wav2vec2-gpt2-wandb-grid-search / trainer_state.json
sanchit-gandhi's picture
End of training
3350b71
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 0.9987389659520807,
"global_step": 297,
"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": 4.767,
"step": 1
},
{
"epoch": 0.01,
"learning_rate": 1.2000000000000002e-07,
"loss": 4.8415,
"step": 2
},
{
"epoch": 0.01,
"learning_rate": 1.8e-07,
"loss": 4.9166,
"step": 3
},
{
"epoch": 0.01,
"learning_rate": 2.4000000000000003e-07,
"loss": 4.8629,
"step": 4
},
{
"epoch": 0.02,
"learning_rate": 3.0000000000000004e-07,
"loss": 4.7452,
"step": 5
},
{
"epoch": 0.02,
"learning_rate": 3.0000000000000004e-07,
"loss": 4.8808,
"step": 6
},
{
"epoch": 0.02,
"learning_rate": 3.6e-07,
"loss": 4.8603,
"step": 7
},
{
"epoch": 0.03,
"learning_rate": 4.2e-07,
"loss": 4.9323,
"step": 8
},
{
"epoch": 0.03,
"learning_rate": 4.800000000000001e-07,
"loss": 4.8445,
"step": 9
},
{
"epoch": 0.03,
"learning_rate": 5.4e-07,
"loss": 4.7068,
"step": 10
},
{
"epoch": 0.04,
"learning_rate": 6.000000000000001e-07,
"loss": 4.8473,
"step": 11
},
{
"epoch": 0.04,
"learning_rate": 6.599999999999999e-07,
"loss": 4.8768,
"step": 12
},
{
"epoch": 0.04,
"learning_rate": 7.2e-07,
"loss": 4.7908,
"step": 13
},
{
"epoch": 0.05,
"learning_rate": 7.799999999999999e-07,
"loss": 4.7452,
"step": 14
},
{
"epoch": 0.05,
"learning_rate": 8.4e-07,
"loss": 4.7585,
"step": 15
},
{
"epoch": 0.05,
"learning_rate": 9e-07,
"loss": 4.6847,
"step": 16
},
{
"epoch": 0.06,
"learning_rate": 9.600000000000001e-07,
"loss": 4.7876,
"step": 17
},
{
"epoch": 0.06,
"learning_rate": 1.0200000000000002e-06,
"loss": 4.7936,
"step": 18
},
{
"epoch": 0.06,
"learning_rate": 1.08e-06,
"loss": 4.7411,
"step": 19
},
{
"epoch": 0.07,
"learning_rate": 1.14e-06,
"loss": 4.8271,
"step": 20
},
{
"epoch": 0.07,
"learning_rate": 1.2000000000000002e-06,
"loss": 4.828,
"step": 21
},
{
"epoch": 0.07,
"learning_rate": 1.26e-06,
"loss": 4.7725,
"step": 22
},
{
"epoch": 0.08,
"learning_rate": 1.3199999999999999e-06,
"loss": 4.6722,
"step": 23
},
{
"epoch": 0.08,
"learning_rate": 1.38e-06,
"loss": 4.8135,
"step": 24
},
{
"epoch": 0.08,
"learning_rate": 1.44e-06,
"loss": 4.6361,
"step": 25
},
{
"epoch": 0.09,
"learning_rate": 1.5e-06,
"loss": 4.9185,
"step": 26
},
{
"epoch": 0.09,
"learning_rate": 1.5599999999999999e-06,
"loss": 4.654,
"step": 27
},
{
"epoch": 0.09,
"learning_rate": 1.62e-06,
"loss": 4.6688,
"step": 28
},
{
"epoch": 0.1,
"learning_rate": 1.68e-06,
"loss": 4.5584,
"step": 29
},
{
"epoch": 0.1,
"learning_rate": 1.74e-06,
"loss": 4.7119,
"step": 30
},
{
"epoch": 0.1,
"learning_rate": 1.8e-06,
"loss": 4.6957,
"step": 31
},
{
"epoch": 0.11,
"learning_rate": 1.86e-06,
"loss": 4.6116,
"step": 32
},
{
"epoch": 0.11,
"learning_rate": 1.9200000000000003e-06,
"loss": 4.6472,
"step": 33
},
{
"epoch": 0.11,
"learning_rate": 1.98e-06,
"loss": 4.5855,
"step": 34
},
{
"epoch": 0.12,
"learning_rate": 2.0400000000000004e-06,
"loss": 4.6201,
"step": 35
},
{
"epoch": 0.12,
"learning_rate": 2.1000000000000002e-06,
"loss": 4.5679,
"step": 36
},
{
"epoch": 0.12,
"learning_rate": 2.16e-06,
"loss": 4.6958,
"step": 37
},
{
"epoch": 0.13,
"learning_rate": 2.22e-06,
"loss": 4.6541,
"step": 38
},
{
"epoch": 0.13,
"learning_rate": 2.28e-06,
"loss": 4.7043,
"step": 39
},
{
"epoch": 0.13,
"learning_rate": 2.34e-06,
"loss": 4.5367,
"step": 40
},
{
"epoch": 0.14,
"learning_rate": 2.4000000000000003e-06,
"loss": 4.6178,
"step": 41
},
{
"epoch": 0.14,
"learning_rate": 2.46e-06,
"loss": 4.6936,
"step": 42
},
{
"epoch": 0.14,
"learning_rate": 2.52e-06,
"loss": 4.4532,
"step": 43
},
{
"epoch": 0.15,
"learning_rate": 2.58e-06,
"loss": 4.7663,
"step": 44
},
{
"epoch": 0.15,
"learning_rate": 2.6399999999999997e-06,
"loss": 4.7255,
"step": 45
},
{
"epoch": 0.15,
"learning_rate": 2.7e-06,
"loss": 4.7645,
"step": 46
},
{
"epoch": 0.16,
"learning_rate": 2.76e-06,
"loss": 4.7834,
"step": 47
},
{
"epoch": 0.16,
"learning_rate": 2.82e-06,
"loss": 4.8407,
"step": 48
},
{
"epoch": 0.16,
"learning_rate": 2.88e-06,
"loss": 4.9009,
"step": 49
},
{
"epoch": 0.17,
"learning_rate": 2.9400000000000002e-06,
"loss": 4.706,
"step": 50
},
{
"epoch": 0.17,
"learning_rate": 3e-06,
"loss": 4.4758,
"step": 51
},
{
"epoch": 0.17,
"learning_rate": 3.06e-06,
"loss": 4.3677,
"step": 52
},
{
"epoch": 0.18,
"learning_rate": 3.1199999999999998e-06,
"loss": 4.3162,
"step": 53
},
{
"epoch": 0.18,
"learning_rate": 3.18e-06,
"loss": 4.3327,
"step": 54
},
{
"epoch": 0.18,
"learning_rate": 3.24e-06,
"loss": 4.269,
"step": 55
},
{
"epoch": 0.19,
"learning_rate": 3.3e-06,
"loss": 4.2999,
"step": 56
},
{
"epoch": 0.19,
"learning_rate": 3.36e-06,
"loss": 4.4273,
"step": 57
},
{
"epoch": 0.2,
"learning_rate": 3.4200000000000003e-06,
"loss": 4.4309,
"step": 58
},
{
"epoch": 0.2,
"learning_rate": 3.48e-06,
"loss": 4.3935,
"step": 59
},
{
"epoch": 0.2,
"learning_rate": 3.54e-06,
"loss": 4.3246,
"step": 60
},
{
"epoch": 0.21,
"learning_rate": 3.6e-06,
"loss": 4.3995,
"step": 61
},
{
"epoch": 0.21,
"learning_rate": 3.66e-06,
"loss": 4.2675,
"step": 62
},
{
"epoch": 0.21,
"learning_rate": 3.72e-06,
"loss": 4.4125,
"step": 63
},
{
"epoch": 0.22,
"learning_rate": 3.7800000000000002e-06,
"loss": 4.3521,
"step": 64
},
{
"epoch": 0.22,
"learning_rate": 3.8400000000000005e-06,
"loss": 4.4045,
"step": 65
},
{
"epoch": 0.22,
"learning_rate": 3.9e-06,
"loss": 4.406,
"step": 66
},
{
"epoch": 0.23,
"learning_rate": 3.96e-06,
"loss": 4.3524,
"step": 67
},
{
"epoch": 0.23,
"learning_rate": 4.0200000000000005e-06,
"loss": 4.3942,
"step": 68
},
{
"epoch": 0.23,
"learning_rate": 4.080000000000001e-06,
"loss": 4.3343,
"step": 69
},
{
"epoch": 0.24,
"learning_rate": 4.14e-06,
"loss": 4.266,
"step": 70
},
{
"epoch": 0.24,
"learning_rate": 4.2000000000000004e-06,
"loss": 4.2738,
"step": 71
},
{
"epoch": 0.24,
"learning_rate": 4.26e-06,
"loss": 4.3974,
"step": 72
},
{
"epoch": 0.25,
"learning_rate": 4.32e-06,
"loss": 4.3188,
"step": 73
},
{
"epoch": 0.25,
"learning_rate": 4.3799999999999996e-06,
"loss": 4.3084,
"step": 74
},
{
"epoch": 0.25,
"learning_rate": 4.44e-06,
"loss": 4.4146,
"step": 75
},
{
"epoch": 0.26,
"learning_rate": 4.5e-06,
"loss": 4.2865,
"step": 76
},
{
"epoch": 0.26,
"learning_rate": 4.56e-06,
"loss": 4.2502,
"step": 77
},
{
"epoch": 0.26,
"learning_rate": 4.62e-06,
"loss": 4.3103,
"step": 78
},
{
"epoch": 0.27,
"learning_rate": 4.68e-06,
"loss": 4.1937,
"step": 79
},
{
"epoch": 0.27,
"learning_rate": 4.74e-06,
"loss": 4.3765,
"step": 80
},
{
"epoch": 0.27,
"learning_rate": 4.800000000000001e-06,
"loss": 4.3016,
"step": 81
},
{
"epoch": 0.28,
"learning_rate": 4.86e-06,
"loss": 4.3088,
"step": 82
},
{
"epoch": 0.28,
"learning_rate": 4.92e-06,
"loss": 4.2784,
"step": 83
},
{
"epoch": 0.28,
"learning_rate": 4.980000000000001e-06,
"loss": 4.3069,
"step": 84
},
{
"epoch": 0.29,
"learning_rate": 5.04e-06,
"loss": 4.4127,
"step": 85
},
{
"epoch": 0.29,
"learning_rate": 5.1e-06,
"loss": 4.2931,
"step": 86
},
{
"epoch": 0.29,
"learning_rate": 5.16e-06,
"loss": 4.3402,
"step": 87
},
{
"epoch": 0.3,
"learning_rate": 5.22e-06,
"loss": 4.3675,
"step": 88
},
{
"epoch": 0.3,
"learning_rate": 5.279999999999999e-06,
"loss": 4.3851,
"step": 89
},
{
"epoch": 0.3,
"learning_rate": 5.34e-06,
"loss": 4.4114,
"step": 90
},
{
"epoch": 0.31,
"learning_rate": 5.4e-06,
"loss": 4.4333,
"step": 91
},
{
"epoch": 0.31,
"learning_rate": 5.46e-06,
"loss": 4.4915,
"step": 92
},
{
"epoch": 0.31,
"learning_rate": 5.52e-06,
"loss": 4.4213,
"step": 93
},
{
"epoch": 0.32,
"learning_rate": 5.58e-06,
"loss": 4.3693,
"step": 94
},
{
"epoch": 0.32,
"learning_rate": 5.64e-06,
"loss": 4.3271,
"step": 95
},
{
"epoch": 0.32,
"learning_rate": 5.7000000000000005e-06,
"loss": 4.4958,
"step": 96
},
{
"epoch": 0.33,
"learning_rate": 5.76e-06,
"loss": 4.6801,
"step": 97
},
{
"epoch": 0.33,
"learning_rate": 5.82e-06,
"loss": 4.6378,
"step": 98
},
{
"epoch": 0.33,
"learning_rate": 5.8800000000000005e-06,
"loss": 4.6153,
"step": 99
},
{
"epoch": 0.34,
"learning_rate": 5.940000000000001e-06,
"loss": 4.6683,
"step": 100
},
{
"epoch": 0.34,
"learning_rate": 6e-06,
"loss": 4.2246,
"step": 101
},
{
"epoch": 0.34,
"learning_rate": 6.0600000000000004e-06,
"loss": 4.2737,
"step": 102
},
{
"epoch": 0.35,
"learning_rate": 6.12e-06,
"loss": 4.2711,
"step": 103
},
{
"epoch": 0.35,
"learning_rate": 6.18e-06,
"loss": 4.1781,
"step": 104
},
{
"epoch": 0.35,
"learning_rate": 6.2399999999999995e-06,
"loss": 4.2846,
"step": 105
},
{
"epoch": 0.36,
"learning_rate": 6.3e-06,
"loss": 4.2148,
"step": 106
},
{
"epoch": 0.36,
"learning_rate": 6.36e-06,
"loss": 4.1611,
"step": 107
},
{
"epoch": 0.36,
"learning_rate": 6.42e-06,
"loss": 4.2767,
"step": 108
},
{
"epoch": 0.37,
"learning_rate": 6.48e-06,
"loss": 4.2806,
"step": 109
},
{
"epoch": 0.37,
"learning_rate": 6.54e-06,
"loss": 4.2561,
"step": 110
},
{
"epoch": 0.37,
"learning_rate": 6.6e-06,
"loss": 4.3238,
"step": 111
},
{
"epoch": 0.38,
"learning_rate": 6.660000000000001e-06,
"loss": 4.2542,
"step": 112
},
{
"epoch": 0.38,
"learning_rate": 6.72e-06,
"loss": 4.1891,
"step": 113
},
{
"epoch": 0.38,
"learning_rate": 6.78e-06,
"loss": 4.2587,
"step": 114
},
{
"epoch": 0.39,
"learning_rate": 6.840000000000001e-06,
"loss": 4.2102,
"step": 115
},
{
"epoch": 0.39,
"learning_rate": 6.900000000000001e-06,
"loss": 4.2241,
"step": 116
},
{
"epoch": 0.39,
"learning_rate": 6.96e-06,
"loss": 4.1748,
"step": 117
},
{
"epoch": 0.4,
"learning_rate": 7.0200000000000006e-06,
"loss": 4.2224,
"step": 118
},
{
"epoch": 0.4,
"learning_rate": 7.08e-06,
"loss": 4.2624,
"step": 119
},
{
"epoch": 0.4,
"learning_rate": 7.14e-06,
"loss": 4.2724,
"step": 120
},
{
"epoch": 0.41,
"learning_rate": 7.2e-06,
"loss": 4.2884,
"step": 121
},
{
"epoch": 0.41,
"learning_rate": 7.26e-06,
"loss": 4.3434,
"step": 122
},
{
"epoch": 0.41,
"learning_rate": 7.32e-06,
"loss": 4.1558,
"step": 123
},
{
"epoch": 0.42,
"learning_rate": 7.3800000000000005e-06,
"loss": 4.2972,
"step": 124
},
{
"epoch": 0.42,
"learning_rate": 7.44e-06,
"loss": 4.3484,
"step": 125
},
{
"epoch": 0.42,
"learning_rate": 7.5e-06,
"loss": 4.2669,
"step": 126
},
{
"epoch": 0.43,
"learning_rate": 7.5600000000000005e-06,
"loss": 4.3126,
"step": 127
},
{
"epoch": 0.43,
"learning_rate": 7.62e-06,
"loss": 4.168,
"step": 128
},
{
"epoch": 0.43,
"learning_rate": 7.680000000000001e-06,
"loss": 4.3161,
"step": 129
},
{
"epoch": 0.44,
"learning_rate": 7.74e-06,
"loss": 4.4017,
"step": 130
},
{
"epoch": 0.44,
"learning_rate": 7.8e-06,
"loss": 4.3215,
"step": 131
},
{
"epoch": 0.44,
"learning_rate": 7.860000000000001e-06,
"loss": 4.1507,
"step": 132
},
{
"epoch": 0.45,
"learning_rate": 7.92e-06,
"loss": 4.3154,
"step": 133
},
{
"epoch": 0.45,
"learning_rate": 7.98e-06,
"loss": 4.1907,
"step": 134
},
{
"epoch": 0.45,
"learning_rate": 8.040000000000001e-06,
"loss": 4.3897,
"step": 135
},
{
"epoch": 0.46,
"learning_rate": 8.1e-06,
"loss": 4.2992,
"step": 136
},
{
"epoch": 0.46,
"learning_rate": 8.160000000000001e-06,
"loss": 4.2309,
"step": 137
},
{
"epoch": 0.46,
"learning_rate": 8.220000000000001e-06,
"loss": 4.1562,
"step": 138
},
{
"epoch": 0.47,
"learning_rate": 8.28e-06,
"loss": 4.1999,
"step": 139
},
{
"epoch": 0.47,
"learning_rate": 8.340000000000001e-06,
"loss": 4.2273,
"step": 140
},
{
"epoch": 0.47,
"learning_rate": 8.400000000000001e-06,
"loss": 4.4316,
"step": 141
},
{
"epoch": 0.48,
"learning_rate": 8.459999999999999e-06,
"loss": 4.197,
"step": 142
},
{
"epoch": 0.48,
"learning_rate": 8.52e-06,
"loss": 4.3164,
"step": 143
},
{
"epoch": 0.48,
"learning_rate": 8.58e-06,
"loss": 4.3355,
"step": 144
},
{
"epoch": 0.49,
"learning_rate": 8.64e-06,
"loss": 4.4157,
"step": 145
},
{
"epoch": 0.49,
"learning_rate": 8.7e-06,
"loss": 4.2614,
"step": 146
},
{
"epoch": 0.49,
"learning_rate": 8.759999999999999e-06,
"loss": 4.3818,
"step": 147
},
{
"epoch": 0.5,
"learning_rate": 8.82e-06,
"loss": 4.5368,
"step": 148
},
{
"epoch": 0.5,
"learning_rate": 8.88e-06,
"loss": 4.534,
"step": 149
},
{
"epoch": 0.5,
"learning_rate": 8.939999999999999e-06,
"loss": 4.642,
"step": 150
},
{
"epoch": 0.51,
"learning_rate": 9e-06,
"loss": 4.26,
"step": 151
},
{
"epoch": 0.51,
"learning_rate": 9.06e-06,
"loss": 4.2318,
"step": 152
},
{
"epoch": 0.51,
"learning_rate": 9.12e-06,
"loss": 4.2651,
"step": 153
},
{
"epoch": 0.52,
"learning_rate": 9.18e-06,
"loss": 4.1934,
"step": 154
},
{
"epoch": 0.52,
"learning_rate": 9.24e-06,
"loss": 4.2266,
"step": 155
},
{
"epoch": 0.52,
"learning_rate": 9.3e-06,
"loss": 4.2535,
"step": 156
},
{
"epoch": 0.53,
"learning_rate": 9.36e-06,
"loss": 4.1637,
"step": 157
},
{
"epoch": 0.53,
"learning_rate": 9.42e-06,
"loss": 4.2322,
"step": 158
},
{
"epoch": 0.53,
"learning_rate": 9.48e-06,
"loss": 4.2244,
"step": 159
},
{
"epoch": 0.54,
"learning_rate": 9.54e-06,
"loss": 4.2042,
"step": 160
},
{
"epoch": 0.54,
"learning_rate": 9.600000000000001e-06,
"loss": 4.2152,
"step": 161
},
{
"epoch": 0.54,
"learning_rate": 9.66e-06,
"loss": 4.1874,
"step": 162
},
{
"epoch": 0.55,
"learning_rate": 9.72e-06,
"loss": 4.2569,
"step": 163
},
{
"epoch": 0.55,
"learning_rate": 9.780000000000001e-06,
"loss": 4.1769,
"step": 164
},
{
"epoch": 0.55,
"learning_rate": 9.84e-06,
"loss": 4.2435,
"step": 165
},
{
"epoch": 0.56,
"learning_rate": 9.9e-06,
"loss": 4.206,
"step": 166
},
{
"epoch": 0.56,
"learning_rate": 9.960000000000001e-06,
"loss": 4.223,
"step": 167
},
{
"epoch": 0.56,
"learning_rate": 1.002e-05,
"loss": 4.2185,
"step": 168
},
{
"epoch": 0.57,
"learning_rate": 1.008e-05,
"loss": 4.1909,
"step": 169
},
{
"epoch": 0.57,
"learning_rate": 1.0140000000000001e-05,
"loss": 4.1373,
"step": 170
},
{
"epoch": 0.58,
"learning_rate": 1.02e-05,
"loss": 4.1232,
"step": 171
},
{
"epoch": 0.58,
"learning_rate": 1.0260000000000002e-05,
"loss": 4.2279,
"step": 172
},
{
"epoch": 0.58,
"learning_rate": 1.032e-05,
"loss": 4.0904,
"step": 173
},
{
"epoch": 0.59,
"learning_rate": 1.0379999999999999e-05,
"loss": 4.1484,
"step": 174
},
{
"epoch": 0.59,
"learning_rate": 1.044e-05,
"loss": 4.296,
"step": 175
},
{
"epoch": 0.59,
"learning_rate": 1.05e-05,
"loss": 4.1508,
"step": 176
},
{
"epoch": 0.6,
"learning_rate": 1.0559999999999999e-05,
"loss": 4.2833,
"step": 177
},
{
"epoch": 0.6,
"learning_rate": 1.062e-05,
"loss": 4.1359,
"step": 178
},
{
"epoch": 0.6,
"learning_rate": 1.068e-05,
"loss": 4.1837,
"step": 179
},
{
"epoch": 0.61,
"learning_rate": 1.074e-05,
"loss": 4.3284,
"step": 180
},
{
"epoch": 0.61,
"learning_rate": 1.08e-05,
"loss": 4.196,
"step": 181
},
{
"epoch": 0.61,
"learning_rate": 1.086e-05,
"loss": 4.1404,
"step": 182
},
{
"epoch": 0.62,
"learning_rate": 1.092e-05,
"loss": 4.1248,
"step": 183
},
{
"epoch": 0.62,
"learning_rate": 1.098e-05,
"loss": 4.2255,
"step": 184
},
{
"epoch": 0.62,
"learning_rate": 1.104e-05,
"loss": 4.2625,
"step": 185
},
{
"epoch": 0.63,
"learning_rate": 1.11e-05,
"loss": 4.2148,
"step": 186
},
{
"epoch": 0.63,
"learning_rate": 1.116e-05,
"loss": 4.2413,
"step": 187
},
{
"epoch": 0.63,
"learning_rate": 1.1220000000000001e-05,
"loss": 4.2147,
"step": 188
},
{
"epoch": 0.64,
"learning_rate": 1.128e-05,
"loss": 4.1846,
"step": 189
},
{
"epoch": 0.64,
"learning_rate": 1.134e-05,
"loss": 4.2799,
"step": 190
},
{
"epoch": 0.64,
"learning_rate": 1.1400000000000001e-05,
"loss": 4.3126,
"step": 191
},
{
"epoch": 0.65,
"learning_rate": 1.146e-05,
"loss": 4.3646,
"step": 192
},
{
"epoch": 0.65,
"learning_rate": 1.152e-05,
"loss": 4.4346,
"step": 193
},
{
"epoch": 0.65,
"learning_rate": 1.1580000000000001e-05,
"loss": 4.3561,
"step": 194
},
{
"epoch": 0.66,
"learning_rate": 1.164e-05,
"loss": 4.3478,
"step": 195
},
{
"epoch": 0.66,
"learning_rate": 1.1700000000000001e-05,
"loss": 4.3004,
"step": 196
},
{
"epoch": 0.66,
"learning_rate": 1.1760000000000001e-05,
"loss": 4.1971,
"step": 197
},
{
"epoch": 0.67,
"learning_rate": 1.182e-05,
"loss": 4.6767,
"step": 198
},
{
"epoch": 0.67,
"learning_rate": 1.1880000000000001e-05,
"loss": 4.4133,
"step": 199
},
{
"epoch": 0.67,
"learning_rate": 1.1940000000000001e-05,
"loss": 4.6429,
"step": 200
},
{
"epoch": 0.68,
"learning_rate": 1.2e-05,
"loss": 4.2574,
"step": 201
},
{
"epoch": 0.68,
"learning_rate": 1.2060000000000001e-05,
"loss": 4.1977,
"step": 202
},
{
"epoch": 0.68,
"learning_rate": 1.2120000000000001e-05,
"loss": 4.1876,
"step": 203
},
{
"epoch": 0.69,
"learning_rate": 1.2180000000000002e-05,
"loss": 4.1156,
"step": 204
},
{
"epoch": 0.69,
"learning_rate": 1.224e-05,
"loss": 4.1713,
"step": 205
},
{
"epoch": 0.69,
"learning_rate": 1.2299999999999999e-05,
"loss": 4.1696,
"step": 206
},
{
"epoch": 0.7,
"learning_rate": 1.236e-05,
"loss": 4.2188,
"step": 207
},
{
"epoch": 0.7,
"learning_rate": 1.242e-05,
"loss": 4.1895,
"step": 208
},
{
"epoch": 0.7,
"learning_rate": 1.2479999999999999e-05,
"loss": 4.1555,
"step": 209
},
{
"epoch": 0.71,
"learning_rate": 1.254e-05,
"loss": 4.2614,
"step": 210
},
{
"epoch": 0.71,
"learning_rate": 1.26e-05,
"loss": 4.0539,
"step": 211
},
{
"epoch": 0.71,
"learning_rate": 1.2659999999999999e-05,
"loss": 4.1863,
"step": 212
},
{
"epoch": 0.72,
"learning_rate": 1.272e-05,
"loss": 4.2161,
"step": 213
},
{
"epoch": 0.72,
"learning_rate": 1.278e-05,
"loss": 4.2059,
"step": 214
},
{
"epoch": 0.72,
"learning_rate": 1.284e-05,
"loss": 4.2133,
"step": 215
},
{
"epoch": 0.73,
"learning_rate": 1.29e-05,
"loss": 4.1991,
"step": 216
},
{
"epoch": 0.73,
"learning_rate": 1.296e-05,
"loss": 4.1563,
"step": 217
},
{
"epoch": 0.73,
"learning_rate": 1.302e-05,
"loss": 4.1303,
"step": 218
},
{
"epoch": 0.74,
"learning_rate": 1.308e-05,
"loss": 4.1368,
"step": 219
},
{
"epoch": 0.74,
"learning_rate": 1.314e-05,
"loss": 4.1971,
"step": 220
},
{
"epoch": 0.74,
"learning_rate": 1.32e-05,
"loss": 4.1136,
"step": 221
},
{
"epoch": 0.75,
"learning_rate": 1.326e-05,
"loss": 4.1069,
"step": 222
},
{
"epoch": 0.75,
"learning_rate": 1.3320000000000001e-05,
"loss": 4.1314,
"step": 223
},
{
"epoch": 0.75,
"learning_rate": 1.338e-05,
"loss": 4.1782,
"step": 224
},
{
"epoch": 0.76,
"learning_rate": 1.344e-05,
"loss": 4.1584,
"step": 225
},
{
"epoch": 0.76,
"learning_rate": 1.3500000000000001e-05,
"loss": 4.1714,
"step": 226
},
{
"epoch": 0.76,
"learning_rate": 1.356e-05,
"loss": 4.1961,
"step": 227
},
{
"epoch": 0.77,
"learning_rate": 1.362e-05,
"loss": 4.1251,
"step": 228
},
{
"epoch": 0.77,
"learning_rate": 1.3680000000000001e-05,
"loss": 4.1515,
"step": 229
},
{
"epoch": 0.77,
"learning_rate": 1.374e-05,
"loss": 4.1599,
"step": 230
},
{
"epoch": 0.78,
"learning_rate": 1.3800000000000002e-05,
"loss": 4.2587,
"step": 231
},
{
"epoch": 0.78,
"learning_rate": 1.3860000000000001e-05,
"loss": 4.1317,
"step": 232
},
{
"epoch": 0.78,
"learning_rate": 1.392e-05,
"loss": 4.2434,
"step": 233
},
{
"epoch": 0.79,
"learning_rate": 1.3980000000000002e-05,
"loss": 4.2162,
"step": 234
},
{
"epoch": 0.79,
"learning_rate": 1.4040000000000001e-05,
"loss": 4.2583,
"step": 235
},
{
"epoch": 0.79,
"learning_rate": 1.4099999999999999e-05,
"loss": 4.0917,
"step": 236
},
{
"epoch": 0.8,
"learning_rate": 1.416e-05,
"loss": 4.1941,
"step": 237
},
{
"epoch": 0.8,
"learning_rate": 1.422e-05,
"loss": 4.3278,
"step": 238
},
{
"epoch": 0.8,
"learning_rate": 1.428e-05,
"loss": 4.1576,
"step": 239
},
{
"epoch": 0.81,
"learning_rate": 1.434e-05,
"loss": 4.2114,
"step": 240
},
{
"epoch": 0.81,
"learning_rate": 1.44e-05,
"loss": 4.3157,
"step": 241
},
{
"epoch": 0.81,
"learning_rate": 1.446e-05,
"loss": 4.2419,
"step": 242
},
{
"epoch": 0.82,
"learning_rate": 1.452e-05,
"loss": 4.1953,
"step": 243
},
{
"epoch": 0.82,
"learning_rate": 1.458e-05,
"loss": 4.2287,
"step": 244
},
{
"epoch": 0.82,
"learning_rate": 1.464e-05,
"loss": 4.2023,
"step": 245
},
{
"epoch": 0.83,
"learning_rate": 1.47e-05,
"loss": 4.3791,
"step": 246
},
{
"epoch": 0.83,
"learning_rate": 1.4760000000000001e-05,
"loss": 4.3207,
"step": 247
},
{
"epoch": 0.83,
"learning_rate": 1.482e-05,
"loss": 4.323,
"step": 248
},
{
"epoch": 0.84,
"learning_rate": 1.488e-05,
"loss": 4.4303,
"step": 249
},
{
"epoch": 0.84,
"learning_rate": 1.4940000000000001e-05,
"loss": 4.4593,
"step": 250
},
{
"epoch": 0.84,
"learning_rate": 1.5e-05,
"loss": 4.2375,
"step": 251
},
{
"epoch": 0.85,
"learning_rate": 1.506e-05,
"loss": 4.1519,
"step": 252
},
{
"epoch": 0.85,
"learning_rate": 1.5120000000000001e-05,
"loss": 4.115,
"step": 253
},
{
"epoch": 0.85,
"learning_rate": 1.518e-05,
"loss": 4.1146,
"step": 254
},
{
"epoch": 0.86,
"learning_rate": 1.524e-05,
"loss": 4.1397,
"step": 255
},
{
"epoch": 0.86,
"learning_rate": 1.53e-05,
"loss": 4.1209,
"step": 256
},
{
"epoch": 0.86,
"learning_rate": 1.5360000000000002e-05,
"loss": 4.2665,
"step": 257
},
{
"epoch": 0.87,
"learning_rate": 1.542e-05,
"loss": 4.1351,
"step": 258
},
{
"epoch": 0.87,
"learning_rate": 1.548e-05,
"loss": 4.1725,
"step": 259
},
{
"epoch": 0.87,
"learning_rate": 1.554e-05,
"loss": 4.0934,
"step": 260
},
{
"epoch": 0.88,
"learning_rate": 1.56e-05,
"loss": 4.1963,
"step": 261
},
{
"epoch": 0.88,
"learning_rate": 1.5660000000000003e-05,
"loss": 4.2262,
"step": 262
},
{
"epoch": 0.88,
"learning_rate": 1.5720000000000002e-05,
"loss": 4.0871,
"step": 263
},
{
"epoch": 0.89,
"learning_rate": 1.578e-05,
"loss": 4.1276,
"step": 264
},
{
"epoch": 0.89,
"learning_rate": 1.584e-05,
"loss": 4.137,
"step": 265
},
{
"epoch": 0.89,
"learning_rate": 1.59e-05,
"loss": 4.2361,
"step": 266
},
{
"epoch": 0.9,
"learning_rate": 1.596e-05,
"loss": 4.1436,
"step": 267
},
{
"epoch": 0.9,
"learning_rate": 1.6020000000000002e-05,
"loss": 4.1081,
"step": 268
},
{
"epoch": 0.9,
"learning_rate": 1.6080000000000002e-05,
"loss": 4.0701,
"step": 269
},
{
"epoch": 0.91,
"learning_rate": 1.614e-05,
"loss": 4.2026,
"step": 270
},
{
"epoch": 0.91,
"learning_rate": 1.62e-05,
"loss": 4.0701,
"step": 271
},
{
"epoch": 0.91,
"learning_rate": 1.626e-05,
"loss": 4.0502,
"step": 272
},
{
"epoch": 0.92,
"learning_rate": 1.6320000000000003e-05,
"loss": 4.1113,
"step": 273
},
{
"epoch": 0.92,
"learning_rate": 1.6380000000000002e-05,
"loss": 4.0909,
"step": 274
},
{
"epoch": 0.92,
"learning_rate": 1.6440000000000002e-05,
"loss": 4.2243,
"step": 275
},
{
"epoch": 0.93,
"learning_rate": 1.65e-05,
"loss": 4.1479,
"step": 276
},
{
"epoch": 0.93,
"learning_rate": 1.656e-05,
"loss": 4.1566,
"step": 277
},
{
"epoch": 0.93,
"learning_rate": 1.6620000000000004e-05,
"loss": 4.0795,
"step": 278
},
{
"epoch": 0.94,
"learning_rate": 1.6680000000000003e-05,
"loss": 4.1255,
"step": 279
},
{
"epoch": 0.94,
"learning_rate": 1.6740000000000002e-05,
"loss": 4.0884,
"step": 280
},
{
"epoch": 0.94,
"learning_rate": 1.6800000000000002e-05,
"loss": 4.1399,
"step": 281
},
{
"epoch": 0.95,
"learning_rate": 1.686e-05,
"loss": 4.1306,
"step": 282
},
{
"epoch": 0.95,
"learning_rate": 1.6919999999999997e-05,
"loss": 4.2975,
"step": 283
},
{
"epoch": 0.96,
"learning_rate": 1.698e-05,
"loss": 4.0595,
"step": 284
},
{
"epoch": 0.96,
"learning_rate": 1.704e-05,
"loss": 4.1804,
"step": 285
},
{
"epoch": 0.96,
"learning_rate": 1.71e-05,
"loss": 4.1973,
"step": 286
},
{
"epoch": 0.97,
"learning_rate": 1.716e-05,
"loss": 4.1023,
"step": 287
},
{
"epoch": 0.97,
"learning_rate": 1.7219999999999998e-05,
"loss": 4.1242,
"step": 288
},
{
"epoch": 0.97,
"learning_rate": 1.728e-05,
"loss": 4.1294,
"step": 289
},
{
"epoch": 0.98,
"learning_rate": 1.734e-05,
"loss": 4.2188,
"step": 290
},
{
"epoch": 0.98,
"learning_rate": 1.74e-05,
"loss": 4.2035,
"step": 291
},
{
"epoch": 0.98,
"learning_rate": 1.746e-05,
"loss": 4.2955,
"step": 292
},
{
"epoch": 0.99,
"learning_rate": 1.7519999999999998e-05,
"loss": 4.262,
"step": 293
},
{
"epoch": 0.99,
"learning_rate": 1.758e-05,
"loss": 4.4573,
"step": 294
},
{
"epoch": 0.99,
"learning_rate": 1.764e-05,
"loss": 4.227,
"step": 295
},
{
"epoch": 1.0,
"learning_rate": 1.77e-05,
"loss": 4.5969,
"step": 296
},
{
"epoch": 1.0,
"learning_rate": 1.776e-05,
"loss": 4.4712,
"step": 297
},
{
"epoch": 1.0,
"step": 297,
"total_flos": 0.0,
"train_loss": 4.3407110846805255,
"train_runtime": 4813.5869,
"train_samples_per_second": 5.929,
"train_steps_per_second": 0.062
}
],
"max_steps": 297,
"num_train_epochs": 1,
"total_flos": 0.0,
"trial_name": null,
"trial_params": null
}