wav2vec2-2-gpt2-grid-search / trainer_state.json
sanchit-gandhi's picture
End of training
d59ea5f
raw
history blame
No virus
51.1 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": 2e-08,
"loss": 4.7633,
"step": 1
},
{
"epoch": 0.0,
"learning_rate": 4e-08,
"loss": 4.9111,
"step": 2
},
{
"epoch": 0.01,
"learning_rate": 4e-08,
"loss": 4.8796,
"step": 3
},
{
"epoch": 0.01,
"learning_rate": 6.000000000000001e-08,
"loss": 4.8776,
"step": 4
},
{
"epoch": 0.01,
"learning_rate": 8e-08,
"loss": 4.8182,
"step": 5
},
{
"epoch": 0.01,
"learning_rate": 1.0000000000000001e-07,
"loss": 4.8851,
"step": 6
},
{
"epoch": 0.02,
"learning_rate": 1.2000000000000002e-07,
"loss": 4.8575,
"step": 7
},
{
"epoch": 0.02,
"learning_rate": 1.4e-07,
"loss": 4.9088,
"step": 8
},
{
"epoch": 0.02,
"learning_rate": 1.6e-07,
"loss": 4.9109,
"step": 9
},
{
"epoch": 0.02,
"learning_rate": 1.8e-07,
"loss": 4.7878,
"step": 10
},
{
"epoch": 0.02,
"learning_rate": 2.0000000000000002e-07,
"loss": 4.9099,
"step": 11
},
{
"epoch": 0.03,
"learning_rate": 2.2e-07,
"loss": 4.8879,
"step": 12
},
{
"epoch": 0.03,
"learning_rate": 2.4000000000000003e-07,
"loss": 4.8415,
"step": 13
},
{
"epoch": 0.03,
"learning_rate": 2.6e-07,
"loss": 4.8503,
"step": 14
},
{
"epoch": 0.03,
"learning_rate": 2.8e-07,
"loss": 4.7343,
"step": 15
},
{
"epoch": 0.04,
"learning_rate": 3.0000000000000004e-07,
"loss": 4.8936,
"step": 16
},
{
"epoch": 0.04,
"learning_rate": 3.2e-07,
"loss": 4.8205,
"step": 17
},
{
"epoch": 0.04,
"learning_rate": 3.4000000000000003e-07,
"loss": 4.9299,
"step": 18
},
{
"epoch": 0.04,
"learning_rate": 3.6e-07,
"loss": 4.7331,
"step": 19
},
{
"epoch": 0.04,
"learning_rate": 3.8e-07,
"loss": 4.9161,
"step": 20
},
{
"epoch": 0.05,
"learning_rate": 4.0000000000000003e-07,
"loss": 4.91,
"step": 21
},
{
"epoch": 0.05,
"learning_rate": 4.2000000000000006e-07,
"loss": 4.8262,
"step": 22
},
{
"epoch": 0.05,
"learning_rate": 4.4e-07,
"loss": 4.9479,
"step": 23
},
{
"epoch": 0.05,
"learning_rate": 4.6000000000000004e-07,
"loss": 4.942,
"step": 24
},
{
"epoch": 0.06,
"learning_rate": 4.800000000000001e-07,
"loss": 4.7356,
"step": 25
},
{
"epoch": 0.06,
"learning_rate": 5.000000000000001e-07,
"loss": 5.0038,
"step": 26
},
{
"epoch": 0.06,
"learning_rate": 5.2e-07,
"loss": 4.8565,
"step": 27
},
{
"epoch": 0.06,
"learning_rate": 5.4e-07,
"loss": 4.7324,
"step": 28
},
{
"epoch": 0.07,
"learning_rate": 5.6e-07,
"loss": 4.7126,
"step": 29
},
{
"epoch": 0.07,
"learning_rate": 5.800000000000001e-07,
"loss": 4.9015,
"step": 30
},
{
"epoch": 0.07,
"learning_rate": 6.000000000000001e-07,
"loss": 4.869,
"step": 31
},
{
"epoch": 0.07,
"learning_rate": 6.200000000000001e-07,
"loss": 4.7292,
"step": 32
},
{
"epoch": 0.07,
"learning_rate": 6.4e-07,
"loss": 4.8362,
"step": 33
},
{
"epoch": 0.08,
"learning_rate": 6.6e-07,
"loss": 4.75,
"step": 34
},
{
"epoch": 0.08,
"learning_rate": 6.800000000000001e-07,
"loss": 4.8524,
"step": 35
},
{
"epoch": 0.08,
"learning_rate": 7.000000000000001e-07,
"loss": 4.7962,
"step": 36
},
{
"epoch": 0.08,
"learning_rate": 7.2e-07,
"loss": 4.859,
"step": 37
},
{
"epoch": 0.09,
"learning_rate": 7.4e-07,
"loss": 4.7945,
"step": 38
},
{
"epoch": 0.09,
"learning_rate": 7.6e-07,
"loss": 4.9263,
"step": 39
},
{
"epoch": 0.09,
"learning_rate": 7.8e-07,
"loss": 4.7248,
"step": 40
},
{
"epoch": 0.09,
"learning_rate": 8.000000000000001e-07,
"loss": 4.803,
"step": 41
},
{
"epoch": 0.09,
"learning_rate": 8.200000000000001e-07,
"loss": 4.9282,
"step": 42
},
{
"epoch": 0.1,
"learning_rate": 8.400000000000001e-07,
"loss": 4.7007,
"step": 43
},
{
"epoch": 0.1,
"learning_rate": 8.6e-07,
"loss": 4.8718,
"step": 44
},
{
"epoch": 0.1,
"learning_rate": 8.8e-07,
"loss": 5.0068,
"step": 45
},
{
"epoch": 0.1,
"learning_rate": 9.000000000000001e-07,
"loss": 5.1512,
"step": 46
},
{
"epoch": 0.11,
"learning_rate": 9.200000000000001e-07,
"loss": 5.0423,
"step": 47
},
{
"epoch": 0.11,
"learning_rate": 9.400000000000001e-07,
"loss": 5.3227,
"step": 48
},
{
"epoch": 0.11,
"learning_rate": 9.600000000000001e-07,
"loss": 5.4297,
"step": 49
},
{
"epoch": 0.11,
"learning_rate": 9.600000000000001e-07,
"loss": 5.339,
"step": 50
},
{
"epoch": 0.11,
"learning_rate": 9.800000000000001e-07,
"loss": 4.657,
"step": 51
},
{
"epoch": 0.12,
"learning_rate": 1.0000000000000002e-06,
"loss": 4.5155,
"step": 52
},
{
"epoch": 0.12,
"learning_rate": 1.02e-06,
"loss": 4.6886,
"step": 53
},
{
"epoch": 0.12,
"learning_rate": 1.04e-06,
"loss": 4.5949,
"step": 54
},
{
"epoch": 0.12,
"learning_rate": 1.06e-06,
"loss": 4.4149,
"step": 55
},
{
"epoch": 0.13,
"learning_rate": 1.08e-06,
"loss": 4.6266,
"step": 56
},
{
"epoch": 0.13,
"learning_rate": 1.1e-06,
"loss": 4.6458,
"step": 57
},
{
"epoch": 0.13,
"learning_rate": 1.12e-06,
"loss": 4.7385,
"step": 58
},
{
"epoch": 0.13,
"learning_rate": 1.14e-06,
"loss": 4.6687,
"step": 59
},
{
"epoch": 0.13,
"learning_rate": 1.1600000000000001e-06,
"loss": 4.5012,
"step": 60
},
{
"epoch": 0.14,
"learning_rate": 1.1800000000000001e-06,
"loss": 4.6079,
"step": 61
},
{
"epoch": 0.14,
"learning_rate": 1.2000000000000002e-06,
"loss": 4.5349,
"step": 62
},
{
"epoch": 0.14,
"learning_rate": 1.2200000000000002e-06,
"loss": 4.5074,
"step": 63
},
{
"epoch": 0.14,
"learning_rate": 1.2400000000000002e-06,
"loss": 4.611,
"step": 64
},
{
"epoch": 0.15,
"learning_rate": 1.26e-06,
"loss": 4.565,
"step": 65
},
{
"epoch": 0.15,
"learning_rate": 1.28e-06,
"loss": 4.4438,
"step": 66
},
{
"epoch": 0.15,
"learning_rate": 1.3e-06,
"loss": 4.6036,
"step": 67
},
{
"epoch": 0.15,
"learning_rate": 1.32e-06,
"loss": 4.4427,
"step": 68
},
{
"epoch": 0.15,
"learning_rate": 1.34e-06,
"loss": 4.5095,
"step": 69
},
{
"epoch": 0.16,
"learning_rate": 1.3600000000000001e-06,
"loss": 4.4858,
"step": 70
},
{
"epoch": 0.16,
"learning_rate": 1.3800000000000001e-06,
"loss": 4.618,
"step": 71
},
{
"epoch": 0.16,
"learning_rate": 1.4000000000000001e-06,
"loss": 4.6007,
"step": 72
},
{
"epoch": 0.16,
"learning_rate": 1.42e-06,
"loss": 4.453,
"step": 73
},
{
"epoch": 0.17,
"learning_rate": 1.44e-06,
"loss": 4.519,
"step": 74
},
{
"epoch": 0.17,
"learning_rate": 1.46e-06,
"loss": 4.5242,
"step": 75
},
{
"epoch": 0.17,
"learning_rate": 1.48e-06,
"loss": 4.4702,
"step": 76
},
{
"epoch": 0.17,
"learning_rate": 1.5e-06,
"loss": 4.5255,
"step": 77
},
{
"epoch": 0.17,
"learning_rate": 1.52e-06,
"loss": 4.4652,
"step": 78
},
{
"epoch": 0.18,
"learning_rate": 1.54e-06,
"loss": 4.4498,
"step": 79
},
{
"epoch": 0.18,
"learning_rate": 1.56e-06,
"loss": 4.5705,
"step": 80
},
{
"epoch": 0.18,
"learning_rate": 1.5800000000000001e-06,
"loss": 4.5202,
"step": 81
},
{
"epoch": 0.18,
"learning_rate": 1.6000000000000001e-06,
"loss": 4.4593,
"step": 82
},
{
"epoch": 0.19,
"learning_rate": 1.6200000000000002e-06,
"loss": 4.5733,
"step": 83
},
{
"epoch": 0.19,
"learning_rate": 1.6400000000000002e-06,
"loss": 4.45,
"step": 84
},
{
"epoch": 0.19,
"learning_rate": 1.6600000000000002e-06,
"loss": 4.4868,
"step": 85
},
{
"epoch": 0.19,
"learning_rate": 1.6800000000000002e-06,
"loss": 4.48,
"step": 86
},
{
"epoch": 0.2,
"learning_rate": 1.7000000000000002e-06,
"loss": 4.5859,
"step": 87
},
{
"epoch": 0.2,
"learning_rate": 1.72e-06,
"loss": 4.4624,
"step": 88
},
{
"epoch": 0.2,
"learning_rate": 1.74e-06,
"loss": 4.6002,
"step": 89
},
{
"epoch": 0.2,
"learning_rate": 1.76e-06,
"loss": 4.5497,
"step": 90
},
{
"epoch": 0.2,
"learning_rate": 1.7800000000000001e-06,
"loss": 4.4896,
"step": 91
},
{
"epoch": 0.21,
"learning_rate": 1.8000000000000001e-06,
"loss": 4.7272,
"step": 92
},
{
"epoch": 0.21,
"learning_rate": 1.8200000000000002e-06,
"loss": 4.5785,
"step": 93
},
{
"epoch": 0.21,
"learning_rate": 1.8400000000000002e-06,
"loss": 4.5895,
"step": 94
},
{
"epoch": 0.21,
"learning_rate": 1.8600000000000002e-06,
"loss": 4.5834,
"step": 95
},
{
"epoch": 0.22,
"learning_rate": 1.8800000000000002e-06,
"loss": 4.7481,
"step": 96
},
{
"epoch": 0.22,
"learning_rate": 1.9000000000000002e-06,
"loss": 4.7947,
"step": 97
},
{
"epoch": 0.22,
"learning_rate": 1.9200000000000003e-06,
"loss": 4.567,
"step": 98
},
{
"epoch": 0.22,
"learning_rate": 1.94e-06,
"loss": 4.8742,
"step": 99
},
{
"epoch": 0.22,
"learning_rate": 1.9600000000000003e-06,
"loss": 4.6728,
"step": 100
},
{
"epoch": 0.23,
"learning_rate": 1.98e-06,
"loss": 4.4148,
"step": 101
},
{
"epoch": 0.23,
"learning_rate": 2.0000000000000003e-06,
"loss": 4.3493,
"step": 102
},
{
"epoch": 0.23,
"learning_rate": 2.02e-06,
"loss": 4.2165,
"step": 103
},
{
"epoch": 0.23,
"learning_rate": 2.04e-06,
"loss": 4.285,
"step": 104
},
{
"epoch": 0.24,
"learning_rate": 2.06e-06,
"loss": 4.2356,
"step": 105
},
{
"epoch": 0.24,
"learning_rate": 2.08e-06,
"loss": 4.278,
"step": 106
},
{
"epoch": 0.24,
"learning_rate": 2.1000000000000002e-06,
"loss": 4.38,
"step": 107
},
{
"epoch": 0.24,
"learning_rate": 2.12e-06,
"loss": 4.2958,
"step": 108
},
{
"epoch": 0.24,
"learning_rate": 2.1400000000000003e-06,
"loss": 4.2908,
"step": 109
},
{
"epoch": 0.25,
"learning_rate": 2.16e-06,
"loss": 4.3311,
"step": 110
},
{
"epoch": 0.25,
"learning_rate": 2.1800000000000003e-06,
"loss": 4.3749,
"step": 111
},
{
"epoch": 0.25,
"learning_rate": 2.2e-06,
"loss": 4.3236,
"step": 112
},
{
"epoch": 0.25,
"learning_rate": 2.2200000000000003e-06,
"loss": 4.348,
"step": 113
},
{
"epoch": 0.26,
"learning_rate": 2.24e-06,
"loss": 4.3211,
"step": 114
},
{
"epoch": 0.26,
"learning_rate": 2.2600000000000004e-06,
"loss": 4.3697,
"step": 115
},
{
"epoch": 0.26,
"learning_rate": 2.28e-06,
"loss": 4.4212,
"step": 116
},
{
"epoch": 0.26,
"learning_rate": 2.3000000000000004e-06,
"loss": 4.3025,
"step": 117
},
{
"epoch": 0.26,
"learning_rate": 2.3200000000000002e-06,
"loss": 4.4696,
"step": 118
},
{
"epoch": 0.27,
"learning_rate": 2.3400000000000005e-06,
"loss": 4.3551,
"step": 119
},
{
"epoch": 0.27,
"learning_rate": 2.3600000000000003e-06,
"loss": 4.3851,
"step": 120
},
{
"epoch": 0.27,
"learning_rate": 2.38e-06,
"loss": 4.2232,
"step": 121
},
{
"epoch": 0.27,
"learning_rate": 2.4000000000000003e-06,
"loss": 4.3045,
"step": 122
},
{
"epoch": 0.28,
"learning_rate": 2.42e-06,
"loss": 4.33,
"step": 123
},
{
"epoch": 0.28,
"learning_rate": 2.4400000000000004e-06,
"loss": 4.3141,
"step": 124
},
{
"epoch": 0.28,
"learning_rate": 2.46e-06,
"loss": 4.4295,
"step": 125
},
{
"epoch": 0.28,
"learning_rate": 2.4800000000000004e-06,
"loss": 4.44,
"step": 126
},
{
"epoch": 0.28,
"learning_rate": 2.5e-06,
"loss": 4.3421,
"step": 127
},
{
"epoch": 0.29,
"learning_rate": 2.52e-06,
"loss": 4.3164,
"step": 128
},
{
"epoch": 0.29,
"learning_rate": 2.5400000000000002e-06,
"loss": 4.2223,
"step": 129
},
{
"epoch": 0.29,
"learning_rate": 2.56e-06,
"loss": 4.3852,
"step": 130
},
{
"epoch": 0.29,
"learning_rate": 2.5800000000000003e-06,
"loss": 4.2854,
"step": 131
},
{
"epoch": 0.3,
"learning_rate": 2.6e-06,
"loss": 4.3269,
"step": 132
},
{
"epoch": 0.3,
"learning_rate": 2.6200000000000003e-06,
"loss": 4.2724,
"step": 133
},
{
"epoch": 0.3,
"learning_rate": 2.64e-06,
"loss": 4.2822,
"step": 134
},
{
"epoch": 0.3,
"learning_rate": 2.6600000000000004e-06,
"loss": 4.4477,
"step": 135
},
{
"epoch": 0.3,
"learning_rate": 2.68e-06,
"loss": 4.3023,
"step": 136
},
{
"epoch": 0.31,
"learning_rate": 2.7000000000000004e-06,
"loss": 4.4109,
"step": 137
},
{
"epoch": 0.31,
"learning_rate": 2.7200000000000002e-06,
"loss": 4.4397,
"step": 138
},
{
"epoch": 0.31,
"learning_rate": 2.7400000000000004e-06,
"loss": 4.4502,
"step": 139
},
{
"epoch": 0.31,
"learning_rate": 2.7600000000000003e-06,
"loss": 4.4306,
"step": 140
},
{
"epoch": 0.32,
"learning_rate": 2.7800000000000005e-06,
"loss": 4.5436,
"step": 141
},
{
"epoch": 0.32,
"learning_rate": 2.8000000000000003e-06,
"loss": 4.4715,
"step": 142
},
{
"epoch": 0.32,
"learning_rate": 2.82e-06,
"loss": 4.3932,
"step": 143
},
{
"epoch": 0.32,
"learning_rate": 2.84e-06,
"loss": 4.4735,
"step": 144
},
{
"epoch": 0.33,
"learning_rate": 2.86e-06,
"loss": 4.3691,
"step": 145
},
{
"epoch": 0.33,
"learning_rate": 2.88e-06,
"loss": 4.4599,
"step": 146
},
{
"epoch": 0.33,
"learning_rate": 2.9e-06,
"loss": 4.7192,
"step": 147
},
{
"epoch": 0.33,
"learning_rate": 2.92e-06,
"loss": 4.8634,
"step": 148
},
{
"epoch": 0.33,
"learning_rate": 2.9400000000000002e-06,
"loss": 4.5975,
"step": 149
},
{
"epoch": 0.34,
"learning_rate": 2.96e-06,
"loss": 4.7077,
"step": 150
},
{
"epoch": 0.34,
"learning_rate": 2.9800000000000003e-06,
"loss": 4.2145,
"step": 151
},
{
"epoch": 0.34,
"learning_rate": 3e-06,
"loss": 4.2845,
"step": 152
},
{
"epoch": 0.34,
"learning_rate": 3.0200000000000003e-06,
"loss": 4.2969,
"step": 153
},
{
"epoch": 0.35,
"learning_rate": 3.04e-06,
"loss": 4.2126,
"step": 154
},
{
"epoch": 0.35,
"learning_rate": 3.0600000000000003e-06,
"loss": 4.3056,
"step": 155
},
{
"epoch": 0.35,
"learning_rate": 3.08e-06,
"loss": 4.172,
"step": 156
},
{
"epoch": 0.35,
"learning_rate": 3.1000000000000004e-06,
"loss": 4.2029,
"step": 157
},
{
"epoch": 0.35,
"learning_rate": 3.12e-06,
"loss": 4.2939,
"step": 158
},
{
"epoch": 0.36,
"learning_rate": 3.1400000000000004e-06,
"loss": 4.3015,
"step": 159
},
{
"epoch": 0.36,
"learning_rate": 3.1600000000000002e-06,
"loss": 4.3041,
"step": 160
},
{
"epoch": 0.36,
"learning_rate": 3.1800000000000005e-06,
"loss": 4.3576,
"step": 161
},
{
"epoch": 0.36,
"learning_rate": 3.2000000000000003e-06,
"loss": 4.3136,
"step": 162
},
{
"epoch": 0.37,
"learning_rate": 3.2200000000000005e-06,
"loss": 4.2696,
"step": 163
},
{
"epoch": 0.37,
"learning_rate": 3.2400000000000003e-06,
"loss": 4.3336,
"step": 164
},
{
"epoch": 0.37,
"learning_rate": 3.2600000000000006e-06,
"loss": 4.1924,
"step": 165
},
{
"epoch": 0.37,
"learning_rate": 3.2800000000000004e-06,
"loss": 4.2559,
"step": 166
},
{
"epoch": 0.37,
"learning_rate": 3.3000000000000006e-06,
"loss": 4.1719,
"step": 167
},
{
"epoch": 0.38,
"learning_rate": 3.3200000000000004e-06,
"loss": 4.2586,
"step": 168
},
{
"epoch": 0.38,
"learning_rate": 3.3400000000000006e-06,
"loss": 4.3229,
"step": 169
},
{
"epoch": 0.38,
"learning_rate": 3.3600000000000004e-06,
"loss": 4.3361,
"step": 170
},
{
"epoch": 0.38,
"learning_rate": 3.3800000000000007e-06,
"loss": 4.3238,
"step": 171
},
{
"epoch": 0.39,
"learning_rate": 3.4000000000000005e-06,
"loss": 4.3695,
"step": 172
},
{
"epoch": 0.39,
"learning_rate": 3.4200000000000007e-06,
"loss": 4.1802,
"step": 173
},
{
"epoch": 0.39,
"learning_rate": 3.44e-06,
"loss": 4.3182,
"step": 174
},
{
"epoch": 0.39,
"learning_rate": 3.46e-06,
"loss": 4.3777,
"step": 175
},
{
"epoch": 0.39,
"learning_rate": 3.48e-06,
"loss": 4.2564,
"step": 176
},
{
"epoch": 0.4,
"learning_rate": 3.5e-06,
"loss": 4.2225,
"step": 177
},
{
"epoch": 0.4,
"learning_rate": 3.52e-06,
"loss": 4.3079,
"step": 178
},
{
"epoch": 0.4,
"learning_rate": 3.54e-06,
"loss": 4.3581,
"step": 179
},
{
"epoch": 0.4,
"learning_rate": 3.5600000000000002e-06,
"loss": 4.3614,
"step": 180
},
{
"epoch": 0.41,
"learning_rate": 3.58e-06,
"loss": 4.2853,
"step": 181
},
{
"epoch": 0.41,
"learning_rate": 3.6000000000000003e-06,
"loss": 4.3422,
"step": 182
},
{
"epoch": 0.41,
"learning_rate": 3.62e-06,
"loss": 4.348,
"step": 183
},
{
"epoch": 0.41,
"learning_rate": 3.6400000000000003e-06,
"loss": 4.3491,
"step": 184
},
{
"epoch": 0.41,
"learning_rate": 3.66e-06,
"loss": 4.3238,
"step": 185
},
{
"epoch": 0.42,
"learning_rate": 3.6800000000000003e-06,
"loss": 4.1929,
"step": 186
},
{
"epoch": 0.42,
"learning_rate": 3.7e-06,
"loss": 4.21,
"step": 187
},
{
"epoch": 0.42,
"learning_rate": 3.7200000000000004e-06,
"loss": 4.164,
"step": 188
},
{
"epoch": 0.42,
"learning_rate": 3.74e-06,
"loss": 4.2805,
"step": 189
},
{
"epoch": 0.43,
"learning_rate": 3.7600000000000004e-06,
"loss": 4.405,
"step": 190
},
{
"epoch": 0.43,
"learning_rate": 3.7800000000000002e-06,
"loss": 4.3344,
"step": 191
},
{
"epoch": 0.43,
"learning_rate": 3.8000000000000005e-06,
"loss": 4.2669,
"step": 192
},
{
"epoch": 0.43,
"learning_rate": 3.820000000000001e-06,
"loss": 4.2899,
"step": 193
},
{
"epoch": 0.43,
"learning_rate": 3.8400000000000005e-06,
"loss": 4.3873,
"step": 194
},
{
"epoch": 0.44,
"learning_rate": 3.86e-06,
"loss": 4.4804,
"step": 195
},
{
"epoch": 0.44,
"learning_rate": 3.88e-06,
"loss": 4.4503,
"step": 196
},
{
"epoch": 0.44,
"learning_rate": 3.900000000000001e-06,
"loss": 4.3873,
"step": 197
},
{
"epoch": 0.44,
"learning_rate": 3.920000000000001e-06,
"loss": 4.6998,
"step": 198
},
{
"epoch": 0.45,
"learning_rate": 3.94e-06,
"loss": 4.6039,
"step": 199
},
{
"epoch": 0.45,
"learning_rate": 3.96e-06,
"loss": 4.5695,
"step": 200
},
{
"epoch": 0.45,
"learning_rate": 3.980000000000001e-06,
"loss": 4.3181,
"step": 201
},
{
"epoch": 0.45,
"learning_rate": 4.000000000000001e-06,
"loss": 4.2565,
"step": 202
},
{
"epoch": 0.46,
"learning_rate": 4.0200000000000005e-06,
"loss": 4.2566,
"step": 203
},
{
"epoch": 0.46,
"learning_rate": 4.04e-06,
"loss": 4.2488,
"step": 204
},
{
"epoch": 0.46,
"learning_rate": 4.060000000000001e-06,
"loss": 4.3213,
"step": 205
},
{
"epoch": 0.46,
"learning_rate": 4.08e-06,
"loss": 4.3263,
"step": 206
},
{
"epoch": 0.46,
"learning_rate": 4.1e-06,
"loss": 4.1808,
"step": 207
},
{
"epoch": 0.47,
"learning_rate": 4.12e-06,
"loss": 4.2721,
"step": 208
},
{
"epoch": 0.47,
"learning_rate": 4.14e-06,
"loss": 4.3141,
"step": 209
},
{
"epoch": 0.47,
"learning_rate": 4.16e-06,
"loss": 4.285,
"step": 210
},
{
"epoch": 0.47,
"learning_rate": 4.18e-06,
"loss": 4.1726,
"step": 211
},
{
"epoch": 0.48,
"learning_rate": 4.2000000000000004e-06,
"loss": 4.2216,
"step": 212
},
{
"epoch": 0.48,
"learning_rate": 4.22e-06,
"loss": 4.1285,
"step": 213
},
{
"epoch": 0.48,
"learning_rate": 4.24e-06,
"loss": 4.2122,
"step": 214
},
{
"epoch": 0.48,
"learning_rate": 4.26e-06,
"loss": 4.2578,
"step": 215
},
{
"epoch": 0.48,
"learning_rate": 4.2800000000000005e-06,
"loss": 4.2336,
"step": 216
},
{
"epoch": 0.49,
"learning_rate": 4.3e-06,
"loss": 4.2851,
"step": 217
},
{
"epoch": 0.49,
"learning_rate": 4.32e-06,
"loss": 4.2353,
"step": 218
},
{
"epoch": 0.49,
"learning_rate": 4.34e-06,
"loss": 4.1564,
"step": 219
},
{
"epoch": 0.49,
"learning_rate": 4.360000000000001e-06,
"loss": 4.2741,
"step": 220
},
{
"epoch": 0.5,
"learning_rate": 4.38e-06,
"loss": 4.2666,
"step": 221
},
{
"epoch": 0.5,
"learning_rate": 4.4e-06,
"loss": 4.2329,
"step": 222
},
{
"epoch": 0.5,
"learning_rate": 4.42e-06,
"loss": 4.2691,
"step": 223
},
{
"epoch": 0.5,
"learning_rate": 4.440000000000001e-06,
"loss": 4.1904,
"step": 224
},
{
"epoch": 0.5,
"learning_rate": 4.4600000000000005e-06,
"loss": 4.3606,
"step": 225
},
{
"epoch": 0.51,
"learning_rate": 4.48e-06,
"loss": 4.2548,
"step": 226
},
{
"epoch": 0.51,
"learning_rate": 4.5e-06,
"loss": 4.3776,
"step": 227
},
{
"epoch": 0.51,
"learning_rate": 4.520000000000001e-06,
"loss": 4.3096,
"step": 228
},
{
"epoch": 0.51,
"learning_rate": 4.540000000000001e-06,
"loss": 4.1924,
"step": 229
},
{
"epoch": 0.52,
"learning_rate": 4.56e-06,
"loss": 4.2802,
"step": 230
},
{
"epoch": 0.52,
"learning_rate": 4.58e-06,
"loss": 4.3578,
"step": 231
},
{
"epoch": 0.52,
"learning_rate": 4.600000000000001e-06,
"loss": 4.1489,
"step": 232
},
{
"epoch": 0.52,
"learning_rate": 4.620000000000001e-06,
"loss": 4.1684,
"step": 233
},
{
"epoch": 0.52,
"learning_rate": 4.6400000000000005e-06,
"loss": 4.2505,
"step": 234
},
{
"epoch": 0.53,
"learning_rate": 4.66e-06,
"loss": 4.3324,
"step": 235
},
{
"epoch": 0.53,
"learning_rate": 4.680000000000001e-06,
"loss": 4.2627,
"step": 236
},
{
"epoch": 0.53,
"learning_rate": 4.7e-06,
"loss": 4.304,
"step": 237
},
{
"epoch": 0.53,
"learning_rate": 4.7200000000000005e-06,
"loss": 4.2457,
"step": 238
},
{
"epoch": 0.54,
"learning_rate": 4.74e-06,
"loss": 4.1868,
"step": 239
},
{
"epoch": 0.54,
"learning_rate": 4.76e-06,
"loss": 4.3028,
"step": 240
},
{
"epoch": 0.54,
"learning_rate": 4.78e-06,
"loss": 4.2704,
"step": 241
},
{
"epoch": 0.54,
"learning_rate": 4.800000000000001e-06,
"loss": 4.3031,
"step": 242
},
{
"epoch": 0.54,
"learning_rate": 4.8200000000000004e-06,
"loss": 4.5193,
"step": 243
},
{
"epoch": 0.55,
"learning_rate": 4.84e-06,
"loss": 4.3524,
"step": 244
},
{
"epoch": 0.55,
"learning_rate": 4.86e-06,
"loss": 4.3437,
"step": 245
},
{
"epoch": 0.55,
"learning_rate": 4.880000000000001e-06,
"loss": 4.2623,
"step": 246
},
{
"epoch": 0.55,
"learning_rate": 4.9000000000000005e-06,
"loss": 4.1807,
"step": 247
},
{
"epoch": 0.56,
"learning_rate": 4.92e-06,
"loss": 4.5587,
"step": 248
},
{
"epoch": 0.56,
"learning_rate": 4.94e-06,
"loss": 4.6662,
"step": 249
},
{
"epoch": 0.56,
"learning_rate": 4.960000000000001e-06,
"loss": 4.6559,
"step": 250
},
{
"epoch": 0.56,
"learning_rate": 4.980000000000001e-06,
"loss": 4.2338,
"step": 251
},
{
"epoch": 0.57,
"learning_rate": 5e-06,
"loss": 4.2654,
"step": 252
},
{
"epoch": 0.57,
"learning_rate": 5.02e-06,
"loss": 4.311,
"step": 253
},
{
"epoch": 0.57,
"learning_rate": 5.04e-06,
"loss": 4.1525,
"step": 254
},
{
"epoch": 0.57,
"learning_rate": 5.060000000000001e-06,
"loss": 4.2531,
"step": 255
},
{
"epoch": 0.57,
"learning_rate": 5.0800000000000005e-06,
"loss": 4.223,
"step": 256
},
{
"epoch": 0.58,
"learning_rate": 5.1e-06,
"loss": 4.1962,
"step": 257
},
{
"epoch": 0.58,
"learning_rate": 5.12e-06,
"loss": 4.2138,
"step": 258
},
{
"epoch": 0.58,
"learning_rate": 5.140000000000001e-06,
"loss": 4.2155,
"step": 259
},
{
"epoch": 0.58,
"learning_rate": 5.1600000000000006e-06,
"loss": 4.1582,
"step": 260
},
{
"epoch": 0.59,
"learning_rate": 5.18e-06,
"loss": 4.2736,
"step": 261
},
{
"epoch": 0.59,
"learning_rate": 5.2e-06,
"loss": 4.2103,
"step": 262
},
{
"epoch": 0.59,
"learning_rate": 5.220000000000001e-06,
"loss": 4.3473,
"step": 263
},
{
"epoch": 0.59,
"learning_rate": 5.240000000000001e-06,
"loss": 4.2133,
"step": 264
},
{
"epoch": 0.59,
"learning_rate": 5.2600000000000005e-06,
"loss": 4.2476,
"step": 265
},
{
"epoch": 0.6,
"learning_rate": 5.28e-06,
"loss": 4.2244,
"step": 266
},
{
"epoch": 0.6,
"learning_rate": 5.300000000000001e-06,
"loss": 4.2517,
"step": 267
},
{
"epoch": 0.6,
"learning_rate": 5.320000000000001e-06,
"loss": 4.2084,
"step": 268
},
{
"epoch": 0.6,
"learning_rate": 5.3400000000000005e-06,
"loss": 4.2887,
"step": 269
},
{
"epoch": 0.61,
"learning_rate": 5.36e-06,
"loss": 4.1665,
"step": 270
},
{
"epoch": 0.61,
"learning_rate": 5.380000000000001e-06,
"loss": 4.0903,
"step": 271
},
{
"epoch": 0.61,
"learning_rate": 5.400000000000001e-06,
"loss": 4.2757,
"step": 272
},
{
"epoch": 0.61,
"learning_rate": 5.420000000000001e-06,
"loss": 4.0794,
"step": 273
},
{
"epoch": 0.61,
"learning_rate": 5.4400000000000004e-06,
"loss": 4.1791,
"step": 274
},
{
"epoch": 0.62,
"learning_rate": 5.460000000000001e-06,
"loss": 4.2729,
"step": 275
},
{
"epoch": 0.62,
"learning_rate": 5.480000000000001e-06,
"loss": 4.2107,
"step": 276
},
{
"epoch": 0.62,
"learning_rate": 5.500000000000001e-06,
"loss": 4.1623,
"step": 277
},
{
"epoch": 0.62,
"learning_rate": 5.5200000000000005e-06,
"loss": 4.2513,
"step": 278
},
{
"epoch": 0.63,
"learning_rate": 5.540000000000001e-06,
"loss": 4.235,
"step": 279
},
{
"epoch": 0.63,
"learning_rate": 5.560000000000001e-06,
"loss": 4.3647,
"step": 280
},
{
"epoch": 0.63,
"learning_rate": 5.580000000000001e-06,
"loss": 4.2633,
"step": 281
},
{
"epoch": 0.63,
"learning_rate": 5.600000000000001e-06,
"loss": 4.12,
"step": 282
},
{
"epoch": 0.63,
"learning_rate": 5.620000000000001e-06,
"loss": 4.1451,
"step": 283
},
{
"epoch": 0.64,
"learning_rate": 5.64e-06,
"loss": 4.2252,
"step": 284
},
{
"epoch": 0.64,
"learning_rate": 5.66e-06,
"loss": 4.3205,
"step": 285
},
{
"epoch": 0.64,
"learning_rate": 5.68e-06,
"loss": 4.2208,
"step": 286
},
{
"epoch": 0.64,
"learning_rate": 5.7e-06,
"loss": 4.3254,
"step": 287
},
{
"epoch": 0.65,
"learning_rate": 5.72e-06,
"loss": 4.2817,
"step": 288
},
{
"epoch": 0.65,
"learning_rate": 5.74e-06,
"loss": 4.192,
"step": 289
},
{
"epoch": 0.65,
"learning_rate": 5.76e-06,
"loss": 4.2462,
"step": 290
},
{
"epoch": 0.65,
"learning_rate": 5.78e-06,
"loss": 4.4973,
"step": 291
},
{
"epoch": 0.65,
"learning_rate": 5.8e-06,
"loss": 4.3751,
"step": 292
},
{
"epoch": 0.66,
"learning_rate": 5.82e-06,
"loss": 4.441,
"step": 293
},
{
"epoch": 0.66,
"learning_rate": 5.84e-06,
"loss": 4.4067,
"step": 294
},
{
"epoch": 0.66,
"learning_rate": 5.86e-06,
"loss": 4.3823,
"step": 295
},
{
"epoch": 0.66,
"learning_rate": 5.8800000000000005e-06,
"loss": 4.3868,
"step": 296
},
{
"epoch": 0.67,
"learning_rate": 5.9e-06,
"loss": 4.2941,
"step": 297
},
{
"epoch": 0.67,
"learning_rate": 5.92e-06,
"loss": 4.6034,
"step": 298
},
{
"epoch": 0.67,
"learning_rate": 5.94e-06,
"loss": 4.37,
"step": 299
},
{
"epoch": 0.67,
"learning_rate": 5.9600000000000005e-06,
"loss": 4.7808,
"step": 300
},
{
"epoch": 0.67,
"learning_rate": 5.98e-06,
"loss": 4.268,
"step": 301
},
{
"epoch": 0.68,
"learning_rate": 6e-06,
"loss": 4.2326,
"step": 302
},
{
"epoch": 0.68,
"learning_rate": 6.02e-06,
"loss": 4.2418,
"step": 303
},
{
"epoch": 0.68,
"learning_rate": 6.040000000000001e-06,
"loss": 4.2376,
"step": 304
},
{
"epoch": 0.68,
"learning_rate": 6.0600000000000004e-06,
"loss": 4.2091,
"step": 305
},
{
"epoch": 0.69,
"learning_rate": 6.08e-06,
"loss": 4.2111,
"step": 306
},
{
"epoch": 0.69,
"learning_rate": 6.1e-06,
"loss": 4.2004,
"step": 307
},
{
"epoch": 0.69,
"learning_rate": 6.120000000000001e-06,
"loss": 4.1965,
"step": 308
},
{
"epoch": 0.69,
"learning_rate": 6.1400000000000005e-06,
"loss": 4.1974,
"step": 309
},
{
"epoch": 0.7,
"learning_rate": 6.16e-06,
"loss": 4.2748,
"step": 310
},
{
"epoch": 0.7,
"learning_rate": 6.18e-06,
"loss": 4.0933,
"step": 311
},
{
"epoch": 0.7,
"learning_rate": 6.200000000000001e-06,
"loss": 4.2521,
"step": 312
},
{
"epoch": 0.7,
"learning_rate": 6.220000000000001e-06,
"loss": 4.2651,
"step": 313
},
{
"epoch": 0.7,
"learning_rate": 6.24e-06,
"loss": 4.182,
"step": 314
},
{
"epoch": 0.71,
"learning_rate": 6.26e-06,
"loss": 4.2096,
"step": 315
},
{
"epoch": 0.71,
"learning_rate": 6.280000000000001e-06,
"loss": 4.1797,
"step": 316
},
{
"epoch": 0.71,
"learning_rate": 6.300000000000001e-06,
"loss": 4.1667,
"step": 317
},
{
"epoch": 0.71,
"learning_rate": 6.3200000000000005e-06,
"loss": 4.2119,
"step": 318
},
{
"epoch": 0.72,
"learning_rate": 6.34e-06,
"loss": 4.1995,
"step": 319
},
{
"epoch": 0.72,
"learning_rate": 6.360000000000001e-06,
"loss": 4.2797,
"step": 320
},
{
"epoch": 0.72,
"learning_rate": 6.380000000000001e-06,
"loss": 4.1701,
"step": 321
},
{
"epoch": 0.72,
"learning_rate": 6.4000000000000006e-06,
"loss": 4.0619,
"step": 322
},
{
"epoch": 0.72,
"learning_rate": 6.42e-06,
"loss": 4.2404,
"step": 323
},
{
"epoch": 0.73,
"learning_rate": 6.440000000000001e-06,
"loss": 4.1039,
"step": 324
},
{
"epoch": 0.73,
"learning_rate": 6.460000000000001e-06,
"loss": 4.2368,
"step": 325
},
{
"epoch": 0.73,
"learning_rate": 6.480000000000001e-06,
"loss": 4.1967,
"step": 326
},
{
"epoch": 0.73,
"learning_rate": 6.5000000000000004e-06,
"loss": 4.1744,
"step": 327
},
{
"epoch": 0.74,
"learning_rate": 6.520000000000001e-06,
"loss": 4.1501,
"step": 328
},
{
"epoch": 0.74,
"learning_rate": 6.540000000000001e-06,
"loss": 4.1861,
"step": 329
},
{
"epoch": 0.74,
"learning_rate": 6.560000000000001e-06,
"loss": 4.2375,
"step": 330
},
{
"epoch": 0.74,
"learning_rate": 6.5800000000000005e-06,
"loss": 4.2348,
"step": 331
},
{
"epoch": 0.74,
"learning_rate": 6.600000000000001e-06,
"loss": 4.133,
"step": 332
},
{
"epoch": 0.75,
"learning_rate": 6.620000000000001e-06,
"loss": 4.2921,
"step": 333
},
{
"epoch": 0.75,
"learning_rate": 6.640000000000001e-06,
"loss": 4.1948,
"step": 334
},
{
"epoch": 0.75,
"learning_rate": 6.660000000000001e-06,
"loss": 4.3708,
"step": 335
},
{
"epoch": 0.75,
"learning_rate": 6.680000000000001e-06,
"loss": 4.1533,
"step": 336
},
{
"epoch": 0.76,
"learning_rate": 6.700000000000001e-06,
"loss": 4.1618,
"step": 337
},
{
"epoch": 0.76,
"learning_rate": 6.720000000000001e-06,
"loss": 4.267,
"step": 338
},
{
"epoch": 0.76,
"learning_rate": 6.740000000000001e-06,
"loss": 4.2544,
"step": 339
},
{
"epoch": 0.76,
"learning_rate": 6.760000000000001e-06,
"loss": 4.3039,
"step": 340
},
{
"epoch": 0.76,
"learning_rate": 6.780000000000001e-06,
"loss": 4.2502,
"step": 341
},
{
"epoch": 0.77,
"learning_rate": 6.800000000000001e-06,
"loss": 4.4135,
"step": 342
},
{
"epoch": 0.77,
"learning_rate": 6.820000000000001e-06,
"loss": 4.3116,
"step": 343
},
{
"epoch": 0.77,
"learning_rate": 6.8400000000000014e-06,
"loss": 4.3977,
"step": 344
},
{
"epoch": 0.77,
"learning_rate": 6.860000000000001e-06,
"loss": 4.135,
"step": 345
},
{
"epoch": 0.78,
"learning_rate": 6.88e-06,
"loss": 4.3846,
"step": 346
},
{
"epoch": 0.78,
"learning_rate": 6.9e-06,
"loss": 4.3741,
"step": 347
},
{
"epoch": 0.78,
"learning_rate": 6.92e-06,
"loss": 4.3161,
"step": 348
},
{
"epoch": 0.78,
"learning_rate": 6.9400000000000005e-06,
"loss": 4.5413,
"step": 349
},
{
"epoch": 0.78,
"learning_rate": 6.96e-06,
"loss": 4.5532,
"step": 350
},
{
"epoch": 0.79,
"learning_rate": 6.98e-06,
"loss": 4.275,
"step": 351
},
{
"epoch": 0.79,
"learning_rate": 7e-06,
"loss": 4.2207,
"step": 352
},
{
"epoch": 0.79,
"learning_rate": 7.0200000000000006e-06,
"loss": 4.1401,
"step": 353
},
{
"epoch": 0.79,
"learning_rate": 7.04e-06,
"loss": 4.0229,
"step": 354
},
{
"epoch": 0.8,
"learning_rate": 7.06e-06,
"loss": 4.2197,
"step": 355
},
{
"epoch": 0.8,
"learning_rate": 7.08e-06,
"loss": 4.3189,
"step": 356
},
{
"epoch": 0.8,
"learning_rate": 7.100000000000001e-06,
"loss": 4.1967,
"step": 357
},
{
"epoch": 0.8,
"learning_rate": 7.1200000000000004e-06,
"loss": 4.2343,
"step": 358
},
{
"epoch": 0.8,
"learning_rate": 7.14e-06,
"loss": 4.1164,
"step": 359
},
{
"epoch": 0.81,
"learning_rate": 7.16e-06,
"loss": 4.1943,
"step": 360
},
{
"epoch": 0.81,
"learning_rate": 7.180000000000001e-06,
"loss": 4.1379,
"step": 361
},
{
"epoch": 0.81,
"learning_rate": 7.2000000000000005e-06,
"loss": 4.1926,
"step": 362
},
{
"epoch": 0.81,
"learning_rate": 7.22e-06,
"loss": 4.1157,
"step": 363
},
{
"epoch": 0.82,
"learning_rate": 7.24e-06,
"loss": 4.2101,
"step": 364
},
{
"epoch": 0.82,
"learning_rate": 7.260000000000001e-06,
"loss": 4.3345,
"step": 365
},
{
"epoch": 0.82,
"learning_rate": 7.280000000000001e-06,
"loss": 4.1492,
"step": 366
},
{
"epoch": 0.82,
"learning_rate": 7.3e-06,
"loss": 4.2292,
"step": 367
},
{
"epoch": 0.83,
"learning_rate": 7.32e-06,
"loss": 4.212,
"step": 368
},
{
"epoch": 0.83,
"learning_rate": 7.340000000000001e-06,
"loss": 4.1305,
"step": 369
},
{
"epoch": 0.83,
"learning_rate": 7.360000000000001e-06,
"loss": 4.141,
"step": 370
},
{
"epoch": 0.83,
"learning_rate": 7.3800000000000005e-06,
"loss": 4.1366,
"step": 371
},
{
"epoch": 0.83,
"learning_rate": 7.4e-06,
"loss": 4.2152,
"step": 372
},
{
"epoch": 0.84,
"learning_rate": 7.420000000000001e-06,
"loss": 4.2039,
"step": 373
},
{
"epoch": 0.84,
"learning_rate": 7.440000000000001e-06,
"loss": 4.0068,
"step": 374
},
{
"epoch": 0.84,
"learning_rate": 7.4600000000000006e-06,
"loss": 4.2379,
"step": 375
},
{
"epoch": 0.84,
"learning_rate": 7.48e-06,
"loss": 4.0893,
"step": 376
},
{
"epoch": 0.85,
"learning_rate": 7.500000000000001e-06,
"loss": 4.2137,
"step": 377
},
{
"epoch": 0.85,
"learning_rate": 7.520000000000001e-06,
"loss": 4.1977,
"step": 378
},
{
"epoch": 0.85,
"learning_rate": 7.540000000000001e-06,
"loss": 4.1133,
"step": 379
},
{
"epoch": 0.85,
"learning_rate": 7.5600000000000005e-06,
"loss": 4.1347,
"step": 380
},
{
"epoch": 0.85,
"learning_rate": 7.58e-06,
"loss": 4.1941,
"step": 381
},
{
"epoch": 0.86,
"learning_rate": 7.600000000000001e-06,
"loss": 4.2719,
"step": 382
},
{
"epoch": 0.86,
"learning_rate": 7.620000000000001e-06,
"loss": 4.1831,
"step": 383
},
{
"epoch": 0.86,
"learning_rate": 7.640000000000001e-06,
"loss": 4.1165,
"step": 384
},
{
"epoch": 0.86,
"learning_rate": 7.660000000000001e-06,
"loss": 4.2029,
"step": 385
},
{
"epoch": 0.87,
"learning_rate": 7.680000000000001e-06,
"loss": 4.0781,
"step": 386
},
{
"epoch": 0.87,
"learning_rate": 7.7e-06,
"loss": 4.2651,
"step": 387
},
{
"epoch": 0.87,
"learning_rate": 7.72e-06,
"loss": 4.2909,
"step": 388
},
{
"epoch": 0.87,
"learning_rate": 7.74e-06,
"loss": 4.1941,
"step": 389
},
{
"epoch": 0.87,
"learning_rate": 7.76e-06,
"loss": 4.172,
"step": 390
},
{
"epoch": 0.88,
"learning_rate": 7.78e-06,
"loss": 4.2985,
"step": 391
},
{
"epoch": 0.88,
"learning_rate": 7.800000000000002e-06,
"loss": 4.1451,
"step": 392
},
{
"epoch": 0.88,
"learning_rate": 7.820000000000001e-06,
"loss": 4.1401,
"step": 393
},
{
"epoch": 0.88,
"learning_rate": 7.840000000000001e-06,
"loss": 4.2258,
"step": 394
},
{
"epoch": 0.89,
"learning_rate": 7.860000000000001e-06,
"loss": 4.3488,
"step": 395
},
{
"epoch": 0.89,
"learning_rate": 7.88e-06,
"loss": 4.3106,
"step": 396
},
{
"epoch": 0.89,
"learning_rate": 7.9e-06,
"loss": 4.4439,
"step": 397
},
{
"epoch": 0.89,
"learning_rate": 7.92e-06,
"loss": 4.4548,
"step": 398
},
{
"epoch": 0.89,
"learning_rate": 7.94e-06,
"loss": 4.5917,
"step": 399
},
{
"epoch": 0.9,
"learning_rate": 7.960000000000002e-06,
"loss": 4.5111,
"step": 400
},
{
"epoch": 0.9,
"learning_rate": 7.980000000000002e-06,
"loss": 4.2176,
"step": 401
},
{
"epoch": 0.9,
"learning_rate": 8.000000000000001e-06,
"loss": 4.0665,
"step": 402
},
{
"epoch": 0.9,
"learning_rate": 8.020000000000001e-06,
"loss": 4.1243,
"step": 403
},
{
"epoch": 0.91,
"learning_rate": 8.040000000000001e-06,
"loss": 4.1244,
"step": 404
},
{
"epoch": 0.91,
"learning_rate": 8.06e-06,
"loss": 4.0896,
"step": 405
},
{
"epoch": 0.91,
"learning_rate": 8.08e-06,
"loss": 4.1743,
"step": 406
},
{
"epoch": 0.91,
"learning_rate": 8.1e-06,
"loss": 4.1469,
"step": 407
},
{
"epoch": 0.91,
"learning_rate": 8.120000000000002e-06,
"loss": 4.2224,
"step": 408
},
{
"epoch": 0.92,
"learning_rate": 8.14e-06,
"loss": 4.1593,
"step": 409
},
{
"epoch": 0.92,
"learning_rate": 8.16e-06,
"loss": 4.1418,
"step": 410
},
{
"epoch": 0.92,
"learning_rate": 8.18e-06,
"loss": 4.3698,
"step": 411
},
{
"epoch": 0.92,
"learning_rate": 8.2e-06,
"loss": 4.1025,
"step": 412
},
{
"epoch": 0.93,
"learning_rate": 8.220000000000001e-06,
"loss": 4.0313,
"step": 413
},
{
"epoch": 0.93,
"learning_rate": 8.24e-06,
"loss": 4.192,
"step": 414
},
{
"epoch": 0.93,
"learning_rate": 8.26e-06,
"loss": 4.2583,
"step": 415
},
{
"epoch": 0.93,
"learning_rate": 8.28e-06,
"loss": 4.1274,
"step": 416
},
{
"epoch": 0.93,
"learning_rate": 8.3e-06,
"loss": 4.152,
"step": 417
},
{
"epoch": 0.94,
"learning_rate": 8.32e-06,
"loss": 4.0524,
"step": 418
},
{
"epoch": 0.94,
"learning_rate": 8.34e-06,
"loss": 4.1577,
"step": 419
},
{
"epoch": 0.94,
"learning_rate": 8.36e-06,
"loss": 4.0665,
"step": 420
},
{
"epoch": 0.94,
"learning_rate": 8.380000000000001e-06,
"loss": 4.2451,
"step": 421
},
{
"epoch": 0.95,
"learning_rate": 8.400000000000001e-06,
"loss": 4.0875,
"step": 422
},
{
"epoch": 0.95,
"learning_rate": 8.42e-06,
"loss": 4.1906,
"step": 423
},
{
"epoch": 0.95,
"learning_rate": 8.44e-06,
"loss": 4.2461,
"step": 424
},
{
"epoch": 0.95,
"learning_rate": 8.46e-06,
"loss": 4.1716,
"step": 425
},
{
"epoch": 0.96,
"learning_rate": 8.48e-06,
"loss": 4.2216,
"step": 426
},
{
"epoch": 0.96,
"learning_rate": 8.5e-06,
"loss": 4.18,
"step": 427
},
{
"epoch": 0.96,
"learning_rate": 8.52e-06,
"loss": 4.1531,
"step": 428
},
{
"epoch": 0.96,
"learning_rate": 8.540000000000001e-06,
"loss": 4.104,
"step": 429
},
{
"epoch": 0.96,
"learning_rate": 8.560000000000001e-06,
"loss": 4.1649,
"step": 430
},
{
"epoch": 0.97,
"learning_rate": 8.580000000000001e-06,
"loss": 4.2599,
"step": 431
},
{
"epoch": 0.97,
"learning_rate": 8.6e-06,
"loss": 4.3766,
"step": 432
},
{
"epoch": 0.97,
"learning_rate": 8.62e-06,
"loss": 4.1677,
"step": 433
},
{
"epoch": 0.97,
"learning_rate": 8.64e-06,
"loss": 4.1718,
"step": 434
},
{
"epoch": 0.98,
"learning_rate": 8.66e-06,
"loss": 4.2196,
"step": 435
},
{
"epoch": 0.98,
"learning_rate": 8.68e-06,
"loss": 4.1337,
"step": 436
},
{
"epoch": 0.98,
"learning_rate": 8.700000000000001e-06,
"loss": 4.1335,
"step": 437
},
{
"epoch": 0.98,
"learning_rate": 8.720000000000001e-06,
"loss": 4.3058,
"step": 438
},
{
"epoch": 0.98,
"learning_rate": 8.740000000000001e-06,
"loss": 4.2738,
"step": 439
},
{
"epoch": 0.99,
"learning_rate": 8.76e-06,
"loss": 4.3081,
"step": 440
},
{
"epoch": 0.99,
"learning_rate": 8.78e-06,
"loss": 4.2582,
"step": 441
},
{
"epoch": 0.99,
"learning_rate": 8.8e-06,
"loss": 4.4641,
"step": 442
},
{
"epoch": 0.99,
"learning_rate": 8.82e-06,
"loss": 4.3681,
"step": 443
},
{
"epoch": 1.0,
"learning_rate": 8.84e-06,
"loss": 4.3675,
"step": 444
},
{
"epoch": 1.0,
"learning_rate": 8.860000000000002e-06,
"loss": 4.6441,
"step": 445
},
{
"epoch": 1.0,
"learning_rate": 8.860000000000002e-06,
"loss": 4.3047,
"step": 446
},
{
"epoch": 1.0,
"step": 446,
"total_flos": 0.0,
"train_loss": 4.377411050112258,
"train_runtime": 4925.13,
"train_samples_per_second": 5.794,
"train_steps_per_second": 0.091
}
],
"max_steps": 446,
"num_train_epochs": 1,
"total_flos": 0.0,
"trial_name": null,
"trial_params": null
}