wav2vec2-2-gpt2-grid-search / trainer_state.json
sanchit-gandhi's picture
End of training
728482e
raw history blame
No virus
34.2 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 0.9987389659520807,
"global_step": 297,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 2e-08,
"loss": 4.7862,
"step": 1
},
{
"epoch": 0.01,
"learning_rate": 2e-08,
"loss": 4.8712,
"step": 2
},
{
"epoch": 0.01,
"learning_rate": 4e-08,
"loss": 4.9393,
"step": 3
},
{
"epoch": 0.01,
"learning_rate": 6.000000000000001e-08,
"loss": 4.8916,
"step": 4
},
{
"epoch": 0.02,
"learning_rate": 8e-08,
"loss": 4.7806,
"step": 5
},
{
"epoch": 0.02,
"learning_rate": 1.0000000000000001e-07,
"loss": 4.9123,
"step": 6
},
{
"epoch": 0.02,
"learning_rate": 1.2000000000000002e-07,
"loss": 4.8899,
"step": 7
},
{
"epoch": 0.03,
"learning_rate": 1.4e-07,
"loss": 4.9651,
"step": 8
},
{
"epoch": 0.03,
"learning_rate": 1.6e-07,
"loss": 4.8742,
"step": 9
},
{
"epoch": 0.03,
"learning_rate": 1.8e-07,
"loss": 4.749,
"step": 10
},
{
"epoch": 0.04,
"learning_rate": 2.0000000000000002e-07,
"loss": 4.8902,
"step": 11
},
{
"epoch": 0.04,
"learning_rate": 2.2e-07,
"loss": 4.9384,
"step": 12
},
{
"epoch": 0.04,
"learning_rate": 2.4000000000000003e-07,
"loss": 4.8481,
"step": 13
},
{
"epoch": 0.05,
"learning_rate": 2.6e-07,
"loss": 4.8025,
"step": 14
},
{
"epoch": 0.05,
"learning_rate": 2.8e-07,
"loss": 4.8327,
"step": 15
},
{
"epoch": 0.05,
"learning_rate": 3.0000000000000004e-07,
"loss": 4.7626,
"step": 16
},
{
"epoch": 0.06,
"learning_rate": 3.2e-07,
"loss": 4.8658,
"step": 17
},
{
"epoch": 0.06,
"learning_rate": 3.4000000000000003e-07,
"loss": 4.8883,
"step": 18
},
{
"epoch": 0.06,
"learning_rate": 3.6e-07,
"loss": 4.8363,
"step": 19
},
{
"epoch": 0.07,
"learning_rate": 3.8e-07,
"loss": 4.9294,
"step": 20
},
{
"epoch": 0.07,
"learning_rate": 4.0000000000000003e-07,
"loss": 4.9403,
"step": 21
},
{
"epoch": 0.07,
"learning_rate": 4.2000000000000006e-07,
"loss": 4.8762,
"step": 22
},
{
"epoch": 0.08,
"learning_rate": 4.4e-07,
"loss": 4.8012,
"step": 23
},
{
"epoch": 0.08,
"learning_rate": 4.6000000000000004e-07,
"loss": 4.9476,
"step": 24
},
{
"epoch": 0.08,
"learning_rate": 4.800000000000001e-07,
"loss": 4.7731,
"step": 25
},
{
"epoch": 0.09,
"learning_rate": 5.000000000000001e-07,
"loss": 5.086,
"step": 26
},
{
"epoch": 0.09,
"learning_rate": 5.2e-07,
"loss": 4.8115,
"step": 27
},
{
"epoch": 0.09,
"learning_rate": 5.4e-07,
"loss": 4.8335,
"step": 28
},
{
"epoch": 0.1,
"learning_rate": 5.6e-07,
"loss": 4.7309,
"step": 29
},
{
"epoch": 0.1,
"learning_rate": 5.800000000000001e-07,
"loss": 4.8824,
"step": 30
},
{
"epoch": 0.1,
"learning_rate": 6.000000000000001e-07,
"loss": 4.875,
"step": 31
},
{
"epoch": 0.11,
"learning_rate": 6.200000000000001e-07,
"loss": 4.8,
"step": 32
},
{
"epoch": 0.11,
"learning_rate": 6.4e-07,
"loss": 4.8361,
"step": 33
},
{
"epoch": 0.11,
"learning_rate": 6.6e-07,
"loss": 4.7767,
"step": 34
},
{
"epoch": 0.12,
"learning_rate": 6.800000000000001e-07,
"loss": 4.8105,
"step": 35
},
{
"epoch": 0.12,
"learning_rate": 7.000000000000001e-07,
"loss": 4.7742,
"step": 36
},
{
"epoch": 0.12,
"learning_rate": 7.2e-07,
"loss": 4.9031,
"step": 37
},
{
"epoch": 0.13,
"learning_rate": 7.4e-07,
"loss": 4.8678,
"step": 38
},
{
"epoch": 0.13,
"learning_rate": 7.6e-07,
"loss": 4.9272,
"step": 39
},
{
"epoch": 0.13,
"learning_rate": 7.8e-07,
"loss": 4.764,
"step": 40
},
{
"epoch": 0.14,
"learning_rate": 8.000000000000001e-07,
"loss": 4.8599,
"step": 41
},
{
"epoch": 0.14,
"learning_rate": 8.200000000000001e-07,
"loss": 4.9507,
"step": 42
},
{
"epoch": 0.14,
"learning_rate": 8.400000000000001e-07,
"loss": 4.7209,
"step": 43
},
{
"epoch": 0.15,
"learning_rate": 8.6e-07,
"loss": 5.0393,
"step": 44
},
{
"epoch": 0.15,
"learning_rate": 8.8e-07,
"loss": 5.0139,
"step": 45
},
{
"epoch": 0.15,
"learning_rate": 9.000000000000001e-07,
"loss": 5.0851,
"step": 46
},
{
"epoch": 0.16,
"learning_rate": 9.200000000000001e-07,
"loss": 5.1161,
"step": 47
},
{
"epoch": 0.16,
"learning_rate": 9.400000000000001e-07,
"loss": 5.2155,
"step": 48
},
{
"epoch": 0.16,
"learning_rate": 9.600000000000001e-07,
"loss": 5.2669,
"step": 49
},
{
"epoch": 0.17,
"learning_rate": 9.800000000000001e-07,
"loss": 5.1875,
"step": 50
},
{
"epoch": 0.17,
"learning_rate": 1.0000000000000002e-06,
"loss": 4.673,
"step": 51
},
{
"epoch": 0.17,
"learning_rate": 1.02e-06,
"loss": 4.5594,
"step": 52
},
{
"epoch": 0.18,
"learning_rate": 1.04e-06,
"loss": 4.5074,
"step": 53
},
{
"epoch": 0.18,
"learning_rate": 1.06e-06,
"loss": 4.5297,
"step": 54
},
{
"epoch": 0.18,
"learning_rate": 1.08e-06,
"loss": 4.4645,
"step": 55
},
{
"epoch": 0.19,
"learning_rate": 1.1e-06,
"loss": 4.4743,
"step": 56
},
{
"epoch": 0.19,
"learning_rate": 1.12e-06,
"loss": 4.6112,
"step": 57
},
{
"epoch": 0.2,
"learning_rate": 1.14e-06,
"loss": 4.6401,
"step": 58
},
{
"epoch": 0.2,
"learning_rate": 1.1600000000000001e-06,
"loss": 4.6055,
"step": 59
},
{
"epoch": 0.2,
"learning_rate": 1.1800000000000001e-06,
"loss": 4.5181,
"step": 60
},
{
"epoch": 0.21,
"learning_rate": 1.2000000000000002e-06,
"loss": 4.5917,
"step": 61
},
{
"epoch": 0.21,
"learning_rate": 1.2200000000000002e-06,
"loss": 4.4533,
"step": 62
},
{
"epoch": 0.21,
"learning_rate": 1.2400000000000002e-06,
"loss": 4.5889,
"step": 63
},
{
"epoch": 0.22,
"learning_rate": 1.26e-06,
"loss": 4.5351,
"step": 64
},
{
"epoch": 0.22,
"learning_rate": 1.28e-06,
"loss": 4.5995,
"step": 65
},
{
"epoch": 0.22,
"learning_rate": 1.3e-06,
"loss": 4.5842,
"step": 66
},
{
"epoch": 0.23,
"learning_rate": 1.32e-06,
"loss": 4.5496,
"step": 67
},
{
"epoch": 0.23,
"learning_rate": 1.34e-06,
"loss": 4.5796,
"step": 68
},
{
"epoch": 0.23,
"learning_rate": 1.3600000000000001e-06,
"loss": 4.5204,
"step": 69
},
{
"epoch": 0.24,
"learning_rate": 1.3800000000000001e-06,
"loss": 4.4595,
"step": 70
},
{
"epoch": 0.24,
"learning_rate": 1.4000000000000001e-06,
"loss": 4.4492,
"step": 71
},
{
"epoch": 0.24,
"learning_rate": 1.42e-06,
"loss": 4.5821,
"step": 72
},
{
"epoch": 0.25,
"learning_rate": 1.44e-06,
"loss": 4.4998,
"step": 73
},
{
"epoch": 0.25,
"learning_rate": 1.46e-06,
"loss": 4.4812,
"step": 74
},
{
"epoch": 0.25,
"learning_rate": 1.48e-06,
"loss": 4.5778,
"step": 75
},
{
"epoch": 0.26,
"learning_rate": 1.5e-06,
"loss": 4.4696,
"step": 76
},
{
"epoch": 0.26,
"learning_rate": 1.52e-06,
"loss": 4.4308,
"step": 77
},
{
"epoch": 0.26,
"learning_rate": 1.54e-06,
"loss": 4.4694,
"step": 78
},
{
"epoch": 0.27,
"learning_rate": 1.56e-06,
"loss": 4.3769,
"step": 79
},
{
"epoch": 0.27,
"learning_rate": 1.5800000000000001e-06,
"loss": 4.5486,
"step": 80
},
{
"epoch": 0.27,
"learning_rate": 1.6000000000000001e-06,
"loss": 4.4755,
"step": 81
},
{
"epoch": 0.28,
"learning_rate": 1.6200000000000002e-06,
"loss": 4.4662,
"step": 82
},
{
"epoch": 0.28,
"learning_rate": 1.6400000000000002e-06,
"loss": 4.4412,
"step": 83
},
{
"epoch": 0.28,
"learning_rate": 1.6600000000000002e-06,
"loss": 4.4749,
"step": 84
},
{
"epoch": 0.29,
"learning_rate": 1.6800000000000002e-06,
"loss": 4.5525,
"step": 85
},
{
"epoch": 0.29,
"learning_rate": 1.7000000000000002e-06,
"loss": 4.4525,
"step": 86
},
{
"epoch": 0.29,
"learning_rate": 1.72e-06,
"loss": 4.5006,
"step": 87
},
{
"epoch": 0.3,
"learning_rate": 1.74e-06,
"loss": 4.5137,
"step": 88
},
{
"epoch": 0.3,
"learning_rate": 1.76e-06,
"loss": 4.549,
"step": 89
},
{
"epoch": 0.3,
"learning_rate": 1.7800000000000001e-06,
"loss": 4.5675,
"step": 90
},
{
"epoch": 0.31,
"learning_rate": 1.8000000000000001e-06,
"loss": 4.595,
"step": 91
},
{
"epoch": 0.31,
"learning_rate": 1.8200000000000002e-06,
"loss": 4.6441,
"step": 92
},
{
"epoch": 0.31,
"learning_rate": 1.8400000000000002e-06,
"loss": 4.5773,
"step": 93
},
{
"epoch": 0.32,
"learning_rate": 1.8600000000000002e-06,
"loss": 4.522,
"step": 94
},
{
"epoch": 0.32,
"learning_rate": 1.8800000000000002e-06,
"loss": 4.4908,
"step": 95
},
{
"epoch": 0.32,
"learning_rate": 1.9000000000000002e-06,
"loss": 4.6946,
"step": 96
},
{
"epoch": 0.33,
"learning_rate": 1.9200000000000003e-06,
"loss": 4.8747,
"step": 97
},
{
"epoch": 0.33,
"learning_rate": 1.94e-06,
"loss": 4.8247,
"step": 98
},
{
"epoch": 0.33,
"learning_rate": 1.9600000000000003e-06,
"loss": 4.786,
"step": 99
},
{
"epoch": 0.34,
"learning_rate": 1.98e-06,
"loss": 4.8753,
"step": 100
},
{
"epoch": 0.34,
"learning_rate": 2.0000000000000003e-06,
"loss": 4.332,
"step": 101
},
{
"epoch": 0.34,
"learning_rate": 2.02e-06,
"loss": 4.3795,
"step": 102
},
{
"epoch": 0.35,
"learning_rate": 2.04e-06,
"loss": 4.3807,
"step": 103
},
{
"epoch": 0.35,
"learning_rate": 2.06e-06,
"loss": 4.2868,
"step": 104
},
{
"epoch": 0.35,
"learning_rate": 2.08e-06,
"loss": 4.3938,
"step": 105
},
{
"epoch": 0.36,
"learning_rate": 2.1000000000000002e-06,
"loss": 4.3391,
"step": 106
},
{
"epoch": 0.36,
"learning_rate": 2.12e-06,
"loss": 4.2794,
"step": 107
},
{
"epoch": 0.36,
"learning_rate": 2.1400000000000003e-06,
"loss": 4.3783,
"step": 108
},
{
"epoch": 0.37,
"learning_rate": 2.16e-06,
"loss": 4.382,
"step": 109
},
{
"epoch": 0.37,
"learning_rate": 2.1800000000000003e-06,
"loss": 4.3473,
"step": 110
},
{
"epoch": 0.37,
"learning_rate": 2.2e-06,
"loss": 4.4273,
"step": 111
},
{
"epoch": 0.38,
"learning_rate": 2.2200000000000003e-06,
"loss": 4.3599,
"step": 112
},
{
"epoch": 0.38,
"learning_rate": 2.24e-06,
"loss": 4.2758,
"step": 113
},
{
"epoch": 0.38,
"learning_rate": 2.2600000000000004e-06,
"loss": 4.3514,
"step": 114
},
{
"epoch": 0.39,
"learning_rate": 2.28e-06,
"loss": 4.3079,
"step": 115
},
{
"epoch": 0.39,
"learning_rate": 2.3000000000000004e-06,
"loss": 4.3189,
"step": 116
},
{
"epoch": 0.39,
"learning_rate": 2.3200000000000002e-06,
"loss": 4.2784,
"step": 117
},
{
"epoch": 0.4,
"learning_rate": 2.3400000000000005e-06,
"loss": 4.3314,
"step": 118
},
{
"epoch": 0.4,
"learning_rate": 2.3600000000000003e-06,
"loss": 4.3653,
"step": 119
},
{
"epoch": 0.4,
"learning_rate": 2.38e-06,
"loss": 4.36,
"step": 120
},
{
"epoch": 0.41,
"learning_rate": 2.4000000000000003e-06,
"loss": 4.3902,
"step": 121
},
{
"epoch": 0.41,
"learning_rate": 2.42e-06,
"loss": 4.4351,
"step": 122
},
{
"epoch": 0.41,
"learning_rate": 2.4400000000000004e-06,
"loss": 4.2534,
"step": 123
},
{
"epoch": 0.42,
"learning_rate": 2.46e-06,
"loss": 4.3887,
"step": 124
},
{
"epoch": 0.42,
"learning_rate": 2.4800000000000004e-06,
"loss": 4.4345,
"step": 125
},
{
"epoch": 0.42,
"learning_rate": 2.5e-06,
"loss": 4.3609,
"step": 126
},
{
"epoch": 0.43,
"learning_rate": 2.52e-06,
"loss": 4.4055,
"step": 127
},
{
"epoch": 0.43,
"learning_rate": 2.5400000000000002e-06,
"loss": 4.2582,
"step": 128
},
{
"epoch": 0.43,
"learning_rate": 2.56e-06,
"loss": 4.4031,
"step": 129
},
{
"epoch": 0.44,
"learning_rate": 2.5800000000000003e-06,
"loss": 4.4855,
"step": 130
},
{
"epoch": 0.44,
"learning_rate": 2.6e-06,
"loss": 4.4102,
"step": 131
},
{
"epoch": 0.44,
"learning_rate": 2.6200000000000003e-06,
"loss": 4.2256,
"step": 132
},
{
"epoch": 0.45,
"learning_rate": 2.64e-06,
"loss": 4.4007,
"step": 133
},
{
"epoch": 0.45,
"learning_rate": 2.6600000000000004e-06,
"loss": 4.2806,
"step": 134
},
{
"epoch": 0.45,
"learning_rate": 2.68e-06,
"loss": 4.4708,
"step": 135
},
{
"epoch": 0.46,
"learning_rate": 2.7000000000000004e-06,
"loss": 4.3876,
"step": 136
},
{
"epoch": 0.46,
"learning_rate": 2.7200000000000002e-06,
"loss": 4.3317,
"step": 137
},
{
"epoch": 0.46,
"learning_rate": 2.7400000000000004e-06,
"loss": 4.2371,
"step": 138
},
{
"epoch": 0.47,
"learning_rate": 2.7600000000000003e-06,
"loss": 4.2725,
"step": 139
},
{
"epoch": 0.47,
"learning_rate": 2.7800000000000005e-06,
"loss": 4.3226,
"step": 140
},
{
"epoch": 0.47,
"learning_rate": 2.8000000000000003e-06,
"loss": 4.5303,
"step": 141
},
{
"epoch": 0.48,
"learning_rate": 2.82e-06,
"loss": 4.2812,
"step": 142
},
{
"epoch": 0.48,
"learning_rate": 2.84e-06,
"loss": 4.3971,
"step": 143
},
{
"epoch": 0.48,
"learning_rate": 2.86e-06,
"loss": 4.426,
"step": 144
},
{
"epoch": 0.49,
"learning_rate": 2.88e-06,
"loss": 4.4879,
"step": 145
},
{
"epoch": 0.49,
"learning_rate": 2.9e-06,
"loss": 4.3671,
"step": 146
},
{
"epoch": 0.49,
"learning_rate": 2.92e-06,
"loss": 4.4797,
"step": 147
},
{
"epoch": 0.5,
"learning_rate": 2.9400000000000002e-06,
"loss": 4.6502,
"step": 148
},
{
"epoch": 0.5,
"learning_rate": 2.96e-06,
"loss": 4.6188,
"step": 149
},
{
"epoch": 0.5,
"learning_rate": 2.9800000000000003e-06,
"loss": 4.7733,
"step": 150
},
{
"epoch": 0.51,
"learning_rate": 3e-06,
"loss": 4.3361,
"step": 151
},
{
"epoch": 0.51,
"learning_rate": 3.0200000000000003e-06,
"loss": 4.3216,
"step": 152
},
{
"epoch": 0.51,
"learning_rate": 3.04e-06,
"loss": 4.3243,
"step": 153
},
{
"epoch": 0.52,
"learning_rate": 3.0600000000000003e-06,
"loss": 4.2681,
"step": 154
},
{
"epoch": 0.52,
"learning_rate": 3.08e-06,
"loss": 4.2981,
"step": 155
},
{
"epoch": 0.52,
"learning_rate": 3.1000000000000004e-06,
"loss": 4.3336,
"step": 156
},
{
"epoch": 0.53,
"learning_rate": 3.12e-06,
"loss": 4.2307,
"step": 157
},
{
"epoch": 0.53,
"learning_rate": 3.1400000000000004e-06,
"loss": 4.3043,
"step": 158
},
{
"epoch": 0.53,
"learning_rate": 3.1600000000000002e-06,
"loss": 4.28,
"step": 159
},
{
"epoch": 0.54,
"learning_rate": 3.1800000000000005e-06,
"loss": 4.2736,
"step": 160
},
{
"epoch": 0.54,
"learning_rate": 3.2000000000000003e-06,
"loss": 4.2756,
"step": 161
},
{
"epoch": 0.54,
"learning_rate": 3.2200000000000005e-06,
"loss": 4.2625,
"step": 162
},
{
"epoch": 0.55,
"learning_rate": 3.2400000000000003e-06,
"loss": 4.3129,
"step": 163
},
{
"epoch": 0.55,
"learning_rate": 3.2600000000000006e-06,
"loss": 4.2453,
"step": 164
},
{
"epoch": 0.55,
"learning_rate": 3.2800000000000004e-06,
"loss": 4.3079,
"step": 165
},
{
"epoch": 0.56,
"learning_rate": 3.3000000000000006e-06,
"loss": 4.2754,
"step": 166
},
{
"epoch": 0.56,
"learning_rate": 3.3200000000000004e-06,
"loss": 4.3026,
"step": 167
},
{
"epoch": 0.56,
"learning_rate": 3.3400000000000006e-06,
"loss": 4.2798,
"step": 168
},
{
"epoch": 0.57,
"learning_rate": 3.3600000000000004e-06,
"loss": 4.2466,
"step": 169
},
{
"epoch": 0.57,
"learning_rate": 3.3800000000000007e-06,
"loss": 4.1995,
"step": 170
},
{
"epoch": 0.58,
"learning_rate": 3.4000000000000005e-06,
"loss": 4.1824,
"step": 171
},
{
"epoch": 0.58,
"learning_rate": 3.4200000000000007e-06,
"loss": 4.2936,
"step": 172
},
{
"epoch": 0.58,
"learning_rate": 3.44e-06,
"loss": 4.1523,
"step": 173
},
{
"epoch": 0.59,
"learning_rate": 3.46e-06,
"loss": 4.2063,
"step": 174
},
{
"epoch": 0.59,
"learning_rate": 3.48e-06,
"loss": 4.3818,
"step": 175
},
{
"epoch": 0.59,
"learning_rate": 3.5e-06,
"loss": 4.2121,
"step": 176
},
{
"epoch": 0.6,
"learning_rate": 3.52e-06,
"loss": 4.35,
"step": 177
},
{
"epoch": 0.6,
"learning_rate": 3.54e-06,
"loss": 4.2109,
"step": 178
},
{
"epoch": 0.6,
"learning_rate": 3.5600000000000002e-06,
"loss": 4.2574,
"step": 179
},
{
"epoch": 0.61,
"learning_rate": 3.58e-06,
"loss": 4.3989,
"step": 180
},
{
"epoch": 0.61,
"learning_rate": 3.6000000000000003e-06,
"loss": 4.2638,
"step": 181
},
{
"epoch": 0.61,
"learning_rate": 3.62e-06,
"loss": 4.2118,
"step": 182
},
{
"epoch": 0.62,
"learning_rate": 3.6400000000000003e-06,
"loss": 4.1813,
"step": 183
},
{
"epoch": 0.62,
"learning_rate": 3.66e-06,
"loss": 4.2778,
"step": 184
},
{
"epoch": 0.62,
"learning_rate": 3.6800000000000003e-06,
"loss": 4.3297,
"step": 185
},
{
"epoch": 0.63,
"learning_rate": 3.7e-06,
"loss": 4.272,
"step": 186
},
{
"epoch": 0.63,
"learning_rate": 3.7200000000000004e-06,
"loss": 4.3077,
"step": 187
},
{
"epoch": 0.63,
"learning_rate": 3.74e-06,
"loss": 4.2812,
"step": 188
},
{
"epoch": 0.64,
"learning_rate": 3.7600000000000004e-06,
"loss": 4.2457,
"step": 189
},
{
"epoch": 0.64,
"learning_rate": 3.7800000000000002e-06,
"loss": 4.3328,
"step": 190
},
{
"epoch": 0.64,
"learning_rate": 3.8000000000000005e-06,
"loss": 4.3663,
"step": 191
},
{
"epoch": 0.65,
"learning_rate": 3.820000000000001e-06,
"loss": 4.4388,
"step": 192
},
{
"epoch": 0.65,
"learning_rate": 3.8400000000000005e-06,
"loss": 4.5001,
"step": 193
},
{
"epoch": 0.65,
"learning_rate": 3.86e-06,
"loss": 4.4342,
"step": 194
},
{
"epoch": 0.66,
"learning_rate": 3.88e-06,
"loss": 4.4113,
"step": 195
},
{
"epoch": 0.66,
"learning_rate": 3.900000000000001e-06,
"loss": 4.3612,
"step": 196
},
{
"epoch": 0.66,
"learning_rate": 3.920000000000001e-06,
"loss": 4.2716,
"step": 197
},
{
"epoch": 0.67,
"learning_rate": 3.94e-06,
"loss": 4.7501,
"step": 198
},
{
"epoch": 0.67,
"learning_rate": 3.96e-06,
"loss": 4.5016,
"step": 199
},
{
"epoch": 0.67,
"learning_rate": 3.980000000000001e-06,
"loss": 4.7116,
"step": 200
},
{
"epoch": 0.68,
"learning_rate": 4.000000000000001e-06,
"loss": 4.3106,
"step": 201
},
{
"epoch": 0.68,
"learning_rate": 4.0200000000000005e-06,
"loss": 4.2625,
"step": 202
},
{
"epoch": 0.68,
"learning_rate": 4.04e-06,
"loss": 4.2393,
"step": 203
},
{
"epoch": 0.69,
"learning_rate": 4.060000000000001e-06,
"loss": 4.1657,
"step": 204
},
{
"epoch": 0.69,
"learning_rate": 4.08e-06,
"loss": 4.2338,
"step": 205
},
{
"epoch": 0.69,
"learning_rate": 4.1e-06,
"loss": 4.229,
"step": 206
},
{
"epoch": 0.7,
"learning_rate": 4.12e-06,
"loss": 4.2768,
"step": 207
},
{
"epoch": 0.7,
"learning_rate": 4.14e-06,
"loss": 4.245,
"step": 208
},
{
"epoch": 0.7,
"learning_rate": 4.16e-06,
"loss": 4.2213,
"step": 209
},
{
"epoch": 0.71,
"learning_rate": 4.18e-06,
"loss": 4.3171,
"step": 210
},
{
"epoch": 0.71,
"learning_rate": 4.2000000000000004e-06,
"loss": 4.1127,
"step": 211
},
{
"epoch": 0.71,
"learning_rate": 4.22e-06,
"loss": 4.2402,
"step": 212
},
{
"epoch": 0.72,
"learning_rate": 4.24e-06,
"loss": 4.265,
"step": 213
},
{
"epoch": 0.72,
"learning_rate": 4.26e-06,
"loss": 4.2684,
"step": 214
},
{
"epoch": 0.72,
"learning_rate": 4.2800000000000005e-06,
"loss": 4.2735,
"step": 215
},
{
"epoch": 0.73,
"learning_rate": 4.3e-06,
"loss": 4.2555,
"step": 216
},
{
"epoch": 0.73,
"learning_rate": 4.32e-06,
"loss": 4.2211,
"step": 217
},
{
"epoch": 0.73,
"learning_rate": 4.34e-06,
"loss": 4.1925,
"step": 218
},
{
"epoch": 0.74,
"learning_rate": 4.360000000000001e-06,
"loss": 4.1934,
"step": 219
},
{
"epoch": 0.74,
"learning_rate": 4.38e-06,
"loss": 4.2549,
"step": 220
},
{
"epoch": 0.74,
"learning_rate": 4.4e-06,
"loss": 4.1733,
"step": 221
},
{
"epoch": 0.75,
"learning_rate": 4.42e-06,
"loss": 4.1752,
"step": 222
},
{
"epoch": 0.75,
"learning_rate": 4.440000000000001e-06,
"loss": 4.1934,
"step": 223
},
{
"epoch": 0.75,
"learning_rate": 4.4600000000000005e-06,
"loss": 4.2419,
"step": 224
},
{
"epoch": 0.76,
"learning_rate": 4.48e-06,
"loss": 4.2159,
"step": 225
},
{
"epoch": 0.76,
"learning_rate": 4.5e-06,
"loss": 4.226,
"step": 226
},
{
"epoch": 0.76,
"learning_rate": 4.520000000000001e-06,
"loss": 4.2576,
"step": 227
},
{
"epoch": 0.77,
"learning_rate": 4.540000000000001e-06,
"loss": 4.1938,
"step": 228
},
{
"epoch": 0.77,
"learning_rate": 4.56e-06,
"loss": 4.2063,
"step": 229
},
{
"epoch": 0.77,
"learning_rate": 4.58e-06,
"loss": 4.2194,
"step": 230
},
{
"epoch": 0.78,
"learning_rate": 4.600000000000001e-06,
"loss": 4.3171,
"step": 231
},
{
"epoch": 0.78,
"learning_rate": 4.620000000000001e-06,
"loss": 4.1888,
"step": 232
},
{
"epoch": 0.78,
"learning_rate": 4.6400000000000005e-06,
"loss": 4.2994,
"step": 233
},
{
"epoch": 0.79,
"learning_rate": 4.66e-06,
"loss": 4.2743,
"step": 234
},
{
"epoch": 0.79,
"learning_rate": 4.680000000000001e-06,
"loss": 4.3265,
"step": 235
},
{
"epoch": 0.79,
"learning_rate": 4.7e-06,
"loss": 4.146,
"step": 236
},
{
"epoch": 0.8,
"learning_rate": 4.7200000000000005e-06,
"loss": 4.2531,
"step": 237
},
{
"epoch": 0.8,
"learning_rate": 4.74e-06,
"loss": 4.3796,
"step": 238
},
{
"epoch": 0.8,
"learning_rate": 4.76e-06,
"loss": 4.2271,
"step": 239
},
{
"epoch": 0.81,
"learning_rate": 4.78e-06,
"loss": 4.2738,
"step": 240
},
{
"epoch": 0.81,
"learning_rate": 4.800000000000001e-06,
"loss": 4.3837,
"step": 241
},
{
"epoch": 0.81,
"learning_rate": 4.8200000000000004e-06,
"loss": 4.3178,
"step": 242
},
{
"epoch": 0.82,
"learning_rate": 4.84e-06,
"loss": 4.2609,
"step": 243
},
{
"epoch": 0.82,
"learning_rate": 4.86e-06,
"loss": 4.2858,
"step": 244
},
{
"epoch": 0.82,
"learning_rate": 4.880000000000001e-06,
"loss": 4.2618,
"step": 245
},
{
"epoch": 0.83,
"learning_rate": 4.9000000000000005e-06,
"loss": 4.4445,
"step": 246
},
{
"epoch": 0.83,
"learning_rate": 4.92e-06,
"loss": 4.3901,
"step": 247
},
{
"epoch": 0.83,
"learning_rate": 4.94e-06,
"loss": 4.3951,
"step": 248
},
{
"epoch": 0.84,
"learning_rate": 4.960000000000001e-06,
"loss": 4.5076,
"step": 249
},
{
"epoch": 0.84,
"learning_rate": 4.980000000000001e-06,
"loss": 4.5527,
"step": 250
},
{
"epoch": 0.84,
"learning_rate": 5e-06,
"loss": 4.2833,
"step": 251
},
{
"epoch": 0.85,
"learning_rate": 5.02e-06,
"loss": 4.206,
"step": 252
},
{
"epoch": 0.85,
"learning_rate": 5.04e-06,
"loss": 4.1542,
"step": 253
},
{
"epoch": 0.85,
"learning_rate": 5.060000000000001e-06,
"loss": 4.1541,
"step": 254
},
{
"epoch": 0.86,
"learning_rate": 5.0800000000000005e-06,
"loss": 4.1914,
"step": 255
},
{
"epoch": 0.86,
"learning_rate": 5.1e-06,
"loss": 4.1824,
"step": 256
},
{
"epoch": 0.86,
"learning_rate": 5.12e-06,
"loss": 4.3117,
"step": 257
},
{
"epoch": 0.87,
"learning_rate": 5.140000000000001e-06,
"loss": 4.192,
"step": 258
},
{
"epoch": 0.87,
"learning_rate": 5.1600000000000006e-06,
"loss": 4.2298,
"step": 259
},
{
"epoch": 0.87,
"learning_rate": 5.18e-06,
"loss": 4.1561,
"step": 260
},
{
"epoch": 0.88,
"learning_rate": 5.2e-06,
"loss": 4.2479,
"step": 261
},
{
"epoch": 0.88,
"learning_rate": 5.220000000000001e-06,
"loss": 4.2713,
"step": 262
},
{
"epoch": 0.88,
"learning_rate": 5.240000000000001e-06,
"loss": 4.1502,
"step": 263
},
{
"epoch": 0.89,
"learning_rate": 5.2600000000000005e-06,
"loss": 4.1789,
"step": 264
},
{
"epoch": 0.89,
"learning_rate": 5.28e-06,
"loss": 4.2032,
"step": 265
},
{
"epoch": 0.89,
"learning_rate": 5.300000000000001e-06,
"loss": 4.2824,
"step": 266
},
{
"epoch": 0.9,
"learning_rate": 5.320000000000001e-06,
"loss": 4.2033,
"step": 267
},
{
"epoch": 0.9,
"learning_rate": 5.3400000000000005e-06,
"loss": 4.1714,
"step": 268
},
{
"epoch": 0.9,
"learning_rate": 5.36e-06,
"loss": 4.1364,
"step": 269
},
{
"epoch": 0.91,
"learning_rate": 5.380000000000001e-06,
"loss": 4.261,
"step": 270
},
{
"epoch": 0.91,
"learning_rate": 5.400000000000001e-06,
"loss": 4.1305,
"step": 271
},
{
"epoch": 0.91,
"learning_rate": 5.420000000000001e-06,
"loss": 4.1106,
"step": 272
},
{
"epoch": 0.92,
"learning_rate": 5.4400000000000004e-06,
"loss": 4.1723,
"step": 273
},
{
"epoch": 0.92,
"learning_rate": 5.460000000000001e-06,
"loss": 4.1495,
"step": 274
},
{
"epoch": 0.92,
"learning_rate": 5.480000000000001e-06,
"loss": 4.2855,
"step": 275
},
{
"epoch": 0.93,
"learning_rate": 5.500000000000001e-06,
"loss": 4.199,
"step": 276
},
{
"epoch": 0.93,
"learning_rate": 5.5200000000000005e-06,
"loss": 4.2175,
"step": 277
},
{
"epoch": 0.93,
"learning_rate": 5.540000000000001e-06,
"loss": 4.152,
"step": 278
},
{
"epoch": 0.94,
"learning_rate": 5.560000000000001e-06,
"loss": 4.1857,
"step": 279
},
{
"epoch": 0.94,
"learning_rate": 5.580000000000001e-06,
"loss": 4.1679,
"step": 280
},
{
"epoch": 0.94,
"learning_rate": 5.600000000000001e-06,
"loss": 4.2036,
"step": 281
},
{
"epoch": 0.95,
"learning_rate": 5.620000000000001e-06,
"loss": 4.2049,
"step": 282
},
{
"epoch": 0.95,
"learning_rate": 5.64e-06,
"loss": 4.3554,
"step": 283
},
{
"epoch": 0.96,
"learning_rate": 5.66e-06,
"loss": 4.1291,
"step": 284
},
{
"epoch": 0.96,
"learning_rate": 5.68e-06,
"loss": 4.2472,
"step": 285
},
{
"epoch": 0.96,
"learning_rate": 5.7e-06,
"loss": 4.2736,
"step": 286
},
{
"epoch": 0.97,
"learning_rate": 5.72e-06,
"loss": 4.1614,
"step": 287
},
{
"epoch": 0.97,
"learning_rate": 5.74e-06,
"loss": 4.2049,
"step": 288
},
{
"epoch": 0.97,
"learning_rate": 5.76e-06,
"loss": 4.206,
"step": 289
},
{
"epoch": 0.98,
"learning_rate": 5.78e-06,
"loss": 4.2975,
"step": 290
},
{
"epoch": 0.98,
"learning_rate": 5.8e-06,
"loss": 4.2788,
"step": 291
},
{
"epoch": 0.98,
"learning_rate": 5.82e-06,
"loss": 4.3599,
"step": 292
},
{
"epoch": 0.99,
"learning_rate": 5.84e-06,
"loss": 4.3441,
"step": 293
},
{
"epoch": 0.99,
"learning_rate": 5.86e-06,
"loss": 4.5265,
"step": 294
},
{
"epoch": 0.99,
"learning_rate": 5.8800000000000005e-06,
"loss": 4.3277,
"step": 295
},
{
"epoch": 1.0,
"learning_rate": 5.9e-06,
"loss": 4.6971,
"step": 296
},
{
"epoch": 1.0,
"learning_rate": 5.92e-06,
"loss": 4.5581,
"step": 297
},
{
"epoch": 1.0,
"step": 297,
"total_flos": 0.0,
"train_loss": 4.444549621556343,
"train_runtime": 4661.3109,
"train_samples_per_second": 6.122,
"train_steps_per_second": 0.064
}
],
"max_steps": 297,
"num_train_epochs": 1,
"total_flos": 0.0,
"trial_name": null,
"trial_params": null
}