wav2vec2-gpt2-wandb-grid-search / trainer_state.json
sanchit-gandhi's picture
End of training
a9f1502
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": 4.8042,
"step": 1
},
{
"epoch": 0.0,
"learning_rate": 2e-08,
"loss": 4.9828,
"step": 2
},
{
"epoch": 0.01,
"learning_rate": 4e-08,
"loss": 4.8735,
"step": 3
},
{
"epoch": 0.01,
"learning_rate": 6.000000000000001e-08,
"loss": 4.8238,
"step": 4
},
{
"epoch": 0.01,
"learning_rate": 8e-08,
"loss": 4.7747,
"step": 5
},
{
"epoch": 0.01,
"learning_rate": 1.0000000000000001e-07,
"loss": 4.7918,
"step": 6
},
{
"epoch": 0.01,
"learning_rate": 1.2000000000000002e-07,
"loss": 4.8798,
"step": 7
},
{
"epoch": 0.01,
"learning_rate": 1.4e-07,
"loss": 4.7786,
"step": 8
},
{
"epoch": 0.02,
"learning_rate": 1.6e-07,
"loss": 4.9028,
"step": 9
},
{
"epoch": 0.02,
"learning_rate": 1.8e-07,
"loss": 4.8233,
"step": 10
},
{
"epoch": 0.02,
"learning_rate": 2.0000000000000002e-07,
"loss": 4.8176,
"step": 11
},
{
"epoch": 0.02,
"learning_rate": 2.2e-07,
"loss": 4.8817,
"step": 12
},
{
"epoch": 0.02,
"learning_rate": 2.4000000000000003e-07,
"loss": 4.8089,
"step": 13
},
{
"epoch": 0.02,
"learning_rate": 2.6e-07,
"loss": 4.7724,
"step": 14
},
{
"epoch": 0.03,
"learning_rate": 2.8e-07,
"loss": 4.662,
"step": 15
},
{
"epoch": 0.03,
"learning_rate": 3.0000000000000004e-07,
"loss": 4.8532,
"step": 16
},
{
"epoch": 0.03,
"learning_rate": 3.2e-07,
"loss": 4.8698,
"step": 17
},
{
"epoch": 0.03,
"learning_rate": 3.4000000000000003e-07,
"loss": 4.9494,
"step": 18
},
{
"epoch": 0.03,
"learning_rate": 3.6e-07,
"loss": 4.7211,
"step": 19
},
{
"epoch": 0.03,
"learning_rate": 3.8e-07,
"loss": 4.9558,
"step": 20
},
{
"epoch": 0.04,
"learning_rate": 4.0000000000000003e-07,
"loss": 4.9076,
"step": 21
},
{
"epoch": 0.04,
"learning_rate": 4.2000000000000006e-07,
"loss": 4.8373,
"step": 22
},
{
"epoch": 0.04,
"learning_rate": 4.4e-07,
"loss": 4.8605,
"step": 23
},
{
"epoch": 0.04,
"learning_rate": 4.6000000000000004e-07,
"loss": 4.9537,
"step": 24
},
{
"epoch": 0.04,
"learning_rate": 4.800000000000001e-07,
"loss": 4.7198,
"step": 25
},
{
"epoch": 0.04,
"learning_rate": 5.000000000000001e-07,
"loss": 5.1327,
"step": 26
},
{
"epoch": 0.05,
"learning_rate": 5.2e-07,
"loss": 4.8064,
"step": 27
},
{
"epoch": 0.05,
"learning_rate": 5.4e-07,
"loss": 4.7093,
"step": 28
},
{
"epoch": 0.05,
"learning_rate": 5.4e-07,
"loss": 4.7345,
"step": 29
},
{
"epoch": 0.05,
"learning_rate": 5.6e-07,
"loss": 4.8727,
"step": 30
},
{
"epoch": 0.05,
"learning_rate": 5.800000000000001e-07,
"loss": 4.8892,
"step": 31
},
{
"epoch": 0.05,
"learning_rate": 6.000000000000001e-07,
"loss": 4.7728,
"step": 32
},
{
"epoch": 0.06,
"learning_rate": 6.200000000000001e-07,
"loss": 4.78,
"step": 33
},
{
"epoch": 0.06,
"learning_rate": 6.4e-07,
"loss": 4.6903,
"step": 34
},
{
"epoch": 0.06,
"learning_rate": 6.6e-07,
"loss": 4.7677,
"step": 35
},
{
"epoch": 0.06,
"learning_rate": 6.800000000000001e-07,
"loss": 4.9281,
"step": 36
},
{
"epoch": 0.06,
"learning_rate": 7.000000000000001e-07,
"loss": 4.7253,
"step": 37
},
{
"epoch": 0.06,
"learning_rate": 7.2e-07,
"loss": 4.8761,
"step": 38
},
{
"epoch": 0.07,
"learning_rate": 7.4e-07,
"loss": 5.0624,
"step": 39
},
{
"epoch": 0.07,
"learning_rate": 7.6e-07,
"loss": 4.6726,
"step": 40
},
{
"epoch": 0.07,
"learning_rate": 7.8e-07,
"loss": 4.7354,
"step": 41
},
{
"epoch": 0.07,
"learning_rate": 8.000000000000001e-07,
"loss": 4.987,
"step": 42
},
{
"epoch": 0.07,
"learning_rate": 8.200000000000001e-07,
"loss": 4.739,
"step": 43
},
{
"epoch": 0.07,
"learning_rate": 8.400000000000001e-07,
"loss": 4.8959,
"step": 44
},
{
"epoch": 0.08,
"learning_rate": 8.6e-07,
"loss": 5.0197,
"step": 45
},
{
"epoch": 0.08,
"learning_rate": 8.8e-07,
"loss": 5.1426,
"step": 46
},
{
"epoch": 0.08,
"learning_rate": 9.000000000000001e-07,
"loss": 5.0508,
"step": 47
},
{
"epoch": 0.08,
"learning_rate": 9.200000000000001e-07,
"loss": 5.4743,
"step": 48
},
{
"epoch": 0.08,
"learning_rate": 9.400000000000001e-07,
"loss": 5.5167,
"step": 49
},
{
"epoch": 0.08,
"learning_rate": 9.400000000000001e-07,
"loss": 5.3789,
"step": 50
},
{
"epoch": 0.09,
"learning_rate": 9.600000000000001e-07,
"loss": 4.5449,
"step": 51
},
{
"epoch": 0.09,
"learning_rate": 9.800000000000001e-07,
"loss": 4.5448,
"step": 52
},
{
"epoch": 0.09,
"learning_rate": 1.0000000000000002e-06,
"loss": 4.609,
"step": 53
},
{
"epoch": 0.09,
"learning_rate": 1.02e-06,
"loss": 4.6805,
"step": 54
},
{
"epoch": 0.09,
"learning_rate": 1.04e-06,
"loss": 4.5197,
"step": 55
},
{
"epoch": 0.09,
"learning_rate": 1.06e-06,
"loss": 4.7056,
"step": 56
},
{
"epoch": 0.1,
"learning_rate": 1.08e-06,
"loss": 4.5713,
"step": 57
},
{
"epoch": 0.1,
"learning_rate": 1.1e-06,
"loss": 4.8102,
"step": 58
},
{
"epoch": 0.1,
"learning_rate": 1.12e-06,
"loss": 4.6042,
"step": 59
},
{
"epoch": 0.1,
"learning_rate": 1.14e-06,
"loss": 4.3496,
"step": 60
},
{
"epoch": 0.1,
"learning_rate": 1.1600000000000001e-06,
"loss": 4.6199,
"step": 61
},
{
"epoch": 0.1,
"learning_rate": 1.1800000000000001e-06,
"loss": 4.675,
"step": 62
},
{
"epoch": 0.11,
"learning_rate": 1.2000000000000002e-06,
"loss": 4.5535,
"step": 63
},
{
"epoch": 0.11,
"learning_rate": 1.2200000000000002e-06,
"loss": 4.4975,
"step": 64
},
{
"epoch": 0.11,
"learning_rate": 1.2400000000000002e-06,
"loss": 4.5554,
"step": 65
},
{
"epoch": 0.11,
"learning_rate": 1.26e-06,
"loss": 4.4926,
"step": 66
},
{
"epoch": 0.11,
"learning_rate": 1.28e-06,
"loss": 4.4971,
"step": 67
},
{
"epoch": 0.11,
"learning_rate": 1.3e-06,
"loss": 4.4337,
"step": 68
},
{
"epoch": 0.12,
"learning_rate": 1.32e-06,
"loss": 4.5496,
"step": 69
},
{
"epoch": 0.12,
"learning_rate": 1.34e-06,
"loss": 4.5592,
"step": 70
},
{
"epoch": 0.12,
"learning_rate": 1.3600000000000001e-06,
"loss": 4.6406,
"step": 71
},
{
"epoch": 0.12,
"learning_rate": 1.3800000000000001e-06,
"loss": 4.5955,
"step": 72
},
{
"epoch": 0.12,
"learning_rate": 1.4000000000000001e-06,
"loss": 4.3985,
"step": 73
},
{
"epoch": 0.12,
"learning_rate": 1.42e-06,
"loss": 4.582,
"step": 74
},
{
"epoch": 0.13,
"learning_rate": 1.44e-06,
"loss": 4.5618,
"step": 75
},
{
"epoch": 0.13,
"learning_rate": 1.46e-06,
"loss": 4.561,
"step": 76
},
{
"epoch": 0.13,
"learning_rate": 1.48e-06,
"loss": 4.5283,
"step": 77
},
{
"epoch": 0.13,
"learning_rate": 1.5e-06,
"loss": 4.5355,
"step": 78
},
{
"epoch": 0.13,
"learning_rate": 1.52e-06,
"loss": 4.412,
"step": 79
},
{
"epoch": 0.13,
"learning_rate": 1.54e-06,
"loss": 4.5652,
"step": 80
},
{
"epoch": 0.14,
"learning_rate": 1.56e-06,
"loss": 4.5097,
"step": 81
},
{
"epoch": 0.14,
"learning_rate": 1.5800000000000001e-06,
"loss": 4.5247,
"step": 82
},
{
"epoch": 0.14,
"learning_rate": 1.6000000000000001e-06,
"loss": 4.5395,
"step": 83
},
{
"epoch": 0.14,
"learning_rate": 1.6200000000000002e-06,
"loss": 4.5561,
"step": 84
},
{
"epoch": 0.14,
"learning_rate": 1.6400000000000002e-06,
"loss": 4.4865,
"step": 85
},
{
"epoch": 0.14,
"learning_rate": 1.6600000000000002e-06,
"loss": 4.3575,
"step": 86
},
{
"epoch": 0.15,
"learning_rate": 1.6800000000000002e-06,
"loss": 4.6712,
"step": 87
},
{
"epoch": 0.15,
"learning_rate": 1.7000000000000002e-06,
"loss": 4.5597,
"step": 88
},
{
"epoch": 0.15,
"learning_rate": 1.72e-06,
"loss": 4.5239,
"step": 89
},
{
"epoch": 0.15,
"learning_rate": 1.74e-06,
"loss": 4.568,
"step": 90
},
{
"epoch": 0.15,
"learning_rate": 1.76e-06,
"loss": 4.5041,
"step": 91
},
{
"epoch": 0.15,
"learning_rate": 1.7800000000000001e-06,
"loss": 4.6208,
"step": 92
},
{
"epoch": 0.16,
"learning_rate": 1.8000000000000001e-06,
"loss": 4.4879,
"step": 93
},
{
"epoch": 0.16,
"learning_rate": 1.8200000000000002e-06,
"loss": 4.7273,
"step": 94
},
{
"epoch": 0.16,
"learning_rate": 1.8400000000000002e-06,
"loss": 4.6565,
"step": 95
},
{
"epoch": 0.16,
"learning_rate": 1.8600000000000002e-06,
"loss": 4.694,
"step": 96
},
{
"epoch": 0.16,
"learning_rate": 1.8800000000000002e-06,
"loss": 4.7673,
"step": 97
},
{
"epoch": 0.16,
"learning_rate": 1.9000000000000002e-06,
"loss": 4.5242,
"step": 98
},
{
"epoch": 0.17,
"learning_rate": 1.9200000000000003e-06,
"loss": 4.7109,
"step": 99
},
{
"epoch": 0.17,
"learning_rate": 1.94e-06,
"loss": 4.4853,
"step": 100
},
{
"epoch": 0.17,
"learning_rate": 1.9600000000000003e-06,
"loss": 4.5437,
"step": 101
},
{
"epoch": 0.17,
"learning_rate": 1.98e-06,
"loss": 4.3413,
"step": 102
},
{
"epoch": 0.17,
"learning_rate": 2.0000000000000003e-06,
"loss": 4.3155,
"step": 103
},
{
"epoch": 0.17,
"learning_rate": 2.02e-06,
"loss": 4.2955,
"step": 104
},
{
"epoch": 0.18,
"learning_rate": 2.04e-06,
"loss": 4.2886,
"step": 105
},
{
"epoch": 0.18,
"learning_rate": 2.06e-06,
"loss": 4.325,
"step": 106
},
{
"epoch": 0.18,
"learning_rate": 2.08e-06,
"loss": 4.4709,
"step": 107
},
{
"epoch": 0.18,
"learning_rate": 2.1000000000000002e-06,
"loss": 4.4975,
"step": 108
},
{
"epoch": 0.18,
"learning_rate": 2.12e-06,
"loss": 4.5738,
"step": 109
},
{
"epoch": 0.18,
"learning_rate": 2.1400000000000003e-06,
"loss": 4.3586,
"step": 110
},
{
"epoch": 0.19,
"learning_rate": 2.16e-06,
"loss": 4.3516,
"step": 111
},
{
"epoch": 0.19,
"learning_rate": 2.1800000000000003e-06,
"loss": 4.4207,
"step": 112
},
{
"epoch": 0.19,
"learning_rate": 2.2e-06,
"loss": 4.4198,
"step": 113
},
{
"epoch": 0.19,
"learning_rate": 2.2200000000000003e-06,
"loss": 4.441,
"step": 114
},
{
"epoch": 0.19,
"learning_rate": 2.24e-06,
"loss": 4.3794,
"step": 115
},
{
"epoch": 0.2,
"learning_rate": 2.2600000000000004e-06,
"loss": 4.3902,
"step": 116
},
{
"epoch": 0.2,
"learning_rate": 2.28e-06,
"loss": 4.4838,
"step": 117
},
{
"epoch": 0.2,
"learning_rate": 2.3000000000000004e-06,
"loss": 4.2504,
"step": 118
},
{
"epoch": 0.2,
"learning_rate": 2.3200000000000002e-06,
"loss": 4.228,
"step": 119
},
{
"epoch": 0.2,
"learning_rate": 2.3400000000000005e-06,
"loss": 4.2372,
"step": 120
},
{
"epoch": 0.2,
"learning_rate": 2.3600000000000003e-06,
"loss": 4.4673,
"step": 121
},
{
"epoch": 0.21,
"learning_rate": 2.38e-06,
"loss": 4.553,
"step": 122
},
{
"epoch": 0.21,
"learning_rate": 2.4000000000000003e-06,
"loss": 4.315,
"step": 123
},
{
"epoch": 0.21,
"learning_rate": 2.42e-06,
"loss": 4.3808,
"step": 124
},
{
"epoch": 0.21,
"learning_rate": 2.4400000000000004e-06,
"loss": 4.3912,
"step": 125
},
{
"epoch": 0.21,
"learning_rate": 2.46e-06,
"loss": 4.2075,
"step": 126
},
{
"epoch": 0.21,
"learning_rate": 2.4800000000000004e-06,
"loss": 4.266,
"step": 127
},
{
"epoch": 0.22,
"learning_rate": 2.5e-06,
"loss": 4.2219,
"step": 128
},
{
"epoch": 0.22,
"learning_rate": 2.52e-06,
"loss": 4.3792,
"step": 129
},
{
"epoch": 0.22,
"learning_rate": 2.5400000000000002e-06,
"loss": 4.4051,
"step": 130
},
{
"epoch": 0.22,
"learning_rate": 2.56e-06,
"loss": 4.3441,
"step": 131
},
{
"epoch": 0.22,
"learning_rate": 2.5800000000000003e-06,
"loss": 4.2955,
"step": 132
},
{
"epoch": 0.22,
"learning_rate": 2.6e-06,
"loss": 4.3969,
"step": 133
},
{
"epoch": 0.23,
"learning_rate": 2.6200000000000003e-06,
"loss": 4.402,
"step": 134
},
{
"epoch": 0.23,
"learning_rate": 2.64e-06,
"loss": 4.4528,
"step": 135
},
{
"epoch": 0.23,
"learning_rate": 2.6600000000000004e-06,
"loss": 4.3123,
"step": 136
},
{
"epoch": 0.23,
"learning_rate": 2.68e-06,
"loss": 4.3315,
"step": 137
},
{
"epoch": 0.23,
"learning_rate": 2.7000000000000004e-06,
"loss": 4.329,
"step": 138
},
{
"epoch": 0.23,
"learning_rate": 2.7200000000000002e-06,
"loss": 4.3091,
"step": 139
},
{
"epoch": 0.24,
"learning_rate": 2.7400000000000004e-06,
"loss": 4.5347,
"step": 140
},
{
"epoch": 0.24,
"learning_rate": 2.7600000000000003e-06,
"loss": 4.453,
"step": 141
},
{
"epoch": 0.24,
"learning_rate": 2.7800000000000005e-06,
"loss": 4.4934,
"step": 142
},
{
"epoch": 0.24,
"learning_rate": 2.8000000000000003e-06,
"loss": 4.5347,
"step": 143
},
{
"epoch": 0.24,
"learning_rate": 2.82e-06,
"loss": 4.3281,
"step": 144
},
{
"epoch": 0.24,
"learning_rate": 2.84e-06,
"loss": 4.4223,
"step": 145
},
{
"epoch": 0.25,
"learning_rate": 2.86e-06,
"loss": 4.7379,
"step": 146
},
{
"epoch": 0.25,
"learning_rate": 2.88e-06,
"loss": 4.7251,
"step": 147
},
{
"epoch": 0.25,
"learning_rate": 2.9e-06,
"loss": 4.7122,
"step": 148
},
{
"epoch": 0.25,
"learning_rate": 2.92e-06,
"loss": 4.6427,
"step": 149
},
{
"epoch": 0.25,
"learning_rate": 2.9400000000000002e-06,
"loss": 4.7168,
"step": 150
},
{
"epoch": 0.25,
"learning_rate": 2.96e-06,
"loss": 4.3009,
"step": 151
},
{
"epoch": 0.26,
"learning_rate": 2.9800000000000003e-06,
"loss": 4.2353,
"step": 152
},
{
"epoch": 0.26,
"learning_rate": 3e-06,
"loss": 4.2273,
"step": 153
},
{
"epoch": 0.26,
"learning_rate": 3.0200000000000003e-06,
"loss": 4.1507,
"step": 154
},
{
"epoch": 0.26,
"learning_rate": 3.04e-06,
"loss": 4.1878,
"step": 155
},
{
"epoch": 0.26,
"learning_rate": 3.0600000000000003e-06,
"loss": 4.1888,
"step": 156
},
{
"epoch": 0.26,
"learning_rate": 3.08e-06,
"loss": 4.3586,
"step": 157
},
{
"epoch": 0.27,
"learning_rate": 3.1000000000000004e-06,
"loss": 4.1796,
"step": 158
},
{
"epoch": 0.27,
"learning_rate": 3.12e-06,
"loss": 4.1439,
"step": 159
},
{
"epoch": 0.27,
"learning_rate": 3.1400000000000004e-06,
"loss": 4.2612,
"step": 160
},
{
"epoch": 0.27,
"learning_rate": 3.1600000000000002e-06,
"loss": 4.3082,
"step": 161
},
{
"epoch": 0.27,
"learning_rate": 3.1800000000000005e-06,
"loss": 4.1563,
"step": 162
},
{
"epoch": 0.27,
"learning_rate": 3.2000000000000003e-06,
"loss": 4.2179,
"step": 163
},
{
"epoch": 0.28,
"learning_rate": 3.2200000000000005e-06,
"loss": 4.2514,
"step": 164
},
{
"epoch": 0.28,
"learning_rate": 3.2400000000000003e-06,
"loss": 4.2873,
"step": 165
},
{
"epoch": 0.28,
"learning_rate": 3.2600000000000006e-06,
"loss": 4.3925,
"step": 166
},
{
"epoch": 0.28,
"learning_rate": 3.2800000000000004e-06,
"loss": 4.2303,
"step": 167
},
{
"epoch": 0.28,
"learning_rate": 3.3000000000000006e-06,
"loss": 4.4668,
"step": 168
},
{
"epoch": 0.28,
"learning_rate": 3.3200000000000004e-06,
"loss": 4.2762,
"step": 169
},
{
"epoch": 0.29,
"learning_rate": 3.3400000000000006e-06,
"loss": 4.3279,
"step": 170
},
{
"epoch": 0.29,
"learning_rate": 3.3600000000000004e-06,
"loss": 4.1924,
"step": 171
},
{
"epoch": 0.29,
"learning_rate": 3.3800000000000007e-06,
"loss": 4.1893,
"step": 172
},
{
"epoch": 0.29,
"learning_rate": 3.4000000000000005e-06,
"loss": 4.2197,
"step": 173
},
{
"epoch": 0.29,
"learning_rate": 3.4200000000000007e-06,
"loss": 4.1982,
"step": 174
},
{
"epoch": 0.29,
"learning_rate": 3.44e-06,
"loss": 4.3419,
"step": 175
},
{
"epoch": 0.3,
"learning_rate": 3.46e-06,
"loss": 4.3705,
"step": 176
},
{
"epoch": 0.3,
"learning_rate": 3.48e-06,
"loss": 4.3055,
"step": 177
},
{
"epoch": 0.3,
"learning_rate": 3.5e-06,
"loss": 4.1993,
"step": 178
},
{
"epoch": 0.3,
"learning_rate": 3.52e-06,
"loss": 4.3358,
"step": 179
},
{
"epoch": 0.3,
"learning_rate": 3.54e-06,
"loss": 4.1873,
"step": 180
},
{
"epoch": 0.3,
"learning_rate": 3.5600000000000002e-06,
"loss": 4.2644,
"step": 181
},
{
"epoch": 0.31,
"learning_rate": 3.58e-06,
"loss": 4.2642,
"step": 182
},
{
"epoch": 0.31,
"learning_rate": 3.6000000000000003e-06,
"loss": 4.2311,
"step": 183
},
{
"epoch": 0.31,
"learning_rate": 3.62e-06,
"loss": 4.1317,
"step": 184
},
{
"epoch": 0.31,
"learning_rate": 3.6400000000000003e-06,
"loss": 4.4816,
"step": 185
},
{
"epoch": 0.31,
"learning_rate": 3.66e-06,
"loss": 4.268,
"step": 186
},
{
"epoch": 0.31,
"learning_rate": 3.6800000000000003e-06,
"loss": 4.3489,
"step": 187
},
{
"epoch": 0.32,
"learning_rate": 3.7e-06,
"loss": 4.3619,
"step": 188
},
{
"epoch": 0.32,
"learning_rate": 3.7200000000000004e-06,
"loss": 4.5345,
"step": 189
},
{
"epoch": 0.32,
"learning_rate": 3.74e-06,
"loss": 4.2482,
"step": 190
},
{
"epoch": 0.32,
"learning_rate": 3.7600000000000004e-06,
"loss": 4.5152,
"step": 191
},
{
"epoch": 0.32,
"learning_rate": 3.7800000000000002e-06,
"loss": 4.4431,
"step": 192
},
{
"epoch": 0.32,
"learning_rate": 3.8000000000000005e-06,
"loss": 4.3379,
"step": 193
},
{
"epoch": 0.33,
"learning_rate": 3.820000000000001e-06,
"loss": 4.4033,
"step": 194
},
{
"epoch": 0.33,
"learning_rate": 3.8400000000000005e-06,
"loss": 4.3606,
"step": 195
},
{
"epoch": 0.33,
"learning_rate": 3.86e-06,
"loss": 4.2561,
"step": 196
},
{
"epoch": 0.33,
"learning_rate": 3.88e-06,
"loss": 4.6605,
"step": 197
},
{
"epoch": 0.33,
"learning_rate": 3.900000000000001e-06,
"loss": 4.6509,
"step": 198
},
{
"epoch": 0.33,
"learning_rate": 3.920000000000001e-06,
"loss": 4.5575,
"step": 199
},
{
"epoch": 0.34,
"learning_rate": 3.94e-06,
"loss": 4.7293,
"step": 200
},
{
"epoch": 0.34,
"learning_rate": 3.96e-06,
"loss": 4.119,
"step": 201
},
{
"epoch": 0.34,
"learning_rate": 3.980000000000001e-06,
"loss": 4.2329,
"step": 202
},
{
"epoch": 0.34,
"learning_rate": 4.000000000000001e-06,
"loss": 4.2809,
"step": 203
},
{
"epoch": 0.34,
"learning_rate": 4.0200000000000005e-06,
"loss": 4.1888,
"step": 204
},
{
"epoch": 0.34,
"learning_rate": 4.04e-06,
"loss": 4.2201,
"step": 205
},
{
"epoch": 0.35,
"learning_rate": 4.060000000000001e-06,
"loss": 4.1575,
"step": 206
},
{
"epoch": 0.35,
"learning_rate": 4.08e-06,
"loss": 4.2356,
"step": 207
},
{
"epoch": 0.35,
"learning_rate": 4.1e-06,
"loss": 4.1999,
"step": 208
},
{
"epoch": 0.35,
"learning_rate": 4.12e-06,
"loss": 4.2149,
"step": 209
},
{
"epoch": 0.35,
"learning_rate": 4.14e-06,
"loss": 4.3318,
"step": 210
},
{
"epoch": 0.35,
"learning_rate": 4.16e-06,
"loss": 4.299,
"step": 211
},
{
"epoch": 0.36,
"learning_rate": 4.18e-06,
"loss": 4.2981,
"step": 212
},
{
"epoch": 0.36,
"learning_rate": 4.2000000000000004e-06,
"loss": 4.2578,
"step": 213
},
{
"epoch": 0.36,
"learning_rate": 4.22e-06,
"loss": 4.353,
"step": 214
},
{
"epoch": 0.36,
"learning_rate": 4.24e-06,
"loss": 4.2134,
"step": 215
},
{
"epoch": 0.36,
"learning_rate": 4.26e-06,
"loss": 4.24,
"step": 216
},
{
"epoch": 0.36,
"learning_rate": 4.2800000000000005e-06,
"loss": 4.1222,
"step": 217
},
{
"epoch": 0.37,
"learning_rate": 4.3e-06,
"loss": 4.2403,
"step": 218
},
{
"epoch": 0.37,
"learning_rate": 4.32e-06,
"loss": 4.2177,
"step": 219
},
{
"epoch": 0.37,
"learning_rate": 4.34e-06,
"loss": 4.3008,
"step": 220
},
{
"epoch": 0.37,
"learning_rate": 4.360000000000001e-06,
"loss": 4.3402,
"step": 221
},
{
"epoch": 0.37,
"learning_rate": 4.38e-06,
"loss": 4.3275,
"step": 222
},
{
"epoch": 0.37,
"learning_rate": 4.4e-06,
"loss": 4.1779,
"step": 223
},
{
"epoch": 0.38,
"learning_rate": 4.42e-06,
"loss": 4.2863,
"step": 224
},
{
"epoch": 0.38,
"learning_rate": 4.440000000000001e-06,
"loss": 4.4608,
"step": 225
},
{
"epoch": 0.38,
"learning_rate": 4.4600000000000005e-06,
"loss": 4.2403,
"step": 226
},
{
"epoch": 0.38,
"learning_rate": 4.48e-06,
"loss": 4.1853,
"step": 227
},
{
"epoch": 0.38,
"learning_rate": 4.5e-06,
"loss": 4.2403,
"step": 228
},
{
"epoch": 0.39,
"learning_rate": 4.520000000000001e-06,
"loss": 4.3736,
"step": 229
},
{
"epoch": 0.39,
"learning_rate": 4.540000000000001e-06,
"loss": 4.3332,
"step": 230
},
{
"epoch": 0.39,
"learning_rate": 4.56e-06,
"loss": 4.2064,
"step": 231
},
{
"epoch": 0.39,
"learning_rate": 4.58e-06,
"loss": 4.2761,
"step": 232
},
{
"epoch": 0.39,
"learning_rate": 4.600000000000001e-06,
"loss": 4.3082,
"step": 233
},
{
"epoch": 0.39,
"learning_rate": 4.620000000000001e-06,
"loss": 4.2818,
"step": 234
},
{
"epoch": 0.4,
"learning_rate": 4.6400000000000005e-06,
"loss": 4.4403,
"step": 235
},
{
"epoch": 0.4,
"learning_rate": 4.66e-06,
"loss": 4.2043,
"step": 236
},
{
"epoch": 0.4,
"learning_rate": 4.680000000000001e-06,
"loss": 4.2007,
"step": 237
},
{
"epoch": 0.4,
"learning_rate": 4.7e-06,
"loss": 4.2285,
"step": 238
},
{
"epoch": 0.4,
"learning_rate": 4.7200000000000005e-06,
"loss": 4.1442,
"step": 239
},
{
"epoch": 0.4,
"learning_rate": 4.74e-06,
"loss": 4.3297,
"step": 240
},
{
"epoch": 0.41,
"learning_rate": 4.76e-06,
"loss": 4.253,
"step": 241
},
{
"epoch": 0.41,
"learning_rate": 4.78e-06,
"loss": 4.2696,
"step": 242
},
{
"epoch": 0.41,
"learning_rate": 4.800000000000001e-06,
"loss": 4.2858,
"step": 243
},
{
"epoch": 0.41,
"learning_rate": 4.8200000000000004e-06,
"loss": 4.3687,
"step": 244
},
{
"epoch": 0.41,
"learning_rate": 4.84e-06,
"loss": 4.4333,
"step": 245
},
{
"epoch": 0.41,
"learning_rate": 4.86e-06,
"loss": 4.476,
"step": 246
},
{
"epoch": 0.42,
"learning_rate": 4.880000000000001e-06,
"loss": 4.4581,
"step": 247
},
{
"epoch": 0.42,
"learning_rate": 4.9000000000000005e-06,
"loss": 4.5935,
"step": 248
},
{
"epoch": 0.42,
"learning_rate": 4.92e-06,
"loss": 4.5562,
"step": 249
},
{
"epoch": 0.42,
"learning_rate": 4.94e-06,
"loss": 4.5708,
"step": 250
},
{
"epoch": 0.42,
"learning_rate": 4.960000000000001e-06,
"loss": 4.2715,
"step": 251
},
{
"epoch": 0.42,
"learning_rate": 4.980000000000001e-06,
"loss": 4.2646,
"step": 252
},
{
"epoch": 0.43,
"learning_rate": 5e-06,
"loss": 4.2305,
"step": 253
},
{
"epoch": 0.43,
"learning_rate": 5.02e-06,
"loss": 4.1202,
"step": 254
},
{
"epoch": 0.43,
"learning_rate": 5.04e-06,
"loss": 4.2932,
"step": 255
},
{
"epoch": 0.43,
"learning_rate": 5.060000000000001e-06,
"loss": 4.2446,
"step": 256
},
{
"epoch": 0.43,
"learning_rate": 5.0800000000000005e-06,
"loss": 4.1774,
"step": 257
},
{
"epoch": 0.43,
"learning_rate": 5.1e-06,
"loss": 4.203,
"step": 258
},
{
"epoch": 0.44,
"learning_rate": 5.12e-06,
"loss": 4.2629,
"step": 259
},
{
"epoch": 0.44,
"learning_rate": 5.140000000000001e-06,
"loss": 4.2674,
"step": 260
},
{
"epoch": 0.44,
"learning_rate": 5.1600000000000006e-06,
"loss": 4.2434,
"step": 261
},
{
"epoch": 0.44,
"learning_rate": 5.18e-06,
"loss": 4.1899,
"step": 262
},
{
"epoch": 0.44,
"learning_rate": 5.2e-06,
"loss": 4.2941,
"step": 263
},
{
"epoch": 0.44,
"learning_rate": 5.220000000000001e-06,
"loss": 4.0175,
"step": 264
},
{
"epoch": 0.45,
"learning_rate": 5.240000000000001e-06,
"loss": 4.1044,
"step": 265
},
{
"epoch": 0.45,
"learning_rate": 5.2600000000000005e-06,
"loss": 4.2199,
"step": 266
},
{
"epoch": 0.45,
"learning_rate": 5.28e-06,
"loss": 4.2392,
"step": 267
},
{
"epoch": 0.45,
"learning_rate": 5.300000000000001e-06,
"loss": 4.1779,
"step": 268
},
{
"epoch": 0.45,
"learning_rate": 5.320000000000001e-06,
"loss": 4.2911,
"step": 269
},
{
"epoch": 0.45,
"learning_rate": 5.3400000000000005e-06,
"loss": 4.2804,
"step": 270
},
{
"epoch": 0.46,
"learning_rate": 5.36e-06,
"loss": 4.2093,
"step": 271
},
{
"epoch": 0.46,
"learning_rate": 5.380000000000001e-06,
"loss": 4.3397,
"step": 272
},
{
"epoch": 0.46,
"learning_rate": 5.400000000000001e-06,
"loss": 4.2085,
"step": 273
},
{
"epoch": 0.46,
"learning_rate": 5.420000000000001e-06,
"loss": 4.2158,
"step": 274
},
{
"epoch": 0.46,
"learning_rate": 5.4400000000000004e-06,
"loss": 4.2876,
"step": 275
},
{
"epoch": 0.46,
"learning_rate": 5.460000000000001e-06,
"loss": 4.2118,
"step": 276
},
{
"epoch": 0.47,
"learning_rate": 5.480000000000001e-06,
"loss": 4.3481,
"step": 277
},
{
"epoch": 0.47,
"learning_rate": 5.500000000000001e-06,
"loss": 4.1363,
"step": 278
},
{
"epoch": 0.47,
"learning_rate": 5.5200000000000005e-06,
"loss": 4.3209,
"step": 279
},
{
"epoch": 0.47,
"learning_rate": 5.540000000000001e-06,
"loss": 4.3781,
"step": 280
},
{
"epoch": 0.47,
"learning_rate": 5.560000000000001e-06,
"loss": 4.3857,
"step": 281
},
{
"epoch": 0.47,
"learning_rate": 5.580000000000001e-06,
"loss": 4.2594,
"step": 282
},
{
"epoch": 0.48,
"learning_rate": 5.600000000000001e-06,
"loss": 4.161,
"step": 283
},
{
"epoch": 0.48,
"learning_rate": 5.620000000000001e-06,
"loss": 4.1539,
"step": 284
},
{
"epoch": 0.48,
"learning_rate": 5.64e-06,
"loss": 4.4102,
"step": 285
},
{
"epoch": 0.48,
"learning_rate": 5.66e-06,
"loss": 4.166,
"step": 286
},
{
"epoch": 0.48,
"learning_rate": 5.68e-06,
"loss": 4.5148,
"step": 287
},
{
"epoch": 0.48,
"learning_rate": 5.7e-06,
"loss": 4.0003,
"step": 288
},
{
"epoch": 0.49,
"learning_rate": 5.72e-06,
"loss": 4.2587,
"step": 289
},
{
"epoch": 0.49,
"learning_rate": 5.74e-06,
"loss": 4.2268,
"step": 290
},
{
"epoch": 0.49,
"learning_rate": 5.76e-06,
"loss": 4.4759,
"step": 291
},
{
"epoch": 0.49,
"learning_rate": 5.78e-06,
"loss": 4.1496,
"step": 292
},
{
"epoch": 0.49,
"learning_rate": 5.8e-06,
"loss": 4.3539,
"step": 293
},
{
"epoch": 0.49,
"learning_rate": 5.82e-06,
"loss": 4.3933,
"step": 294
},
{
"epoch": 0.5,
"learning_rate": 5.84e-06,
"loss": 4.3932,
"step": 295
},
{
"epoch": 0.5,
"learning_rate": 5.86e-06,
"loss": 4.2008,
"step": 296
},
{
"epoch": 0.5,
"learning_rate": 5.8800000000000005e-06,
"loss": 4.2785,
"step": 297
},
{
"epoch": 0.5,
"learning_rate": 5.9e-06,
"loss": 4.4644,
"step": 298
},
{
"epoch": 0.5,
"learning_rate": 5.92e-06,
"loss": 4.4931,
"step": 299
},
{
"epoch": 0.5,
"learning_rate": 5.94e-06,
"loss": 4.7331,
"step": 300
},
{
"epoch": 0.51,
"learning_rate": 5.9600000000000005e-06,
"loss": 4.3267,
"step": 301
},
{
"epoch": 0.51,
"learning_rate": 5.98e-06,
"loss": 4.2696,
"step": 302
},
{
"epoch": 0.51,
"learning_rate": 6e-06,
"loss": 4.3485,
"step": 303
},
{
"epoch": 0.51,
"learning_rate": 6.02e-06,
"loss": 4.1293,
"step": 304
},
{
"epoch": 0.51,
"learning_rate": 6.040000000000001e-06,
"loss": 4.3256,
"step": 305
},
{
"epoch": 0.51,
"learning_rate": 6.0600000000000004e-06,
"loss": 4.313,
"step": 306
},
{
"epoch": 0.52,
"learning_rate": 6.08e-06,
"loss": 4.1031,
"step": 307
},
{
"epoch": 0.52,
"learning_rate": 6.1e-06,
"loss": 4.3191,
"step": 308
},
{
"epoch": 0.52,
"learning_rate": 6.120000000000001e-06,
"loss": 4.2111,
"step": 309
},
{
"epoch": 0.52,
"learning_rate": 6.1400000000000005e-06,
"loss": 4.3018,
"step": 310
},
{
"epoch": 0.52,
"learning_rate": 6.16e-06,
"loss": 4.1658,
"step": 311
},
{
"epoch": 0.52,
"learning_rate": 6.18e-06,
"loss": 4.2064,
"step": 312
},
{
"epoch": 0.53,
"learning_rate": 6.200000000000001e-06,
"loss": 4.2251,
"step": 313
},
{
"epoch": 0.53,
"learning_rate": 6.220000000000001e-06,
"loss": 4.2262,
"step": 314
},
{
"epoch": 0.53,
"learning_rate": 6.24e-06,
"loss": 4.2471,
"step": 315
},
{
"epoch": 0.53,
"learning_rate": 6.26e-06,
"loss": 4.2381,
"step": 316
},
{
"epoch": 0.53,
"learning_rate": 6.280000000000001e-06,
"loss": 4.1972,
"step": 317
},
{
"epoch": 0.53,
"learning_rate": 6.300000000000001e-06,
"loss": 4.2808,
"step": 318
},
{
"epoch": 0.54,
"learning_rate": 6.3200000000000005e-06,
"loss": 4.1035,
"step": 319
},
{
"epoch": 0.54,
"learning_rate": 6.34e-06,
"loss": 4.265,
"step": 320
},
{
"epoch": 0.54,
"learning_rate": 6.360000000000001e-06,
"loss": 4.1205,
"step": 321
},
{
"epoch": 0.54,
"learning_rate": 6.380000000000001e-06,
"loss": 4.1692,
"step": 322
},
{
"epoch": 0.54,
"learning_rate": 6.4000000000000006e-06,
"loss": 4.1871,
"step": 323
},
{
"epoch": 0.54,
"learning_rate": 6.42e-06,
"loss": 4.2543,
"step": 324
},
{
"epoch": 0.55,
"learning_rate": 6.440000000000001e-06,
"loss": 4.2072,
"step": 325
},
{
"epoch": 0.55,
"learning_rate": 6.460000000000001e-06,
"loss": 4.1326,
"step": 326
},
{
"epoch": 0.55,
"learning_rate": 6.480000000000001e-06,
"loss": 4.4151,
"step": 327
},
{
"epoch": 0.55,
"learning_rate": 6.5000000000000004e-06,
"loss": 4.1615,
"step": 328
},
{
"epoch": 0.55,
"learning_rate": 6.520000000000001e-06,
"loss": 4.0782,
"step": 329
},
{
"epoch": 0.55,
"learning_rate": 6.540000000000001e-06,
"loss": 4.3004,
"step": 330
},
{
"epoch": 0.56,
"learning_rate": 6.560000000000001e-06,
"loss": 4.1869,
"step": 331
},
{
"epoch": 0.56,
"learning_rate": 6.5800000000000005e-06,
"loss": 4.137,
"step": 332
},
{
"epoch": 0.56,
"learning_rate": 6.600000000000001e-06,
"loss": 4.1984,
"step": 333
},
{
"epoch": 0.56,
"learning_rate": 6.620000000000001e-06,
"loss": 4.2389,
"step": 334
},
{
"epoch": 0.56,
"learning_rate": 6.640000000000001e-06,
"loss": 4.2559,
"step": 335
},
{
"epoch": 0.56,
"learning_rate": 6.660000000000001e-06,
"loss": 4.1806,
"step": 336
},
{
"epoch": 0.57,
"learning_rate": 6.680000000000001e-06,
"loss": 4.1488,
"step": 337
},
{
"epoch": 0.57,
"learning_rate": 6.700000000000001e-06,
"loss": 4.2468,
"step": 338
},
{
"epoch": 0.57,
"learning_rate": 6.720000000000001e-06,
"loss": 4.2594,
"step": 339
},
{
"epoch": 0.57,
"learning_rate": 6.740000000000001e-06,
"loss": 4.287,
"step": 340
},
{
"epoch": 0.57,
"learning_rate": 6.760000000000001e-06,
"loss": 4.2912,
"step": 341
},
{
"epoch": 0.58,
"learning_rate": 6.780000000000001e-06,
"loss": 4.3932,
"step": 342
},
{
"epoch": 0.58,
"learning_rate": 6.800000000000001e-06,
"loss": 4.418,
"step": 343
},
{
"epoch": 0.58,
"learning_rate": 6.820000000000001e-06,
"loss": 4.2976,
"step": 344
},
{
"epoch": 0.58,
"learning_rate": 6.8400000000000014e-06,
"loss": 4.4477,
"step": 345
},
{
"epoch": 0.58,
"learning_rate": 6.860000000000001e-06,
"loss": 4.1469,
"step": 346
},
{
"epoch": 0.58,
"learning_rate": 6.88e-06,
"loss": 4.2839,
"step": 347
},
{
"epoch": 0.59,
"learning_rate": 6.9e-06,
"loss": 4.6134,
"step": 348
},
{
"epoch": 0.59,
"learning_rate": 6.92e-06,
"loss": 4.4709,
"step": 349
},
{
"epoch": 0.59,
"learning_rate": 6.9400000000000005e-06,
"loss": 4.7728,
"step": 350
},
{
"epoch": 0.59,
"learning_rate": 6.96e-06,
"loss": 4.1291,
"step": 351
},
{
"epoch": 0.59,
"learning_rate": 6.98e-06,
"loss": 4.237,
"step": 352
},
{
"epoch": 0.59,
"learning_rate": 7e-06,
"loss": 4.2614,
"step": 353
},
{
"epoch": 0.6,
"learning_rate": 7.0200000000000006e-06,
"loss": 4.1032,
"step": 354
},
{
"epoch": 0.6,
"learning_rate": 7.04e-06,
"loss": 4.1884,
"step": 355
},
{
"epoch": 0.6,
"learning_rate": 7.06e-06,
"loss": 4.2002,
"step": 356
},
{
"epoch": 0.6,
"learning_rate": 7.08e-06,
"loss": 4.2122,
"step": 357
},
{
"epoch": 0.6,
"learning_rate": 7.100000000000001e-06,
"loss": 4.1464,
"step": 358
},
{
"epoch": 0.6,
"learning_rate": 7.1200000000000004e-06,
"loss": 4.1748,
"step": 359
},
{
"epoch": 0.61,
"learning_rate": 7.14e-06,
"loss": 4.1205,
"step": 360
},
{
"epoch": 0.61,
"learning_rate": 7.16e-06,
"loss": 4.2075,
"step": 361
},
{
"epoch": 0.61,
"learning_rate": 7.180000000000001e-06,
"loss": 4.2108,
"step": 362
},
{
"epoch": 0.61,
"learning_rate": 7.2000000000000005e-06,
"loss": 4.2577,
"step": 363
},
{
"epoch": 0.61,
"learning_rate": 7.22e-06,
"loss": 4.1437,
"step": 364
},
{
"epoch": 0.61,
"learning_rate": 7.24e-06,
"loss": 4.2128,
"step": 365
},
{
"epoch": 0.62,
"learning_rate": 7.260000000000001e-06,
"loss": 4.1773,
"step": 366
},
{
"epoch": 0.62,
"learning_rate": 7.280000000000001e-06,
"loss": 4.179,
"step": 367
},
{
"epoch": 0.62,
"learning_rate": 7.3e-06,
"loss": 4.164,
"step": 368
},
{
"epoch": 0.62,
"learning_rate": 7.32e-06,
"loss": 4.2615,
"step": 369
},
{
"epoch": 0.62,
"learning_rate": 7.340000000000001e-06,
"loss": 4.1449,
"step": 370
},
{
"epoch": 0.62,
"learning_rate": 7.360000000000001e-06,
"loss": 4.0527,
"step": 371
},
{
"epoch": 0.63,
"learning_rate": 7.3800000000000005e-06,
"loss": 4.2189,
"step": 372
},
{
"epoch": 0.63,
"learning_rate": 7.4e-06,
"loss": 4.0695,
"step": 373
},
{
"epoch": 0.63,
"learning_rate": 7.420000000000001e-06,
"loss": 4.0884,
"step": 374
},
{
"epoch": 0.63,
"learning_rate": 7.440000000000001e-06,
"loss": 4.2513,
"step": 375
},
{
"epoch": 0.63,
"learning_rate": 7.4600000000000006e-06,
"loss": 4.1336,
"step": 376
},
{
"epoch": 0.63,
"learning_rate": 7.48e-06,
"loss": 4.2286,
"step": 377
},
{
"epoch": 0.64,
"learning_rate": 7.500000000000001e-06,
"loss": 4.2841,
"step": 378
},
{
"epoch": 0.64,
"learning_rate": 7.520000000000001e-06,
"loss": 4.1134,
"step": 379
},
{
"epoch": 0.64,
"learning_rate": 7.540000000000001e-06,
"loss": 4.337,
"step": 380
},
{
"epoch": 0.64,
"learning_rate": 7.5600000000000005e-06,
"loss": 4.2637,
"step": 381
},
{
"epoch": 0.64,
"learning_rate": 7.58e-06,
"loss": 4.1276,
"step": 382
},
{
"epoch": 0.64,
"learning_rate": 7.600000000000001e-06,
"loss": 4.0296,
"step": 383
},
{
"epoch": 0.65,
"learning_rate": 7.620000000000001e-06,
"loss": 4.2207,
"step": 384
},
{
"epoch": 0.65,
"learning_rate": 7.640000000000001e-06,
"loss": 4.4024,
"step": 385
},
{
"epoch": 0.65,
"learning_rate": 7.660000000000001e-06,
"loss": 4.1624,
"step": 386
},
{
"epoch": 0.65,
"learning_rate": 7.680000000000001e-06,
"loss": 4.3254,
"step": 387
},
{
"epoch": 0.65,
"learning_rate": 7.7e-06,
"loss": 4.2542,
"step": 388
},
{
"epoch": 0.65,
"learning_rate": 7.72e-06,
"loss": 4.1433,
"step": 389
},
{
"epoch": 0.66,
"learning_rate": 7.74e-06,
"loss": 4.1481,
"step": 390
},
{
"epoch": 0.66,
"learning_rate": 7.76e-06,
"loss": 4.3634,
"step": 391
},
{
"epoch": 0.66,
"learning_rate": 7.78e-06,
"loss": 4.2915,
"step": 392
},
{
"epoch": 0.66,
"learning_rate": 7.800000000000002e-06,
"loss": 4.5435,
"step": 393
},
{
"epoch": 0.66,
"learning_rate": 7.820000000000001e-06,
"loss": 4.3662,
"step": 394
},
{
"epoch": 0.66,
"learning_rate": 7.840000000000001e-06,
"loss": 4.2978,
"step": 395
},
{
"epoch": 0.67,
"learning_rate": 7.860000000000001e-06,
"loss": 4.2279,
"step": 396
},
{
"epoch": 0.67,
"learning_rate": 7.88e-06,
"loss": 4.4285,
"step": 397
},
{
"epoch": 0.67,
"learning_rate": 7.9e-06,
"loss": 4.6224,
"step": 398
},
{
"epoch": 0.67,
"learning_rate": 7.92e-06,
"loss": 4.4336,
"step": 399
},
{
"epoch": 0.67,
"learning_rate": 7.94e-06,
"loss": 4.5344,
"step": 400
},
{
"epoch": 0.67,
"learning_rate": 7.960000000000002e-06,
"loss": 4.2421,
"step": 401
},
{
"epoch": 0.68,
"learning_rate": 7.980000000000002e-06,
"loss": 4.1492,
"step": 402
},
{
"epoch": 0.68,
"learning_rate": 8.000000000000001e-06,
"loss": 4.211,
"step": 403
},
{
"epoch": 0.68,
"learning_rate": 8.020000000000001e-06,
"loss": 4.1921,
"step": 404
},
{
"epoch": 0.68,
"learning_rate": 8.040000000000001e-06,
"loss": 4.1333,
"step": 405
},
{
"epoch": 0.68,
"learning_rate": 8.06e-06,
"loss": 4.1795,
"step": 406
},
{
"epoch": 0.68,
"learning_rate": 8.08e-06,
"loss": 4.1875,
"step": 407
},
{
"epoch": 0.69,
"learning_rate": 8.1e-06,
"loss": 4.1004,
"step": 408
},
{
"epoch": 0.69,
"learning_rate": 8.120000000000002e-06,
"loss": 4.1633,
"step": 409
},
{
"epoch": 0.69,
"learning_rate": 8.14e-06,
"loss": 4.1125,
"step": 410
},
{
"epoch": 0.69,
"learning_rate": 8.16e-06,
"loss": 4.1233,
"step": 411
},
{
"epoch": 0.69,
"learning_rate": 8.18e-06,
"loss": 4.2896,
"step": 412
},
{
"epoch": 0.69,
"learning_rate": 8.2e-06,
"loss": 4.1791,
"step": 413
},
{
"epoch": 0.7,
"learning_rate": 8.220000000000001e-06,
"loss": 4.2289,
"step": 414
},
{
"epoch": 0.7,
"learning_rate": 8.24e-06,
"loss": 4.2061,
"step": 415
},
{
"epoch": 0.7,
"learning_rate": 8.26e-06,
"loss": 4.1171,
"step": 416
},
{
"epoch": 0.7,
"learning_rate": 8.28e-06,
"loss": 4.1973,
"step": 417
},
{
"epoch": 0.7,
"learning_rate": 8.3e-06,
"loss": 4.0318,
"step": 418
},
{
"epoch": 0.7,
"learning_rate": 8.32e-06,
"loss": 4.2931,
"step": 419
},
{
"epoch": 0.71,
"learning_rate": 8.34e-06,
"loss": 4.1999,
"step": 420
},
{
"epoch": 0.71,
"learning_rate": 8.36e-06,
"loss": 3.9695,
"step": 421
},
{
"epoch": 0.71,
"learning_rate": 8.380000000000001e-06,
"loss": 4.1124,
"step": 422
},
{
"epoch": 0.71,
"learning_rate": 8.400000000000001e-06,
"loss": 4.2425,
"step": 423
},
{
"epoch": 0.71,
"learning_rate": 8.42e-06,
"loss": 4.1725,
"step": 424
},
{
"epoch": 0.71,
"learning_rate": 8.44e-06,
"loss": 4.166,
"step": 425
},
{
"epoch": 0.72,
"learning_rate": 8.46e-06,
"loss": 4.2335,
"step": 426
},
{
"epoch": 0.72,
"learning_rate": 8.48e-06,
"loss": 4.1832,
"step": 427
},
{
"epoch": 0.72,
"learning_rate": 8.5e-06,
"loss": 4.1219,
"step": 428
},
{
"epoch": 0.72,
"learning_rate": 8.52e-06,
"loss": 4.237,
"step": 429
},
{
"epoch": 0.72,
"learning_rate": 8.540000000000001e-06,
"loss": 4.1834,
"step": 430
},
{
"epoch": 0.72,
"learning_rate": 8.560000000000001e-06,
"loss": 4.2492,
"step": 431
},
{
"epoch": 0.73,
"learning_rate": 8.580000000000001e-06,
"loss": 4.0503,
"step": 432
},
{
"epoch": 0.73,
"learning_rate": 8.6e-06,
"loss": 4.2888,
"step": 433
},
{
"epoch": 0.73,
"learning_rate": 8.62e-06,
"loss": 4.2068,
"step": 434
},
{
"epoch": 0.73,
"learning_rate": 8.64e-06,
"loss": 4.2792,
"step": 435
},
{
"epoch": 0.73,
"learning_rate": 8.66e-06,
"loss": 4.1248,
"step": 436
},
{
"epoch": 0.73,
"learning_rate": 8.68e-06,
"loss": 4.0674,
"step": 437
},
{
"epoch": 0.74,
"learning_rate": 8.700000000000001e-06,
"loss": 4.2741,
"step": 438
},
{
"epoch": 0.74,
"learning_rate": 8.720000000000001e-06,
"loss": 4.1833,
"step": 439
},
{
"epoch": 0.74,
"learning_rate": 8.740000000000001e-06,
"loss": 4.3683,
"step": 440
},
{
"epoch": 0.74,
"learning_rate": 8.76e-06,
"loss": 4.2469,
"step": 441
},
{
"epoch": 0.74,
"learning_rate": 8.78e-06,
"loss": 4.4203,
"step": 442
},
{
"epoch": 0.74,
"learning_rate": 8.8e-06,
"loss": 4.3338,
"step": 443
},
{
"epoch": 0.75,
"learning_rate": 8.82e-06,
"loss": 4.2684,
"step": 444
},
{
"epoch": 0.75,
"learning_rate": 8.84e-06,
"loss": 4.0929,
"step": 445
},
{
"epoch": 0.75,
"learning_rate": 8.860000000000002e-06,
"loss": 4.3938,
"step": 446
},
{
"epoch": 0.75,
"learning_rate": 8.880000000000001e-06,
"loss": 4.3174,
"step": 447
},
{
"epoch": 0.75,
"learning_rate": 8.900000000000001e-06,
"loss": 4.2561,
"step": 448
},
{
"epoch": 0.75,
"learning_rate": 8.920000000000001e-06,
"loss": 4.444,
"step": 449
},
{
"epoch": 0.76,
"learning_rate": 8.94e-06,
"loss": 4.5666,
"step": 450
},
{
"epoch": 0.76,
"learning_rate": 8.96e-06,
"loss": 4.2167,
"step": 451
},
{
"epoch": 0.76,
"learning_rate": 8.98e-06,
"loss": 4.2266,
"step": 452
},
{
"epoch": 0.76,
"learning_rate": 9e-06,
"loss": 4.1614,
"step": 453
},
{
"epoch": 0.76,
"learning_rate": 9.020000000000002e-06,
"loss": 4.0323,
"step": 454
},
{
"epoch": 0.77,
"learning_rate": 9.040000000000002e-06,
"loss": 4.214,
"step": 455
},
{
"epoch": 0.77,
"learning_rate": 9.060000000000001e-06,
"loss": 4.152,
"step": 456
},
{
"epoch": 0.77,
"learning_rate": 9.080000000000001e-06,
"loss": 4.2426,
"step": 457
},
{
"epoch": 0.77,
"learning_rate": 9.100000000000001e-06,
"loss": 4.2681,
"step": 458
},
{
"epoch": 0.77,
"learning_rate": 9.12e-06,
"loss": 4.1891,
"step": 459
},
{
"epoch": 0.77,
"learning_rate": 9.14e-06,
"loss": 4.2,
"step": 460
},
{
"epoch": 0.78,
"learning_rate": 9.16e-06,
"loss": 4.1147,
"step": 461
},
{
"epoch": 0.78,
"learning_rate": 9.180000000000002e-06,
"loss": 4.1357,
"step": 462
},
{
"epoch": 0.78,
"learning_rate": 9.200000000000002e-06,
"loss": 4.1446,
"step": 463
},
{
"epoch": 0.78,
"learning_rate": 9.220000000000002e-06,
"loss": 4.2087,
"step": 464
},
{
"epoch": 0.78,
"learning_rate": 9.240000000000001e-06,
"loss": 4.2335,
"step": 465
},
{
"epoch": 0.78,
"learning_rate": 9.260000000000001e-06,
"loss": 4.1698,
"step": 466
},
{
"epoch": 0.79,
"learning_rate": 9.280000000000001e-06,
"loss": 4.1785,
"step": 467
},
{
"epoch": 0.79,
"learning_rate": 9.3e-06,
"loss": 4.1921,
"step": 468
},
{
"epoch": 0.79,
"learning_rate": 9.32e-06,
"loss": 4.1838,
"step": 469
},
{
"epoch": 0.79,
"learning_rate": 9.340000000000002e-06,
"loss": 4.1536,
"step": 470
},
{
"epoch": 0.79,
"learning_rate": 9.360000000000002e-06,
"loss": 4.081,
"step": 471
},
{
"epoch": 0.79,
"learning_rate": 9.38e-06,
"loss": 4.1874,
"step": 472
},
{
"epoch": 0.8,
"learning_rate": 9.4e-06,
"loss": 4.1796,
"step": 473
},
{
"epoch": 0.8,
"learning_rate": 9.42e-06,
"loss": 4.0716,
"step": 474
},
{
"epoch": 0.8,
"learning_rate": 9.440000000000001e-06,
"loss": 4.1163,
"step": 475
},
{
"epoch": 0.8,
"learning_rate": 9.460000000000001e-06,
"loss": 4.1431,
"step": 476
},
{
"epoch": 0.8,
"learning_rate": 9.48e-06,
"loss": 4.1631,
"step": 477
},
{
"epoch": 0.8,
"learning_rate": 9.5e-06,
"loss": 4.1923,
"step": 478
},
{
"epoch": 0.81,
"learning_rate": 9.52e-06,
"loss": 4.0699,
"step": 479
},
{
"epoch": 0.81,
"learning_rate": 9.54e-06,
"loss": 4.0712,
"step": 480
},
{
"epoch": 0.81,
"learning_rate": 9.56e-06,
"loss": 4.3093,
"step": 481
},
{
"epoch": 0.81,
"learning_rate": 9.58e-06,
"loss": 4.1374,
"step": 482
},
{
"epoch": 0.81,
"learning_rate": 9.600000000000001e-06,
"loss": 4.3051,
"step": 483
},
{
"epoch": 0.81,
"learning_rate": 9.620000000000001e-06,
"loss": 4.2504,
"step": 484
},
{
"epoch": 0.82,
"learning_rate": 9.640000000000001e-06,
"loss": 4.2339,
"step": 485
},
{
"epoch": 0.82,
"learning_rate": 9.66e-06,
"loss": 4.1411,
"step": 486
},
{
"epoch": 0.82,
"learning_rate": 9.68e-06,
"loss": 4.276,
"step": 487
},
{
"epoch": 0.82,
"learning_rate": 9.7e-06,
"loss": 4.2906,
"step": 488
},
{
"epoch": 0.82,
"learning_rate": 9.72e-06,
"loss": 4.216,
"step": 489
},
{
"epoch": 0.82,
"learning_rate": 9.74e-06,
"loss": 4.1313,
"step": 490
},
{
"epoch": 0.83,
"learning_rate": 9.760000000000001e-06,
"loss": 4.3359,
"step": 491
},
{
"epoch": 0.83,
"learning_rate": 9.780000000000001e-06,
"loss": 4.1545,
"step": 492
},
{
"epoch": 0.83,
"learning_rate": 9.800000000000001e-06,
"loss": 4.036,
"step": 493
},
{
"epoch": 0.83,
"learning_rate": 9.820000000000001e-06,
"loss": 4.2527,
"step": 494
},
{
"epoch": 0.83,
"learning_rate": 9.84e-06,
"loss": 4.3611,
"step": 495
},
{
"epoch": 0.83,
"learning_rate": 9.86e-06,
"loss": 4.3937,
"step": 496
},
{
"epoch": 0.84,
"learning_rate": 9.88e-06,
"loss": 4.2911,
"step": 497
},
{
"epoch": 0.84,
"learning_rate": 9.9e-06,
"loss": 4.4337,
"step": 498
},
{
"epoch": 0.84,
"learning_rate": 9.920000000000002e-06,
"loss": 4.4938,
"step": 499
},
{
"epoch": 0.84,
"learning_rate": 9.940000000000001e-06,
"loss": 4.4108,
"step": 500
},
{
"epoch": 0.84,
"eval_loss": 4.200125694274902,
"eval_runtime": 786.8447,
"eval_samples_per_second": 3.358,
"eval_steps_per_second": 0.281,
"eval_wer": 1.9598968663228877,
"step": 500
},
{
"epoch": 0.84,
"learning_rate": 9.960000000000001e-06,
"loss": 4.2338,
"step": 501
},
{
"epoch": 0.84,
"learning_rate": 9.980000000000001e-06,
"loss": 4.1417,
"step": 502
},
{
"epoch": 0.85,
"learning_rate": 1e-05,
"loss": 4.1053,
"step": 503
},
{
"epoch": 0.85,
"learning_rate": 9.893617021276596e-06,
"loss": 4.1975,
"step": 504
},
{
"epoch": 0.85,
"learning_rate": 9.787234042553192e-06,
"loss": 4.1617,
"step": 505
},
{
"epoch": 0.85,
"learning_rate": 9.680851063829787e-06,
"loss": 4.2979,
"step": 506
},
{
"epoch": 0.85,
"learning_rate": 9.574468085106385e-06,
"loss": 4.1643,
"step": 507
},
{
"epoch": 0.85,
"learning_rate": 9.46808510638298e-06,
"loss": 4.158,
"step": 508
},
{
"epoch": 0.86,
"learning_rate": 9.361702127659576e-06,
"loss": 4.0941,
"step": 509
},
{
"epoch": 0.86,
"learning_rate": 9.255319148936171e-06,
"loss": 4.17,
"step": 510
},
{
"epoch": 0.86,
"learning_rate": 9.148936170212767e-06,
"loss": 4.2423,
"step": 511
},
{
"epoch": 0.86,
"learning_rate": 9.042553191489362e-06,
"loss": 4.1385,
"step": 512
},
{
"epoch": 0.86,
"learning_rate": 8.936170212765958e-06,
"loss": 4.1332,
"step": 513
},
{
"epoch": 0.86,
"learning_rate": 8.829787234042555e-06,
"loss": 4.0833,
"step": 514
},
{
"epoch": 0.87,
"learning_rate": 8.72340425531915e-06,
"loss": 4.2107,
"step": 515
},
{
"epoch": 0.87,
"learning_rate": 8.617021276595746e-06,
"loss": 4.1385,
"step": 516
},
{
"epoch": 0.87,
"learning_rate": 8.510638297872341e-06,
"loss": 4.1791,
"step": 517
},
{
"epoch": 0.87,
"learning_rate": 8.404255319148937e-06,
"loss": 4.2385,
"step": 518
},
{
"epoch": 0.87,
"learning_rate": 8.297872340425532e-06,
"loss": 4.0488,
"step": 519
},
{
"epoch": 0.87,
"learning_rate": 8.191489361702128e-06,
"loss": 4.1701,
"step": 520
},
{
"epoch": 0.88,
"learning_rate": 8.085106382978723e-06,
"loss": 4.1984,
"step": 521
},
{
"epoch": 0.88,
"learning_rate": 7.97872340425532e-06,
"loss": 4.2571,
"step": 522
},
{
"epoch": 0.88,
"learning_rate": 7.872340425531916e-06,
"loss": 4.0696,
"step": 523
},
{
"epoch": 0.88,
"learning_rate": 7.765957446808511e-06,
"loss": 3.9969,
"step": 524
},
{
"epoch": 0.88,
"learning_rate": 7.659574468085107e-06,
"loss": 4.0892,
"step": 525
},
{
"epoch": 0.88,
"learning_rate": 7.553191489361703e-06,
"loss": 4.0812,
"step": 526
},
{
"epoch": 0.89,
"learning_rate": 7.446808510638298e-06,
"loss": 4.2357,
"step": 527
},
{
"epoch": 0.89,
"learning_rate": 7.340425531914894e-06,
"loss": 4.1053,
"step": 528
},
{
"epoch": 0.89,
"learning_rate": 7.234042553191491e-06,
"loss": 4.056,
"step": 529
},
{
"epoch": 0.89,
"learning_rate": 7.127659574468085e-06,
"loss": 4.0824,
"step": 530
},
{
"epoch": 0.89,
"learning_rate": 7.021276595744682e-06,
"loss": 4.0396,
"step": 531
},
{
"epoch": 0.89,
"learning_rate": 6.914893617021278e-06,
"loss": 4.2931,
"step": 532
},
{
"epoch": 0.9,
"learning_rate": 6.808510638297873e-06,
"loss": 4.1061,
"step": 533
},
{
"epoch": 0.9,
"learning_rate": 6.702127659574469e-06,
"loss": 4.124,
"step": 534
},
{
"epoch": 0.9,
"learning_rate": 6.595744680851064e-06,
"loss": 4.149,
"step": 535
},
{
"epoch": 0.9,
"learning_rate": 6.48936170212766e-06,
"loss": 4.0031,
"step": 536
},
{
"epoch": 0.9,
"learning_rate": 6.382978723404256e-06,
"loss": 4.2344,
"step": 537
},
{
"epoch": 0.9,
"learning_rate": 6.276595744680851e-06,
"loss": 4.2397,
"step": 538
},
{
"epoch": 0.91,
"learning_rate": 6.170212765957447e-06,
"loss": 4.0717,
"step": 539
},
{
"epoch": 0.91,
"learning_rate": 6.063829787234044e-06,
"loss": 4.156,
"step": 540
},
{
"epoch": 0.91,
"learning_rate": 5.957446808510638e-06,
"loss": 4.151,
"step": 541
},
{
"epoch": 0.91,
"learning_rate": 5.851063829787235e-06,
"loss": 4.1377,
"step": 542
},
{
"epoch": 0.91,
"learning_rate": 5.744680851063831e-06,
"loss": 4.3309,
"step": 543
},
{
"epoch": 0.91,
"learning_rate": 5.638297872340426e-06,
"loss": 4.161,
"step": 544
},
{
"epoch": 0.92,
"learning_rate": 5.531914893617022e-06,
"loss": 4.2825,
"step": 545
},
{
"epoch": 0.92,
"learning_rate": 5.425531914893617e-06,
"loss": 4.4817,
"step": 546
},
{
"epoch": 0.92,
"learning_rate": 5.319148936170213e-06,
"loss": 4.2491,
"step": 547
},
{
"epoch": 0.92,
"learning_rate": 5.212765957446809e-06,
"loss": 4.5033,
"step": 548
},
{
"epoch": 0.92,
"learning_rate": 5.106382978723404e-06,
"loss": 4.6914,
"step": 549
},
{
"epoch": 0.92,
"learning_rate": 5e-06,
"loss": 4.3848,
"step": 550
},
{
"epoch": 0.93,
"learning_rate": 4.893617021276596e-06,
"loss": 4.1878,
"step": 551
},
{
"epoch": 0.93,
"learning_rate": 4.787234042553192e-06,
"loss": 4.1017,
"step": 552
},
{
"epoch": 0.93,
"learning_rate": 4.680851063829788e-06,
"loss": 4.091,
"step": 553
},
{
"epoch": 0.93,
"learning_rate": 4.574468085106383e-06,
"loss": 4.0357,
"step": 554
},
{
"epoch": 0.93,
"learning_rate": 4.468085106382979e-06,
"loss": 4.0749,
"step": 555
},
{
"epoch": 0.93,
"learning_rate": 4.361702127659575e-06,
"loss": 4.106,
"step": 556
},
{
"epoch": 0.94,
"learning_rate": 4.255319148936171e-06,
"loss": 4.1974,
"step": 557
},
{
"epoch": 0.94,
"learning_rate": 4.148936170212766e-06,
"loss": 4.2221,
"step": 558
},
{
"epoch": 0.94,
"learning_rate": 4.042553191489362e-06,
"loss": 4.0609,
"step": 559
},
{
"epoch": 0.94,
"learning_rate": 3.936170212765958e-06,
"loss": 4.1074,
"step": 560
},
{
"epoch": 0.94,
"learning_rate": 3.8297872340425535e-06,
"loss": 4.4338,
"step": 561
},
{
"epoch": 0.94,
"learning_rate": 3.723404255319149e-06,
"loss": 4.041,
"step": 562
},
{
"epoch": 0.95,
"learning_rate": 3.6170212765957453e-06,
"loss": 4.0901,
"step": 563
},
{
"epoch": 0.95,
"learning_rate": 3.510638297872341e-06,
"loss": 4.1839,
"step": 564
},
{
"epoch": 0.95,
"learning_rate": 3.4042553191489363e-06,
"loss": 4.1574,
"step": 565
},
{
"epoch": 0.95,
"learning_rate": 3.297872340425532e-06,
"loss": 4.0428,
"step": 566
},
{
"epoch": 0.95,
"learning_rate": 3.191489361702128e-06,
"loss": 4.1384,
"step": 567
},
{
"epoch": 0.96,
"learning_rate": 3.0851063829787237e-06,
"loss": 4.0906,
"step": 568
},
{
"epoch": 0.96,
"learning_rate": 2.978723404255319e-06,
"loss": 3.9581,
"step": 569
},
{
"epoch": 0.96,
"learning_rate": 2.8723404255319155e-06,
"loss": 4.1344,
"step": 570
},
{
"epoch": 0.96,
"learning_rate": 2.765957446808511e-06,
"loss": 4.1249,
"step": 571
},
{
"epoch": 0.96,
"learning_rate": 2.6595744680851065e-06,
"loss": 4.1715,
"step": 572
},
{
"epoch": 0.96,
"learning_rate": 2.553191489361702e-06,
"loss": 4.2385,
"step": 573
},
{
"epoch": 0.97,
"learning_rate": 2.446808510638298e-06,
"loss": 4.1462,
"step": 574
},
{
"epoch": 0.97,
"learning_rate": 2.340425531914894e-06,
"loss": 4.258,
"step": 575
},
{
"epoch": 0.97,
"learning_rate": 2.2340425531914894e-06,
"loss": 4.1278,
"step": 576
},
{
"epoch": 0.97,
"learning_rate": 2.1276595744680853e-06,
"loss": 4.1169,
"step": 577
},
{
"epoch": 0.97,
"learning_rate": 2.021276595744681e-06,
"loss": 4.0495,
"step": 578
},
{
"epoch": 0.97,
"learning_rate": 1.9148936170212767e-06,
"loss": 4.004,
"step": 579
},
{
"epoch": 0.98,
"learning_rate": 1.8085106382978727e-06,
"loss": 4.2692,
"step": 580
},
{
"epoch": 0.98,
"learning_rate": 1.7021276595744682e-06,
"loss": 4.3664,
"step": 581
},
{
"epoch": 0.98,
"learning_rate": 1.595744680851064e-06,
"loss": 4.2793,
"step": 582
},
{
"epoch": 0.98,
"learning_rate": 1.4893617021276596e-06,
"loss": 4.0669,
"step": 583
},
{
"epoch": 0.98,
"learning_rate": 1.3829787234042555e-06,
"loss": 4.2035,
"step": 584
},
{
"epoch": 0.98,
"learning_rate": 1.276595744680851e-06,
"loss": 4.1755,
"step": 585
},
{
"epoch": 0.99,
"learning_rate": 1.170212765957447e-06,
"loss": 4.0676,
"step": 586
},
{
"epoch": 0.99,
"learning_rate": 1.0638297872340427e-06,
"loss": 4.2543,
"step": 587
},
{
"epoch": 0.99,
"learning_rate": 9.574468085106384e-07,
"loss": 4.2331,
"step": 588
},
{
"epoch": 0.99,
"learning_rate": 8.510638297872341e-07,
"loss": 4.3553,
"step": 589
},
{
"epoch": 0.99,
"learning_rate": 7.446808510638298e-07,
"loss": 4.2031,
"step": 590
},
{
"epoch": 0.99,
"learning_rate": 6.382978723404255e-07,
"loss": 4.4084,
"step": 591
},
{
"epoch": 1.0,
"learning_rate": 5.319148936170213e-07,
"loss": 4.2994,
"step": 592
},
{
"epoch": 1.0,
"learning_rate": 4.2553191489361704e-07,
"loss": 4.5066,
"step": 593
},
{
"epoch": 1.0,
"learning_rate": 3.1914893617021275e-07,
"loss": 4.4907,
"step": 594
},
{
"epoch": 1.0,
"step": 594,
"total_flos": 0.0,
"train_loss": 4.336673439953865,
"train_runtime": 5718.4377,
"train_samples_per_second": 4.991,
"train_steps_per_second": 0.104
}
],
"max_steps": 594,
"num_train_epochs": 1,
"total_flos": 0.0,
"trial_name": null,
"trial_params": null
}