wav2vec2-2-bert-grid-search / trainer_state.json
sanchit-gandhi's picture
End of training
623bc86
raw
history blame
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": 9.136,
"step": 1
},
{
"epoch": 0.0,
"learning_rate": 0.0,
"loss": 9.0353,
"step": 2
},
{
"epoch": 0.01,
"learning_rate": 0.0,
"loss": 8.8891,
"step": 3
},
{
"epoch": 0.01,
"learning_rate": 0.0,
"loss": 8.7961,
"step": 4
},
{
"epoch": 0.01,
"learning_rate": 2e-08,
"loss": 8.8132,
"step": 5
},
{
"epoch": 0.01,
"learning_rate": 4e-08,
"loss": 8.6436,
"step": 6
},
{
"epoch": 0.01,
"learning_rate": 6.000000000000001e-08,
"loss": 9.0931,
"step": 7
},
{
"epoch": 0.01,
"learning_rate": 8e-08,
"loss": 8.869,
"step": 8
},
{
"epoch": 0.02,
"learning_rate": 1.0000000000000001e-07,
"loss": 8.878,
"step": 9
},
{
"epoch": 0.02,
"learning_rate": 1.2000000000000002e-07,
"loss": 8.7189,
"step": 10
},
{
"epoch": 0.02,
"learning_rate": 1.4e-07,
"loss": 8.8518,
"step": 11
},
{
"epoch": 0.02,
"learning_rate": 1.6e-07,
"loss": 8.7859,
"step": 12
},
{
"epoch": 0.02,
"learning_rate": 1.8e-07,
"loss": 8.6734,
"step": 13
},
{
"epoch": 0.02,
"learning_rate": 2.0000000000000002e-07,
"loss": 8.723,
"step": 14
},
{
"epoch": 0.03,
"learning_rate": 2.2e-07,
"loss": 8.7489,
"step": 15
},
{
"epoch": 0.03,
"learning_rate": 2.4000000000000003e-07,
"loss": 8.7691,
"step": 16
},
{
"epoch": 0.03,
"learning_rate": 2.6e-07,
"loss": 8.6419,
"step": 17
},
{
"epoch": 0.03,
"learning_rate": 2.8e-07,
"loss": 8.5175,
"step": 18
},
{
"epoch": 0.03,
"learning_rate": 2.8e-07,
"loss": 8.4803,
"step": 19
},
{
"epoch": 0.03,
"learning_rate": 3.0000000000000004e-07,
"loss": 8.7215,
"step": 20
},
{
"epoch": 0.04,
"learning_rate": 3.2e-07,
"loss": 8.6841,
"step": 21
},
{
"epoch": 0.04,
"learning_rate": 3.4000000000000003e-07,
"loss": 8.5702,
"step": 22
},
{
"epoch": 0.04,
"learning_rate": 3.6e-07,
"loss": 8.7331,
"step": 23
},
{
"epoch": 0.04,
"learning_rate": 3.8e-07,
"loss": 8.4915,
"step": 24
},
{
"epoch": 0.04,
"learning_rate": 4.0000000000000003e-07,
"loss": 8.3233,
"step": 25
},
{
"epoch": 0.04,
"learning_rate": 4.2000000000000006e-07,
"loss": 8.5035,
"step": 26
},
{
"epoch": 0.05,
"learning_rate": 4.4e-07,
"loss": 8.4372,
"step": 27
},
{
"epoch": 0.05,
"learning_rate": 4.6000000000000004e-07,
"loss": 8.2875,
"step": 28
},
{
"epoch": 0.05,
"learning_rate": 4.800000000000001e-07,
"loss": 8.3858,
"step": 29
},
{
"epoch": 0.05,
"learning_rate": 5.000000000000001e-07,
"loss": 8.204,
"step": 30
},
{
"epoch": 0.05,
"learning_rate": 5.2e-07,
"loss": 8.3033,
"step": 31
},
{
"epoch": 0.05,
"learning_rate": 5.4e-07,
"loss": 8.2837,
"step": 32
},
{
"epoch": 0.06,
"learning_rate": 5.6e-07,
"loss": 8.0819,
"step": 33
},
{
"epoch": 0.06,
"learning_rate": 5.800000000000001e-07,
"loss": 8.105,
"step": 34
},
{
"epoch": 0.06,
"learning_rate": 6.000000000000001e-07,
"loss": 7.9917,
"step": 35
},
{
"epoch": 0.06,
"learning_rate": 6.200000000000001e-07,
"loss": 8.0203,
"step": 36
},
{
"epoch": 0.06,
"learning_rate": 6.4e-07,
"loss": 8.0964,
"step": 37
},
{
"epoch": 0.06,
"learning_rate": 6.6e-07,
"loss": 8.1488,
"step": 38
},
{
"epoch": 0.07,
"learning_rate": 6.800000000000001e-07,
"loss": 8.1559,
"step": 39
},
{
"epoch": 0.07,
"learning_rate": 7.000000000000001e-07,
"loss": 7.9231,
"step": 40
},
{
"epoch": 0.07,
"learning_rate": 7.2e-07,
"loss": 7.9967,
"step": 41
},
{
"epoch": 0.07,
"learning_rate": 7.4e-07,
"loss": 8.0424,
"step": 42
},
{
"epoch": 0.07,
"learning_rate": 7.6e-07,
"loss": 7.8849,
"step": 43
},
{
"epoch": 0.07,
"learning_rate": 7.8e-07,
"loss": 8.0896,
"step": 44
},
{
"epoch": 0.08,
"learning_rate": 8.000000000000001e-07,
"loss": 8.1857,
"step": 45
},
{
"epoch": 0.08,
"learning_rate": 8.200000000000001e-07,
"loss": 8.0947,
"step": 46
},
{
"epoch": 0.08,
"learning_rate": 8.400000000000001e-07,
"loss": 8.104,
"step": 47
},
{
"epoch": 0.08,
"learning_rate": 8.6e-07,
"loss": 8.1007,
"step": 48
},
{
"epoch": 0.08,
"learning_rate": 8.8e-07,
"loss": 8.2011,
"step": 49
},
{
"epoch": 0.08,
"learning_rate": 9.000000000000001e-07,
"loss": 8.2088,
"step": 50
},
{
"epoch": 0.09,
"learning_rate": 9.200000000000001e-07,
"loss": 7.6965,
"step": 51
},
{
"epoch": 0.09,
"learning_rate": 9.400000000000001e-07,
"loss": 7.8332,
"step": 52
},
{
"epoch": 0.09,
"learning_rate": 9.600000000000001e-07,
"loss": 7.6262,
"step": 53
},
{
"epoch": 0.09,
"learning_rate": 9.800000000000001e-07,
"loss": 7.6614,
"step": 54
},
{
"epoch": 0.09,
"learning_rate": 1.0000000000000002e-06,
"loss": 7.5486,
"step": 55
},
{
"epoch": 0.09,
"learning_rate": 1.02e-06,
"loss": 7.6535,
"step": 56
},
{
"epoch": 0.1,
"learning_rate": 1.04e-06,
"loss": 7.514,
"step": 57
},
{
"epoch": 0.1,
"learning_rate": 1.06e-06,
"loss": 7.6361,
"step": 58
},
{
"epoch": 0.1,
"learning_rate": 1.08e-06,
"loss": 7.6413,
"step": 59
},
{
"epoch": 0.1,
"learning_rate": 1.1e-06,
"loss": 7.476,
"step": 60
},
{
"epoch": 0.1,
"learning_rate": 1.12e-06,
"loss": 7.6203,
"step": 61
},
{
"epoch": 0.1,
"learning_rate": 1.14e-06,
"loss": 7.5593,
"step": 62
},
{
"epoch": 0.11,
"learning_rate": 1.1600000000000001e-06,
"loss": 7.4873,
"step": 63
},
{
"epoch": 0.11,
"learning_rate": 1.1800000000000001e-06,
"loss": 7.4919,
"step": 64
},
{
"epoch": 0.11,
"learning_rate": 1.2000000000000002e-06,
"loss": 7.4833,
"step": 65
},
{
"epoch": 0.11,
"learning_rate": 1.2200000000000002e-06,
"loss": 7.434,
"step": 66
},
{
"epoch": 0.11,
"learning_rate": 1.2400000000000002e-06,
"loss": 7.5032,
"step": 67
},
{
"epoch": 0.11,
"learning_rate": 1.26e-06,
"loss": 7.309,
"step": 68
},
{
"epoch": 0.12,
"learning_rate": 1.28e-06,
"loss": 7.4619,
"step": 69
},
{
"epoch": 0.12,
"learning_rate": 1.3e-06,
"loss": 7.438,
"step": 70
},
{
"epoch": 0.12,
"learning_rate": 1.32e-06,
"loss": 7.468,
"step": 71
},
{
"epoch": 0.12,
"learning_rate": 1.34e-06,
"loss": 7.515,
"step": 72
},
{
"epoch": 0.12,
"learning_rate": 1.3600000000000001e-06,
"loss": 7.2502,
"step": 73
},
{
"epoch": 0.12,
"learning_rate": 1.3800000000000001e-06,
"loss": 7.3669,
"step": 74
},
{
"epoch": 0.13,
"learning_rate": 1.4000000000000001e-06,
"loss": 7.4405,
"step": 75
},
{
"epoch": 0.13,
"learning_rate": 1.42e-06,
"loss": 7.4281,
"step": 76
},
{
"epoch": 0.13,
"learning_rate": 1.44e-06,
"loss": 7.4472,
"step": 77
},
{
"epoch": 0.13,
"learning_rate": 1.46e-06,
"loss": 7.4338,
"step": 78
},
{
"epoch": 0.13,
"learning_rate": 1.48e-06,
"loss": 7.3441,
"step": 79
},
{
"epoch": 0.13,
"learning_rate": 1.5e-06,
"loss": 7.3553,
"step": 80
},
{
"epoch": 0.14,
"learning_rate": 1.52e-06,
"loss": 7.3386,
"step": 81
},
{
"epoch": 0.14,
"learning_rate": 1.54e-06,
"loss": 7.418,
"step": 82
},
{
"epoch": 0.14,
"learning_rate": 1.56e-06,
"loss": 7.3155,
"step": 83
},
{
"epoch": 0.14,
"learning_rate": 1.5800000000000001e-06,
"loss": 7.398,
"step": 84
},
{
"epoch": 0.14,
"learning_rate": 1.6000000000000001e-06,
"loss": 7.2936,
"step": 85
},
{
"epoch": 0.14,
"learning_rate": 1.6200000000000002e-06,
"loss": 7.1857,
"step": 86
},
{
"epoch": 0.15,
"learning_rate": 1.6400000000000002e-06,
"loss": 7.3915,
"step": 87
},
{
"epoch": 0.15,
"learning_rate": 1.6600000000000002e-06,
"loss": 7.3553,
"step": 88
},
{
"epoch": 0.15,
"learning_rate": 1.6800000000000002e-06,
"loss": 7.2862,
"step": 89
},
{
"epoch": 0.15,
"learning_rate": 1.7000000000000002e-06,
"loss": 7.4464,
"step": 90
},
{
"epoch": 0.15,
"learning_rate": 1.72e-06,
"loss": 7.3681,
"step": 91
},
{
"epoch": 0.15,
"learning_rate": 1.74e-06,
"loss": 7.4004,
"step": 92
},
{
"epoch": 0.16,
"learning_rate": 1.76e-06,
"loss": 7.3562,
"step": 93
},
{
"epoch": 0.16,
"learning_rate": 1.7800000000000001e-06,
"loss": 7.388,
"step": 94
},
{
"epoch": 0.16,
"learning_rate": 1.8000000000000001e-06,
"loss": 7.2885,
"step": 95
},
{
"epoch": 0.16,
"learning_rate": 1.8200000000000002e-06,
"loss": 7.39,
"step": 96
},
{
"epoch": 0.16,
"learning_rate": 1.8400000000000002e-06,
"loss": 7.5075,
"step": 97
},
{
"epoch": 0.16,
"learning_rate": 1.8600000000000002e-06,
"loss": 7.1727,
"step": 98
},
{
"epoch": 0.17,
"learning_rate": 1.8800000000000002e-06,
"loss": 7.4873,
"step": 99
},
{
"epoch": 0.17,
"learning_rate": 1.9000000000000002e-06,
"loss": 7.2502,
"step": 100
},
{
"epoch": 0.17,
"learning_rate": 1.9200000000000003e-06,
"loss": 7.2948,
"step": 101
},
{
"epoch": 0.17,
"learning_rate": 1.94e-06,
"loss": 7.2064,
"step": 102
},
{
"epoch": 0.17,
"learning_rate": 1.9600000000000003e-06,
"loss": 7.207,
"step": 103
},
{
"epoch": 0.17,
"learning_rate": 1.98e-06,
"loss": 7.1543,
"step": 104
},
{
"epoch": 0.18,
"learning_rate": 2.0000000000000003e-06,
"loss": 7.1772,
"step": 105
},
{
"epoch": 0.18,
"learning_rate": 2.02e-06,
"loss": 7.2445,
"step": 106
},
{
"epoch": 0.18,
"learning_rate": 2.04e-06,
"loss": 7.3242,
"step": 107
},
{
"epoch": 0.18,
"learning_rate": 2.06e-06,
"loss": 7.3011,
"step": 108
},
{
"epoch": 0.18,
"learning_rate": 2.08e-06,
"loss": 7.2736,
"step": 109
},
{
"epoch": 0.18,
"learning_rate": 2.1000000000000002e-06,
"loss": 7.2264,
"step": 110
},
{
"epoch": 0.19,
"learning_rate": 2.12e-06,
"loss": 7.1384,
"step": 111
},
{
"epoch": 0.19,
"learning_rate": 2.1400000000000003e-06,
"loss": 7.1728,
"step": 112
},
{
"epoch": 0.19,
"learning_rate": 2.16e-06,
"loss": 7.1274,
"step": 113
},
{
"epoch": 0.19,
"learning_rate": 2.1800000000000003e-06,
"loss": 7.2443,
"step": 114
},
{
"epoch": 0.19,
"learning_rate": 2.2e-06,
"loss": 7.1811,
"step": 115
},
{
"epoch": 0.2,
"learning_rate": 2.2200000000000003e-06,
"loss": 7.1871,
"step": 116
},
{
"epoch": 0.2,
"learning_rate": 2.24e-06,
"loss": 7.215,
"step": 117
},
{
"epoch": 0.2,
"learning_rate": 2.2600000000000004e-06,
"loss": 7.142,
"step": 118
},
{
"epoch": 0.2,
"learning_rate": 2.28e-06,
"loss": 7.1879,
"step": 119
},
{
"epoch": 0.2,
"learning_rate": 2.3000000000000004e-06,
"loss": 7.1609,
"step": 120
},
{
"epoch": 0.2,
"learning_rate": 2.3200000000000002e-06,
"loss": 7.3064,
"step": 121
},
{
"epoch": 0.21,
"learning_rate": 2.3400000000000005e-06,
"loss": 7.3407,
"step": 122
},
{
"epoch": 0.21,
"learning_rate": 2.3600000000000003e-06,
"loss": 7.2515,
"step": 123
},
{
"epoch": 0.21,
"learning_rate": 2.38e-06,
"loss": 7.2728,
"step": 124
},
{
"epoch": 0.21,
"learning_rate": 2.4000000000000003e-06,
"loss": 7.1747,
"step": 125
},
{
"epoch": 0.21,
"learning_rate": 2.42e-06,
"loss": 7.2339,
"step": 126
},
{
"epoch": 0.21,
"learning_rate": 2.4400000000000004e-06,
"loss": 7.1446,
"step": 127
},
{
"epoch": 0.22,
"learning_rate": 2.46e-06,
"loss": 7.2047,
"step": 128
},
{
"epoch": 0.22,
"learning_rate": 2.4800000000000004e-06,
"loss": 7.1819,
"step": 129
},
{
"epoch": 0.22,
"learning_rate": 2.5e-06,
"loss": 7.2608,
"step": 130
},
{
"epoch": 0.22,
"learning_rate": 2.52e-06,
"loss": 7.1753,
"step": 131
},
{
"epoch": 0.22,
"learning_rate": 2.5400000000000002e-06,
"loss": 7.1165,
"step": 132
},
{
"epoch": 0.22,
"learning_rate": 2.56e-06,
"loss": 7.1877,
"step": 133
},
{
"epoch": 0.23,
"learning_rate": 2.5800000000000003e-06,
"loss": 7.3196,
"step": 134
},
{
"epoch": 0.23,
"learning_rate": 2.6e-06,
"loss": 7.2101,
"step": 135
},
{
"epoch": 0.23,
"learning_rate": 2.6200000000000003e-06,
"loss": 7.2951,
"step": 136
},
{
"epoch": 0.23,
"learning_rate": 2.64e-06,
"loss": 7.0824,
"step": 137
},
{
"epoch": 0.23,
"learning_rate": 2.6600000000000004e-06,
"loss": 7.189,
"step": 138
},
{
"epoch": 0.23,
"learning_rate": 2.68e-06,
"loss": 7.1444,
"step": 139
},
{
"epoch": 0.24,
"learning_rate": 2.7000000000000004e-06,
"loss": 7.1674,
"step": 140
},
{
"epoch": 0.24,
"learning_rate": 2.7200000000000002e-06,
"loss": 7.1791,
"step": 141
},
{
"epoch": 0.24,
"learning_rate": 2.7400000000000004e-06,
"loss": 7.3197,
"step": 142
},
{
"epoch": 0.24,
"learning_rate": 2.7600000000000003e-06,
"loss": 7.2949,
"step": 143
},
{
"epoch": 0.24,
"learning_rate": 2.7800000000000005e-06,
"loss": 6.9477,
"step": 144
},
{
"epoch": 0.24,
"learning_rate": 2.8000000000000003e-06,
"loss": 7.0258,
"step": 145
},
{
"epoch": 0.25,
"learning_rate": 2.82e-06,
"loss": 7.4421,
"step": 146
},
{
"epoch": 0.25,
"learning_rate": 2.84e-06,
"loss": 7.1155,
"step": 147
},
{
"epoch": 0.25,
"learning_rate": 2.86e-06,
"loss": 7.1012,
"step": 148
},
{
"epoch": 0.25,
"learning_rate": 2.88e-06,
"loss": 7.0881,
"step": 149
},
{
"epoch": 0.25,
"learning_rate": 2.9e-06,
"loss": 6.9097,
"step": 150
},
{
"epoch": 0.25,
"learning_rate": 2.92e-06,
"loss": 7.1856,
"step": 151
},
{
"epoch": 0.26,
"learning_rate": 2.9400000000000002e-06,
"loss": 7.2058,
"step": 152
},
{
"epoch": 0.26,
"learning_rate": 2.96e-06,
"loss": 7.2105,
"step": 153
},
{
"epoch": 0.26,
"learning_rate": 2.9800000000000003e-06,
"loss": 7.0837,
"step": 154
},
{
"epoch": 0.26,
"learning_rate": 3e-06,
"loss": 7.1757,
"step": 155
},
{
"epoch": 0.26,
"learning_rate": 3.0200000000000003e-06,
"loss": 7.1905,
"step": 156
},
{
"epoch": 0.26,
"learning_rate": 3.04e-06,
"loss": 7.1309,
"step": 157
},
{
"epoch": 0.27,
"learning_rate": 3.0600000000000003e-06,
"loss": 7.1657,
"step": 158
},
{
"epoch": 0.27,
"learning_rate": 3.08e-06,
"loss": 7.0089,
"step": 159
},
{
"epoch": 0.27,
"learning_rate": 3.1000000000000004e-06,
"loss": 7.1945,
"step": 160
},
{
"epoch": 0.27,
"learning_rate": 3.12e-06,
"loss": 7.0095,
"step": 161
},
{
"epoch": 0.27,
"learning_rate": 3.1400000000000004e-06,
"loss": 7.0303,
"step": 162
},
{
"epoch": 0.27,
"learning_rate": 3.1600000000000002e-06,
"loss": 7.1481,
"step": 163
},
{
"epoch": 0.28,
"learning_rate": 3.1800000000000005e-06,
"loss": 7.1551,
"step": 164
},
{
"epoch": 0.28,
"learning_rate": 3.2000000000000003e-06,
"loss": 7.0461,
"step": 165
},
{
"epoch": 0.28,
"learning_rate": 3.2200000000000005e-06,
"loss": 7.2556,
"step": 166
},
{
"epoch": 0.28,
"learning_rate": 3.2400000000000003e-06,
"loss": 7.2093,
"step": 167
},
{
"epoch": 0.28,
"learning_rate": 3.2600000000000006e-06,
"loss": 7.1808,
"step": 168
},
{
"epoch": 0.28,
"learning_rate": 3.2800000000000004e-06,
"loss": 7.069,
"step": 169
},
{
"epoch": 0.29,
"learning_rate": 3.3000000000000006e-06,
"loss": 7.1427,
"step": 170
},
{
"epoch": 0.29,
"learning_rate": 3.3200000000000004e-06,
"loss": 7.0845,
"step": 171
},
{
"epoch": 0.29,
"learning_rate": 3.3400000000000006e-06,
"loss": 6.9713,
"step": 172
},
{
"epoch": 0.29,
"learning_rate": 3.3600000000000004e-06,
"loss": 7.0579,
"step": 173
},
{
"epoch": 0.29,
"learning_rate": 3.3800000000000007e-06,
"loss": 6.9732,
"step": 174
},
{
"epoch": 0.29,
"learning_rate": 3.4000000000000005e-06,
"loss": 7.0666,
"step": 175
},
{
"epoch": 0.3,
"learning_rate": 3.4200000000000007e-06,
"loss": 7.1507,
"step": 176
},
{
"epoch": 0.3,
"learning_rate": 3.44e-06,
"loss": 7.1627,
"step": 177
},
{
"epoch": 0.3,
"learning_rate": 3.46e-06,
"loss": 7.056,
"step": 178
},
{
"epoch": 0.3,
"learning_rate": 3.48e-06,
"loss": 7.1262,
"step": 179
},
{
"epoch": 0.3,
"learning_rate": 3.5e-06,
"loss": 7.1895,
"step": 180
},
{
"epoch": 0.3,
"learning_rate": 3.52e-06,
"loss": 7.102,
"step": 181
},
{
"epoch": 0.31,
"learning_rate": 3.54e-06,
"loss": 7.1646,
"step": 182
},
{
"epoch": 0.31,
"learning_rate": 3.5600000000000002e-06,
"loss": 7.1235,
"step": 183
},
{
"epoch": 0.31,
"learning_rate": 3.58e-06,
"loss": 6.9836,
"step": 184
},
{
"epoch": 0.31,
"learning_rate": 3.6000000000000003e-06,
"loss": 7.2477,
"step": 185
},
{
"epoch": 0.31,
"learning_rate": 3.62e-06,
"loss": 7.0128,
"step": 186
},
{
"epoch": 0.31,
"learning_rate": 3.6400000000000003e-06,
"loss": 7.126,
"step": 187
},
{
"epoch": 0.32,
"learning_rate": 3.66e-06,
"loss": 7.0905,
"step": 188
},
{
"epoch": 0.32,
"learning_rate": 3.6800000000000003e-06,
"loss": 7.2065,
"step": 189
},
{
"epoch": 0.32,
"learning_rate": 3.7e-06,
"loss": 7.1866,
"step": 190
},
{
"epoch": 0.32,
"learning_rate": 3.7200000000000004e-06,
"loss": 7.1504,
"step": 191
},
{
"epoch": 0.32,
"learning_rate": 3.74e-06,
"loss": 7.2071,
"step": 192
},
{
"epoch": 0.32,
"learning_rate": 3.7600000000000004e-06,
"loss": 6.9719,
"step": 193
},
{
"epoch": 0.33,
"learning_rate": 3.7800000000000002e-06,
"loss": 7.0764,
"step": 194
},
{
"epoch": 0.33,
"learning_rate": 3.8000000000000005e-06,
"loss": 7.0092,
"step": 195
},
{
"epoch": 0.33,
"learning_rate": 3.820000000000001e-06,
"loss": 6.9236,
"step": 196
},
{
"epoch": 0.33,
"learning_rate": 3.8400000000000005e-06,
"loss": 7.0731,
"step": 197
},
{
"epoch": 0.33,
"learning_rate": 3.86e-06,
"loss": 7.048,
"step": 198
},
{
"epoch": 0.33,
"learning_rate": 3.88e-06,
"loss": 6.7116,
"step": 199
},
{
"epoch": 0.34,
"learning_rate": 3.900000000000001e-06,
"loss": 6.7435,
"step": 200
},
{
"epoch": 0.34,
"learning_rate": 3.920000000000001e-06,
"loss": 7.1091,
"step": 201
},
{
"epoch": 0.34,
"learning_rate": 3.94e-06,
"loss": 7.1345,
"step": 202
},
{
"epoch": 0.34,
"learning_rate": 3.96e-06,
"loss": 7.1767,
"step": 203
},
{
"epoch": 0.34,
"learning_rate": 3.980000000000001e-06,
"loss": 7.0125,
"step": 204
},
{
"epoch": 0.34,
"learning_rate": 4.000000000000001e-06,
"loss": 7.0674,
"step": 205
},
{
"epoch": 0.35,
"learning_rate": 4.0200000000000005e-06,
"loss": 7.0564,
"step": 206
},
{
"epoch": 0.35,
"learning_rate": 4.04e-06,
"loss": 7.149,
"step": 207
},
{
"epoch": 0.35,
"learning_rate": 4.060000000000001e-06,
"loss": 7.1658,
"step": 208
},
{
"epoch": 0.35,
"learning_rate": 4.08e-06,
"loss": 7.071,
"step": 209
},
{
"epoch": 0.35,
"learning_rate": 4.1e-06,
"loss": 7.1684,
"step": 210
},
{
"epoch": 0.35,
"learning_rate": 4.12e-06,
"loss": 7.1056,
"step": 211
},
{
"epoch": 0.36,
"learning_rate": 4.14e-06,
"loss": 7.1056,
"step": 212
},
{
"epoch": 0.36,
"learning_rate": 4.16e-06,
"loss": 7.0879,
"step": 213
},
{
"epoch": 0.36,
"learning_rate": 4.18e-06,
"loss": 7.1096,
"step": 214
},
{
"epoch": 0.36,
"learning_rate": 4.2000000000000004e-06,
"loss": 7.0527,
"step": 215
},
{
"epoch": 0.36,
"learning_rate": 4.22e-06,
"loss": 7.1545,
"step": 216
},
{
"epoch": 0.36,
"learning_rate": 4.24e-06,
"loss": 6.9937,
"step": 217
},
{
"epoch": 0.37,
"learning_rate": 4.26e-06,
"loss": 7.0043,
"step": 218
},
{
"epoch": 0.37,
"learning_rate": 4.2800000000000005e-06,
"loss": 6.9814,
"step": 219
},
{
"epoch": 0.37,
"learning_rate": 4.3e-06,
"loss": 7.1756,
"step": 220
},
{
"epoch": 0.37,
"learning_rate": 4.32e-06,
"loss": 7.0563,
"step": 221
},
{
"epoch": 0.37,
"learning_rate": 4.34e-06,
"loss": 7.0901,
"step": 222
},
{
"epoch": 0.37,
"learning_rate": 4.360000000000001e-06,
"loss": 7.0863,
"step": 223
},
{
"epoch": 0.38,
"learning_rate": 4.38e-06,
"loss": 7.1361,
"step": 224
},
{
"epoch": 0.38,
"learning_rate": 4.4e-06,
"loss": 7.1468,
"step": 225
},
{
"epoch": 0.38,
"learning_rate": 4.42e-06,
"loss": 6.9863,
"step": 226
},
{
"epoch": 0.38,
"learning_rate": 4.440000000000001e-06,
"loss": 6.9697,
"step": 227
},
{
"epoch": 0.38,
"learning_rate": 4.4600000000000005e-06,
"loss": 7.0623,
"step": 228
},
{
"epoch": 0.39,
"learning_rate": 4.48e-06,
"loss": 7.1303,
"step": 229
},
{
"epoch": 0.39,
"learning_rate": 4.5e-06,
"loss": 7.1264,
"step": 230
},
{
"epoch": 0.39,
"learning_rate": 4.520000000000001e-06,
"loss": 7.0863,
"step": 231
},
{
"epoch": 0.39,
"learning_rate": 4.540000000000001e-06,
"loss": 6.9809,
"step": 232
},
{
"epoch": 0.39,
"learning_rate": 4.56e-06,
"loss": 7.0967,
"step": 233
},
{
"epoch": 0.39,
"learning_rate": 4.58e-06,
"loss": 7.114,
"step": 234
},
{
"epoch": 0.4,
"learning_rate": 4.600000000000001e-06,
"loss": 7.1581,
"step": 235
},
{
"epoch": 0.4,
"learning_rate": 4.620000000000001e-06,
"loss": 6.9555,
"step": 236
},
{
"epoch": 0.4,
"learning_rate": 4.6400000000000005e-06,
"loss": 7.0181,
"step": 237
},
{
"epoch": 0.4,
"learning_rate": 4.66e-06,
"loss": 7.0655,
"step": 238
},
{
"epoch": 0.4,
"learning_rate": 4.680000000000001e-06,
"loss": 7.0629,
"step": 239
},
{
"epoch": 0.4,
"learning_rate": 4.7e-06,
"loss": 7.1218,
"step": 240
},
{
"epoch": 0.41,
"learning_rate": 4.7200000000000005e-06,
"loss": 7.1858,
"step": 241
},
{
"epoch": 0.41,
"learning_rate": 4.74e-06,
"loss": 7.0461,
"step": 242
},
{
"epoch": 0.41,
"learning_rate": 4.76e-06,
"loss": 7.1017,
"step": 243
},
{
"epoch": 0.41,
"learning_rate": 4.78e-06,
"loss": 7.0139,
"step": 244
},
{
"epoch": 0.41,
"learning_rate": 4.800000000000001e-06,
"loss": 6.9299,
"step": 245
},
{
"epoch": 0.41,
"learning_rate": 4.8200000000000004e-06,
"loss": 6.9379,
"step": 246
},
{
"epoch": 0.42,
"learning_rate": 4.84e-06,
"loss": 6.8947,
"step": 247
},
{
"epoch": 0.42,
"learning_rate": 4.86e-06,
"loss": 6.9973,
"step": 248
},
{
"epoch": 0.42,
"learning_rate": 4.880000000000001e-06,
"loss": 6.8035,
"step": 249
},
{
"epoch": 0.42,
"learning_rate": 4.9000000000000005e-06,
"loss": 6.6622,
"step": 250
},
{
"epoch": 0.42,
"learning_rate": 4.92e-06,
"loss": 7.178,
"step": 251
},
{
"epoch": 0.42,
"learning_rate": 4.94e-06,
"loss": 7.2078,
"step": 252
},
{
"epoch": 0.43,
"learning_rate": 4.960000000000001e-06,
"loss": 7.0564,
"step": 253
},
{
"epoch": 0.43,
"learning_rate": 4.980000000000001e-06,
"loss": 7.0769,
"step": 254
},
{
"epoch": 0.43,
"learning_rate": 5e-06,
"loss": 7.1539,
"step": 255
},
{
"epoch": 0.43,
"learning_rate": 5.02e-06,
"loss": 7.1108,
"step": 256
},
{
"epoch": 0.43,
"learning_rate": 5.04e-06,
"loss": 7.0991,
"step": 257
},
{
"epoch": 0.43,
"learning_rate": 5.060000000000001e-06,
"loss": 7.0253,
"step": 258
},
{
"epoch": 0.44,
"learning_rate": 5.0800000000000005e-06,
"loss": 7.138,
"step": 259
},
{
"epoch": 0.44,
"learning_rate": 5.1e-06,
"loss": 7.055,
"step": 260
},
{
"epoch": 0.44,
"learning_rate": 5.12e-06,
"loss": 7.0462,
"step": 261
},
{
"epoch": 0.44,
"learning_rate": 5.140000000000001e-06,
"loss": 7.0532,
"step": 262
},
{
"epoch": 0.44,
"learning_rate": 5.1600000000000006e-06,
"loss": 7.0895,
"step": 263
},
{
"epoch": 0.44,
"learning_rate": 5.18e-06,
"loss": 6.9003,
"step": 264
},
{
"epoch": 0.45,
"learning_rate": 5.2e-06,
"loss": 6.9966,
"step": 265
},
{
"epoch": 0.45,
"learning_rate": 5.220000000000001e-06,
"loss": 7.0119,
"step": 266
},
{
"epoch": 0.45,
"learning_rate": 5.240000000000001e-06,
"loss": 7.091,
"step": 267
},
{
"epoch": 0.45,
"learning_rate": 5.2600000000000005e-06,
"loss": 6.9715,
"step": 268
},
{
"epoch": 0.45,
"learning_rate": 5.28e-06,
"loss": 7.1006,
"step": 269
},
{
"epoch": 0.45,
"learning_rate": 5.300000000000001e-06,
"loss": 7.0264,
"step": 270
},
{
"epoch": 0.46,
"learning_rate": 5.320000000000001e-06,
"loss": 7.0949,
"step": 271
},
{
"epoch": 0.46,
"learning_rate": 5.3400000000000005e-06,
"loss": 7.1155,
"step": 272
},
{
"epoch": 0.46,
"learning_rate": 5.36e-06,
"loss": 7.0131,
"step": 273
},
{
"epoch": 0.46,
"learning_rate": 5.380000000000001e-06,
"loss": 7.0184,
"step": 274
},
{
"epoch": 0.46,
"learning_rate": 5.400000000000001e-06,
"loss": 7.0663,
"step": 275
},
{
"epoch": 0.46,
"learning_rate": 5.420000000000001e-06,
"loss": 7.0464,
"step": 276
},
{
"epoch": 0.47,
"learning_rate": 5.4400000000000004e-06,
"loss": 7.1456,
"step": 277
},
{
"epoch": 0.47,
"learning_rate": 5.460000000000001e-06,
"loss": 7.0548,
"step": 278
},
{
"epoch": 0.47,
"learning_rate": 5.480000000000001e-06,
"loss": 7.0388,
"step": 279
},
{
"epoch": 0.47,
"learning_rate": 5.500000000000001e-06,
"loss": 7.0853,
"step": 280
},
{
"epoch": 0.47,
"learning_rate": 5.5200000000000005e-06,
"loss": 7.0649,
"step": 281
},
{
"epoch": 0.47,
"learning_rate": 5.540000000000001e-06,
"loss": 7.0957,
"step": 282
},
{
"epoch": 0.48,
"learning_rate": 5.560000000000001e-06,
"loss": 6.9417,
"step": 283
},
{
"epoch": 0.48,
"learning_rate": 5.580000000000001e-06,
"loss": 7.0445,
"step": 284
},
{
"epoch": 0.48,
"learning_rate": 5.600000000000001e-06,
"loss": 7.0993,
"step": 285
},
{
"epoch": 0.48,
"learning_rate": 5.620000000000001e-06,
"loss": 6.9807,
"step": 286
},
{
"epoch": 0.48,
"learning_rate": 5.64e-06,
"loss": 7.1612,
"step": 287
},
{
"epoch": 0.48,
"learning_rate": 5.66e-06,
"loss": 6.9164,
"step": 288
},
{
"epoch": 0.49,
"learning_rate": 5.68e-06,
"loss": 6.9623,
"step": 289
},
{
"epoch": 0.49,
"learning_rate": 5.7e-06,
"loss": 7.1175,
"step": 290
},
{
"epoch": 0.49,
"learning_rate": 5.72e-06,
"loss": 7.1494,
"step": 291
},
{
"epoch": 0.49,
"learning_rate": 5.74e-06,
"loss": 6.9501,
"step": 292
},
{
"epoch": 0.49,
"learning_rate": 5.76e-06,
"loss": 7.0356,
"step": 293
},
{
"epoch": 0.49,
"learning_rate": 5.78e-06,
"loss": 7.0778,
"step": 294
},
{
"epoch": 0.5,
"learning_rate": 5.8e-06,
"loss": 6.8589,
"step": 295
},
{
"epoch": 0.5,
"learning_rate": 5.82e-06,
"loss": 6.8831,
"step": 296
},
{
"epoch": 0.5,
"learning_rate": 5.84e-06,
"loss": 6.7751,
"step": 297
},
{
"epoch": 0.5,
"learning_rate": 5.86e-06,
"loss": 6.8866,
"step": 298
},
{
"epoch": 0.5,
"learning_rate": 5.8800000000000005e-06,
"loss": 6.7705,
"step": 299
},
{
"epoch": 0.5,
"learning_rate": 5.9e-06,
"loss": 6.6378,
"step": 300
},
{
"epoch": 0.51,
"learning_rate": 5.92e-06,
"loss": 7.1476,
"step": 301
},
{
"epoch": 0.51,
"learning_rate": 5.94e-06,
"loss": 7.1939,
"step": 302
},
{
"epoch": 0.51,
"learning_rate": 5.9600000000000005e-06,
"loss": 7.1753,
"step": 303
},
{
"epoch": 0.51,
"learning_rate": 5.98e-06,
"loss": 7.1118,
"step": 304
},
{
"epoch": 0.51,
"learning_rate": 6e-06,
"loss": 7.1377,
"step": 305
},
{
"epoch": 0.51,
"learning_rate": 6.02e-06,
"loss": 7.191,
"step": 306
},
{
"epoch": 0.52,
"learning_rate": 6.040000000000001e-06,
"loss": 7.0375,
"step": 307
},
{
"epoch": 0.52,
"learning_rate": 6.0600000000000004e-06,
"loss": 7.1247,
"step": 308
},
{
"epoch": 0.52,
"learning_rate": 6.08e-06,
"loss": 7.009,
"step": 309
},
{
"epoch": 0.52,
"learning_rate": 6.1e-06,
"loss": 7.0828,
"step": 310
},
{
"epoch": 0.52,
"learning_rate": 6.120000000000001e-06,
"loss": 6.9627,
"step": 311
},
{
"epoch": 0.52,
"learning_rate": 6.1400000000000005e-06,
"loss": 7.1735,
"step": 312
},
{
"epoch": 0.53,
"learning_rate": 6.16e-06,
"loss": 6.9515,
"step": 313
},
{
"epoch": 0.53,
"learning_rate": 6.18e-06,
"loss": 7.0213,
"step": 314
},
{
"epoch": 0.53,
"learning_rate": 6.200000000000001e-06,
"loss": 7.0429,
"step": 315
},
{
"epoch": 0.53,
"learning_rate": 6.220000000000001e-06,
"loss": 6.9728,
"step": 316
},
{
"epoch": 0.53,
"learning_rate": 6.24e-06,
"loss": 6.9617,
"step": 317
},
{
"epoch": 0.53,
"learning_rate": 6.26e-06,
"loss": 7.1781,
"step": 318
},
{
"epoch": 0.54,
"learning_rate": 6.280000000000001e-06,
"loss": 7.0572,
"step": 319
},
{
"epoch": 0.54,
"learning_rate": 6.300000000000001e-06,
"loss": 7.0887,
"step": 320
},
{
"epoch": 0.54,
"learning_rate": 6.3200000000000005e-06,
"loss": 6.9978,
"step": 321
},
{
"epoch": 0.54,
"learning_rate": 6.34e-06,
"loss": 6.9942,
"step": 322
},
{
"epoch": 0.54,
"learning_rate": 6.360000000000001e-06,
"loss": 6.9522,
"step": 323
},
{
"epoch": 0.54,
"learning_rate": 6.380000000000001e-06,
"loss": 6.9385,
"step": 324
},
{
"epoch": 0.55,
"learning_rate": 6.4000000000000006e-06,
"loss": 7.1572,
"step": 325
},
{
"epoch": 0.55,
"learning_rate": 6.42e-06,
"loss": 7.0409,
"step": 326
},
{
"epoch": 0.55,
"learning_rate": 6.440000000000001e-06,
"loss": 7.2063,
"step": 327
},
{
"epoch": 0.55,
"learning_rate": 6.460000000000001e-06,
"loss": 7.0266,
"step": 328
},
{
"epoch": 0.55,
"learning_rate": 6.480000000000001e-06,
"loss": 6.9566,
"step": 329
},
{
"epoch": 0.55,
"learning_rate": 6.5000000000000004e-06,
"loss": 7.0507,
"step": 330
},
{
"epoch": 0.56,
"learning_rate": 6.520000000000001e-06,
"loss": 7.0022,
"step": 331
},
{
"epoch": 0.56,
"learning_rate": 6.540000000000001e-06,
"loss": 6.9985,
"step": 332
},
{
"epoch": 0.56,
"learning_rate": 6.560000000000001e-06,
"loss": 6.929,
"step": 333
},
{
"epoch": 0.56,
"learning_rate": 6.5800000000000005e-06,
"loss": 7.0259,
"step": 334
},
{
"epoch": 0.56,
"learning_rate": 6.600000000000001e-06,
"loss": 7.065,
"step": 335
},
{
"epoch": 0.56,
"learning_rate": 6.620000000000001e-06,
"loss": 6.9689,
"step": 336
},
{
"epoch": 0.57,
"learning_rate": 6.640000000000001e-06,
"loss": 7.0373,
"step": 337
},
{
"epoch": 0.57,
"learning_rate": 6.660000000000001e-06,
"loss": 7.0155,
"step": 338
},
{
"epoch": 0.57,
"learning_rate": 6.680000000000001e-06,
"loss": 7.1228,
"step": 339
},
{
"epoch": 0.57,
"learning_rate": 6.700000000000001e-06,
"loss": 7.1218,
"step": 340
},
{
"epoch": 0.57,
"learning_rate": 6.720000000000001e-06,
"loss": 6.9836,
"step": 341
},
{
"epoch": 0.58,
"learning_rate": 6.740000000000001e-06,
"loss": 6.9952,
"step": 342
},
{
"epoch": 0.58,
"learning_rate": 6.760000000000001e-06,
"loss": 6.8988,
"step": 343
},
{
"epoch": 0.58,
"learning_rate": 6.780000000000001e-06,
"loss": 7.0253,
"step": 344
},
{
"epoch": 0.58,
"learning_rate": 6.800000000000001e-06,
"loss": 6.9399,
"step": 345
},
{
"epoch": 0.58,
"learning_rate": 6.820000000000001e-06,
"loss": 6.8285,
"step": 346
},
{
"epoch": 0.58,
"learning_rate": 6.8400000000000014e-06,
"loss": 6.825,
"step": 347
},
{
"epoch": 0.59,
"learning_rate": 6.860000000000001e-06,
"loss": 7.0763,
"step": 348
},
{
"epoch": 0.59,
"learning_rate": 6.88e-06,
"loss": 6.8147,
"step": 349
},
{
"epoch": 0.59,
"learning_rate": 6.9e-06,
"loss": 6.745,
"step": 350
},
{
"epoch": 0.59,
"learning_rate": 6.92e-06,
"loss": 7.204,
"step": 351
},
{
"epoch": 0.59,
"learning_rate": 6.9400000000000005e-06,
"loss": 7.2319,
"step": 352
},
{
"epoch": 0.59,
"learning_rate": 6.96e-06,
"loss": 7.2049,
"step": 353
},
{
"epoch": 0.6,
"learning_rate": 6.98e-06,
"loss": 7.1772,
"step": 354
},
{
"epoch": 0.6,
"learning_rate": 7e-06,
"loss": 7.1362,
"step": 355
},
{
"epoch": 0.6,
"learning_rate": 7.0200000000000006e-06,
"loss": 7.1392,
"step": 356
},
{
"epoch": 0.6,
"learning_rate": 7.04e-06,
"loss": 7.0229,
"step": 357
},
{
"epoch": 0.6,
"learning_rate": 7.06e-06,
"loss": 7.0526,
"step": 358
},
{
"epoch": 0.6,
"learning_rate": 7.08e-06,
"loss": 7.0012,
"step": 359
},
{
"epoch": 0.61,
"learning_rate": 7.100000000000001e-06,
"loss": 7.0498,
"step": 360
},
{
"epoch": 0.61,
"learning_rate": 7.1200000000000004e-06,
"loss": 7.0708,
"step": 361
},
{
"epoch": 0.61,
"learning_rate": 7.14e-06,
"loss": 6.9313,
"step": 362
},
{
"epoch": 0.61,
"learning_rate": 7.16e-06,
"loss": 7.1228,
"step": 363
},
{
"epoch": 0.61,
"learning_rate": 7.180000000000001e-06,
"loss": 7.0275,
"step": 364
},
{
"epoch": 0.61,
"learning_rate": 7.2000000000000005e-06,
"loss": 7.1301,
"step": 365
},
{
"epoch": 0.62,
"learning_rate": 7.22e-06,
"loss": 6.8894,
"step": 366
},
{
"epoch": 0.62,
"learning_rate": 7.24e-06,
"loss": 7.1013,
"step": 367
},
{
"epoch": 0.62,
"learning_rate": 7.260000000000001e-06,
"loss": 6.9501,
"step": 368
},
{
"epoch": 0.62,
"learning_rate": 7.280000000000001e-06,
"loss": 6.9579,
"step": 369
},
{
"epoch": 0.62,
"learning_rate": 7.3e-06,
"loss": 7.114,
"step": 370
},
{
"epoch": 0.62,
"learning_rate": 7.32e-06,
"loss": 6.9799,
"step": 371
},
{
"epoch": 0.63,
"learning_rate": 7.340000000000001e-06,
"loss": 6.9302,
"step": 372
},
{
"epoch": 0.63,
"learning_rate": 7.360000000000001e-06,
"loss": 6.9814,
"step": 373
},
{
"epoch": 0.63,
"learning_rate": 7.3800000000000005e-06,
"loss": 6.8834,
"step": 374
},
{
"epoch": 0.63,
"learning_rate": 7.4e-06,
"loss": 6.9998,
"step": 375
},
{
"epoch": 0.63,
"learning_rate": 7.420000000000001e-06,
"loss": 6.9156,
"step": 376
},
{
"epoch": 0.63,
"learning_rate": 7.440000000000001e-06,
"loss": 6.9523,
"step": 377
},
{
"epoch": 0.64,
"learning_rate": 7.4600000000000006e-06,
"loss": 7.0801,
"step": 378
},
{
"epoch": 0.64,
"learning_rate": 7.48e-06,
"loss": 6.9825,
"step": 379
},
{
"epoch": 0.64,
"learning_rate": 7.500000000000001e-06,
"loss": 7.0974,
"step": 380
},
{
"epoch": 0.64,
"learning_rate": 7.520000000000001e-06,
"loss": 7.0487,
"step": 381
},
{
"epoch": 0.64,
"learning_rate": 7.540000000000001e-06,
"loss": 6.9902,
"step": 382
},
{
"epoch": 0.64,
"learning_rate": 7.5600000000000005e-06,
"loss": 6.8837,
"step": 383
},
{
"epoch": 0.65,
"learning_rate": 7.58e-06,
"loss": 7.0034,
"step": 384
},
{
"epoch": 0.65,
"learning_rate": 7.600000000000001e-06,
"loss": 7.0496,
"step": 385
},
{
"epoch": 0.65,
"learning_rate": 7.620000000000001e-06,
"loss": 7.017,
"step": 386
},
{
"epoch": 0.65,
"learning_rate": 7.640000000000001e-06,
"loss": 7.0979,
"step": 387
},
{
"epoch": 0.65,
"learning_rate": 7.660000000000001e-06,
"loss": 7.027,
"step": 388
},
{
"epoch": 0.65,
"learning_rate": 7.680000000000001e-06,
"loss": 6.921,
"step": 389
},
{
"epoch": 0.66,
"learning_rate": 7.7e-06,
"loss": 6.9699,
"step": 390
},
{
"epoch": 0.66,
"learning_rate": 7.72e-06,
"loss": 7.0364,
"step": 391
},
{
"epoch": 0.66,
"learning_rate": 7.74e-06,
"loss": 6.9877,
"step": 392
},
{
"epoch": 0.66,
"learning_rate": 7.76e-06,
"loss": 6.9522,
"step": 393
},
{
"epoch": 0.66,
"learning_rate": 7.78e-06,
"loss": 6.9362,
"step": 394
},
{
"epoch": 0.66,
"learning_rate": 7.800000000000002e-06,
"loss": 6.7218,
"step": 395
},
{
"epoch": 0.67,
"learning_rate": 7.820000000000001e-06,
"loss": 6.7015,
"step": 396
},
{
"epoch": 0.67,
"learning_rate": 7.840000000000001e-06,
"loss": 6.8373,
"step": 397
},
{
"epoch": 0.67,
"learning_rate": 7.860000000000001e-06,
"loss": 6.8885,
"step": 398
},
{
"epoch": 0.67,
"learning_rate": 7.88e-06,
"loss": 6.5494,
"step": 399
},
{
"epoch": 0.67,
"learning_rate": 7.9e-06,
"loss": 6.3983,
"step": 400
},
{
"epoch": 0.67,
"learning_rate": 7.92e-06,
"loss": 7.2372,
"step": 401
},
{
"epoch": 0.68,
"learning_rate": 7.94e-06,
"loss": 7.1236,
"step": 402
},
{
"epoch": 0.68,
"learning_rate": 7.960000000000002e-06,
"loss": 7.2272,
"step": 403
},
{
"epoch": 0.68,
"learning_rate": 7.980000000000002e-06,
"loss": 7.1397,
"step": 404
},
{
"epoch": 0.68,
"learning_rate": 8.000000000000001e-06,
"loss": 7.1128,
"step": 405
},
{
"epoch": 0.68,
"learning_rate": 8.020000000000001e-06,
"loss": 6.9502,
"step": 406
},
{
"epoch": 0.68,
"learning_rate": 8.040000000000001e-06,
"loss": 6.8923,
"step": 407
},
{
"epoch": 0.69,
"learning_rate": 8.06e-06,
"loss": 7.0852,
"step": 408
},
{
"epoch": 0.69,
"learning_rate": 8.08e-06,
"loss": 6.9212,
"step": 409
},
{
"epoch": 0.69,
"learning_rate": 8.1e-06,
"loss": 6.9236,
"step": 410
},
{
"epoch": 0.69,
"learning_rate": 8.120000000000002e-06,
"loss": 6.9754,
"step": 411
},
{
"epoch": 0.69,
"learning_rate": 8.14e-06,
"loss": 6.9512,
"step": 412
},
{
"epoch": 0.69,
"learning_rate": 8.16e-06,
"loss": 6.8273,
"step": 413
},
{
"epoch": 0.7,
"learning_rate": 8.18e-06,
"loss": 6.9514,
"step": 414
},
{
"epoch": 0.7,
"learning_rate": 8.2e-06,
"loss": 7.1138,
"step": 415
},
{
"epoch": 0.7,
"learning_rate": 8.220000000000001e-06,
"loss": 6.8356,
"step": 416
},
{
"epoch": 0.7,
"learning_rate": 8.24e-06,
"loss": 6.9481,
"step": 417
},
{
"epoch": 0.7,
"learning_rate": 8.26e-06,
"loss": 6.7483,
"step": 418
},
{
"epoch": 0.7,
"learning_rate": 8.28e-06,
"loss": 6.8859,
"step": 419
},
{
"epoch": 0.71,
"learning_rate": 8.3e-06,
"loss": 6.9719,
"step": 420
},
{
"epoch": 0.71,
"learning_rate": 8.32e-06,
"loss": 6.8554,
"step": 421
},
{
"epoch": 0.71,
"learning_rate": 8.34e-06,
"loss": 6.8961,
"step": 422
},
{
"epoch": 0.71,
"learning_rate": 8.36e-06,
"loss": 6.915,
"step": 423
},
{
"epoch": 0.71,
"learning_rate": 8.380000000000001e-06,
"loss": 6.9823,
"step": 424
},
{
"epoch": 0.71,
"learning_rate": 8.400000000000001e-06,
"loss": 6.9163,
"step": 425
},
{
"epoch": 0.72,
"learning_rate": 8.42e-06,
"loss": 6.9953,
"step": 426
},
{
"epoch": 0.72,
"learning_rate": 8.44e-06,
"loss": 6.9091,
"step": 427
},
{
"epoch": 0.72,
"learning_rate": 8.46e-06,
"loss": 6.8866,
"step": 428
},
{
"epoch": 0.72,
"learning_rate": 8.48e-06,
"loss": 7.0465,
"step": 429
},
{
"epoch": 0.72,
"learning_rate": 8.5e-06,
"loss": 6.8745,
"step": 430
},
{
"epoch": 0.72,
"learning_rate": 8.52e-06,
"loss": 6.8777,
"step": 431
},
{
"epoch": 0.73,
"learning_rate": 8.540000000000001e-06,
"loss": 6.8274,
"step": 432
},
{
"epoch": 0.73,
"learning_rate": 8.560000000000001e-06,
"loss": 6.9802,
"step": 433
},
{
"epoch": 0.73,
"learning_rate": 8.580000000000001e-06,
"loss": 6.9284,
"step": 434
},
{
"epoch": 0.73,
"learning_rate": 8.6e-06,
"loss": 7.008,
"step": 435
},
{
"epoch": 0.73,
"learning_rate": 8.62e-06,
"loss": 6.8494,
"step": 436
},
{
"epoch": 0.73,
"learning_rate": 8.64e-06,
"loss": 6.8169,
"step": 437
},
{
"epoch": 0.74,
"learning_rate": 8.66e-06,
"loss": 6.8364,
"step": 438
},
{
"epoch": 0.74,
"learning_rate": 8.68e-06,
"loss": 6.8554,
"step": 439
},
{
"epoch": 0.74,
"learning_rate": 8.700000000000001e-06,
"loss": 6.9575,
"step": 440
},
{
"epoch": 0.74,
"learning_rate": 8.720000000000001e-06,
"loss": 6.737,
"step": 441
},
{
"epoch": 0.74,
"learning_rate": 8.740000000000001e-06,
"loss": 6.9307,
"step": 442
},
{
"epoch": 0.74,
"learning_rate": 8.76e-06,
"loss": 6.7614,
"step": 443
},
{
"epoch": 0.75,
"learning_rate": 8.78e-06,
"loss": 6.7005,
"step": 444
},
{
"epoch": 0.75,
"learning_rate": 8.8e-06,
"loss": 6.6435,
"step": 445
},
{
"epoch": 0.75,
"learning_rate": 8.82e-06,
"loss": 6.8544,
"step": 446
},
{
"epoch": 0.75,
"learning_rate": 8.84e-06,
"loss": 6.6232,
"step": 447
},
{
"epoch": 0.75,
"learning_rate": 8.860000000000002e-06,
"loss": 6.4901,
"step": 448
},
{
"epoch": 0.75,
"learning_rate": 8.880000000000001e-06,
"loss": 6.6787,
"step": 449
},
{
"epoch": 0.76,
"learning_rate": 8.900000000000001e-06,
"loss": 6.3934,
"step": 450
},
{
"epoch": 0.76,
"learning_rate": 8.920000000000001e-06,
"loss": 6.8428,
"step": 451
},
{
"epoch": 0.76,
"learning_rate": 8.94e-06,
"loss": 6.7912,
"step": 452
},
{
"epoch": 0.76,
"learning_rate": 8.96e-06,
"loss": 6.7057,
"step": 453
},
{
"epoch": 0.76,
"learning_rate": 8.98e-06,
"loss": 6.5682,
"step": 454
},
{
"epoch": 0.77,
"learning_rate": 9e-06,
"loss": 6.7899,
"step": 455
},
{
"epoch": 0.77,
"learning_rate": 9.020000000000002e-06,
"loss": 6.7245,
"step": 456
},
{
"epoch": 0.77,
"learning_rate": 9.040000000000002e-06,
"loss": 6.6646,
"step": 457
},
{
"epoch": 0.77,
"learning_rate": 9.060000000000001e-06,
"loss": 6.838,
"step": 458
},
{
"epoch": 0.77,
"learning_rate": 9.080000000000001e-06,
"loss": 6.5864,
"step": 459
},
{
"epoch": 0.77,
"learning_rate": 9.100000000000001e-06,
"loss": 6.5744,
"step": 460
},
{
"epoch": 0.78,
"learning_rate": 9.12e-06,
"loss": 6.649,
"step": 461
},
{
"epoch": 0.78,
"learning_rate": 9.14e-06,
"loss": 6.5776,
"step": 462
},
{
"epoch": 0.78,
"learning_rate": 9.16e-06,
"loss": 6.5176,
"step": 463
},
{
"epoch": 0.78,
"learning_rate": 9.180000000000002e-06,
"loss": 6.5483,
"step": 464
},
{
"epoch": 0.78,
"learning_rate": 9.200000000000002e-06,
"loss": 6.611,
"step": 465
},
{
"epoch": 0.78,
"learning_rate": 9.220000000000002e-06,
"loss": 6.5864,
"step": 466
},
{
"epoch": 0.79,
"learning_rate": 9.240000000000001e-06,
"loss": 6.6553,
"step": 467
},
{
"epoch": 0.79,
"learning_rate": 9.260000000000001e-06,
"loss": 6.5245,
"step": 468
},
{
"epoch": 0.79,
"learning_rate": 9.280000000000001e-06,
"loss": 6.4231,
"step": 469
},
{
"epoch": 0.79,
"learning_rate": 9.3e-06,
"loss": 6.4863,
"step": 470
},
{
"epoch": 0.79,
"learning_rate": 9.32e-06,
"loss": 6.4302,
"step": 471
},
{
"epoch": 0.79,
"learning_rate": 9.340000000000002e-06,
"loss": 6.4779,
"step": 472
},
{
"epoch": 0.8,
"learning_rate": 9.360000000000002e-06,
"loss": 6.463,
"step": 473
},
{
"epoch": 0.8,
"learning_rate": 9.38e-06,
"loss": 6.4628,
"step": 474
},
{
"epoch": 0.8,
"learning_rate": 9.4e-06,
"loss": 6.3983,
"step": 475
},
{
"epoch": 0.8,
"learning_rate": 9.42e-06,
"loss": 6.3678,
"step": 476
},
{
"epoch": 0.8,
"learning_rate": 9.440000000000001e-06,
"loss": 6.4204,
"step": 477
},
{
"epoch": 0.8,
"learning_rate": 9.460000000000001e-06,
"loss": 6.3695,
"step": 478
},
{
"epoch": 0.81,
"learning_rate": 9.48e-06,
"loss": 6.3032,
"step": 479
},
{
"epoch": 0.81,
"learning_rate": 9.5e-06,
"loss": 6.2476,
"step": 480
},
{
"epoch": 0.81,
"learning_rate": 9.52e-06,
"loss": 6.447,
"step": 481
},
{
"epoch": 0.81,
"learning_rate": 9.54e-06,
"loss": 6.4076,
"step": 482
},
{
"epoch": 0.81,
"learning_rate": 9.56e-06,
"loss": 6.4206,
"step": 483
},
{
"epoch": 0.81,
"learning_rate": 9.58e-06,
"loss": 6.3157,
"step": 484
},
{
"epoch": 0.82,
"learning_rate": 9.600000000000001e-06,
"loss": 6.4491,
"step": 485
},
{
"epoch": 0.82,
"learning_rate": 9.620000000000001e-06,
"loss": 6.3701,
"step": 486
},
{
"epoch": 0.82,
"learning_rate": 9.640000000000001e-06,
"loss": 6.332,
"step": 487
},
{
"epoch": 0.82,
"learning_rate": 9.66e-06,
"loss": 6.408,
"step": 488
},
{
"epoch": 0.82,
"learning_rate": 9.68e-06,
"loss": 6.23,
"step": 489
},
{
"epoch": 0.82,
"learning_rate": 9.7e-06,
"loss": 6.1728,
"step": 490
},
{
"epoch": 0.83,
"learning_rate": 9.72e-06,
"loss": 6.4291,
"step": 491
},
{
"epoch": 0.83,
"learning_rate": 9.74e-06,
"loss": 6.3717,
"step": 492
},
{
"epoch": 0.83,
"learning_rate": 9.760000000000001e-06,
"loss": 6.2002,
"step": 493
},
{
"epoch": 0.83,
"learning_rate": 9.780000000000001e-06,
"loss": 6.2088,
"step": 494
},
{
"epoch": 0.83,
"learning_rate": 9.800000000000001e-06,
"loss": 6.4054,
"step": 495
},
{
"epoch": 0.83,
"learning_rate": 9.820000000000001e-06,
"loss": 6.5151,
"step": 496
},
{
"epoch": 0.84,
"learning_rate": 9.84e-06,
"loss": 6.237,
"step": 497
},
{
"epoch": 0.84,
"learning_rate": 9.86e-06,
"loss": 6.3757,
"step": 498
},
{
"epoch": 0.84,
"learning_rate": 9.88e-06,
"loss": 6.2802,
"step": 499
},
{
"epoch": 0.84,
"learning_rate": 9.9e-06,
"loss": 6.255,
"step": 500
},
{
"epoch": 0.84,
"eval_loss": 6.07191801071167,
"eval_runtime": 429.7022,
"eval_samples_per_second": 6.148,
"eval_steps_per_second": 0.514,
"eval_wer": 0.9409885509058079,
"step": 500
},
{
"epoch": 0.84,
"learning_rate": 9.920000000000002e-06,
"loss": 6.3112,
"step": 501
},
{
"epoch": 0.84,
"learning_rate": 9.940000000000001e-06,
"loss": 6.3351,
"step": 502
},
{
"epoch": 0.85,
"learning_rate": 9.960000000000001e-06,
"loss": 6.1704,
"step": 503
},
{
"epoch": 0.85,
"learning_rate": 9.980000000000001e-06,
"loss": 6.2321,
"step": 504
},
{
"epoch": 0.85,
"learning_rate": 1e-05,
"loss": 6.2253,
"step": 505
},
{
"epoch": 0.85,
"learning_rate": 9.893617021276596e-06,
"loss": 6.1916,
"step": 506
},
{
"epoch": 0.85,
"learning_rate": 9.787234042553192e-06,
"loss": 6.0859,
"step": 507
},
{
"epoch": 0.85,
"learning_rate": 9.680851063829787e-06,
"loss": 6.097,
"step": 508
},
{
"epoch": 0.86,
"learning_rate": 9.574468085106385e-06,
"loss": 6.0029,
"step": 509
},
{
"epoch": 0.86,
"learning_rate": 9.46808510638298e-06,
"loss": 6.2107,
"step": 510
},
{
"epoch": 0.86,
"learning_rate": 9.361702127659576e-06,
"loss": 6.1482,
"step": 511
},
{
"epoch": 0.86,
"learning_rate": 9.255319148936171e-06,
"loss": 6.0517,
"step": 512
},
{
"epoch": 0.86,
"learning_rate": 9.148936170212767e-06,
"loss": 6.0906,
"step": 513
},
{
"epoch": 0.86,
"learning_rate": 9.042553191489362e-06,
"loss": 5.9753,
"step": 514
},
{
"epoch": 0.87,
"learning_rate": 8.936170212765958e-06,
"loss": 6.1427,
"step": 515
},
{
"epoch": 0.87,
"learning_rate": 8.829787234042555e-06,
"loss": 6.0517,
"step": 516
},
{
"epoch": 0.87,
"learning_rate": 8.72340425531915e-06,
"loss": 6.0664,
"step": 517
},
{
"epoch": 0.87,
"learning_rate": 8.617021276595746e-06,
"loss": 6.1817,
"step": 518
},
{
"epoch": 0.87,
"learning_rate": 8.510638297872341e-06,
"loss": 6.0093,
"step": 519
},
{
"epoch": 0.87,
"learning_rate": 8.404255319148937e-06,
"loss": 6.1801,
"step": 520
},
{
"epoch": 0.88,
"learning_rate": 8.297872340425532e-06,
"loss": 6.1169,
"step": 521
},
{
"epoch": 0.88,
"learning_rate": 8.191489361702128e-06,
"loss": 6.0626,
"step": 522
},
{
"epoch": 0.88,
"learning_rate": 8.085106382978723e-06,
"loss": 6.0347,
"step": 523
},
{
"epoch": 0.88,
"learning_rate": 7.97872340425532e-06,
"loss": 5.8924,
"step": 524
},
{
"epoch": 0.88,
"learning_rate": 7.872340425531916e-06,
"loss": 5.9368,
"step": 525
},
{
"epoch": 0.88,
"learning_rate": 7.765957446808511e-06,
"loss": 5.8941,
"step": 526
},
{
"epoch": 0.89,
"learning_rate": 7.659574468085107e-06,
"loss": 6.0771,
"step": 527
},
{
"epoch": 0.89,
"learning_rate": 7.553191489361703e-06,
"loss": 6.0417,
"step": 528
},
{
"epoch": 0.89,
"learning_rate": 7.446808510638298e-06,
"loss": 5.9173,
"step": 529
},
{
"epoch": 0.89,
"learning_rate": 7.340425531914894e-06,
"loss": 5.9807,
"step": 530
},
{
"epoch": 0.89,
"learning_rate": 7.234042553191491e-06,
"loss": 5.8555,
"step": 531
},
{
"epoch": 0.89,
"learning_rate": 7.127659574468085e-06,
"loss": 6.0597,
"step": 532
},
{
"epoch": 0.9,
"learning_rate": 7.021276595744682e-06,
"loss": 5.9052,
"step": 533
},
{
"epoch": 0.9,
"learning_rate": 6.914893617021278e-06,
"loss": 5.9172,
"step": 534
},
{
"epoch": 0.9,
"learning_rate": 6.808510638297873e-06,
"loss": 6.0196,
"step": 535
},
{
"epoch": 0.9,
"learning_rate": 6.702127659574469e-06,
"loss": 5.9001,
"step": 536
},
{
"epoch": 0.9,
"learning_rate": 6.595744680851064e-06,
"loss": 6.0601,
"step": 537
},
{
"epoch": 0.9,
"learning_rate": 6.48936170212766e-06,
"loss": 5.934,
"step": 538
},
{
"epoch": 0.91,
"learning_rate": 6.382978723404256e-06,
"loss": 5.9092,
"step": 539
},
{
"epoch": 0.91,
"learning_rate": 6.276595744680851e-06,
"loss": 5.8966,
"step": 540
},
{
"epoch": 0.91,
"learning_rate": 6.170212765957447e-06,
"loss": 5.7967,
"step": 541
},
{
"epoch": 0.91,
"learning_rate": 6.063829787234044e-06,
"loss": 5.8617,
"step": 542
},
{
"epoch": 0.91,
"learning_rate": 5.957446808510638e-06,
"loss": 5.828,
"step": 543
},
{
"epoch": 0.91,
"learning_rate": 5.851063829787235e-06,
"loss": 5.9159,
"step": 544
},
{
"epoch": 0.92,
"learning_rate": 5.744680851063831e-06,
"loss": 6.0056,
"step": 545
},
{
"epoch": 0.92,
"learning_rate": 5.638297872340426e-06,
"loss": 6.0341,
"step": 546
},
{
"epoch": 0.92,
"learning_rate": 5.531914893617022e-06,
"loss": 5.8778,
"step": 547
},
{
"epoch": 0.92,
"learning_rate": 5.425531914893617e-06,
"loss": 6.0156,
"step": 548
},
{
"epoch": 0.92,
"learning_rate": 5.319148936170213e-06,
"loss": 6.2097,
"step": 549
},
{
"epoch": 0.92,
"learning_rate": 5.212765957446809e-06,
"loss": 5.6322,
"step": 550
},
{
"epoch": 0.93,
"learning_rate": 5.106382978723404e-06,
"loss": 5.8747,
"step": 551
},
{
"epoch": 0.93,
"learning_rate": 5e-06,
"loss": 5.8933,
"step": 552
},
{
"epoch": 0.93,
"learning_rate": 4.893617021276596e-06,
"loss": 6.0209,
"step": 553
},
{
"epoch": 0.93,
"learning_rate": 4.787234042553192e-06,
"loss": 5.807,
"step": 554
},
{
"epoch": 0.93,
"learning_rate": 4.680851063829788e-06,
"loss": 5.7649,
"step": 555
},
{
"epoch": 0.93,
"learning_rate": 4.574468085106383e-06,
"loss": 5.9386,
"step": 556
},
{
"epoch": 0.94,
"learning_rate": 4.468085106382979e-06,
"loss": 5.9375,
"step": 557
},
{
"epoch": 0.94,
"learning_rate": 4.361702127659575e-06,
"loss": 5.8553,
"step": 558
},
{
"epoch": 0.94,
"learning_rate": 4.255319148936171e-06,
"loss": 5.6976,
"step": 559
},
{
"epoch": 0.94,
"learning_rate": 4.148936170212766e-06,
"loss": 5.8118,
"step": 560
},
{
"epoch": 0.94,
"learning_rate": 4.042553191489362e-06,
"loss": 6.0723,
"step": 561
},
{
"epoch": 0.94,
"learning_rate": 3.936170212765958e-06,
"loss": 5.733,
"step": 562
},
{
"epoch": 0.95,
"learning_rate": 3.8297872340425535e-06,
"loss": 5.7999,
"step": 563
},
{
"epoch": 0.95,
"learning_rate": 3.723404255319149e-06,
"loss": 5.9407,
"step": 564
},
{
"epoch": 0.95,
"learning_rate": 3.6170212765957453e-06,
"loss": 5.8233,
"step": 565
},
{
"epoch": 0.95,
"learning_rate": 3.510638297872341e-06,
"loss": 5.8286,
"step": 566
},
{
"epoch": 0.95,
"learning_rate": 3.4042553191489363e-06,
"loss": 5.7935,
"step": 567
},
{
"epoch": 0.96,
"learning_rate": 3.297872340425532e-06,
"loss": 5.7849,
"step": 568
},
{
"epoch": 0.96,
"learning_rate": 3.191489361702128e-06,
"loss": 5.7043,
"step": 569
},
{
"epoch": 0.96,
"learning_rate": 3.0851063829787237e-06,
"loss": 5.8238,
"step": 570
},
{
"epoch": 0.96,
"learning_rate": 2.978723404255319e-06,
"loss": 5.7665,
"step": 571
},
{
"epoch": 0.96,
"learning_rate": 2.8723404255319155e-06,
"loss": 5.7923,
"step": 572
},
{
"epoch": 0.96,
"learning_rate": 2.765957446808511e-06,
"loss": 5.8948,
"step": 573
},
{
"epoch": 0.97,
"learning_rate": 2.6595744680851065e-06,
"loss": 5.7315,
"step": 574
},
{
"epoch": 0.97,
"learning_rate": 2.553191489361702e-06,
"loss": 5.884,
"step": 575
},
{
"epoch": 0.97,
"learning_rate": 2.446808510638298e-06,
"loss": 5.7937,
"step": 576
},
{
"epoch": 0.97,
"learning_rate": 2.340425531914894e-06,
"loss": 5.7292,
"step": 577
},
{
"epoch": 0.97,
"learning_rate": 2.2340425531914894e-06,
"loss": 5.6651,
"step": 578
},
{
"epoch": 0.97,
"learning_rate": 2.1276595744680853e-06,
"loss": 5.5988,
"step": 579
},
{
"epoch": 0.98,
"learning_rate": 2.021276595744681e-06,
"loss": 5.876,
"step": 580
},
{
"epoch": 0.98,
"learning_rate": 1.9148936170212767e-06,
"loss": 5.9059,
"step": 581
},
{
"epoch": 0.98,
"learning_rate": 1.8085106382978727e-06,
"loss": 5.8383,
"step": 582
},
{
"epoch": 0.98,
"learning_rate": 1.7021276595744682e-06,
"loss": 5.6995,
"step": 583
},
{
"epoch": 0.98,
"learning_rate": 1.595744680851064e-06,
"loss": 5.7479,
"step": 584
},
{
"epoch": 0.98,
"learning_rate": 1.4893617021276596e-06,
"loss": 5.7665,
"step": 585
},
{
"epoch": 0.99,
"learning_rate": 1.3829787234042555e-06,
"loss": 5.7168,
"step": 586
},
{
"epoch": 0.99,
"learning_rate": 1.276595744680851e-06,
"loss": 5.7855,
"step": 587
},
{
"epoch": 0.99,
"learning_rate": 1.170212765957447e-06,
"loss": 5.8604,
"step": 588
},
{
"epoch": 0.99,
"learning_rate": 1.0638297872340427e-06,
"loss": 5.833,
"step": 589
},
{
"epoch": 0.99,
"learning_rate": 9.574468085106384e-07,
"loss": 5.7115,
"step": 590
},
{
"epoch": 0.99,
"learning_rate": 8.510638297872341e-07,
"loss": 6.0308,
"step": 591
},
{
"epoch": 1.0,
"learning_rate": 7.446808510638298e-07,
"loss": 5.5846,
"step": 592
},
{
"epoch": 1.0,
"learning_rate": 6.382978723404255e-07,
"loss": 6.0995,
"step": 593
},
{
"epoch": 1.0,
"learning_rate": 5.319148936170213e-07,
"loss": 5.8076,
"step": 594
},
{
"epoch": 1.0,
"step": 594,
"total_flos": 0.0,
"train_loss": 6.969648663845127,
"train_runtime": 5241.1822,
"train_samples_per_second": 5.445,
"train_steps_per_second": 0.113
}
],
"max_steps": 594,
"num_train_epochs": 1,
"total_flos": 0.0,
"trial_name": null,
"trial_params": null
}