wav2vec2-2-gpt2-grid-search / trainer_state.json
sanchit-gandhi's picture
End of training
bd63150
raw
history blame
No virus
50.1 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": 6.000000000000001e-08,
"loss": 4.7467,
"step": 1
},
{
"epoch": 0.0,
"learning_rate": 1.2000000000000002e-07,
"loss": 4.8989,
"step": 2
},
{
"epoch": 0.01,
"learning_rate": 1.2000000000000002e-07,
"loss": 4.8697,
"step": 3
},
{
"epoch": 0.01,
"learning_rate": 1.8e-07,
"loss": 4.8677,
"step": 4
},
{
"epoch": 0.01,
"learning_rate": 2.4000000000000003e-07,
"loss": 4.7891,
"step": 5
},
{
"epoch": 0.01,
"learning_rate": 3.0000000000000004e-07,
"loss": 4.8594,
"step": 6
},
{
"epoch": 0.02,
"learning_rate": 3.6e-07,
"loss": 4.8379,
"step": 7
},
{
"epoch": 0.02,
"learning_rate": 4.2e-07,
"loss": 4.884,
"step": 8
},
{
"epoch": 0.02,
"learning_rate": 4.800000000000001e-07,
"loss": 4.8896,
"step": 9
},
{
"epoch": 0.02,
"learning_rate": 5.4e-07,
"loss": 4.7506,
"step": 10
},
{
"epoch": 0.02,
"learning_rate": 6.000000000000001e-07,
"loss": 4.8671,
"step": 11
},
{
"epoch": 0.03,
"learning_rate": 6.599999999999999e-07,
"loss": 4.8392,
"step": 12
},
{
"epoch": 0.03,
"learning_rate": 7.2e-07,
"loss": 4.7912,
"step": 13
},
{
"epoch": 0.03,
"learning_rate": 7.799999999999999e-07,
"loss": 4.8021,
"step": 14
},
{
"epoch": 0.03,
"learning_rate": 8.4e-07,
"loss": 4.6805,
"step": 15
},
{
"epoch": 0.04,
"learning_rate": 9e-07,
"loss": 4.8372,
"step": 16
},
{
"epoch": 0.04,
"learning_rate": 9.600000000000001e-07,
"loss": 4.7554,
"step": 17
},
{
"epoch": 0.04,
"learning_rate": 1.0200000000000002e-06,
"loss": 4.8458,
"step": 18
},
{
"epoch": 0.04,
"learning_rate": 1.08e-06,
"loss": 4.6487,
"step": 19
},
{
"epoch": 0.04,
"learning_rate": 1.14e-06,
"loss": 4.814,
"step": 20
},
{
"epoch": 0.05,
"learning_rate": 1.2000000000000002e-06,
"loss": 4.8036,
"step": 21
},
{
"epoch": 0.05,
"learning_rate": 1.26e-06,
"loss": 4.7237,
"step": 22
},
{
"epoch": 0.05,
"learning_rate": 1.3199999999999999e-06,
"loss": 4.8262,
"step": 23
},
{
"epoch": 0.05,
"learning_rate": 1.38e-06,
"loss": 4.8156,
"step": 24
},
{
"epoch": 0.06,
"learning_rate": 1.44e-06,
"loss": 4.6007,
"step": 25
},
{
"epoch": 0.06,
"learning_rate": 1.5e-06,
"loss": 4.8689,
"step": 26
},
{
"epoch": 0.06,
"learning_rate": 1.5599999999999999e-06,
"loss": 4.7013,
"step": 27
},
{
"epoch": 0.06,
"learning_rate": 1.62e-06,
"loss": 4.5816,
"step": 28
},
{
"epoch": 0.07,
"learning_rate": 1.68e-06,
"loss": 4.5434,
"step": 29
},
{
"epoch": 0.07,
"learning_rate": 1.74e-06,
"loss": 4.7481,
"step": 30
},
{
"epoch": 0.07,
"learning_rate": 1.8e-06,
"loss": 4.7113,
"step": 31
},
{
"epoch": 0.07,
"learning_rate": 1.86e-06,
"loss": 4.5551,
"step": 32
},
{
"epoch": 0.07,
"learning_rate": 1.9200000000000003e-06,
"loss": 4.6482,
"step": 33
},
{
"epoch": 0.08,
"learning_rate": 1.98e-06,
"loss": 4.5717,
"step": 34
},
{
"epoch": 0.08,
"learning_rate": 2.0400000000000004e-06,
"loss": 4.6616,
"step": 35
},
{
"epoch": 0.08,
"learning_rate": 2.1000000000000002e-06,
"loss": 4.604,
"step": 36
},
{
"epoch": 0.08,
"learning_rate": 2.16e-06,
"loss": 4.6679,
"step": 37
},
{
"epoch": 0.09,
"learning_rate": 2.22e-06,
"loss": 4.6088,
"step": 38
},
{
"epoch": 0.09,
"learning_rate": 2.28e-06,
"loss": 4.7293,
"step": 39
},
{
"epoch": 0.09,
"learning_rate": 2.34e-06,
"loss": 4.528,
"step": 40
},
{
"epoch": 0.09,
"learning_rate": 2.4000000000000003e-06,
"loss": 4.5742,
"step": 41
},
{
"epoch": 0.09,
"learning_rate": 2.46e-06,
"loss": 4.6963,
"step": 42
},
{
"epoch": 0.1,
"learning_rate": 2.52e-06,
"loss": 4.4568,
"step": 43
},
{
"epoch": 0.1,
"learning_rate": 2.58e-06,
"loss": 4.6075,
"step": 44
},
{
"epoch": 0.1,
"learning_rate": 2.6399999999999997e-06,
"loss": 4.7206,
"step": 45
},
{
"epoch": 0.1,
"learning_rate": 2.7e-06,
"loss": 4.8138,
"step": 46
},
{
"epoch": 0.11,
"learning_rate": 2.76e-06,
"loss": 4.7123,
"step": 47
},
{
"epoch": 0.11,
"learning_rate": 2.82e-06,
"loss": 4.9597,
"step": 48
},
{
"epoch": 0.11,
"learning_rate": 2.88e-06,
"loss": 5.0472,
"step": 49
},
{
"epoch": 0.11,
"learning_rate": 2.88e-06,
"loss": 4.8181,
"step": 50
},
{
"epoch": 0.11,
"learning_rate": 2.9400000000000002e-06,
"loss": 4.4758,
"step": 51
},
{
"epoch": 0.12,
"learning_rate": 3e-06,
"loss": 4.3338,
"step": 52
},
{
"epoch": 0.12,
"learning_rate": 3.06e-06,
"loss": 4.4911,
"step": 53
},
{
"epoch": 0.12,
"learning_rate": 3.1199999999999998e-06,
"loss": 4.4091,
"step": 54
},
{
"epoch": 0.12,
"learning_rate": 3.18e-06,
"loss": 4.2127,
"step": 55
},
{
"epoch": 0.13,
"learning_rate": 3.24e-06,
"loss": 4.4448,
"step": 56
},
{
"epoch": 0.13,
"learning_rate": 3.3e-06,
"loss": 4.4563,
"step": 57
},
{
"epoch": 0.13,
"learning_rate": 3.36e-06,
"loss": 4.5467,
"step": 58
},
{
"epoch": 0.13,
"learning_rate": 3.4200000000000003e-06,
"loss": 4.4715,
"step": 59
},
{
"epoch": 0.13,
"learning_rate": 3.48e-06,
"loss": 4.3147,
"step": 60
},
{
"epoch": 0.14,
"learning_rate": 3.54e-06,
"loss": 4.4139,
"step": 61
},
{
"epoch": 0.14,
"learning_rate": 3.6e-06,
"loss": 4.3598,
"step": 62
},
{
"epoch": 0.14,
"learning_rate": 3.66e-06,
"loss": 4.3187,
"step": 63
},
{
"epoch": 0.14,
"learning_rate": 3.72e-06,
"loss": 4.4192,
"step": 64
},
{
"epoch": 0.15,
"learning_rate": 3.7800000000000002e-06,
"loss": 4.3839,
"step": 65
},
{
"epoch": 0.15,
"learning_rate": 3.8400000000000005e-06,
"loss": 4.2728,
"step": 66
},
{
"epoch": 0.15,
"learning_rate": 3.9e-06,
"loss": 4.4006,
"step": 67
},
{
"epoch": 0.15,
"learning_rate": 3.96e-06,
"loss": 4.2707,
"step": 68
},
{
"epoch": 0.15,
"learning_rate": 4.0200000000000005e-06,
"loss": 4.3321,
"step": 69
},
{
"epoch": 0.16,
"learning_rate": 4.080000000000001e-06,
"loss": 4.2972,
"step": 70
},
{
"epoch": 0.16,
"learning_rate": 4.14e-06,
"loss": 4.4408,
"step": 71
},
{
"epoch": 0.16,
"learning_rate": 4.2000000000000004e-06,
"loss": 4.4262,
"step": 72
},
{
"epoch": 0.16,
"learning_rate": 4.26e-06,
"loss": 4.2577,
"step": 73
},
{
"epoch": 0.17,
"learning_rate": 4.32e-06,
"loss": 4.3613,
"step": 74
},
{
"epoch": 0.17,
"learning_rate": 4.3799999999999996e-06,
"loss": 4.3558,
"step": 75
},
{
"epoch": 0.17,
"learning_rate": 4.44e-06,
"loss": 4.298,
"step": 76
},
{
"epoch": 0.17,
"learning_rate": 4.5e-06,
"loss": 4.3474,
"step": 77
},
{
"epoch": 0.17,
"learning_rate": 4.56e-06,
"loss": 4.2967,
"step": 78
},
{
"epoch": 0.18,
"learning_rate": 4.62e-06,
"loss": 4.2636,
"step": 79
},
{
"epoch": 0.18,
"learning_rate": 4.68e-06,
"loss": 4.4215,
"step": 80
},
{
"epoch": 0.18,
"learning_rate": 4.74e-06,
"loss": 4.3629,
"step": 81
},
{
"epoch": 0.18,
"learning_rate": 4.800000000000001e-06,
"loss": 4.3059,
"step": 82
},
{
"epoch": 0.19,
"learning_rate": 4.86e-06,
"loss": 4.4145,
"step": 83
},
{
"epoch": 0.19,
"learning_rate": 4.92e-06,
"loss": 4.2857,
"step": 84
},
{
"epoch": 0.19,
"learning_rate": 4.980000000000001e-06,
"loss": 4.3572,
"step": 85
},
{
"epoch": 0.19,
"learning_rate": 5.04e-06,
"loss": 4.3283,
"step": 86
},
{
"epoch": 0.2,
"learning_rate": 5.1e-06,
"loss": 4.4223,
"step": 87
},
{
"epoch": 0.2,
"learning_rate": 5.16e-06,
"loss": 4.2989,
"step": 88
},
{
"epoch": 0.2,
"learning_rate": 5.22e-06,
"loss": 4.4382,
"step": 89
},
{
"epoch": 0.2,
"learning_rate": 5.279999999999999e-06,
"loss": 4.4153,
"step": 90
},
{
"epoch": 0.2,
"learning_rate": 5.34e-06,
"loss": 4.3058,
"step": 91
},
{
"epoch": 0.21,
"learning_rate": 5.4e-06,
"loss": 4.5648,
"step": 92
},
{
"epoch": 0.21,
"learning_rate": 5.46e-06,
"loss": 4.4436,
"step": 93
},
{
"epoch": 0.21,
"learning_rate": 5.52e-06,
"loss": 4.4195,
"step": 94
},
{
"epoch": 0.21,
"learning_rate": 5.58e-06,
"loss": 4.3846,
"step": 95
},
{
"epoch": 0.22,
"learning_rate": 5.64e-06,
"loss": 4.5868,
"step": 96
},
{
"epoch": 0.22,
"learning_rate": 5.7000000000000005e-06,
"loss": 4.5338,
"step": 97
},
{
"epoch": 0.22,
"learning_rate": 5.76e-06,
"loss": 4.3898,
"step": 98
},
{
"epoch": 0.22,
"learning_rate": 5.82e-06,
"loss": 4.7102,
"step": 99
},
{
"epoch": 0.22,
"learning_rate": 5.8800000000000005e-06,
"loss": 4.4897,
"step": 100
},
{
"epoch": 0.23,
"learning_rate": 5.940000000000001e-06,
"loss": 4.3043,
"step": 101
},
{
"epoch": 0.23,
"learning_rate": 6e-06,
"loss": 4.2462,
"step": 102
},
{
"epoch": 0.23,
"learning_rate": 6.0600000000000004e-06,
"loss": 4.118,
"step": 103
},
{
"epoch": 0.23,
"learning_rate": 6.12e-06,
"loss": 4.1814,
"step": 104
},
{
"epoch": 0.24,
"learning_rate": 6.18e-06,
"loss": 4.1373,
"step": 105
},
{
"epoch": 0.24,
"learning_rate": 6.2399999999999995e-06,
"loss": 4.1883,
"step": 106
},
{
"epoch": 0.24,
"learning_rate": 6.3e-06,
"loss": 4.2876,
"step": 107
},
{
"epoch": 0.24,
"learning_rate": 6.36e-06,
"loss": 4.1793,
"step": 108
},
{
"epoch": 0.24,
"learning_rate": 6.42e-06,
"loss": 4.1962,
"step": 109
},
{
"epoch": 0.25,
"learning_rate": 6.48e-06,
"loss": 4.2323,
"step": 110
},
{
"epoch": 0.25,
"learning_rate": 6.54e-06,
"loss": 4.273,
"step": 111
},
{
"epoch": 0.25,
"learning_rate": 6.6e-06,
"loss": 4.2356,
"step": 112
},
{
"epoch": 0.25,
"learning_rate": 6.660000000000001e-06,
"loss": 4.2405,
"step": 113
},
{
"epoch": 0.26,
"learning_rate": 6.72e-06,
"loss": 4.2016,
"step": 114
},
{
"epoch": 0.26,
"learning_rate": 6.78e-06,
"loss": 4.2613,
"step": 115
},
{
"epoch": 0.26,
"learning_rate": 6.840000000000001e-06,
"loss": 4.3399,
"step": 116
},
{
"epoch": 0.26,
"learning_rate": 6.900000000000001e-06,
"loss": 4.2145,
"step": 117
},
{
"epoch": 0.26,
"learning_rate": 6.96e-06,
"loss": 4.373,
"step": 118
},
{
"epoch": 0.27,
"learning_rate": 7.0200000000000006e-06,
"loss": 4.2657,
"step": 119
},
{
"epoch": 0.27,
"learning_rate": 7.08e-06,
"loss": 4.2946,
"step": 120
},
{
"epoch": 0.27,
"learning_rate": 7.14e-06,
"loss": 4.1299,
"step": 121
},
{
"epoch": 0.27,
"learning_rate": 7.2e-06,
"loss": 4.2265,
"step": 122
},
{
"epoch": 0.28,
"learning_rate": 7.26e-06,
"loss": 4.2383,
"step": 123
},
{
"epoch": 0.28,
"learning_rate": 7.32e-06,
"loss": 4.2282,
"step": 124
},
{
"epoch": 0.28,
"learning_rate": 7.3800000000000005e-06,
"loss": 4.3391,
"step": 125
},
{
"epoch": 0.28,
"learning_rate": 7.44e-06,
"loss": 4.3396,
"step": 126
},
{
"epoch": 0.28,
"learning_rate": 7.5e-06,
"loss": 4.2597,
"step": 127
},
{
"epoch": 0.29,
"learning_rate": 7.5600000000000005e-06,
"loss": 4.2329,
"step": 128
},
{
"epoch": 0.29,
"learning_rate": 7.62e-06,
"loss": 4.1393,
"step": 129
},
{
"epoch": 0.29,
"learning_rate": 7.680000000000001e-06,
"loss": 4.2915,
"step": 130
},
{
"epoch": 0.29,
"learning_rate": 7.74e-06,
"loss": 4.1948,
"step": 131
},
{
"epoch": 0.3,
"learning_rate": 7.8e-06,
"loss": 4.2495,
"step": 132
},
{
"epoch": 0.3,
"learning_rate": 7.860000000000001e-06,
"loss": 4.1773,
"step": 133
},
{
"epoch": 0.3,
"learning_rate": 7.92e-06,
"loss": 4.2126,
"step": 134
},
{
"epoch": 0.3,
"learning_rate": 7.98e-06,
"loss": 4.3701,
"step": 135
},
{
"epoch": 0.3,
"learning_rate": 8.040000000000001e-06,
"loss": 4.2294,
"step": 136
},
{
"epoch": 0.31,
"learning_rate": 8.1e-06,
"loss": 4.3182,
"step": 137
},
{
"epoch": 0.31,
"learning_rate": 8.160000000000001e-06,
"loss": 4.3569,
"step": 138
},
{
"epoch": 0.31,
"learning_rate": 8.220000000000001e-06,
"loss": 4.3725,
"step": 139
},
{
"epoch": 0.31,
"learning_rate": 8.28e-06,
"loss": 4.3333,
"step": 140
},
{
"epoch": 0.32,
"learning_rate": 8.340000000000001e-06,
"loss": 4.453,
"step": 141
},
{
"epoch": 0.32,
"learning_rate": 8.400000000000001e-06,
"loss": 4.3793,
"step": 142
},
{
"epoch": 0.32,
"learning_rate": 8.459999999999999e-06,
"loss": 4.3205,
"step": 143
},
{
"epoch": 0.32,
"learning_rate": 8.52e-06,
"loss": 4.3848,
"step": 144
},
{
"epoch": 0.33,
"learning_rate": 8.58e-06,
"loss": 4.2916,
"step": 145
},
{
"epoch": 0.33,
"learning_rate": 8.64e-06,
"loss": 4.3768,
"step": 146
},
{
"epoch": 0.33,
"learning_rate": 8.7e-06,
"loss": 4.6325,
"step": 147
},
{
"epoch": 0.33,
"learning_rate": 8.759999999999999e-06,
"loss": 4.7625,
"step": 148
},
{
"epoch": 0.33,
"learning_rate": 8.82e-06,
"loss": 4.5033,
"step": 149
},
{
"epoch": 0.34,
"learning_rate": 8.88e-06,
"loss": 4.6029,
"step": 150
},
{
"epoch": 0.34,
"learning_rate": 8.939999999999999e-06,
"loss": 4.1388,
"step": 151
},
{
"epoch": 0.34,
"learning_rate": 9e-06,
"loss": 4.2244,
"step": 152
},
{
"epoch": 0.34,
"learning_rate": 9.06e-06,
"loss": 4.24,
"step": 153
},
{
"epoch": 0.35,
"learning_rate": 9.12e-06,
"loss": 4.1295,
"step": 154
},
{
"epoch": 0.35,
"learning_rate": 9.18e-06,
"loss": 4.2362,
"step": 155
},
{
"epoch": 0.35,
"learning_rate": 9.24e-06,
"loss": 4.0947,
"step": 156
},
{
"epoch": 0.35,
"learning_rate": 9.3e-06,
"loss": 4.1343,
"step": 157
},
{
"epoch": 0.35,
"learning_rate": 9.36e-06,
"loss": 4.2288,
"step": 158
},
{
"epoch": 0.36,
"learning_rate": 9.42e-06,
"loss": 4.226,
"step": 159
},
{
"epoch": 0.36,
"learning_rate": 9.48e-06,
"loss": 4.246,
"step": 160
},
{
"epoch": 0.36,
"learning_rate": 9.54e-06,
"loss": 4.2936,
"step": 161
},
{
"epoch": 0.36,
"learning_rate": 9.600000000000001e-06,
"loss": 4.2505,
"step": 162
},
{
"epoch": 0.37,
"learning_rate": 9.66e-06,
"loss": 4.2031,
"step": 163
},
{
"epoch": 0.37,
"learning_rate": 9.72e-06,
"loss": 4.2617,
"step": 164
},
{
"epoch": 0.37,
"learning_rate": 9.780000000000001e-06,
"loss": 4.1315,
"step": 165
},
{
"epoch": 0.37,
"learning_rate": 9.84e-06,
"loss": 4.1966,
"step": 166
},
{
"epoch": 0.37,
"learning_rate": 9.9e-06,
"loss": 4.1076,
"step": 167
},
{
"epoch": 0.38,
"learning_rate": 9.960000000000001e-06,
"loss": 4.1875,
"step": 168
},
{
"epoch": 0.38,
"learning_rate": 1.002e-05,
"loss": 4.2511,
"step": 169
},
{
"epoch": 0.38,
"learning_rate": 1.008e-05,
"loss": 4.2712,
"step": 170
},
{
"epoch": 0.38,
"learning_rate": 1.0140000000000001e-05,
"loss": 4.2481,
"step": 171
},
{
"epoch": 0.39,
"learning_rate": 1.02e-05,
"loss": 4.3051,
"step": 172
},
{
"epoch": 0.39,
"learning_rate": 1.0260000000000002e-05,
"loss": 4.1154,
"step": 173
},
{
"epoch": 0.39,
"learning_rate": 1.032e-05,
"loss": 4.2669,
"step": 174
},
{
"epoch": 0.39,
"learning_rate": 1.0379999999999999e-05,
"loss": 4.3157,
"step": 175
},
{
"epoch": 0.39,
"learning_rate": 1.044e-05,
"loss": 4.191,
"step": 176
},
{
"epoch": 0.4,
"learning_rate": 1.05e-05,
"loss": 4.153,
"step": 177
},
{
"epoch": 0.4,
"learning_rate": 1.0559999999999999e-05,
"loss": 4.2413,
"step": 178
},
{
"epoch": 0.4,
"learning_rate": 1.062e-05,
"loss": 4.3099,
"step": 179
},
{
"epoch": 0.4,
"learning_rate": 1.068e-05,
"loss": 4.2988,
"step": 180
},
{
"epoch": 0.41,
"learning_rate": 1.074e-05,
"loss": 4.2251,
"step": 181
},
{
"epoch": 0.41,
"learning_rate": 1.08e-05,
"loss": 4.2882,
"step": 182
},
{
"epoch": 0.41,
"learning_rate": 1.086e-05,
"loss": 4.2774,
"step": 183
},
{
"epoch": 0.41,
"learning_rate": 1.092e-05,
"loss": 4.2853,
"step": 184
},
{
"epoch": 0.41,
"learning_rate": 1.098e-05,
"loss": 4.2599,
"step": 185
},
{
"epoch": 0.42,
"learning_rate": 1.104e-05,
"loss": 4.1288,
"step": 186
},
{
"epoch": 0.42,
"learning_rate": 1.11e-05,
"loss": 4.1495,
"step": 187
},
{
"epoch": 0.42,
"learning_rate": 1.116e-05,
"loss": 4.1223,
"step": 188
},
{
"epoch": 0.42,
"learning_rate": 1.1220000000000001e-05,
"loss": 4.2413,
"step": 189
},
{
"epoch": 0.43,
"learning_rate": 1.128e-05,
"loss": 4.3383,
"step": 190
},
{
"epoch": 0.43,
"learning_rate": 1.134e-05,
"loss": 4.2579,
"step": 191
},
{
"epoch": 0.43,
"learning_rate": 1.1400000000000001e-05,
"loss": 4.2088,
"step": 192
},
{
"epoch": 0.43,
"learning_rate": 1.146e-05,
"loss": 4.2129,
"step": 193
},
{
"epoch": 0.43,
"learning_rate": 1.152e-05,
"loss": 4.3219,
"step": 194
},
{
"epoch": 0.44,
"learning_rate": 1.1580000000000001e-05,
"loss": 4.4336,
"step": 195
},
{
"epoch": 0.44,
"learning_rate": 1.164e-05,
"loss": 4.3798,
"step": 196
},
{
"epoch": 0.44,
"learning_rate": 1.1700000000000001e-05,
"loss": 4.3175,
"step": 197
},
{
"epoch": 0.44,
"learning_rate": 1.1760000000000001e-05,
"loss": 4.6016,
"step": 198
},
{
"epoch": 0.45,
"learning_rate": 1.182e-05,
"loss": 4.5392,
"step": 199
},
{
"epoch": 0.45,
"learning_rate": 1.1880000000000001e-05,
"loss": 4.4927,
"step": 200
},
{
"epoch": 0.45,
"learning_rate": 1.1940000000000001e-05,
"loss": 4.2635,
"step": 201
},
{
"epoch": 0.45,
"learning_rate": 1.2e-05,
"loss": 4.1927,
"step": 202
},
{
"epoch": 0.46,
"learning_rate": 1.2060000000000001e-05,
"loss": 4.2133,
"step": 203
},
{
"epoch": 0.46,
"learning_rate": 1.2120000000000001e-05,
"loss": 4.204,
"step": 204
},
{
"epoch": 0.46,
"learning_rate": 1.2180000000000002e-05,
"loss": 4.2708,
"step": 205
},
{
"epoch": 0.46,
"learning_rate": 1.224e-05,
"loss": 4.2503,
"step": 206
},
{
"epoch": 0.46,
"learning_rate": 1.2299999999999999e-05,
"loss": 4.1237,
"step": 207
},
{
"epoch": 0.47,
"learning_rate": 1.236e-05,
"loss": 4.218,
"step": 208
},
{
"epoch": 0.47,
"learning_rate": 1.242e-05,
"loss": 4.2768,
"step": 209
},
{
"epoch": 0.47,
"learning_rate": 1.2479999999999999e-05,
"loss": 4.2393,
"step": 210
},
{
"epoch": 0.47,
"learning_rate": 1.254e-05,
"loss": 4.1165,
"step": 211
},
{
"epoch": 0.48,
"learning_rate": 1.26e-05,
"loss": 4.1624,
"step": 212
},
{
"epoch": 0.48,
"learning_rate": 1.2659999999999999e-05,
"loss": 4.0918,
"step": 213
},
{
"epoch": 0.48,
"learning_rate": 1.272e-05,
"loss": 4.1649,
"step": 214
},
{
"epoch": 0.48,
"learning_rate": 1.278e-05,
"loss": 4.2075,
"step": 215
},
{
"epoch": 0.48,
"learning_rate": 1.284e-05,
"loss": 4.1771,
"step": 216
},
{
"epoch": 0.49,
"learning_rate": 1.29e-05,
"loss": 4.2251,
"step": 217
},
{
"epoch": 0.49,
"learning_rate": 1.296e-05,
"loss": 4.1744,
"step": 218
},
{
"epoch": 0.49,
"learning_rate": 1.302e-05,
"loss": 4.1068,
"step": 219
},
{
"epoch": 0.49,
"learning_rate": 1.308e-05,
"loss": 4.2239,
"step": 220
},
{
"epoch": 0.5,
"learning_rate": 1.314e-05,
"loss": 4.2007,
"step": 221
},
{
"epoch": 0.5,
"learning_rate": 1.32e-05,
"loss": 4.1825,
"step": 222
},
{
"epoch": 0.5,
"learning_rate": 1.326e-05,
"loss": 4.2203,
"step": 223
},
{
"epoch": 0.5,
"learning_rate": 1.3320000000000001e-05,
"loss": 4.1415,
"step": 224
},
{
"epoch": 0.5,
"learning_rate": 1.338e-05,
"loss": 4.3006,
"step": 225
},
{
"epoch": 0.51,
"learning_rate": 1.344e-05,
"loss": 4.1949,
"step": 226
},
{
"epoch": 0.51,
"learning_rate": 1.3500000000000001e-05,
"loss": 4.3143,
"step": 227
},
{
"epoch": 0.51,
"learning_rate": 1.356e-05,
"loss": 4.2632,
"step": 228
},
{
"epoch": 0.51,
"learning_rate": 1.362e-05,
"loss": 4.1447,
"step": 229
},
{
"epoch": 0.52,
"learning_rate": 1.3680000000000001e-05,
"loss": 4.2359,
"step": 230
},
{
"epoch": 0.52,
"learning_rate": 1.374e-05,
"loss": 4.3002,
"step": 231
},
{
"epoch": 0.52,
"learning_rate": 1.3800000000000002e-05,
"loss": 4.0972,
"step": 232
},
{
"epoch": 0.52,
"learning_rate": 1.3860000000000001e-05,
"loss": 4.1221,
"step": 233
},
{
"epoch": 0.52,
"learning_rate": 1.392e-05,
"loss": 4.2082,
"step": 234
},
{
"epoch": 0.53,
"learning_rate": 1.3980000000000002e-05,
"loss": 4.2862,
"step": 235
},
{
"epoch": 0.53,
"learning_rate": 1.4040000000000001e-05,
"loss": 4.2141,
"step": 236
},
{
"epoch": 0.53,
"learning_rate": 1.4099999999999999e-05,
"loss": 4.2371,
"step": 237
},
{
"epoch": 0.53,
"learning_rate": 1.416e-05,
"loss": 4.1934,
"step": 238
},
{
"epoch": 0.54,
"learning_rate": 1.422e-05,
"loss": 4.1346,
"step": 239
},
{
"epoch": 0.54,
"learning_rate": 1.428e-05,
"loss": 4.2417,
"step": 240
},
{
"epoch": 0.54,
"learning_rate": 1.434e-05,
"loss": 4.2216,
"step": 241
},
{
"epoch": 0.54,
"learning_rate": 1.44e-05,
"loss": 4.2632,
"step": 242
},
{
"epoch": 0.54,
"learning_rate": 1.446e-05,
"loss": 4.4629,
"step": 243
},
{
"epoch": 0.55,
"learning_rate": 1.452e-05,
"loss": 4.2896,
"step": 244
},
{
"epoch": 0.55,
"learning_rate": 1.458e-05,
"loss": 4.2854,
"step": 245
},
{
"epoch": 0.55,
"learning_rate": 1.464e-05,
"loss": 4.1776,
"step": 246
},
{
"epoch": 0.55,
"learning_rate": 1.47e-05,
"loss": 4.1134,
"step": 247
},
{
"epoch": 0.56,
"learning_rate": 1.4760000000000001e-05,
"loss": 4.4902,
"step": 248
},
{
"epoch": 0.56,
"learning_rate": 1.482e-05,
"loss": 4.5911,
"step": 249
},
{
"epoch": 0.56,
"learning_rate": 1.488e-05,
"loss": 4.5827,
"step": 250
},
{
"epoch": 0.56,
"learning_rate": 1.4940000000000001e-05,
"loss": 4.1924,
"step": 251
},
{
"epoch": 0.57,
"learning_rate": 1.5e-05,
"loss": 4.2254,
"step": 252
},
{
"epoch": 0.57,
"learning_rate": 1.506e-05,
"loss": 4.2717,
"step": 253
},
{
"epoch": 0.57,
"learning_rate": 1.5120000000000001e-05,
"loss": 4.1103,
"step": 254
},
{
"epoch": 0.57,
"learning_rate": 1.518e-05,
"loss": 4.2008,
"step": 255
},
{
"epoch": 0.57,
"learning_rate": 1.524e-05,
"loss": 4.1802,
"step": 256
},
{
"epoch": 0.58,
"learning_rate": 1.53e-05,
"loss": 4.1619,
"step": 257
},
{
"epoch": 0.58,
"learning_rate": 1.5360000000000002e-05,
"loss": 4.1629,
"step": 258
},
{
"epoch": 0.58,
"learning_rate": 1.542e-05,
"loss": 4.175,
"step": 259
},
{
"epoch": 0.58,
"learning_rate": 1.548e-05,
"loss": 4.1092,
"step": 260
},
{
"epoch": 0.59,
"learning_rate": 1.554e-05,
"loss": 4.2349,
"step": 261
},
{
"epoch": 0.59,
"learning_rate": 1.56e-05,
"loss": 4.1544,
"step": 262
},
{
"epoch": 0.59,
"learning_rate": 1.5660000000000003e-05,
"loss": 4.2932,
"step": 263
},
{
"epoch": 0.59,
"learning_rate": 1.5720000000000002e-05,
"loss": 4.1587,
"step": 264
},
{
"epoch": 0.59,
"learning_rate": 1.578e-05,
"loss": 4.2105,
"step": 265
},
{
"epoch": 0.6,
"learning_rate": 1.584e-05,
"loss": 4.1676,
"step": 266
},
{
"epoch": 0.6,
"learning_rate": 1.59e-05,
"loss": 4.2021,
"step": 267
},
{
"epoch": 0.6,
"learning_rate": 1.596e-05,
"loss": 4.1491,
"step": 268
},
{
"epoch": 0.6,
"learning_rate": 1.6020000000000002e-05,
"loss": 4.2414,
"step": 269
},
{
"epoch": 0.61,
"learning_rate": 1.6080000000000002e-05,
"loss": 4.1202,
"step": 270
},
{
"epoch": 0.61,
"learning_rate": 1.614e-05,
"loss": 4.0542,
"step": 271
},
{
"epoch": 0.61,
"learning_rate": 1.62e-05,
"loss": 4.2317,
"step": 272
},
{
"epoch": 0.61,
"learning_rate": 1.626e-05,
"loss": 4.0302,
"step": 273
},
{
"epoch": 0.61,
"learning_rate": 1.6320000000000003e-05,
"loss": 4.1353,
"step": 274
},
{
"epoch": 0.62,
"learning_rate": 1.6380000000000002e-05,
"loss": 4.2181,
"step": 275
},
{
"epoch": 0.62,
"learning_rate": 1.6440000000000002e-05,
"loss": 4.162,
"step": 276
},
{
"epoch": 0.62,
"learning_rate": 1.65e-05,
"loss": 4.1127,
"step": 277
},
{
"epoch": 0.62,
"learning_rate": 1.656e-05,
"loss": 4.1905,
"step": 278
},
{
"epoch": 0.63,
"learning_rate": 1.6620000000000004e-05,
"loss": 4.1745,
"step": 279
},
{
"epoch": 0.63,
"learning_rate": 1.6680000000000003e-05,
"loss": 4.309,
"step": 280
},
{
"epoch": 0.63,
"learning_rate": 1.6740000000000002e-05,
"loss": 4.2104,
"step": 281
},
{
"epoch": 0.63,
"learning_rate": 1.6800000000000002e-05,
"loss": 4.0622,
"step": 282
},
{
"epoch": 0.63,
"learning_rate": 1.686e-05,
"loss": 4.0876,
"step": 283
},
{
"epoch": 0.64,
"learning_rate": 1.6919999999999997e-05,
"loss": 4.1781,
"step": 284
},
{
"epoch": 0.64,
"learning_rate": 1.698e-05,
"loss": 4.2782,
"step": 285
},
{
"epoch": 0.64,
"learning_rate": 1.704e-05,
"loss": 4.1623,
"step": 286
},
{
"epoch": 0.64,
"learning_rate": 1.71e-05,
"loss": 4.2786,
"step": 287
},
{
"epoch": 0.65,
"learning_rate": 1.716e-05,
"loss": 4.2257,
"step": 288
},
{
"epoch": 0.65,
"learning_rate": 1.7219999999999998e-05,
"loss": 4.1352,
"step": 289
},
{
"epoch": 0.65,
"learning_rate": 1.728e-05,
"loss": 4.1937,
"step": 290
},
{
"epoch": 0.65,
"learning_rate": 1.734e-05,
"loss": 4.4352,
"step": 291
},
{
"epoch": 0.65,
"learning_rate": 1.74e-05,
"loss": 4.3165,
"step": 292
},
{
"epoch": 0.66,
"learning_rate": 1.746e-05,
"loss": 4.3812,
"step": 293
},
{
"epoch": 0.66,
"learning_rate": 1.7519999999999998e-05,
"loss": 4.3448,
"step": 294
},
{
"epoch": 0.66,
"learning_rate": 1.758e-05,
"loss": 4.3412,
"step": 295
},
{
"epoch": 0.66,
"learning_rate": 1.764e-05,
"loss": 4.346,
"step": 296
},
{
"epoch": 0.67,
"learning_rate": 1.77e-05,
"loss": 4.2396,
"step": 297
},
{
"epoch": 0.67,
"learning_rate": 1.776e-05,
"loss": 4.5625,
"step": 298
},
{
"epoch": 0.67,
"learning_rate": 1.782e-05,
"loss": 4.304,
"step": 299
},
{
"epoch": 0.67,
"learning_rate": 1.7879999999999998e-05,
"loss": 4.7425,
"step": 300
},
{
"epoch": 0.67,
"learning_rate": 1.794e-05,
"loss": 4.2219,
"step": 301
},
{
"epoch": 0.68,
"learning_rate": 1.8e-05,
"loss": 4.1874,
"step": 302
},
{
"epoch": 0.68,
"learning_rate": 1.806e-05,
"loss": 4.1883,
"step": 303
},
{
"epoch": 0.68,
"learning_rate": 1.812e-05,
"loss": 4.2025,
"step": 304
},
{
"epoch": 0.68,
"learning_rate": 1.818e-05,
"loss": 4.1717,
"step": 305
},
{
"epoch": 0.69,
"learning_rate": 1.824e-05,
"loss": 4.1519,
"step": 306
},
{
"epoch": 0.69,
"learning_rate": 1.83e-05,
"loss": 4.1405,
"step": 307
},
{
"epoch": 0.69,
"learning_rate": 1.836e-05,
"loss": 4.1463,
"step": 308
},
{
"epoch": 0.69,
"learning_rate": 1.842e-05,
"loss": 4.1423,
"step": 309
},
{
"epoch": 0.7,
"learning_rate": 1.848e-05,
"loss": 4.2402,
"step": 310
},
{
"epoch": 0.7,
"learning_rate": 1.854e-05,
"loss": 4.0354,
"step": 311
},
{
"epoch": 0.7,
"learning_rate": 1.86e-05,
"loss": 4.2098,
"step": 312
},
{
"epoch": 0.7,
"learning_rate": 1.866e-05,
"loss": 4.2141,
"step": 313
},
{
"epoch": 0.7,
"learning_rate": 1.872e-05,
"loss": 4.1336,
"step": 314
},
{
"epoch": 0.71,
"learning_rate": 1.878e-05,
"loss": 4.1595,
"step": 315
},
{
"epoch": 0.71,
"learning_rate": 1.884e-05,
"loss": 4.1353,
"step": 316
},
{
"epoch": 0.71,
"learning_rate": 1.8900000000000002e-05,
"loss": 4.131,
"step": 317
},
{
"epoch": 0.71,
"learning_rate": 1.896e-05,
"loss": 4.1589,
"step": 318
},
{
"epoch": 0.72,
"learning_rate": 1.902e-05,
"loss": 4.1548,
"step": 319
},
{
"epoch": 0.72,
"learning_rate": 1.908e-05,
"loss": 4.2273,
"step": 320
},
{
"epoch": 0.72,
"learning_rate": 1.914e-05,
"loss": 4.1213,
"step": 321
},
{
"epoch": 0.72,
"learning_rate": 1.9200000000000003e-05,
"loss": 3.9964,
"step": 322
},
{
"epoch": 0.72,
"learning_rate": 1.9260000000000002e-05,
"loss": 4.1827,
"step": 323
},
{
"epoch": 0.73,
"learning_rate": 1.932e-05,
"loss": 4.05,
"step": 324
},
{
"epoch": 0.73,
"learning_rate": 1.938e-05,
"loss": 4.1939,
"step": 325
},
{
"epoch": 0.73,
"learning_rate": 1.944e-05,
"loss": 4.1657,
"step": 326
},
{
"epoch": 0.73,
"learning_rate": 1.95e-05,
"loss": 4.1165,
"step": 327
},
{
"epoch": 0.74,
"learning_rate": 1.9560000000000002e-05,
"loss": 4.0746,
"step": 328
},
{
"epoch": 0.74,
"learning_rate": 1.9620000000000002e-05,
"loss": 4.132,
"step": 329
},
{
"epoch": 0.74,
"learning_rate": 1.968e-05,
"loss": 4.1953,
"step": 330
},
{
"epoch": 0.74,
"learning_rate": 1.974e-05,
"loss": 4.1936,
"step": 331
},
{
"epoch": 0.74,
"learning_rate": 1.98e-05,
"loss": 4.0789,
"step": 332
},
{
"epoch": 0.75,
"learning_rate": 1.9860000000000003e-05,
"loss": 4.249,
"step": 333
},
{
"epoch": 0.75,
"learning_rate": 1.9920000000000002e-05,
"loss": 4.1472,
"step": 334
},
{
"epoch": 0.75,
"learning_rate": 1.9980000000000002e-05,
"loss": 4.2981,
"step": 335
},
{
"epoch": 0.75,
"learning_rate": 2.004e-05,
"loss": 4.109,
"step": 336
},
{
"epoch": 0.76,
"learning_rate": 2.01e-05,
"loss": 4.1043,
"step": 337
},
{
"epoch": 0.76,
"learning_rate": 2.016e-05,
"loss": 4.2191,
"step": 338
},
{
"epoch": 0.76,
"learning_rate": 2.0220000000000003e-05,
"loss": 4.1956,
"step": 339
},
{
"epoch": 0.76,
"learning_rate": 2.0280000000000002e-05,
"loss": 4.256,
"step": 340
},
{
"epoch": 0.76,
"learning_rate": 2.0340000000000002e-05,
"loss": 4.1683,
"step": 341
},
{
"epoch": 0.77,
"learning_rate": 2.04e-05,
"loss": 4.3209,
"step": 342
},
{
"epoch": 0.77,
"learning_rate": 2.046e-05,
"loss": 4.2487,
"step": 343
},
{
"epoch": 0.77,
"learning_rate": 2.0520000000000003e-05,
"loss": 4.3359,
"step": 344
},
{
"epoch": 0.77,
"learning_rate": 2.0580000000000003e-05,
"loss": 4.0818,
"step": 345
},
{
"epoch": 0.78,
"learning_rate": 2.064e-05,
"loss": 4.3366,
"step": 346
},
{
"epoch": 0.78,
"learning_rate": 2.07e-05,
"loss": 4.3017,
"step": 347
},
{
"epoch": 0.78,
"learning_rate": 2.0759999999999998e-05,
"loss": 4.24,
"step": 348
},
{
"epoch": 0.78,
"learning_rate": 2.082e-05,
"loss": 4.4577,
"step": 349
},
{
"epoch": 0.78,
"learning_rate": 2.088e-05,
"loss": 4.4505,
"step": 350
},
{
"epoch": 0.79,
"learning_rate": 2.094e-05,
"loss": 4.2355,
"step": 351
},
{
"epoch": 0.79,
"learning_rate": 2.1e-05,
"loss": 4.1619,
"step": 352
},
{
"epoch": 0.79,
"learning_rate": 2.1059999999999998e-05,
"loss": 4.0909,
"step": 353
},
{
"epoch": 0.79,
"learning_rate": 2.1119999999999998e-05,
"loss": 3.9861,
"step": 354
},
{
"epoch": 0.8,
"learning_rate": 2.118e-05,
"loss": 4.1762,
"step": 355
},
{
"epoch": 0.8,
"learning_rate": 2.124e-05,
"loss": 4.2691,
"step": 356
},
{
"epoch": 0.8,
"learning_rate": 2.13e-05,
"loss": 4.1392,
"step": 357
},
{
"epoch": 0.8,
"learning_rate": 2.136e-05,
"loss": 4.1894,
"step": 358
},
{
"epoch": 0.8,
"learning_rate": 2.1419999999999998e-05,
"loss": 4.0681,
"step": 359
},
{
"epoch": 0.81,
"learning_rate": 2.148e-05,
"loss": 4.1466,
"step": 360
},
{
"epoch": 0.81,
"learning_rate": 2.154e-05,
"loss": 4.0732,
"step": 361
},
{
"epoch": 0.81,
"learning_rate": 2.16e-05,
"loss": 4.1424,
"step": 362
},
{
"epoch": 0.81,
"learning_rate": 2.166e-05,
"loss": 4.0726,
"step": 363
},
{
"epoch": 0.82,
"learning_rate": 2.172e-05,
"loss": 4.1657,
"step": 364
},
{
"epoch": 0.82,
"learning_rate": 2.178e-05,
"loss": 4.2799,
"step": 365
},
{
"epoch": 0.82,
"learning_rate": 2.184e-05,
"loss": 4.0866,
"step": 366
},
{
"epoch": 0.82,
"learning_rate": 2.19e-05,
"loss": 4.1687,
"step": 367
},
{
"epoch": 0.83,
"learning_rate": 2.196e-05,
"loss": 4.1556,
"step": 368
},
{
"epoch": 0.83,
"learning_rate": 2.202e-05,
"loss": 4.0718,
"step": 369
},
{
"epoch": 0.83,
"learning_rate": 2.208e-05,
"loss": 4.088,
"step": 370
},
{
"epoch": 0.83,
"learning_rate": 2.214e-05,
"loss": 4.0683,
"step": 371
},
{
"epoch": 0.83,
"learning_rate": 2.22e-05,
"loss": 4.1558,
"step": 372
},
{
"epoch": 0.84,
"learning_rate": 2.226e-05,
"loss": 4.1561,
"step": 373
},
{
"epoch": 0.84,
"learning_rate": 2.232e-05,
"loss": 3.9638,
"step": 374
},
{
"epoch": 0.84,
"learning_rate": 2.238e-05,
"loss": 4.1671,
"step": 375
},
{
"epoch": 0.84,
"learning_rate": 2.2440000000000002e-05,
"loss": 4.0345,
"step": 376
},
{
"epoch": 0.85,
"learning_rate": 2.25e-05,
"loss": 4.1402,
"step": 377
},
{
"epoch": 0.85,
"learning_rate": 2.256e-05,
"loss": 4.1447,
"step": 378
},
{
"epoch": 0.85,
"learning_rate": 2.262e-05,
"loss": 4.0566,
"step": 379
},
{
"epoch": 0.85,
"learning_rate": 2.268e-05,
"loss": 4.074,
"step": 380
},
{
"epoch": 0.85,
"learning_rate": 2.274e-05,
"loss": 4.1216,
"step": 381
},
{
"epoch": 0.86,
"learning_rate": 2.2800000000000002e-05,
"loss": 4.1987,
"step": 382
},
{
"epoch": 0.86,
"learning_rate": 2.286e-05,
"loss": 4.1338,
"step": 383
},
{
"epoch": 0.86,
"learning_rate": 2.292e-05,
"loss": 4.0442,
"step": 384
},
{
"epoch": 0.86,
"learning_rate": 2.298e-05,
"loss": 4.1502,
"step": 385
},
{
"epoch": 0.87,
"learning_rate": 2.304e-05,
"loss": 4.0203,
"step": 386
},
{
"epoch": 0.87,
"learning_rate": 2.3100000000000002e-05,
"loss": 4.2042,
"step": 387
},
{
"epoch": 0.87,
"learning_rate": 2.3160000000000002e-05,
"loss": 4.2338,
"step": 388
},
{
"epoch": 0.87,
"learning_rate": 2.322e-05,
"loss": 4.1295,
"step": 389
},
{
"epoch": 0.87,
"learning_rate": 2.328e-05,
"loss": 4.1054,
"step": 390
},
{
"epoch": 0.88,
"learning_rate": 2.334e-05,
"loss": 4.2366,
"step": 391
},
{
"epoch": 0.88,
"learning_rate": 2.3400000000000003e-05,
"loss": 4.0665,
"step": 392
},
{
"epoch": 0.88,
"learning_rate": 2.3460000000000002e-05,
"loss": 4.0721,
"step": 393
},
{
"epoch": 0.88,
"learning_rate": 2.3520000000000002e-05,
"loss": 4.1672,
"step": 394
},
{
"epoch": 0.89,
"learning_rate": 2.358e-05,
"loss": 4.3088,
"step": 395
},
{
"epoch": 0.89,
"learning_rate": 2.364e-05,
"loss": 4.2441,
"step": 396
},
{
"epoch": 0.89,
"learning_rate": 2.37e-05,
"loss": 4.386,
"step": 397
},
{
"epoch": 0.89,
"learning_rate": 2.3760000000000003e-05,
"loss": 4.3494,
"step": 398
},
{
"epoch": 0.89,
"learning_rate": 2.3820000000000002e-05,
"loss": 4.4878,
"step": 399
},
{
"epoch": 0.9,
"learning_rate": 2.3880000000000002e-05,
"loss": 4.4159,
"step": 400
},
{
"epoch": 0.9,
"learning_rate": 2.394e-05,
"loss": 4.1722,
"step": 401
},
{
"epoch": 0.9,
"learning_rate": 2.4e-05,
"loss": 4.0105,
"step": 402
},
{
"epoch": 0.9,
"learning_rate": 2.4060000000000003e-05,
"loss": 4.0866,
"step": 403
},
{
"epoch": 0.91,
"learning_rate": 2.4120000000000003e-05,
"loss": 4.0665,
"step": 404
},
{
"epoch": 0.91,
"learning_rate": 2.4180000000000002e-05,
"loss": 4.0443,
"step": 405
},
{
"epoch": 0.91,
"learning_rate": 2.4240000000000002e-05,
"loss": 4.1285,
"step": 406
},
{
"epoch": 0.91,
"learning_rate": 2.43e-05,
"loss": 4.0993,
"step": 407
},
{
"epoch": 0.91,
"learning_rate": 2.4360000000000004e-05,
"loss": 4.1537,
"step": 408
},
{
"epoch": 0.92,
"learning_rate": 2.442e-05,
"loss": 4.1082,
"step": 409
},
{
"epoch": 0.92,
"learning_rate": 2.448e-05,
"loss": 4.0856,
"step": 410
},
{
"epoch": 0.92,
"learning_rate": 2.454e-05,
"loss": 4.3065,
"step": 411
},
{
"epoch": 0.92,
"learning_rate": 2.4599999999999998e-05,
"loss": 4.0353,
"step": 412
},
{
"epoch": 0.93,
"learning_rate": 2.4659999999999998e-05,
"loss": 3.992,
"step": 413
},
{
"epoch": 0.93,
"learning_rate": 2.472e-05,
"loss": 4.1374,
"step": 414
},
{
"epoch": 0.93,
"learning_rate": 2.478e-05,
"loss": 4.2158,
"step": 415
},
{
"epoch": 0.93,
"learning_rate": 2.484e-05,
"loss": 4.0603,
"step": 416
},
{
"epoch": 0.93,
"learning_rate": 2.49e-05,
"loss": 4.0957,
"step": 417
},
{
"epoch": 0.94,
"learning_rate": 2.4959999999999998e-05,
"loss": 3.9952,
"step": 418
},
{
"epoch": 0.94,
"learning_rate": 2.502e-05,
"loss": 4.0886,
"step": 419
},
{
"epoch": 0.94,
"learning_rate": 2.508e-05,
"loss": 4.0119,
"step": 420
},
{
"epoch": 0.94,
"learning_rate": 2.514e-05,
"loss": 4.1779,
"step": 421
},
{
"epoch": 0.95,
"learning_rate": 2.52e-05,
"loss": 4.0311,
"step": 422
},
{
"epoch": 0.95,
"learning_rate": 2.526e-05,
"loss": 4.1313,
"step": 423
},
{
"epoch": 0.95,
"learning_rate": 2.5319999999999998e-05,
"loss": 4.1833,
"step": 424
},
{
"epoch": 0.95,
"learning_rate": 2.538e-05,
"loss": 4.1238,
"step": 425
},
{
"epoch": 0.96,
"learning_rate": 2.544e-05,
"loss": 4.1676,
"step": 426
},
{
"epoch": 0.96,
"learning_rate": 2.55e-05,
"loss": 4.1039,
"step": 427
},
{
"epoch": 0.96,
"learning_rate": 2.556e-05,
"loss": 4.0975,
"step": 428
},
{
"epoch": 0.96,
"learning_rate": 2.562e-05,
"loss": 4.0224,
"step": 429
},
{
"epoch": 0.96,
"learning_rate": 2.568e-05,
"loss": 4.1026,
"step": 430
},
{
"epoch": 0.97,
"learning_rate": 2.574e-05,
"loss": 4.194,
"step": 431
},
{
"epoch": 0.97,
"learning_rate": 2.58e-05,
"loss": 4.305,
"step": 432
},
{
"epoch": 0.97,
"learning_rate": 2.586e-05,
"loss": 4.1124,
"step": 433
},
{
"epoch": 0.97,
"learning_rate": 2.592e-05,
"loss": 4.11,
"step": 434
},
{
"epoch": 0.98,
"learning_rate": 2.5980000000000002e-05,
"loss": 4.1639,
"step": 435
},
{
"epoch": 0.98,
"learning_rate": 2.604e-05,
"loss": 4.0736,
"step": 436
},
{
"epoch": 0.98,
"learning_rate": 2.61e-05,
"loss": 4.0552,
"step": 437
},
{
"epoch": 0.98,
"learning_rate": 2.616e-05,
"loss": 4.235,
"step": 438
},
{
"epoch": 0.98,
"learning_rate": 2.622e-05,
"loss": 4.1872,
"step": 439
},
{
"epoch": 0.99,
"learning_rate": 2.628e-05,
"loss": 4.2574,
"step": 440
},
{
"epoch": 0.99,
"learning_rate": 2.6340000000000002e-05,
"loss": 4.1761,
"step": 441
},
{
"epoch": 0.99,
"learning_rate": 2.64e-05,
"loss": 4.3863,
"step": 442
},
{
"epoch": 0.99,
"learning_rate": 2.646e-05,
"loss": 4.3228,
"step": 443
},
{
"epoch": 1.0,
"learning_rate": 2.652e-05,
"loss": 4.2573,
"step": 444
},
{
"epoch": 1.0,
"learning_rate": 2.658e-05,
"loss": 4.5512,
"step": 445
},
{
"epoch": 1.0,
"learning_rate": 2.6640000000000002e-05,
"loss": 4.2274,
"step": 446
},
{
"epoch": 1.0,
"step": 446,
"total_flos": 0.0,
"train_loss": 4.292909586376139,
"train_runtime": 4936.2943,
"train_samples_per_second": 5.781,
"train_steps_per_second": 0.09
}
],
"max_steps": 446,
"num_train_epochs": 1,
"total_flos": 0.0,
"trial_name": null,
"trial_params": null
}