wav2vec2-gpt2-wandb-grid-search / trainer_state.json
sanchit-gandhi's picture
End of training
eeafcdb
raw history blame
No virus
67.6 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": 4.825,
"step": 1
},
{
"epoch": 0.0,
"learning_rate": 6.000000000000001e-08,
"loss": 4.9943,
"step": 2
},
{
"epoch": 0.01,
"learning_rate": 1.2000000000000002e-07,
"loss": 4.8848,
"step": 3
},
{
"epoch": 0.01,
"learning_rate": 1.8e-07,
"loss": 4.8356,
"step": 4
},
{
"epoch": 0.01,
"learning_rate": 2.4000000000000003e-07,
"loss": 4.786,
"step": 5
},
{
"epoch": 0.01,
"learning_rate": 3.0000000000000004e-07,
"loss": 4.8026,
"step": 6
},
{
"epoch": 0.01,
"learning_rate": 3.6e-07,
"loss": 4.8925,
"step": 7
},
{
"epoch": 0.01,
"learning_rate": 4.2e-07,
"loss": 4.7958,
"step": 8
},
{
"epoch": 0.02,
"learning_rate": 4.800000000000001e-07,
"loss": 4.9053,
"step": 9
},
{
"epoch": 0.02,
"learning_rate": 5.4e-07,
"loss": 4.825,
"step": 10
},
{
"epoch": 0.02,
"learning_rate": 6.000000000000001e-07,
"loss": 4.8168,
"step": 11
},
{
"epoch": 0.02,
"learning_rate": 6.599999999999999e-07,
"loss": 4.8753,
"step": 12
},
{
"epoch": 0.02,
"learning_rate": 7.2e-07,
"loss": 4.7954,
"step": 13
},
{
"epoch": 0.02,
"learning_rate": 7.799999999999999e-07,
"loss": 4.7457,
"step": 14
},
{
"epoch": 0.03,
"learning_rate": 8.4e-07,
"loss": 4.6363,
"step": 15
},
{
"epoch": 0.03,
"learning_rate": 9e-07,
"loss": 4.8226,
"step": 16
},
{
"epoch": 0.03,
"learning_rate": 9.600000000000001e-07,
"loss": 4.8358,
"step": 17
},
{
"epoch": 0.03,
"learning_rate": 1.0200000000000002e-06,
"loss": 4.8984,
"step": 18
},
{
"epoch": 0.03,
"learning_rate": 1.08e-06,
"loss": 4.6559,
"step": 19
},
{
"epoch": 0.03,
"learning_rate": 1.14e-06,
"loss": 4.8867,
"step": 20
},
{
"epoch": 0.04,
"learning_rate": 1.2000000000000002e-06,
"loss": 4.8371,
"step": 21
},
{
"epoch": 0.04,
"learning_rate": 1.26e-06,
"loss": 4.7583,
"step": 22
},
{
"epoch": 0.04,
"learning_rate": 1.3199999999999999e-06,
"loss": 4.7754,
"step": 23
},
{
"epoch": 0.04,
"learning_rate": 1.38e-06,
"loss": 4.8551,
"step": 24
},
{
"epoch": 0.04,
"learning_rate": 1.44e-06,
"loss": 4.6186,
"step": 25
},
{
"epoch": 0.04,
"learning_rate": 1.5e-06,
"loss": 5.0174,
"step": 26
},
{
"epoch": 0.05,
"learning_rate": 1.5599999999999999e-06,
"loss": 4.6709,
"step": 27
},
{
"epoch": 0.05,
"learning_rate": 1.62e-06,
"loss": 4.5875,
"step": 28
},
{
"epoch": 0.05,
"learning_rate": 1.68e-06,
"loss": 4.5901,
"step": 29
},
{
"epoch": 0.05,
"learning_rate": 1.74e-06,
"loss": 4.7274,
"step": 30
},
{
"epoch": 0.05,
"learning_rate": 1.8e-06,
"loss": 4.7431,
"step": 31
},
{
"epoch": 0.05,
"learning_rate": 1.86e-06,
"loss": 4.6088,
"step": 32
},
{
"epoch": 0.06,
"learning_rate": 1.86e-06,
"loss": 4.615,
"step": 33
},
{
"epoch": 0.06,
"learning_rate": 1.9200000000000003e-06,
"loss": 4.5482,
"step": 34
},
{
"epoch": 0.06,
"learning_rate": 1.98e-06,
"loss": 4.6133,
"step": 35
},
{
"epoch": 0.06,
"learning_rate": 2.0400000000000004e-06,
"loss": 4.7532,
"step": 36
},
{
"epoch": 0.06,
"learning_rate": 2.1000000000000002e-06,
"loss": 4.5558,
"step": 37
},
{
"epoch": 0.06,
"learning_rate": 2.16e-06,
"loss": 4.7056,
"step": 38
},
{
"epoch": 0.07,
"learning_rate": 2.22e-06,
"loss": 4.8751,
"step": 39
},
{
"epoch": 0.07,
"learning_rate": 2.28e-06,
"loss": 4.4975,
"step": 40
},
{
"epoch": 0.07,
"learning_rate": 2.34e-06,
"loss": 4.55,
"step": 41
},
{
"epoch": 0.07,
"learning_rate": 2.4000000000000003e-06,
"loss": 4.7879,
"step": 42
},
{
"epoch": 0.07,
"learning_rate": 2.46e-06,
"loss": 4.5138,
"step": 43
},
{
"epoch": 0.07,
"learning_rate": 2.52e-06,
"loss": 4.6415,
"step": 44
},
{
"epoch": 0.08,
"learning_rate": 2.58e-06,
"loss": 4.7682,
"step": 45
},
{
"epoch": 0.08,
"learning_rate": 2.6399999999999997e-06,
"loss": 4.8274,
"step": 46
},
{
"epoch": 0.08,
"learning_rate": 2.7e-06,
"loss": 4.7302,
"step": 47
},
{
"epoch": 0.08,
"learning_rate": 2.76e-06,
"loss": 5.1127,
"step": 48
},
{
"epoch": 0.08,
"learning_rate": 2.82e-06,
"loss": 5.1541,
"step": 49
},
{
"epoch": 0.08,
"learning_rate": 2.82e-06,
"loss": 4.9245,
"step": 50
},
{
"epoch": 0.09,
"learning_rate": 2.88e-06,
"loss": 4.3791,
"step": 51
},
{
"epoch": 0.09,
"learning_rate": 2.9400000000000002e-06,
"loss": 4.3854,
"step": 52
},
{
"epoch": 0.09,
"learning_rate": 3e-06,
"loss": 4.4257,
"step": 53
},
{
"epoch": 0.09,
"learning_rate": 3.06e-06,
"loss": 4.4946,
"step": 54
},
{
"epoch": 0.09,
"learning_rate": 3.1199999999999998e-06,
"loss": 4.3393,
"step": 55
},
{
"epoch": 0.09,
"learning_rate": 3.18e-06,
"loss": 4.5326,
"step": 56
},
{
"epoch": 0.1,
"learning_rate": 3.24e-06,
"loss": 4.3733,
"step": 57
},
{
"epoch": 0.1,
"learning_rate": 3.3e-06,
"loss": 4.6388,
"step": 58
},
{
"epoch": 0.1,
"learning_rate": 3.36e-06,
"loss": 4.4159,
"step": 59
},
{
"epoch": 0.1,
"learning_rate": 3.4200000000000003e-06,
"loss": 4.1852,
"step": 60
},
{
"epoch": 0.1,
"learning_rate": 3.48e-06,
"loss": 4.4377,
"step": 61
},
{
"epoch": 0.1,
"learning_rate": 3.54e-06,
"loss": 4.5062,
"step": 62
},
{
"epoch": 0.11,
"learning_rate": 3.6e-06,
"loss": 4.3738,
"step": 63
},
{
"epoch": 0.11,
"learning_rate": 3.66e-06,
"loss": 4.3337,
"step": 64
},
{
"epoch": 0.11,
"learning_rate": 3.72e-06,
"loss": 4.3739,
"step": 65
},
{
"epoch": 0.11,
"learning_rate": 3.7800000000000002e-06,
"loss": 4.3296,
"step": 66
},
{
"epoch": 0.11,
"learning_rate": 3.8400000000000005e-06,
"loss": 4.2975,
"step": 67
},
{
"epoch": 0.11,
"learning_rate": 3.9e-06,
"loss": 4.2616,
"step": 68
},
{
"epoch": 0.12,
"learning_rate": 3.96e-06,
"loss": 4.3797,
"step": 69
},
{
"epoch": 0.12,
"learning_rate": 4.0200000000000005e-06,
"loss": 4.3733,
"step": 70
},
{
"epoch": 0.12,
"learning_rate": 4.080000000000001e-06,
"loss": 4.4484,
"step": 71
},
{
"epoch": 0.12,
"learning_rate": 4.14e-06,
"loss": 4.4206,
"step": 72
},
{
"epoch": 0.12,
"learning_rate": 4.2000000000000004e-06,
"loss": 4.2009,
"step": 73
},
{
"epoch": 0.12,
"learning_rate": 4.26e-06,
"loss": 4.4137,
"step": 74
},
{
"epoch": 0.13,
"learning_rate": 4.32e-06,
"loss": 4.3882,
"step": 75
},
{
"epoch": 0.13,
"learning_rate": 4.3799999999999996e-06,
"loss": 4.3882,
"step": 76
},
{
"epoch": 0.13,
"learning_rate": 4.44e-06,
"loss": 4.3548,
"step": 77
},
{
"epoch": 0.13,
"learning_rate": 4.5e-06,
"loss": 4.3271,
"step": 78
},
{
"epoch": 0.13,
"learning_rate": 4.56e-06,
"loss": 4.2483,
"step": 79
},
{
"epoch": 0.13,
"learning_rate": 4.62e-06,
"loss": 4.4255,
"step": 80
},
{
"epoch": 0.14,
"learning_rate": 4.68e-06,
"loss": 4.3572,
"step": 81
},
{
"epoch": 0.14,
"learning_rate": 4.74e-06,
"loss": 4.3565,
"step": 82
},
{
"epoch": 0.14,
"learning_rate": 4.800000000000001e-06,
"loss": 4.3942,
"step": 83
},
{
"epoch": 0.14,
"learning_rate": 4.86e-06,
"loss": 4.4026,
"step": 84
},
{
"epoch": 0.14,
"learning_rate": 4.92e-06,
"loss": 4.3234,
"step": 85
},
{
"epoch": 0.14,
"learning_rate": 4.980000000000001e-06,
"loss": 4.2278,
"step": 86
},
{
"epoch": 0.15,
"learning_rate": 5.04e-06,
"loss": 4.5104,
"step": 87
},
{
"epoch": 0.15,
"learning_rate": 5.1e-06,
"loss": 4.3936,
"step": 88
},
{
"epoch": 0.15,
"learning_rate": 5.16e-06,
"loss": 4.3561,
"step": 89
},
{
"epoch": 0.15,
"learning_rate": 5.22e-06,
"loss": 4.4134,
"step": 90
},
{
"epoch": 0.15,
"learning_rate": 5.279999999999999e-06,
"loss": 4.3388,
"step": 91
},
{
"epoch": 0.15,
"learning_rate": 5.34e-06,
"loss": 4.4584,
"step": 92
},
{
"epoch": 0.16,
"learning_rate": 5.4e-06,
"loss": 4.3285,
"step": 93
},
{
"epoch": 0.16,
"learning_rate": 5.46e-06,
"loss": 4.5666,
"step": 94
},
{
"epoch": 0.16,
"learning_rate": 5.52e-06,
"loss": 4.4606,
"step": 95
},
{
"epoch": 0.16,
"learning_rate": 5.58e-06,
"loss": 4.5187,
"step": 96
},
{
"epoch": 0.16,
"learning_rate": 5.64e-06,
"loss": 4.5188,
"step": 97
},
{
"epoch": 0.16,
"learning_rate": 5.7000000000000005e-06,
"loss": 4.3568,
"step": 98
},
{
"epoch": 0.17,
"learning_rate": 5.76e-06,
"loss": 4.5727,
"step": 99
},
{
"epoch": 0.17,
"learning_rate": 5.82e-06,
"loss": 4.3026,
"step": 100
},
{
"epoch": 0.17,
"learning_rate": 5.8800000000000005e-06,
"loss": 4.4187,
"step": 101
},
{
"epoch": 0.17,
"learning_rate": 5.940000000000001e-06,
"loss": 4.232,
"step": 102
},
{
"epoch": 0.17,
"learning_rate": 6e-06,
"loss": 4.2246,
"step": 103
},
{
"epoch": 0.17,
"learning_rate": 6.0600000000000004e-06,
"loss": 4.1768,
"step": 104
},
{
"epoch": 0.18,
"learning_rate": 6.12e-06,
"loss": 4.2036,
"step": 105
},
{
"epoch": 0.18,
"learning_rate": 6.18e-06,
"loss": 4.2184,
"step": 106
},
{
"epoch": 0.18,
"learning_rate": 6.2399999999999995e-06,
"loss": 4.3674,
"step": 107
},
{
"epoch": 0.18,
"learning_rate": 6.3e-06,
"loss": 4.383,
"step": 108
},
{
"epoch": 0.18,
"learning_rate": 6.36e-06,
"loss": 4.4528,
"step": 109
},
{
"epoch": 0.18,
"learning_rate": 6.42e-06,
"loss": 4.2543,
"step": 110
},
{
"epoch": 0.19,
"learning_rate": 6.48e-06,
"loss": 4.2407,
"step": 111
},
{
"epoch": 0.19,
"learning_rate": 6.54e-06,
"loss": 4.3249,
"step": 112
},
{
"epoch": 0.19,
"learning_rate": 6.6e-06,
"loss": 4.3045,
"step": 113
},
{
"epoch": 0.19,
"learning_rate": 6.660000000000001e-06,
"loss": 4.3379,
"step": 114
},
{
"epoch": 0.19,
"learning_rate": 6.72e-06,
"loss": 4.2925,
"step": 115
},
{
"epoch": 0.2,
"learning_rate": 6.78e-06,
"loss": 4.2865,
"step": 116
},
{
"epoch": 0.2,
"learning_rate": 6.840000000000001e-06,
"loss": 4.3821,
"step": 117
},
{
"epoch": 0.2,
"learning_rate": 6.900000000000001e-06,
"loss": 4.1678,
"step": 118
},
{
"epoch": 0.2,
"learning_rate": 6.96e-06,
"loss": 4.1453,
"step": 119
},
{
"epoch": 0.2,
"learning_rate": 7.0200000000000006e-06,
"loss": 4.1488,
"step": 120
},
{
"epoch": 0.2,
"learning_rate": 7.08e-06,
"loss": 4.3728,
"step": 121
},
{
"epoch": 0.21,
"learning_rate": 7.14e-06,
"loss": 4.462,
"step": 122
},
{
"epoch": 0.21,
"learning_rate": 7.2e-06,
"loss": 4.2362,
"step": 123
},
{
"epoch": 0.21,
"learning_rate": 7.26e-06,
"loss": 4.2868,
"step": 124
},
{
"epoch": 0.21,
"learning_rate": 7.32e-06,
"loss": 4.31,
"step": 125
},
{
"epoch": 0.21,
"learning_rate": 7.3800000000000005e-06,
"loss": 4.1183,
"step": 126
},
{
"epoch": 0.21,
"learning_rate": 7.44e-06,
"loss": 4.1797,
"step": 127
},
{
"epoch": 0.22,
"learning_rate": 7.5e-06,
"loss": 4.1409,
"step": 128
},
{
"epoch": 0.22,
"learning_rate": 7.5600000000000005e-06,
"loss": 4.2855,
"step": 129
},
{
"epoch": 0.22,
"learning_rate": 7.62e-06,
"loss": 4.3087,
"step": 130
},
{
"epoch": 0.22,
"learning_rate": 7.680000000000001e-06,
"loss": 4.2624,
"step": 131
},
{
"epoch": 0.22,
"learning_rate": 7.74e-06,
"loss": 4.2154,
"step": 132
},
{
"epoch": 0.22,
"learning_rate": 7.8e-06,
"loss": 4.3051,
"step": 133
},
{
"epoch": 0.23,
"learning_rate": 7.860000000000001e-06,
"loss": 4.3184,
"step": 134
},
{
"epoch": 0.23,
"learning_rate": 7.92e-06,
"loss": 4.3726,
"step": 135
},
{
"epoch": 0.23,
"learning_rate": 7.98e-06,
"loss": 4.2253,
"step": 136
},
{
"epoch": 0.23,
"learning_rate": 8.040000000000001e-06,
"loss": 4.24,
"step": 137
},
{
"epoch": 0.23,
"learning_rate": 8.1e-06,
"loss": 4.2433,
"step": 138
},
{
"epoch": 0.23,
"learning_rate": 8.160000000000001e-06,
"loss": 4.2364,
"step": 139
},
{
"epoch": 0.24,
"learning_rate": 8.220000000000001e-06,
"loss": 4.4473,
"step": 140
},
{
"epoch": 0.24,
"learning_rate": 8.28e-06,
"loss": 4.3516,
"step": 141
},
{
"epoch": 0.24,
"learning_rate": 8.340000000000001e-06,
"loss": 4.4041,
"step": 142
},
{
"epoch": 0.24,
"learning_rate": 8.400000000000001e-06,
"loss": 4.4607,
"step": 143
},
{
"epoch": 0.24,
"learning_rate": 8.459999999999999e-06,
"loss": 4.2525,
"step": 144
},
{
"epoch": 0.24,
"learning_rate": 8.52e-06,
"loss": 4.3223,
"step": 145
},
{
"epoch": 0.25,
"learning_rate": 8.58e-06,
"loss": 4.6555,
"step": 146
},
{
"epoch": 0.25,
"learning_rate": 8.64e-06,
"loss": 4.6114,
"step": 147
},
{
"epoch": 0.25,
"learning_rate": 8.7e-06,
"loss": 4.597,
"step": 148
},
{
"epoch": 0.25,
"learning_rate": 8.759999999999999e-06,
"loss": 4.5454,
"step": 149
},
{
"epoch": 0.25,
"learning_rate": 8.82e-06,
"loss": 4.6048,
"step": 150
},
{
"epoch": 0.25,
"learning_rate": 8.88e-06,
"loss": 4.2387,
"step": 151
},
{
"epoch": 0.26,
"learning_rate": 8.939999999999999e-06,
"loss": 4.175,
"step": 152
},
{
"epoch": 0.26,
"learning_rate": 9e-06,
"loss": 4.1509,
"step": 153
},
{
"epoch": 0.26,
"learning_rate": 9.06e-06,
"loss": 4.0987,
"step": 154
},
{
"epoch": 0.26,
"learning_rate": 9.12e-06,
"loss": 4.1229,
"step": 155
},
{
"epoch": 0.26,
"learning_rate": 9.18e-06,
"loss": 4.1271,
"step": 156
},
{
"epoch": 0.26,
"learning_rate": 9.24e-06,
"loss": 4.2867,
"step": 157
},
{
"epoch": 0.27,
"learning_rate": 9.3e-06,
"loss": 4.1166,
"step": 158
},
{
"epoch": 0.27,
"learning_rate": 9.36e-06,
"loss": 4.0816,
"step": 159
},
{
"epoch": 0.27,
"learning_rate": 9.42e-06,
"loss": 4.1966,
"step": 160
},
{
"epoch": 0.27,
"learning_rate": 9.48e-06,
"loss": 4.2313,
"step": 161
},
{
"epoch": 0.27,
"learning_rate": 9.54e-06,
"loss": 4.1073,
"step": 162
},
{
"epoch": 0.27,
"learning_rate": 9.600000000000001e-06,
"loss": 4.1444,
"step": 163
},
{
"epoch": 0.28,
"learning_rate": 9.66e-06,
"loss": 4.1647,
"step": 164
},
{
"epoch": 0.28,
"learning_rate": 9.72e-06,
"loss": 4.2105,
"step": 165
},
{
"epoch": 0.28,
"learning_rate": 9.780000000000001e-06,
"loss": 4.3272,
"step": 166
},
{
"epoch": 0.28,
"learning_rate": 9.84e-06,
"loss": 4.1639,
"step": 167
},
{
"epoch": 0.28,
"learning_rate": 9.9e-06,
"loss": 4.3927,
"step": 168
},
{
"epoch": 0.28,
"learning_rate": 9.960000000000001e-06,
"loss": 4.2094,
"step": 169
},
{
"epoch": 0.29,
"learning_rate": 1.002e-05,
"loss": 4.2621,
"step": 170
},
{
"epoch": 0.29,
"learning_rate": 1.008e-05,
"loss": 4.1226,
"step": 171
},
{
"epoch": 0.29,
"learning_rate": 1.0140000000000001e-05,
"loss": 4.1309,
"step": 172
},
{
"epoch": 0.29,
"learning_rate": 1.02e-05,
"loss": 4.1452,
"step": 173
},
{
"epoch": 0.29,
"learning_rate": 1.0260000000000002e-05,
"loss": 4.1409,
"step": 174
},
{
"epoch": 0.29,
"learning_rate": 1.032e-05,
"loss": 4.2783,
"step": 175
},
{
"epoch": 0.3,
"learning_rate": 1.0379999999999999e-05,
"loss": 4.3033,
"step": 176
},
{
"epoch": 0.3,
"learning_rate": 1.044e-05,
"loss": 4.2395,
"step": 177
},
{
"epoch": 0.3,
"learning_rate": 1.05e-05,
"loss": 4.1372,
"step": 178
},
{
"epoch": 0.3,
"learning_rate": 1.0559999999999999e-05,
"loss": 4.2603,
"step": 179
},
{
"epoch": 0.3,
"learning_rate": 1.062e-05,
"loss": 4.1358,
"step": 180
},
{
"epoch": 0.3,
"learning_rate": 1.068e-05,
"loss": 4.2092,
"step": 181
},
{
"epoch": 0.31,
"learning_rate": 1.074e-05,
"loss": 4.1969,
"step": 182
},
{
"epoch": 0.31,
"learning_rate": 1.08e-05,
"loss": 4.183,
"step": 183
},
{
"epoch": 0.31,
"learning_rate": 1.086e-05,
"loss": 4.0783,
"step": 184
},
{
"epoch": 0.31,
"learning_rate": 1.092e-05,
"loss": 4.4204,
"step": 185
},
{
"epoch": 0.31,
"learning_rate": 1.098e-05,
"loss": 4.2284,
"step": 186
},
{
"epoch": 0.31,
"learning_rate": 1.104e-05,
"loss": 4.2885,
"step": 187
},
{
"epoch": 0.32,
"learning_rate": 1.11e-05,
"loss": 4.316,
"step": 188
},
{
"epoch": 0.32,
"learning_rate": 1.116e-05,
"loss": 4.4701,
"step": 189
},
{
"epoch": 0.32,
"learning_rate": 1.1220000000000001e-05,
"loss": 4.1684,
"step": 190
},
{
"epoch": 0.32,
"learning_rate": 1.128e-05,
"loss": 4.4552,
"step": 191
},
{
"epoch": 0.32,
"learning_rate": 1.134e-05,
"loss": 4.3806,
"step": 192
},
{
"epoch": 0.32,
"learning_rate": 1.1400000000000001e-05,
"loss": 4.2893,
"step": 193
},
{
"epoch": 0.33,
"learning_rate": 1.146e-05,
"loss": 4.3502,
"step": 194
},
{
"epoch": 0.33,
"learning_rate": 1.152e-05,
"loss": 4.3027,
"step": 195
},
{
"epoch": 0.33,
"learning_rate": 1.1580000000000001e-05,
"loss": 4.1918,
"step": 196
},
{
"epoch": 0.33,
"learning_rate": 1.164e-05,
"loss": 4.6005,
"step": 197
},
{
"epoch": 0.33,
"learning_rate": 1.1700000000000001e-05,
"loss": 4.5839,
"step": 198
},
{
"epoch": 0.33,
"learning_rate": 1.1760000000000001e-05,
"loss": 4.4992,
"step": 199
},
{
"epoch": 0.34,
"learning_rate": 1.182e-05,
"loss": 4.6242,
"step": 200
},
{
"epoch": 0.34,
"learning_rate": 1.1880000000000001e-05,
"loss": 4.0522,
"step": 201
},
{
"epoch": 0.34,
"learning_rate": 1.1940000000000001e-05,
"loss": 4.191,
"step": 202
},
{
"epoch": 0.34,
"learning_rate": 1.2e-05,
"loss": 4.2412,
"step": 203
},
{
"epoch": 0.34,
"learning_rate": 1.2060000000000001e-05,
"loss": 4.1275,
"step": 204
},
{
"epoch": 0.34,
"learning_rate": 1.2120000000000001e-05,
"loss": 4.1794,
"step": 205
},
{
"epoch": 0.35,
"learning_rate": 1.2180000000000002e-05,
"loss": 4.1126,
"step": 206
},
{
"epoch": 0.35,
"learning_rate": 1.224e-05,
"loss": 4.1722,
"step": 207
},
{
"epoch": 0.35,
"learning_rate": 1.2299999999999999e-05,
"loss": 4.1502,
"step": 208
},
{
"epoch": 0.35,
"learning_rate": 1.236e-05,
"loss": 4.1738,
"step": 209
},
{
"epoch": 0.35,
"learning_rate": 1.242e-05,
"loss": 4.2742,
"step": 210
},
{
"epoch": 0.35,
"learning_rate": 1.2479999999999999e-05,
"loss": 4.264,
"step": 211
},
{
"epoch": 0.36,
"learning_rate": 1.254e-05,
"loss": 4.2409,
"step": 212
},
{
"epoch": 0.36,
"learning_rate": 1.26e-05,
"loss": 4.2049,
"step": 213
},
{
"epoch": 0.36,
"learning_rate": 1.2659999999999999e-05,
"loss": 4.2889,
"step": 214
},
{
"epoch": 0.36,
"learning_rate": 1.272e-05,
"loss": 4.1661,
"step": 215
},
{
"epoch": 0.36,
"learning_rate": 1.278e-05,
"loss": 4.1836,
"step": 216
},
{
"epoch": 0.36,
"learning_rate": 1.284e-05,
"loss": 4.0784,
"step": 217
},
{
"epoch": 0.37,
"learning_rate": 1.29e-05,
"loss": 4.1899,
"step": 218
},
{
"epoch": 0.37,
"learning_rate": 1.296e-05,
"loss": 4.1644,
"step": 219
},
{
"epoch": 0.37,
"learning_rate": 1.302e-05,
"loss": 4.2537,
"step": 220
},
{
"epoch": 0.37,
"learning_rate": 1.308e-05,
"loss": 4.2828,
"step": 221
},
{
"epoch": 0.37,
"learning_rate": 1.314e-05,
"loss": 4.2814,
"step": 222
},
{
"epoch": 0.37,
"learning_rate": 1.32e-05,
"loss": 4.1264,
"step": 223
},
{
"epoch": 0.38,
"learning_rate": 1.326e-05,
"loss": 4.2383,
"step": 224
},
{
"epoch": 0.38,
"learning_rate": 1.3320000000000001e-05,
"loss": 4.4086,
"step": 225
},
{
"epoch": 0.38,
"learning_rate": 1.338e-05,
"loss": 4.1876,
"step": 226
},
{
"epoch": 0.38,
"learning_rate": 1.344e-05,
"loss": 4.1319,
"step": 227
},
{
"epoch": 0.38,
"learning_rate": 1.3500000000000001e-05,
"loss": 4.1903,
"step": 228
},
{
"epoch": 0.39,
"learning_rate": 1.356e-05,
"loss": 4.3229,
"step": 229
},
{
"epoch": 0.39,
"learning_rate": 1.362e-05,
"loss": 4.2784,
"step": 230
},
{
"epoch": 0.39,
"learning_rate": 1.3680000000000001e-05,
"loss": 4.1548,
"step": 231
},
{
"epoch": 0.39,
"learning_rate": 1.374e-05,
"loss": 4.2373,
"step": 232
},
{
"epoch": 0.39,
"learning_rate": 1.3800000000000002e-05,
"loss": 4.2429,
"step": 233
},
{
"epoch": 0.39,
"learning_rate": 1.3860000000000001e-05,
"loss": 4.2312,
"step": 234
},
{
"epoch": 0.4,
"learning_rate": 1.392e-05,
"loss": 4.3897,
"step": 235
},
{
"epoch": 0.4,
"learning_rate": 1.3980000000000002e-05,
"loss": 4.1516,
"step": 236
},
{
"epoch": 0.4,
"learning_rate": 1.4040000000000001e-05,
"loss": 4.1352,
"step": 237
},
{
"epoch": 0.4,
"learning_rate": 1.4099999999999999e-05,
"loss": 4.1851,
"step": 238
},
{
"epoch": 0.4,
"learning_rate": 1.416e-05,
"loss": 4.1067,
"step": 239
},
{
"epoch": 0.4,
"learning_rate": 1.422e-05,
"loss": 4.2594,
"step": 240
},
{
"epoch": 0.41,
"learning_rate": 1.428e-05,
"loss": 4.1948,
"step": 241
},
{
"epoch": 0.41,
"learning_rate": 1.434e-05,
"loss": 4.2121,
"step": 242
},
{
"epoch": 0.41,
"learning_rate": 1.44e-05,
"loss": 4.2277,
"step": 243
},
{
"epoch": 0.41,
"learning_rate": 1.446e-05,
"loss": 4.3072,
"step": 244
},
{
"epoch": 0.41,
"learning_rate": 1.452e-05,
"loss": 4.406,
"step": 245
},
{
"epoch": 0.41,
"learning_rate": 1.458e-05,
"loss": 4.4097,
"step": 246
},
{
"epoch": 0.42,
"learning_rate": 1.464e-05,
"loss": 4.404,
"step": 247
},
{
"epoch": 0.42,
"learning_rate": 1.47e-05,
"loss": 4.5215,
"step": 248
},
{
"epoch": 0.42,
"learning_rate": 1.4760000000000001e-05,
"loss": 4.4901,
"step": 249
},
{
"epoch": 0.42,
"learning_rate": 1.482e-05,
"loss": 4.5015,
"step": 250
},
{
"epoch": 0.42,
"learning_rate": 1.488e-05,
"loss": 4.2261,
"step": 251
},
{
"epoch": 0.42,
"learning_rate": 1.4940000000000001e-05,
"loss": 4.209,
"step": 252
},
{
"epoch": 0.43,
"learning_rate": 1.5e-05,
"loss": 4.1874,
"step": 253
},
{
"epoch": 0.43,
"learning_rate": 1.506e-05,
"loss": 4.0856,
"step": 254
},
{
"epoch": 0.43,
"learning_rate": 1.5120000000000001e-05,
"loss": 4.2503,
"step": 255
},
{
"epoch": 0.43,
"learning_rate": 1.518e-05,
"loss": 4.1818,
"step": 256
},
{
"epoch": 0.43,
"learning_rate": 1.524e-05,
"loss": 4.1302,
"step": 257
},
{
"epoch": 0.43,
"learning_rate": 1.53e-05,
"loss": 4.1573,
"step": 258
},
{
"epoch": 0.44,
"learning_rate": 1.5360000000000002e-05,
"loss": 4.2211,
"step": 259
},
{
"epoch": 0.44,
"learning_rate": 1.542e-05,
"loss": 4.2337,
"step": 260
},
{
"epoch": 0.44,
"learning_rate": 1.548e-05,
"loss": 4.1873,
"step": 261
},
{
"epoch": 0.44,
"learning_rate": 1.554e-05,
"loss": 4.1336,
"step": 262
},
{
"epoch": 0.44,
"learning_rate": 1.56e-05,
"loss": 4.2485,
"step": 263
},
{
"epoch": 0.44,
"learning_rate": 1.5660000000000003e-05,
"loss": 3.9808,
"step": 264
},
{
"epoch": 0.45,
"learning_rate": 1.5720000000000002e-05,
"loss": 4.0666,
"step": 265
},
{
"epoch": 0.45,
"learning_rate": 1.578e-05,
"loss": 4.1813,
"step": 266
},
{
"epoch": 0.45,
"learning_rate": 1.584e-05,
"loss": 4.1908,
"step": 267
},
{
"epoch": 0.45,
"learning_rate": 1.59e-05,
"loss": 4.1232,
"step": 268
},
{
"epoch": 0.45,
"learning_rate": 1.596e-05,
"loss": 4.2437,
"step": 269
},
{
"epoch": 0.45,
"learning_rate": 1.6020000000000002e-05,
"loss": 4.2354,
"step": 270
},
{
"epoch": 0.46,
"learning_rate": 1.6080000000000002e-05,
"loss": 4.1458,
"step": 271
},
{
"epoch": 0.46,
"learning_rate": 1.614e-05,
"loss": 4.3086,
"step": 272
},
{
"epoch": 0.46,
"learning_rate": 1.62e-05,
"loss": 4.165,
"step": 273
},
{
"epoch": 0.46,
"learning_rate": 1.626e-05,
"loss": 4.1709,
"step": 274
},
{
"epoch": 0.46,
"learning_rate": 1.6320000000000003e-05,
"loss": 4.2307,
"step": 275
},
{
"epoch": 0.46,
"learning_rate": 1.6380000000000002e-05,
"loss": 4.1744,
"step": 276
},
{
"epoch": 0.47,
"learning_rate": 1.6440000000000002e-05,
"loss": 4.2963,
"step": 277
},
{
"epoch": 0.47,
"learning_rate": 1.65e-05,
"loss": 4.0863,
"step": 278
},
{
"epoch": 0.47,
"learning_rate": 1.656e-05,
"loss": 4.2849,
"step": 279
},
{
"epoch": 0.47,
"learning_rate": 1.6620000000000004e-05,
"loss": 4.3489,
"step": 280
},
{
"epoch": 0.47,
"learning_rate": 1.6680000000000003e-05,
"loss": 4.347,
"step": 281
},
{
"epoch": 0.47,
"learning_rate": 1.6740000000000002e-05,
"loss": 4.2189,
"step": 282
},
{
"epoch": 0.48,
"learning_rate": 1.6800000000000002e-05,
"loss": 4.13,
"step": 283
},
{
"epoch": 0.48,
"learning_rate": 1.686e-05,
"loss": 4.1049,
"step": 284
},
{
"epoch": 0.48,
"learning_rate": 1.6919999999999997e-05,
"loss": 4.3792,
"step": 285
},
{
"epoch": 0.48,
"learning_rate": 1.698e-05,
"loss": 4.115,
"step": 286
},
{
"epoch": 0.48,
"learning_rate": 1.704e-05,
"loss": 4.469,
"step": 287
},
{
"epoch": 0.48,
"learning_rate": 1.71e-05,
"loss": 3.9562,
"step": 288
},
{
"epoch": 0.49,
"learning_rate": 1.716e-05,
"loss": 4.209,
"step": 289
},
{
"epoch": 0.49,
"learning_rate": 1.7219999999999998e-05,
"loss": 4.1822,
"step": 290
},
{
"epoch": 0.49,
"learning_rate": 1.728e-05,
"loss": 4.4213,
"step": 291
},
{
"epoch": 0.49,
"learning_rate": 1.734e-05,
"loss": 4.1106,
"step": 292
},
{
"epoch": 0.49,
"learning_rate": 1.74e-05,
"loss": 4.2934,
"step": 293
},
{
"epoch": 0.49,
"learning_rate": 1.746e-05,
"loss": 4.3631,
"step": 294
},
{
"epoch": 0.5,
"learning_rate": 1.7519999999999998e-05,
"loss": 4.343,
"step": 295
},
{
"epoch": 0.5,
"learning_rate": 1.758e-05,
"loss": 4.1336,
"step": 296
},
{
"epoch": 0.5,
"learning_rate": 1.764e-05,
"loss": 4.2157,
"step": 297
},
{
"epoch": 0.5,
"learning_rate": 1.77e-05,
"loss": 4.3987,
"step": 298
},
{
"epoch": 0.5,
"learning_rate": 1.776e-05,
"loss": 4.4069,
"step": 299
},
{
"epoch": 0.5,
"learning_rate": 1.782e-05,
"loss": 4.641,
"step": 300
},
{
"epoch": 0.51,
"learning_rate": 1.7879999999999998e-05,
"loss": 4.284,
"step": 301
},
{
"epoch": 0.51,
"learning_rate": 1.794e-05,
"loss": 4.2259,
"step": 302
},
{
"epoch": 0.51,
"learning_rate": 1.8e-05,
"loss": 4.3155,
"step": 303
},
{
"epoch": 0.51,
"learning_rate": 1.806e-05,
"loss": 4.0805,
"step": 304
},
{
"epoch": 0.51,
"learning_rate": 1.812e-05,
"loss": 4.2797,
"step": 305
},
{
"epoch": 0.51,
"learning_rate": 1.818e-05,
"loss": 4.2722,
"step": 306
},
{
"epoch": 0.52,
"learning_rate": 1.824e-05,
"loss": 4.0556,
"step": 307
},
{
"epoch": 0.52,
"learning_rate": 1.83e-05,
"loss": 4.2789,
"step": 308
},
{
"epoch": 0.52,
"learning_rate": 1.836e-05,
"loss": 4.1761,
"step": 309
},
{
"epoch": 0.52,
"learning_rate": 1.842e-05,
"loss": 4.261,
"step": 310
},
{
"epoch": 0.52,
"learning_rate": 1.848e-05,
"loss": 4.1181,
"step": 311
},
{
"epoch": 0.52,
"learning_rate": 1.854e-05,
"loss": 4.1469,
"step": 312
},
{
"epoch": 0.53,
"learning_rate": 1.86e-05,
"loss": 4.187,
"step": 313
},
{
"epoch": 0.53,
"learning_rate": 1.866e-05,
"loss": 4.1858,
"step": 314
},
{
"epoch": 0.53,
"learning_rate": 1.872e-05,
"loss": 4.2071,
"step": 315
},
{
"epoch": 0.53,
"learning_rate": 1.878e-05,
"loss": 4.2038,
"step": 316
},
{
"epoch": 0.53,
"learning_rate": 1.884e-05,
"loss": 4.1419,
"step": 317
},
{
"epoch": 0.53,
"learning_rate": 1.8900000000000002e-05,
"loss": 4.2296,
"step": 318
},
{
"epoch": 0.54,
"learning_rate": 1.896e-05,
"loss": 4.0631,
"step": 319
},
{
"epoch": 0.54,
"learning_rate": 1.902e-05,
"loss": 4.2325,
"step": 320
},
{
"epoch": 0.54,
"learning_rate": 1.908e-05,
"loss": 4.0767,
"step": 321
},
{
"epoch": 0.54,
"learning_rate": 1.914e-05,
"loss": 4.1089,
"step": 322
},
{
"epoch": 0.54,
"learning_rate": 1.9200000000000003e-05,
"loss": 4.1485,
"step": 323
},
{
"epoch": 0.54,
"learning_rate": 1.9260000000000002e-05,
"loss": 4.209,
"step": 324
},
{
"epoch": 0.55,
"learning_rate": 1.932e-05,
"loss": 4.1571,
"step": 325
},
{
"epoch": 0.55,
"learning_rate": 1.938e-05,
"loss": 4.1008,
"step": 326
},
{
"epoch": 0.55,
"learning_rate": 1.944e-05,
"loss": 4.3666,
"step": 327
},
{
"epoch": 0.55,
"learning_rate": 1.95e-05,
"loss": 4.1024,
"step": 328
},
{
"epoch": 0.55,
"learning_rate": 1.9560000000000002e-05,
"loss": 4.0316,
"step": 329
},
{
"epoch": 0.55,
"learning_rate": 1.9620000000000002e-05,
"loss": 4.2485,
"step": 330
},
{
"epoch": 0.56,
"learning_rate": 1.968e-05,
"loss": 4.1414,
"step": 331
},
{
"epoch": 0.56,
"learning_rate": 1.974e-05,
"loss": 4.0656,
"step": 332
},
{
"epoch": 0.56,
"learning_rate": 1.98e-05,
"loss": 4.1471,
"step": 333
},
{
"epoch": 0.56,
"learning_rate": 1.9860000000000003e-05,
"loss": 4.2051,
"step": 334
},
{
"epoch": 0.56,
"learning_rate": 1.9920000000000002e-05,
"loss": 4.2217,
"step": 335
},
{
"epoch": 0.56,
"learning_rate": 1.9980000000000002e-05,
"loss": 4.1121,
"step": 336
},
{
"epoch": 0.57,
"learning_rate": 2.004e-05,
"loss": 4.0834,
"step": 337
},
{
"epoch": 0.57,
"learning_rate": 2.01e-05,
"loss": 4.1913,
"step": 338
},
{
"epoch": 0.57,
"learning_rate": 2.016e-05,
"loss": 4.1983,
"step": 339
},
{
"epoch": 0.57,
"learning_rate": 2.0220000000000003e-05,
"loss": 4.2388,
"step": 340
},
{
"epoch": 0.57,
"learning_rate": 2.0280000000000002e-05,
"loss": 4.2476,
"step": 341
},
{
"epoch": 0.58,
"learning_rate": 2.0340000000000002e-05,
"loss": 4.3317,
"step": 342
},
{
"epoch": 0.58,
"learning_rate": 2.04e-05,
"loss": 4.3618,
"step": 343
},
{
"epoch": 0.58,
"learning_rate": 2.046e-05,
"loss": 4.2219,
"step": 344
},
{
"epoch": 0.58,
"learning_rate": 2.0520000000000003e-05,
"loss": 4.3897,
"step": 345
},
{
"epoch": 0.58,
"learning_rate": 2.0580000000000003e-05,
"loss": 4.0839,
"step": 346
},
{
"epoch": 0.58,
"learning_rate": 2.064e-05,
"loss": 4.2025,
"step": 347
},
{
"epoch": 0.59,
"learning_rate": 2.07e-05,
"loss": 4.5123,
"step": 348
},
{
"epoch": 0.59,
"learning_rate": 2.0759999999999998e-05,
"loss": 4.4112,
"step": 349
},
{
"epoch": 0.59,
"learning_rate": 2.082e-05,
"loss": 4.6823,
"step": 350
},
{
"epoch": 0.59,
"learning_rate": 2.088e-05,
"loss": 4.1024,
"step": 351
},
{
"epoch": 0.59,
"learning_rate": 2.094e-05,
"loss": 4.1924,
"step": 352
},
{
"epoch": 0.59,
"learning_rate": 2.1e-05,
"loss": 4.2308,
"step": 353
},
{
"epoch": 0.6,
"learning_rate": 2.1059999999999998e-05,
"loss": 4.0928,
"step": 354
},
{
"epoch": 0.6,
"learning_rate": 2.1119999999999998e-05,
"loss": 4.1424,
"step": 355
},
{
"epoch": 0.6,
"learning_rate": 2.118e-05,
"loss": 4.1515,
"step": 356
},
{
"epoch": 0.6,
"learning_rate": 2.124e-05,
"loss": 4.1723,
"step": 357
},
{
"epoch": 0.6,
"learning_rate": 2.13e-05,
"loss": 4.0917,
"step": 358
},
{
"epoch": 0.6,
"learning_rate": 2.136e-05,
"loss": 4.1245,
"step": 359
},
{
"epoch": 0.61,
"learning_rate": 2.1419999999999998e-05,
"loss": 4.0757,
"step": 360
},
{
"epoch": 0.61,
"learning_rate": 2.148e-05,
"loss": 4.162,
"step": 361
},
{
"epoch": 0.61,
"learning_rate": 2.154e-05,
"loss": 4.1548,
"step": 362
},
{
"epoch": 0.61,
"learning_rate": 2.16e-05,
"loss": 4.221,
"step": 363
},
{
"epoch": 0.61,
"learning_rate": 2.166e-05,
"loss": 4.0905,
"step": 364
},
{
"epoch": 0.61,
"learning_rate": 2.172e-05,
"loss": 4.1585,
"step": 365
},
{
"epoch": 0.62,
"learning_rate": 2.178e-05,
"loss": 4.1137,
"step": 366
},
{
"epoch": 0.62,
"learning_rate": 2.184e-05,
"loss": 4.1436,
"step": 367
},
{
"epoch": 0.62,
"learning_rate": 2.19e-05,
"loss": 4.0909,
"step": 368
},
{
"epoch": 0.62,
"learning_rate": 2.196e-05,
"loss": 4.2206,
"step": 369
},
{
"epoch": 0.62,
"learning_rate": 2.202e-05,
"loss": 4.1105,
"step": 370
},
{
"epoch": 0.62,
"learning_rate": 2.208e-05,
"loss": 4.0323,
"step": 371
},
{
"epoch": 0.63,
"learning_rate": 2.214e-05,
"loss": 4.1716,
"step": 372
},
{
"epoch": 0.63,
"learning_rate": 2.22e-05,
"loss": 4.0105,
"step": 373
},
{
"epoch": 0.63,
"learning_rate": 2.226e-05,
"loss": 4.0406,
"step": 374
},
{
"epoch": 0.63,
"learning_rate": 2.232e-05,
"loss": 4.2098,
"step": 375
},
{
"epoch": 0.63,
"learning_rate": 2.238e-05,
"loss": 4.095,
"step": 376
},
{
"epoch": 0.63,
"learning_rate": 2.2440000000000002e-05,
"loss": 4.1806,
"step": 377
},
{
"epoch": 0.64,
"learning_rate": 2.25e-05,
"loss": 4.2243,
"step": 378
},
{
"epoch": 0.64,
"learning_rate": 2.256e-05,
"loss": 4.0377,
"step": 379
},
{
"epoch": 0.64,
"learning_rate": 2.262e-05,
"loss": 4.2748,
"step": 380
},
{
"epoch": 0.64,
"learning_rate": 2.268e-05,
"loss": 4.203,
"step": 381
},
{
"epoch": 0.64,
"learning_rate": 2.274e-05,
"loss": 4.0632,
"step": 382
},
{
"epoch": 0.64,
"learning_rate": 2.2800000000000002e-05,
"loss": 3.9685,
"step": 383
},
{
"epoch": 0.65,
"learning_rate": 2.286e-05,
"loss": 4.1863,
"step": 384
},
{
"epoch": 0.65,
"learning_rate": 2.292e-05,
"loss": 4.3516,
"step": 385
},
{
"epoch": 0.65,
"learning_rate": 2.298e-05,
"loss": 4.1114,
"step": 386
},
{
"epoch": 0.65,
"learning_rate": 2.304e-05,
"loss": 4.2782,
"step": 387
},
{
"epoch": 0.65,
"learning_rate": 2.3100000000000002e-05,
"loss": 4.1808,
"step": 388
},
{
"epoch": 0.65,
"learning_rate": 2.3160000000000002e-05,
"loss": 4.0981,
"step": 389
},
{
"epoch": 0.66,
"learning_rate": 2.322e-05,
"loss": 4.0864,
"step": 390
},
{
"epoch": 0.66,
"learning_rate": 2.328e-05,
"loss": 4.2972,
"step": 391
},
{
"epoch": 0.66,
"learning_rate": 2.334e-05,
"loss": 4.2071,
"step": 392
},
{
"epoch": 0.66,
"learning_rate": 2.3400000000000003e-05,
"loss": 4.4654,
"step": 393
},
{
"epoch": 0.66,
"learning_rate": 2.3460000000000002e-05,
"loss": 4.2892,
"step": 394
},
{
"epoch": 0.66,
"learning_rate": 2.3520000000000002e-05,
"loss": 4.2635,
"step": 395
},
{
"epoch": 0.67,
"learning_rate": 2.358e-05,
"loss": 4.183,
"step": 396
},
{
"epoch": 0.67,
"learning_rate": 2.364e-05,
"loss": 4.3497,
"step": 397
},
{
"epoch": 0.67,
"learning_rate": 2.37e-05,
"loss": 4.5072,
"step": 398
},
{
"epoch": 0.67,
"learning_rate": 2.3760000000000003e-05,
"loss": 4.3209,
"step": 399
},
{
"epoch": 0.67,
"learning_rate": 2.3820000000000002e-05,
"loss": 4.3901,
"step": 400
},
{
"epoch": 0.67,
"learning_rate": 2.3880000000000002e-05,
"loss": 4.2026,
"step": 401
},
{
"epoch": 0.68,
"learning_rate": 2.394e-05,
"loss": 4.1082,
"step": 402
},
{
"epoch": 0.68,
"learning_rate": 2.4e-05,
"loss": 4.164,
"step": 403
},
{
"epoch": 0.68,
"learning_rate": 2.4060000000000003e-05,
"loss": 4.1493,
"step": 404
},
{
"epoch": 0.68,
"learning_rate": 2.4120000000000003e-05,
"loss": 4.0781,
"step": 405
},
{
"epoch": 0.68,
"learning_rate": 2.4180000000000002e-05,
"loss": 4.1192,
"step": 406
},
{
"epoch": 0.68,
"learning_rate": 2.4240000000000002e-05,
"loss": 4.1224,
"step": 407
},
{
"epoch": 0.69,
"learning_rate": 2.43e-05,
"loss": 4.0526,
"step": 408
},
{
"epoch": 0.69,
"learning_rate": 2.4360000000000004e-05,
"loss": 4.1111,
"step": 409
},
{
"epoch": 0.69,
"learning_rate": 2.442e-05,
"loss": 4.0723,
"step": 410
},
{
"epoch": 0.69,
"learning_rate": 2.448e-05,
"loss": 4.0744,
"step": 411
},
{
"epoch": 0.69,
"learning_rate": 2.454e-05,
"loss": 4.2666,
"step": 412
},
{
"epoch": 0.69,
"learning_rate": 2.4599999999999998e-05,
"loss": 4.1323,
"step": 413
},
{
"epoch": 0.7,
"learning_rate": 2.4659999999999998e-05,
"loss": 4.1745,
"step": 414
},
{
"epoch": 0.7,
"learning_rate": 2.472e-05,
"loss": 4.1703,
"step": 415
},
{
"epoch": 0.7,
"learning_rate": 2.478e-05,
"loss": 4.08,
"step": 416
},
{
"epoch": 0.7,
"learning_rate": 2.484e-05,
"loss": 4.1442,
"step": 417
},
{
"epoch": 0.7,
"learning_rate": 2.49e-05,
"loss": 3.9723,
"step": 418
},
{
"epoch": 0.7,
"learning_rate": 2.4959999999999998e-05,
"loss": 4.2404,
"step": 419
},
{
"epoch": 0.71,
"learning_rate": 2.502e-05,
"loss": 4.1607,
"step": 420
},
{
"epoch": 0.71,
"learning_rate": 2.508e-05,
"loss": 3.907,
"step": 421
},
{
"epoch": 0.71,
"learning_rate": 2.514e-05,
"loss": 4.0577,
"step": 422
},
{
"epoch": 0.71,
"learning_rate": 2.52e-05,
"loss": 4.2103,
"step": 423
},
{
"epoch": 0.71,
"learning_rate": 2.526e-05,
"loss": 4.1175,
"step": 424
},
{
"epoch": 0.71,
"learning_rate": 2.5319999999999998e-05,
"loss": 4.1346,
"step": 425
},
{
"epoch": 0.72,
"learning_rate": 2.538e-05,
"loss": 4.1944,
"step": 426
},
{
"epoch": 0.72,
"learning_rate": 2.544e-05,
"loss": 4.1165,
"step": 427
},
{
"epoch": 0.72,
"learning_rate": 2.55e-05,
"loss": 4.0602,
"step": 428
},
{
"epoch": 0.72,
"learning_rate": 2.556e-05,
"loss": 4.181,
"step": 429
},
{
"epoch": 0.72,
"learning_rate": 2.562e-05,
"loss": 4.131,
"step": 430
},
{
"epoch": 0.72,
"learning_rate": 2.568e-05,
"loss": 4.2406,
"step": 431
},
{
"epoch": 0.73,
"learning_rate": 2.574e-05,
"loss": 4.0029,
"step": 432
},
{
"epoch": 0.73,
"learning_rate": 2.58e-05,
"loss": 4.2316,
"step": 433
},
{
"epoch": 0.73,
"learning_rate": 2.586e-05,
"loss": 4.1736,
"step": 434
},
{
"epoch": 0.73,
"learning_rate": 2.592e-05,
"loss": 4.216,
"step": 435
},
{
"epoch": 0.73,
"learning_rate": 2.5980000000000002e-05,
"loss": 4.1108,
"step": 436
},
{
"epoch": 0.73,
"learning_rate": 2.604e-05,
"loss": 4.0038,
"step": 437
},
{
"epoch": 0.74,
"learning_rate": 2.61e-05,
"loss": 4.2346,
"step": 438
},
{
"epoch": 0.74,
"learning_rate": 2.616e-05,
"loss": 4.1184,
"step": 439
},
{
"epoch": 0.74,
"learning_rate": 2.622e-05,
"loss": 4.3308,
"step": 440
},
{
"epoch": 0.74,
"learning_rate": 2.628e-05,
"loss": 4.1971,
"step": 441
},
{
"epoch": 0.74,
"learning_rate": 2.6340000000000002e-05,
"loss": 4.3519,
"step": 442
},
{
"epoch": 0.74,
"learning_rate": 2.64e-05,
"loss": 4.27,
"step": 443
},
{
"epoch": 0.75,
"learning_rate": 2.646e-05,
"loss": 4.2281,
"step": 444
},
{
"epoch": 0.75,
"learning_rate": 2.652e-05,
"loss": 4.0387,
"step": 445
},
{
"epoch": 0.75,
"learning_rate": 2.658e-05,
"loss": 4.3265,
"step": 446
},
{
"epoch": 0.75,
"learning_rate": 2.6640000000000002e-05,
"loss": 4.259,
"step": 447
},
{
"epoch": 0.75,
"learning_rate": 2.6700000000000002e-05,
"loss": 4.1865,
"step": 448
},
{
"epoch": 0.75,
"learning_rate": 2.676e-05,
"loss": 4.3621,
"step": 449
},
{
"epoch": 0.76,
"learning_rate": 2.682e-05,
"loss": 4.3646,
"step": 450
},
{
"epoch": 0.76,
"learning_rate": 2.688e-05,
"loss": 4.1923,
"step": 451
},
{
"epoch": 0.76,
"learning_rate": 2.6940000000000003e-05,
"loss": 4.1942,
"step": 452
},
{
"epoch": 0.76,
"learning_rate": 2.7000000000000002e-05,
"loss": 4.1208,
"step": 453
},
{
"epoch": 0.76,
"learning_rate": 2.7060000000000002e-05,
"loss": 4.0062,
"step": 454
},
{
"epoch": 0.77,
"learning_rate": 2.712e-05,
"loss": 4.1999,
"step": 455
},
{
"epoch": 0.77,
"learning_rate": 2.718e-05,
"loss": 4.1307,
"step": 456
},
{
"epoch": 0.77,
"learning_rate": 2.724e-05,
"loss": 4.2131,
"step": 457
},
{
"epoch": 0.77,
"learning_rate": 2.7300000000000003e-05,
"loss": 4.2255,
"step": 458
},
{
"epoch": 0.77,
"learning_rate": 2.7360000000000002e-05,
"loss": 4.142,
"step": 459
},
{
"epoch": 0.77,
"learning_rate": 2.7420000000000002e-05,
"loss": 4.16,
"step": 460
},
{
"epoch": 0.78,
"learning_rate": 2.748e-05,
"loss": 4.0841,
"step": 461
},
{
"epoch": 0.78,
"learning_rate": 2.754e-05,
"loss": 4.081,
"step": 462
},
{
"epoch": 0.78,
"learning_rate": 2.7600000000000003e-05,
"loss": 4.0801,
"step": 463
},
{
"epoch": 0.78,
"learning_rate": 2.7660000000000003e-05,
"loss": 4.1737,
"step": 464
},
{
"epoch": 0.78,
"learning_rate": 2.7720000000000002e-05,
"loss": 4.1929,
"step": 465
},
{
"epoch": 0.78,
"learning_rate": 2.778e-05,
"loss": 4.1173,
"step": 466
},
{
"epoch": 0.79,
"learning_rate": 2.784e-05,
"loss": 4.1316,
"step": 467
},
{
"epoch": 0.79,
"learning_rate": 2.79e-05,
"loss": 4.1334,
"step": 468
},
{
"epoch": 0.79,
"learning_rate": 2.7960000000000003e-05,
"loss": 4.1245,
"step": 469
},
{
"epoch": 0.79,
"learning_rate": 2.8020000000000003e-05,
"loss": 4.1056,
"step": 470
},
{
"epoch": 0.79,
"learning_rate": 2.8080000000000002e-05,
"loss": 4.045,
"step": 471
},
{
"epoch": 0.79,
"learning_rate": 2.8139999999999998e-05,
"loss": 4.1275,
"step": 472
},
{
"epoch": 0.8,
"learning_rate": 2.8199999999999998e-05,
"loss": 4.126,
"step": 473
},
{
"epoch": 0.8,
"learning_rate": 2.826e-05,
"loss": 4.0065,
"step": 474
},
{
"epoch": 0.8,
"learning_rate": 2.832e-05,
"loss": 4.0684,
"step": 475
},
{
"epoch": 0.8,
"learning_rate": 2.838e-05,
"loss": 4.0925,
"step": 476
},
{
"epoch": 0.8,
"learning_rate": 2.844e-05,
"loss": 4.1197,
"step": 477
},
{
"epoch": 0.8,
"learning_rate": 2.8499999999999998e-05,
"loss": 4.1422,
"step": 478
},
{
"epoch": 0.81,
"learning_rate": 2.856e-05,
"loss": 4.0224,
"step": 479
},
{
"epoch": 0.81,
"learning_rate": 2.862e-05,
"loss": 4.0076,
"step": 480
},
{
"epoch": 0.81,
"learning_rate": 2.868e-05,
"loss": 4.2809,
"step": 481
},
{
"epoch": 0.81,
"learning_rate": 2.874e-05,
"loss": 4.0911,
"step": 482
},
{
"epoch": 0.81,
"learning_rate": 2.88e-05,
"loss": 4.2538,
"step": 483
},
{
"epoch": 0.81,
"learning_rate": 2.8859999999999998e-05,
"loss": 4.1925,
"step": 484
},
{
"epoch": 0.82,
"learning_rate": 2.892e-05,
"loss": 4.1767,
"step": 485
},
{
"epoch": 0.82,
"learning_rate": 2.898e-05,
"loss": 4.1299,
"step": 486
},
{
"epoch": 0.82,
"learning_rate": 2.904e-05,
"loss": 4.2235,
"step": 487
},
{
"epoch": 0.82,
"learning_rate": 2.91e-05,
"loss": 4.2217,
"step": 488
},
{
"epoch": 0.82,
"learning_rate": 2.916e-05,
"loss": 4.1539,
"step": 489
},
{
"epoch": 0.82,
"learning_rate": 2.922e-05,
"loss": 4.0713,
"step": 490
},
{
"epoch": 0.83,
"learning_rate": 2.928e-05,
"loss": 4.2944,
"step": 491
},
{
"epoch": 0.83,
"learning_rate": 2.934e-05,
"loss": 4.0968,
"step": 492
},
{
"epoch": 0.83,
"learning_rate": 2.94e-05,
"loss": 3.9586,
"step": 493
},
{
"epoch": 0.83,
"learning_rate": 2.946e-05,
"loss": 4.174,
"step": 494
},
{
"epoch": 0.83,
"learning_rate": 2.9520000000000002e-05,
"loss": 4.3646,
"step": 495
},
{
"epoch": 0.83,
"learning_rate": 2.958e-05,
"loss": 4.3606,
"step": 496
},
{
"epoch": 0.84,
"learning_rate": 2.964e-05,
"loss": 4.3115,
"step": 497
},
{
"epoch": 0.84,
"learning_rate": 2.97e-05,
"loss": 4.3132,
"step": 498
},
{
"epoch": 0.84,
"learning_rate": 2.976e-05,
"loss": 4.436,
"step": 499
},
{
"epoch": 0.84,
"learning_rate": 2.982e-05,
"loss": 4.3047,
"step": 500
},
{
"epoch": 0.84,
"eval_loss": 4.170833587646484,
"eval_runtime": 786.0537,
"eval_samples_per_second": 3.361,
"eval_steps_per_second": 0.281,
"eval_wer": 1.9509520031733438,
"step": 500
},
{
"epoch": 0.84,
"learning_rate": 2.9880000000000002e-05,
"loss": 4.2247,
"step": 501
},
{
"epoch": 0.84,
"learning_rate": 2.994e-05,
"loss": 4.0796,
"step": 502
},
{
"epoch": 0.85,
"learning_rate": 3e-05,
"loss": 4.064,
"step": 503
},
{
"epoch": 0.85,
"learning_rate": 2.968085106382979e-05,
"loss": 4.1893,
"step": 504
},
{
"epoch": 0.85,
"learning_rate": 2.9361702127659574e-05,
"loss": 4.125,
"step": 505
},
{
"epoch": 0.85,
"learning_rate": 2.9042553191489362e-05,
"loss": 4.256,
"step": 506
},
{
"epoch": 0.85,
"learning_rate": 2.872340425531915e-05,
"loss": 4.1286,
"step": 507
},
{
"epoch": 0.85,
"learning_rate": 2.8404255319148935e-05,
"loss": 4.1184,
"step": 508
},
{
"epoch": 0.86,
"learning_rate": 2.8085106382978723e-05,
"loss": 4.0473,
"step": 509
},
{
"epoch": 0.86,
"learning_rate": 2.776595744680851e-05,
"loss": 4.1273,
"step": 510
},
{
"epoch": 0.86,
"learning_rate": 2.74468085106383e-05,
"loss": 4.1812,
"step": 511
},
{
"epoch": 0.86,
"learning_rate": 2.7127659574468088e-05,
"loss": 4.0927,
"step": 512
},
{
"epoch": 0.86,
"learning_rate": 2.6808510638297873e-05,
"loss": 4.096,
"step": 513
},
{
"epoch": 0.86,
"learning_rate": 2.648936170212766e-05,
"loss": 4.0449,
"step": 514
},
{
"epoch": 0.87,
"learning_rate": 2.617021276595745e-05,
"loss": 4.1582,
"step": 515
},
{
"epoch": 0.87,
"learning_rate": 2.5851063829787234e-05,
"loss": 4.1007,
"step": 516
},
{
"epoch": 0.87,
"learning_rate": 2.5531914893617022e-05,
"loss": 4.1424,
"step": 517
},
{
"epoch": 0.87,
"learning_rate": 2.521276595744681e-05,
"loss": 4.2051,
"step": 518
},
{
"epoch": 0.87,
"learning_rate": 2.4893617021276595e-05,
"loss": 4.0328,
"step": 519
},
{
"epoch": 0.87,
"learning_rate": 2.4574468085106383e-05,
"loss": 4.1308,
"step": 520
},
{
"epoch": 0.88,
"learning_rate": 2.4255319148936168e-05,
"loss": 4.1889,
"step": 521
},
{
"epoch": 0.88,
"learning_rate": 2.3936170212765956e-05,
"loss": 4.2244,
"step": 522
},
{
"epoch": 0.88,
"learning_rate": 2.3617021276595744e-05,
"loss": 4.0055,
"step": 523
},
{
"epoch": 0.88,
"learning_rate": 2.3297872340425533e-05,
"loss": 3.9468,
"step": 524
},
{
"epoch": 0.88,
"learning_rate": 2.297872340425532e-05,
"loss": 4.0308,
"step": 525
},
{
"epoch": 0.88,
"learning_rate": 2.265957446808511e-05,
"loss": 4.0522,
"step": 526
},
{
"epoch": 0.89,
"learning_rate": 2.2340425531914894e-05,
"loss": 4.2188,
"step": 527
},
{
"epoch": 0.89,
"learning_rate": 2.2021276595744682e-05,
"loss": 4.0726,
"step": 528
},
{
"epoch": 0.89,
"learning_rate": 2.170212765957447e-05,
"loss": 4.0272,
"step": 529
},
{
"epoch": 0.89,
"learning_rate": 2.1382978723404255e-05,
"loss": 3.9968,
"step": 530
},
{
"epoch": 0.89,
"learning_rate": 2.1063829787234043e-05,
"loss": 4.0069,
"step": 531
},
{
"epoch": 0.89,
"learning_rate": 2.074468085106383e-05,
"loss": 4.2255,
"step": 532
},
{
"epoch": 0.9,
"learning_rate": 2.0425531914893616e-05,
"loss": 4.0596,
"step": 533
},
{
"epoch": 0.9,
"learning_rate": 2.0106382978723404e-05,
"loss": 4.0921,
"step": 534
},
{
"epoch": 0.9,
"learning_rate": 1.978723404255319e-05,
"loss": 4.0877,
"step": 535
},
{
"epoch": 0.9,
"learning_rate": 1.9468085106382977e-05,
"loss": 4.0089,
"step": 536
},
{
"epoch": 0.9,
"learning_rate": 1.914893617021277e-05,
"loss": 4.1551,
"step": 537
},
{
"epoch": 0.9,
"learning_rate": 1.8829787234042554e-05,
"loss": 4.1675,
"step": 538
},
{
"epoch": 0.91,
"learning_rate": 1.8510638297872342e-05,
"loss": 4.0228,
"step": 539
},
{
"epoch": 0.91,
"learning_rate": 1.819148936170213e-05,
"loss": 4.1119,
"step": 540
},
{
"epoch": 0.91,
"learning_rate": 1.7872340425531915e-05,
"loss": 4.075,
"step": 541
},
{
"epoch": 0.91,
"learning_rate": 1.7553191489361703e-05,
"loss": 4.1492,
"step": 542
},
{
"epoch": 0.91,
"learning_rate": 1.723404255319149e-05,
"loss": 4.249,
"step": 543
},
{
"epoch": 0.91,
"learning_rate": 1.6914893617021276e-05,
"loss": 4.1364,
"step": 544
},
{
"epoch": 0.92,
"learning_rate": 1.6595744680851064e-05,
"loss": 4.233,
"step": 545
},
{
"epoch": 0.92,
"learning_rate": 1.627659574468085e-05,
"loss": 4.4023,
"step": 546
},
{
"epoch": 0.92,
"learning_rate": 1.5957446808510637e-05,
"loss": 4.1616,
"step": 547
},
{
"epoch": 0.92,
"learning_rate": 1.5638297872340426e-05,
"loss": 4.3976,
"step": 548
},
{
"epoch": 0.92,
"learning_rate": 1.531914893617021e-05,
"loss": 4.6723,
"step": 549
},
{
"epoch": 0.92,
"learning_rate": 1.5e-05,
"loss": 4.292,
"step": 550
},
{
"epoch": 0.93,
"learning_rate": 1.4680851063829787e-05,
"loss": 4.1422,
"step": 551
},
{
"epoch": 0.93,
"learning_rate": 1.4361702127659575e-05,
"loss": 4.0512,
"step": 552
},
{
"epoch": 0.93,
"learning_rate": 1.4042553191489362e-05,
"loss": 4.0554,
"step": 553
},
{
"epoch": 0.93,
"learning_rate": 1.372340425531915e-05,
"loss": 4.0038,
"step": 554
},
{
"epoch": 0.93,
"learning_rate": 1.3404255319148936e-05,
"loss": 4.0264,
"step": 555
},
{
"epoch": 0.93,
"learning_rate": 1.3085106382978724e-05,
"loss": 4.0999,
"step": 556
},
{
"epoch": 0.94,
"learning_rate": 1.2765957446808511e-05,
"loss": 4.1626,
"step": 557
},
{
"epoch": 0.94,
"learning_rate": 1.2446808510638298e-05,
"loss": 4.1767,
"step": 558
},
{
"epoch": 0.94,
"learning_rate": 1.2127659574468084e-05,
"loss": 4.0166,
"step": 559
},
{
"epoch": 0.94,
"learning_rate": 1.1808510638297872e-05,
"loss": 4.0534,
"step": 560
},
{
"epoch": 0.94,
"learning_rate": 1.148936170212766e-05,
"loss": 4.3846,
"step": 561
},
{
"epoch": 0.94,
"learning_rate": 1.1170212765957447e-05,
"loss": 4.0051,
"step": 562
},
{
"epoch": 0.95,
"learning_rate": 1.0851063829787235e-05,
"loss": 4.0467,
"step": 563
},
{
"epoch": 0.95,
"learning_rate": 1.0531914893617022e-05,
"loss": 4.1727,
"step": 564
},
{
"epoch": 0.95,
"learning_rate": 1.0212765957446808e-05,
"loss": 4.1383,
"step": 565
},
{
"epoch": 0.95,
"learning_rate": 9.893617021276595e-06,
"loss": 4.0209,
"step": 566
},
{
"epoch": 0.95,
"learning_rate": 9.574468085106385e-06,
"loss": 4.0899,
"step": 567
},
{
"epoch": 0.96,
"learning_rate": 9.255319148936171e-06,
"loss": 4.0305,
"step": 568
},
{
"epoch": 0.96,
"learning_rate": 8.936170212765958e-06,
"loss": 3.9252,
"step": 569
},
{
"epoch": 0.96,
"learning_rate": 8.617021276595746e-06,
"loss": 4.0809,
"step": 570
},
{
"epoch": 0.96,
"learning_rate": 8.297872340425532e-06,
"loss": 4.069,
"step": 571
},
{
"epoch": 0.96,
"learning_rate": 7.978723404255319e-06,
"loss": 4.114,
"step": 572
},
{
"epoch": 0.96,
"learning_rate": 7.659574468085105e-06,
"loss": 4.1565,
"step": 573
},
{
"epoch": 0.97,
"learning_rate": 7.3404255319148934e-06,
"loss": 4.0973,
"step": 574
},
{
"epoch": 0.97,
"learning_rate": 7.021276595744681e-06,
"loss": 4.2048,
"step": 575
},
{
"epoch": 0.97,
"learning_rate": 6.702127659574468e-06,
"loss": 4.084,
"step": 576
},
{
"epoch": 0.97,
"learning_rate": 6.3829787234042555e-06,
"loss": 4.0916,
"step": 577
},
{
"epoch": 0.97,
"learning_rate": 6.063829787234042e-06,
"loss": 3.9743,
"step": 578
},
{
"epoch": 0.97,
"learning_rate": 5.74468085106383e-06,
"loss": 3.9492,
"step": 579
},
{
"epoch": 0.98,
"learning_rate": 5.4255319148936176e-06,
"loss": 4.209,
"step": 580
},
{
"epoch": 0.98,
"learning_rate": 5.106382978723404e-06,
"loss": 4.3008,
"step": 581
},
{
"epoch": 0.98,
"learning_rate": 4.787234042553192e-06,
"loss": 4.2066,
"step": 582
},
{
"epoch": 0.98,
"learning_rate": 4.468085106382979e-06,
"loss": 3.9929,
"step": 583
},
{
"epoch": 0.98,
"learning_rate": 4.148936170212766e-06,
"loss": 4.1489,
"step": 584
},
{
"epoch": 0.98,
"learning_rate": 3.829787234042553e-06,
"loss": 4.112,
"step": 585
},
{
"epoch": 0.99,
"learning_rate": 3.5106382978723404e-06,
"loss": 3.9953,
"step": 586
},
{
"epoch": 0.99,
"learning_rate": 3.1914893617021277e-06,
"loss": 4.188,
"step": 587
},
{
"epoch": 0.99,
"learning_rate": 2.872340425531915e-06,
"loss": 4.1903,
"step": 588
},
{
"epoch": 0.99,
"learning_rate": 2.553191489361702e-06,
"loss": 4.2876,
"step": 589
},
{
"epoch": 0.99,
"learning_rate": 2.2340425531914894e-06,
"loss": 4.1514,
"step": 590
},
{
"epoch": 0.99,
"learning_rate": 1.9148936170212763e-06,
"loss": 4.34,
"step": 591
},
{
"epoch": 1.0,
"learning_rate": 1.5957446808510639e-06,
"loss": 4.2715,
"step": 592
},
{
"epoch": 1.0,
"learning_rate": 1.276595744680851e-06,
"loss": 4.4617,
"step": 593
},
{
"epoch": 1.0,
"learning_rate": 9.574468085106382e-07,
"loss": 4.3369,
"step": 594
},
{
"epoch": 1.0,
"step": 594,
"total_flos": 0.0,
"train_loss": 4.265389429599749,
"train_runtime": 5681.1629,
"train_samples_per_second": 5.023,
"train_steps_per_second": 0.105
}
],
"max_steps": 594,
"num_train_epochs": 1,
"total_flos": 0.0,
"trial_name": null,
"trial_params": null
}