wav2vec2-2-gpt2-grid-search / trainer_state.json
sanchit-gandhi's picture
End of training
8d4e1a2
raw
history blame
No virus
68.7 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 0.9987389659520807,
"global_step": 594,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 0.0,
"loss": 4.8043,
"step": 1
},
{
"epoch": 0.0,
"learning_rate": 2e-08,
"loss": 4.9816,
"step": 2
},
{
"epoch": 0.01,
"learning_rate": 4e-08,
"loss": 4.8553,
"step": 3
},
{
"epoch": 0.01,
"learning_rate": 6.000000000000001e-08,
"loss": 4.8133,
"step": 4
},
{
"epoch": 0.01,
"learning_rate": 8e-08,
"loss": 4.7678,
"step": 5
},
{
"epoch": 0.01,
"learning_rate": 1.0000000000000001e-07,
"loss": 4.7823,
"step": 6
},
{
"epoch": 0.01,
"learning_rate": 1.2000000000000002e-07,
"loss": 4.8813,
"step": 7
},
{
"epoch": 0.01,
"learning_rate": 1.4e-07,
"loss": 4.7725,
"step": 8
},
{
"epoch": 0.02,
"learning_rate": 1.6e-07,
"loss": 4.8957,
"step": 9
},
{
"epoch": 0.02,
"learning_rate": 1.8e-07,
"loss": 4.8176,
"step": 10
},
{
"epoch": 0.02,
"learning_rate": 2.0000000000000002e-07,
"loss": 4.8163,
"step": 11
},
{
"epoch": 0.02,
"learning_rate": 2.2e-07,
"loss": 4.8674,
"step": 12
},
{
"epoch": 0.02,
"learning_rate": 2.4000000000000003e-07,
"loss": 4.8031,
"step": 13
},
{
"epoch": 0.02,
"learning_rate": 2.6e-07,
"loss": 4.7586,
"step": 14
},
{
"epoch": 0.03,
"learning_rate": 2.8e-07,
"loss": 4.6561,
"step": 15
},
{
"epoch": 0.03,
"learning_rate": 3.0000000000000004e-07,
"loss": 4.8397,
"step": 16
},
{
"epoch": 0.03,
"learning_rate": 3.2e-07,
"loss": 4.8695,
"step": 17
},
{
"epoch": 0.03,
"learning_rate": 3.4000000000000003e-07,
"loss": 4.9448,
"step": 18
},
{
"epoch": 0.03,
"learning_rate": 3.6e-07,
"loss": 4.7123,
"step": 19
},
{
"epoch": 0.03,
"learning_rate": 3.8e-07,
"loss": 4.9487,
"step": 20
},
{
"epoch": 0.04,
"learning_rate": 4.0000000000000003e-07,
"loss": 4.898,
"step": 21
},
{
"epoch": 0.04,
"learning_rate": 4.2000000000000006e-07,
"loss": 4.8324,
"step": 22
},
{
"epoch": 0.04,
"learning_rate": 4.4e-07,
"loss": 4.8531,
"step": 23
},
{
"epoch": 0.04,
"learning_rate": 4.6000000000000004e-07,
"loss": 4.9432,
"step": 24
},
{
"epoch": 0.04,
"learning_rate": 4.800000000000001e-07,
"loss": 4.7127,
"step": 25
},
{
"epoch": 0.04,
"learning_rate": 5.000000000000001e-07,
"loss": 5.1345,
"step": 26
},
{
"epoch": 0.05,
"learning_rate": 5.2e-07,
"loss": 4.7968,
"step": 27
},
{
"epoch": 0.05,
"learning_rate": 5.4e-07,
"loss": 4.6922,
"step": 28
},
{
"epoch": 0.05,
"learning_rate": 5.6e-07,
"loss": 4.7146,
"step": 29
},
{
"epoch": 0.05,
"learning_rate": 5.800000000000001e-07,
"loss": 4.8548,
"step": 30
},
{
"epoch": 0.05,
"learning_rate": 6.000000000000001e-07,
"loss": 4.8666,
"step": 31
},
{
"epoch": 0.05,
"learning_rate": 6.200000000000001e-07,
"loss": 4.7535,
"step": 32
},
{
"epoch": 0.06,
"learning_rate": 6.200000000000001e-07,
"loss": 4.7578,
"step": 33
},
{
"epoch": 0.06,
"learning_rate": 6.4e-07,
"loss": 4.6922,
"step": 34
},
{
"epoch": 0.06,
"learning_rate": 6.6e-07,
"loss": 4.7539,
"step": 35
},
{
"epoch": 0.06,
"learning_rate": 6.800000000000001e-07,
"loss": 4.9275,
"step": 36
},
{
"epoch": 0.06,
"learning_rate": 7.000000000000001e-07,
"loss": 4.7236,
"step": 37
},
{
"epoch": 0.06,
"learning_rate": 7.2e-07,
"loss": 4.8574,
"step": 38
},
{
"epoch": 0.07,
"learning_rate": 7.4e-07,
"loss": 5.0691,
"step": 39
},
{
"epoch": 0.07,
"learning_rate": 7.6e-07,
"loss": 4.6674,
"step": 40
},
{
"epoch": 0.07,
"learning_rate": 7.8e-07,
"loss": 4.7318,
"step": 41
},
{
"epoch": 0.07,
"learning_rate": 8.000000000000001e-07,
"loss": 4.9842,
"step": 42
},
{
"epoch": 0.07,
"learning_rate": 8.200000000000001e-07,
"loss": 4.7206,
"step": 43
},
{
"epoch": 0.07,
"learning_rate": 8.400000000000001e-07,
"loss": 4.8842,
"step": 44
},
{
"epoch": 0.08,
"learning_rate": 8.6e-07,
"loss": 5.0102,
"step": 45
},
{
"epoch": 0.08,
"learning_rate": 8.8e-07,
"loss": 5.1112,
"step": 46
},
{
"epoch": 0.08,
"learning_rate": 9.000000000000001e-07,
"loss": 5.0286,
"step": 47
},
{
"epoch": 0.08,
"learning_rate": 9.200000000000001e-07,
"loss": 5.4426,
"step": 48
},
{
"epoch": 0.08,
"learning_rate": 9.400000000000001e-07,
"loss": 5.4768,
"step": 49
},
{
"epoch": 0.08,
"learning_rate": 9.400000000000001e-07,
"loss": 5.3523,
"step": 50
},
{
"epoch": 0.09,
"learning_rate": 9.600000000000001e-07,
"loss": 4.5321,
"step": 51
},
{
"epoch": 0.09,
"learning_rate": 9.800000000000001e-07,
"loss": 4.5271,
"step": 52
},
{
"epoch": 0.09,
"learning_rate": 1.0000000000000002e-06,
"loss": 4.5945,
"step": 53
},
{
"epoch": 0.09,
"learning_rate": 1.02e-06,
"loss": 4.674,
"step": 54
},
{
"epoch": 0.09,
"learning_rate": 1.04e-06,
"loss": 4.5153,
"step": 55
},
{
"epoch": 0.09,
"learning_rate": 1.06e-06,
"loss": 4.6963,
"step": 56
},
{
"epoch": 0.1,
"learning_rate": 1.08e-06,
"loss": 4.56,
"step": 57
},
{
"epoch": 0.1,
"learning_rate": 1.1e-06,
"loss": 4.8043,
"step": 58
},
{
"epoch": 0.1,
"learning_rate": 1.12e-06,
"loss": 4.5976,
"step": 59
},
{
"epoch": 0.1,
"learning_rate": 1.14e-06,
"loss": 4.3466,
"step": 60
},
{
"epoch": 0.1,
"learning_rate": 1.1600000000000001e-06,
"loss": 4.6048,
"step": 61
},
{
"epoch": 0.1,
"learning_rate": 1.1800000000000001e-06,
"loss": 4.6691,
"step": 62
},
{
"epoch": 0.11,
"learning_rate": 1.2000000000000002e-06,
"loss": 4.5441,
"step": 63
},
{
"epoch": 0.11,
"learning_rate": 1.2200000000000002e-06,
"loss": 4.4973,
"step": 64
},
{
"epoch": 0.11,
"learning_rate": 1.2400000000000002e-06,
"loss": 4.5558,
"step": 65
},
{
"epoch": 0.11,
"learning_rate": 1.26e-06,
"loss": 4.4949,
"step": 66
},
{
"epoch": 0.11,
"learning_rate": 1.28e-06,
"loss": 4.4888,
"step": 67
},
{
"epoch": 0.11,
"learning_rate": 1.3e-06,
"loss": 4.4224,
"step": 68
},
{
"epoch": 0.12,
"learning_rate": 1.32e-06,
"loss": 4.544,
"step": 69
},
{
"epoch": 0.12,
"learning_rate": 1.34e-06,
"loss": 4.5557,
"step": 70
},
{
"epoch": 0.12,
"learning_rate": 1.3600000000000001e-06,
"loss": 4.6351,
"step": 71
},
{
"epoch": 0.12,
"learning_rate": 1.3800000000000001e-06,
"loss": 4.5891,
"step": 72
},
{
"epoch": 0.12,
"learning_rate": 1.4000000000000001e-06,
"loss": 4.385,
"step": 73
},
{
"epoch": 0.12,
"learning_rate": 1.42e-06,
"loss": 4.5821,
"step": 74
},
{
"epoch": 0.13,
"learning_rate": 1.44e-06,
"loss": 4.554,
"step": 75
},
{
"epoch": 0.13,
"learning_rate": 1.46e-06,
"loss": 4.5606,
"step": 76
},
{
"epoch": 0.13,
"learning_rate": 1.48e-06,
"loss": 4.5196,
"step": 77
},
{
"epoch": 0.13,
"learning_rate": 1.5e-06,
"loss": 4.5308,
"step": 78
},
{
"epoch": 0.13,
"learning_rate": 1.52e-06,
"loss": 4.4185,
"step": 79
},
{
"epoch": 0.13,
"learning_rate": 1.54e-06,
"loss": 4.555,
"step": 80
},
{
"epoch": 0.14,
"learning_rate": 1.56e-06,
"loss": 4.508,
"step": 81
},
{
"epoch": 0.14,
"learning_rate": 1.5800000000000001e-06,
"loss": 4.5176,
"step": 82
},
{
"epoch": 0.14,
"learning_rate": 1.6000000000000001e-06,
"loss": 4.5365,
"step": 83
},
{
"epoch": 0.14,
"learning_rate": 1.6200000000000002e-06,
"loss": 4.5456,
"step": 84
},
{
"epoch": 0.14,
"learning_rate": 1.6400000000000002e-06,
"loss": 4.4706,
"step": 85
},
{
"epoch": 0.14,
"learning_rate": 1.6600000000000002e-06,
"loss": 4.3598,
"step": 86
},
{
"epoch": 0.15,
"learning_rate": 1.6800000000000002e-06,
"loss": 4.6727,
"step": 87
},
{
"epoch": 0.15,
"learning_rate": 1.7000000000000002e-06,
"loss": 4.5677,
"step": 88
},
{
"epoch": 0.15,
"learning_rate": 1.72e-06,
"loss": 4.5185,
"step": 89
},
{
"epoch": 0.15,
"learning_rate": 1.74e-06,
"loss": 4.5671,
"step": 90
},
{
"epoch": 0.15,
"learning_rate": 1.76e-06,
"loss": 4.4984,
"step": 91
},
{
"epoch": 0.15,
"learning_rate": 1.7800000000000001e-06,
"loss": 4.6209,
"step": 92
},
{
"epoch": 0.16,
"learning_rate": 1.8000000000000001e-06,
"loss": 4.4814,
"step": 93
},
{
"epoch": 0.16,
"learning_rate": 1.8200000000000002e-06,
"loss": 4.7069,
"step": 94
},
{
"epoch": 0.16,
"learning_rate": 1.8400000000000002e-06,
"loss": 4.6537,
"step": 95
},
{
"epoch": 0.16,
"learning_rate": 1.8600000000000002e-06,
"loss": 4.6972,
"step": 96
},
{
"epoch": 0.16,
"learning_rate": 1.8800000000000002e-06,
"loss": 4.7654,
"step": 97
},
{
"epoch": 0.16,
"learning_rate": 1.9000000000000002e-06,
"loss": 4.5225,
"step": 98
},
{
"epoch": 0.17,
"learning_rate": 1.9200000000000003e-06,
"loss": 4.694,
"step": 99
},
{
"epoch": 0.17,
"learning_rate": 1.94e-06,
"loss": 4.487,
"step": 100
},
{
"epoch": 0.17,
"learning_rate": 1.9600000000000003e-06,
"loss": 4.539,
"step": 101
},
{
"epoch": 0.17,
"learning_rate": 1.98e-06,
"loss": 4.3454,
"step": 102
},
{
"epoch": 0.17,
"learning_rate": 2.0000000000000003e-06,
"loss": 4.3138,
"step": 103
},
{
"epoch": 0.17,
"learning_rate": 2.02e-06,
"loss": 4.2958,
"step": 104
},
{
"epoch": 0.18,
"learning_rate": 2.04e-06,
"loss": 4.2833,
"step": 105
},
{
"epoch": 0.18,
"learning_rate": 2.06e-06,
"loss": 4.3276,
"step": 106
},
{
"epoch": 0.18,
"learning_rate": 2.08e-06,
"loss": 4.4692,
"step": 107
},
{
"epoch": 0.18,
"learning_rate": 2.1000000000000002e-06,
"loss": 4.4968,
"step": 108
},
{
"epoch": 0.18,
"learning_rate": 2.12e-06,
"loss": 4.5626,
"step": 109
},
{
"epoch": 0.18,
"learning_rate": 2.1400000000000003e-06,
"loss": 4.3518,
"step": 110
},
{
"epoch": 0.19,
"learning_rate": 2.16e-06,
"loss": 4.3489,
"step": 111
},
{
"epoch": 0.19,
"learning_rate": 2.1800000000000003e-06,
"loss": 4.4177,
"step": 112
},
{
"epoch": 0.19,
"learning_rate": 2.2e-06,
"loss": 4.4097,
"step": 113
},
{
"epoch": 0.19,
"learning_rate": 2.2200000000000003e-06,
"loss": 4.4427,
"step": 114
},
{
"epoch": 0.19,
"learning_rate": 2.24e-06,
"loss": 4.3799,
"step": 115
},
{
"epoch": 0.2,
"learning_rate": 2.2600000000000004e-06,
"loss": 4.3865,
"step": 116
},
{
"epoch": 0.2,
"learning_rate": 2.28e-06,
"loss": 4.4844,
"step": 117
},
{
"epoch": 0.2,
"learning_rate": 2.3000000000000004e-06,
"loss": 4.2464,
"step": 118
},
{
"epoch": 0.2,
"learning_rate": 2.3200000000000002e-06,
"loss": 4.2288,
"step": 119
},
{
"epoch": 0.2,
"learning_rate": 2.3400000000000005e-06,
"loss": 4.2251,
"step": 120
},
{
"epoch": 0.2,
"learning_rate": 2.3600000000000003e-06,
"loss": 4.4691,
"step": 121
},
{
"epoch": 0.21,
"learning_rate": 2.38e-06,
"loss": 4.5464,
"step": 122
},
{
"epoch": 0.21,
"learning_rate": 2.4000000000000003e-06,
"loss": 4.3148,
"step": 123
},
{
"epoch": 0.21,
"learning_rate": 2.42e-06,
"loss": 4.3815,
"step": 124
},
{
"epoch": 0.21,
"learning_rate": 2.4400000000000004e-06,
"loss": 4.3865,
"step": 125
},
{
"epoch": 0.21,
"learning_rate": 2.46e-06,
"loss": 4.2059,
"step": 126
},
{
"epoch": 0.21,
"learning_rate": 2.4800000000000004e-06,
"loss": 4.264,
"step": 127
},
{
"epoch": 0.22,
"learning_rate": 2.5e-06,
"loss": 4.2287,
"step": 128
},
{
"epoch": 0.22,
"learning_rate": 2.52e-06,
"loss": 4.3775,
"step": 129
},
{
"epoch": 0.22,
"learning_rate": 2.5400000000000002e-06,
"loss": 4.403,
"step": 130
},
{
"epoch": 0.22,
"learning_rate": 2.56e-06,
"loss": 4.3432,
"step": 131
},
{
"epoch": 0.22,
"learning_rate": 2.5800000000000003e-06,
"loss": 4.2978,
"step": 132
},
{
"epoch": 0.22,
"learning_rate": 2.6e-06,
"loss": 4.3968,
"step": 133
},
{
"epoch": 0.23,
"learning_rate": 2.6200000000000003e-06,
"loss": 4.4003,
"step": 134
},
{
"epoch": 0.23,
"learning_rate": 2.64e-06,
"loss": 4.449,
"step": 135
},
{
"epoch": 0.23,
"learning_rate": 2.6600000000000004e-06,
"loss": 4.3175,
"step": 136
},
{
"epoch": 0.23,
"learning_rate": 2.68e-06,
"loss": 4.3273,
"step": 137
},
{
"epoch": 0.23,
"learning_rate": 2.7000000000000004e-06,
"loss": 4.3344,
"step": 138
},
{
"epoch": 0.23,
"learning_rate": 2.7200000000000002e-06,
"loss": 4.32,
"step": 139
},
{
"epoch": 0.24,
"learning_rate": 2.7400000000000004e-06,
"loss": 4.5413,
"step": 140
},
{
"epoch": 0.24,
"learning_rate": 2.7600000000000003e-06,
"loss": 4.4599,
"step": 141
},
{
"epoch": 0.24,
"learning_rate": 2.7800000000000005e-06,
"loss": 4.495,
"step": 142
},
{
"epoch": 0.24,
"learning_rate": 2.8000000000000003e-06,
"loss": 4.538,
"step": 143
},
{
"epoch": 0.24,
"learning_rate": 2.82e-06,
"loss": 4.3266,
"step": 144
},
{
"epoch": 0.24,
"learning_rate": 2.84e-06,
"loss": 4.429,
"step": 145
},
{
"epoch": 0.25,
"learning_rate": 2.86e-06,
"loss": 4.7391,
"step": 146
},
{
"epoch": 0.25,
"learning_rate": 2.88e-06,
"loss": 4.7207,
"step": 147
},
{
"epoch": 0.25,
"learning_rate": 2.9e-06,
"loss": 4.7066,
"step": 148
},
{
"epoch": 0.25,
"learning_rate": 2.92e-06,
"loss": 4.6515,
"step": 149
},
{
"epoch": 0.25,
"learning_rate": 2.9400000000000002e-06,
"loss": 4.7034,
"step": 150
},
{
"epoch": 0.25,
"learning_rate": 2.96e-06,
"loss": 4.2933,
"step": 151
},
{
"epoch": 0.26,
"learning_rate": 2.9800000000000003e-06,
"loss": 4.2286,
"step": 152
},
{
"epoch": 0.26,
"learning_rate": 3e-06,
"loss": 4.22,
"step": 153
},
{
"epoch": 0.26,
"learning_rate": 3.0200000000000003e-06,
"loss": 4.1533,
"step": 154
},
{
"epoch": 0.26,
"learning_rate": 3.04e-06,
"loss": 4.196,
"step": 155
},
{
"epoch": 0.26,
"learning_rate": 3.0600000000000003e-06,
"loss": 4.184,
"step": 156
},
{
"epoch": 0.26,
"learning_rate": 3.08e-06,
"loss": 4.3548,
"step": 157
},
{
"epoch": 0.27,
"learning_rate": 3.1000000000000004e-06,
"loss": 4.1817,
"step": 158
},
{
"epoch": 0.27,
"learning_rate": 3.12e-06,
"loss": 4.1367,
"step": 159
},
{
"epoch": 0.27,
"learning_rate": 3.1400000000000004e-06,
"loss": 4.2649,
"step": 160
},
{
"epoch": 0.27,
"learning_rate": 3.1600000000000002e-06,
"loss": 4.3083,
"step": 161
},
{
"epoch": 0.27,
"learning_rate": 3.1800000000000005e-06,
"loss": 4.1601,
"step": 162
},
{
"epoch": 0.27,
"learning_rate": 3.2000000000000003e-06,
"loss": 4.2107,
"step": 163
},
{
"epoch": 0.28,
"learning_rate": 3.2200000000000005e-06,
"loss": 4.2489,
"step": 164
},
{
"epoch": 0.28,
"learning_rate": 3.2400000000000003e-06,
"loss": 4.2803,
"step": 165
},
{
"epoch": 0.28,
"learning_rate": 3.2600000000000006e-06,
"loss": 4.3913,
"step": 166
},
{
"epoch": 0.28,
"learning_rate": 3.2800000000000004e-06,
"loss": 4.228,
"step": 167
},
{
"epoch": 0.28,
"learning_rate": 3.3000000000000006e-06,
"loss": 4.4536,
"step": 168
},
{
"epoch": 0.28,
"learning_rate": 3.3200000000000004e-06,
"loss": 4.2729,
"step": 169
},
{
"epoch": 0.29,
"learning_rate": 3.3400000000000006e-06,
"loss": 4.3269,
"step": 170
},
{
"epoch": 0.29,
"learning_rate": 3.3600000000000004e-06,
"loss": 4.1987,
"step": 171
},
{
"epoch": 0.29,
"learning_rate": 3.3800000000000007e-06,
"loss": 4.1876,
"step": 172
},
{
"epoch": 0.29,
"learning_rate": 3.4000000000000005e-06,
"loss": 4.2146,
"step": 173
},
{
"epoch": 0.29,
"learning_rate": 3.4200000000000007e-06,
"loss": 4.1946,
"step": 174
},
{
"epoch": 0.29,
"learning_rate": 3.44e-06,
"loss": 4.3494,
"step": 175
},
{
"epoch": 0.3,
"learning_rate": 3.46e-06,
"loss": 4.3716,
"step": 176
},
{
"epoch": 0.3,
"learning_rate": 3.48e-06,
"loss": 4.3028,
"step": 177
},
{
"epoch": 0.3,
"learning_rate": 3.5e-06,
"loss": 4.1935,
"step": 178
},
{
"epoch": 0.3,
"learning_rate": 3.52e-06,
"loss": 4.331,
"step": 179
},
{
"epoch": 0.3,
"learning_rate": 3.54e-06,
"loss": 4.1889,
"step": 180
},
{
"epoch": 0.3,
"learning_rate": 3.5600000000000002e-06,
"loss": 4.2626,
"step": 181
},
{
"epoch": 0.31,
"learning_rate": 3.58e-06,
"loss": 4.261,
"step": 182
},
{
"epoch": 0.31,
"learning_rate": 3.6000000000000003e-06,
"loss": 4.239,
"step": 183
},
{
"epoch": 0.31,
"learning_rate": 3.62e-06,
"loss": 4.1259,
"step": 184
},
{
"epoch": 0.31,
"learning_rate": 3.6400000000000003e-06,
"loss": 4.4795,
"step": 185
},
{
"epoch": 0.31,
"learning_rate": 3.66e-06,
"loss": 4.2646,
"step": 186
},
{
"epoch": 0.31,
"learning_rate": 3.6800000000000003e-06,
"loss": 4.3469,
"step": 187
},
{
"epoch": 0.32,
"learning_rate": 3.7e-06,
"loss": 4.3586,
"step": 188
},
{
"epoch": 0.32,
"learning_rate": 3.7200000000000004e-06,
"loss": 4.5375,
"step": 189
},
{
"epoch": 0.32,
"learning_rate": 3.74e-06,
"loss": 4.2441,
"step": 190
},
{
"epoch": 0.32,
"learning_rate": 3.7600000000000004e-06,
"loss": 4.5268,
"step": 191
},
{
"epoch": 0.32,
"learning_rate": 3.7800000000000002e-06,
"loss": 4.4442,
"step": 192
},
{
"epoch": 0.32,
"learning_rate": 3.8000000000000005e-06,
"loss": 4.3399,
"step": 193
},
{
"epoch": 0.33,
"learning_rate": 3.820000000000001e-06,
"loss": 4.4032,
"step": 194
},
{
"epoch": 0.33,
"learning_rate": 3.8400000000000005e-06,
"loss": 4.3523,
"step": 195
},
{
"epoch": 0.33,
"learning_rate": 3.86e-06,
"loss": 4.259,
"step": 196
},
{
"epoch": 0.33,
"learning_rate": 3.88e-06,
"loss": 4.666,
"step": 197
},
{
"epoch": 0.33,
"learning_rate": 3.900000000000001e-06,
"loss": 4.6439,
"step": 198
},
{
"epoch": 0.33,
"learning_rate": 3.920000000000001e-06,
"loss": 4.5596,
"step": 199
},
{
"epoch": 0.34,
"learning_rate": 3.94e-06,
"loss": 4.7306,
"step": 200
},
{
"epoch": 0.34,
"learning_rate": 3.96e-06,
"loss": 4.1171,
"step": 201
},
{
"epoch": 0.34,
"learning_rate": 3.980000000000001e-06,
"loss": 4.2316,
"step": 202
},
{
"epoch": 0.34,
"learning_rate": 4.000000000000001e-06,
"loss": 4.2785,
"step": 203
},
{
"epoch": 0.34,
"learning_rate": 4.0200000000000005e-06,
"loss": 4.1903,
"step": 204
},
{
"epoch": 0.34,
"learning_rate": 4.04e-06,
"loss": 4.2281,
"step": 205
},
{
"epoch": 0.35,
"learning_rate": 4.060000000000001e-06,
"loss": 4.1607,
"step": 206
},
{
"epoch": 0.35,
"learning_rate": 4.08e-06,
"loss": 4.2343,
"step": 207
},
{
"epoch": 0.35,
"learning_rate": 4.1e-06,
"loss": 4.1985,
"step": 208
},
{
"epoch": 0.35,
"learning_rate": 4.12e-06,
"loss": 4.2208,
"step": 209
},
{
"epoch": 0.35,
"learning_rate": 4.14e-06,
"loss": 4.3262,
"step": 210
},
{
"epoch": 0.35,
"learning_rate": 4.16e-06,
"loss": 4.2993,
"step": 211
},
{
"epoch": 0.36,
"learning_rate": 4.18e-06,
"loss": 4.2977,
"step": 212
},
{
"epoch": 0.36,
"learning_rate": 4.2000000000000004e-06,
"loss": 4.2572,
"step": 213
},
{
"epoch": 0.36,
"learning_rate": 4.22e-06,
"loss": 4.3531,
"step": 214
},
{
"epoch": 0.36,
"learning_rate": 4.24e-06,
"loss": 4.2164,
"step": 215
},
{
"epoch": 0.36,
"learning_rate": 4.26e-06,
"loss": 4.2395,
"step": 216
},
{
"epoch": 0.36,
"learning_rate": 4.2800000000000005e-06,
"loss": 4.125,
"step": 217
},
{
"epoch": 0.37,
"learning_rate": 4.3e-06,
"loss": 4.2381,
"step": 218
},
{
"epoch": 0.37,
"learning_rate": 4.32e-06,
"loss": 4.2245,
"step": 219
},
{
"epoch": 0.37,
"learning_rate": 4.34e-06,
"loss": 4.3043,
"step": 220
},
{
"epoch": 0.37,
"learning_rate": 4.360000000000001e-06,
"loss": 4.3443,
"step": 221
},
{
"epoch": 0.37,
"learning_rate": 4.38e-06,
"loss": 4.3346,
"step": 222
},
{
"epoch": 0.37,
"learning_rate": 4.4e-06,
"loss": 4.1726,
"step": 223
},
{
"epoch": 0.38,
"learning_rate": 4.42e-06,
"loss": 4.2873,
"step": 224
},
{
"epoch": 0.38,
"learning_rate": 4.440000000000001e-06,
"loss": 4.4616,
"step": 225
},
{
"epoch": 0.38,
"learning_rate": 4.4600000000000005e-06,
"loss": 4.2406,
"step": 226
},
{
"epoch": 0.38,
"learning_rate": 4.48e-06,
"loss": 4.1887,
"step": 227
},
{
"epoch": 0.38,
"learning_rate": 4.5e-06,
"loss": 4.2441,
"step": 228
},
{
"epoch": 0.39,
"learning_rate": 4.520000000000001e-06,
"loss": 4.377,
"step": 229
},
{
"epoch": 0.39,
"learning_rate": 4.540000000000001e-06,
"loss": 4.3398,
"step": 230
},
{
"epoch": 0.39,
"learning_rate": 4.56e-06,
"loss": 4.2106,
"step": 231
},
{
"epoch": 0.39,
"learning_rate": 4.58e-06,
"loss": 4.2779,
"step": 232
},
{
"epoch": 0.39,
"learning_rate": 4.600000000000001e-06,
"loss": 4.3087,
"step": 233
},
{
"epoch": 0.39,
"learning_rate": 4.620000000000001e-06,
"loss": 4.2863,
"step": 234
},
{
"epoch": 0.4,
"learning_rate": 4.6400000000000005e-06,
"loss": 4.4418,
"step": 235
},
{
"epoch": 0.4,
"learning_rate": 4.66e-06,
"loss": 4.2034,
"step": 236
},
{
"epoch": 0.4,
"learning_rate": 4.680000000000001e-06,
"loss": 4.2032,
"step": 237
},
{
"epoch": 0.4,
"learning_rate": 4.7e-06,
"loss": 4.2322,
"step": 238
},
{
"epoch": 0.4,
"learning_rate": 4.7200000000000005e-06,
"loss": 4.1456,
"step": 239
},
{
"epoch": 0.4,
"learning_rate": 4.74e-06,
"loss": 4.3317,
"step": 240
},
{
"epoch": 0.41,
"learning_rate": 4.76e-06,
"loss": 4.2461,
"step": 241
},
{
"epoch": 0.41,
"learning_rate": 4.78e-06,
"loss": 4.2655,
"step": 242
},
{
"epoch": 0.41,
"learning_rate": 4.800000000000001e-06,
"loss": 4.2899,
"step": 243
},
{
"epoch": 0.41,
"learning_rate": 4.8200000000000004e-06,
"loss": 4.3705,
"step": 244
},
{
"epoch": 0.41,
"learning_rate": 4.84e-06,
"loss": 4.439,
"step": 245
},
{
"epoch": 0.41,
"learning_rate": 4.86e-06,
"loss": 4.4711,
"step": 246
},
{
"epoch": 0.42,
"learning_rate": 4.880000000000001e-06,
"loss": 4.4541,
"step": 247
},
{
"epoch": 0.42,
"learning_rate": 4.9000000000000005e-06,
"loss": 4.5902,
"step": 248
},
{
"epoch": 0.42,
"learning_rate": 4.92e-06,
"loss": 4.5564,
"step": 249
},
{
"epoch": 0.42,
"learning_rate": 4.94e-06,
"loss": 4.5735,
"step": 250
},
{
"epoch": 0.42,
"learning_rate": 4.960000000000001e-06,
"loss": 4.2696,
"step": 251
},
{
"epoch": 0.42,
"learning_rate": 4.980000000000001e-06,
"loss": 4.2687,
"step": 252
},
{
"epoch": 0.43,
"learning_rate": 5e-06,
"loss": 4.2286,
"step": 253
},
{
"epoch": 0.43,
"learning_rate": 5.02e-06,
"loss": 4.1151,
"step": 254
},
{
"epoch": 0.43,
"learning_rate": 5.04e-06,
"loss": 4.294,
"step": 255
},
{
"epoch": 0.43,
"learning_rate": 5.060000000000001e-06,
"loss": 4.244,
"step": 256
},
{
"epoch": 0.43,
"learning_rate": 5.0800000000000005e-06,
"loss": 4.1804,
"step": 257
},
{
"epoch": 0.43,
"learning_rate": 5.1e-06,
"loss": 4.2024,
"step": 258
},
{
"epoch": 0.44,
"learning_rate": 5.12e-06,
"loss": 4.2631,
"step": 259
},
{
"epoch": 0.44,
"learning_rate": 5.140000000000001e-06,
"loss": 4.269,
"step": 260
},
{
"epoch": 0.44,
"learning_rate": 5.1600000000000006e-06,
"loss": 4.2433,
"step": 261
},
{
"epoch": 0.44,
"learning_rate": 5.18e-06,
"loss": 4.1892,
"step": 262
},
{
"epoch": 0.44,
"learning_rate": 5.2e-06,
"loss": 4.2933,
"step": 263
},
{
"epoch": 0.44,
"learning_rate": 5.220000000000001e-06,
"loss": 4.0133,
"step": 264
},
{
"epoch": 0.45,
"learning_rate": 5.240000000000001e-06,
"loss": 4.1051,
"step": 265
},
{
"epoch": 0.45,
"learning_rate": 5.2600000000000005e-06,
"loss": 4.2157,
"step": 266
},
{
"epoch": 0.45,
"learning_rate": 5.28e-06,
"loss": 4.2397,
"step": 267
},
{
"epoch": 0.45,
"learning_rate": 5.300000000000001e-06,
"loss": 4.177,
"step": 268
},
{
"epoch": 0.45,
"learning_rate": 5.320000000000001e-06,
"loss": 4.292,
"step": 269
},
{
"epoch": 0.45,
"learning_rate": 5.3400000000000005e-06,
"loss": 4.2874,
"step": 270
},
{
"epoch": 0.46,
"learning_rate": 5.36e-06,
"loss": 4.2115,
"step": 271
},
{
"epoch": 0.46,
"learning_rate": 5.380000000000001e-06,
"loss": 4.3472,
"step": 272
},
{
"epoch": 0.46,
"learning_rate": 5.400000000000001e-06,
"loss": 4.2087,
"step": 273
},
{
"epoch": 0.46,
"learning_rate": 5.420000000000001e-06,
"loss": 4.2198,
"step": 274
},
{
"epoch": 0.46,
"learning_rate": 5.4400000000000004e-06,
"loss": 4.2855,
"step": 275
},
{
"epoch": 0.46,
"learning_rate": 5.460000000000001e-06,
"loss": 4.2137,
"step": 276
},
{
"epoch": 0.47,
"learning_rate": 5.480000000000001e-06,
"loss": 4.3496,
"step": 277
},
{
"epoch": 0.47,
"learning_rate": 5.500000000000001e-06,
"loss": 4.1356,
"step": 278
},
{
"epoch": 0.47,
"learning_rate": 5.5200000000000005e-06,
"loss": 4.317,
"step": 279
},
{
"epoch": 0.47,
"learning_rate": 5.540000000000001e-06,
"loss": 4.3791,
"step": 280
},
{
"epoch": 0.47,
"learning_rate": 5.560000000000001e-06,
"loss": 4.3842,
"step": 281
},
{
"epoch": 0.47,
"learning_rate": 5.580000000000001e-06,
"loss": 4.2575,
"step": 282
},
{
"epoch": 0.48,
"learning_rate": 5.600000000000001e-06,
"loss": 4.1577,
"step": 283
},
{
"epoch": 0.48,
"learning_rate": 5.620000000000001e-06,
"loss": 4.153,
"step": 284
},
{
"epoch": 0.48,
"learning_rate": 5.64e-06,
"loss": 4.4077,
"step": 285
},
{
"epoch": 0.48,
"learning_rate": 5.66e-06,
"loss": 4.1658,
"step": 286
},
{
"epoch": 0.48,
"learning_rate": 5.68e-06,
"loss": 4.5108,
"step": 287
},
{
"epoch": 0.48,
"learning_rate": 5.7e-06,
"loss": 4.0048,
"step": 288
},
{
"epoch": 0.49,
"learning_rate": 5.72e-06,
"loss": 4.2594,
"step": 289
},
{
"epoch": 0.49,
"learning_rate": 5.74e-06,
"loss": 4.2277,
"step": 290
},
{
"epoch": 0.49,
"learning_rate": 5.76e-06,
"loss": 4.4854,
"step": 291
},
{
"epoch": 0.49,
"learning_rate": 5.78e-06,
"loss": 4.1498,
"step": 292
},
{
"epoch": 0.49,
"learning_rate": 5.8e-06,
"loss": 4.3567,
"step": 293
},
{
"epoch": 0.49,
"learning_rate": 5.82e-06,
"loss": 4.3946,
"step": 294
},
{
"epoch": 0.5,
"learning_rate": 5.84e-06,
"loss": 4.3868,
"step": 295
},
{
"epoch": 0.5,
"learning_rate": 5.86e-06,
"loss": 4.2016,
"step": 296
},
{
"epoch": 0.5,
"learning_rate": 5.8800000000000005e-06,
"loss": 4.2791,
"step": 297
},
{
"epoch": 0.5,
"learning_rate": 5.9e-06,
"loss": 4.4595,
"step": 298
},
{
"epoch": 0.5,
"learning_rate": 5.92e-06,
"loss": 4.4781,
"step": 299
},
{
"epoch": 0.5,
"learning_rate": 5.94e-06,
"loss": 4.7221,
"step": 300
},
{
"epoch": 0.51,
"learning_rate": 5.9600000000000005e-06,
"loss": 4.3254,
"step": 301
},
{
"epoch": 0.51,
"learning_rate": 5.98e-06,
"loss": 4.266,
"step": 302
},
{
"epoch": 0.51,
"learning_rate": 6e-06,
"loss": 4.3527,
"step": 303
},
{
"epoch": 0.51,
"learning_rate": 6.02e-06,
"loss": 4.1289,
"step": 304
},
{
"epoch": 0.51,
"learning_rate": 6.040000000000001e-06,
"loss": 4.3227,
"step": 305
},
{
"epoch": 0.51,
"learning_rate": 6.0600000000000004e-06,
"loss": 4.3129,
"step": 306
},
{
"epoch": 0.52,
"learning_rate": 6.08e-06,
"loss": 4.1029,
"step": 307
},
{
"epoch": 0.52,
"learning_rate": 6.1e-06,
"loss": 4.3159,
"step": 308
},
{
"epoch": 0.52,
"learning_rate": 6.120000000000001e-06,
"loss": 4.2134,
"step": 309
},
{
"epoch": 0.52,
"learning_rate": 6.1400000000000005e-06,
"loss": 4.3056,
"step": 310
},
{
"epoch": 0.52,
"learning_rate": 6.16e-06,
"loss": 4.1592,
"step": 311
},
{
"epoch": 0.52,
"learning_rate": 6.18e-06,
"loss": 4.2038,
"step": 312
},
{
"epoch": 0.53,
"learning_rate": 6.200000000000001e-06,
"loss": 4.2221,
"step": 313
},
{
"epoch": 0.53,
"learning_rate": 6.220000000000001e-06,
"loss": 4.2282,
"step": 314
},
{
"epoch": 0.53,
"learning_rate": 6.24e-06,
"loss": 4.2493,
"step": 315
},
{
"epoch": 0.53,
"learning_rate": 6.26e-06,
"loss": 4.2448,
"step": 316
},
{
"epoch": 0.53,
"learning_rate": 6.280000000000001e-06,
"loss": 4.1976,
"step": 317
},
{
"epoch": 0.53,
"learning_rate": 6.300000000000001e-06,
"loss": 4.2818,
"step": 318
},
{
"epoch": 0.54,
"learning_rate": 6.3200000000000005e-06,
"loss": 4.1041,
"step": 319
},
{
"epoch": 0.54,
"learning_rate": 6.34e-06,
"loss": 4.2671,
"step": 320
},
{
"epoch": 0.54,
"learning_rate": 6.360000000000001e-06,
"loss": 4.1219,
"step": 321
},
{
"epoch": 0.54,
"learning_rate": 6.380000000000001e-06,
"loss": 4.1702,
"step": 322
},
{
"epoch": 0.54,
"learning_rate": 6.4000000000000006e-06,
"loss": 4.187,
"step": 323
},
{
"epoch": 0.54,
"learning_rate": 6.42e-06,
"loss": 4.2524,
"step": 324
},
{
"epoch": 0.55,
"learning_rate": 6.440000000000001e-06,
"loss": 4.2123,
"step": 325
},
{
"epoch": 0.55,
"learning_rate": 6.460000000000001e-06,
"loss": 4.1287,
"step": 326
},
{
"epoch": 0.55,
"learning_rate": 6.480000000000001e-06,
"loss": 4.4177,
"step": 327
},
{
"epoch": 0.55,
"learning_rate": 6.5000000000000004e-06,
"loss": 4.1621,
"step": 328
},
{
"epoch": 0.55,
"learning_rate": 6.520000000000001e-06,
"loss": 4.0732,
"step": 329
},
{
"epoch": 0.55,
"learning_rate": 6.540000000000001e-06,
"loss": 4.3004,
"step": 330
},
{
"epoch": 0.56,
"learning_rate": 6.560000000000001e-06,
"loss": 4.1879,
"step": 331
},
{
"epoch": 0.56,
"learning_rate": 6.5800000000000005e-06,
"loss": 4.1324,
"step": 332
},
{
"epoch": 0.56,
"learning_rate": 6.600000000000001e-06,
"loss": 4.2003,
"step": 333
},
{
"epoch": 0.56,
"learning_rate": 6.620000000000001e-06,
"loss": 4.2383,
"step": 334
},
{
"epoch": 0.56,
"learning_rate": 6.640000000000001e-06,
"loss": 4.2517,
"step": 335
},
{
"epoch": 0.56,
"learning_rate": 6.660000000000001e-06,
"loss": 4.1818,
"step": 336
},
{
"epoch": 0.57,
"learning_rate": 6.680000000000001e-06,
"loss": 4.148,
"step": 337
},
{
"epoch": 0.57,
"learning_rate": 6.700000000000001e-06,
"loss": 4.2455,
"step": 338
},
{
"epoch": 0.57,
"learning_rate": 6.720000000000001e-06,
"loss": 4.2638,
"step": 339
},
{
"epoch": 0.57,
"learning_rate": 6.740000000000001e-06,
"loss": 4.2869,
"step": 340
},
{
"epoch": 0.57,
"learning_rate": 6.760000000000001e-06,
"loss": 4.2918,
"step": 341
},
{
"epoch": 0.58,
"learning_rate": 6.780000000000001e-06,
"loss": 4.3944,
"step": 342
},
{
"epoch": 0.58,
"learning_rate": 6.800000000000001e-06,
"loss": 4.4212,
"step": 343
},
{
"epoch": 0.58,
"learning_rate": 6.820000000000001e-06,
"loss": 4.2959,
"step": 344
},
{
"epoch": 0.58,
"learning_rate": 6.8400000000000014e-06,
"loss": 4.4497,
"step": 345
},
{
"epoch": 0.58,
"learning_rate": 6.860000000000001e-06,
"loss": 4.1427,
"step": 346
},
{
"epoch": 0.58,
"learning_rate": 6.88e-06,
"loss": 4.2791,
"step": 347
},
{
"epoch": 0.59,
"learning_rate": 6.9e-06,
"loss": 4.6095,
"step": 348
},
{
"epoch": 0.59,
"learning_rate": 6.92e-06,
"loss": 4.4704,
"step": 349
},
{
"epoch": 0.59,
"learning_rate": 6.9400000000000005e-06,
"loss": 4.7726,
"step": 350
},
{
"epoch": 0.59,
"learning_rate": 6.96e-06,
"loss": 4.1317,
"step": 351
},
{
"epoch": 0.59,
"learning_rate": 6.98e-06,
"loss": 4.2371,
"step": 352
},
{
"epoch": 0.59,
"learning_rate": 7e-06,
"loss": 4.2641,
"step": 353
},
{
"epoch": 0.6,
"learning_rate": 7.0200000000000006e-06,
"loss": 4.1023,
"step": 354
},
{
"epoch": 0.6,
"learning_rate": 7.04e-06,
"loss": 4.1948,
"step": 355
},
{
"epoch": 0.6,
"learning_rate": 7.06e-06,
"loss": 4.1926,
"step": 356
},
{
"epoch": 0.6,
"learning_rate": 7.08e-06,
"loss": 4.215,
"step": 357
},
{
"epoch": 0.6,
"learning_rate": 7.100000000000001e-06,
"loss": 4.1484,
"step": 358
},
{
"epoch": 0.6,
"learning_rate": 7.1200000000000004e-06,
"loss": 4.1754,
"step": 359
},
{
"epoch": 0.61,
"learning_rate": 7.14e-06,
"loss": 4.1181,
"step": 360
},
{
"epoch": 0.61,
"learning_rate": 7.16e-06,
"loss": 4.2056,
"step": 361
},
{
"epoch": 0.61,
"learning_rate": 7.180000000000001e-06,
"loss": 4.2102,
"step": 362
},
{
"epoch": 0.61,
"learning_rate": 7.2000000000000005e-06,
"loss": 4.2621,
"step": 363
},
{
"epoch": 0.61,
"learning_rate": 7.22e-06,
"loss": 4.1432,
"step": 364
},
{
"epoch": 0.61,
"learning_rate": 7.24e-06,
"loss": 4.2162,
"step": 365
},
{
"epoch": 0.62,
"learning_rate": 7.260000000000001e-06,
"loss": 4.1775,
"step": 366
},
{
"epoch": 0.62,
"learning_rate": 7.280000000000001e-06,
"loss": 4.1788,
"step": 367
},
{
"epoch": 0.62,
"learning_rate": 7.3e-06,
"loss": 4.1636,
"step": 368
},
{
"epoch": 0.62,
"learning_rate": 7.32e-06,
"loss": 4.2602,
"step": 369
},
{
"epoch": 0.62,
"learning_rate": 7.340000000000001e-06,
"loss": 4.1437,
"step": 370
},
{
"epoch": 0.62,
"learning_rate": 7.360000000000001e-06,
"loss": 4.0553,
"step": 371
},
{
"epoch": 0.63,
"learning_rate": 7.3800000000000005e-06,
"loss": 4.219,
"step": 372
},
{
"epoch": 0.63,
"learning_rate": 7.4e-06,
"loss": 4.0653,
"step": 373
},
{
"epoch": 0.63,
"learning_rate": 7.420000000000001e-06,
"loss": 4.0878,
"step": 374
},
{
"epoch": 0.63,
"learning_rate": 7.440000000000001e-06,
"loss": 4.2545,
"step": 375
},
{
"epoch": 0.63,
"learning_rate": 7.4600000000000006e-06,
"loss": 4.1381,
"step": 376
},
{
"epoch": 0.63,
"learning_rate": 7.48e-06,
"loss": 4.2293,
"step": 377
},
{
"epoch": 0.64,
"learning_rate": 7.500000000000001e-06,
"loss": 4.2858,
"step": 378
},
{
"epoch": 0.64,
"learning_rate": 7.520000000000001e-06,
"loss": 4.1157,
"step": 379
},
{
"epoch": 0.64,
"learning_rate": 7.540000000000001e-06,
"loss": 4.3363,
"step": 380
},
{
"epoch": 0.64,
"learning_rate": 7.5600000000000005e-06,
"loss": 4.2621,
"step": 381
},
{
"epoch": 0.64,
"learning_rate": 7.58e-06,
"loss": 4.1278,
"step": 382
},
{
"epoch": 0.64,
"learning_rate": 7.600000000000001e-06,
"loss": 4.0316,
"step": 383
},
{
"epoch": 0.65,
"learning_rate": 7.620000000000001e-06,
"loss": 4.2188,
"step": 384
},
{
"epoch": 0.65,
"learning_rate": 7.640000000000001e-06,
"loss": 4.4046,
"step": 385
},
{
"epoch": 0.65,
"learning_rate": 7.660000000000001e-06,
"loss": 4.1662,
"step": 386
},
{
"epoch": 0.65,
"learning_rate": 7.680000000000001e-06,
"loss": 4.3235,
"step": 387
},
{
"epoch": 0.65,
"learning_rate": 7.7e-06,
"loss": 4.2542,
"step": 388
},
{
"epoch": 0.65,
"learning_rate": 7.72e-06,
"loss": 4.142,
"step": 389
},
{
"epoch": 0.66,
"learning_rate": 7.74e-06,
"loss": 4.1514,
"step": 390
},
{
"epoch": 0.66,
"learning_rate": 7.76e-06,
"loss": 4.366,
"step": 391
},
{
"epoch": 0.66,
"learning_rate": 7.78e-06,
"loss": 4.2973,
"step": 392
},
{
"epoch": 0.66,
"learning_rate": 7.800000000000002e-06,
"loss": 4.5381,
"step": 393
},
{
"epoch": 0.66,
"learning_rate": 7.820000000000001e-06,
"loss": 4.3709,
"step": 394
},
{
"epoch": 0.66,
"learning_rate": 7.840000000000001e-06,
"loss": 4.2935,
"step": 395
},
{
"epoch": 0.67,
"learning_rate": 7.860000000000001e-06,
"loss": 4.2228,
"step": 396
},
{
"epoch": 0.67,
"learning_rate": 7.88e-06,
"loss": 4.4299,
"step": 397
},
{
"epoch": 0.67,
"learning_rate": 7.9e-06,
"loss": 4.6182,
"step": 398
},
{
"epoch": 0.67,
"learning_rate": 7.92e-06,
"loss": 4.43,
"step": 399
},
{
"epoch": 0.67,
"learning_rate": 7.94e-06,
"loss": 4.5297,
"step": 400
},
{
"epoch": 0.67,
"learning_rate": 7.960000000000002e-06,
"loss": 4.2478,
"step": 401
},
{
"epoch": 0.68,
"learning_rate": 7.980000000000002e-06,
"loss": 4.15,
"step": 402
},
{
"epoch": 0.68,
"learning_rate": 8.000000000000001e-06,
"loss": 4.2147,
"step": 403
},
{
"epoch": 0.68,
"learning_rate": 8.020000000000001e-06,
"loss": 4.1922,
"step": 404
},
{
"epoch": 0.68,
"learning_rate": 8.040000000000001e-06,
"loss": 4.1304,
"step": 405
},
{
"epoch": 0.68,
"learning_rate": 8.06e-06,
"loss": 4.1817,
"step": 406
},
{
"epoch": 0.68,
"learning_rate": 8.08e-06,
"loss": 4.1808,
"step": 407
},
{
"epoch": 0.69,
"learning_rate": 8.1e-06,
"loss": 4.0998,
"step": 408
},
{
"epoch": 0.69,
"learning_rate": 8.120000000000002e-06,
"loss": 4.1643,
"step": 409
},
{
"epoch": 0.69,
"learning_rate": 8.14e-06,
"loss": 4.1098,
"step": 410
},
{
"epoch": 0.69,
"learning_rate": 8.16e-06,
"loss": 4.1225,
"step": 411
},
{
"epoch": 0.69,
"learning_rate": 8.18e-06,
"loss": 4.2878,
"step": 412
},
{
"epoch": 0.69,
"learning_rate": 8.2e-06,
"loss": 4.1765,
"step": 413
},
{
"epoch": 0.7,
"learning_rate": 8.220000000000001e-06,
"loss": 4.2272,
"step": 414
},
{
"epoch": 0.7,
"learning_rate": 8.24e-06,
"loss": 4.2027,
"step": 415
},
{
"epoch": 0.7,
"learning_rate": 8.26e-06,
"loss": 4.1183,
"step": 416
},
{
"epoch": 0.7,
"learning_rate": 8.28e-06,
"loss": 4.1938,
"step": 417
},
{
"epoch": 0.7,
"learning_rate": 8.3e-06,
"loss": 4.0316,
"step": 418
},
{
"epoch": 0.7,
"learning_rate": 8.32e-06,
"loss": 4.2936,
"step": 419
},
{
"epoch": 0.71,
"learning_rate": 8.34e-06,
"loss": 4.1993,
"step": 420
},
{
"epoch": 0.71,
"learning_rate": 8.36e-06,
"loss": 3.973,
"step": 421
},
{
"epoch": 0.71,
"learning_rate": 8.380000000000001e-06,
"loss": 4.1127,
"step": 422
},
{
"epoch": 0.71,
"learning_rate": 8.400000000000001e-06,
"loss": 4.2403,
"step": 423
},
{
"epoch": 0.71,
"learning_rate": 8.42e-06,
"loss": 4.1748,
"step": 424
},
{
"epoch": 0.71,
"learning_rate": 8.44e-06,
"loss": 4.1619,
"step": 425
},
{
"epoch": 0.72,
"learning_rate": 8.46e-06,
"loss": 4.2359,
"step": 426
},
{
"epoch": 0.72,
"learning_rate": 8.48e-06,
"loss": 4.1836,
"step": 427
},
{
"epoch": 0.72,
"learning_rate": 8.5e-06,
"loss": 4.1234,
"step": 428
},
{
"epoch": 0.72,
"learning_rate": 8.52e-06,
"loss": 4.2324,
"step": 429
},
{
"epoch": 0.72,
"learning_rate": 8.540000000000001e-06,
"loss": 4.1858,
"step": 430
},
{
"epoch": 0.72,
"learning_rate": 8.560000000000001e-06,
"loss": 4.2511,
"step": 431
},
{
"epoch": 0.73,
"learning_rate": 8.580000000000001e-06,
"loss": 4.0518,
"step": 432
},
{
"epoch": 0.73,
"learning_rate": 8.6e-06,
"loss": 4.2866,
"step": 433
},
{
"epoch": 0.73,
"learning_rate": 8.62e-06,
"loss": 4.2093,
"step": 434
},
{
"epoch": 0.73,
"learning_rate": 8.64e-06,
"loss": 4.2793,
"step": 435
},
{
"epoch": 0.73,
"learning_rate": 8.66e-06,
"loss": 4.1262,
"step": 436
},
{
"epoch": 0.73,
"learning_rate": 8.68e-06,
"loss": 4.0688,
"step": 437
},
{
"epoch": 0.74,
"learning_rate": 8.700000000000001e-06,
"loss": 4.2754,
"step": 438
},
{
"epoch": 0.74,
"learning_rate": 8.720000000000001e-06,
"loss": 4.1836,
"step": 439
},
{
"epoch": 0.74,
"learning_rate": 8.740000000000001e-06,
"loss": 4.3704,
"step": 440
},
{
"epoch": 0.74,
"learning_rate": 8.76e-06,
"loss": 4.2427,
"step": 441
},
{
"epoch": 0.74,
"learning_rate": 8.78e-06,
"loss": 4.422,
"step": 442
},
{
"epoch": 0.74,
"learning_rate": 8.8e-06,
"loss": 4.3321,
"step": 443
},
{
"epoch": 0.75,
"learning_rate": 8.82e-06,
"loss": 4.2648,
"step": 444
},
{
"epoch": 0.75,
"learning_rate": 8.84e-06,
"loss": 4.089,
"step": 445
},
{
"epoch": 0.75,
"learning_rate": 8.860000000000002e-06,
"loss": 4.3901,
"step": 446
},
{
"epoch": 0.75,
"learning_rate": 8.880000000000001e-06,
"loss": 4.3074,
"step": 447
},
{
"epoch": 0.75,
"learning_rate": 8.900000000000001e-06,
"loss": 4.2438,
"step": 448
},
{
"epoch": 0.75,
"learning_rate": 8.920000000000001e-06,
"loss": 4.4442,
"step": 449
},
{
"epoch": 0.76,
"learning_rate": 8.94e-06,
"loss": 4.5537,
"step": 450
},
{
"epoch": 0.76,
"learning_rate": 8.96e-06,
"loss": 4.2158,
"step": 451
},
{
"epoch": 0.76,
"learning_rate": 8.98e-06,
"loss": 4.2222,
"step": 452
},
{
"epoch": 0.76,
"learning_rate": 9e-06,
"loss": 4.1596,
"step": 453
},
{
"epoch": 0.76,
"learning_rate": 9.020000000000002e-06,
"loss": 4.0322,
"step": 454
},
{
"epoch": 0.77,
"learning_rate": 9.040000000000002e-06,
"loss": 4.2099,
"step": 455
},
{
"epoch": 0.77,
"learning_rate": 9.060000000000001e-06,
"loss": 4.1521,
"step": 456
},
{
"epoch": 0.77,
"learning_rate": 9.080000000000001e-06,
"loss": 4.2381,
"step": 457
},
{
"epoch": 0.77,
"learning_rate": 9.100000000000001e-06,
"loss": 4.2634,
"step": 458
},
{
"epoch": 0.77,
"learning_rate": 9.12e-06,
"loss": 4.1887,
"step": 459
},
{
"epoch": 0.77,
"learning_rate": 9.14e-06,
"loss": 4.1982,
"step": 460
},
{
"epoch": 0.78,
"learning_rate": 9.16e-06,
"loss": 4.1105,
"step": 461
},
{
"epoch": 0.78,
"learning_rate": 9.180000000000002e-06,
"loss": 4.1386,
"step": 462
},
{
"epoch": 0.78,
"learning_rate": 9.200000000000002e-06,
"loss": 4.1434,
"step": 463
},
{
"epoch": 0.78,
"learning_rate": 9.220000000000002e-06,
"loss": 4.2041,
"step": 464
},
{
"epoch": 0.78,
"learning_rate": 9.240000000000001e-06,
"loss": 4.2266,
"step": 465
},
{
"epoch": 0.78,
"learning_rate": 9.260000000000001e-06,
"loss": 4.1666,
"step": 466
},
{
"epoch": 0.79,
"learning_rate": 9.280000000000001e-06,
"loss": 4.1755,
"step": 467
},
{
"epoch": 0.79,
"learning_rate": 9.3e-06,
"loss": 4.1936,
"step": 468
},
{
"epoch": 0.79,
"learning_rate": 9.32e-06,
"loss": 4.1813,
"step": 469
},
{
"epoch": 0.79,
"learning_rate": 9.340000000000002e-06,
"loss": 4.1513,
"step": 470
},
{
"epoch": 0.79,
"learning_rate": 9.360000000000002e-06,
"loss": 4.0757,
"step": 471
},
{
"epoch": 0.79,
"learning_rate": 9.38e-06,
"loss": 4.1831,
"step": 472
},
{
"epoch": 0.8,
"learning_rate": 9.4e-06,
"loss": 4.1809,
"step": 473
},
{
"epoch": 0.8,
"learning_rate": 9.42e-06,
"loss": 4.0661,
"step": 474
},
{
"epoch": 0.8,
"learning_rate": 9.440000000000001e-06,
"loss": 4.1123,
"step": 475
},
{
"epoch": 0.8,
"learning_rate": 9.460000000000001e-06,
"loss": 4.1363,
"step": 476
},
{
"epoch": 0.8,
"learning_rate": 9.48e-06,
"loss": 4.1625,
"step": 477
},
{
"epoch": 0.8,
"learning_rate": 9.5e-06,
"loss": 4.1857,
"step": 478
},
{
"epoch": 0.81,
"learning_rate": 9.52e-06,
"loss": 4.0617,
"step": 479
},
{
"epoch": 0.81,
"learning_rate": 9.54e-06,
"loss": 4.0652,
"step": 480
},
{
"epoch": 0.81,
"learning_rate": 9.56e-06,
"loss": 4.3014,
"step": 481
},
{
"epoch": 0.81,
"learning_rate": 9.58e-06,
"loss": 4.1331,
"step": 482
},
{
"epoch": 0.81,
"learning_rate": 9.600000000000001e-06,
"loss": 4.2987,
"step": 483
},
{
"epoch": 0.81,
"learning_rate": 9.620000000000001e-06,
"loss": 4.2466,
"step": 484
},
{
"epoch": 0.82,
"learning_rate": 9.640000000000001e-06,
"loss": 4.2273,
"step": 485
},
{
"epoch": 0.82,
"learning_rate": 9.66e-06,
"loss": 4.1424,
"step": 486
},
{
"epoch": 0.82,
"learning_rate": 9.68e-06,
"loss": 4.2659,
"step": 487
},
{
"epoch": 0.82,
"learning_rate": 9.7e-06,
"loss": 4.2862,
"step": 488
},
{
"epoch": 0.82,
"learning_rate": 9.72e-06,
"loss": 4.2092,
"step": 489
},
{
"epoch": 0.82,
"learning_rate": 9.74e-06,
"loss": 4.1234,
"step": 490
},
{
"epoch": 0.83,
"learning_rate": 9.760000000000001e-06,
"loss": 4.3266,
"step": 491
},
{
"epoch": 0.83,
"learning_rate": 9.780000000000001e-06,
"loss": 4.1458,
"step": 492
},
{
"epoch": 0.83,
"learning_rate": 9.800000000000001e-06,
"loss": 4.027,
"step": 493
},
{
"epoch": 0.83,
"learning_rate": 9.820000000000001e-06,
"loss": 4.2423,
"step": 494
},
{
"epoch": 0.83,
"learning_rate": 9.84e-06,
"loss": 4.3469,
"step": 495
},
{
"epoch": 0.83,
"learning_rate": 9.86e-06,
"loss": 4.391,
"step": 496
},
{
"epoch": 0.84,
"learning_rate": 9.88e-06,
"loss": 4.267,
"step": 497
},
{
"epoch": 0.84,
"learning_rate": 9.9e-06,
"loss": 4.4238,
"step": 498
},
{
"epoch": 0.84,
"learning_rate": 9.920000000000002e-06,
"loss": 4.4835,
"step": 499
},
{
"epoch": 0.84,
"learning_rate": 9.940000000000001e-06,
"loss": 4.4132,
"step": 500
},
{
"epoch": 0.84,
"eval_loss": 4.189728260040283,
"eval_runtime": 775.4835,
"eval_samples_per_second": 3.407,
"eval_steps_per_second": 0.285,
"eval_wer": 1.9666600555335185,
"step": 500
},
{
"epoch": 0.84,
"learning_rate": 9.960000000000001e-06,
"loss": 4.2325,
"step": 501
},
{
"epoch": 0.84,
"learning_rate": 9.980000000000001e-06,
"loss": 4.1381,
"step": 502
},
{
"epoch": 0.85,
"learning_rate": 1e-05,
"loss": 4.1025,
"step": 503
},
{
"epoch": 0.85,
"learning_rate": 9.893617021276596e-06,
"loss": 4.1918,
"step": 504
},
{
"epoch": 0.85,
"learning_rate": 9.787234042553192e-06,
"loss": 4.1578,
"step": 505
},
{
"epoch": 0.85,
"learning_rate": 9.680851063829787e-06,
"loss": 4.2885,
"step": 506
},
{
"epoch": 0.85,
"learning_rate": 9.574468085106385e-06,
"loss": 4.1569,
"step": 507
},
{
"epoch": 0.85,
"learning_rate": 9.46808510638298e-06,
"loss": 4.1487,
"step": 508
},
{
"epoch": 0.86,
"learning_rate": 9.361702127659576e-06,
"loss": 4.0881,
"step": 509
},
{
"epoch": 0.86,
"learning_rate": 9.255319148936171e-06,
"loss": 4.1581,
"step": 510
},
{
"epoch": 0.86,
"learning_rate": 9.148936170212767e-06,
"loss": 4.2344,
"step": 511
},
{
"epoch": 0.86,
"learning_rate": 9.042553191489362e-06,
"loss": 4.1285,
"step": 512
},
{
"epoch": 0.86,
"learning_rate": 8.936170212765958e-06,
"loss": 4.1244,
"step": 513
},
{
"epoch": 0.86,
"learning_rate": 8.829787234042555e-06,
"loss": 4.0797,
"step": 514
},
{
"epoch": 0.87,
"learning_rate": 8.72340425531915e-06,
"loss": 4.2064,
"step": 515
},
{
"epoch": 0.87,
"learning_rate": 8.617021276595746e-06,
"loss": 4.1293,
"step": 516
},
{
"epoch": 0.87,
"learning_rate": 8.510638297872341e-06,
"loss": 4.1691,
"step": 517
},
{
"epoch": 0.87,
"learning_rate": 8.404255319148937e-06,
"loss": 4.23,
"step": 518
},
{
"epoch": 0.87,
"learning_rate": 8.297872340425532e-06,
"loss": 4.045,
"step": 519
},
{
"epoch": 0.87,
"learning_rate": 8.191489361702128e-06,
"loss": 4.1642,
"step": 520
},
{
"epoch": 0.88,
"learning_rate": 8.085106382978723e-06,
"loss": 4.1907,
"step": 521
},
{
"epoch": 0.88,
"learning_rate": 7.97872340425532e-06,
"loss": 4.2534,
"step": 522
},
{
"epoch": 0.88,
"learning_rate": 7.872340425531916e-06,
"loss": 4.0643,
"step": 523
},
{
"epoch": 0.88,
"learning_rate": 7.765957446808511e-06,
"loss": 3.9933,
"step": 524
},
{
"epoch": 0.88,
"learning_rate": 7.659574468085107e-06,
"loss": 4.0862,
"step": 525
},
{
"epoch": 0.88,
"learning_rate": 7.553191489361703e-06,
"loss": 4.0739,
"step": 526
},
{
"epoch": 0.89,
"learning_rate": 7.446808510638298e-06,
"loss": 4.2288,
"step": 527
},
{
"epoch": 0.89,
"learning_rate": 7.340425531914894e-06,
"loss": 4.1022,
"step": 528
},
{
"epoch": 0.89,
"learning_rate": 7.234042553191491e-06,
"loss": 4.0489,
"step": 529
},
{
"epoch": 0.89,
"learning_rate": 7.127659574468085e-06,
"loss": 4.0736,
"step": 530
},
{
"epoch": 0.89,
"learning_rate": 7.021276595744682e-06,
"loss": 4.0327,
"step": 531
},
{
"epoch": 0.89,
"learning_rate": 6.914893617021278e-06,
"loss": 4.2801,
"step": 532
},
{
"epoch": 0.9,
"learning_rate": 6.808510638297873e-06,
"loss": 4.0947,
"step": 533
},
{
"epoch": 0.9,
"learning_rate": 6.702127659574469e-06,
"loss": 4.1191,
"step": 534
},
{
"epoch": 0.9,
"learning_rate": 6.595744680851064e-06,
"loss": 4.1445,
"step": 535
},
{
"epoch": 0.9,
"learning_rate": 6.48936170212766e-06,
"loss": 3.9945,
"step": 536
},
{
"epoch": 0.9,
"learning_rate": 6.382978723404256e-06,
"loss": 4.2223,
"step": 537
},
{
"epoch": 0.9,
"learning_rate": 6.276595744680851e-06,
"loss": 4.2336,
"step": 538
},
{
"epoch": 0.91,
"learning_rate": 6.170212765957447e-06,
"loss": 4.0644,
"step": 539
},
{
"epoch": 0.91,
"learning_rate": 6.063829787234044e-06,
"loss": 4.1494,
"step": 540
},
{
"epoch": 0.91,
"learning_rate": 5.957446808510638e-06,
"loss": 4.1364,
"step": 541
},
{
"epoch": 0.91,
"learning_rate": 5.851063829787235e-06,
"loss": 4.1308,
"step": 542
},
{
"epoch": 0.91,
"learning_rate": 5.744680851063831e-06,
"loss": 4.3163,
"step": 543
},
{
"epoch": 0.91,
"learning_rate": 5.638297872340426e-06,
"loss": 4.156,
"step": 544
},
{
"epoch": 0.92,
"learning_rate": 5.531914893617022e-06,
"loss": 4.2632,
"step": 545
},
{
"epoch": 0.92,
"learning_rate": 5.425531914893617e-06,
"loss": 4.4659,
"step": 546
},
{
"epoch": 0.92,
"learning_rate": 5.319148936170213e-06,
"loss": 4.2235,
"step": 547
},
{
"epoch": 0.92,
"learning_rate": 5.212765957446809e-06,
"loss": 4.4768,
"step": 548
},
{
"epoch": 0.92,
"learning_rate": 5.106382978723404e-06,
"loss": 4.6745,
"step": 549
},
{
"epoch": 0.92,
"learning_rate": 5e-06,
"loss": 4.3527,
"step": 550
},
{
"epoch": 0.93,
"learning_rate": 4.893617021276596e-06,
"loss": 4.1747,
"step": 551
},
{
"epoch": 0.93,
"learning_rate": 4.787234042553192e-06,
"loss": 4.0934,
"step": 552
},
{
"epoch": 0.93,
"learning_rate": 4.680851063829788e-06,
"loss": 4.082,
"step": 553
},
{
"epoch": 0.93,
"learning_rate": 4.574468085106383e-06,
"loss": 4.0284,
"step": 554
},
{
"epoch": 0.93,
"learning_rate": 4.468085106382979e-06,
"loss": 4.0693,
"step": 555
},
{
"epoch": 0.93,
"learning_rate": 4.361702127659575e-06,
"loss": 4.1033,
"step": 556
},
{
"epoch": 0.94,
"learning_rate": 4.255319148936171e-06,
"loss": 4.1894,
"step": 557
},
{
"epoch": 0.94,
"learning_rate": 4.148936170212766e-06,
"loss": 4.2196,
"step": 558
},
{
"epoch": 0.94,
"learning_rate": 4.042553191489362e-06,
"loss": 4.0573,
"step": 559
},
{
"epoch": 0.94,
"learning_rate": 3.936170212765958e-06,
"loss": 4.098,
"step": 560
},
{
"epoch": 0.94,
"learning_rate": 3.8297872340425535e-06,
"loss": 4.4263,
"step": 561
},
{
"epoch": 0.94,
"learning_rate": 3.723404255319149e-06,
"loss": 4.0279,
"step": 562
},
{
"epoch": 0.95,
"learning_rate": 3.6170212765957453e-06,
"loss": 4.0855,
"step": 563
},
{
"epoch": 0.95,
"learning_rate": 3.510638297872341e-06,
"loss": 4.1792,
"step": 564
},
{
"epoch": 0.95,
"learning_rate": 3.4042553191489363e-06,
"loss": 4.1505,
"step": 565
},
{
"epoch": 0.95,
"learning_rate": 3.297872340425532e-06,
"loss": 4.0332,
"step": 566
},
{
"epoch": 0.95,
"learning_rate": 3.191489361702128e-06,
"loss": 4.1336,
"step": 567
},
{
"epoch": 0.96,
"learning_rate": 3.0851063829787237e-06,
"loss": 4.0822,
"step": 568
},
{
"epoch": 0.96,
"learning_rate": 2.978723404255319e-06,
"loss": 3.9525,
"step": 569
},
{
"epoch": 0.96,
"learning_rate": 2.8723404255319155e-06,
"loss": 4.1345,
"step": 570
},
{
"epoch": 0.96,
"learning_rate": 2.765957446808511e-06,
"loss": 4.1153,
"step": 571
},
{
"epoch": 0.96,
"learning_rate": 2.6595744680851065e-06,
"loss": 4.1605,
"step": 572
},
{
"epoch": 0.96,
"learning_rate": 2.553191489361702e-06,
"loss": 4.2209,
"step": 573
},
{
"epoch": 0.97,
"learning_rate": 2.446808510638298e-06,
"loss": 4.1386,
"step": 574
},
{
"epoch": 0.97,
"learning_rate": 2.340425531914894e-06,
"loss": 4.2482,
"step": 575
},
{
"epoch": 0.97,
"learning_rate": 2.2340425531914894e-06,
"loss": 4.1276,
"step": 576
},
{
"epoch": 0.97,
"learning_rate": 2.1276595744680853e-06,
"loss": 4.1114,
"step": 577
},
{
"epoch": 0.97,
"learning_rate": 2.021276595744681e-06,
"loss": 4.0383,
"step": 578
},
{
"epoch": 0.97,
"learning_rate": 1.9148936170212767e-06,
"loss": 3.9893,
"step": 579
},
{
"epoch": 0.98,
"learning_rate": 1.8085106382978727e-06,
"loss": 4.2624,
"step": 580
},
{
"epoch": 0.98,
"learning_rate": 1.7021276595744682e-06,
"loss": 4.3654,
"step": 581
},
{
"epoch": 0.98,
"learning_rate": 1.595744680851064e-06,
"loss": 4.2641,
"step": 582
},
{
"epoch": 0.98,
"learning_rate": 1.4893617021276596e-06,
"loss": 4.0602,
"step": 583
},
{
"epoch": 0.98,
"learning_rate": 1.3829787234042555e-06,
"loss": 4.1982,
"step": 584
},
{
"epoch": 0.98,
"learning_rate": 1.276595744680851e-06,
"loss": 4.1675,
"step": 585
},
{
"epoch": 0.99,
"learning_rate": 1.170212765957447e-06,
"loss": 4.063,
"step": 586
},
{
"epoch": 0.99,
"learning_rate": 1.0638297872340427e-06,
"loss": 4.2433,
"step": 587
},
{
"epoch": 0.99,
"learning_rate": 9.574468085106384e-07,
"loss": 4.2181,
"step": 588
},
{
"epoch": 0.99,
"learning_rate": 8.510638297872341e-07,
"loss": 4.3395,
"step": 589
},
{
"epoch": 0.99,
"learning_rate": 7.446808510638298e-07,
"loss": 4.1872,
"step": 590
},
{
"epoch": 0.99,
"learning_rate": 6.382978723404255e-07,
"loss": 4.399,
"step": 591
},
{
"epoch": 1.0,
"learning_rate": 5.319148936170213e-07,
"loss": 4.2918,
"step": 592
},
{
"epoch": 1.0,
"learning_rate": 4.2553191489361704e-07,
"loss": 4.4872,
"step": 593
},
{
"epoch": 1.0,
"learning_rate": 3.1914893617021275e-07,
"loss": 4.4735,
"step": 594
},
{
"epoch": 1.0,
"step": 594,
"total_flos": 0.0,
"train_loss": 4.333276915630508,
"train_runtime": 5595.6802,
"train_samples_per_second": 5.1,
"train_steps_per_second": 0.106
}
],
"max_steps": 594,
"num_train_epochs": 1,
"total_flos": 0.0,
"trial_name": null,
"trial_params": null
}