wav2vec2-2-gpt2-grid-search / trainer_state.json
sanchit-gandhi's picture
End of training
0aa5433
raw
history blame
No virus
69.1 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": 4.8251,
"step": 1
},
{
"epoch": 0.0,
"learning_rate": 6e-07,
"loss": 4.9877,
"step": 2
},
{
"epoch": 0.01,
"learning_rate": 1.2e-06,
"loss": 4.8754,
"step": 3
},
{
"epoch": 0.01,
"learning_rate": 1.8e-06,
"loss": 4.8216,
"step": 4
},
{
"epoch": 0.01,
"learning_rate": 2.4e-06,
"loss": 4.7633,
"step": 5
},
{
"epoch": 0.01,
"learning_rate": 2.9999999999999997e-06,
"loss": 4.7562,
"step": 6
},
{
"epoch": 0.01,
"learning_rate": 3.6e-06,
"loss": 4.8158,
"step": 7
},
{
"epoch": 0.01,
"learning_rate": 4.2e-06,
"loss": 4.6936,
"step": 8
},
{
"epoch": 0.02,
"learning_rate": 4.8e-06,
"loss": 4.7747,
"step": 9
},
{
"epoch": 0.02,
"learning_rate": 5.399999999999999e-06,
"loss": 4.6495,
"step": 10
},
{
"epoch": 0.02,
"learning_rate": 5.999999999999999e-06,
"loss": 4.6263,
"step": 11
},
{
"epoch": 0.02,
"learning_rate": 6.599999999999999e-06,
"loss": 4.6435,
"step": 12
},
{
"epoch": 0.02,
"learning_rate": 7.2e-06,
"loss": 4.5393,
"step": 13
},
{
"epoch": 0.02,
"learning_rate": 7.799999999999998e-06,
"loss": 4.4934,
"step": 14
},
{
"epoch": 0.03,
"learning_rate": 8.4e-06,
"loss": 4.3827,
"step": 15
},
{
"epoch": 0.03,
"learning_rate": 8.999999999999999e-06,
"loss": 4.528,
"step": 16
},
{
"epoch": 0.03,
"learning_rate": 9.6e-06,
"loss": 4.5121,
"step": 17
},
{
"epoch": 0.03,
"learning_rate": 1.02e-05,
"loss": 4.5744,
"step": 18
},
{
"epoch": 0.03,
"learning_rate": 1.0799999999999998e-05,
"loss": 4.3183,
"step": 19
},
{
"epoch": 0.03,
"learning_rate": 1.14e-05,
"loss": 4.5438,
"step": 20
},
{
"epoch": 0.04,
"learning_rate": 1.1999999999999999e-05,
"loss": 4.4622,
"step": 21
},
{
"epoch": 0.04,
"learning_rate": 1.26e-05,
"loss": 4.4327,
"step": 22
},
{
"epoch": 0.04,
"learning_rate": 1.3199999999999997e-05,
"loss": 4.4276,
"step": 23
},
{
"epoch": 0.04,
"learning_rate": 1.3799999999999998e-05,
"loss": 4.4719,
"step": 24
},
{
"epoch": 0.04,
"learning_rate": 1.44e-05,
"loss": 4.2464,
"step": 25
},
{
"epoch": 0.04,
"learning_rate": 1.4999999999999999e-05,
"loss": 4.664,
"step": 26
},
{
"epoch": 0.05,
"learning_rate": 1.5599999999999996e-05,
"loss": 4.3188,
"step": 27
},
{
"epoch": 0.05,
"learning_rate": 1.6199999999999997e-05,
"loss": 4.202,
"step": 28
},
{
"epoch": 0.05,
"learning_rate": 1.68e-05,
"loss": 4.226,
"step": 29
},
{
"epoch": 0.05,
"learning_rate": 1.74e-05,
"loss": 4.3916,
"step": 30
},
{
"epoch": 0.05,
"learning_rate": 1.7999999999999997e-05,
"loss": 4.4153,
"step": 31
},
{
"epoch": 0.05,
"learning_rate": 1.8599999999999998e-05,
"loss": 4.2902,
"step": 32
},
{
"epoch": 0.06,
"learning_rate": 1.92e-05,
"loss": 4.2797,
"step": 33
},
{
"epoch": 0.06,
"learning_rate": 1.98e-05,
"loss": 4.2387,
"step": 34
},
{
"epoch": 0.06,
"learning_rate": 2.04e-05,
"loss": 4.2934,
"step": 35
},
{
"epoch": 0.06,
"learning_rate": 2.1e-05,
"loss": 4.4521,
"step": 36
},
{
"epoch": 0.06,
"learning_rate": 2.1599999999999996e-05,
"loss": 4.2516,
"step": 37
},
{
"epoch": 0.06,
"learning_rate": 2.2199999999999998e-05,
"loss": 4.4088,
"step": 38
},
{
"epoch": 0.07,
"learning_rate": 2.28e-05,
"loss": 4.5498,
"step": 39
},
{
"epoch": 0.07,
"learning_rate": 2.34e-05,
"loss": 4.2121,
"step": 40
},
{
"epoch": 0.07,
"learning_rate": 2.3999999999999997e-05,
"loss": 4.2337,
"step": 41
},
{
"epoch": 0.07,
"learning_rate": 2.4599999999999998e-05,
"loss": 4.4752,
"step": 42
},
{
"epoch": 0.07,
"learning_rate": 2.52e-05,
"loss": 4.1934,
"step": 43
},
{
"epoch": 0.07,
"learning_rate": 2.5799999999999997e-05,
"loss": 4.3419,
"step": 44
},
{
"epoch": 0.08,
"learning_rate": 2.6399999999999995e-05,
"loss": 4.4241,
"step": 45
},
{
"epoch": 0.08,
"learning_rate": 2.6999999999999996e-05,
"loss": 4.4737,
"step": 46
},
{
"epoch": 0.08,
"learning_rate": 2.7599999999999997e-05,
"loss": 4.3846,
"step": 47
},
{
"epoch": 0.08,
"learning_rate": 2.8199999999999998e-05,
"loss": 4.7292,
"step": 48
},
{
"epoch": 0.08,
"learning_rate": 2.8199999999999998e-05,
"loss": 4.7684,
"step": 49
},
{
"epoch": 0.08,
"learning_rate": 2.88e-05,
"loss": 4.4745,
"step": 50
},
{
"epoch": 0.09,
"learning_rate": 2.94e-05,
"loss": 4.193,
"step": 51
},
{
"epoch": 0.09,
"learning_rate": 2.9999999999999997e-05,
"loss": 4.2115,
"step": 52
},
{
"epoch": 0.09,
"learning_rate": 3.06e-05,
"loss": 4.2324,
"step": 53
},
{
"epoch": 0.09,
"learning_rate": 3.119999999999999e-05,
"loss": 4.3067,
"step": 54
},
{
"epoch": 0.09,
"learning_rate": 3.1799999999999994e-05,
"loss": 4.167,
"step": 55
},
{
"epoch": 0.09,
"learning_rate": 3.2399999999999995e-05,
"loss": 4.3703,
"step": 56
},
{
"epoch": 0.1,
"learning_rate": 3.2999999999999996e-05,
"loss": 4.1812,
"step": 57
},
{
"epoch": 0.1,
"learning_rate": 3.36e-05,
"loss": 4.4778,
"step": 58
},
{
"epoch": 0.1,
"learning_rate": 3.42e-05,
"loss": 4.2518,
"step": 59
},
{
"epoch": 0.1,
"learning_rate": 3.48e-05,
"loss": 4.0458,
"step": 60
},
{
"epoch": 0.1,
"learning_rate": 3.539999999999999e-05,
"loss": 4.2996,
"step": 61
},
{
"epoch": 0.1,
"learning_rate": 3.5999999999999994e-05,
"loss": 4.3556,
"step": 62
},
{
"epoch": 0.11,
"learning_rate": 3.6599999999999995e-05,
"loss": 4.2308,
"step": 63
},
{
"epoch": 0.11,
"learning_rate": 3.7199999999999996e-05,
"loss": 4.2086,
"step": 64
},
{
"epoch": 0.11,
"learning_rate": 3.78e-05,
"loss": 4.2349,
"step": 65
},
{
"epoch": 0.11,
"learning_rate": 3.84e-05,
"loss": 4.1813,
"step": 66
},
{
"epoch": 0.11,
"learning_rate": 3.9e-05,
"loss": 4.1444,
"step": 67
},
{
"epoch": 0.11,
"learning_rate": 3.96e-05,
"loss": 4.1388,
"step": 68
},
{
"epoch": 0.12,
"learning_rate": 4.02e-05,
"loss": 4.2267,
"step": 69
},
{
"epoch": 0.12,
"learning_rate": 4.08e-05,
"loss": 4.2124,
"step": 70
},
{
"epoch": 0.12,
"learning_rate": 4.14e-05,
"loss": 4.3128,
"step": 71
},
{
"epoch": 0.12,
"learning_rate": 4.2e-05,
"loss": 4.2821,
"step": 72
},
{
"epoch": 0.12,
"learning_rate": 4.259999999999999e-05,
"loss": 4.0836,
"step": 73
},
{
"epoch": 0.12,
"learning_rate": 4.319999999999999e-05,
"loss": 4.2887,
"step": 74
},
{
"epoch": 0.13,
"learning_rate": 4.3799999999999994e-05,
"loss": 4.2636,
"step": 75
},
{
"epoch": 0.13,
"learning_rate": 4.4399999999999995e-05,
"loss": 4.2405,
"step": 76
},
{
"epoch": 0.13,
"learning_rate": 4.4999999999999996e-05,
"loss": 4.2182,
"step": 77
},
{
"epoch": 0.13,
"learning_rate": 4.56e-05,
"loss": 4.2002,
"step": 78
},
{
"epoch": 0.13,
"learning_rate": 4.62e-05,
"loss": 4.1308,
"step": 79
},
{
"epoch": 0.13,
"learning_rate": 4.68e-05,
"loss": 4.3264,
"step": 80
},
{
"epoch": 0.14,
"learning_rate": 4.7399999999999993e-05,
"loss": 4.2728,
"step": 81
},
{
"epoch": 0.14,
"learning_rate": 4.7999999999999994e-05,
"loss": 4.2262,
"step": 82
},
{
"epoch": 0.14,
"learning_rate": 4.8599999999999995e-05,
"loss": 4.3323,
"step": 83
},
{
"epoch": 0.14,
"learning_rate": 4.9199999999999997e-05,
"loss": 4.2886,
"step": 84
},
{
"epoch": 0.14,
"learning_rate": 4.98e-05,
"loss": 4.2094,
"step": 85
},
{
"epoch": 0.14,
"learning_rate": 5.04e-05,
"loss": 4.1616,
"step": 86
},
{
"epoch": 0.15,
"learning_rate": 5.1e-05,
"loss": 4.4014,
"step": 87
},
{
"epoch": 0.15,
"learning_rate": 5.1599999999999994e-05,
"loss": 4.2411,
"step": 88
},
{
"epoch": 0.15,
"learning_rate": 5.2199999999999995e-05,
"loss": 4.2378,
"step": 89
},
{
"epoch": 0.15,
"learning_rate": 5.279999999999999e-05,
"loss": 4.2782,
"step": 90
},
{
"epoch": 0.15,
"learning_rate": 5.339999999999999e-05,
"loss": 4.2438,
"step": 91
},
{
"epoch": 0.15,
"learning_rate": 5.399999999999999e-05,
"loss": 4.3678,
"step": 92
},
{
"epoch": 0.16,
"learning_rate": 5.459999999999999e-05,
"loss": 4.2209,
"step": 93
},
{
"epoch": 0.16,
"learning_rate": 5.519999999999999e-05,
"loss": 4.4583,
"step": 94
},
{
"epoch": 0.16,
"learning_rate": 5.5799999999999994e-05,
"loss": 4.3543,
"step": 95
},
{
"epoch": 0.16,
"learning_rate": 5.6399999999999995e-05,
"loss": 4.3987,
"step": 96
},
{
"epoch": 0.16,
"learning_rate": 5.6999999999999996e-05,
"loss": 4.3792,
"step": 97
},
{
"epoch": 0.16,
"learning_rate": 5.76e-05,
"loss": 4.2989,
"step": 98
},
{
"epoch": 0.17,
"learning_rate": 5.82e-05,
"loss": 4.5377,
"step": 99
},
{
"epoch": 0.17,
"learning_rate": 5.88e-05,
"loss": 4.1863,
"step": 100
},
{
"epoch": 0.17,
"learning_rate": 5.94e-05,
"loss": 4.3579,
"step": 101
},
{
"epoch": 0.17,
"learning_rate": 5.9999999999999995e-05,
"loss": 4.1935,
"step": 102
},
{
"epoch": 0.17,
"learning_rate": 6.0599999999999996e-05,
"loss": 4.206,
"step": 103
},
{
"epoch": 0.17,
"learning_rate": 6.12e-05,
"loss": 4.1263,
"step": 104
},
{
"epoch": 0.18,
"learning_rate": 6.18e-05,
"loss": 4.1788,
"step": 105
},
{
"epoch": 0.18,
"learning_rate": 6.239999999999999e-05,
"loss": 4.1409,
"step": 106
},
{
"epoch": 0.18,
"learning_rate": 6.299999999999999e-05,
"loss": 4.2692,
"step": 107
},
{
"epoch": 0.18,
"learning_rate": 6.359999999999999e-05,
"loss": 4.3054,
"step": 108
},
{
"epoch": 0.18,
"learning_rate": 6.419999999999999e-05,
"loss": 4.4333,
"step": 109
},
{
"epoch": 0.18,
"learning_rate": 6.479999999999999e-05,
"loss": 4.2029,
"step": 110
},
{
"epoch": 0.19,
"learning_rate": 6.539999999999999e-05,
"loss": 4.2235,
"step": 111
},
{
"epoch": 0.19,
"learning_rate": 6.599999999999999e-05,
"loss": 4.256,
"step": 112
},
{
"epoch": 0.19,
"learning_rate": 6.659999999999999e-05,
"loss": 4.2251,
"step": 113
},
{
"epoch": 0.19,
"learning_rate": 6.72e-05,
"loss": 4.2755,
"step": 114
},
{
"epoch": 0.19,
"learning_rate": 6.78e-05,
"loss": 4.269,
"step": 115
},
{
"epoch": 0.2,
"learning_rate": 6.84e-05,
"loss": 4.2454,
"step": 116
},
{
"epoch": 0.2,
"learning_rate": 6.9e-05,
"loss": 4.3376,
"step": 117
},
{
"epoch": 0.2,
"learning_rate": 6.96e-05,
"loss": 4.1226,
"step": 118
},
{
"epoch": 0.2,
"learning_rate": 7.02e-05,
"loss": 4.1287,
"step": 119
},
{
"epoch": 0.2,
"learning_rate": 7.079999999999999e-05,
"loss": 4.1267,
"step": 120
},
{
"epoch": 0.2,
"learning_rate": 7.139999999999999e-05,
"loss": 4.2983,
"step": 121
},
{
"epoch": 0.21,
"learning_rate": 7.199999999999999e-05,
"loss": 4.4717,
"step": 122
},
{
"epoch": 0.21,
"learning_rate": 7.259999999999999e-05,
"loss": 4.193,
"step": 123
},
{
"epoch": 0.21,
"learning_rate": 7.319999999999999e-05,
"loss": 4.2596,
"step": 124
},
{
"epoch": 0.21,
"learning_rate": 7.379999999999999e-05,
"loss": 4.2809,
"step": 125
},
{
"epoch": 0.21,
"learning_rate": 7.439999999999999e-05,
"loss": 4.1158,
"step": 126
},
{
"epoch": 0.21,
"learning_rate": 7.5e-05,
"loss": 4.1375,
"step": 127
},
{
"epoch": 0.22,
"learning_rate": 7.56e-05,
"loss": 4.1286,
"step": 128
},
{
"epoch": 0.22,
"learning_rate": 7.62e-05,
"loss": 4.2348,
"step": 129
},
{
"epoch": 0.22,
"learning_rate": 7.68e-05,
"loss": 4.2981,
"step": 130
},
{
"epoch": 0.22,
"learning_rate": 7.74e-05,
"loss": 4.2467,
"step": 131
},
{
"epoch": 0.22,
"learning_rate": 7.8e-05,
"loss": 4.2083,
"step": 132
},
{
"epoch": 0.22,
"learning_rate": 7.86e-05,
"loss": 4.2554,
"step": 133
},
{
"epoch": 0.23,
"learning_rate": 7.92e-05,
"loss": 4.3453,
"step": 134
},
{
"epoch": 0.23,
"learning_rate": 7.98e-05,
"loss": 4.3405,
"step": 135
},
{
"epoch": 0.23,
"learning_rate": 8.04e-05,
"loss": 4.2022,
"step": 136
},
{
"epoch": 0.23,
"learning_rate": 8.1e-05,
"loss": 4.182,
"step": 137
},
{
"epoch": 0.23,
"learning_rate": 8.16e-05,
"loss": 4.284,
"step": 138
},
{
"epoch": 0.23,
"learning_rate": 8.22e-05,
"loss": 4.224,
"step": 139
},
{
"epoch": 0.24,
"learning_rate": 8.28e-05,
"loss": 4.4008,
"step": 140
},
{
"epoch": 0.24,
"learning_rate": 8.34e-05,
"loss": 4.2829,
"step": 141
},
{
"epoch": 0.24,
"learning_rate": 8.4e-05,
"loss": 4.3976,
"step": 142
},
{
"epoch": 0.24,
"learning_rate": 8.459999999999998e-05,
"loss": 4.4763,
"step": 143
},
{
"epoch": 0.24,
"learning_rate": 8.519999999999998e-05,
"loss": 4.3198,
"step": 144
},
{
"epoch": 0.24,
"learning_rate": 8.579999999999998e-05,
"loss": 4.3517,
"step": 145
},
{
"epoch": 0.25,
"learning_rate": 8.639999999999999e-05,
"loss": 4.7129,
"step": 146
},
{
"epoch": 0.25,
"learning_rate": 8.699999999999999e-05,
"loss": 4.6215,
"step": 147
},
{
"epoch": 0.25,
"learning_rate": 8.759999999999999e-05,
"loss": 4.5996,
"step": 148
},
{
"epoch": 0.25,
"learning_rate": 8.819999999999999e-05,
"loss": 4.6987,
"step": 149
},
{
"epoch": 0.25,
"learning_rate": 8.879999999999999e-05,
"loss": 4.577,
"step": 150
},
{
"epoch": 0.25,
"learning_rate": 8.939999999999999e-05,
"loss": 4.3062,
"step": 151
},
{
"epoch": 0.26,
"learning_rate": 8.999999999999999e-05,
"loss": 4.2302,
"step": 152
},
{
"epoch": 0.26,
"learning_rate": 9.059999999999999e-05,
"loss": 4.2216,
"step": 153
},
{
"epoch": 0.26,
"learning_rate": 9.12e-05,
"loss": 4.12,
"step": 154
},
{
"epoch": 0.26,
"learning_rate": 9.18e-05,
"loss": 4.1832,
"step": 155
},
{
"epoch": 0.26,
"learning_rate": 9.24e-05,
"loss": 4.1754,
"step": 156
},
{
"epoch": 0.26,
"learning_rate": 9.3e-05,
"loss": 4.274,
"step": 157
},
{
"epoch": 0.27,
"learning_rate": 9.36e-05,
"loss": 4.1414,
"step": 158
},
{
"epoch": 0.27,
"learning_rate": 9.419999999999999e-05,
"loss": 4.0708,
"step": 159
},
{
"epoch": 0.27,
"learning_rate": 9.479999999999999e-05,
"loss": 4.2012,
"step": 160
},
{
"epoch": 0.27,
"learning_rate": 9.539999999999999e-05,
"loss": 4.2286,
"step": 161
},
{
"epoch": 0.27,
"learning_rate": 9.599999999999999e-05,
"loss": 4.1608,
"step": 162
},
{
"epoch": 0.27,
"learning_rate": 9.659999999999999e-05,
"loss": 4.1439,
"step": 163
},
{
"epoch": 0.28,
"learning_rate": 9.719999999999999e-05,
"loss": 4.1617,
"step": 164
},
{
"epoch": 0.28,
"learning_rate": 9.779999999999999e-05,
"loss": 4.2501,
"step": 165
},
{
"epoch": 0.28,
"learning_rate": 9.839999999999999e-05,
"loss": 4.322,
"step": 166
},
{
"epoch": 0.28,
"learning_rate": 9.9e-05,
"loss": 4.1637,
"step": 167
},
{
"epoch": 0.28,
"learning_rate": 9.96e-05,
"loss": 4.3875,
"step": 168
},
{
"epoch": 0.28,
"learning_rate": 0.0001002,
"loss": 4.1984,
"step": 169
},
{
"epoch": 0.29,
"learning_rate": 0.0001008,
"loss": 4.2334,
"step": 170
},
{
"epoch": 0.29,
"learning_rate": 0.0001014,
"loss": 4.0895,
"step": 171
},
{
"epoch": 0.29,
"learning_rate": 0.000102,
"loss": 4.1526,
"step": 172
},
{
"epoch": 0.29,
"learning_rate": 0.0001026,
"loss": 4.1554,
"step": 173
},
{
"epoch": 0.29,
"learning_rate": 0.00010319999999999999,
"loss": 4.174,
"step": 174
},
{
"epoch": 0.29,
"learning_rate": 0.00010379999999999999,
"loss": 4.3213,
"step": 175
},
{
"epoch": 0.3,
"learning_rate": 0.00010439999999999999,
"loss": 4.3565,
"step": 176
},
{
"epoch": 0.3,
"learning_rate": 0.00010499999999999999,
"loss": 4.2848,
"step": 177
},
{
"epoch": 0.3,
"learning_rate": 0.00010559999999999998,
"loss": 4.1378,
"step": 178
},
{
"epoch": 0.3,
"learning_rate": 0.00010619999999999998,
"loss": 4.2663,
"step": 179
},
{
"epoch": 0.3,
"learning_rate": 0.00010679999999999998,
"loss": 4.1552,
"step": 180
},
{
"epoch": 0.3,
"learning_rate": 0.00010739999999999998,
"loss": 4.4146,
"step": 181
},
{
"epoch": 0.31,
"learning_rate": 0.00010799999999999998,
"loss": 4.212,
"step": 182
},
{
"epoch": 0.31,
"learning_rate": 0.00010859999999999998,
"loss": 4.2198,
"step": 183
},
{
"epoch": 0.31,
"learning_rate": 0.00010919999999999998,
"loss": 4.0986,
"step": 184
},
{
"epoch": 0.31,
"learning_rate": 0.00010979999999999999,
"loss": 4.4803,
"step": 185
},
{
"epoch": 0.31,
"learning_rate": 0.00011039999999999999,
"loss": 4.2863,
"step": 186
},
{
"epoch": 0.31,
"learning_rate": 0.00011099999999999999,
"loss": 4.292,
"step": 187
},
{
"epoch": 0.32,
"learning_rate": 0.00011159999999999999,
"loss": 4.3438,
"step": 188
},
{
"epoch": 0.32,
"learning_rate": 0.00011219999999999999,
"loss": 4.4794,
"step": 189
},
{
"epoch": 0.32,
"learning_rate": 0.00011279999999999999,
"loss": 4.1904,
"step": 190
},
{
"epoch": 0.32,
"learning_rate": 0.00011339999999999999,
"loss": 4.4741,
"step": 191
},
{
"epoch": 0.32,
"learning_rate": 0.00011399999999999999,
"loss": 4.3827,
"step": 192
},
{
"epoch": 0.32,
"learning_rate": 0.0001146,
"loss": 4.5862,
"step": 193
},
{
"epoch": 0.33,
"learning_rate": 0.0001152,
"loss": 4.5182,
"step": 194
},
{
"epoch": 0.33,
"learning_rate": 0.0001158,
"loss": 4.306,
"step": 195
},
{
"epoch": 0.33,
"learning_rate": 0.0001164,
"loss": 4.3188,
"step": 196
},
{
"epoch": 0.33,
"learning_rate": 0.000117,
"loss": 4.7074,
"step": 197
},
{
"epoch": 0.33,
"learning_rate": 0.0001176,
"loss": 4.5491,
"step": 198
},
{
"epoch": 0.33,
"learning_rate": 0.0001182,
"loss": 4.8521,
"step": 199
},
{
"epoch": 0.34,
"learning_rate": 0.0001188,
"loss": 4.7298,
"step": 200
},
{
"epoch": 0.34,
"learning_rate": 0.0001194,
"loss": 4.2078,
"step": 201
},
{
"epoch": 0.34,
"learning_rate": 0.00011999999999999999,
"loss": 4.2913,
"step": 202
},
{
"epoch": 0.34,
"learning_rate": 0.00012059999999999999,
"loss": 4.4183,
"step": 203
},
{
"epoch": 0.34,
"learning_rate": 0.00012119999999999999,
"loss": 4.1983,
"step": 204
},
{
"epoch": 0.34,
"learning_rate": 0.00012179999999999999,
"loss": 4.2838,
"step": 205
},
{
"epoch": 0.35,
"learning_rate": 0.0001224,
"loss": 4.2377,
"step": 206
},
{
"epoch": 0.35,
"learning_rate": 0.00012299999999999998,
"loss": 4.3051,
"step": 207
},
{
"epoch": 0.35,
"learning_rate": 0.0001236,
"loss": 4.681,
"step": 208
},
{
"epoch": 0.35,
"learning_rate": 0.00012419999999999998,
"loss": 4.3501,
"step": 209
},
{
"epoch": 0.35,
"learning_rate": 0.00012479999999999997,
"loss": 4.3868,
"step": 210
},
{
"epoch": 0.35,
"learning_rate": 0.00012539999999999999,
"loss": 4.3917,
"step": 211
},
{
"epoch": 0.36,
"learning_rate": 0.00012599999999999997,
"loss": 4.3078,
"step": 212
},
{
"epoch": 0.36,
"learning_rate": 0.0001266,
"loss": 4.3063,
"step": 213
},
{
"epoch": 0.36,
"learning_rate": 0.00012719999999999997,
"loss": 4.3981,
"step": 214
},
{
"epoch": 0.36,
"learning_rate": 0.0001278,
"loss": 4.2747,
"step": 215
},
{
"epoch": 0.36,
"learning_rate": 0.00012839999999999998,
"loss": 4.3503,
"step": 216
},
{
"epoch": 0.36,
"learning_rate": 0.000129,
"loss": 4.182,
"step": 217
},
{
"epoch": 0.37,
"learning_rate": 0.00012959999999999998,
"loss": 4.437,
"step": 218
},
{
"epoch": 0.37,
"learning_rate": 0.0001302,
"loss": 4.2125,
"step": 219
},
{
"epoch": 0.37,
"learning_rate": 0.00013079999999999998,
"loss": 4.3465,
"step": 220
},
{
"epoch": 0.37,
"learning_rate": 0.0001314,
"loss": 4.3094,
"step": 221
},
{
"epoch": 0.37,
"learning_rate": 0.00013199999999999998,
"loss": 4.386,
"step": 222
},
{
"epoch": 0.37,
"learning_rate": 0.0001326,
"loss": 4.3321,
"step": 223
},
{
"epoch": 0.38,
"learning_rate": 0.00013319999999999999,
"loss": 4.351,
"step": 224
},
{
"epoch": 0.38,
"learning_rate": 0.0001338,
"loss": 4.6107,
"step": 225
},
{
"epoch": 0.38,
"learning_rate": 0.0001344,
"loss": 4.2463,
"step": 226
},
{
"epoch": 0.38,
"learning_rate": 0.000135,
"loss": 4.1284,
"step": 227
},
{
"epoch": 0.38,
"learning_rate": 0.0001356,
"loss": 4.2243,
"step": 228
},
{
"epoch": 0.39,
"learning_rate": 0.0001362,
"loss": 4.491,
"step": 229
},
{
"epoch": 0.39,
"learning_rate": 0.0001368,
"loss": 4.4179,
"step": 230
},
{
"epoch": 0.39,
"learning_rate": 0.0001374,
"loss": 4.1993,
"step": 231
},
{
"epoch": 0.39,
"learning_rate": 0.000138,
"loss": 4.291,
"step": 232
},
{
"epoch": 0.39,
"learning_rate": 0.0001386,
"loss": 4.7274,
"step": 233
},
{
"epoch": 0.39,
"learning_rate": 0.0001392,
"loss": 4.2559,
"step": 234
},
{
"epoch": 0.4,
"learning_rate": 0.00013979999999999998,
"loss": 4.5068,
"step": 235
},
{
"epoch": 0.4,
"learning_rate": 0.0001404,
"loss": 4.2703,
"step": 236
},
{
"epoch": 0.4,
"learning_rate": 0.00014099999999999998,
"loss": 4.2862,
"step": 237
},
{
"epoch": 0.4,
"learning_rate": 0.00014159999999999997,
"loss": 4.2675,
"step": 238
},
{
"epoch": 0.4,
"learning_rate": 0.0001422,
"loss": 4.3457,
"step": 239
},
{
"epoch": 0.4,
"learning_rate": 0.00014279999999999997,
"loss": 4.3633,
"step": 240
},
{
"epoch": 0.41,
"learning_rate": 0.0001434,
"loss": 4.5277,
"step": 241
},
{
"epoch": 0.41,
"learning_rate": 0.00014399999999999998,
"loss": 4.4545,
"step": 242
},
{
"epoch": 0.41,
"learning_rate": 0.0001446,
"loss": 4.3021,
"step": 243
},
{
"epoch": 0.41,
"learning_rate": 0.00014519999999999998,
"loss": 4.6181,
"step": 244
},
{
"epoch": 0.41,
"learning_rate": 0.0001458,
"loss": 4.681,
"step": 245
},
{
"epoch": 0.41,
"learning_rate": 0.00014639999999999998,
"loss": 4.5904,
"step": 246
},
{
"epoch": 0.42,
"learning_rate": 0.000147,
"loss": 4.819,
"step": 247
},
{
"epoch": 0.42,
"learning_rate": 0.00014759999999999998,
"loss": 4.7076,
"step": 248
},
{
"epoch": 0.42,
"learning_rate": 0.0001482,
"loss": 4.5274,
"step": 249
},
{
"epoch": 0.42,
"learning_rate": 0.00014879999999999998,
"loss": 4.6464,
"step": 250
},
{
"epoch": 0.42,
"learning_rate": 0.0001494,
"loss": 4.3735,
"step": 251
},
{
"epoch": 0.42,
"learning_rate": 0.00015,
"loss": 4.4111,
"step": 252
},
{
"epoch": 0.43,
"learning_rate": 0.00015059999999999997,
"loss": 4.5762,
"step": 253
},
{
"epoch": 0.43,
"learning_rate": 0.0001512,
"loss": 4.3417,
"step": 254
},
{
"epoch": 0.43,
"learning_rate": 0.00015179999999999998,
"loss": 4.5584,
"step": 255
},
{
"epoch": 0.43,
"learning_rate": 0.0001524,
"loss": 4.5966,
"step": 256
},
{
"epoch": 0.43,
"learning_rate": 0.00015299999999999998,
"loss": 4.3418,
"step": 257
},
{
"epoch": 0.43,
"learning_rate": 0.0001536,
"loss": 4.3593,
"step": 258
},
{
"epoch": 0.44,
"learning_rate": 0.00015419999999999998,
"loss": 4.3958,
"step": 259
},
{
"epoch": 0.44,
"learning_rate": 0.0001548,
"loss": 4.558,
"step": 260
},
{
"epoch": 0.44,
"learning_rate": 0.00015539999999999998,
"loss": 4.4078,
"step": 261
},
{
"epoch": 0.44,
"learning_rate": 0.000156,
"loss": 4.3026,
"step": 262
},
{
"epoch": 0.44,
"learning_rate": 0.00015659999999999998,
"loss": 4.6461,
"step": 263
},
{
"epoch": 0.44,
"learning_rate": 0.0001572,
"loss": 4.0745,
"step": 264
},
{
"epoch": 0.45,
"learning_rate": 0.0001578,
"loss": 4.3821,
"step": 265
},
{
"epoch": 0.45,
"learning_rate": 0.0001584,
"loss": 4.3248,
"step": 266
},
{
"epoch": 0.45,
"learning_rate": 0.000159,
"loss": 4.4178,
"step": 267
},
{
"epoch": 0.45,
"learning_rate": 0.0001596,
"loss": 4.3037,
"step": 268
},
{
"epoch": 0.45,
"learning_rate": 0.0001602,
"loss": 4.4218,
"step": 269
},
{
"epoch": 0.45,
"learning_rate": 0.0001608,
"loss": 4.3435,
"step": 270
},
{
"epoch": 0.46,
"learning_rate": 0.0001614,
"loss": 4.2526,
"step": 271
},
{
"epoch": 0.46,
"learning_rate": 0.000162,
"loss": 4.5418,
"step": 272
},
{
"epoch": 0.46,
"learning_rate": 0.0001626,
"loss": 4.4274,
"step": 273
},
{
"epoch": 0.46,
"learning_rate": 0.0001632,
"loss": 4.3375,
"step": 274
},
{
"epoch": 0.46,
"learning_rate": 0.0001638,
"loss": 4.4932,
"step": 275
},
{
"epoch": 0.46,
"learning_rate": 0.0001644,
"loss": 4.3847,
"step": 276
},
{
"epoch": 0.47,
"learning_rate": 0.000165,
"loss": 4.5048,
"step": 277
},
{
"epoch": 0.47,
"learning_rate": 0.0001656,
"loss": 4.4597,
"step": 278
},
{
"epoch": 0.47,
"learning_rate": 0.0001662,
"loss": 4.4021,
"step": 279
},
{
"epoch": 0.47,
"learning_rate": 0.0001668,
"loss": 4.4614,
"step": 280
},
{
"epoch": 0.47,
"learning_rate": 0.0001674,
"loss": 4.5178,
"step": 281
},
{
"epoch": 0.47,
"learning_rate": 0.000168,
"loss": 4.4294,
"step": 282
},
{
"epoch": 0.48,
"learning_rate": 0.0001686,
"loss": 4.3767,
"step": 283
},
{
"epoch": 0.48,
"learning_rate": 0.00016919999999999997,
"loss": 4.3318,
"step": 284
},
{
"epoch": 0.48,
"learning_rate": 0.00016979999999999998,
"loss": 4.594,
"step": 285
},
{
"epoch": 0.48,
"learning_rate": 0.00017039999999999997,
"loss": 4.2769,
"step": 286
},
{
"epoch": 0.48,
"learning_rate": 0.00017099999999999998,
"loss": 4.8188,
"step": 287
},
{
"epoch": 0.48,
"learning_rate": 0.00017159999999999997,
"loss": 4.1727,
"step": 288
},
{
"epoch": 0.49,
"learning_rate": 0.00017219999999999998,
"loss": 4.4774,
"step": 289
},
{
"epoch": 0.49,
"learning_rate": 0.00017279999999999997,
"loss": 4.6156,
"step": 290
},
{
"epoch": 0.49,
"learning_rate": 0.00017339999999999996,
"loss": 4.5715,
"step": 291
},
{
"epoch": 0.49,
"learning_rate": 0.00017399999999999997,
"loss": 4.3004,
"step": 292
},
{
"epoch": 0.49,
"learning_rate": 0.00017459999999999996,
"loss": 4.457,
"step": 293
},
{
"epoch": 0.49,
"learning_rate": 0.00017519999999999998,
"loss": 4.5484,
"step": 294
},
{
"epoch": 0.5,
"learning_rate": 0.00017579999999999996,
"loss": 4.6633,
"step": 295
},
{
"epoch": 0.5,
"learning_rate": 0.00017639999999999998,
"loss": 4.4057,
"step": 296
},
{
"epoch": 0.5,
"learning_rate": 0.00017699999999999997,
"loss": 5.0306,
"step": 297
},
{
"epoch": 0.5,
"learning_rate": 0.00017759999999999998,
"loss": 4.8782,
"step": 298
},
{
"epoch": 0.5,
"learning_rate": 0.00017819999999999997,
"loss": 4.7462,
"step": 299
},
{
"epoch": 0.5,
"learning_rate": 0.00017879999999999998,
"loss": 5.1954,
"step": 300
},
{
"epoch": 0.51,
"learning_rate": 0.00017939999999999997,
"loss": 4.601,
"step": 301
},
{
"epoch": 0.51,
"learning_rate": 0.00017999999999999998,
"loss": 4.5284,
"step": 302
},
{
"epoch": 0.51,
"learning_rate": 0.00018059999999999997,
"loss": 4.6252,
"step": 303
},
{
"epoch": 0.51,
"learning_rate": 0.00018119999999999999,
"loss": 4.521,
"step": 304
},
{
"epoch": 0.51,
"learning_rate": 0.00018179999999999997,
"loss": 4.5921,
"step": 305
},
{
"epoch": 0.51,
"learning_rate": 0.0001824,
"loss": 4.5414,
"step": 306
},
{
"epoch": 0.52,
"learning_rate": 0.00018299999999999998,
"loss": 4.5384,
"step": 307
},
{
"epoch": 0.52,
"learning_rate": 0.0001836,
"loss": 4.7024,
"step": 308
},
{
"epoch": 0.52,
"learning_rate": 0.00018419999999999998,
"loss": 4.5066,
"step": 309
},
{
"epoch": 0.52,
"learning_rate": 0.0001848,
"loss": 4.4578,
"step": 310
},
{
"epoch": 0.52,
"learning_rate": 0.00018539999999999998,
"loss": 4.2657,
"step": 311
},
{
"epoch": 0.52,
"learning_rate": 0.000186,
"loss": 4.5986,
"step": 312
},
{
"epoch": 0.53,
"learning_rate": 0.00018659999999999998,
"loss": 4.5689,
"step": 313
},
{
"epoch": 0.53,
"learning_rate": 0.0001872,
"loss": 4.624,
"step": 314
},
{
"epoch": 0.53,
"learning_rate": 0.00018779999999999998,
"loss": 4.3777,
"step": 315
},
{
"epoch": 0.53,
"learning_rate": 0.00018839999999999997,
"loss": 4.6881,
"step": 316
},
{
"epoch": 0.53,
"learning_rate": 0.00018899999999999999,
"loss": 4.4102,
"step": 317
},
{
"epoch": 0.53,
"learning_rate": 0.00018959999999999997,
"loss": 4.8632,
"step": 318
},
{
"epoch": 0.54,
"learning_rate": 0.0001902,
"loss": 4.5565,
"step": 319
},
{
"epoch": 0.54,
"learning_rate": 0.00019079999999999998,
"loss": 4.6379,
"step": 320
},
{
"epoch": 0.54,
"learning_rate": 0.0001914,
"loss": 4.2919,
"step": 321
},
{
"epoch": 0.54,
"learning_rate": 0.00019199999999999998,
"loss": 4.4188,
"step": 322
},
{
"epoch": 0.54,
"learning_rate": 0.0001926,
"loss": 4.6448,
"step": 323
},
{
"epoch": 0.54,
"learning_rate": 0.00019319999999999998,
"loss": 4.4343,
"step": 324
},
{
"epoch": 0.55,
"learning_rate": 0.0001938,
"loss": 4.8242,
"step": 325
},
{
"epoch": 0.55,
"learning_rate": 0.00019439999999999998,
"loss": 4.3897,
"step": 326
},
{
"epoch": 0.55,
"learning_rate": 0.000195,
"loss": 4.7531,
"step": 327
},
{
"epoch": 0.55,
"learning_rate": 0.00019559999999999998,
"loss": 4.5004,
"step": 328
},
{
"epoch": 0.55,
"learning_rate": 0.0001962,
"loss": 4.3962,
"step": 329
},
{
"epoch": 0.55,
"learning_rate": 0.00019679999999999999,
"loss": 4.4023,
"step": 330
},
{
"epoch": 0.56,
"learning_rate": 0.0001974,
"loss": 4.5184,
"step": 331
},
{
"epoch": 0.56,
"learning_rate": 0.000198,
"loss": 4.3142,
"step": 332
},
{
"epoch": 0.56,
"learning_rate": 0.0001986,
"loss": 4.5157,
"step": 333
},
{
"epoch": 0.56,
"learning_rate": 0.0001992,
"loss": 4.5152,
"step": 334
},
{
"epoch": 0.56,
"learning_rate": 0.0001998,
"loss": 4.4877,
"step": 335
},
{
"epoch": 0.56,
"learning_rate": 0.0002004,
"loss": 4.44,
"step": 336
},
{
"epoch": 0.57,
"learning_rate": 0.000201,
"loss": 4.6629,
"step": 337
},
{
"epoch": 0.57,
"learning_rate": 0.0002016,
"loss": 4.4032,
"step": 338
},
{
"epoch": 0.57,
"learning_rate": 0.0002022,
"loss": 4.4355,
"step": 339
},
{
"epoch": 0.57,
"learning_rate": 0.0002028,
"loss": 4.5324,
"step": 340
},
{
"epoch": 0.57,
"learning_rate": 0.00020339999999999998,
"loss": 4.4673,
"step": 341
},
{
"epoch": 0.58,
"learning_rate": 0.000204,
"loss": 4.8003,
"step": 342
},
{
"epoch": 0.58,
"learning_rate": 0.00020459999999999999,
"loss": 4.6395,
"step": 343
},
{
"epoch": 0.58,
"learning_rate": 0.0002052,
"loss": 4.6675,
"step": 344
},
{
"epoch": 0.58,
"learning_rate": 0.0002058,
"loss": 4.8611,
"step": 345
},
{
"epoch": 0.58,
"learning_rate": 0.00020639999999999998,
"loss": 4.363,
"step": 346
},
{
"epoch": 0.58,
"learning_rate": 0.00020699999999999996,
"loss": 4.3723,
"step": 347
},
{
"epoch": 0.59,
"learning_rate": 0.00020759999999999998,
"loss": 4.7011,
"step": 348
},
{
"epoch": 0.59,
"learning_rate": 0.00020819999999999996,
"loss": 4.5094,
"step": 349
},
{
"epoch": 0.59,
"learning_rate": 0.00020879999999999998,
"loss": 4.6712,
"step": 350
},
{
"epoch": 0.59,
"learning_rate": 0.00020939999999999997,
"loss": 4.629,
"step": 351
},
{
"epoch": 0.59,
"learning_rate": 0.00020999999999999998,
"loss": 4.5946,
"step": 352
},
{
"epoch": 0.59,
"learning_rate": 0.00021059999999999997,
"loss": 4.517,
"step": 353
},
{
"epoch": 0.6,
"learning_rate": 0.00021119999999999996,
"loss": 4.4386,
"step": 354
},
{
"epoch": 0.6,
"learning_rate": 0.00021179999999999997,
"loss": 4.5275,
"step": 355
},
{
"epoch": 0.6,
"learning_rate": 0.00021239999999999996,
"loss": 4.3692,
"step": 356
},
{
"epoch": 0.6,
"learning_rate": 0.00021299999999999997,
"loss": 4.5466,
"step": 357
},
{
"epoch": 0.6,
"learning_rate": 0.00021359999999999996,
"loss": 4.9401,
"step": 358
},
{
"epoch": 0.6,
"learning_rate": 0.00021419999999999998,
"loss": 4.3218,
"step": 359
},
{
"epoch": 0.61,
"learning_rate": 0.00021479999999999996,
"loss": 4.6933,
"step": 360
},
{
"epoch": 0.61,
"learning_rate": 0.00021539999999999998,
"loss": 4.6379,
"step": 361
},
{
"epoch": 0.61,
"learning_rate": 0.00021599999999999996,
"loss": 4.377,
"step": 362
},
{
"epoch": 0.61,
"learning_rate": 0.00021659999999999998,
"loss": 4.898,
"step": 363
},
{
"epoch": 0.61,
"learning_rate": 0.00021719999999999997,
"loss": 4.4985,
"step": 364
},
{
"epoch": 0.61,
"learning_rate": 0.00021779999999999998,
"loss": 4.5904,
"step": 365
},
{
"epoch": 0.62,
"learning_rate": 0.00021839999999999997,
"loss": 4.4725,
"step": 366
},
{
"epoch": 0.62,
"learning_rate": 0.00021899999999999998,
"loss": 4.4717,
"step": 367
},
{
"epoch": 0.62,
"learning_rate": 0.00021959999999999997,
"loss": 4.3563,
"step": 368
},
{
"epoch": 0.62,
"learning_rate": 0.00022019999999999999,
"loss": 4.4058,
"step": 369
},
{
"epoch": 0.62,
"learning_rate": 0.00022079999999999997,
"loss": 4.7998,
"step": 370
},
{
"epoch": 0.62,
"learning_rate": 0.0002214,
"loss": 4.5314,
"step": 371
},
{
"epoch": 0.63,
"learning_rate": 0.00022199999999999998,
"loss": 4.398,
"step": 372
},
{
"epoch": 0.63,
"learning_rate": 0.0002226,
"loss": 4.4327,
"step": 373
},
{
"epoch": 0.63,
"learning_rate": 0.00022319999999999998,
"loss": 4.7163,
"step": 374
},
{
"epoch": 0.63,
"learning_rate": 0.0002238,
"loss": 4.7412,
"step": 375
},
{
"epoch": 0.63,
"learning_rate": 0.00022439999999999998,
"loss": 4.6339,
"step": 376
},
{
"epoch": 0.63,
"learning_rate": 0.000225,
"loss": 4.5151,
"step": 377
},
{
"epoch": 0.64,
"learning_rate": 0.00022559999999999998,
"loss": 4.7596,
"step": 378
},
{
"epoch": 0.64,
"learning_rate": 0.00022619999999999997,
"loss": 4.3111,
"step": 379
},
{
"epoch": 0.64,
"learning_rate": 0.00022679999999999998,
"loss": 4.5822,
"step": 380
},
{
"epoch": 0.64,
"learning_rate": 0.00022739999999999997,
"loss": 4.573,
"step": 381
},
{
"epoch": 0.64,
"learning_rate": 0.00022799999999999999,
"loss": 4.3831,
"step": 382
},
{
"epoch": 0.64,
"learning_rate": 0.00022859999999999997,
"loss": 4.5107,
"step": 383
},
{
"epoch": 0.65,
"learning_rate": 0.0002292,
"loss": 4.4021,
"step": 384
},
{
"epoch": 0.65,
"learning_rate": 0.00022979999999999997,
"loss": 4.7495,
"step": 385
},
{
"epoch": 0.65,
"learning_rate": 0.0002304,
"loss": 4.8891,
"step": 386
},
{
"epoch": 0.65,
"learning_rate": 0.00023099999999999998,
"loss": 5.0482,
"step": 387
},
{
"epoch": 0.65,
"learning_rate": 0.0002316,
"loss": 4.5507,
"step": 388
},
{
"epoch": 0.65,
"learning_rate": 0.00023219999999999998,
"loss": 4.5954,
"step": 389
},
{
"epoch": 0.66,
"learning_rate": 0.0002328,
"loss": 4.4654,
"step": 390
},
{
"epoch": 0.66,
"learning_rate": 0.00023339999999999998,
"loss": 4.5147,
"step": 391
},
{
"epoch": 0.66,
"learning_rate": 0.000234,
"loss": 4.6645,
"step": 392
},
{
"epoch": 0.66,
"learning_rate": 0.00023459999999999998,
"loss": 4.6974,
"step": 393
},
{
"epoch": 0.66,
"learning_rate": 0.0002352,
"loss": 4.4298,
"step": 394
},
{
"epoch": 0.66,
"learning_rate": 0.00023579999999999999,
"loss": 4.8421,
"step": 395
},
{
"epoch": 0.67,
"learning_rate": 0.0002364,
"loss": 4.6812,
"step": 396
},
{
"epoch": 0.67,
"learning_rate": 0.000237,
"loss": 5.1311,
"step": 397
},
{
"epoch": 0.67,
"learning_rate": 0.0002376,
"loss": 5.5849,
"step": 398
},
{
"epoch": 0.67,
"learning_rate": 0.0002382,
"loss": 4.9822,
"step": 399
},
{
"epoch": 0.67,
"learning_rate": 0.0002388,
"loss": 5.4755,
"step": 400
},
{
"epoch": 0.67,
"learning_rate": 0.0002394,
"loss": 5.0482,
"step": 401
},
{
"epoch": 0.68,
"learning_rate": 0.00023999999999999998,
"loss": 4.7492,
"step": 402
},
{
"epoch": 0.68,
"learning_rate": 0.0002406,
"loss": 4.5452,
"step": 403
},
{
"epoch": 0.68,
"learning_rate": 0.00024119999999999998,
"loss": 4.8564,
"step": 404
},
{
"epoch": 0.68,
"learning_rate": 0.0002418,
"loss": 4.568,
"step": 405
},
{
"epoch": 0.68,
"learning_rate": 0.00024239999999999998,
"loss": 4.5792,
"step": 406
},
{
"epoch": 0.68,
"learning_rate": 0.000243,
"loss": 4.8879,
"step": 407
},
{
"epoch": 0.69,
"learning_rate": 0.00024359999999999999,
"loss": 4.5071,
"step": 408
},
{
"epoch": 0.69,
"learning_rate": 0.00024419999999999997,
"loss": 4.9798,
"step": 409
},
{
"epoch": 0.69,
"learning_rate": 0.0002448,
"loss": 4.5812,
"step": 410
},
{
"epoch": 0.69,
"learning_rate": 0.00024539999999999995,
"loss": 4.4708,
"step": 411
},
{
"epoch": 0.69,
"learning_rate": 0.00024599999999999996,
"loss": 4.7339,
"step": 412
},
{
"epoch": 0.69,
"learning_rate": 0.0002466,
"loss": 4.467,
"step": 413
},
{
"epoch": 0.7,
"learning_rate": 0.0002472,
"loss": 4.7394,
"step": 414
},
{
"epoch": 0.7,
"learning_rate": 0.00024779999999999995,
"loss": 4.5679,
"step": 415
},
{
"epoch": 0.7,
"learning_rate": 0.00024839999999999997,
"loss": 4.5506,
"step": 416
},
{
"epoch": 0.7,
"learning_rate": 0.000249,
"loss": 4.5689,
"step": 417
},
{
"epoch": 0.7,
"learning_rate": 0.00024959999999999994,
"loss": 4.1599,
"step": 418
},
{
"epoch": 0.7,
"learning_rate": 0.00025019999999999996,
"loss": 4.8572,
"step": 419
},
{
"epoch": 0.71,
"learning_rate": 0.00025079999999999997,
"loss": 4.722,
"step": 420
},
{
"epoch": 0.71,
"learning_rate": 0.0002514,
"loss": 4.2483,
"step": 421
},
{
"epoch": 0.71,
"learning_rate": 0.00025199999999999995,
"loss": 4.3537,
"step": 422
},
{
"epoch": 0.71,
"learning_rate": 0.00025259999999999996,
"loss": 4.6733,
"step": 423
},
{
"epoch": 0.71,
"learning_rate": 0.0002532,
"loss": 4.4947,
"step": 424
},
{
"epoch": 0.71,
"learning_rate": 0.0002538,
"loss": 4.6257,
"step": 425
},
{
"epoch": 0.72,
"learning_rate": 0.00025439999999999995,
"loss": 4.5752,
"step": 426
},
{
"epoch": 0.72,
"learning_rate": 0.00025499999999999996,
"loss": 4.286,
"step": 427
},
{
"epoch": 0.72,
"learning_rate": 0.0002556,
"loss": 4.2402,
"step": 428
},
{
"epoch": 0.72,
"learning_rate": 0.0002562,
"loss": 4.8508,
"step": 429
},
{
"epoch": 0.72,
"learning_rate": 0.00025679999999999995,
"loss": 4.4226,
"step": 430
},
{
"epoch": 0.72,
"learning_rate": 0.00025739999999999997,
"loss": 4.8964,
"step": 431
},
{
"epoch": 0.73,
"learning_rate": 0.000258,
"loss": 4.3863,
"step": 432
},
{
"epoch": 0.73,
"learning_rate": 0.0002586,
"loss": 4.5225,
"step": 433
},
{
"epoch": 0.73,
"learning_rate": 0.00025919999999999996,
"loss": 4.6005,
"step": 434
},
{
"epoch": 0.73,
"learning_rate": 0.00025979999999999997,
"loss": 4.6333,
"step": 435
},
{
"epoch": 0.73,
"learning_rate": 0.0002604,
"loss": 4.4055,
"step": 436
},
{
"epoch": 0.73,
"learning_rate": 0.000261,
"loss": 4.5249,
"step": 437
},
{
"epoch": 0.74,
"learning_rate": 0.00026159999999999996,
"loss": 4.528,
"step": 438
},
{
"epoch": 0.74,
"learning_rate": 0.0002622,
"loss": 4.4824,
"step": 439
},
{
"epoch": 0.74,
"learning_rate": 0.0002628,
"loss": 4.9304,
"step": 440
},
{
"epoch": 0.74,
"learning_rate": 0.00026339999999999995,
"loss": 4.9839,
"step": 441
},
{
"epoch": 0.74,
"learning_rate": 0.00026399999999999997,
"loss": 4.8169,
"step": 442
},
{
"epoch": 0.74,
"learning_rate": 0.0002646,
"loss": 4.9251,
"step": 443
},
{
"epoch": 0.75,
"learning_rate": 0.0002652,
"loss": 4.7229,
"step": 444
},
{
"epoch": 0.75,
"learning_rate": 0.00026579999999999996,
"loss": 4.453,
"step": 445
},
{
"epoch": 0.75,
"learning_rate": 0.00026639999999999997,
"loss": 4.8808,
"step": 446
},
{
"epoch": 0.75,
"learning_rate": 0.000267,
"loss": 4.5888,
"step": 447
},
{
"epoch": 0.75,
"learning_rate": 0.0002676,
"loss": 4.671,
"step": 448
},
{
"epoch": 0.75,
"learning_rate": 0.00026819999999999996,
"loss": 5.2919,
"step": 449
},
{
"epoch": 0.76,
"learning_rate": 0.0002688,
"loss": 4.4541,
"step": 450
},
{
"epoch": 0.76,
"learning_rate": 0.0002694,
"loss": 4.5015,
"step": 451
},
{
"epoch": 0.76,
"learning_rate": 0.00027,
"loss": 4.6347,
"step": 452
},
{
"epoch": 0.76,
"learning_rate": 0.00027059999999999996,
"loss": 4.4529,
"step": 453
},
{
"epoch": 0.76,
"learning_rate": 0.0002712,
"loss": 4.7072,
"step": 454
},
{
"epoch": 0.77,
"learning_rate": 0.0002718,
"loss": 4.6178,
"step": 455
},
{
"epoch": 0.77,
"learning_rate": 0.0002724,
"loss": 4.7541,
"step": 456
},
{
"epoch": 0.77,
"learning_rate": 0.00027299999999999997,
"loss": 4.8252,
"step": 457
},
{
"epoch": 0.77,
"learning_rate": 0.0002736,
"loss": 4.7205,
"step": 458
},
{
"epoch": 0.77,
"learning_rate": 0.0002742,
"loss": 4.7718,
"step": 459
},
{
"epoch": 0.77,
"learning_rate": 0.0002748,
"loss": 4.5784,
"step": 460
},
{
"epoch": 0.78,
"learning_rate": 0.00027539999999999997,
"loss": 4.7469,
"step": 461
},
{
"epoch": 0.78,
"learning_rate": 0.000276,
"loss": 4.4427,
"step": 462
},
{
"epoch": 0.78,
"learning_rate": 0.0002766,
"loss": 4.5062,
"step": 463
},
{
"epoch": 0.78,
"learning_rate": 0.0002772,
"loss": 4.6286,
"step": 464
},
{
"epoch": 0.78,
"learning_rate": 0.0002778,
"loss": 4.8309,
"step": 465
},
{
"epoch": 0.78,
"learning_rate": 0.0002784,
"loss": 4.5841,
"step": 466
},
{
"epoch": 0.79,
"learning_rate": 0.000279,
"loss": 4.581,
"step": 467
},
{
"epoch": 0.79,
"learning_rate": 0.00027959999999999997,
"loss": 4.5023,
"step": 468
},
{
"epoch": 0.79,
"learning_rate": 0.0002802,
"loss": 4.5518,
"step": 469
},
{
"epoch": 0.79,
"learning_rate": 0.0002808,
"loss": 4.3897,
"step": 470
},
{
"epoch": 0.79,
"learning_rate": 0.00028139999999999996,
"loss": 4.4474,
"step": 471
},
{
"epoch": 0.79,
"learning_rate": 0.00028199999999999997,
"loss": 4.4435,
"step": 472
},
{
"epoch": 0.8,
"learning_rate": 0.0002826,
"loss": 4.5926,
"step": 473
},
{
"epoch": 0.8,
"learning_rate": 0.00028319999999999994,
"loss": 4.5936,
"step": 474
},
{
"epoch": 0.8,
"learning_rate": 0.00028379999999999996,
"loss": 4.5352,
"step": 475
},
{
"epoch": 0.8,
"learning_rate": 0.0002844,
"loss": 4.7309,
"step": 476
},
{
"epoch": 0.8,
"learning_rate": 0.000285,
"loss": 4.6586,
"step": 477
},
{
"epoch": 0.8,
"learning_rate": 0.00028559999999999995,
"loss": 4.6348,
"step": 478
},
{
"epoch": 0.81,
"learning_rate": 0.00028619999999999996,
"loss": 4.3622,
"step": 479
},
{
"epoch": 0.81,
"learning_rate": 0.0002868,
"loss": 4.3913,
"step": 480
},
{
"epoch": 0.81,
"learning_rate": 0.00028739999999999994,
"loss": 4.8266,
"step": 481
},
{
"epoch": 0.81,
"learning_rate": 0.00028799999999999995,
"loss": 4.8246,
"step": 482
},
{
"epoch": 0.81,
"learning_rate": 0.00028859999999999997,
"loss": 4.9268,
"step": 483
},
{
"epoch": 0.81,
"learning_rate": 0.0002892,
"loss": 4.881,
"step": 484
},
{
"epoch": 0.82,
"learning_rate": 0.00028979999999999994,
"loss": 4.6537,
"step": 485
},
{
"epoch": 0.82,
"learning_rate": 0.00029039999999999996,
"loss": 4.5566,
"step": 486
},
{
"epoch": 0.82,
"learning_rate": 0.00029099999999999997,
"loss": 4.5838,
"step": 487
},
{
"epoch": 0.82,
"learning_rate": 0.0002916,
"loss": 4.5166,
"step": 488
},
{
"epoch": 0.82,
"learning_rate": 0.00029219999999999995,
"loss": 4.6484,
"step": 489
},
{
"epoch": 0.82,
"learning_rate": 0.00029279999999999996,
"loss": 4.3926,
"step": 490
},
{
"epoch": 0.83,
"learning_rate": 0.0002934,
"loss": 4.9063,
"step": 491
},
{
"epoch": 0.83,
"learning_rate": 0.000294,
"loss": 4.5744,
"step": 492
},
{
"epoch": 0.83,
"learning_rate": 0.00029459999999999995,
"loss": 4.2873,
"step": 493
},
{
"epoch": 0.83,
"learning_rate": 0.00029519999999999997,
"loss": 4.6052,
"step": 494
},
{
"epoch": 0.83,
"learning_rate": 0.0002958,
"loss": 4.7127,
"step": 495
},
{
"epoch": 0.83,
"learning_rate": 0.0002964,
"loss": 4.8242,
"step": 496
},
{
"epoch": 0.84,
"learning_rate": 0.00029699999999999996,
"loss": 4.4981,
"step": 497
},
{
"epoch": 0.84,
"learning_rate": 0.00029759999999999997,
"loss": 4.5186,
"step": 498
},
{
"epoch": 0.84,
"learning_rate": 0.0002982,
"loss": 5.0847,
"step": 499
},
{
"epoch": 0.84,
"learning_rate": 0.0002988,
"loss": 4.7686,
"step": 500
},
{
"epoch": 0.84,
"eval_loss": 4.900937080383301,
"eval_runtime": 769.6154,
"eval_samples_per_second": 3.433,
"eval_steps_per_second": 0.287,
"eval_wer": 1.9500793335977786,
"step": 500
},
{
"epoch": 0.84,
"learning_rate": 0.00029939999999999996,
"loss": 4.7283,
"step": 501
},
{
"epoch": 0.84,
"learning_rate": 0.0003,
"loss": 4.4173,
"step": 502
},
{
"epoch": 0.85,
"learning_rate": 0.00029680851063829784,
"loss": 4.5777,
"step": 503
},
{
"epoch": 0.85,
"learning_rate": 0.0002936170212765957,
"loss": 4.8135,
"step": 504
},
{
"epoch": 0.85,
"learning_rate": 0.0002904255319148936,
"loss": 4.7635,
"step": 505
},
{
"epoch": 0.85,
"learning_rate": 0.0002872340425531915,
"loss": 4.7398,
"step": 506
},
{
"epoch": 0.85,
"learning_rate": 0.00028404255319148934,
"loss": 4.7005,
"step": 507
},
{
"epoch": 0.85,
"learning_rate": 0.0002808510638297872,
"loss": 4.7717,
"step": 508
},
{
"epoch": 0.86,
"learning_rate": 0.00027765957446808506,
"loss": 4.5064,
"step": 509
},
{
"epoch": 0.86,
"learning_rate": 0.000274468085106383,
"loss": 4.806,
"step": 510
},
{
"epoch": 0.86,
"learning_rate": 0.00027127659574468084,
"loss": 4.6994,
"step": 511
},
{
"epoch": 0.86,
"learning_rate": 0.0002680851063829787,
"loss": 4.5931,
"step": 512
},
{
"epoch": 0.86,
"learning_rate": 0.00026489361702127656,
"loss": 4.4888,
"step": 513
},
{
"epoch": 0.86,
"learning_rate": 0.0002617021276595745,
"loss": 4.4274,
"step": 514
},
{
"epoch": 0.87,
"learning_rate": 0.0002585106382978723,
"loss": 4.5197,
"step": 515
},
{
"epoch": 0.87,
"learning_rate": 0.0002553191489361702,
"loss": 4.4446,
"step": 516
},
{
"epoch": 0.87,
"learning_rate": 0.00025212765957446806,
"loss": 4.7041,
"step": 517
},
{
"epoch": 0.87,
"learning_rate": 0.0002489361702127659,
"loss": 4.5883,
"step": 518
},
{
"epoch": 0.87,
"learning_rate": 0.00024574468085106384,
"loss": 4.5084,
"step": 519
},
{
"epoch": 0.87,
"learning_rate": 0.00024255319148936167,
"loss": 4.5682,
"step": 520
},
{
"epoch": 0.88,
"learning_rate": 0.00023936170212765956,
"loss": 4.7571,
"step": 521
},
{
"epoch": 0.88,
"learning_rate": 0.00023617021276595742,
"loss": 4.9358,
"step": 522
},
{
"epoch": 0.88,
"learning_rate": 0.00023297872340425529,
"loss": 4.4314,
"step": 523
},
{
"epoch": 0.88,
"learning_rate": 0.00022978723404255317,
"loss": 4.1877,
"step": 524
},
{
"epoch": 0.88,
"learning_rate": 0.00022659574468085106,
"loss": 4.7371,
"step": 525
},
{
"epoch": 0.88,
"learning_rate": 0.0002234042553191489,
"loss": 4.6611,
"step": 526
},
{
"epoch": 0.89,
"learning_rate": 0.00022021276595744679,
"loss": 4.6147,
"step": 527
},
{
"epoch": 0.89,
"learning_rate": 0.00021702127659574468,
"loss": 4.4807,
"step": 528
},
{
"epoch": 0.89,
"learning_rate": 0.00021382978723404254,
"loss": 4.8237,
"step": 529
},
{
"epoch": 0.89,
"learning_rate": 0.0002106382978723404,
"loss": 4.3209,
"step": 530
},
{
"epoch": 0.89,
"learning_rate": 0.0002074468085106383,
"loss": 4.4169,
"step": 531
},
{
"epoch": 0.89,
"learning_rate": 0.00020425531914893615,
"loss": 4.6349,
"step": 532
},
{
"epoch": 0.9,
"learning_rate": 0.00020106382978723404,
"loss": 4.6437,
"step": 533
},
{
"epoch": 0.9,
"learning_rate": 0.00019787234042553187,
"loss": 4.4318,
"step": 534
},
{
"epoch": 0.9,
"learning_rate": 0.00019468085106382976,
"loss": 4.3929,
"step": 535
},
{
"epoch": 0.9,
"learning_rate": 0.00019148936170212765,
"loss": 4.5963,
"step": 536
},
{
"epoch": 0.9,
"learning_rate": 0.0001882978723404255,
"loss": 4.474,
"step": 537
},
{
"epoch": 0.9,
"learning_rate": 0.0001851063829787234,
"loss": 4.396,
"step": 538
},
{
"epoch": 0.91,
"learning_rate": 0.00018191489361702126,
"loss": 4.7928,
"step": 539
},
{
"epoch": 0.91,
"learning_rate": 0.00017872340425531912,
"loss": 4.569,
"step": 540
},
{
"epoch": 0.91,
"learning_rate": 0.000175531914893617,
"loss": 4.4839,
"step": 541
},
{
"epoch": 0.91,
"learning_rate": 0.0001723404255319149,
"loss": 4.6734,
"step": 542
},
{
"epoch": 0.91,
"learning_rate": 0.00016914893617021274,
"loss": 4.405,
"step": 543
},
{
"epoch": 0.91,
"learning_rate": 0.00016595744680851062,
"loss": 4.5196,
"step": 544
},
{
"epoch": 0.92,
"learning_rate": 0.00016276595744680849,
"loss": 4.4101,
"step": 545
},
{
"epoch": 0.92,
"learning_rate": 0.00015957446808510637,
"loss": 4.9518,
"step": 546
},
{
"epoch": 0.92,
"learning_rate": 0.00015638297872340426,
"loss": 4.4228,
"step": 547
},
{
"epoch": 0.92,
"learning_rate": 0.0001531914893617021,
"loss": 5.044,
"step": 548
},
{
"epoch": 0.92,
"learning_rate": 0.00015,
"loss": 5.1572,
"step": 549
},
{
"epoch": 0.92,
"learning_rate": 0.00014680851063829785,
"loss": 4.4465,
"step": 550
},
{
"epoch": 0.93,
"learning_rate": 0.00014361702127659574,
"loss": 4.5287,
"step": 551
},
{
"epoch": 0.93,
"learning_rate": 0.0001404255319148936,
"loss": 4.616,
"step": 552
},
{
"epoch": 0.93,
"learning_rate": 0.0001372340425531915,
"loss": 4.5176,
"step": 553
},
{
"epoch": 0.93,
"learning_rate": 0.00013404255319148935,
"loss": 4.3372,
"step": 554
},
{
"epoch": 0.93,
"learning_rate": 0.00013085106382978724,
"loss": 4.4698,
"step": 555
},
{
"epoch": 0.93,
"learning_rate": 0.0001276595744680851,
"loss": 4.4738,
"step": 556
},
{
"epoch": 0.94,
"learning_rate": 0.00012446808510638296,
"loss": 4.6455,
"step": 557
},
{
"epoch": 0.94,
"learning_rate": 0.00012127659574468084,
"loss": 4.5513,
"step": 558
},
{
"epoch": 0.94,
"learning_rate": 0.00011808510638297871,
"loss": 4.3848,
"step": 559
},
{
"epoch": 0.94,
"learning_rate": 0.00011489361702127659,
"loss": 4.3127,
"step": 560
},
{
"epoch": 0.94,
"learning_rate": 0.00011170212765957445,
"loss": 4.7444,
"step": 561
},
{
"epoch": 0.94,
"learning_rate": 0.00010851063829787234,
"loss": 4.4381,
"step": 562
},
{
"epoch": 0.95,
"learning_rate": 0.0001053191489361702,
"loss": 4.4416,
"step": 563
},
{
"epoch": 0.95,
"learning_rate": 0.00010212765957446807,
"loss": 4.8204,
"step": 564
},
{
"epoch": 0.95,
"learning_rate": 9.893617021276594e-05,
"loss": 4.5756,
"step": 565
},
{
"epoch": 0.95,
"learning_rate": 9.574468085106382e-05,
"loss": 4.5617,
"step": 566
},
{
"epoch": 0.95,
"learning_rate": 9.25531914893617e-05,
"loss": 4.4047,
"step": 567
},
{
"epoch": 0.96,
"learning_rate": 8.936170212765956e-05,
"loss": 4.4505,
"step": 568
},
{
"epoch": 0.96,
"learning_rate": 8.617021276595745e-05,
"loss": 4.356,
"step": 569
},
{
"epoch": 0.96,
"learning_rate": 8.297872340425531e-05,
"loss": 4.4439,
"step": 570
},
{
"epoch": 0.96,
"learning_rate": 7.978723404255319e-05,
"loss": 4.4821,
"step": 571
},
{
"epoch": 0.96,
"learning_rate": 7.659574468085105e-05,
"loss": 4.3846,
"step": 572
},
{
"epoch": 0.96,
"learning_rate": 7.340425531914892e-05,
"loss": 4.4297,
"step": 573
},
{
"epoch": 0.97,
"learning_rate": 7.02127659574468e-05,
"loss": 4.3423,
"step": 574
},
{
"epoch": 0.97,
"learning_rate": 6.702127659574467e-05,
"loss": 4.3857,
"step": 575
},
{
"epoch": 0.97,
"learning_rate": 6.382978723404255e-05,
"loss": 4.3895,
"step": 576
},
{
"epoch": 0.97,
"learning_rate": 6.063829787234042e-05,
"loss": 4.5989,
"step": 577
},
{
"epoch": 0.97,
"learning_rate": 5.7446808510638294e-05,
"loss": 4.2233,
"step": 578
},
{
"epoch": 0.97,
"learning_rate": 5.425531914893617e-05,
"loss": 4.3258,
"step": 579
},
{
"epoch": 0.98,
"learning_rate": 5.106382978723404e-05,
"loss": 4.4218,
"step": 580
},
{
"epoch": 0.98,
"learning_rate": 4.787234042553191e-05,
"loss": 4.5631,
"step": 581
},
{
"epoch": 0.98,
"learning_rate": 4.468085106382978e-05,
"loss": 4.3969,
"step": 582
},
{
"epoch": 0.98,
"learning_rate": 4.1489361702127656e-05,
"loss": 4.203,
"step": 583
},
{
"epoch": 0.98,
"learning_rate": 3.8297872340425525e-05,
"loss": 4.3834,
"step": 584
},
{
"epoch": 0.98,
"learning_rate": 3.51063829787234e-05,
"loss": 4.4053,
"step": 585
},
{
"epoch": 0.99,
"learning_rate": 3.1914893617021275e-05,
"loss": 4.3702,
"step": 586
},
{
"epoch": 0.99,
"learning_rate": 2.8723404255319147e-05,
"loss": 4.4538,
"step": 587
},
{
"epoch": 0.99,
"learning_rate": 2.553191489361702e-05,
"loss": 4.617,
"step": 588
},
{
"epoch": 0.99,
"learning_rate": 2.234042553191489e-05,
"loss": 4.4198,
"step": 589
},
{
"epoch": 0.99,
"learning_rate": 1.9148936170212762e-05,
"loss": 4.2778,
"step": 590
},
{
"epoch": 0.99,
"learning_rate": 1.5957446808510637e-05,
"loss": 4.5074,
"step": 591
},
{
"epoch": 1.0,
"learning_rate": 1.276595744680851e-05,
"loss": 4.901,
"step": 592
},
{
"epoch": 1.0,
"learning_rate": 9.574468085106381e-06,
"loss": 4.704,
"step": 593
},
{
"epoch": 1.0,
"learning_rate": 6.382978723404255e-06,
"loss": 4.5343,
"step": 594
},
{
"epoch": 1.0,
"step": 594,
"total_flos": 0.0,
"train_loss": 4.477476079857309,
"train_runtime": 5476.9762,
"train_samples_per_second": 5.211,
"train_steps_per_second": 0.108
}
],
"max_steps": 594,
"num_train_epochs": 1,
"total_flos": 0.0,
"trial_name": null,
"trial_params": null
}