wav2vec2-gpt2-wandb-grid-search / trainer_state.json
sanchit-gandhi's picture
End of training
137974c
raw
history blame
No virus
67.8 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": 4.8225,
"step": 1
},
{
"epoch": 0.0,
"learning_rate": 0.0,
"loss": 5.0072,
"step": 2
},
{
"epoch": 0.01,
"learning_rate": 2.0000000000000002e-07,
"loss": 4.8898,
"step": 3
},
{
"epoch": 0.01,
"learning_rate": 4.0000000000000003e-07,
"loss": 4.839,
"step": 4
},
{
"epoch": 0.01,
"learning_rate": 6.000000000000001e-07,
"loss": 4.7915,
"step": 5
},
{
"epoch": 0.01,
"learning_rate": 8.000000000000001e-07,
"loss": 4.8052,
"step": 6
},
{
"epoch": 0.01,
"learning_rate": 1.0000000000000002e-06,
"loss": 4.8753,
"step": 7
},
{
"epoch": 0.01,
"learning_rate": 1.2000000000000002e-06,
"loss": 4.7787,
"step": 8
},
{
"epoch": 0.02,
"learning_rate": 1.4000000000000001e-06,
"loss": 4.8916,
"step": 9
},
{
"epoch": 0.02,
"learning_rate": 1.6000000000000001e-06,
"loss": 4.7974,
"step": 10
},
{
"epoch": 0.02,
"learning_rate": 1.8e-06,
"loss": 4.7885,
"step": 11
},
{
"epoch": 0.02,
"learning_rate": 2.0000000000000003e-06,
"loss": 4.8247,
"step": 12
},
{
"epoch": 0.02,
"learning_rate": 2.2e-06,
"loss": 4.7332,
"step": 13
},
{
"epoch": 0.02,
"learning_rate": 2.4000000000000003e-06,
"loss": 4.6684,
"step": 14
},
{
"epoch": 0.03,
"learning_rate": 2.6e-06,
"loss": 4.5584,
"step": 15
},
{
"epoch": 0.03,
"learning_rate": 2.8000000000000003e-06,
"loss": 4.7161,
"step": 16
},
{
"epoch": 0.03,
"learning_rate": 3e-06,
"loss": 4.7131,
"step": 17
},
{
"epoch": 0.03,
"learning_rate": 3.2000000000000003e-06,
"loss": 4.7672,
"step": 18
},
{
"epoch": 0.03,
"learning_rate": 3.4000000000000005e-06,
"loss": 4.5319,
"step": 19
},
{
"epoch": 0.03,
"learning_rate": 3.6e-06,
"loss": 4.7361,
"step": 20
},
{
"epoch": 0.04,
"learning_rate": 3.8e-06,
"loss": 4.6663,
"step": 21
},
{
"epoch": 0.04,
"learning_rate": 4.000000000000001e-06,
"loss": 4.6147,
"step": 22
},
{
"epoch": 0.04,
"learning_rate": 4.2000000000000004e-06,
"loss": 4.6091,
"step": 23
},
{
"epoch": 0.04,
"learning_rate": 4.4e-06,
"loss": 4.662,
"step": 24
},
{
"epoch": 0.04,
"learning_rate": 4.6e-06,
"loss": 4.4311,
"step": 25
},
{
"epoch": 0.04,
"learning_rate": 4.800000000000001e-06,
"loss": 4.8399,
"step": 26
},
{
"epoch": 0.05,
"learning_rate": 5e-06,
"loss": 4.474,
"step": 27
},
{
"epoch": 0.05,
"learning_rate": 5.2e-06,
"loss": 4.379,
"step": 28
},
{
"epoch": 0.05,
"learning_rate": 5.4e-06,
"loss": 4.3821,
"step": 29
},
{
"epoch": 0.05,
"learning_rate": 5.600000000000001e-06,
"loss": 4.5227,
"step": 30
},
{
"epoch": 0.05,
"learning_rate": 5.8e-06,
"loss": 4.5345,
"step": 31
},
{
"epoch": 0.05,
"learning_rate": 6e-06,
"loss": 4.4068,
"step": 32
},
{
"epoch": 0.06,
"learning_rate": 6.2e-06,
"loss": 4.394,
"step": 33
},
{
"epoch": 0.06,
"learning_rate": 6.4000000000000006e-06,
"loss": 4.3383,
"step": 34
},
{
"epoch": 0.06,
"learning_rate": 6.6e-06,
"loss": 4.3907,
"step": 35
},
{
"epoch": 0.06,
"learning_rate": 6.800000000000001e-06,
"loss": 4.5407,
"step": 36
},
{
"epoch": 0.06,
"learning_rate": 7.000000000000001e-06,
"loss": 4.3381,
"step": 37
},
{
"epoch": 0.06,
"learning_rate": 7.2e-06,
"loss": 4.4952,
"step": 38
},
{
"epoch": 0.07,
"learning_rate": 7.4e-06,
"loss": 4.6521,
"step": 39
},
{
"epoch": 0.07,
"learning_rate": 7.6e-06,
"loss": 4.3231,
"step": 40
},
{
"epoch": 0.07,
"learning_rate": 7.8e-06,
"loss": 4.3164,
"step": 41
},
{
"epoch": 0.07,
"learning_rate": 8.000000000000001e-06,
"loss": 4.5604,
"step": 42
},
{
"epoch": 0.07,
"learning_rate": 8.200000000000001e-06,
"loss": 4.2994,
"step": 43
},
{
"epoch": 0.07,
"learning_rate": 8.400000000000001e-06,
"loss": 4.4294,
"step": 44
},
{
"epoch": 0.08,
"learning_rate": 8.599999999999999e-06,
"loss": 4.5385,
"step": 45
},
{
"epoch": 0.08,
"learning_rate": 8.8e-06,
"loss": 4.5593,
"step": 46
},
{
"epoch": 0.08,
"learning_rate": 9e-06,
"loss": 4.4697,
"step": 47
},
{
"epoch": 0.08,
"learning_rate": 9.2e-06,
"loss": 4.8313,
"step": 48
},
{
"epoch": 0.08,
"learning_rate": 9.4e-06,
"loss": 4.9122,
"step": 49
},
{
"epoch": 0.08,
"learning_rate": 9.600000000000001e-06,
"loss": 4.5792,
"step": 50
},
{
"epoch": 0.09,
"learning_rate": 9.800000000000001e-06,
"loss": 4.2615,
"step": 51
},
{
"epoch": 0.09,
"learning_rate": 1e-05,
"loss": 4.2647,
"step": 52
},
{
"epoch": 0.09,
"learning_rate": 1.02e-05,
"loss": 4.2893,
"step": 53
},
{
"epoch": 0.09,
"learning_rate": 1.04e-05,
"loss": 4.3664,
"step": 54
},
{
"epoch": 0.09,
"learning_rate": 1.06e-05,
"loss": 4.2194,
"step": 55
},
{
"epoch": 0.09,
"learning_rate": 1.08e-05,
"loss": 4.4206,
"step": 56
},
{
"epoch": 0.1,
"learning_rate": 1.1000000000000001e-05,
"loss": 4.2292,
"step": 57
},
{
"epoch": 0.1,
"learning_rate": 1.1200000000000001e-05,
"loss": 4.5213,
"step": 58
},
{
"epoch": 0.1,
"learning_rate": 1.1400000000000001e-05,
"loss": 4.2824,
"step": 59
},
{
"epoch": 0.1,
"learning_rate": 1.16e-05,
"loss": 4.0693,
"step": 60
},
{
"epoch": 0.1,
"learning_rate": 1.18e-05,
"loss": 4.3305,
"step": 61
},
{
"epoch": 0.1,
"learning_rate": 1.2e-05,
"loss": 4.3873,
"step": 62
},
{
"epoch": 0.11,
"learning_rate": 1.22e-05,
"loss": 4.2663,
"step": 63
},
{
"epoch": 0.11,
"learning_rate": 1.24e-05,
"loss": 4.2255,
"step": 64
},
{
"epoch": 0.11,
"learning_rate": 1.2600000000000001e-05,
"loss": 4.28,
"step": 65
},
{
"epoch": 0.11,
"learning_rate": 1.2800000000000001e-05,
"loss": 4.2312,
"step": 66
},
{
"epoch": 0.11,
"learning_rate": 1.3000000000000001e-05,
"loss": 4.1834,
"step": 67
},
{
"epoch": 0.11,
"learning_rate": 1.32e-05,
"loss": 4.1685,
"step": 68
},
{
"epoch": 0.12,
"learning_rate": 1.3400000000000002e-05,
"loss": 4.2772,
"step": 69
},
{
"epoch": 0.12,
"learning_rate": 1.3600000000000002e-05,
"loss": 4.2668,
"step": 70
},
{
"epoch": 0.12,
"learning_rate": 1.3800000000000002e-05,
"loss": 4.3441,
"step": 71
},
{
"epoch": 0.12,
"learning_rate": 1.4000000000000001e-05,
"loss": 4.3228,
"step": 72
},
{
"epoch": 0.12,
"learning_rate": 1.42e-05,
"loss": 4.1117,
"step": 73
},
{
"epoch": 0.12,
"learning_rate": 1.44e-05,
"loss": 4.3191,
"step": 74
},
{
"epoch": 0.13,
"learning_rate": 1.4599999999999999e-05,
"loss": 4.2973,
"step": 75
},
{
"epoch": 0.13,
"learning_rate": 1.48e-05,
"loss": 4.273,
"step": 76
},
{
"epoch": 0.13,
"learning_rate": 1.5e-05,
"loss": 4.2649,
"step": 77
},
{
"epoch": 0.13,
"learning_rate": 1.52e-05,
"loss": 4.2333,
"step": 78
},
{
"epoch": 0.13,
"learning_rate": 1.54e-05,
"loss": 4.1605,
"step": 79
},
{
"epoch": 0.13,
"learning_rate": 1.56e-05,
"loss": 4.3374,
"step": 80
},
{
"epoch": 0.14,
"learning_rate": 1.58e-05,
"loss": 4.2885,
"step": 81
},
{
"epoch": 0.14,
"learning_rate": 1.6000000000000003e-05,
"loss": 4.2615,
"step": 82
},
{
"epoch": 0.14,
"learning_rate": 1.62e-05,
"loss": 4.342,
"step": 83
},
{
"epoch": 0.14,
"learning_rate": 1.6400000000000002e-05,
"loss": 4.3158,
"step": 84
},
{
"epoch": 0.14,
"learning_rate": 1.66e-05,
"loss": 4.2401,
"step": 85
},
{
"epoch": 0.14,
"learning_rate": 1.6800000000000002e-05,
"loss": 4.1713,
"step": 86
},
{
"epoch": 0.15,
"learning_rate": 1.7000000000000003e-05,
"loss": 4.4382,
"step": 87
},
{
"epoch": 0.15,
"learning_rate": 1.7199999999999998e-05,
"loss": 4.2933,
"step": 88
},
{
"epoch": 0.15,
"learning_rate": 1.74e-05,
"loss": 4.2668,
"step": 89
},
{
"epoch": 0.15,
"learning_rate": 1.76e-05,
"loss": 4.3194,
"step": 90
},
{
"epoch": 0.15,
"learning_rate": 1.78e-05,
"loss": 4.2625,
"step": 91
},
{
"epoch": 0.15,
"learning_rate": 1.8e-05,
"loss": 4.3701,
"step": 92
},
{
"epoch": 0.16,
"learning_rate": 1.8200000000000002e-05,
"loss": 4.2481,
"step": 93
},
{
"epoch": 0.16,
"learning_rate": 1.84e-05,
"loss": 4.4905,
"step": 94
},
{
"epoch": 0.16,
"learning_rate": 1.86e-05,
"loss": 4.3724,
"step": 95
},
{
"epoch": 0.16,
"learning_rate": 1.88e-05,
"loss": 4.4327,
"step": 96
},
{
"epoch": 0.16,
"learning_rate": 1.9e-05,
"loss": 4.4057,
"step": 97
},
{
"epoch": 0.16,
"learning_rate": 1.9200000000000003e-05,
"loss": 4.2809,
"step": 98
},
{
"epoch": 0.17,
"learning_rate": 1.94e-05,
"loss": 4.519,
"step": 99
},
{
"epoch": 0.17,
"learning_rate": 1.9600000000000002e-05,
"loss": 4.209,
"step": 100
},
{
"epoch": 0.17,
"learning_rate": 1.9800000000000004e-05,
"loss": 4.3489,
"step": 101
},
{
"epoch": 0.17,
"learning_rate": 2e-05,
"loss": 4.1833,
"step": 102
},
{
"epoch": 0.17,
"learning_rate": 2.0200000000000003e-05,
"loss": 4.1774,
"step": 103
},
{
"epoch": 0.17,
"learning_rate": 2.04e-05,
"loss": 4.126,
"step": 104
},
{
"epoch": 0.18,
"learning_rate": 2.06e-05,
"loss": 4.1726,
"step": 105
},
{
"epoch": 0.18,
"learning_rate": 2.08e-05,
"loss": 4.1543,
"step": 106
},
{
"epoch": 0.18,
"learning_rate": 2.1e-05,
"loss": 4.2942,
"step": 107
},
{
"epoch": 0.18,
"learning_rate": 2.12e-05,
"loss": 4.3168,
"step": 108
},
{
"epoch": 0.18,
"learning_rate": 2.1400000000000002e-05,
"loss": 4.3971,
"step": 109
},
{
"epoch": 0.18,
"learning_rate": 2.16e-05,
"loss": 4.206,
"step": 110
},
{
"epoch": 0.19,
"learning_rate": 2.18e-05,
"loss": 4.2,
"step": 111
},
{
"epoch": 0.19,
"learning_rate": 2.2000000000000003e-05,
"loss": 4.2552,
"step": 112
},
{
"epoch": 0.19,
"learning_rate": 2.22e-05,
"loss": 4.2447,
"step": 113
},
{
"epoch": 0.19,
"learning_rate": 2.2400000000000002e-05,
"loss": 4.281,
"step": 114
},
{
"epoch": 0.19,
"learning_rate": 2.26e-05,
"loss": 4.2514,
"step": 115
},
{
"epoch": 0.2,
"learning_rate": 2.2800000000000002e-05,
"loss": 4.2428,
"step": 116
},
{
"epoch": 0.2,
"learning_rate": 2.3000000000000003e-05,
"loss": 4.3366,
"step": 117
},
{
"epoch": 0.2,
"learning_rate": 2.32e-05,
"loss": 4.1153,
"step": 118
},
{
"epoch": 0.2,
"learning_rate": 2.3400000000000003e-05,
"loss": 4.1132,
"step": 119
},
{
"epoch": 0.2,
"learning_rate": 2.36e-05,
"loss": 4.1045,
"step": 120
},
{
"epoch": 0.2,
"learning_rate": 2.38e-05,
"loss": 4.3175,
"step": 121
},
{
"epoch": 0.21,
"learning_rate": 2.4e-05,
"loss": 4.4242,
"step": 122
},
{
"epoch": 0.21,
"learning_rate": 2.4200000000000002e-05,
"loss": 4.1832,
"step": 123
},
{
"epoch": 0.21,
"learning_rate": 2.44e-05,
"loss": 4.2404,
"step": 124
},
{
"epoch": 0.21,
"learning_rate": 2.46e-05,
"loss": 4.2611,
"step": 125
},
{
"epoch": 0.21,
"learning_rate": 2.48e-05,
"loss": 4.0726,
"step": 126
},
{
"epoch": 0.21,
"learning_rate": 2.5e-05,
"loss": 4.1251,
"step": 127
},
{
"epoch": 0.22,
"learning_rate": 2.5200000000000003e-05,
"loss": 4.1081,
"step": 128
},
{
"epoch": 0.22,
"learning_rate": 2.54e-05,
"loss": 4.2353,
"step": 129
},
{
"epoch": 0.22,
"learning_rate": 2.5600000000000002e-05,
"loss": 4.2703,
"step": 130
},
{
"epoch": 0.22,
"learning_rate": 2.58e-05,
"loss": 4.2179,
"step": 131
},
{
"epoch": 0.22,
"learning_rate": 2.6000000000000002e-05,
"loss": 4.1872,
"step": 132
},
{
"epoch": 0.22,
"learning_rate": 2.6200000000000003e-05,
"loss": 4.2557,
"step": 133
},
{
"epoch": 0.23,
"learning_rate": 2.64e-05,
"loss": 4.2838,
"step": 134
},
{
"epoch": 0.23,
"learning_rate": 2.6600000000000003e-05,
"loss": 4.3264,
"step": 135
},
{
"epoch": 0.23,
"learning_rate": 2.6800000000000004e-05,
"loss": 4.1789,
"step": 136
},
{
"epoch": 0.23,
"learning_rate": 2.7000000000000002e-05,
"loss": 4.1824,
"step": 137
},
{
"epoch": 0.23,
"learning_rate": 2.7200000000000004e-05,
"loss": 4.2021,
"step": 138
},
{
"epoch": 0.23,
"learning_rate": 2.7400000000000002e-05,
"loss": 4.2113,
"step": 139
},
{
"epoch": 0.24,
"learning_rate": 2.7600000000000003e-05,
"loss": 4.3841,
"step": 140
},
{
"epoch": 0.24,
"learning_rate": 2.7800000000000005e-05,
"loss": 4.2928,
"step": 141
},
{
"epoch": 0.24,
"learning_rate": 2.8000000000000003e-05,
"loss": 4.3591,
"step": 142
},
{
"epoch": 0.24,
"learning_rate": 2.8199999999999998e-05,
"loss": 4.3945,
"step": 143
},
{
"epoch": 0.24,
"learning_rate": 2.84e-05,
"loss": 4.2167,
"step": 144
},
{
"epoch": 0.24,
"learning_rate": 2.86e-05,
"loss": 4.2745,
"step": 145
},
{
"epoch": 0.25,
"learning_rate": 2.88e-05,
"loss": 4.6106,
"step": 146
},
{
"epoch": 0.25,
"learning_rate": 2.9e-05,
"loss": 4.5692,
"step": 147
},
{
"epoch": 0.25,
"learning_rate": 2.9199999999999998e-05,
"loss": 4.5427,
"step": 148
},
{
"epoch": 0.25,
"learning_rate": 2.94e-05,
"loss": 4.5165,
"step": 149
},
{
"epoch": 0.25,
"learning_rate": 2.96e-05,
"loss": 4.5584,
"step": 150
},
{
"epoch": 0.25,
"learning_rate": 2.98e-05,
"loss": 4.2149,
"step": 151
},
{
"epoch": 0.26,
"learning_rate": 3e-05,
"loss": 4.1459,
"step": 152
},
{
"epoch": 0.26,
"learning_rate": 3.02e-05,
"loss": 4.1111,
"step": 153
},
{
"epoch": 0.26,
"learning_rate": 3.04e-05,
"loss": 4.0712,
"step": 154
},
{
"epoch": 0.26,
"learning_rate": 3.06e-05,
"loss": 4.0859,
"step": 155
},
{
"epoch": 0.26,
"learning_rate": 3.08e-05,
"loss": 4.1,
"step": 156
},
{
"epoch": 0.26,
"learning_rate": 3.1e-05,
"loss": 4.2382,
"step": 157
},
{
"epoch": 0.27,
"learning_rate": 3.12e-05,
"loss": 4.0741,
"step": 158
},
{
"epoch": 0.27,
"learning_rate": 3.1400000000000004e-05,
"loss": 4.038,
"step": 159
},
{
"epoch": 0.27,
"learning_rate": 3.16e-05,
"loss": 4.1515,
"step": 160
},
{
"epoch": 0.27,
"learning_rate": 3.18e-05,
"loss": 4.1819,
"step": 161
},
{
"epoch": 0.27,
"learning_rate": 3.2000000000000005e-05,
"loss": 4.0791,
"step": 162
},
{
"epoch": 0.27,
"learning_rate": 3.2200000000000003e-05,
"loss": 4.0986,
"step": 163
},
{
"epoch": 0.28,
"learning_rate": 3.24e-05,
"loss": 4.1264,
"step": 164
},
{
"epoch": 0.28,
"learning_rate": 3.26e-05,
"loss": 4.1626,
"step": 165
},
{
"epoch": 0.28,
"learning_rate": 3.2800000000000004e-05,
"loss": 4.2766,
"step": 166
},
{
"epoch": 0.28,
"learning_rate": 3.3e-05,
"loss": 4.1187,
"step": 167
},
{
"epoch": 0.28,
"learning_rate": 3.32e-05,
"loss": 4.3421,
"step": 168
},
{
"epoch": 0.28,
"learning_rate": 3.3400000000000005e-05,
"loss": 4.1639,
"step": 169
},
{
"epoch": 0.29,
"learning_rate": 3.3600000000000004e-05,
"loss": 4.2129,
"step": 170
},
{
"epoch": 0.29,
"learning_rate": 3.38e-05,
"loss": 4.0701,
"step": 171
},
{
"epoch": 0.29,
"learning_rate": 3.4000000000000007e-05,
"loss": 4.0963,
"step": 172
},
{
"epoch": 0.29,
"learning_rate": 3.4200000000000005e-05,
"loss": 4.1,
"step": 173
},
{
"epoch": 0.29,
"learning_rate": 3.4399999999999996e-05,
"loss": 4.1168,
"step": 174
},
{
"epoch": 0.29,
"learning_rate": 3.46e-05,
"loss": 4.2434,
"step": 175
},
{
"epoch": 0.3,
"learning_rate": 3.48e-05,
"loss": 4.2658,
"step": 176
},
{
"epoch": 0.3,
"learning_rate": 3.5e-05,
"loss": 4.1915,
"step": 177
},
{
"epoch": 0.3,
"learning_rate": 3.52e-05,
"loss": 4.0929,
"step": 178
},
{
"epoch": 0.3,
"learning_rate": 3.54e-05,
"loss": 4.2164,
"step": 179
},
{
"epoch": 0.3,
"learning_rate": 3.56e-05,
"loss": 4.0857,
"step": 180
},
{
"epoch": 0.3,
"learning_rate": 3.58e-05,
"loss": 4.1816,
"step": 181
},
{
"epoch": 0.31,
"learning_rate": 3.6e-05,
"loss": 4.153,
"step": 182
},
{
"epoch": 0.31,
"learning_rate": 3.62e-05,
"loss": 4.1573,
"step": 183
},
{
"epoch": 0.31,
"learning_rate": 3.6400000000000004e-05,
"loss": 4.0508,
"step": 184
},
{
"epoch": 0.31,
"learning_rate": 3.66e-05,
"loss": 4.401,
"step": 185
},
{
"epoch": 0.31,
"learning_rate": 3.68e-05,
"loss": 4.1857,
"step": 186
},
{
"epoch": 0.31,
"learning_rate": 3.7e-05,
"loss": 4.2432,
"step": 187
},
{
"epoch": 0.32,
"learning_rate": 3.72e-05,
"loss": 4.2723,
"step": 188
},
{
"epoch": 0.32,
"learning_rate": 3.74e-05,
"loss": 4.4056,
"step": 189
},
{
"epoch": 0.32,
"learning_rate": 3.76e-05,
"loss": 4.1092,
"step": 190
},
{
"epoch": 0.32,
"learning_rate": 3.7800000000000004e-05,
"loss": 4.4034,
"step": 191
},
{
"epoch": 0.32,
"learning_rate": 3.8e-05,
"loss": 4.3196,
"step": 192
},
{
"epoch": 0.32,
"learning_rate": 3.82e-05,
"loss": 4.3128,
"step": 193
},
{
"epoch": 0.33,
"learning_rate": 3.8400000000000005e-05,
"loss": 4.3066,
"step": 194
},
{
"epoch": 0.33,
"learning_rate": 3.86e-05,
"loss": 4.2342,
"step": 195
},
{
"epoch": 0.33,
"learning_rate": 3.88e-05,
"loss": 4.1387,
"step": 196
},
{
"epoch": 0.33,
"learning_rate": 3.9000000000000006e-05,
"loss": 4.5583,
"step": 197
},
{
"epoch": 0.33,
"learning_rate": 3.9200000000000004e-05,
"loss": 4.4962,
"step": 198
},
{
"epoch": 0.33,
"learning_rate": 3.94e-05,
"loss": 4.4661,
"step": 199
},
{
"epoch": 0.34,
"learning_rate": 3.960000000000001e-05,
"loss": 4.5158,
"step": 200
},
{
"epoch": 0.34,
"learning_rate": 3.9800000000000005e-05,
"loss": 4.035,
"step": 201
},
{
"epoch": 0.34,
"learning_rate": 4e-05,
"loss": 4.1883,
"step": 202
},
{
"epoch": 0.34,
"learning_rate": 4.02e-05,
"loss": 4.2433,
"step": 203
},
{
"epoch": 0.34,
"learning_rate": 4.0400000000000006e-05,
"loss": 4.0817,
"step": 204
},
{
"epoch": 0.34,
"learning_rate": 4.0600000000000004e-05,
"loss": 4.1543,
"step": 205
},
{
"epoch": 0.35,
"learning_rate": 4.08e-05,
"loss": 4.102,
"step": 206
},
{
"epoch": 0.35,
"learning_rate": 4.1e-05,
"loss": 4.1483,
"step": 207
},
{
"epoch": 0.35,
"learning_rate": 4.12e-05,
"loss": 4.1526,
"step": 208
},
{
"epoch": 0.35,
"learning_rate": 4.14e-05,
"loss": 4.175,
"step": 209
},
{
"epoch": 0.35,
"learning_rate": 4.16e-05,
"loss": 4.2503,
"step": 210
},
{
"epoch": 0.35,
"learning_rate": 4.18e-05,
"loss": 4.2906,
"step": 211
},
{
"epoch": 0.36,
"learning_rate": 4.2e-05,
"loss": 4.203,
"step": 212
},
{
"epoch": 0.36,
"learning_rate": 4.22e-05,
"loss": 4.1712,
"step": 213
},
{
"epoch": 0.36,
"learning_rate": 4.24e-05,
"loss": 4.2543,
"step": 214
},
{
"epoch": 0.36,
"learning_rate": 4.26e-05,
"loss": 4.1719,
"step": 215
},
{
"epoch": 0.36,
"learning_rate": 4.2800000000000004e-05,
"loss": 4.149,
"step": 216
},
{
"epoch": 0.36,
"learning_rate": 4.3e-05,
"loss": 4.0823,
"step": 217
},
{
"epoch": 0.37,
"learning_rate": 4.32e-05,
"loss": 4.1403,
"step": 218
},
{
"epoch": 0.37,
"learning_rate": 4.3400000000000005e-05,
"loss": 4.124,
"step": 219
},
{
"epoch": 0.37,
"learning_rate": 4.36e-05,
"loss": 4.2078,
"step": 220
},
{
"epoch": 0.37,
"learning_rate": 4.38e-05,
"loss": 4.2419,
"step": 221
},
{
"epoch": 0.37,
"learning_rate": 4.4000000000000006e-05,
"loss": 4.2563,
"step": 222
},
{
"epoch": 0.37,
"learning_rate": 4.4200000000000004e-05,
"loss": 4.1024,
"step": 223
},
{
"epoch": 0.38,
"learning_rate": 4.44e-05,
"loss": 4.2445,
"step": 224
},
{
"epoch": 0.38,
"learning_rate": 4.46e-05,
"loss": 4.3883,
"step": 225
},
{
"epoch": 0.38,
"learning_rate": 4.4800000000000005e-05,
"loss": 4.1431,
"step": 226
},
{
"epoch": 0.38,
"learning_rate": 4.5e-05,
"loss": 4.0658,
"step": 227
},
{
"epoch": 0.38,
"learning_rate": 4.52e-05,
"loss": 4.1397,
"step": 228
},
{
"epoch": 0.39,
"learning_rate": 4.5400000000000006e-05,
"loss": 4.2887,
"step": 229
},
{
"epoch": 0.39,
"learning_rate": 4.5600000000000004e-05,
"loss": 4.2302,
"step": 230
},
{
"epoch": 0.39,
"learning_rate": 4.58e-05,
"loss": 4.1176,
"step": 231
},
{
"epoch": 0.39,
"learning_rate": 4.600000000000001e-05,
"loss": 4.2021,
"step": 232
},
{
"epoch": 0.39,
"learning_rate": 4.6200000000000005e-05,
"loss": 4.2752,
"step": 233
},
{
"epoch": 0.39,
"learning_rate": 4.64e-05,
"loss": 4.1686,
"step": 234
},
{
"epoch": 0.4,
"learning_rate": 4.660000000000001e-05,
"loss": 4.3288,
"step": 235
},
{
"epoch": 0.4,
"learning_rate": 4.6800000000000006e-05,
"loss": 4.121,
"step": 236
},
{
"epoch": 0.4,
"learning_rate": 4.7e-05,
"loss": 4.0721,
"step": 237
},
{
"epoch": 0.4,
"learning_rate": 4.72e-05,
"loss": 4.1238,
"step": 238
},
{
"epoch": 0.4,
"learning_rate": 4.74e-05,
"loss": 4.0866,
"step": 239
},
{
"epoch": 0.4,
"learning_rate": 4.76e-05,
"loss": 4.2226,
"step": 240
},
{
"epoch": 0.41,
"learning_rate": 4.78e-05,
"loss": 4.1383,
"step": 241
},
{
"epoch": 0.41,
"learning_rate": 4.8e-05,
"loss": 4.1834,
"step": 242
},
{
"epoch": 0.41,
"learning_rate": 4.82e-05,
"loss": 4.1741,
"step": 243
},
{
"epoch": 0.41,
"learning_rate": 4.8400000000000004e-05,
"loss": 4.3256,
"step": 244
},
{
"epoch": 0.41,
"learning_rate": 4.86e-05,
"loss": 4.4218,
"step": 245
},
{
"epoch": 0.41,
"learning_rate": 4.88e-05,
"loss": 4.388,
"step": 246
},
{
"epoch": 0.42,
"learning_rate": 4.9e-05,
"loss": 4.4811,
"step": 247
},
{
"epoch": 0.42,
"learning_rate": 4.92e-05,
"loss": 4.4647,
"step": 248
},
{
"epoch": 0.42,
"learning_rate": 4.94e-05,
"loss": 4.3753,
"step": 249
},
{
"epoch": 0.42,
"learning_rate": 4.96e-05,
"loss": 4.4577,
"step": 250
},
{
"epoch": 0.42,
"learning_rate": 4.9800000000000004e-05,
"loss": 4.2267,
"step": 251
},
{
"epoch": 0.42,
"learning_rate": 5e-05,
"loss": 4.1948,
"step": 252
},
{
"epoch": 0.43,
"learning_rate": 5.02e-05,
"loss": 4.1944,
"step": 253
},
{
"epoch": 0.43,
"learning_rate": 5.0400000000000005e-05,
"loss": 4.1107,
"step": 254
},
{
"epoch": 0.43,
"learning_rate": 5.0600000000000003e-05,
"loss": 4.2676,
"step": 255
},
{
"epoch": 0.43,
"learning_rate": 5.08e-05,
"loss": 4.1532,
"step": 256
},
{
"epoch": 0.43,
"learning_rate": 5.1000000000000006e-05,
"loss": 4.1267,
"step": 257
},
{
"epoch": 0.43,
"learning_rate": 5.1200000000000004e-05,
"loss": 4.1397,
"step": 258
},
{
"epoch": 0.44,
"learning_rate": 5.14e-05,
"loss": 4.2355,
"step": 259
},
{
"epoch": 0.44,
"learning_rate": 5.16e-05,
"loss": 4.2747,
"step": 260
},
{
"epoch": 0.44,
"learning_rate": 5.1800000000000005e-05,
"loss": 4.1574,
"step": 261
},
{
"epoch": 0.44,
"learning_rate": 5.2000000000000004e-05,
"loss": 4.1059,
"step": 262
},
{
"epoch": 0.44,
"learning_rate": 5.22e-05,
"loss": 4.2589,
"step": 263
},
{
"epoch": 0.44,
"learning_rate": 5.2400000000000007e-05,
"loss": 3.9687,
"step": 264
},
{
"epoch": 0.45,
"learning_rate": 5.2600000000000005e-05,
"loss": 4.0608,
"step": 265
},
{
"epoch": 0.45,
"learning_rate": 5.28e-05,
"loss": 4.156,
"step": 266
},
{
"epoch": 0.45,
"learning_rate": 5.300000000000001e-05,
"loss": 4.1761,
"step": 267
},
{
"epoch": 0.45,
"learning_rate": 5.3200000000000006e-05,
"loss": 4.0868,
"step": 268
},
{
"epoch": 0.45,
"learning_rate": 5.3400000000000004e-05,
"loss": 4.2123,
"step": 269
},
{
"epoch": 0.45,
"learning_rate": 5.360000000000001e-05,
"loss": 4.2234,
"step": 270
},
{
"epoch": 0.46,
"learning_rate": 5.380000000000001e-05,
"loss": 4.1167,
"step": 271
},
{
"epoch": 0.46,
"learning_rate": 5.4000000000000005e-05,
"loss": 4.291,
"step": 272
},
{
"epoch": 0.46,
"learning_rate": 5.420000000000001e-05,
"loss": 4.2129,
"step": 273
},
{
"epoch": 0.46,
"learning_rate": 5.440000000000001e-05,
"loss": 4.1663,
"step": 274
},
{
"epoch": 0.46,
"learning_rate": 5.4600000000000006e-05,
"loss": 4.2445,
"step": 275
},
{
"epoch": 0.46,
"learning_rate": 5.4800000000000004e-05,
"loss": 4.1665,
"step": 276
},
{
"epoch": 0.47,
"learning_rate": 5.500000000000001e-05,
"loss": 4.2805,
"step": 277
},
{
"epoch": 0.47,
"learning_rate": 5.520000000000001e-05,
"loss": 4.1222,
"step": 278
},
{
"epoch": 0.47,
"learning_rate": 5.5400000000000005e-05,
"loss": 4.3026,
"step": 279
},
{
"epoch": 0.47,
"learning_rate": 5.560000000000001e-05,
"loss": 4.3612,
"step": 280
},
{
"epoch": 0.47,
"learning_rate": 5.580000000000001e-05,
"loss": 4.3457,
"step": 281
},
{
"epoch": 0.47,
"learning_rate": 5.6000000000000006e-05,
"loss": 4.2449,
"step": 282
},
{
"epoch": 0.48,
"learning_rate": 5.620000000000001e-05,
"loss": 4.1166,
"step": 283
},
{
"epoch": 0.48,
"learning_rate": 5.6399999999999995e-05,
"loss": 4.1394,
"step": 284
},
{
"epoch": 0.48,
"learning_rate": 5.66e-05,
"loss": 4.357,
"step": 285
},
{
"epoch": 0.48,
"learning_rate": 5.68e-05,
"loss": 4.1017,
"step": 286
},
{
"epoch": 0.48,
"learning_rate": 5.6999999999999996e-05,
"loss": 4.4652,
"step": 287
},
{
"epoch": 0.48,
"learning_rate": 5.72e-05,
"loss": 3.9479,
"step": 288
},
{
"epoch": 0.49,
"learning_rate": 5.74e-05,
"loss": 4.2049,
"step": 289
},
{
"epoch": 0.49,
"learning_rate": 5.76e-05,
"loss": 4.1827,
"step": 290
},
{
"epoch": 0.49,
"learning_rate": 5.7799999999999995e-05,
"loss": 4.3733,
"step": 291
},
{
"epoch": 0.49,
"learning_rate": 5.8e-05,
"loss": 4.0827,
"step": 292
},
{
"epoch": 0.49,
"learning_rate": 5.82e-05,
"loss": 4.3009,
"step": 293
},
{
"epoch": 0.49,
"learning_rate": 5.8399999999999997e-05,
"loss": 4.3702,
"step": 294
},
{
"epoch": 0.5,
"learning_rate": 5.86e-05,
"loss": 4.3088,
"step": 295
},
{
"epoch": 0.5,
"learning_rate": 5.88e-05,
"loss": 4.0949,
"step": 296
},
{
"epoch": 0.5,
"learning_rate": 5.9e-05,
"loss": 4.4279,
"step": 297
},
{
"epoch": 0.5,
"learning_rate": 5.92e-05,
"loss": 4.5189,
"step": 298
},
{
"epoch": 0.5,
"learning_rate": 5.94e-05,
"loss": 4.4628,
"step": 299
},
{
"epoch": 0.5,
"learning_rate": 5.96e-05,
"loss": 4.7137,
"step": 300
},
{
"epoch": 0.51,
"learning_rate": 5.9800000000000003e-05,
"loss": 4.3368,
"step": 301
},
{
"epoch": 0.51,
"learning_rate": 6e-05,
"loss": 4.231,
"step": 302
},
{
"epoch": 0.51,
"learning_rate": 6.02e-05,
"loss": 4.3327,
"step": 303
},
{
"epoch": 0.51,
"learning_rate": 6.04e-05,
"loss": 4.1077,
"step": 304
},
{
"epoch": 0.51,
"learning_rate": 6.06e-05,
"loss": 4.3066,
"step": 305
},
{
"epoch": 0.51,
"learning_rate": 6.08e-05,
"loss": 4.2898,
"step": 306
},
{
"epoch": 0.52,
"learning_rate": 6.1e-05,
"loss": 4.1,
"step": 307
},
{
"epoch": 0.52,
"learning_rate": 6.12e-05,
"loss": 4.2885,
"step": 308
},
{
"epoch": 0.52,
"learning_rate": 6.14e-05,
"loss": 4.2159,
"step": 309
},
{
"epoch": 0.52,
"learning_rate": 6.16e-05,
"loss": 4.2852,
"step": 310
},
{
"epoch": 0.52,
"learning_rate": 6.18e-05,
"loss": 4.1498,
"step": 311
},
{
"epoch": 0.52,
"learning_rate": 6.2e-05,
"loss": 4.1133,
"step": 312
},
{
"epoch": 0.53,
"learning_rate": 6.220000000000001e-05,
"loss": 4.196,
"step": 313
},
{
"epoch": 0.53,
"learning_rate": 6.24e-05,
"loss": 4.2193,
"step": 314
},
{
"epoch": 0.53,
"learning_rate": 6.26e-05,
"loss": 4.1943,
"step": 315
},
{
"epoch": 0.53,
"learning_rate": 6.280000000000001e-05,
"loss": 4.2541,
"step": 316
},
{
"epoch": 0.53,
"learning_rate": 6.3e-05,
"loss": 4.1273,
"step": 317
},
{
"epoch": 0.53,
"learning_rate": 6.32e-05,
"loss": 4.2596,
"step": 318
},
{
"epoch": 0.54,
"learning_rate": 6.340000000000001e-05,
"loss": 4.1504,
"step": 319
},
{
"epoch": 0.54,
"learning_rate": 6.36e-05,
"loss": 4.2921,
"step": 320
},
{
"epoch": 0.54,
"learning_rate": 6.38e-05,
"loss": 4.0708,
"step": 321
},
{
"epoch": 0.54,
"learning_rate": 6.400000000000001e-05,
"loss": 4.1126,
"step": 322
},
{
"epoch": 0.54,
"learning_rate": 6.42e-05,
"loss": 4.274,
"step": 323
},
{
"epoch": 0.54,
"learning_rate": 6.440000000000001e-05,
"loss": 4.2403,
"step": 324
},
{
"epoch": 0.55,
"learning_rate": 6.460000000000001e-05,
"loss": 4.1817,
"step": 325
},
{
"epoch": 0.55,
"learning_rate": 6.48e-05,
"loss": 4.1604,
"step": 326
},
{
"epoch": 0.55,
"learning_rate": 6.500000000000001e-05,
"loss": 4.413,
"step": 327
},
{
"epoch": 0.55,
"learning_rate": 6.52e-05,
"loss": 4.1049,
"step": 328
},
{
"epoch": 0.55,
"learning_rate": 6.54e-05,
"loss": 4.0653,
"step": 329
},
{
"epoch": 0.55,
"learning_rate": 6.560000000000001e-05,
"loss": 4.2674,
"step": 330
},
{
"epoch": 0.56,
"learning_rate": 6.58e-05,
"loss": 4.2443,
"step": 331
},
{
"epoch": 0.56,
"learning_rate": 6.6e-05,
"loss": 4.035,
"step": 332
},
{
"epoch": 0.56,
"learning_rate": 6.620000000000001e-05,
"loss": 4.2306,
"step": 333
},
{
"epoch": 0.56,
"learning_rate": 6.64e-05,
"loss": 4.2563,
"step": 334
},
{
"epoch": 0.56,
"learning_rate": 6.66e-05,
"loss": 4.2434,
"step": 335
},
{
"epoch": 0.56,
"learning_rate": 6.680000000000001e-05,
"loss": 4.1627,
"step": 336
},
{
"epoch": 0.57,
"learning_rate": 6.7e-05,
"loss": 4.1118,
"step": 337
},
{
"epoch": 0.57,
"learning_rate": 6.720000000000001e-05,
"loss": 4.2157,
"step": 338
},
{
"epoch": 0.57,
"learning_rate": 6.740000000000001e-05,
"loss": 4.1961,
"step": 339
},
{
"epoch": 0.57,
"learning_rate": 6.76e-05,
"loss": 4.2993,
"step": 340
},
{
"epoch": 0.57,
"learning_rate": 6.780000000000001e-05,
"loss": 4.2616,
"step": 341
},
{
"epoch": 0.58,
"learning_rate": 6.800000000000001e-05,
"loss": 4.3949,
"step": 342
},
{
"epoch": 0.58,
"learning_rate": 6.82e-05,
"loss": 4.4184,
"step": 343
},
{
"epoch": 0.58,
"learning_rate": 6.840000000000001e-05,
"loss": 4.2724,
"step": 344
},
{
"epoch": 0.58,
"learning_rate": 6.860000000000001e-05,
"loss": 4.3862,
"step": 345
},
{
"epoch": 0.58,
"learning_rate": 6.879999999999999e-05,
"loss": 4.0938,
"step": 346
},
{
"epoch": 0.58,
"learning_rate": 6.9e-05,
"loss": 4.2047,
"step": 347
},
{
"epoch": 0.59,
"learning_rate": 6.92e-05,
"loss": 4.5291,
"step": 348
},
{
"epoch": 0.59,
"learning_rate": 6.939999999999999e-05,
"loss": 4.4491,
"step": 349
},
{
"epoch": 0.59,
"learning_rate": 6.96e-05,
"loss": 4.6159,
"step": 350
},
{
"epoch": 0.59,
"learning_rate": 6.98e-05,
"loss": 4.3333,
"step": 351
},
{
"epoch": 0.59,
"learning_rate": 7e-05,
"loss": 4.2327,
"step": 352
},
{
"epoch": 0.59,
"learning_rate": 7.02e-05,
"loss": 4.3054,
"step": 353
},
{
"epoch": 0.6,
"learning_rate": 7.04e-05,
"loss": 4.1521,
"step": 354
},
{
"epoch": 0.6,
"learning_rate": 7.06e-05,
"loss": 4.2491,
"step": 355
},
{
"epoch": 0.6,
"learning_rate": 7.08e-05,
"loss": 4.21,
"step": 356
},
{
"epoch": 0.6,
"learning_rate": 7.1e-05,
"loss": 4.2289,
"step": 357
},
{
"epoch": 0.6,
"learning_rate": 7.12e-05,
"loss": 4.2476,
"step": 358
},
{
"epoch": 0.6,
"learning_rate": 7.14e-05,
"loss": 4.1769,
"step": 359
},
{
"epoch": 0.61,
"learning_rate": 7.16e-05,
"loss": 4.226,
"step": 360
},
{
"epoch": 0.61,
"learning_rate": 7.18e-05,
"loss": 4.229,
"step": 361
},
{
"epoch": 0.61,
"learning_rate": 7.2e-05,
"loss": 4.1832,
"step": 362
},
{
"epoch": 0.61,
"learning_rate": 7.22e-05,
"loss": 4.339,
"step": 363
},
{
"epoch": 0.61,
"learning_rate": 7.24e-05,
"loss": 4.1442,
"step": 364
},
{
"epoch": 0.61,
"learning_rate": 7.26e-05,
"loss": 4.2575,
"step": 365
},
{
"epoch": 0.62,
"learning_rate": 7.280000000000001e-05,
"loss": 4.1256,
"step": 366
},
{
"epoch": 0.62,
"learning_rate": 7.3e-05,
"loss": 4.1973,
"step": 367
},
{
"epoch": 0.62,
"learning_rate": 7.32e-05,
"loss": 4.1011,
"step": 368
},
{
"epoch": 0.62,
"learning_rate": 7.340000000000001e-05,
"loss": 4.2649,
"step": 369
},
{
"epoch": 0.62,
"learning_rate": 7.36e-05,
"loss": 4.2652,
"step": 370
},
{
"epoch": 0.62,
"learning_rate": 7.38e-05,
"loss": 4.1545,
"step": 371
},
{
"epoch": 0.63,
"learning_rate": 7.4e-05,
"loss": 4.1874,
"step": 372
},
{
"epoch": 0.63,
"learning_rate": 7.42e-05,
"loss": 4.0624,
"step": 373
},
{
"epoch": 0.63,
"learning_rate": 7.44e-05,
"loss": 4.1635,
"step": 374
},
{
"epoch": 0.63,
"learning_rate": 7.46e-05,
"loss": 4.2729,
"step": 375
},
{
"epoch": 0.63,
"learning_rate": 7.48e-05,
"loss": 4.3062,
"step": 376
},
{
"epoch": 0.63,
"learning_rate": 7.500000000000001e-05,
"loss": 4.2091,
"step": 377
},
{
"epoch": 0.64,
"learning_rate": 7.52e-05,
"loss": 4.2752,
"step": 378
},
{
"epoch": 0.64,
"learning_rate": 7.54e-05,
"loss": 4.0568,
"step": 379
},
{
"epoch": 0.64,
"learning_rate": 7.560000000000001e-05,
"loss": 4.3214,
"step": 380
},
{
"epoch": 0.64,
"learning_rate": 7.58e-05,
"loss": 4.3059,
"step": 381
},
{
"epoch": 0.64,
"learning_rate": 7.6e-05,
"loss": 4.1338,
"step": 382
},
{
"epoch": 0.64,
"learning_rate": 7.620000000000001e-05,
"loss": 4.1773,
"step": 383
},
{
"epoch": 0.65,
"learning_rate": 7.64e-05,
"loss": 4.2126,
"step": 384
},
{
"epoch": 0.65,
"learning_rate": 7.66e-05,
"loss": 4.4449,
"step": 385
},
{
"epoch": 0.65,
"learning_rate": 7.680000000000001e-05,
"loss": 4.2583,
"step": 386
},
{
"epoch": 0.65,
"learning_rate": 7.7e-05,
"loss": 4.4383,
"step": 387
},
{
"epoch": 0.65,
"learning_rate": 7.72e-05,
"loss": 4.2463,
"step": 388
},
{
"epoch": 0.65,
"learning_rate": 7.740000000000001e-05,
"loss": 4.2243,
"step": 389
},
{
"epoch": 0.66,
"learning_rate": 7.76e-05,
"loss": 4.2311,
"step": 390
},
{
"epoch": 0.66,
"learning_rate": 7.780000000000001e-05,
"loss": 4.3077,
"step": 391
},
{
"epoch": 0.66,
"learning_rate": 7.800000000000001e-05,
"loss": 4.3871,
"step": 392
},
{
"epoch": 0.66,
"learning_rate": 7.82e-05,
"loss": 4.4987,
"step": 393
},
{
"epoch": 0.66,
"learning_rate": 7.840000000000001e-05,
"loss": 4.2839,
"step": 394
},
{
"epoch": 0.66,
"learning_rate": 7.860000000000001e-05,
"loss": 4.3919,
"step": 395
},
{
"epoch": 0.67,
"learning_rate": 7.88e-05,
"loss": 4.3298,
"step": 396
},
{
"epoch": 0.67,
"learning_rate": 7.900000000000001e-05,
"loss": 4.3474,
"step": 397
},
{
"epoch": 0.67,
"learning_rate": 7.920000000000001e-05,
"loss": 4.7022,
"step": 398
},
{
"epoch": 0.67,
"learning_rate": 7.94e-05,
"loss": 4.4599,
"step": 399
},
{
"epoch": 0.67,
"learning_rate": 7.960000000000001e-05,
"loss": 4.8706,
"step": 400
},
{
"epoch": 0.67,
"learning_rate": 7.98e-05,
"loss": 4.36,
"step": 401
},
{
"epoch": 0.68,
"learning_rate": 8e-05,
"loss": 4.2806,
"step": 402
},
{
"epoch": 0.68,
"learning_rate": 8.020000000000001e-05,
"loss": 4.3386,
"step": 403
},
{
"epoch": 0.68,
"learning_rate": 8.04e-05,
"loss": 4.2763,
"step": 404
},
{
"epoch": 0.68,
"learning_rate": 8.060000000000001e-05,
"loss": 4.2155,
"step": 405
},
{
"epoch": 0.68,
"learning_rate": 8.080000000000001e-05,
"loss": 4.2131,
"step": 406
},
{
"epoch": 0.68,
"learning_rate": 8.1e-05,
"loss": 4.4031,
"step": 407
},
{
"epoch": 0.69,
"learning_rate": 8.120000000000001e-05,
"loss": 4.0873,
"step": 408
},
{
"epoch": 0.69,
"learning_rate": 8.14e-05,
"loss": 4.1883,
"step": 409
},
{
"epoch": 0.69,
"learning_rate": 8.16e-05,
"loss": 4.1759,
"step": 410
},
{
"epoch": 0.69,
"learning_rate": 8.18e-05,
"loss": 4.1658,
"step": 411
},
{
"epoch": 0.69,
"learning_rate": 8.2e-05,
"loss": 4.3362,
"step": 412
},
{
"epoch": 0.69,
"learning_rate": 8.22e-05,
"loss": 4.2107,
"step": 413
},
{
"epoch": 0.7,
"learning_rate": 8.24e-05,
"loss": 4.3678,
"step": 414
},
{
"epoch": 0.7,
"learning_rate": 8.26e-05,
"loss": 4.235,
"step": 415
},
{
"epoch": 0.7,
"learning_rate": 8.28e-05,
"loss": 4.1738,
"step": 416
},
{
"epoch": 0.7,
"learning_rate": 8.3e-05,
"loss": 4.2348,
"step": 417
},
{
"epoch": 0.7,
"learning_rate": 8.32e-05,
"loss": 3.9847,
"step": 418
},
{
"epoch": 0.7,
"learning_rate": 8.34e-05,
"loss": 4.3757,
"step": 419
},
{
"epoch": 0.71,
"learning_rate": 8.36e-05,
"loss": 4.2326,
"step": 420
},
{
"epoch": 0.71,
"learning_rate": 8.38e-05,
"loss": 3.9055,
"step": 421
},
{
"epoch": 0.71,
"learning_rate": 8.4e-05,
"loss": 4.1417,
"step": 422
},
{
"epoch": 0.71,
"learning_rate": 8.42e-05,
"loss": 4.2901,
"step": 423
},
{
"epoch": 0.71,
"learning_rate": 8.44e-05,
"loss": 4.2111,
"step": 424
},
{
"epoch": 0.71,
"learning_rate": 8.46e-05,
"loss": 4.2131,
"step": 425
},
{
"epoch": 0.72,
"learning_rate": 8.48e-05,
"loss": 4.3113,
"step": 426
},
{
"epoch": 0.72,
"learning_rate": 8.5e-05,
"loss": 4.1273,
"step": 427
},
{
"epoch": 0.72,
"learning_rate": 8.52e-05,
"loss": 4.0975,
"step": 428
},
{
"epoch": 0.72,
"learning_rate": 8.54e-05,
"loss": 4.2873,
"step": 429
},
{
"epoch": 0.72,
"learning_rate": 8.560000000000001e-05,
"loss": 4.2203,
"step": 430
},
{
"epoch": 0.72,
"learning_rate": 8.58e-05,
"loss": 4.3555,
"step": 431
},
{
"epoch": 0.73,
"learning_rate": 8.6e-05,
"loss": 4.0412,
"step": 432
},
{
"epoch": 0.73,
"learning_rate": 8.620000000000001e-05,
"loss": 4.2636,
"step": 433
},
{
"epoch": 0.73,
"learning_rate": 8.64e-05,
"loss": 4.3913,
"step": 434
},
{
"epoch": 0.73,
"learning_rate": 8.66e-05,
"loss": 4.334,
"step": 435
},
{
"epoch": 0.73,
"learning_rate": 8.680000000000001e-05,
"loss": 4.3023,
"step": 436
},
{
"epoch": 0.73,
"learning_rate": 8.7e-05,
"loss": 4.1347,
"step": 437
},
{
"epoch": 0.74,
"learning_rate": 8.72e-05,
"loss": 4.2697,
"step": 438
},
{
"epoch": 0.74,
"learning_rate": 8.740000000000001e-05,
"loss": 4.2604,
"step": 439
},
{
"epoch": 0.74,
"learning_rate": 8.76e-05,
"loss": 4.3875,
"step": 440
},
{
"epoch": 0.74,
"learning_rate": 8.78e-05,
"loss": 4.3363,
"step": 441
},
{
"epoch": 0.74,
"learning_rate": 8.800000000000001e-05,
"loss": 4.6421,
"step": 442
},
{
"epoch": 0.74,
"learning_rate": 8.82e-05,
"loss": 4.4132,
"step": 443
},
{
"epoch": 0.75,
"learning_rate": 8.840000000000001e-05,
"loss": 4.3105,
"step": 444
},
{
"epoch": 0.75,
"learning_rate": 8.86e-05,
"loss": 4.1261,
"step": 445
},
{
"epoch": 0.75,
"learning_rate": 8.88e-05,
"loss": 4.3448,
"step": 446
},
{
"epoch": 0.75,
"learning_rate": 8.900000000000001e-05,
"loss": 4.3906,
"step": 447
},
{
"epoch": 0.75,
"learning_rate": 8.92e-05,
"loss": 4.3253,
"step": 448
},
{
"epoch": 0.75,
"learning_rate": 8.94e-05,
"loss": 4.4832,
"step": 449
},
{
"epoch": 0.76,
"learning_rate": 8.960000000000001e-05,
"loss": 4.2745,
"step": 450
},
{
"epoch": 0.76,
"learning_rate": 8.98e-05,
"loss": 4.2774,
"step": 451
},
{
"epoch": 0.76,
"learning_rate": 9e-05,
"loss": 4.2983,
"step": 452
},
{
"epoch": 0.76,
"learning_rate": 9.020000000000001e-05,
"loss": 4.1945,
"step": 453
},
{
"epoch": 0.76,
"learning_rate": 9.04e-05,
"loss": 4.182,
"step": 454
},
{
"epoch": 0.77,
"learning_rate": 9.06e-05,
"loss": 4.3015,
"step": 455
},
{
"epoch": 0.77,
"learning_rate": 9.080000000000001e-05,
"loss": 4.216,
"step": 456
},
{
"epoch": 0.77,
"learning_rate": 9.1e-05,
"loss": 4.4956,
"step": 457
},
{
"epoch": 0.77,
"learning_rate": 9.120000000000001e-05,
"loss": 4.4189,
"step": 458
},
{
"epoch": 0.77,
"learning_rate": 9.140000000000001e-05,
"loss": 4.3564,
"step": 459
},
{
"epoch": 0.77,
"learning_rate": 9.16e-05,
"loss": 4.3126,
"step": 460
},
{
"epoch": 0.78,
"learning_rate": 9.180000000000001e-05,
"loss": 4.3284,
"step": 461
},
{
"epoch": 0.78,
"learning_rate": 9.200000000000001e-05,
"loss": 4.1673,
"step": 462
},
{
"epoch": 0.78,
"learning_rate": 9.22e-05,
"loss": 4.1379,
"step": 463
},
{
"epoch": 0.78,
"learning_rate": 9.240000000000001e-05,
"loss": 4.3931,
"step": 464
},
{
"epoch": 0.78,
"learning_rate": 9.260000000000001e-05,
"loss": 4.3984,
"step": 465
},
{
"epoch": 0.78,
"learning_rate": 9.28e-05,
"loss": 4.2236,
"step": 466
},
{
"epoch": 0.79,
"learning_rate": 9.300000000000001e-05,
"loss": 4.2489,
"step": 467
},
{
"epoch": 0.79,
"learning_rate": 9.320000000000002e-05,
"loss": 4.188,
"step": 468
},
{
"epoch": 0.79,
"learning_rate": 9.340000000000001e-05,
"loss": 4.2623,
"step": 469
},
{
"epoch": 0.79,
"learning_rate": 9.360000000000001e-05,
"loss": 4.1637,
"step": 470
},
{
"epoch": 0.79,
"learning_rate": 9.38e-05,
"loss": 4.1768,
"step": 471
},
{
"epoch": 0.79,
"learning_rate": 9.4e-05,
"loss": 4.2127,
"step": 472
},
{
"epoch": 0.8,
"learning_rate": 9.42e-05,
"loss": 4.3013,
"step": 473
},
{
"epoch": 0.8,
"learning_rate": 9.44e-05,
"loss": 4.111,
"step": 474
},
{
"epoch": 0.8,
"learning_rate": 9.46e-05,
"loss": 4.1387,
"step": 475
},
{
"epoch": 0.8,
"learning_rate": 9.48e-05,
"loss": 4.2175,
"step": 476
},
{
"epoch": 0.8,
"learning_rate": 9.5e-05,
"loss": 4.237,
"step": 477
},
{
"epoch": 0.8,
"learning_rate": 9.52e-05,
"loss": 4.2458,
"step": 478
},
{
"epoch": 0.81,
"learning_rate": 9.54e-05,
"loss": 4.1149,
"step": 479
},
{
"epoch": 0.81,
"learning_rate": 9.56e-05,
"loss": 4.067,
"step": 480
},
{
"epoch": 0.81,
"learning_rate": 9.58e-05,
"loss": 4.4475,
"step": 481
},
{
"epoch": 0.81,
"learning_rate": 9.6e-05,
"loss": 4.3741,
"step": 482
},
{
"epoch": 0.81,
"learning_rate": 9.620000000000001e-05,
"loss": 4.2896,
"step": 483
},
{
"epoch": 0.81,
"learning_rate": 9.64e-05,
"loss": 4.3246,
"step": 484
},
{
"epoch": 0.82,
"learning_rate": 9.66e-05,
"loss": 4.2455,
"step": 485
},
{
"epoch": 0.82,
"learning_rate": 9.680000000000001e-05,
"loss": 4.1892,
"step": 486
},
{
"epoch": 0.82,
"learning_rate": 9.7e-05,
"loss": 4.2892,
"step": 487
},
{
"epoch": 0.82,
"learning_rate": 9.72e-05,
"loss": 4.2748,
"step": 488
},
{
"epoch": 0.82,
"learning_rate": 9.74e-05,
"loss": 4.2966,
"step": 489
},
{
"epoch": 0.82,
"learning_rate": 9.76e-05,
"loss": 4.1103,
"step": 490
},
{
"epoch": 0.83,
"learning_rate": 9.78e-05,
"loss": 4.5372,
"step": 491
},
{
"epoch": 0.83,
"learning_rate": 9.8e-05,
"loss": 4.2285,
"step": 492
},
{
"epoch": 0.83,
"learning_rate": 9.82e-05,
"loss": 3.9966,
"step": 493
},
{
"epoch": 0.83,
"learning_rate": 9.84e-05,
"loss": 4.3058,
"step": 494
},
{
"epoch": 0.83,
"learning_rate": 9.86e-05,
"loss": 4.5724,
"step": 495
},
{
"epoch": 0.83,
"learning_rate": 9.88e-05,
"loss": 4.3851,
"step": 496
},
{
"epoch": 0.84,
"learning_rate": 9.900000000000001e-05,
"loss": 4.3653,
"step": 497
},
{
"epoch": 0.84,
"learning_rate": 9.92e-05,
"loss": 4.3187,
"step": 498
},
{
"epoch": 0.84,
"learning_rate": 9.94e-05,
"loss": 4.5631,
"step": 499
},
{
"epoch": 0.84,
"learning_rate": 9.960000000000001e-05,
"loss": 4.346,
"step": 500
},
{
"epoch": 0.84,
"eval_loss": 4.40821647644043,
"eval_runtime": 786.867,
"eval_samples_per_second": 3.358,
"eval_steps_per_second": 0.281,
"eval_wer": 1.907457358191194,
"step": 500
},
{
"epoch": 0.84,
"learning_rate": 9.98e-05,
"loss": 4.4984,
"step": 501
},
{
"epoch": 0.84,
"learning_rate": 0.0001,
"loss": 4.1902,
"step": 502
},
{
"epoch": 0.85,
"learning_rate": 9.893617021276596e-05,
"loss": 4.2023,
"step": 503
},
{
"epoch": 0.85,
"learning_rate": 9.787234042553192e-05,
"loss": 4.3951,
"step": 504
},
{
"epoch": 0.85,
"learning_rate": 9.680851063829788e-05,
"loss": 4.2746,
"step": 505
},
{
"epoch": 0.85,
"learning_rate": 9.574468085106384e-05,
"loss": 4.3589,
"step": 506
},
{
"epoch": 0.85,
"learning_rate": 9.468085106382978e-05,
"loss": 4.2035,
"step": 507
},
{
"epoch": 0.85,
"learning_rate": 9.361702127659576e-05,
"loss": 4.2748,
"step": 508
},
{
"epoch": 0.86,
"learning_rate": 9.25531914893617e-05,
"loss": 4.1227,
"step": 509
},
{
"epoch": 0.86,
"learning_rate": 9.148936170212766e-05,
"loss": 4.2689,
"step": 510
},
{
"epoch": 0.86,
"learning_rate": 9.042553191489363e-05,
"loss": 4.3483,
"step": 511
},
{
"epoch": 0.86,
"learning_rate": 8.936170212765958e-05,
"loss": 4.1754,
"step": 512
},
{
"epoch": 0.86,
"learning_rate": 8.829787234042553e-05,
"loss": 4.1887,
"step": 513
},
{
"epoch": 0.86,
"learning_rate": 8.723404255319149e-05,
"loss": 4.3354,
"step": 514
},
{
"epoch": 0.87,
"learning_rate": 8.617021276595745e-05,
"loss": 4.241,
"step": 515
},
{
"epoch": 0.87,
"learning_rate": 8.510638297872341e-05,
"loss": 4.2799,
"step": 516
},
{
"epoch": 0.87,
"learning_rate": 8.404255319148937e-05,
"loss": 4.3408,
"step": 517
},
{
"epoch": 0.87,
"learning_rate": 8.297872340425533e-05,
"loss": 4.3558,
"step": 518
},
{
"epoch": 0.87,
"learning_rate": 8.191489361702128e-05,
"loss": 4.2045,
"step": 519
},
{
"epoch": 0.87,
"learning_rate": 8.085106382978723e-05,
"loss": 4.1557,
"step": 520
},
{
"epoch": 0.88,
"learning_rate": 7.978723404255319e-05,
"loss": 4.2368,
"step": 521
},
{
"epoch": 0.88,
"learning_rate": 7.872340425531916e-05,
"loss": 4.3475,
"step": 522
},
{
"epoch": 0.88,
"learning_rate": 7.76595744680851e-05,
"loss": 4.1364,
"step": 523
},
{
"epoch": 0.88,
"learning_rate": 7.659574468085106e-05,
"loss": 3.9837,
"step": 524
},
{
"epoch": 0.88,
"learning_rate": 7.553191489361703e-05,
"loss": 4.0987,
"step": 525
},
{
"epoch": 0.88,
"learning_rate": 7.446808510638298e-05,
"loss": 4.2456,
"step": 526
},
{
"epoch": 0.89,
"learning_rate": 7.340425531914894e-05,
"loss": 4.3864,
"step": 527
},
{
"epoch": 0.89,
"learning_rate": 7.23404255319149e-05,
"loss": 4.148,
"step": 528
},
{
"epoch": 0.89,
"learning_rate": 7.127659574468085e-05,
"loss": 4.0937,
"step": 529
},
{
"epoch": 0.89,
"learning_rate": 7.021276595744681e-05,
"loss": 4.0465,
"step": 530
},
{
"epoch": 0.89,
"learning_rate": 6.914893617021277e-05,
"loss": 4.0592,
"step": 531
},
{
"epoch": 0.89,
"learning_rate": 6.808510638297873e-05,
"loss": 4.3688,
"step": 532
},
{
"epoch": 0.9,
"learning_rate": 6.702127659574469e-05,
"loss": 4.1947,
"step": 533
},
{
"epoch": 0.9,
"learning_rate": 6.595744680851063e-05,
"loss": 4.1332,
"step": 534
},
{
"epoch": 0.9,
"learning_rate": 6.489361702127659e-05,
"loss": 4.1444,
"step": 535
},
{
"epoch": 0.9,
"learning_rate": 6.382978723404256e-05,
"loss": 4.1606,
"step": 536
},
{
"epoch": 0.9,
"learning_rate": 6.276595744680851e-05,
"loss": 4.2486,
"step": 537
},
{
"epoch": 0.9,
"learning_rate": 6.170212765957447e-05,
"loss": 4.1731,
"step": 538
},
{
"epoch": 0.91,
"learning_rate": 6.063829787234043e-05,
"loss": 4.1163,
"step": 539
},
{
"epoch": 0.91,
"learning_rate": 5.9574468085106384e-05,
"loss": 4.1554,
"step": 540
},
{
"epoch": 0.91,
"learning_rate": 5.851063829787234e-05,
"loss": 4.1198,
"step": 541
},
{
"epoch": 0.91,
"learning_rate": 5.744680851063831e-05,
"loss": 4.2323,
"step": 542
},
{
"epoch": 0.91,
"learning_rate": 5.638297872340426e-05,
"loss": 4.2438,
"step": 543
},
{
"epoch": 0.91,
"learning_rate": 5.531914893617022e-05,
"loss": 4.1804,
"step": 544
},
{
"epoch": 0.92,
"learning_rate": 5.425531914893617e-05,
"loss": 4.2424,
"step": 545
},
{
"epoch": 0.92,
"learning_rate": 5.319148936170213e-05,
"loss": 4.4403,
"step": 546
},
{
"epoch": 0.92,
"learning_rate": 5.212765957446809e-05,
"loss": 4.1733,
"step": 547
},
{
"epoch": 0.92,
"learning_rate": 5.1063829787234044e-05,
"loss": 4.382,
"step": 548
},
{
"epoch": 0.92,
"learning_rate": 5e-05,
"loss": 4.7592,
"step": 549
},
{
"epoch": 0.92,
"learning_rate": 4.893617021276596e-05,
"loss": 4.2295,
"step": 550
},
{
"epoch": 0.93,
"learning_rate": 4.787234042553192e-05,
"loss": 4.2583,
"step": 551
},
{
"epoch": 0.93,
"learning_rate": 4.680851063829788e-05,
"loss": 4.2183,
"step": 552
},
{
"epoch": 0.93,
"learning_rate": 4.574468085106383e-05,
"loss": 4.2371,
"step": 553
},
{
"epoch": 0.93,
"learning_rate": 4.468085106382979e-05,
"loss": 4.0849,
"step": 554
},
{
"epoch": 0.93,
"learning_rate": 4.3617021276595746e-05,
"loss": 4.1186,
"step": 555
},
{
"epoch": 0.93,
"learning_rate": 4.2553191489361704e-05,
"loss": 4.1655,
"step": 556
},
{
"epoch": 0.94,
"learning_rate": 4.148936170212766e-05,
"loss": 4.2651,
"step": 557
},
{
"epoch": 0.94,
"learning_rate": 4.0425531914893614e-05,
"loss": 4.2168,
"step": 558
},
{
"epoch": 0.94,
"learning_rate": 3.936170212765958e-05,
"loss": 4.0934,
"step": 559
},
{
"epoch": 0.94,
"learning_rate": 3.829787234042553e-05,
"loss": 4.0671,
"step": 560
},
{
"epoch": 0.94,
"learning_rate": 3.723404255319149e-05,
"loss": 4.4043,
"step": 561
},
{
"epoch": 0.94,
"learning_rate": 3.617021276595745e-05,
"loss": 4.1243,
"step": 562
},
{
"epoch": 0.95,
"learning_rate": 3.5106382978723407e-05,
"loss": 4.0769,
"step": 563
},
{
"epoch": 0.95,
"learning_rate": 3.4042553191489365e-05,
"loss": 4.3662,
"step": 564
},
{
"epoch": 0.95,
"learning_rate": 3.2978723404255317e-05,
"loss": 4.1857,
"step": 565
},
{
"epoch": 0.95,
"learning_rate": 3.191489361702128e-05,
"loss": 4.2426,
"step": 566
},
{
"epoch": 0.95,
"learning_rate": 3.085106382978723e-05,
"loss": 4.2188,
"step": 567
},
{
"epoch": 0.96,
"learning_rate": 2.9787234042553192e-05,
"loss": 4.103,
"step": 568
},
{
"epoch": 0.96,
"learning_rate": 2.8723404255319154e-05,
"loss": 4.1005,
"step": 569
},
{
"epoch": 0.96,
"learning_rate": 2.765957446808511e-05,
"loss": 4.1211,
"step": 570
},
{
"epoch": 0.96,
"learning_rate": 2.6595744680851064e-05,
"loss": 4.3098,
"step": 571
},
{
"epoch": 0.96,
"learning_rate": 2.5531914893617022e-05,
"loss": 4.2227,
"step": 572
},
{
"epoch": 0.96,
"learning_rate": 2.446808510638298e-05,
"loss": 4.2263,
"step": 573
},
{
"epoch": 0.97,
"learning_rate": 2.340425531914894e-05,
"loss": 4.1398,
"step": 574
},
{
"epoch": 0.97,
"learning_rate": 2.2340425531914894e-05,
"loss": 4.2215,
"step": 575
},
{
"epoch": 0.97,
"learning_rate": 2.1276595744680852e-05,
"loss": 4.1775,
"step": 576
},
{
"epoch": 0.97,
"learning_rate": 2.0212765957446807e-05,
"loss": 4.2561,
"step": 577
},
{
"epoch": 0.97,
"learning_rate": 1.9148936170212766e-05,
"loss": 4.0031,
"step": 578
},
{
"epoch": 0.97,
"learning_rate": 1.8085106382978724e-05,
"loss": 3.9507,
"step": 579
},
{
"epoch": 0.98,
"learning_rate": 1.7021276595744682e-05,
"loss": 4.2023,
"step": 580
},
{
"epoch": 0.98,
"learning_rate": 1.595744680851064e-05,
"loss": 4.3914,
"step": 581
},
{
"epoch": 0.98,
"learning_rate": 1.4893617021276596e-05,
"loss": 4.1952,
"step": 582
},
{
"epoch": 0.98,
"learning_rate": 1.3829787234042554e-05,
"loss": 3.9833,
"step": 583
},
{
"epoch": 0.98,
"learning_rate": 1.2765957446808511e-05,
"loss": 4.1362,
"step": 584
},
{
"epoch": 0.98,
"learning_rate": 1.170212765957447e-05,
"loss": 4.1881,
"step": 585
},
{
"epoch": 0.99,
"learning_rate": 1.0638297872340426e-05,
"loss": 4.0666,
"step": 586
},
{
"epoch": 0.99,
"learning_rate": 9.574468085106383e-06,
"loss": 4.2112,
"step": 587
},
{
"epoch": 0.99,
"learning_rate": 8.510638297872341e-06,
"loss": 4.2646,
"step": 588
},
{
"epoch": 0.99,
"learning_rate": 7.446808510638298e-06,
"loss": 4.2727,
"step": 589
},
{
"epoch": 0.99,
"learning_rate": 6.3829787234042555e-06,
"loss": 4.1834,
"step": 590
},
{
"epoch": 0.99,
"learning_rate": 5.319148936170213e-06,
"loss": 4.3415,
"step": 591
},
{
"epoch": 1.0,
"learning_rate": 4.255319148936171e-06,
"loss": 4.4151,
"step": 592
},
{
"epoch": 1.0,
"learning_rate": 3.1914893617021277e-06,
"loss": 4.4448,
"step": 593
},
{
"epoch": 1.0,
"learning_rate": 2.1276595744680853e-06,
"loss": 4.2767,
"step": 594
},
{
"epoch": 1.0,
"step": 594,
"total_flos": 0.0,
"train_loss": 4.277455122382553,
"train_runtime": 5668.958,
"train_samples_per_second": 5.034,
"train_steps_per_second": 0.105
}
],
"max_steps": 594,
"num_train_epochs": 1,
"total_flos": 0.0,
"trial_name": null,
"trial_params": null
}