wav2vec2-2-gpt2-grid-search / trainer_state.json
sanchit-gandhi's picture
End of training
a05afcb
raw
history blame
No virus
51.5 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 1.0,
"global_step": 446,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 6e-07,
"loss": 4.7769,
"step": 1
},
{
"epoch": 0.0,
"learning_rate": 1.2e-06,
"loss": 4.9164,
"step": 2
},
{
"epoch": 0.01,
"learning_rate": 1.8e-06,
"loss": 4.8809,
"step": 3
},
{
"epoch": 0.01,
"learning_rate": 1.8e-06,
"loss": 4.8612,
"step": 4
},
{
"epoch": 0.01,
"learning_rate": 2.4e-06,
"loss": 4.7997,
"step": 5
},
{
"epoch": 0.01,
"learning_rate": 2.9999999999999997e-06,
"loss": 4.8452,
"step": 6
},
{
"epoch": 0.02,
"learning_rate": 3.6e-06,
"loss": 4.7936,
"step": 7
},
{
"epoch": 0.02,
"learning_rate": 4.2e-06,
"loss": 4.8186,
"step": 8
},
{
"epoch": 0.02,
"learning_rate": 4.8e-06,
"loss": 4.7825,
"step": 9
},
{
"epoch": 0.02,
"learning_rate": 5.399999999999999e-06,
"loss": 4.621,
"step": 10
},
{
"epoch": 0.02,
"learning_rate": 5.999999999999999e-06,
"loss": 4.7091,
"step": 11
},
{
"epoch": 0.03,
"learning_rate": 6.599999999999999e-06,
"loss": 4.6398,
"step": 12
},
{
"epoch": 0.03,
"learning_rate": 7.2e-06,
"loss": 4.5605,
"step": 13
},
{
"epoch": 0.03,
"learning_rate": 7.799999999999998e-06,
"loss": 4.5522,
"step": 14
},
{
"epoch": 0.03,
"learning_rate": 8.4e-06,
"loss": 4.4202,
"step": 15
},
{
"epoch": 0.04,
"learning_rate": 8.999999999999999e-06,
"loss": 4.5517,
"step": 16
},
{
"epoch": 0.04,
"learning_rate": 9.6e-06,
"loss": 4.4522,
"step": 17
},
{
"epoch": 0.04,
"learning_rate": 1.02e-05,
"loss": 4.5369,
"step": 18
},
{
"epoch": 0.04,
"learning_rate": 1.0799999999999998e-05,
"loss": 4.3402,
"step": 19
},
{
"epoch": 0.04,
"learning_rate": 1.14e-05,
"loss": 4.4829,
"step": 20
},
{
"epoch": 0.05,
"learning_rate": 1.1999999999999999e-05,
"loss": 4.4559,
"step": 21
},
{
"epoch": 0.05,
"learning_rate": 1.26e-05,
"loss": 4.4034,
"step": 22
},
{
"epoch": 0.05,
"learning_rate": 1.3199999999999997e-05,
"loss": 4.4785,
"step": 23
},
{
"epoch": 0.05,
"learning_rate": 1.3799999999999998e-05,
"loss": 4.4421,
"step": 24
},
{
"epoch": 0.06,
"learning_rate": 1.44e-05,
"loss": 4.2612,
"step": 25
},
{
"epoch": 0.06,
"learning_rate": 1.4999999999999999e-05,
"loss": 4.515,
"step": 26
},
{
"epoch": 0.06,
"learning_rate": 1.5599999999999996e-05,
"loss": 4.382,
"step": 27
},
{
"epoch": 0.06,
"learning_rate": 1.6199999999999997e-05,
"loss": 4.217,
"step": 28
},
{
"epoch": 0.07,
"learning_rate": 1.68e-05,
"loss": 4.1962,
"step": 29
},
{
"epoch": 0.07,
"learning_rate": 1.74e-05,
"loss": 4.4288,
"step": 30
},
{
"epoch": 0.07,
"learning_rate": 1.7999999999999997e-05,
"loss": 4.3932,
"step": 31
},
{
"epoch": 0.07,
"learning_rate": 1.8599999999999998e-05,
"loss": 4.2549,
"step": 32
},
{
"epoch": 0.07,
"learning_rate": 1.92e-05,
"loss": 4.3319,
"step": 33
},
{
"epoch": 0.08,
"learning_rate": 1.98e-05,
"loss": 4.2797,
"step": 34
},
{
"epoch": 0.08,
"learning_rate": 2.04e-05,
"loss": 4.3458,
"step": 35
},
{
"epoch": 0.08,
"learning_rate": 2.1e-05,
"loss": 4.3281,
"step": 36
},
{
"epoch": 0.08,
"learning_rate": 2.1599999999999996e-05,
"loss": 4.3676,
"step": 37
},
{
"epoch": 0.09,
"learning_rate": 2.2199999999999998e-05,
"loss": 4.304,
"step": 38
},
{
"epoch": 0.09,
"learning_rate": 2.28e-05,
"loss": 4.4121,
"step": 39
},
{
"epoch": 0.09,
"learning_rate": 2.34e-05,
"loss": 4.2497,
"step": 40
},
{
"epoch": 0.09,
"learning_rate": 2.3999999999999997e-05,
"loss": 4.3074,
"step": 41
},
{
"epoch": 0.09,
"learning_rate": 2.4599999999999998e-05,
"loss": 4.388,
"step": 42
},
{
"epoch": 0.1,
"learning_rate": 2.52e-05,
"loss": 4.1538,
"step": 43
},
{
"epoch": 0.1,
"learning_rate": 2.5799999999999997e-05,
"loss": 4.3297,
"step": 44
},
{
"epoch": 0.1,
"learning_rate": 2.6399999999999995e-05,
"loss": 4.4247,
"step": 45
},
{
"epoch": 0.1,
"learning_rate": 2.6999999999999996e-05,
"loss": 4.4755,
"step": 46
},
{
"epoch": 0.11,
"learning_rate": 2.7599999999999997e-05,
"loss": 4.3917,
"step": 47
},
{
"epoch": 0.11,
"learning_rate": 2.8199999999999998e-05,
"loss": 4.617,
"step": 48
},
{
"epoch": 0.11,
"learning_rate": 2.88e-05,
"loss": 4.6925,
"step": 49
},
{
"epoch": 0.11,
"learning_rate": 2.88e-05,
"loss": 4.4051,
"step": 50
},
{
"epoch": 0.11,
"learning_rate": 2.94e-05,
"loss": 4.2783,
"step": 51
},
{
"epoch": 0.12,
"learning_rate": 2.9999999999999997e-05,
"loss": 4.1599,
"step": 52
},
{
"epoch": 0.12,
"learning_rate": 3.06e-05,
"loss": 4.3254,
"step": 53
},
{
"epoch": 0.12,
"learning_rate": 3.119999999999999e-05,
"loss": 4.2387,
"step": 54
},
{
"epoch": 0.12,
"learning_rate": 3.1799999999999994e-05,
"loss": 4.0646,
"step": 55
},
{
"epoch": 0.13,
"learning_rate": 3.2399999999999995e-05,
"loss": 4.265,
"step": 56
},
{
"epoch": 0.13,
"learning_rate": 3.2999999999999996e-05,
"loss": 4.286,
"step": 57
},
{
"epoch": 0.13,
"learning_rate": 3.36e-05,
"loss": 4.3781,
"step": 58
},
{
"epoch": 0.13,
"learning_rate": 3.42e-05,
"loss": 4.295,
"step": 59
},
{
"epoch": 0.13,
"learning_rate": 3.48e-05,
"loss": 4.1682,
"step": 60
},
{
"epoch": 0.14,
"learning_rate": 3.539999999999999e-05,
"loss": 4.2437,
"step": 61
},
{
"epoch": 0.14,
"learning_rate": 3.5999999999999994e-05,
"loss": 4.188,
"step": 62
},
{
"epoch": 0.14,
"learning_rate": 3.6599999999999995e-05,
"loss": 4.1751,
"step": 63
},
{
"epoch": 0.14,
"learning_rate": 3.7199999999999996e-05,
"loss": 4.2774,
"step": 64
},
{
"epoch": 0.15,
"learning_rate": 3.78e-05,
"loss": 4.2418,
"step": 65
},
{
"epoch": 0.15,
"learning_rate": 3.84e-05,
"loss": 4.3547,
"step": 66
},
{
"epoch": 0.15,
"learning_rate": 3.9e-05,
"loss": 4.3256,
"step": 67
},
{
"epoch": 0.15,
"learning_rate": 3.96e-05,
"loss": 4.1705,
"step": 68
},
{
"epoch": 0.15,
"learning_rate": 4.02e-05,
"loss": 4.2013,
"step": 69
},
{
"epoch": 0.16,
"learning_rate": 4.08e-05,
"loss": 4.1742,
"step": 70
},
{
"epoch": 0.16,
"learning_rate": 4.14e-05,
"loss": 4.2953,
"step": 71
},
{
"epoch": 0.16,
"learning_rate": 4.2e-05,
"loss": 4.3439,
"step": 72
},
{
"epoch": 0.16,
"learning_rate": 4.259999999999999e-05,
"loss": 4.1583,
"step": 73
},
{
"epoch": 0.17,
"learning_rate": 4.319999999999999e-05,
"loss": 4.2408,
"step": 74
},
{
"epoch": 0.17,
"learning_rate": 4.3799999999999994e-05,
"loss": 4.2264,
"step": 75
},
{
"epoch": 0.17,
"learning_rate": 4.4399999999999995e-05,
"loss": 4.2005,
"step": 76
},
{
"epoch": 0.17,
"learning_rate": 4.4999999999999996e-05,
"loss": 4.2334,
"step": 77
},
{
"epoch": 0.17,
"learning_rate": 4.56e-05,
"loss": 4.1885,
"step": 78
},
{
"epoch": 0.18,
"learning_rate": 4.62e-05,
"loss": 4.1813,
"step": 79
},
{
"epoch": 0.18,
"learning_rate": 4.68e-05,
"loss": 4.3409,
"step": 80
},
{
"epoch": 0.18,
"learning_rate": 4.7399999999999993e-05,
"loss": 4.262,
"step": 81
},
{
"epoch": 0.18,
"learning_rate": 4.7999999999999994e-05,
"loss": 4.2151,
"step": 82
},
{
"epoch": 0.19,
"learning_rate": 4.8599999999999995e-05,
"loss": 4.321,
"step": 83
},
{
"epoch": 0.19,
"learning_rate": 4.9199999999999997e-05,
"loss": 4.1941,
"step": 84
},
{
"epoch": 0.19,
"learning_rate": 4.98e-05,
"loss": 4.2677,
"step": 85
},
{
"epoch": 0.19,
"learning_rate": 5.04e-05,
"loss": 4.2252,
"step": 86
},
{
"epoch": 0.2,
"learning_rate": 5.1e-05,
"loss": 4.2876,
"step": 87
},
{
"epoch": 0.2,
"learning_rate": 5.1599999999999994e-05,
"loss": 4.1668,
"step": 88
},
{
"epoch": 0.2,
"learning_rate": 5.2199999999999995e-05,
"loss": 4.3233,
"step": 89
},
{
"epoch": 0.2,
"learning_rate": 5.279999999999999e-05,
"loss": 4.3072,
"step": 90
},
{
"epoch": 0.2,
"learning_rate": 5.339999999999999e-05,
"loss": 4.1785,
"step": 91
},
{
"epoch": 0.21,
"learning_rate": 5.399999999999999e-05,
"loss": 4.428,
"step": 92
},
{
"epoch": 0.21,
"learning_rate": 5.459999999999999e-05,
"loss": 4.3544,
"step": 93
},
{
"epoch": 0.21,
"learning_rate": 5.519999999999999e-05,
"loss": 4.3594,
"step": 94
},
{
"epoch": 0.21,
"learning_rate": 5.5799999999999994e-05,
"loss": 4.2516,
"step": 95
},
{
"epoch": 0.22,
"learning_rate": 5.6399999999999995e-05,
"loss": 4.4909,
"step": 96
},
{
"epoch": 0.22,
"learning_rate": 5.6999999999999996e-05,
"loss": 4.3805,
"step": 97
},
{
"epoch": 0.22,
"learning_rate": 5.76e-05,
"loss": 4.2646,
"step": 98
},
{
"epoch": 0.22,
"learning_rate": 5.82e-05,
"loss": 4.6111,
"step": 99
},
{
"epoch": 0.22,
"learning_rate": 5.88e-05,
"loss": 4.4095,
"step": 100
},
{
"epoch": 0.23,
"learning_rate": 5.94e-05,
"loss": 4.2562,
"step": 101
},
{
"epoch": 0.23,
"learning_rate": 5.9999999999999995e-05,
"loss": 4.1922,
"step": 102
},
{
"epoch": 0.23,
"learning_rate": 6.0599999999999996e-05,
"loss": 4.0757,
"step": 103
},
{
"epoch": 0.23,
"learning_rate": 6.12e-05,
"loss": 4.1528,
"step": 104
},
{
"epoch": 0.24,
"learning_rate": 6.18e-05,
"loss": 4.0813,
"step": 105
},
{
"epoch": 0.24,
"learning_rate": 6.239999999999999e-05,
"loss": 4.1241,
"step": 106
},
{
"epoch": 0.24,
"learning_rate": 6.299999999999999e-05,
"loss": 4.183,
"step": 107
},
{
"epoch": 0.24,
"learning_rate": 6.359999999999999e-05,
"loss": 4.0929,
"step": 108
},
{
"epoch": 0.24,
"learning_rate": 6.419999999999999e-05,
"loss": 4.1176,
"step": 109
},
{
"epoch": 0.25,
"learning_rate": 6.479999999999999e-05,
"loss": 4.1593,
"step": 110
},
{
"epoch": 0.25,
"learning_rate": 6.539999999999999e-05,
"loss": 4.1886,
"step": 111
},
{
"epoch": 0.25,
"learning_rate": 6.599999999999999e-05,
"loss": 4.1987,
"step": 112
},
{
"epoch": 0.25,
"learning_rate": 6.659999999999999e-05,
"loss": 4.1617,
"step": 113
},
{
"epoch": 0.26,
"learning_rate": 6.72e-05,
"loss": 4.1273,
"step": 114
},
{
"epoch": 0.26,
"learning_rate": 6.78e-05,
"loss": 4.1707,
"step": 115
},
{
"epoch": 0.26,
"learning_rate": 6.84e-05,
"loss": 4.2915,
"step": 116
},
{
"epoch": 0.26,
"learning_rate": 6.9e-05,
"loss": 4.1509,
"step": 117
},
{
"epoch": 0.26,
"learning_rate": 6.96e-05,
"loss": 4.2995,
"step": 118
},
{
"epoch": 0.27,
"learning_rate": 7.02e-05,
"loss": 4.2146,
"step": 119
},
{
"epoch": 0.27,
"learning_rate": 7.079999999999999e-05,
"loss": 4.2528,
"step": 120
},
{
"epoch": 0.27,
"learning_rate": 7.139999999999999e-05,
"loss": 4.0505,
"step": 121
},
{
"epoch": 0.27,
"learning_rate": 7.199999999999999e-05,
"loss": 4.1719,
"step": 122
},
{
"epoch": 0.28,
"learning_rate": 7.259999999999999e-05,
"loss": 4.1813,
"step": 123
},
{
"epoch": 0.28,
"learning_rate": 7.319999999999999e-05,
"loss": 4.1625,
"step": 124
},
{
"epoch": 0.28,
"learning_rate": 7.379999999999999e-05,
"loss": 4.2766,
"step": 125
},
{
"epoch": 0.28,
"learning_rate": 7.439999999999999e-05,
"loss": 4.2836,
"step": 126
},
{
"epoch": 0.28,
"learning_rate": 7.5e-05,
"loss": 4.1918,
"step": 127
},
{
"epoch": 0.29,
"learning_rate": 7.56e-05,
"loss": 4.1589,
"step": 128
},
{
"epoch": 0.29,
"learning_rate": 7.62e-05,
"loss": 4.085,
"step": 129
},
{
"epoch": 0.29,
"learning_rate": 7.68e-05,
"loss": 4.2634,
"step": 130
},
{
"epoch": 0.29,
"learning_rate": 7.74e-05,
"loss": 4.1471,
"step": 131
},
{
"epoch": 0.3,
"learning_rate": 7.8e-05,
"loss": 4.2167,
"step": 132
},
{
"epoch": 0.3,
"learning_rate": 7.86e-05,
"loss": 4.1086,
"step": 133
},
{
"epoch": 0.3,
"learning_rate": 7.92e-05,
"loss": 4.2122,
"step": 134
},
{
"epoch": 0.3,
"learning_rate": 7.98e-05,
"loss": 4.3218,
"step": 135
},
{
"epoch": 0.3,
"learning_rate": 8.04e-05,
"loss": 4.1705,
"step": 136
},
{
"epoch": 0.31,
"learning_rate": 8.1e-05,
"loss": 4.2621,
"step": 137
},
{
"epoch": 0.31,
"learning_rate": 8.16e-05,
"loss": 4.2883,
"step": 138
},
{
"epoch": 0.31,
"learning_rate": 8.22e-05,
"loss": 4.3078,
"step": 139
},
{
"epoch": 0.31,
"learning_rate": 8.28e-05,
"loss": 4.2738,
"step": 140
},
{
"epoch": 0.32,
"learning_rate": 8.34e-05,
"loss": 4.3946,
"step": 141
},
{
"epoch": 0.32,
"learning_rate": 8.4e-05,
"loss": 4.3515,
"step": 142
},
{
"epoch": 0.32,
"learning_rate": 8.459999999999998e-05,
"loss": 4.3548,
"step": 143
},
{
"epoch": 0.32,
"learning_rate": 8.519999999999998e-05,
"loss": 4.348,
"step": 144
},
{
"epoch": 0.33,
"learning_rate": 8.579999999999998e-05,
"loss": 4.2612,
"step": 145
},
{
"epoch": 0.33,
"learning_rate": 8.639999999999999e-05,
"loss": 4.3299,
"step": 146
},
{
"epoch": 0.33,
"learning_rate": 8.699999999999999e-05,
"loss": 4.6279,
"step": 147
},
{
"epoch": 0.33,
"learning_rate": 8.759999999999999e-05,
"loss": 4.6792,
"step": 148
},
{
"epoch": 0.33,
"learning_rate": 8.819999999999999e-05,
"loss": 4.537,
"step": 149
},
{
"epoch": 0.34,
"learning_rate": 8.879999999999999e-05,
"loss": 4.5154,
"step": 150
},
{
"epoch": 0.34,
"learning_rate": 8.939999999999999e-05,
"loss": 4.1244,
"step": 151
},
{
"epoch": 0.34,
"learning_rate": 8.999999999999999e-05,
"loss": 4.2241,
"step": 152
},
{
"epoch": 0.34,
"learning_rate": 9.059999999999999e-05,
"loss": 4.2773,
"step": 153
},
{
"epoch": 0.35,
"learning_rate": 9.12e-05,
"loss": 4.0853,
"step": 154
},
{
"epoch": 0.35,
"learning_rate": 9.18e-05,
"loss": 4.2703,
"step": 155
},
{
"epoch": 0.35,
"learning_rate": 9.24e-05,
"loss": 4.0987,
"step": 156
},
{
"epoch": 0.35,
"learning_rate": 9.3e-05,
"loss": 4.1206,
"step": 157
},
{
"epoch": 0.35,
"learning_rate": 9.36e-05,
"loss": 4.2341,
"step": 158
},
{
"epoch": 0.36,
"learning_rate": 9.419999999999999e-05,
"loss": 4.2542,
"step": 159
},
{
"epoch": 0.36,
"learning_rate": 9.479999999999999e-05,
"loss": 4.2671,
"step": 160
},
{
"epoch": 0.36,
"learning_rate": 9.539999999999999e-05,
"loss": 4.298,
"step": 161
},
{
"epoch": 0.36,
"learning_rate": 9.599999999999999e-05,
"loss": 4.2375,
"step": 162
},
{
"epoch": 0.37,
"learning_rate": 9.659999999999999e-05,
"loss": 4.2119,
"step": 163
},
{
"epoch": 0.37,
"learning_rate": 9.719999999999999e-05,
"loss": 4.271,
"step": 164
},
{
"epoch": 0.37,
"learning_rate": 9.779999999999999e-05,
"loss": 4.1551,
"step": 165
},
{
"epoch": 0.37,
"learning_rate": 9.839999999999999e-05,
"loss": 4.1953,
"step": 166
},
{
"epoch": 0.37,
"learning_rate": 9.9e-05,
"loss": 4.0992,
"step": 167
},
{
"epoch": 0.38,
"learning_rate": 9.96e-05,
"loss": 4.1463,
"step": 168
},
{
"epoch": 0.38,
"learning_rate": 0.0001002,
"loss": 4.367,
"step": 169
},
{
"epoch": 0.38,
"learning_rate": 0.0001008,
"loss": 4.3162,
"step": 170
},
{
"epoch": 0.38,
"learning_rate": 0.0001014,
"loss": 4.2243,
"step": 171
},
{
"epoch": 0.39,
"learning_rate": 0.000102,
"loss": 4.3289,
"step": 172
},
{
"epoch": 0.39,
"learning_rate": 0.0001026,
"loss": 4.1431,
"step": 173
},
{
"epoch": 0.39,
"learning_rate": 0.00010319999999999999,
"loss": 4.3012,
"step": 174
},
{
"epoch": 0.39,
"learning_rate": 0.00010379999999999999,
"loss": 4.3199,
"step": 175
},
{
"epoch": 0.39,
"learning_rate": 0.00010439999999999999,
"loss": 4.1572,
"step": 176
},
{
"epoch": 0.4,
"learning_rate": 0.00010499999999999999,
"loss": 4.1236,
"step": 177
},
{
"epoch": 0.4,
"learning_rate": 0.00010559999999999998,
"loss": 4.2767,
"step": 178
},
{
"epoch": 0.4,
"learning_rate": 0.00010619999999999998,
"loss": 4.3445,
"step": 179
},
{
"epoch": 0.4,
"learning_rate": 0.00010679999999999998,
"loss": 4.2821,
"step": 180
},
{
"epoch": 0.41,
"learning_rate": 0.00010739999999999998,
"loss": 4.228,
"step": 181
},
{
"epoch": 0.41,
"learning_rate": 0.00010799999999999998,
"loss": 4.2785,
"step": 182
},
{
"epoch": 0.41,
"learning_rate": 0.00010859999999999998,
"loss": 4.3994,
"step": 183
},
{
"epoch": 0.41,
"learning_rate": 0.00010919999999999998,
"loss": 4.2999,
"step": 184
},
{
"epoch": 0.41,
"learning_rate": 0.00010979999999999999,
"loss": 4.2597,
"step": 185
},
{
"epoch": 0.42,
"learning_rate": 0.00011039999999999999,
"loss": 4.1352,
"step": 186
},
{
"epoch": 0.42,
"learning_rate": 0.00011099999999999999,
"loss": 4.1805,
"step": 187
},
{
"epoch": 0.42,
"learning_rate": 0.00011159999999999999,
"loss": 4.127,
"step": 188
},
{
"epoch": 0.42,
"learning_rate": 0.00011219999999999999,
"loss": 4.31,
"step": 189
},
{
"epoch": 0.43,
"learning_rate": 0.00011279999999999999,
"loss": 4.3199,
"step": 190
},
{
"epoch": 0.43,
"learning_rate": 0.00011339999999999999,
"loss": 4.2928,
"step": 191
},
{
"epoch": 0.43,
"learning_rate": 0.00011399999999999999,
"loss": 4.2784,
"step": 192
},
{
"epoch": 0.43,
"learning_rate": 0.0001146,
"loss": 4.2223,
"step": 193
},
{
"epoch": 0.43,
"learning_rate": 0.0001152,
"loss": 4.3253,
"step": 194
},
{
"epoch": 0.44,
"learning_rate": 0.0001158,
"loss": 4.4979,
"step": 195
},
{
"epoch": 0.44,
"learning_rate": 0.0001164,
"loss": 4.3866,
"step": 196
},
{
"epoch": 0.44,
"learning_rate": 0.000117,
"loss": 4.5051,
"step": 197
},
{
"epoch": 0.44,
"learning_rate": 0.0001176,
"loss": 4.5433,
"step": 198
},
{
"epoch": 0.45,
"learning_rate": 0.0001182,
"loss": 4.6245,
"step": 199
},
{
"epoch": 0.45,
"learning_rate": 0.0001188,
"loss": 4.5374,
"step": 200
},
{
"epoch": 0.45,
"learning_rate": 0.0001194,
"loss": 4.3464,
"step": 201
},
{
"epoch": 0.45,
"learning_rate": 0.00011999999999999999,
"loss": 4.3094,
"step": 202
},
{
"epoch": 0.46,
"learning_rate": 0.00012059999999999999,
"loss": 4.3137,
"step": 203
},
{
"epoch": 0.46,
"learning_rate": 0.00012119999999999999,
"loss": 4.2578,
"step": 204
},
{
"epoch": 0.46,
"learning_rate": 0.00012179999999999999,
"loss": 4.3459,
"step": 205
},
{
"epoch": 0.46,
"learning_rate": 0.0001224,
"loss": 4.3108,
"step": 206
},
{
"epoch": 0.46,
"learning_rate": 0.00012299999999999998,
"loss": 4.144,
"step": 207
},
{
"epoch": 0.47,
"learning_rate": 0.0001236,
"loss": 4.3351,
"step": 208
},
{
"epoch": 0.47,
"learning_rate": 0.00012419999999999998,
"loss": 4.3923,
"step": 209
},
{
"epoch": 0.47,
"learning_rate": 0.00012479999999999997,
"loss": 4.367,
"step": 210
},
{
"epoch": 0.47,
"learning_rate": 0.00012539999999999999,
"loss": 4.1492,
"step": 211
},
{
"epoch": 0.48,
"learning_rate": 0.00012599999999999997,
"loss": 4.2222,
"step": 212
},
{
"epoch": 0.48,
"learning_rate": 0.0001266,
"loss": 4.1713,
"step": 213
},
{
"epoch": 0.48,
"learning_rate": 0.00012719999999999997,
"loss": 4.1952,
"step": 214
},
{
"epoch": 0.48,
"learning_rate": 0.0001278,
"loss": 4.2625,
"step": 215
},
{
"epoch": 0.48,
"learning_rate": 0.00012839999999999998,
"loss": 4.2425,
"step": 216
},
{
"epoch": 0.49,
"learning_rate": 0.000129,
"loss": 4.2802,
"step": 217
},
{
"epoch": 0.49,
"learning_rate": 0.00012959999999999998,
"loss": 4.2234,
"step": 218
},
{
"epoch": 0.49,
"learning_rate": 0.0001302,
"loss": 4.303,
"step": 219
},
{
"epoch": 0.49,
"learning_rate": 0.00013079999999999998,
"loss": 4.2587,
"step": 220
},
{
"epoch": 0.5,
"learning_rate": 0.0001314,
"loss": 4.2141,
"step": 221
},
{
"epoch": 0.5,
"learning_rate": 0.00013199999999999998,
"loss": 4.2823,
"step": 222
},
{
"epoch": 0.5,
"learning_rate": 0.0001326,
"loss": 4.3153,
"step": 223
},
{
"epoch": 0.5,
"learning_rate": 0.00013319999999999999,
"loss": 4.2467,
"step": 224
},
{
"epoch": 0.5,
"learning_rate": 0.0001338,
"loss": 4.3831,
"step": 225
},
{
"epoch": 0.51,
"learning_rate": 0.0001344,
"loss": 4.2606,
"step": 226
},
{
"epoch": 0.51,
"learning_rate": 0.000135,
"loss": 4.4986,
"step": 227
},
{
"epoch": 0.51,
"learning_rate": 0.0001356,
"loss": 4.3744,
"step": 228
},
{
"epoch": 0.51,
"learning_rate": 0.0001362,
"loss": 4.245,
"step": 229
},
{
"epoch": 0.52,
"learning_rate": 0.0001368,
"loss": 4.4023,
"step": 230
},
{
"epoch": 0.52,
"learning_rate": 0.0001374,
"loss": 4.3228,
"step": 231
},
{
"epoch": 0.52,
"learning_rate": 0.000138,
"loss": 4.1694,
"step": 232
},
{
"epoch": 0.52,
"learning_rate": 0.0001386,
"loss": 4.4445,
"step": 233
},
{
"epoch": 0.52,
"learning_rate": 0.0001392,
"loss": 4.3381,
"step": 234
},
{
"epoch": 0.53,
"learning_rate": 0.00013979999999999998,
"loss": 4.4417,
"step": 235
},
{
"epoch": 0.53,
"learning_rate": 0.0001404,
"loss": 4.2774,
"step": 236
},
{
"epoch": 0.53,
"learning_rate": 0.00014099999999999998,
"loss": 4.3152,
"step": 237
},
{
"epoch": 0.53,
"learning_rate": 0.00014159999999999997,
"loss": 4.2731,
"step": 238
},
{
"epoch": 0.54,
"learning_rate": 0.0001422,
"loss": 4.2984,
"step": 239
},
{
"epoch": 0.54,
"learning_rate": 0.00014279999999999997,
"loss": 4.4744,
"step": 240
},
{
"epoch": 0.54,
"learning_rate": 0.0001434,
"loss": 4.2872,
"step": 241
},
{
"epoch": 0.54,
"learning_rate": 0.00014399999999999998,
"loss": 4.3544,
"step": 242
},
{
"epoch": 0.54,
"learning_rate": 0.0001446,
"loss": 4.5594,
"step": 243
},
{
"epoch": 0.55,
"learning_rate": 0.00014519999999999998,
"loss": 4.4725,
"step": 244
},
{
"epoch": 0.55,
"learning_rate": 0.0001458,
"loss": 4.3135,
"step": 245
},
{
"epoch": 0.55,
"learning_rate": 0.00014639999999999998,
"loss": 4.1827,
"step": 246
},
{
"epoch": 0.55,
"learning_rate": 0.000147,
"loss": 4.0959,
"step": 247
},
{
"epoch": 0.56,
"learning_rate": 0.00014759999999999998,
"loss": 4.9479,
"step": 248
},
{
"epoch": 0.56,
"learning_rate": 0.0001482,
"loss": 4.9316,
"step": 249
},
{
"epoch": 0.56,
"learning_rate": 0.00014879999999999998,
"loss": 4.9687,
"step": 250
},
{
"epoch": 0.56,
"learning_rate": 0.0001494,
"loss": 4.4753,
"step": 251
},
{
"epoch": 0.57,
"learning_rate": 0.00015,
"loss": 4.3512,
"step": 252
},
{
"epoch": 0.57,
"learning_rate": 0.00015059999999999997,
"loss": 4.4217,
"step": 253
},
{
"epoch": 0.57,
"learning_rate": 0.0001512,
"loss": 4.2391,
"step": 254
},
{
"epoch": 0.57,
"learning_rate": 0.00015179999999999998,
"loss": 4.3475,
"step": 255
},
{
"epoch": 0.57,
"learning_rate": 0.0001524,
"loss": 4.251,
"step": 256
},
{
"epoch": 0.58,
"learning_rate": 0.00015299999999999998,
"loss": 4.2616,
"step": 257
},
{
"epoch": 0.58,
"learning_rate": 0.0001536,
"loss": 4.3103,
"step": 258
},
{
"epoch": 0.58,
"learning_rate": 0.00015419999999999998,
"loss": 4.3225,
"step": 259
},
{
"epoch": 0.58,
"learning_rate": 0.0001548,
"loss": 4.573,
"step": 260
},
{
"epoch": 0.59,
"learning_rate": 0.00015539999999999998,
"loss": 4.5106,
"step": 261
},
{
"epoch": 0.59,
"learning_rate": 0.000156,
"loss": 4.28,
"step": 262
},
{
"epoch": 0.59,
"learning_rate": 0.00015659999999999998,
"loss": 4.5133,
"step": 263
},
{
"epoch": 0.59,
"learning_rate": 0.0001572,
"loss": 4.3204,
"step": 264
},
{
"epoch": 0.59,
"learning_rate": 0.0001578,
"loss": 4.4585,
"step": 265
},
{
"epoch": 0.6,
"learning_rate": 0.0001584,
"loss": 4.3421,
"step": 266
},
{
"epoch": 0.6,
"learning_rate": 0.000159,
"loss": 4.3581,
"step": 267
},
{
"epoch": 0.6,
"learning_rate": 0.0001596,
"loss": 4.295,
"step": 268
},
{
"epoch": 0.6,
"learning_rate": 0.0001602,
"loss": 4.5293,
"step": 269
},
{
"epoch": 0.61,
"learning_rate": 0.0001608,
"loss": 4.445,
"step": 270
},
{
"epoch": 0.61,
"learning_rate": 0.0001614,
"loss": 4.2543,
"step": 271
},
{
"epoch": 0.61,
"learning_rate": 0.000162,
"loss": 4.3813,
"step": 272
},
{
"epoch": 0.61,
"learning_rate": 0.0001626,
"loss": 4.3964,
"step": 273
},
{
"epoch": 0.61,
"learning_rate": 0.0001632,
"loss": 4.4097,
"step": 274
},
{
"epoch": 0.62,
"learning_rate": 0.0001638,
"loss": 4.4565,
"step": 275
},
{
"epoch": 0.62,
"learning_rate": 0.0001644,
"loss": 4.2819,
"step": 276
},
{
"epoch": 0.62,
"learning_rate": 0.000165,
"loss": 4.302,
"step": 277
},
{
"epoch": 0.62,
"learning_rate": 0.0001656,
"loss": 4.3931,
"step": 278
},
{
"epoch": 0.63,
"learning_rate": 0.0001662,
"loss": 4.3709,
"step": 279
},
{
"epoch": 0.63,
"learning_rate": 0.0001668,
"loss": 4.4844,
"step": 280
},
{
"epoch": 0.63,
"learning_rate": 0.0001674,
"loss": 4.4591,
"step": 281
},
{
"epoch": 0.63,
"learning_rate": 0.000168,
"loss": 4.3974,
"step": 282
},
{
"epoch": 0.63,
"learning_rate": 0.0001686,
"loss": 4.2969,
"step": 283
},
{
"epoch": 0.64,
"learning_rate": 0.00016919999999999997,
"loss": 4.3746,
"step": 284
},
{
"epoch": 0.64,
"learning_rate": 0.00016979999999999998,
"loss": 4.5457,
"step": 285
},
{
"epoch": 0.64,
"learning_rate": 0.00017039999999999997,
"loss": 4.3635,
"step": 286
},
{
"epoch": 0.64,
"learning_rate": 0.00017099999999999998,
"loss": 4.5762,
"step": 287
},
{
"epoch": 0.65,
"learning_rate": 0.00017159999999999997,
"loss": 4.4033,
"step": 288
},
{
"epoch": 0.65,
"learning_rate": 0.00017219999999999998,
"loss": 4.3375,
"step": 289
},
{
"epoch": 0.65,
"learning_rate": 0.00017279999999999997,
"loss": 4.3955,
"step": 290
},
{
"epoch": 0.65,
"learning_rate": 0.00017339999999999996,
"loss": 4.7462,
"step": 291
},
{
"epoch": 0.65,
"learning_rate": 0.00017399999999999997,
"loss": 4.4709,
"step": 292
},
{
"epoch": 0.66,
"learning_rate": 0.00017459999999999996,
"loss": 4.5451,
"step": 293
},
{
"epoch": 0.66,
"learning_rate": 0.00017519999999999998,
"loss": 4.6423,
"step": 294
},
{
"epoch": 0.66,
"learning_rate": 0.00017579999999999996,
"loss": 4.4917,
"step": 295
},
{
"epoch": 0.66,
"learning_rate": 0.00017639999999999998,
"loss": 4.6416,
"step": 296
},
{
"epoch": 0.67,
"learning_rate": 0.00017699999999999997,
"loss": 4.6124,
"step": 297
},
{
"epoch": 0.67,
"learning_rate": 0.00017759999999999998,
"loss": 5.0896,
"step": 298
},
{
"epoch": 0.67,
"learning_rate": 0.00017819999999999997,
"loss": 4.4072,
"step": 299
},
{
"epoch": 0.67,
"learning_rate": 0.00017879999999999998,
"loss": 5.0349,
"step": 300
},
{
"epoch": 0.67,
"learning_rate": 0.00017939999999999997,
"loss": 4.4267,
"step": 301
},
{
"epoch": 0.68,
"learning_rate": 0.00017999999999999998,
"loss": 4.4898,
"step": 302
},
{
"epoch": 0.68,
"learning_rate": 0.00018059999999999997,
"loss": 4.6272,
"step": 303
},
{
"epoch": 0.68,
"learning_rate": 0.00018119999999999999,
"loss": 4.4936,
"step": 304
},
{
"epoch": 0.68,
"learning_rate": 0.00018179999999999997,
"loss": 4.387,
"step": 305
},
{
"epoch": 0.69,
"learning_rate": 0.0001824,
"loss": 4.3621,
"step": 306
},
{
"epoch": 0.69,
"learning_rate": 0.00018299999999999998,
"loss": 4.5623,
"step": 307
},
{
"epoch": 0.69,
"learning_rate": 0.0001836,
"loss": 4.4278,
"step": 308
},
{
"epoch": 0.69,
"learning_rate": 0.00018419999999999998,
"loss": 4.3864,
"step": 309
},
{
"epoch": 0.7,
"learning_rate": 0.0001848,
"loss": 4.505,
"step": 310
},
{
"epoch": 0.7,
"learning_rate": 0.00018539999999999998,
"loss": 4.2331,
"step": 311
},
{
"epoch": 0.7,
"learning_rate": 0.000186,
"loss": 4.5312,
"step": 312
},
{
"epoch": 0.7,
"learning_rate": 0.00018659999999999998,
"loss": 4.6578,
"step": 313
},
{
"epoch": 0.7,
"learning_rate": 0.0001872,
"loss": 4.3891,
"step": 314
},
{
"epoch": 0.71,
"learning_rate": 0.00018779999999999998,
"loss": 4.3873,
"step": 315
},
{
"epoch": 0.71,
"learning_rate": 0.00018839999999999997,
"loss": 4.2965,
"step": 316
},
{
"epoch": 0.71,
"learning_rate": 0.00018899999999999999,
"loss": 4.3478,
"step": 317
},
{
"epoch": 0.71,
"learning_rate": 0.00018959999999999997,
"loss": 4.3313,
"step": 318
},
{
"epoch": 0.72,
"learning_rate": 0.0001902,
"loss": 4.3946,
"step": 319
},
{
"epoch": 0.72,
"learning_rate": 0.00019079999999999998,
"loss": 4.4888,
"step": 320
},
{
"epoch": 0.72,
"learning_rate": 0.0001914,
"loss": 4.3891,
"step": 321
},
{
"epoch": 0.72,
"learning_rate": 0.00019199999999999998,
"loss": 4.1888,
"step": 322
},
{
"epoch": 0.72,
"learning_rate": 0.0001926,
"loss": 4.4259,
"step": 323
},
{
"epoch": 0.73,
"learning_rate": 0.00019319999999999998,
"loss": 4.2324,
"step": 324
},
{
"epoch": 0.73,
"learning_rate": 0.0001938,
"loss": 4.5447,
"step": 325
},
{
"epoch": 0.73,
"learning_rate": 0.00019439999999999998,
"loss": 4.5354,
"step": 326
},
{
"epoch": 0.73,
"learning_rate": 0.000195,
"loss": 4.244,
"step": 327
},
{
"epoch": 0.74,
"learning_rate": 0.00019559999999999998,
"loss": 4.1891,
"step": 328
},
{
"epoch": 0.74,
"learning_rate": 0.0001962,
"loss": 4.3217,
"step": 329
},
{
"epoch": 0.74,
"learning_rate": 0.00019679999999999999,
"loss": 4.4298,
"step": 330
},
{
"epoch": 0.74,
"learning_rate": 0.0001974,
"loss": 4.3572,
"step": 331
},
{
"epoch": 0.74,
"learning_rate": 0.000198,
"loss": 4.3305,
"step": 332
},
{
"epoch": 0.75,
"learning_rate": 0.0001986,
"loss": 4.4236,
"step": 333
},
{
"epoch": 0.75,
"learning_rate": 0.0001992,
"loss": 4.5337,
"step": 334
},
{
"epoch": 0.75,
"learning_rate": 0.0001998,
"loss": 4.6162,
"step": 335
},
{
"epoch": 0.75,
"learning_rate": 0.0002004,
"loss": 4.3679,
"step": 336
},
{
"epoch": 0.76,
"learning_rate": 0.000201,
"loss": 4.3996,
"step": 337
},
{
"epoch": 0.76,
"learning_rate": 0.0002016,
"loss": 4.3851,
"step": 338
},
{
"epoch": 0.76,
"learning_rate": 0.0002022,
"loss": 4.6605,
"step": 339
},
{
"epoch": 0.76,
"learning_rate": 0.0002028,
"loss": 4.4989,
"step": 340
},
{
"epoch": 0.76,
"learning_rate": 0.00020339999999999998,
"loss": 4.2929,
"step": 341
},
{
"epoch": 0.77,
"learning_rate": 0.000204,
"loss": 4.7402,
"step": 342
},
{
"epoch": 0.77,
"learning_rate": 0.00020459999999999999,
"loss": 4.5554,
"step": 343
},
{
"epoch": 0.77,
"learning_rate": 0.0002052,
"loss": 4.5082,
"step": 344
},
{
"epoch": 0.77,
"learning_rate": 0.0002058,
"loss": 4.4385,
"step": 345
},
{
"epoch": 0.78,
"learning_rate": 0.00020639999999999998,
"loss": 4.399,
"step": 346
},
{
"epoch": 0.78,
"learning_rate": 0.00020699999999999996,
"loss": 4.643,
"step": 347
},
{
"epoch": 0.78,
"learning_rate": 0.00020759999999999998,
"loss": 4.6824,
"step": 348
},
{
"epoch": 0.78,
"learning_rate": 0.00020819999999999996,
"loss": 4.7186,
"step": 349
},
{
"epoch": 0.78,
"learning_rate": 0.00020879999999999998,
"loss": 4.4537,
"step": 350
},
{
"epoch": 0.79,
"learning_rate": 0.00020939999999999997,
"loss": 4.5032,
"step": 351
},
{
"epoch": 0.79,
"learning_rate": 0.00020999999999999998,
"loss": 4.4422,
"step": 352
},
{
"epoch": 0.79,
"learning_rate": 0.00021059999999999997,
"loss": 4.3128,
"step": 353
},
{
"epoch": 0.79,
"learning_rate": 0.00021119999999999996,
"loss": 4.4256,
"step": 354
},
{
"epoch": 0.8,
"learning_rate": 0.00021179999999999997,
"loss": 4.4806,
"step": 355
},
{
"epoch": 0.8,
"learning_rate": 0.00021239999999999996,
"loss": 4.6034,
"step": 356
},
{
"epoch": 0.8,
"learning_rate": 0.00021299999999999997,
"loss": 4.5852,
"step": 357
},
{
"epoch": 0.8,
"learning_rate": 0.00021359999999999996,
"loss": 4.6319,
"step": 358
},
{
"epoch": 0.8,
"learning_rate": 0.00021419999999999998,
"loss": 4.4082,
"step": 359
},
{
"epoch": 0.81,
"learning_rate": 0.00021479999999999996,
"loss": 4.4631,
"step": 360
},
{
"epoch": 0.81,
"learning_rate": 0.00021539999999999998,
"loss": 4.4657,
"step": 361
},
{
"epoch": 0.81,
"learning_rate": 0.00021599999999999996,
"loss": 4.386,
"step": 362
},
{
"epoch": 0.81,
"learning_rate": 0.00021659999999999998,
"loss": 4.2731,
"step": 363
},
{
"epoch": 0.82,
"learning_rate": 0.00021719999999999997,
"loss": 4.5098,
"step": 364
},
{
"epoch": 0.82,
"learning_rate": 0.00021779999999999998,
"loss": 4.5502,
"step": 365
},
{
"epoch": 0.82,
"learning_rate": 0.00021839999999999997,
"loss": 4.3979,
"step": 366
},
{
"epoch": 0.82,
"learning_rate": 0.00021899999999999998,
"loss": 4.4207,
"step": 367
},
{
"epoch": 0.83,
"learning_rate": 0.00021959999999999997,
"loss": 4.3131,
"step": 368
},
{
"epoch": 0.83,
"learning_rate": 0.00022019999999999999,
"loss": 4.3822,
"step": 369
},
{
"epoch": 0.83,
"learning_rate": 0.00022079999999999997,
"loss": 4.4616,
"step": 370
},
{
"epoch": 0.83,
"learning_rate": 0.0002214,
"loss": 4.3793,
"step": 371
},
{
"epoch": 0.83,
"learning_rate": 0.00022199999999999998,
"loss": 4.4134,
"step": 372
},
{
"epoch": 0.84,
"learning_rate": 0.0002226,
"loss": 4.4609,
"step": 373
},
{
"epoch": 0.84,
"learning_rate": 0.00022319999999999998,
"loss": 4.1396,
"step": 374
},
{
"epoch": 0.84,
"learning_rate": 0.0002238,
"loss": 4.3943,
"step": 375
},
{
"epoch": 0.84,
"learning_rate": 0.00022439999999999998,
"loss": 4.3628,
"step": 376
},
{
"epoch": 0.85,
"learning_rate": 0.000225,
"loss": 4.3679,
"step": 377
},
{
"epoch": 0.85,
"learning_rate": 0.00022559999999999998,
"loss": 4.5214,
"step": 378
},
{
"epoch": 0.85,
"learning_rate": 0.00022619999999999997,
"loss": 4.3838,
"step": 379
},
{
"epoch": 0.85,
"learning_rate": 0.00022679999999999998,
"loss": 4.2615,
"step": 380
},
{
"epoch": 0.85,
"learning_rate": 0.00022739999999999997,
"loss": 4.6558,
"step": 381
},
{
"epoch": 0.86,
"learning_rate": 0.00022799999999999999,
"loss": 4.5738,
"step": 382
},
{
"epoch": 0.86,
"learning_rate": 0.00022859999999999997,
"loss": 4.4223,
"step": 383
},
{
"epoch": 0.86,
"learning_rate": 0.0002292,
"loss": 4.2627,
"step": 384
},
{
"epoch": 0.86,
"learning_rate": 0.00022979999999999997,
"loss": 4.4766,
"step": 385
},
{
"epoch": 0.87,
"learning_rate": 0.0002304,
"loss": 4.5283,
"step": 386
},
{
"epoch": 0.87,
"learning_rate": 0.00023099999999999998,
"loss": 4.6417,
"step": 387
},
{
"epoch": 0.87,
"learning_rate": 0.0002316,
"loss": 4.4494,
"step": 388
},
{
"epoch": 0.87,
"learning_rate": 0.00023219999999999998,
"loss": 4.4591,
"step": 389
},
{
"epoch": 0.87,
"learning_rate": 0.0002328,
"loss": 4.3444,
"step": 390
},
{
"epoch": 0.88,
"learning_rate": 0.00023339999999999998,
"loss": 4.5171,
"step": 391
},
{
"epoch": 0.88,
"learning_rate": 0.000234,
"loss": 4.3703,
"step": 392
},
{
"epoch": 0.88,
"learning_rate": 0.00023459999999999998,
"loss": 4.24,
"step": 393
},
{
"epoch": 0.88,
"learning_rate": 0.0002352,
"loss": 4.6065,
"step": 394
},
{
"epoch": 0.89,
"learning_rate": 0.00023579999999999999,
"loss": 4.5494,
"step": 395
},
{
"epoch": 0.89,
"learning_rate": 0.0002364,
"loss": 4.4742,
"step": 396
},
{
"epoch": 0.89,
"learning_rate": 0.000237,
"loss": 4.7762,
"step": 397
},
{
"epoch": 0.89,
"learning_rate": 0.0002376,
"loss": 4.5632,
"step": 398
},
{
"epoch": 0.89,
"learning_rate": 0.0002382,
"loss": 5.0284,
"step": 399
},
{
"epoch": 0.9,
"learning_rate": 0.0002388,
"loss": 4.5484,
"step": 400
},
{
"epoch": 0.9,
"learning_rate": 0.0002394,
"loss": 4.7688,
"step": 401
},
{
"epoch": 0.9,
"learning_rate": 0.00023999999999999998,
"loss": 4.8056,
"step": 402
},
{
"epoch": 0.9,
"learning_rate": 0.0002406,
"loss": 4.6578,
"step": 403
},
{
"epoch": 0.91,
"learning_rate": 0.00024119999999999998,
"loss": 4.5998,
"step": 404
},
{
"epoch": 0.91,
"learning_rate": 0.0002418,
"loss": 4.3381,
"step": 405
},
{
"epoch": 0.91,
"learning_rate": 0.00024239999999999998,
"loss": 4.6957,
"step": 406
},
{
"epoch": 0.91,
"learning_rate": 0.000243,
"loss": 4.3475,
"step": 407
},
{
"epoch": 0.91,
"learning_rate": 0.00024359999999999999,
"loss": 4.4122,
"step": 408
},
{
"epoch": 0.92,
"learning_rate": 0.00024419999999999997,
"loss": 4.2999,
"step": 409
},
{
"epoch": 0.92,
"learning_rate": 0.0002448,
"loss": 4.382,
"step": 410
},
{
"epoch": 0.92,
"learning_rate": 0.00024539999999999995,
"loss": 4.6021,
"step": 411
},
{
"epoch": 0.92,
"learning_rate": 0.00024599999999999996,
"loss": 4.5202,
"step": 412
},
{
"epoch": 0.93,
"learning_rate": 0.0002466,
"loss": 4.2693,
"step": 413
},
{
"epoch": 0.93,
"learning_rate": 0.0002472,
"loss": 4.4864,
"step": 414
},
{
"epoch": 0.93,
"learning_rate": 0.00024779999999999995,
"loss": 4.6728,
"step": 415
},
{
"epoch": 0.93,
"learning_rate": 0.00024839999999999997,
"loss": 4.4965,
"step": 416
},
{
"epoch": 0.93,
"learning_rate": 0.000249,
"loss": 4.5908,
"step": 417
},
{
"epoch": 0.94,
"learning_rate": 0.00024959999999999994,
"loss": 4.323,
"step": 418
},
{
"epoch": 0.94,
"learning_rate": 0.00025019999999999996,
"loss": 4.3817,
"step": 419
},
{
"epoch": 0.94,
"learning_rate": 0.00025079999999999997,
"loss": 4.5368,
"step": 420
},
{
"epoch": 0.94,
"learning_rate": 0.0002514,
"loss": 4.6007,
"step": 421
},
{
"epoch": 0.95,
"learning_rate": 0.00025199999999999995,
"loss": 4.3087,
"step": 422
},
{
"epoch": 0.95,
"learning_rate": 0.00025259999999999996,
"loss": 4.4084,
"step": 423
},
{
"epoch": 0.95,
"learning_rate": 0.0002532,
"loss": 4.4871,
"step": 424
},
{
"epoch": 0.95,
"learning_rate": 0.0002538,
"loss": 4.4925,
"step": 425
},
{
"epoch": 0.96,
"learning_rate": 0.00025439999999999995,
"loss": 4.3477,
"step": 426
},
{
"epoch": 0.96,
"learning_rate": 0.00025499999999999996,
"loss": 4.3819,
"step": 427
},
{
"epoch": 0.96,
"learning_rate": 0.0002556,
"loss": 4.4199,
"step": 428
},
{
"epoch": 0.96,
"learning_rate": 0.0002562,
"loss": 4.2349,
"step": 429
},
{
"epoch": 0.96,
"learning_rate": 0.00025679999999999995,
"loss": 4.6076,
"step": 430
},
{
"epoch": 0.97,
"learning_rate": 0.00025739999999999997,
"loss": 4.4529,
"step": 431
},
{
"epoch": 0.97,
"learning_rate": 0.000258,
"loss": 4.6179,
"step": 432
},
{
"epoch": 0.97,
"learning_rate": 0.0002586,
"loss": 4.3895,
"step": 433
},
{
"epoch": 0.97,
"learning_rate": 0.00025919999999999996,
"loss": 4.389,
"step": 434
},
{
"epoch": 0.98,
"learning_rate": 0.00025979999999999997,
"loss": 4.7689,
"step": 435
},
{
"epoch": 0.98,
"learning_rate": 0.0002604,
"loss": 4.3116,
"step": 436
},
{
"epoch": 0.98,
"learning_rate": 0.000261,
"loss": 4.5383,
"step": 437
},
{
"epoch": 0.98,
"learning_rate": 0.00026159999999999996,
"loss": 4.8375,
"step": 438
},
{
"epoch": 0.98,
"learning_rate": 0.0002622,
"loss": 4.6096,
"step": 439
},
{
"epoch": 0.99,
"learning_rate": 0.0002628,
"loss": 4.5625,
"step": 440
},
{
"epoch": 0.99,
"learning_rate": 0.00026339999999999995,
"loss": 4.3984,
"step": 441
},
{
"epoch": 0.99,
"learning_rate": 0.00026399999999999997,
"loss": 4.8585,
"step": 442
},
{
"epoch": 0.99,
"learning_rate": 0.0002646,
"loss": 5.0958,
"step": 443
},
{
"epoch": 1.0,
"learning_rate": 0.0002652,
"loss": 5.1536,
"step": 444
},
{
"epoch": 1.0,
"learning_rate": 0.00026579999999999996,
"loss": 5.0367,
"step": 445
},
{
"epoch": 1.0,
"learning_rate": 0.00026639999999999997,
"loss": 4.5288,
"step": 446
},
{
"epoch": 1.0,
"step": 446,
"total_flos": 0.0,
"train_loss": 4.388703477756859,
"train_runtime": 4866.733,
"train_samples_per_second": 5.864,
"train_steps_per_second": 0.092
}
],
"max_steps": 446,
"num_train_epochs": 1,
"total_flos": 0.0,
"trial_name": null,
"trial_params": null
}