
"learning_rate": 6.939999999999999e-05,
"loss": 4.3885,
"step": 349
},
{
"epoch": 0.78,
"learning_rate": 6.96e-05,
"loss": 4.3474,
"step": 350
},
{
"epoch": 0.79,
"learning_rate": 6.98e-05,
"loss": 4.2769,
"step": 351
},
{
"epoch": 0.79,
"learning_rate": 7e-05,
"loss": 4.1673,
"step": 352
},
{
"epoch": 0.79,
"learning_rate": 7.02e-05,
"loss": 4.1257,
"step": 353
},
{
"epoch": 0.79,
"learning_rate": 7.04e-05,
"loss": 4.0416,
"step": 354
},
{
"epoch": 0.8,
"learning_rate": 7.06e-05,
"loss": 4.1925,
"step": 355
},
{
"epoch": 0.8,
"learning_rate": 7.08e-05,
"loss": 4.3139,
"step": 356
},
{
"epoch": 0.8,
"learning_rate": 7.1e-05,
"loss": 4.1854,
"step": 357
},
{
"epoch": 0.8,
"learning_rate": 7.12e-05,
"loss": 4.2114,
"step": 358
},
{
"epoch": 0.8,
"learning_rate": 7.14e-05,
"loss": 4.0914,
"step": 359
},
{
"epoch": 0.81,
"learning_rate": 7.16e-05,
"loss": 4.1907,
"step": 360
},
{
"epoch": 0.81,
"learning_rate": 7.18e-05,
"loss": 4.0951,
"step": 361
},
{
"epoch": 0.81,
"learning_rate": 7.2e-05,
"loss": 4.1182,
"step": 362
},
{
"epoch": 0.81,
"learning_rate": 7.22e-05,
"loss": 4.0534,
"step": 363
},
{
"epoch": 0.82,
"learning_rate": 7.24e-05,
"loss": 4.1806,
"step": 364
},
{
"epoch": 0.82,
"learning_rate": 7.26e-05,
"loss": 4.2851,
"step": 365
},
{
"epoch": 0.82,
"learning_rate": 7.280000000000001e-05,
"loss": 4.1213,
"step": 366
},
{
"epoch": 0.82,
"learning_rate": 7.3e-05,
"loss": 4.205,
"step": 367
},
{
"epoch": 0.83,
"learning_rate": 7.32e-05,
"loss": 4.1709,
"step": 368
},
{
"epoch": 0.83,
"learning_rate": 7.340000000000001e-05,
"loss": 4.064,
"step": 369
},
{
"epoch": 0.83,
"learning_rate": 7.36e-05,
"loss": 4.0942,
"step": 370
},
{
"epoch": 0.83,
"learning_rate": 7.38e-05,
"loss": 4.0847,
"step": 371
},
{
"epoch": 0.83,
"learning_rate": 7.4e-05,
"loss": 4.1675,
"step": 372
},
{
"epoch": 0.84,
"learning_rate": 7.42e-05,
"loss": 4.1618,
"step": 373
},
{
"epoch": 0.84,
"learning_rate": 7.44e-05,
"loss": 3.9479,
"step": 374
},
{
"epoch": 0.84,
"learning_rate": 7.46e-05,
"loss": 4.1896,
"step": 375
},
{
"epoch": 0.84,
"learning_rate": 7.48e-05,
"loss": 4.0284,
"step": 376
},
{
"epoch": 0.85,
"learning_rate": 7.500000000000001e-05,
"loss": 4.1114,
"step": 377
},
{
"epoch": 0.85,
"learning_rate": 7.52e-05,
"loss": 4.2131,
"step": 378
},
{
"epoch": 0.85,
"learning_rate": 7.54e-05,
"loss": 4.1098,
"step": 379
},
{
"epoch": 0.85,
"learning_rate": 7.560000000000001e-05,
"loss": 4.0568,
"step": 380
},
{
"epoch": 0.85,
"learning_rate": 7.58e-05,
"loss": 4.1287,
"step": 381
},
{
"epoch": 0.86,
"learning_rate": 7.6e-05,
"loss": 4.2378,
"step": 382
},
{
"epoch": 0.86,
"learning_rate": 7.620000000000001e-05,
"loss": 4.1744,
"step": 383
},
{
"epoch": 0.86,
"learning_rate": 7.64e-05,
"loss": 4.0495,
"step": 384
},
{
"epoch": 0.86,
"learning_rate": 7.66e-05,
"loss": 4.2233,
"step": 385
},
{
"epoch": 0.87,
"learning_rate": 7.680000000000001e-05,
"loss": 4.0447,
"step": 386
},
{
"epoch": 0.87,
"learning_rate": 7.7e-05,
"loss": 4.1888,
"step": 387
},
{
"epoch": 0.87,
"learning_rate": 7.72e-05,
"loss": 4.2306,
"step": 388
},
{
"epoch": 0.87,
"learning_rate": 7.740000000000001e-05,
"loss": 4.1236,
"step": 389
},
{
"epoch": 0.87,
"learning_rate": 7.76e-05,
"loss": 4.1625,
"step": 390
},
{
"epoch": 0.88,
"learning_rate": 7.780000000000001e-05,
"loss": 4.2749,
"step": 391
},
{
"epoch": 0.88,
"learning_rate": 7.800000000000001e-05,
"loss": 4.0711,
"step": 392
},
{
"epoch": 0.88,
"learning_rate": 7.82e-05,
"loss": 4.0423,
"step": 393
},
{
"epoch": 0.88,
"learning_rate": 7.840000000000001e-05,
"loss": 4.1891,
"step": 394
},
{
"epoch": 0.89,
"learning_rate": 7.860000000000001e-05,
"loss": 4.353,
"step": 395
},
{
"epoch": 0.89,
"learning_rate": 7.88e-05,
"loss": 4.2261,
"step": 396
},
{
"epoch": 0.89,
"learning_rate": 7.900000000000001e-05,
"loss": 4.5605,
"step": 397
},
{
"epoch": 0.89,
"learning_rate": 7.920000000000001e-05,
"loss": 4.3021,
"step": 398
},
{
"epoch": 0.89,
"learning_rate": 7.94e-05,
"loss": 4.6879,
"step": 399
},
{
"epoch": 0.9,
"learning_rate": 7.960000000000001e-05,
"loss": 4.4051,
"step": 400
},
{
"epoch": 0.9,
"learning_rate": 7.98e-05,
"loss": 4.2403,
"step": 401
},
{
"epoch": 0.9,
"learning_rate": 8e-05,
"loss": 4.1457,
"step": 402
},
{
"epoch": 0.9,
"learning_rate": 8.020000000000001e-05,
"loss": 4.2249,
"step": 403
},
{
"epoch": 0.91,
"learning_rate": 8.04e-05,
"loss": 4.1852,
"step": 404
},
{
"epoch": 0.91,
"learning_rate": 8.060000000000001e-05,
"loss": 4.0872,
"step": 405
},
{
"epoch": 0.91,
"learning_rate": 8.080000000000001e-05,
"loss": 4.2534,
"step": 406
},
{
"epoch": 0.91,
"learning_rate": 8.1e-05,
"loss": 4.128,
"step": 407
},
{
"epoch": 0.91,
"learning_rate": 8.120000000000001e-05,
"loss": 4.2292,
"step": 408
},
{
"epoch": 0.92,
"learning_rate": 8.14e-05,
"loss": 4.1299,
"step": 409
},
{
"epoch": 0.92,
"learning_rate": 8.16e-05,
"loss": 4.0981,
"step": 410
},
{
"epoch": 0.92,
"learning_rate": 8.18e-05,
"loss": 4.353,
"step": 411
},
{
"epoch": 0.92,
"learning_rate": 8.2e-05,
"loss": 4.0653,
"step": 412
},
{
"epoch": 0.93,
"learning_rate": 8.22e-05,
"loss": 4.0245,
"step": 413
},
{
"epoch": 0.93,
"learning_rate": 8.24e-05,
"loss": 4.2077,
"step": 414
},
{
"epoch": 0.93,
"learning_rate": 8.26e-05,
"loss": 4.3732,
"step": 415
},
{
"epoch": 0.93,
"learning_rate": 8.28e-05,
"loss": 4.1227,
"step": 416
},
{
"epoch": 0.93,
"learning_rate": 8.3e-05,
"loss": 4.1922,
"step": 417
},
{
"epoch": 0.94,
"learning_rate": 8.32e-05,
"loss": 4.0332,
"step": 418
},
{
"epoch": 0.94,
"learning_rate": 8.34e-05,
"loss": 4.1886,
"step": 419
},
{
"epoch": 0.94,
"learning_rate": 8.36e-05,
"loss": 4.0436,
"step": 420
},
{
"epoch": 0.94,
"learning_rate": 8.38e-05,
"loss": 4.2622,
"step": 421
},
{
"epoch": 0.95,
"learning_rate": 8.4e-05,
"loss": 4.0481,
"step": 422
},
{
"epoch": 0.95,
"learning_rate": 8.42e-05,
"loss": 4.1584,
"step": 423
},
{
"epoch": 0.95,
"learning_rate": 8.44e-05,
"loss": 4.2609,
"step": 424
},
{
"epoch": 0.95,
"learning_rate": 8.46e-05,
"loss": 4.153,
"step": 425
},
{
"epoch": 0.96,
"learning_rate": 8.48e-05,
"loss": 4.1653,
"step": 426
},
{
"epoch": 0.96,
"learning_rate": 8.5e-05,
"loss": 4.1399,
"step": 427
},
{
"epoch": 0.96,
"learning_rate": 8.52e-05,
"loss": 4.1541,
"step": 428
},
{
"epoch": 0.96,
"learning_rate": 8.54e-05,
"loss": 4.0411,
"step": 429
},
{
"epoch": 0.96,
"learning_rate": 8.560000000000001e-05,
"loss": 4.2159,
"step": 430
},
{
"epoch": 0.97,
"learning_rate": 8.58e-05,
"loss": 4.203,
"step": 431
},
{
"epoch": 0.97,
"learning_rate": 8.6e-05,
"loss": 4.3561,
"step": 432
},
{
"epoch": 0.97,
"learning_rate": 8.620000000000001e-05,
"loss": 4.147,
"step": 433
},
{
"epoch": 0.97,
"learning_rate": 8.64e-05,
"loss": 4.0882,
"step": 434
},
{
"epoch": 0.98,
"learning_rate": 8.66e-05,
"loss": 4.239,
"step": 435
},
{
"epoch": 0.98,
"learning_rate": 8.680000000000001e-05,
"loss": 4.047,
"step": 436
},
{
"epoch": 0.98,
"learning_rate": 8.7e-05,
"loss": 4.0671,
"step": 437
},
{
"epoch": 0.98,
"learning_rate": 8.72e-05,
"loss": 4.3211,
"step": 438
},
{
"epoch": 0.98,
"learning_rate": 8.740000000000001e-05,
"loss": 4.2255,
"step": 439
},
{
"epoch": 0.99,
"learning_rate": 8.76e-05,
"loss": 4.2234,
"step": 440
},
{
"epoch": 0.99,
"learning_rate": 8.78e-05,
"loss": 4.1753,
"step": 441
},
{
"epoch": 0.99,
"learning_rate": 8.800000000000001e-05,
"loss": 4.4779,
"step": 442
},
{
"epoch": 0.99,
"learning_rate": 8.82e-05,
"loss": 4.4142,
"step": 443
},
{
"epoch": 1.0,
"learning_rate": 8.840000000000001e-05,
"loss": 4.3129,
"step": 444
},
{
"epoch": 1.0,
"learning_rate": 8.86e-05,
"loss": 4.4601,
"step": 445
},
{
"epoch": 1.0,
"learning_rate": 8.88e-05,
"loss": 4.2186,
"step": 446
},
{
"epoch": 1.0,
"step": 446,
"total_flos": 0.0,
"train_loss": 4.259838603537179,
"train_runtime": 4900.3549,
"train_samples_per_second": 5.824,
"train_steps_per_second": 0.091
}
],
"max_steps": 446,
"num_train_epochs": 1,
"total_flos": 0.0,
"trial_name": null,
"trial_params": null
}