wav2vec2-gpt2-wandb-grid-search / trainer_state.json
sanchit-gandhi's picture
End of training
46e3515
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": 0.0,
"loss": 4.7746,
"step": 1
},
{
"epoch": 0.0,
"learning_rate": 6.000000000000001e-08,
"loss": 4.9289,
"step": 2
},
{
"epoch": 0.01,
"learning_rate": 1.2000000000000002e-07,
"loss": 4.8883,
"step": 3
},
{
"epoch": 0.01,
"learning_rate": 1.8e-07,
"loss": 4.8789,
"step": 4
},
{
"epoch": 0.01,
"learning_rate": 2.4000000000000003e-07,
"loss": 4.8296,
"step": 5
},
{
"epoch": 0.01,
"learning_rate": 3.0000000000000004e-07,
"loss": 4.8764,
"step": 6
},
{
"epoch": 0.02,
"learning_rate": 3.6e-07,
"loss": 4.8601,
"step": 7
},
{
"epoch": 0.02,
"learning_rate": 4.2e-07,
"loss": 4.9133,
"step": 8
},
{
"epoch": 0.02,
"learning_rate": 4.800000000000001e-07,
"loss": 4.916,
"step": 9
},
{
"epoch": 0.02,
"learning_rate": 5.4e-07,
"loss": 4.7759,
"step": 10
},
{
"epoch": 0.02,
"learning_rate": 6.000000000000001e-07,
"loss": 4.8958,
"step": 11
},
{
"epoch": 0.03,
"learning_rate": 6.599999999999999e-07,
"loss": 4.8645,
"step": 12
},
{
"epoch": 0.03,
"learning_rate": 7.2e-07,
"loss": 4.8147,
"step": 13
},
{
"epoch": 0.03,
"learning_rate": 7.799999999999999e-07,
"loss": 4.8259,
"step": 14
},
{
"epoch": 0.03,
"learning_rate": 8.4e-07,
"loss": 4.7034,
"step": 15
},
{
"epoch": 0.04,
"learning_rate": 9e-07,
"loss": 4.8451,
"step": 16
},
{
"epoch": 0.04,
"learning_rate": 9.600000000000001e-07,
"loss": 4.7678,
"step": 17
},
{
"epoch": 0.04,
"learning_rate": 1.0200000000000002e-06,
"loss": 4.8715,
"step": 18
},
{
"epoch": 0.04,
"learning_rate": 1.08e-06,
"loss": 4.6656,
"step": 19
},
{
"epoch": 0.04,
"learning_rate": 1.14e-06,
"loss": 4.8439,
"step": 20
},
{
"epoch": 0.05,
"learning_rate": 1.2000000000000002e-06,
"loss": 4.8241,
"step": 21
},
{
"epoch": 0.05,
"learning_rate": 1.26e-06,
"loss": 4.7418,
"step": 22
},
{
"epoch": 0.05,
"learning_rate": 1.3199999999999999e-06,
"loss": 4.8523,
"step": 23
},
{
"epoch": 0.05,
"learning_rate": 1.38e-06,
"loss": 4.8442,
"step": 24
},
{
"epoch": 0.06,
"learning_rate": 1.44e-06,
"loss": 4.6269,
"step": 25
},
{
"epoch": 0.06,
"learning_rate": 1.5e-06,
"loss": 4.8907,
"step": 26
},
{
"epoch": 0.06,
"learning_rate": 1.5599999999999999e-06,
"loss": 4.7369,
"step": 27
},
{
"epoch": 0.06,
"learning_rate": 1.62e-06,
"loss": 4.5892,
"step": 28
},
{
"epoch": 0.07,
"learning_rate": 1.68e-06,
"loss": 4.564,
"step": 29
},
{
"epoch": 0.07,
"learning_rate": 1.74e-06,
"loss": 4.7634,
"step": 30
},
{
"epoch": 0.07,
"learning_rate": 1.8e-06,
"loss": 4.735,
"step": 31
},
{
"epoch": 0.07,
"learning_rate": 1.86e-06,
"loss": 4.5768,
"step": 32
},
{
"epoch": 0.07,
"learning_rate": 1.9200000000000003e-06,
"loss": 4.6719,
"step": 33
},
{
"epoch": 0.08,
"learning_rate": 1.98e-06,
"loss": 4.5934,
"step": 34
},
{
"epoch": 0.08,
"learning_rate": 2.0400000000000004e-06,
"loss": 4.6764,
"step": 35
},
{
"epoch": 0.08,
"learning_rate": 2.1000000000000002e-06,
"loss": 4.6146,
"step": 36
},
{
"epoch": 0.08,
"learning_rate": 2.16e-06,
"loss": 4.6754,
"step": 37
},
{
"epoch": 0.09,
"learning_rate": 2.22e-06,
"loss": 4.6119,
"step": 38
},
{
"epoch": 0.09,
"learning_rate": 2.28e-06,
"loss": 4.7453,
"step": 39
},
{
"epoch": 0.09,
"learning_rate": 2.34e-06,
"loss": 4.5306,
"step": 40
},
{
"epoch": 0.09,
"learning_rate": 2.4000000000000003e-06,
"loss": 4.595,
"step": 41
},
{
"epoch": 0.09,
"learning_rate": 2.46e-06,
"loss": 4.7049,
"step": 42
},
{
"epoch": 0.1,
"learning_rate": 2.52e-06,
"loss": 4.4633,
"step": 43
},
{
"epoch": 0.1,
"learning_rate": 2.58e-06,
"loss": 4.6159,
"step": 44
},
{
"epoch": 0.1,
"learning_rate": 2.6399999999999997e-06,
"loss": 4.7318,
"step": 45
},
{
"epoch": 0.1,
"learning_rate": 2.7e-06,
"loss": 4.8312,
"step": 46
},
{
"epoch": 0.11,
"learning_rate": 2.76e-06,
"loss": 4.7198,
"step": 47
},
{
"epoch": 0.11,
"learning_rate": 2.82e-06,
"loss": 4.9727,
"step": 48
},
{
"epoch": 0.11,
"learning_rate": 2.88e-06,
"loss": 5.057,
"step": 49
},
{
"epoch": 0.11,
"learning_rate": 2.88e-06,
"loss": 4.837,
"step": 50
},
{
"epoch": 0.11,
"learning_rate": 2.9400000000000002e-06,
"loss": 4.4908,
"step": 51
},
{
"epoch": 0.12,
"learning_rate": 3e-06,
"loss": 4.3346,
"step": 52
},
{
"epoch": 0.12,
"learning_rate": 3.06e-06,
"loss": 4.5022,
"step": 53
},
{
"epoch": 0.12,
"learning_rate": 3.1199999999999998e-06,
"loss": 4.4011,
"step": 54
},
{
"epoch": 0.12,
"learning_rate": 3.18e-06,
"loss": 4.2133,
"step": 55
},
{
"epoch": 0.13,
"learning_rate": 3.24e-06,
"loss": 4.4502,
"step": 56
},
{
"epoch": 0.13,
"learning_rate": 3.3e-06,
"loss": 4.4597,
"step": 57
},
{
"epoch": 0.13,
"learning_rate": 3.36e-06,
"loss": 4.5472,
"step": 58
},
{
"epoch": 0.13,
"learning_rate": 3.4200000000000003e-06,
"loss": 4.4784,
"step": 59
},
{
"epoch": 0.13,
"learning_rate": 3.48e-06,
"loss": 4.3192,
"step": 60
},
{
"epoch": 0.14,
"learning_rate": 3.54e-06,
"loss": 4.4132,
"step": 61
},
{
"epoch": 0.14,
"learning_rate": 3.6e-06,
"loss": 4.3723,
"step": 62
},
{
"epoch": 0.14,
"learning_rate": 3.66e-06,
"loss": 4.3143,
"step": 63
},
{
"epoch": 0.14,
"learning_rate": 3.72e-06,
"loss": 4.4111,
"step": 64
},
{
"epoch": 0.15,
"learning_rate": 3.7800000000000002e-06,
"loss": 4.381,
"step": 65
},
{
"epoch": 0.15,
"learning_rate": 3.8400000000000005e-06,
"loss": 4.2726,
"step": 66
},
{
"epoch": 0.15,
"learning_rate": 3.9e-06,
"loss": 4.4058,
"step": 67
},
{
"epoch": 0.15,
"learning_rate": 3.96e-06,
"loss": 4.269,
"step": 68
},
{
"epoch": 0.15,
"learning_rate": 4.0200000000000005e-06,
"loss": 4.3351,
"step": 69
},
{
"epoch": 0.16,
"learning_rate": 4.080000000000001e-06,
"loss": 4.2911,
"step": 70
},
{
"epoch": 0.16,
"learning_rate": 4.14e-06,
"loss": 4.434,
"step": 71
},
{
"epoch": 0.16,
"learning_rate": 4.2000000000000004e-06,
"loss": 4.4303,
"step": 72
},
{
"epoch": 0.16,
"learning_rate": 4.26e-06,
"loss": 4.2569,
"step": 73
},
{
"epoch": 0.17,
"learning_rate": 4.32e-06,
"loss": 4.3597,
"step": 74
},
{
"epoch": 0.17,
"learning_rate": 4.3799999999999996e-06,
"loss": 4.3462,
"step": 75
},
{
"epoch": 0.17,
"learning_rate": 4.44e-06,
"loss": 4.3001,
"step": 76
},
{
"epoch": 0.17,
"learning_rate": 4.5e-06,
"loss": 4.3445,
"step": 77
},
{
"epoch": 0.17,
"learning_rate": 4.56e-06,
"loss": 4.2932,
"step": 78
},
{
"epoch": 0.18,
"learning_rate": 4.62e-06,
"loss": 4.267,
"step": 79
},
{
"epoch": 0.18,
"learning_rate": 4.68e-06,
"loss": 4.4182,
"step": 80
},
{
"epoch": 0.18,
"learning_rate": 4.74e-06,
"loss": 4.3713,
"step": 81
},
{
"epoch": 0.18,
"learning_rate": 4.800000000000001e-06,
"loss": 4.3061,
"step": 82
},
{
"epoch": 0.19,
"learning_rate": 4.86e-06,
"loss": 4.4143,
"step": 83
},
{
"epoch": 0.19,
"learning_rate": 4.92e-06,
"loss": 4.2805,
"step": 84
},
{
"epoch": 0.19,
"learning_rate": 4.980000000000001e-06,
"loss": 4.3571,
"step": 85
},
{
"epoch": 0.19,
"learning_rate": 5.04e-06,
"loss": 4.3305,
"step": 86
},
{
"epoch": 0.2,
"learning_rate": 5.1e-06,
"loss": 4.4209,
"step": 87
},
{
"epoch": 0.2,
"learning_rate": 5.16e-06,
"loss": 4.2915,
"step": 88
},
{
"epoch": 0.2,
"learning_rate": 5.22e-06,
"loss": 4.4348,
"step": 89
},
{
"epoch": 0.2,
"learning_rate": 5.279999999999999e-06,
"loss": 4.4178,
"step": 90
},
{
"epoch": 0.2,
"learning_rate": 5.34e-06,
"loss": 4.3064,
"step": 91
},
{
"epoch": 0.21,
"learning_rate": 5.4e-06,
"loss": 4.5617,
"step": 92
},
{
"epoch": 0.21,
"learning_rate": 5.46e-06,
"loss": 4.4422,
"step": 93
},
{
"epoch": 0.21,
"learning_rate": 5.52e-06,
"loss": 4.4235,
"step": 94
},
{
"epoch": 0.21,
"learning_rate": 5.58e-06,
"loss": 4.3727,
"step": 95
},
{
"epoch": 0.22,
"learning_rate": 5.64e-06,
"loss": 4.5854,
"step": 96
},
{
"epoch": 0.22,
"learning_rate": 5.7000000000000005e-06,
"loss": 4.5272,
"step": 97
},
{
"epoch": 0.22,
"learning_rate": 5.76e-06,
"loss": 4.3841,
"step": 98
},
{
"epoch": 0.22,
"learning_rate": 5.82e-06,
"loss": 4.711,
"step": 99
},
{
"epoch": 0.22,
"learning_rate": 5.8800000000000005e-06,
"loss": 4.4718,
"step": 100
},
{
"epoch": 0.23,
"learning_rate": 5.940000000000001e-06,
"loss": 4.3046,
"step": 101
},
{
"epoch": 0.23,
"learning_rate": 6e-06,
"loss": 4.2435,
"step": 102
},
{
"epoch": 0.23,
"learning_rate": 6.0600000000000004e-06,
"loss": 4.113,
"step": 103
},
{
"epoch": 0.23,
"learning_rate": 6.12e-06,
"loss": 4.1906,
"step": 104
},
{
"epoch": 0.24,
"learning_rate": 6.18e-06,
"loss": 4.1371,
"step": 105
},
{
"epoch": 0.24,
"learning_rate": 6.2399999999999995e-06,
"loss": 4.1835,
"step": 106
},
{
"epoch": 0.24,
"learning_rate": 6.3e-06,
"loss": 4.28,
"step": 107
},
{
"epoch": 0.24,
"learning_rate": 6.36e-06,
"loss": 4.1821,
"step": 108
},
{
"epoch": 0.24,
"learning_rate": 6.42e-06,
"loss": 4.1884,
"step": 109
},
{
"epoch": 0.25,
"learning_rate": 6.48e-06,
"loss": 4.2333,
"step": 110
},
{
"epoch": 0.25,
"learning_rate": 6.54e-06,
"loss": 4.2675,
"step": 111
},
{
"epoch": 0.25,
"learning_rate": 6.6e-06,
"loss": 4.2355,
"step": 112
},
{
"epoch": 0.25,
"learning_rate": 6.660000000000001e-06,
"loss": 4.2424,
"step": 113
},
{
"epoch": 0.26,
"learning_rate": 6.72e-06,
"loss": 4.2007,
"step": 114
},
{
"epoch": 0.26,
"learning_rate": 6.78e-06,
"loss": 4.2562,
"step": 115
},
{
"epoch": 0.26,
"learning_rate": 6.840000000000001e-06,
"loss": 4.3384,
"step": 116
},
{
"epoch": 0.26,
"learning_rate": 6.900000000000001e-06,
"loss": 4.217,
"step": 117
},
{
"epoch": 0.26,
"learning_rate": 6.96e-06,
"loss": 4.37,
"step": 118
},
{
"epoch": 0.27,
"learning_rate": 7.0200000000000006e-06,
"loss": 4.2666,
"step": 119
},
{
"epoch": 0.27,
"learning_rate": 7.08e-06,
"loss": 4.292,
"step": 120
},
{
"epoch": 0.27,
"learning_rate": 7.14e-06,
"loss": 4.1314,
"step": 121
},
{
"epoch": 0.27,
"learning_rate": 7.2e-06,
"loss": 4.2254,
"step": 122
},
{
"epoch": 0.28,
"learning_rate": 7.26e-06,
"loss": 4.236,
"step": 123
},
{
"epoch": 0.28,
"learning_rate": 7.32e-06,
"loss": 4.2277,
"step": 124
},
{
"epoch": 0.28,
"learning_rate": 7.3800000000000005e-06,
"loss": 4.3383,
"step": 125
},
{
"epoch": 0.28,
"learning_rate": 7.44e-06,
"loss": 4.3372,
"step": 126
},
{
"epoch": 0.28,
"learning_rate": 7.5e-06,
"loss": 4.2553,
"step": 127
},
{
"epoch": 0.29,
"learning_rate": 7.5600000000000005e-06,
"loss": 4.2319,
"step": 128
},
{
"epoch": 0.29,
"learning_rate": 7.62e-06,
"loss": 4.1354,
"step": 129
},
{
"epoch": 0.29,
"learning_rate": 7.680000000000001e-06,
"loss": 4.2859,
"step": 130
},
{
"epoch": 0.29,
"learning_rate": 7.74e-06,
"loss": 4.1919,
"step": 131
},
{
"epoch": 0.3,
"learning_rate": 7.8e-06,
"loss": 4.2501,
"step": 132
},
{
"epoch": 0.3,
"learning_rate": 7.860000000000001e-06,
"loss": 4.1723,
"step": 133
},
{
"epoch": 0.3,
"learning_rate": 7.92e-06,
"loss": 4.2081,
"step": 134
},
{
"epoch": 0.3,
"learning_rate": 7.98e-06,
"loss": 4.3765,
"step": 135
},
{
"epoch": 0.3,
"learning_rate": 8.040000000000001e-06,
"loss": 4.2239,
"step": 136
},
{
"epoch": 0.31,
"learning_rate": 8.1e-06,
"loss": 4.3212,
"step": 137
},
{
"epoch": 0.31,
"learning_rate": 8.160000000000001e-06,
"loss": 4.3565,
"step": 138
},
{
"epoch": 0.31,
"learning_rate": 8.220000000000001e-06,
"loss": 4.3689,
"step": 139
},
{
"epoch": 0.31,
"learning_rate": 8.28e-06,
"loss": 4.3316,
"step": 140
},
{
"epoch": 0.32,
"learning_rate": 8.340000000000001e-06,
"loss": 4.4513,
"step": 141
},
{
"epoch": 0.32,
"learning_rate": 8.400000000000001e-06,
"loss": 4.3836,
"step": 142
},
{
"epoch": 0.32,
"learning_rate": 8.459999999999999e-06,
"loss": 4.3245,
"step": 143
},
{
"epoch": 0.32,
"learning_rate": 8.52e-06,
"loss": 4.3805,
"step": 144
},
{
"epoch": 0.33,
"learning_rate": 8.58e-06,
"loss": 4.2924,
"step": 145
},
{
"epoch": 0.33,
"learning_rate": 8.64e-06,
"loss": 4.3797,
"step": 146
},
{
"epoch": 0.33,
"learning_rate": 8.7e-06,
"loss": 4.6306,
"step": 147
},
{
"epoch": 0.33,
"learning_rate": 8.759999999999999e-06,
"loss": 4.77,
"step": 148
},
{
"epoch": 0.33,
"learning_rate": 8.82e-06,
"loss": 4.5055,
"step": 149
},
{
"epoch": 0.34,
"learning_rate": 8.88e-06,
"loss": 4.6005,
"step": 150
},
{
"epoch": 0.34,
"learning_rate": 8.939999999999999e-06,
"loss": 4.1378,
"step": 151
},
{
"epoch": 0.34,
"learning_rate": 9e-06,
"loss": 4.2224,
"step": 152
},
{
"epoch": 0.34,
"learning_rate": 9.06e-06,
"loss": 4.2385,
"step": 153
},
{
"epoch": 0.35,
"learning_rate": 9.12e-06,
"loss": 4.1296,
"step": 154
},
{
"epoch": 0.35,
"learning_rate": 9.18e-06,
"loss": 4.2382,
"step": 155
},
{
"epoch": 0.35,
"learning_rate": 9.24e-06,
"loss": 4.0975,
"step": 156
},
{
"epoch": 0.35,
"learning_rate": 9.3e-06,
"loss": 4.1335,
"step": 157
},
{
"epoch": 0.35,
"learning_rate": 9.36e-06,
"loss": 4.2322,
"step": 158
},
{
"epoch": 0.36,
"learning_rate": 9.42e-06,
"loss": 4.2269,
"step": 159
},
{
"epoch": 0.36,
"learning_rate": 9.48e-06,
"loss": 4.2434,
"step": 160
},
{
"epoch": 0.36,
"learning_rate": 9.54e-06,
"loss": 4.2971,
"step": 161
},
{
"epoch": 0.36,
"learning_rate": 9.600000000000001e-06,
"loss": 4.2479,
"step": 162
},
{
"epoch": 0.37,
"learning_rate": 9.66e-06,
"loss": 4.2019,
"step": 163
},
{
"epoch": 0.37,
"learning_rate": 9.72e-06,
"loss": 4.2686,
"step": 164
},
{
"epoch": 0.37,
"learning_rate": 9.780000000000001e-06,
"loss": 4.1306,
"step": 165
},
{
"epoch": 0.37,
"learning_rate": 9.84e-06,
"loss": 4.1961,
"step": 166
},
{
"epoch": 0.37,
"learning_rate": 9.9e-06,
"loss": 4.1072,
"step": 167
},
{
"epoch": 0.38,
"learning_rate": 9.960000000000001e-06,
"loss": 4.1856,
"step": 168
},
{
"epoch": 0.38,
"learning_rate": 1.002e-05,
"loss": 4.2457,
"step": 169
},
{
"epoch": 0.38,
"learning_rate": 1.008e-05,
"loss": 4.2701,
"step": 170
},
{
"epoch": 0.38,
"learning_rate": 1.0140000000000001e-05,
"loss": 4.2464,
"step": 171
},
{
"epoch": 0.39,
"learning_rate": 1.02e-05,
"loss": 4.3068,
"step": 172
},
{
"epoch": 0.39,
"learning_rate": 1.0260000000000002e-05,
"loss": 4.1162,
"step": 173
},
{
"epoch": 0.39,
"learning_rate": 1.032e-05,
"loss": 4.2632,
"step": 174
},
{
"epoch": 0.39,
"learning_rate": 1.0379999999999999e-05,
"loss": 4.3134,
"step": 175
},
{
"epoch": 0.39,
"learning_rate": 1.044e-05,
"loss": 4.1883,
"step": 176
},
{
"epoch": 0.4,
"learning_rate": 1.05e-05,
"loss": 4.1563,
"step": 177
},
{
"epoch": 0.4,
"learning_rate": 1.0559999999999999e-05,
"loss": 4.2408,
"step": 178
},
{
"epoch": 0.4,
"learning_rate": 1.062e-05,
"loss": 4.3123,
"step": 179
},
{
"epoch": 0.4,
"learning_rate": 1.068e-05,
"loss": 4.301,
"step": 180
},
{
"epoch": 0.41,
"learning_rate": 1.074e-05,
"loss": 4.2258,
"step": 181
},
{
"epoch": 0.41,
"learning_rate": 1.08e-05,
"loss": 4.2896,
"step": 182
},
{
"epoch": 0.41,
"learning_rate": 1.086e-05,
"loss": 4.2775,
"step": 183
},
{
"epoch": 0.41,
"learning_rate": 1.092e-05,
"loss": 4.2849,
"step": 184
},
{
"epoch": 0.41,
"learning_rate": 1.098e-05,
"loss": 4.2654,
"step": 185
},
{
"epoch": 0.42,
"learning_rate": 1.104e-05,
"loss": 4.128,
"step": 186
},
{
"epoch": 0.42,
"learning_rate": 1.11e-05,
"loss": 4.1517,
"step": 187
},
{
"epoch": 0.42,
"learning_rate": 1.116e-05,
"loss": 4.1204,
"step": 188
},
{
"epoch": 0.42,
"learning_rate": 1.1220000000000001e-05,
"loss": 4.2358,
"step": 189
},
{
"epoch": 0.43,
"learning_rate": 1.128e-05,
"loss": 4.3392,
"step": 190
},
{
"epoch": 0.43,
"learning_rate": 1.134e-05,
"loss": 4.254,
"step": 191
},
{
"epoch": 0.43,
"learning_rate": 1.1400000000000001e-05,
"loss": 4.2131,
"step": 192
},
{
"epoch": 0.43,
"learning_rate": 1.146e-05,
"loss": 4.2087,
"step": 193
},
{
"epoch": 0.43,
"learning_rate": 1.152e-05,
"loss": 4.3236,
"step": 194
},
{
"epoch": 0.44,
"learning_rate": 1.1580000000000001e-05,
"loss": 4.4341,
"step": 195
},
{
"epoch": 0.44,
"learning_rate": 1.164e-05,
"loss": 4.3805,
"step": 196
},
{
"epoch": 0.44,
"learning_rate": 1.1700000000000001e-05,
"loss": 4.3194,
"step": 197
},
{
"epoch": 0.44,
"learning_rate": 1.1760000000000001e-05,
"loss": 4.61,
"step": 198
},
{
"epoch": 0.45,
"learning_rate": 1.182e-05,
"loss": 4.5467,
"step": 199
},
{
"epoch": 0.45,
"learning_rate": 1.1880000000000001e-05,
"loss": 4.4941,
"step": 200
},
{
"epoch": 0.45,
"learning_rate": 1.1940000000000001e-05,
"loss": 4.2598,
"step": 201
},
{
"epoch": 0.45,
"learning_rate": 1.2e-05,
"loss": 4.1943,
"step": 202
},
{
"epoch": 0.46,
"learning_rate": 1.2060000000000001e-05,
"loss": 4.2136,
"step": 203
},
{
"epoch": 0.46,
"learning_rate": 1.2120000000000001e-05,
"loss": 4.209,
"step": 204
},
{
"epoch": 0.46,
"learning_rate": 1.2180000000000002e-05,
"loss": 4.2688,
"step": 205
},
{
"epoch": 0.46,
"learning_rate": 1.224e-05,
"loss": 4.2527,
"step": 206
},
{
"epoch": 0.46,
"learning_rate": 1.2299999999999999e-05,
"loss": 4.1229,
"step": 207
},
{
"epoch": 0.47,
"learning_rate": 1.236e-05,
"loss": 4.2149,
"step": 208
},
{
"epoch": 0.47,
"learning_rate": 1.242e-05,
"loss": 4.2757,
"step": 209
},
{
"epoch": 0.47,
"learning_rate": 1.2479999999999999e-05,
"loss": 4.2353,
"step": 210
},
{
"epoch": 0.47,
"learning_rate": 1.254e-05,
"loss": 4.1108,
"step": 211
},
{
"epoch": 0.48,
"learning_rate": 1.26e-05,
"loss": 4.16,
"step": 212
},
{
"epoch": 0.48,
"learning_rate": 1.2659999999999999e-05,
"loss": 4.0885,
"step": 213
},
{
"epoch": 0.48,
"learning_rate": 1.272e-05,
"loss": 4.1628,
"step": 214
},
{
"epoch": 0.48,
"learning_rate": 1.278e-05,
"loss": 4.209,
"step": 215
},
{
"epoch": 0.48,
"learning_rate": 1.284e-05,
"loss": 4.1769,
"step": 216
},
{
"epoch": 0.49,
"learning_rate": 1.29e-05,
"loss": 4.2236,
"step": 217
},
{
"epoch": 0.49,
"learning_rate": 1.296e-05,
"loss": 4.175,
"step": 218
},
{
"epoch": 0.49,
"learning_rate": 1.302e-05,
"loss": 4.1064,
"step": 219
},
{
"epoch": 0.49,
"learning_rate": 1.308e-05,
"loss": 4.2232,
"step": 220
},
{
"epoch": 0.5,
"learning_rate": 1.314e-05,
"loss": 4.1993,
"step": 221
},
{
"epoch": 0.5,
"learning_rate": 1.32e-05,
"loss": 4.1825,
"step": 222
},
{
"epoch": 0.5,
"learning_rate": 1.326e-05,
"loss": 4.2186,
"step": 223
},
{
"epoch": 0.5,
"learning_rate": 1.3320000000000001e-05,
"loss": 4.1414,
"step": 224
},
{
"epoch": 0.5,
"learning_rate": 1.338e-05,
"loss": 4.299,
"step": 225
},
{
"epoch": 0.51,
"learning_rate": 1.344e-05,
"loss": 4.1934,
"step": 226
},
{
"epoch": 0.51,
"learning_rate": 1.3500000000000001e-05,
"loss": 4.3162,
"step": 227
},
{
"epoch": 0.51,
"learning_rate": 1.356e-05,
"loss": 4.262,
"step": 228
},
{
"epoch": 0.51,
"learning_rate": 1.362e-05,
"loss": 4.1437,
"step": 229
},
{
"epoch": 0.52,
"learning_rate": 1.3680000000000001e-05,
"loss": 4.2367,
"step": 230
},
{
"epoch": 0.52,
"learning_rate": 1.374e-05,
"loss": 4.2988,
"step": 231
},
{
"epoch": 0.52,
"learning_rate": 1.3800000000000002e-05,
"loss": 4.0968,
"step": 232
},
{
"epoch": 0.52,
"learning_rate": 1.3860000000000001e-05,
"loss": 4.1227,
"step": 233
},
{
"epoch": 0.52,
"learning_rate": 1.392e-05,
"loss": 4.2133,
"step": 234
},
{
"epoch": 0.53,
"learning_rate": 1.3980000000000002e-05,
"loss": 4.2843,
"step": 235
},
{
"epoch": 0.53,
"learning_rate": 1.4040000000000001e-05,
"loss": 4.2125,
"step": 236
},
{
"epoch": 0.53,
"learning_rate": 1.4099999999999999e-05,
"loss": 4.2384,
"step": 237
},
{
"epoch": 0.53,
"learning_rate": 1.416e-05,
"loss": 4.1947,
"step": 238
},
{
"epoch": 0.54,
"learning_rate": 1.422e-05,
"loss": 4.1355,
"step": 239
},
{
"epoch": 0.54,
"learning_rate": 1.428e-05,
"loss": 4.2412,
"step": 240
},
{
"epoch": 0.54,
"learning_rate": 1.434e-05,
"loss": 4.2237,
"step": 241
},
{
"epoch": 0.54,
"learning_rate": 1.44e-05,
"loss": 4.2662,
"step": 242
},
{
"epoch": 0.54,
"learning_rate": 1.446e-05,
"loss": 4.4641,
"step": 243
},
{
"epoch": 0.55,
"learning_rate": 1.452e-05,
"loss": 4.2923,
"step": 244
},
{
"epoch": 0.55,
"learning_rate": 1.458e-05,
"loss": 4.2855,
"step": 245
},
{
"epoch": 0.55,
"learning_rate": 1.464e-05,
"loss": 4.1766,
"step": 246
},
{
"epoch": 0.55,
"learning_rate": 1.47e-05,
"loss": 4.1232,
"step": 247
},
{
"epoch": 0.56,
"learning_rate": 1.4760000000000001e-05,
"loss": 4.4921,
"step": 248
},
{
"epoch": 0.56,
"learning_rate": 1.482e-05,
"loss": 4.5888,
"step": 249
},
{
"epoch": 0.56,
"learning_rate": 1.488e-05,
"loss": 4.582,
"step": 250
},
{
"epoch": 0.56,
"learning_rate": 1.4940000000000001e-05,
"loss": 4.1951,
"step": 251
},
{
"epoch": 0.57,
"learning_rate": 1.5e-05,
"loss": 4.2269,
"step": 252
},
{
"epoch": 0.57,
"learning_rate": 1.506e-05,
"loss": 4.2679,
"step": 253
},
{
"epoch": 0.57,
"learning_rate": 1.5120000000000001e-05,
"loss": 4.1101,
"step": 254
},
{
"epoch": 0.57,
"learning_rate": 1.518e-05,
"loss": 4.1986,
"step": 255
},
{
"epoch": 0.57,
"learning_rate": 1.524e-05,
"loss": 4.179,
"step": 256
},
{
"epoch": 0.58,
"learning_rate": 1.53e-05,
"loss": 4.1623,
"step": 257
},
{
"epoch": 0.58,
"learning_rate": 1.5360000000000002e-05,
"loss": 4.1636,
"step": 258
},
{
"epoch": 0.58,
"learning_rate": 1.542e-05,
"loss": 4.1744,
"step": 259
},
{
"epoch": 0.58,
"learning_rate": 1.548e-05,
"loss": 4.1098,
"step": 260
},
{
"epoch": 0.59,
"learning_rate": 1.554e-05,
"loss": 4.2371,
"step": 261
},
{
"epoch": 0.59,
"learning_rate": 1.56e-05,
"loss": 4.1562,
"step": 262
},
{
"epoch": 0.59,
"learning_rate": 1.5660000000000003e-05,
"loss": 4.294,
"step": 263
},
{
"epoch": 0.59,
"learning_rate": 1.5720000000000002e-05,
"loss": 4.1616,
"step": 264
},
{
"epoch": 0.59,
"learning_rate": 1.578e-05,
"loss": 4.2108,
"step": 265
},
{
"epoch": 0.6,
"learning_rate": 1.584e-05,
"loss": 4.1689,
"step": 266
},
{
"epoch": 0.6,
"learning_rate": 1.59e-05,
"loss": 4.2019,
"step": 267
},
{
"epoch": 0.6,
"learning_rate": 1.596e-05,
"loss": 4.1509,
"step": 268
},
{
"epoch": 0.6,
"learning_rate": 1.6020000000000002e-05,
"loss": 4.2425,
"step": 269
},
{
"epoch": 0.61,
"learning_rate": 1.6080000000000002e-05,
"loss": 4.1185,
"step": 270
},
{
"epoch": 0.61,
"learning_rate": 1.614e-05,
"loss": 4.0561,
"step": 271
},
{
"epoch": 0.61,
"learning_rate": 1.62e-05,
"loss": 4.2334,
"step": 272
},
{
"epoch": 0.61,
"learning_rate": 1.626e-05,
"loss": 4.03,
"step": 273
},
{
"epoch": 0.61,
"learning_rate": 1.6320000000000003e-05,
"loss": 4.1374,
"step": 274
},
{
"epoch": 0.62,
"learning_rate": 1.6380000000000002e-05,
"loss": 4.2188,
"step": 275
},
{
"epoch": 0.62,
"learning_rate": 1.6440000000000002e-05,
"loss": 4.1667,
"step": 276
},
{
"epoch": 0.62,
"learning_rate": 1.65e-05,
"loss": 4.1154,
"step": 277
},
{
"epoch": 0.62,
"learning_rate": 1.656e-05,
"loss": 4.1921,
"step": 278
},
{
"epoch": 0.63,
"learning_rate": 1.6620000000000004e-05,
"loss": 4.1726,
"step": 279
},
{
"epoch": 0.63,
"learning_rate": 1.6680000000000003e-05,
"loss": 4.3145,
"step": 280
},
{
"epoch": 0.63,
"learning_rate": 1.6740000000000002e-05,
"loss": 4.2087,
"step": 281
},
{
"epoch": 0.63,
"learning_rate": 1.6800000000000002e-05,
"loss": 4.0653,
"step": 282
},
{
"epoch": 0.63,
"learning_rate": 1.686e-05,
"loss": 4.0902,
"step": 283
},
{
"epoch": 0.64,
"learning_rate": 1.6919999999999997e-05,
"loss": 4.1814,
"step": 284
},
{
"epoch": 0.64,
"learning_rate": 1.698e-05,
"loss": 4.2828,
"step": 285
},
{
"epoch": 0.64,
"learning_rate": 1.704e-05,
"loss": 4.1693,
"step": 286
},
{
"epoch": 0.64,
"learning_rate": 1.71e-05,
"loss": 4.2818,
"step": 287
},
{
"epoch": 0.65,
"learning_rate": 1.716e-05,
"loss": 4.2331,
"step": 288
},
{
"epoch": 0.65,
"learning_rate": 1.7219999999999998e-05,
"loss": 4.1357,
"step": 289
},
{
"epoch": 0.65,
"learning_rate": 1.728e-05,
"loss": 4.1985,
"step": 290
},
{
"epoch": 0.65,
"learning_rate": 1.734e-05,
"loss": 4.4431,
"step": 291
},
{
"epoch": 0.65,
"learning_rate": 1.74e-05,
"loss": 4.3144,
"step": 292
},
{
"epoch": 0.66,
"learning_rate": 1.746e-05,
"loss": 4.3839,
"step": 293
},
{
"epoch": 0.66,
"learning_rate": 1.7519999999999998e-05,
"loss": 4.3518,
"step": 294
},
{
"epoch": 0.66,
"learning_rate": 1.758e-05,
"loss": 4.3501,
"step": 295
},
{
"epoch": 0.66,
"learning_rate": 1.764e-05,
"loss": 4.3526,
"step": 296
},
{
"epoch": 0.67,
"learning_rate": 1.77e-05,
"loss": 4.2403,
"step": 297
},
{
"epoch": 0.67,
"learning_rate": 1.776e-05,
"loss": 4.5569,
"step": 298
},
{
"epoch": 0.67,
"learning_rate": 1.782e-05,
"loss": 4.3044,
"step": 299
},
{
"epoch": 0.67,
"learning_rate": 1.7879999999999998e-05,
"loss": 4.7419,
"step": 300
},
{
"epoch": 0.67,
"learning_rate": 1.794e-05,
"loss": 4.2255,
"step": 301
},
{
"epoch": 0.68,
"learning_rate": 1.8e-05,
"loss": 4.1911,
"step": 302
},
{
"epoch": 0.68,
"learning_rate": 1.806e-05,
"loss": 4.1913,
"step": 303
},
{
"epoch": 0.68,
"learning_rate": 1.812e-05,
"loss": 4.2035,
"step": 304
},
{
"epoch": 0.68,
"learning_rate": 1.818e-05,
"loss": 4.1659,
"step": 305
},
{
"epoch": 0.69,
"learning_rate": 1.824e-05,
"loss": 4.1522,
"step": 306
},
{
"epoch": 0.69,
"learning_rate": 1.83e-05,
"loss": 4.1422,
"step": 307
},
{
"epoch": 0.69,
"learning_rate": 1.836e-05,
"loss": 4.148,
"step": 308
},
{
"epoch": 0.69,
"learning_rate": 1.842e-05,
"loss": 4.1458,
"step": 309
},
{
"epoch": 0.7,
"learning_rate": 1.848e-05,
"loss": 4.2421,
"step": 310
},
{
"epoch": 0.7,
"learning_rate": 1.854e-05,
"loss": 4.0367,
"step": 311
},
{
"epoch": 0.7,
"learning_rate": 1.86e-05,
"loss": 4.2137,
"step": 312
},
{
"epoch": 0.7,
"learning_rate": 1.866e-05,
"loss": 4.2195,
"step": 313
},
{
"epoch": 0.7,
"learning_rate": 1.872e-05,
"loss": 4.1313,
"step": 314
},
{
"epoch": 0.71,
"learning_rate": 1.878e-05,
"loss": 4.165,
"step": 315
},
{
"epoch": 0.71,
"learning_rate": 1.884e-05,
"loss": 4.1387,
"step": 316
},
{
"epoch": 0.71,
"learning_rate": 1.8900000000000002e-05,
"loss": 4.1316,
"step": 317
},
{
"epoch": 0.71,
"learning_rate": 1.896e-05,
"loss": 4.164,
"step": 318
},
{
"epoch": 0.72,
"learning_rate": 1.902e-05,
"loss": 4.1583,
"step": 319
},
{
"epoch": 0.72,
"learning_rate": 1.908e-05,
"loss": 4.2331,
"step": 320
},
{
"epoch": 0.72,
"learning_rate": 1.914e-05,
"loss": 4.1241,
"step": 321
},
{
"epoch": 0.72,
"learning_rate": 1.9200000000000003e-05,
"loss": 3.9994,
"step": 322
},
{
"epoch": 0.72,
"learning_rate": 1.9260000000000002e-05,
"loss": 4.1963,
"step": 323
},
{
"epoch": 0.73,
"learning_rate": 1.932e-05,
"loss": 4.05,
"step": 324
},
{
"epoch": 0.73,
"learning_rate": 1.938e-05,
"loss": 4.2003,
"step": 325
},
{
"epoch": 0.73,
"learning_rate": 1.944e-05,
"loss": 4.1707,
"step": 326
},
{
"epoch": 0.73,
"learning_rate": 1.95e-05,
"loss": 4.1238,
"step": 327
},
{
"epoch": 0.74,
"learning_rate": 1.9560000000000002e-05,
"loss": 4.0829,
"step": 328
},
{
"epoch": 0.74,
"learning_rate": 1.9620000000000002e-05,
"loss": 4.1401,
"step": 329
},
{
"epoch": 0.74,
"learning_rate": 1.968e-05,
"loss": 4.1984,
"step": 330
},
{
"epoch": 0.74,
"learning_rate": 1.974e-05,
"loss": 4.2043,
"step": 331
},
{
"epoch": 0.74,
"learning_rate": 1.98e-05,
"loss": 4.0911,
"step": 332
},
{
"epoch": 0.75,
"learning_rate": 1.9860000000000003e-05,
"loss": 4.2525,
"step": 333
},
{
"epoch": 0.75,
"learning_rate": 1.9920000000000002e-05,
"loss": 4.1515,
"step": 334
},
{
"epoch": 0.75,
"learning_rate": 1.9980000000000002e-05,
"loss": 4.3075,
"step": 335
},
{
"epoch": 0.75,
"learning_rate": 2.004e-05,
"loss": 4.1172,
"step": 336
},
{
"epoch": 0.76,
"learning_rate": 2.01e-05,
"loss": 4.1193,
"step": 337
},
{
"epoch": 0.76,
"learning_rate": 2.016e-05,
"loss": 4.2335,
"step": 338
},
{
"epoch": 0.76,
"learning_rate": 2.0220000000000003e-05,
"loss": 4.2131,
"step": 339
},
{
"epoch": 0.76,
"learning_rate": 2.0280000000000002e-05,
"loss": 4.264,
"step": 340
},
{
"epoch": 0.76,
"learning_rate": 2.0340000000000002e-05,
"loss": 4.1833,
"step": 341
},
{
"epoch": 0.77,
"learning_rate": 2.04e-05,
"loss": 4.3401,
"step": 342
},
{
"epoch": 0.77,
"learning_rate": 2.046e-05,
"loss": 4.2661,
"step": 343
},
{
"epoch": 0.77,
"learning_rate": 2.0520000000000003e-05,
"loss": 4.3547,
"step": 344
},
{
"epoch": 0.77,
"learning_rate": 2.0580000000000003e-05,
"loss": 4.096,
"step": 345
},
{
"epoch": 0.78,
"learning_rate": 2.064e-05,
"loss": 4.339,
"step": 346
},
{
"epoch": 0.78,
"learning_rate": 2.07e-05,
"loss": 4.3258,
"step": 347
},
{
"epoch": 0.78,
"learning_rate": 2.0759999999999998e-05,
"loss": 4.2641,
"step": 348
},
{
"epoch": 0.78,
"learning_rate": 2.082e-05,
"loss": 4.4812,
"step": 349
},
{
"epoch": 0.78,
"learning_rate": 2.088e-05,
"loss": 4.4627,
"step": 350
},
{
"epoch": 0.79,
"learning_rate": 2.094e-05,
"loss": 4.2444,
"step": 351
},
{
"epoch": 0.79,
"learning_rate": 2.1e-05,
"loss": 4.1637,
"step": 352
},
{
"epoch": 0.79,
"learning_rate": 2.1059999999999998e-05,
"loss": 4.1012,
"step": 353
},
{
"epoch": 0.79,
"learning_rate": 2.1119999999999998e-05,
"loss": 3.9923,
"step": 354
},
{
"epoch": 0.8,
"learning_rate": 2.118e-05,
"loss": 4.175,
"step": 355
},
{
"epoch": 0.8,
"learning_rate": 2.124e-05,
"loss": 4.2816,
"step": 356
},
{
"epoch": 0.8,
"learning_rate": 2.13e-05,
"loss": 4.1508,
"step": 357
},
{
"epoch": 0.8,
"learning_rate": 2.136e-05,
"loss": 4.2011,
"step": 358
},
{
"epoch": 0.8,
"learning_rate": 2.1419999999999998e-05,
"loss": 4.0786,
"step": 359
},
{
"epoch": 0.81,
"learning_rate": 2.148e-05,
"loss": 4.1572,
"step": 360
},
{
"epoch": 0.81,
"learning_rate": 2.154e-05,
"loss": 4.0844,
"step": 361
},
{
"epoch": 0.81,
"learning_rate": 2.16e-05,
"loss": 4.1476,
"step": 362
},
{
"epoch": 0.81,
"learning_rate": 2.166e-05,
"loss": 4.0828,
"step": 363
},
{
"epoch": 0.82,
"learning_rate": 2.172e-05,
"loss": 4.1726,
"step": 364
},
{
"epoch": 0.82,
"learning_rate": 2.178e-05,
"loss": 4.2892,
"step": 365
},
{
"epoch": 0.82,
"learning_rate": 2.184e-05,
"loss": 4.0957,
"step": 366
},
{
"epoch": 0.82,
"learning_rate": 2.19e-05,
"loss": 4.1822,
"step": 367
},
{
"epoch": 0.83,
"learning_rate": 2.196e-05,
"loss": 4.1647,
"step": 368
},
{
"epoch": 0.83,
"learning_rate": 2.202e-05,
"loss": 4.0817,
"step": 369
},
{
"epoch": 0.83,
"learning_rate": 2.208e-05,
"loss": 4.0894,
"step": 370
},
{
"epoch": 0.83,
"learning_rate": 2.214e-05,
"loss": 4.079,
"step": 371
},
{
"epoch": 0.83,
"learning_rate": 2.22e-05,
"loss": 4.1652,
"step": 372
},
{
"epoch": 0.84,
"learning_rate": 2.226e-05,
"loss": 4.1596,
"step": 373
},
{
"epoch": 0.84,
"learning_rate": 2.232e-05,
"loss": 3.9787,
"step": 374
},
{
"epoch": 0.84,
"learning_rate": 2.238e-05,
"loss": 4.1849,
"step": 375
},
{
"epoch": 0.84,
"learning_rate": 2.2440000000000002e-05,
"loss": 4.0419,
"step": 376
},
{
"epoch": 0.85,
"learning_rate": 2.25e-05,
"loss": 4.145,
"step": 377
},
{
"epoch": 0.85,
"learning_rate": 2.256e-05,
"loss": 4.1548,
"step": 378
},
{
"epoch": 0.85,
"learning_rate": 2.262e-05,
"loss": 4.0737,
"step": 379
},
{
"epoch": 0.85,
"learning_rate": 2.268e-05,
"loss": 4.096,
"step": 380
},
{
"epoch": 0.85,
"learning_rate": 2.274e-05,
"loss": 4.1361,
"step": 381
},
{
"epoch": 0.86,
"learning_rate": 2.2800000000000002e-05,
"loss": 4.2105,
"step": 382
},
{
"epoch": 0.86,
"learning_rate": 2.286e-05,
"loss": 4.1431,
"step": 383
},
{
"epoch": 0.86,
"learning_rate": 2.292e-05,
"loss": 4.0593,
"step": 384
},
{
"epoch": 0.86,
"learning_rate": 2.298e-05,
"loss": 4.16,
"step": 385
},
{
"epoch": 0.87,
"learning_rate": 2.304e-05,
"loss": 4.0336,
"step": 386
},
{
"epoch": 0.87,
"learning_rate": 2.3100000000000002e-05,
"loss": 4.2153,
"step": 387
},
{
"epoch": 0.87,
"learning_rate": 2.3160000000000002e-05,
"loss": 4.2426,
"step": 388
},
{
"epoch": 0.87,
"learning_rate": 2.322e-05,
"loss": 4.1324,
"step": 389
},
{
"epoch": 0.87,
"learning_rate": 2.328e-05,
"loss": 4.126,
"step": 390
},
{
"epoch": 0.88,
"learning_rate": 2.334e-05,
"loss": 4.2552,
"step": 391
},
{
"epoch": 0.88,
"learning_rate": 2.3400000000000003e-05,
"loss": 4.0773,
"step": 392
},
{
"epoch": 0.88,
"learning_rate": 2.3460000000000002e-05,
"loss": 4.0848,
"step": 393
},
{
"epoch": 0.88,
"learning_rate": 2.3520000000000002e-05,
"loss": 4.1795,
"step": 394
},
{
"epoch": 0.89,
"learning_rate": 2.358e-05,
"loss": 4.3042,
"step": 395
},
{
"epoch": 0.89,
"learning_rate": 2.364e-05,
"loss": 4.2524,
"step": 396
},
{
"epoch": 0.89,
"learning_rate": 2.37e-05,
"loss": 4.4209,
"step": 397
},
{
"epoch": 0.89,
"learning_rate": 2.3760000000000003e-05,
"loss": 4.3868,
"step": 398
},
{
"epoch": 0.89,
"learning_rate": 2.3820000000000002e-05,
"loss": 4.5116,
"step": 399
},
{
"epoch": 0.9,
"learning_rate": 2.3880000000000002e-05,
"loss": 4.4561,
"step": 400
},
{
"epoch": 0.9,
"learning_rate": 2.394e-05,
"loss": 4.1715,
"step": 401
},
{
"epoch": 0.9,
"learning_rate": 2.4e-05,
"loss": 4.018,
"step": 402
},
{
"epoch": 0.9,
"learning_rate": 2.4060000000000003e-05,
"loss": 4.0975,
"step": 403
},
{
"epoch": 0.91,
"learning_rate": 2.4120000000000003e-05,
"loss": 4.0795,
"step": 404
},
{
"epoch": 0.91,
"learning_rate": 2.4180000000000002e-05,
"loss": 4.0454,
"step": 405
},
{
"epoch": 0.91,
"learning_rate": 2.4240000000000002e-05,
"loss": 4.1319,
"step": 406
},
{
"epoch": 0.91,
"learning_rate": 2.43e-05,
"loss": 4.1041,
"step": 407
},
{
"epoch": 0.91,
"learning_rate": 2.4360000000000004e-05,
"loss": 4.1619,
"step": 408
},
{
"epoch": 0.92,
"learning_rate": 2.442e-05,
"loss": 4.1178,
"step": 409
},
{
"epoch": 0.92,
"learning_rate": 2.448e-05,
"loss": 4.0931,
"step": 410
},
{
"epoch": 0.92,
"learning_rate": 2.454e-05,
"loss": 4.314,
"step": 411
},
{
"epoch": 0.92,
"learning_rate": 2.4599999999999998e-05,
"loss": 4.0508,
"step": 412
},
{
"epoch": 0.93,
"learning_rate": 2.4659999999999998e-05,
"loss": 4.003,
"step": 413
},
{
"epoch": 0.93,
"learning_rate": 2.472e-05,
"loss": 4.1375,
"step": 414
},
{
"epoch": 0.93,
"learning_rate": 2.478e-05,
"loss": 4.2206,
"step": 415
},
{
"epoch": 0.93,
"learning_rate": 2.484e-05,
"loss": 4.0785,
"step": 416
},
{
"epoch": 0.93,
"learning_rate": 2.49e-05,
"loss": 4.1078,
"step": 417
},
{
"epoch": 0.94,
"learning_rate": 2.4959999999999998e-05,
"loss": 4.0014,
"step": 418
},
{
"epoch": 0.94,
"learning_rate": 2.502e-05,
"loss": 4.0941,
"step": 419
},
{
"epoch": 0.94,
"learning_rate": 2.508e-05,
"loss": 4.0181,
"step": 420
},
{
"epoch": 0.94,
"learning_rate": 2.514e-05,
"loss": 4.1846,
"step": 421
},
{
"epoch": 0.95,
"learning_rate": 2.52e-05,
"loss": 4.0393,
"step": 422
},
{
"epoch": 0.95,
"learning_rate": 2.526e-05,
"loss": 4.1369,
"step": 423
},
{
"epoch": 0.95,
"learning_rate": 2.5319999999999998e-05,
"loss": 4.1978,
"step": 424
},
{
"epoch": 0.95,
"learning_rate": 2.538e-05,
"loss": 4.1363,
"step": 425
},
{
"epoch": 0.96,
"learning_rate": 2.544e-05,
"loss": 4.1793,
"step": 426
},
{
"epoch": 0.96,
"learning_rate": 2.55e-05,
"loss": 4.1114,
"step": 427
},
{
"epoch": 0.96,
"learning_rate": 2.556e-05,
"loss": 4.1015,
"step": 428
},
{
"epoch": 0.96,
"learning_rate": 2.562e-05,
"loss": 4.0305,
"step": 429
},
{
"epoch": 0.96,
"learning_rate": 2.568e-05,
"loss": 4.1179,
"step": 430
},
{
"epoch": 0.97,
"learning_rate": 2.574e-05,
"loss": 4.1957,
"step": 431
},
{
"epoch": 0.97,
"learning_rate": 2.58e-05,
"loss": 4.3161,
"step": 432
},
{
"epoch": 0.97,
"learning_rate": 2.586e-05,
"loss": 4.1238,
"step": 433
},
{
"epoch": 0.97,
"learning_rate": 2.592e-05,
"loss": 4.1152,
"step": 434
},
{
"epoch": 0.98,
"learning_rate": 2.5980000000000002e-05,
"loss": 4.1593,
"step": 435
},
{
"epoch": 0.98,
"learning_rate": 2.604e-05,
"loss": 4.0797,
"step": 436
},
{
"epoch": 0.98,
"learning_rate": 2.61e-05,
"loss": 4.0625,
"step": 437
},
{
"epoch": 0.98,
"learning_rate": 2.616e-05,
"loss": 4.2553,
"step": 438
},
{
"epoch": 0.98,
"learning_rate": 2.622e-05,
"loss": 4.1998,
"step": 439
},
{
"epoch": 0.99,
"learning_rate": 2.628e-05,
"loss": 4.2637,
"step": 440
},
{
"epoch": 0.99,
"learning_rate": 2.6340000000000002e-05,
"loss": 4.2109,
"step": 441
},
{
"epoch": 0.99,
"learning_rate": 2.64e-05,
"loss": 4.4059,
"step": 442
},
{
"epoch": 0.99,
"learning_rate": 2.646e-05,
"loss": 4.3216,
"step": 443
},
{
"epoch": 1.0,
"learning_rate": 2.652e-05,
"loss": 4.2793,
"step": 444
},
{
"epoch": 1.0,
"learning_rate": 2.658e-05,
"loss": 4.5485,
"step": 445
},
{
"epoch": 1.0,
"learning_rate": 2.6640000000000002e-05,
"loss": 4.2807,
"step": 446
},
{
"epoch": 1.0,
"step": 446,
"total_flos": 0.0,
"train_loss": 4.298295111399595,
"train_runtime": 5163.9194,
"train_samples_per_second": 5.526,
"train_steps_per_second": 0.086
}
],
"max_steps": 446,
"num_train_epochs": 1,
"total_flos": 0.0,
"trial_name": null,
"trial_params": null
}