wav2vec2-2-gpt2-grid-search / trainer_state.json
sanchit-gandhi's picture
End of training
c2c7a1c
raw history blame
No virus
33.6 kB
{
"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": 4.7628,
"step": 1
},
{
"epoch": 0.01,
"learning_rate": 1.2000000000000002e-07,
"loss": 4.8361,
"step": 2
},
{
"epoch": 0.01,
"learning_rate": 1.8e-07,
"loss": 4.9126,
"step": 3
},
{
"epoch": 0.01,
"learning_rate": 2.4000000000000003e-07,
"loss": 4.8568,
"step": 4
},
{
"epoch": 0.02,
"learning_rate": 3.0000000000000004e-07,
"loss": 4.7498,
"step": 5
},
{
"epoch": 0.02,
"learning_rate": 3.0000000000000004e-07,
"loss": 4.8731,
"step": 6
},
{
"epoch": 0.02,
"learning_rate": 3.6e-07,
"loss": 4.8567,
"step": 7
},
{
"epoch": 0.03,
"learning_rate": 4.2e-07,
"loss": 4.9357,
"step": 8
},
{
"epoch": 0.03,
"learning_rate": 4.800000000000001e-07,
"loss": 4.8471,
"step": 9
},
{
"epoch": 0.03,
"learning_rate": 5.4e-07,
"loss": 4.7009,
"step": 10
},
{
"epoch": 0.04,
"learning_rate": 6.000000000000001e-07,
"loss": 4.8436,
"step": 11
},
{
"epoch": 0.04,
"learning_rate": 6.599999999999999e-07,
"loss": 4.8838,
"step": 12
},
{
"epoch": 0.04,
"learning_rate": 7.2e-07,
"loss": 4.7957,
"step": 13
},
{
"epoch": 0.05,
"learning_rate": 7.799999999999999e-07,
"loss": 4.7472,
"step": 14
},
{
"epoch": 0.05,
"learning_rate": 8.4e-07,
"loss": 4.7638,
"step": 15
},
{
"epoch": 0.05,
"learning_rate": 9e-07,
"loss": 4.6964,
"step": 16
},
{
"epoch": 0.06,
"learning_rate": 9.600000000000001e-07,
"loss": 4.7848,
"step": 17
},
{
"epoch": 0.06,
"learning_rate": 1.0200000000000002e-06,
"loss": 4.7928,
"step": 18
},
{
"epoch": 0.06,
"learning_rate": 1.08e-06,
"loss": 4.7468,
"step": 19
},
{
"epoch": 0.07,
"learning_rate": 1.14e-06,
"loss": 4.8219,
"step": 20
},
{
"epoch": 0.07,
"learning_rate": 1.2000000000000002e-06,
"loss": 4.8385,
"step": 21
},
{
"epoch": 0.07,
"learning_rate": 1.26e-06,
"loss": 4.7717,
"step": 22
},
{
"epoch": 0.08,
"learning_rate": 1.3199999999999999e-06,
"loss": 4.6769,
"step": 23
},
{
"epoch": 0.08,
"learning_rate": 1.38e-06,
"loss": 4.8167,
"step": 24
},
{
"epoch": 0.08,
"learning_rate": 1.44e-06,
"loss": 4.6377,
"step": 25
},
{
"epoch": 0.09,
"learning_rate": 1.5e-06,
"loss": 4.9288,
"step": 26
},
{
"epoch": 0.09,
"learning_rate": 1.5599999999999999e-06,
"loss": 4.6606,
"step": 27
},
{
"epoch": 0.09,
"learning_rate": 1.62e-06,
"loss": 4.6691,
"step": 28
},
{
"epoch": 0.1,
"learning_rate": 1.68e-06,
"loss": 4.5663,
"step": 29
},
{
"epoch": 0.1,
"learning_rate": 1.74e-06,
"loss": 4.7152,
"step": 30
},
{
"epoch": 0.1,
"learning_rate": 1.8e-06,
"loss": 4.7019,
"step": 31
},
{
"epoch": 0.11,
"learning_rate": 1.86e-06,
"loss": 4.6205,
"step": 32
},
{
"epoch": 0.11,
"learning_rate": 1.9200000000000003e-06,
"loss": 4.6544,
"step": 33
},
{
"epoch": 0.11,
"learning_rate": 1.98e-06,
"loss": 4.5924,
"step": 34
},
{
"epoch": 0.12,
"learning_rate": 2.0400000000000004e-06,
"loss": 4.6187,
"step": 35
},
{
"epoch": 0.12,
"learning_rate": 2.1000000000000002e-06,
"loss": 4.5789,
"step": 36
},
{
"epoch": 0.12,
"learning_rate": 2.16e-06,
"loss": 4.6971,
"step": 37
},
{
"epoch": 0.13,
"learning_rate": 2.22e-06,
"loss": 4.6569,
"step": 38
},
{
"epoch": 0.13,
"learning_rate": 2.28e-06,
"loss": 4.7059,
"step": 39
},
{
"epoch": 0.13,
"learning_rate": 2.34e-06,
"loss": 4.5452,
"step": 40
},
{
"epoch": 0.14,
"learning_rate": 2.4000000000000003e-06,
"loss": 4.6199,
"step": 41
},
{
"epoch": 0.14,
"learning_rate": 2.46e-06,
"loss": 4.7043,
"step": 42
},
{
"epoch": 0.14,
"learning_rate": 2.52e-06,
"loss": 4.4592,
"step": 43
},
{
"epoch": 0.15,
"learning_rate": 2.58e-06,
"loss": 4.7637,
"step": 44
},
{
"epoch": 0.15,
"learning_rate": 2.6399999999999997e-06,
"loss": 4.721,
"step": 45
},
{
"epoch": 0.15,
"learning_rate": 2.7e-06,
"loss": 4.7594,
"step": 46
},
{
"epoch": 0.16,
"learning_rate": 2.76e-06,
"loss": 4.7838,
"step": 47
},
{
"epoch": 0.16,
"learning_rate": 2.82e-06,
"loss": 4.8429,
"step": 48
},
{
"epoch": 0.16,
"learning_rate": 2.88e-06,
"loss": 4.9107,
"step": 49
},
{
"epoch": 0.17,
"learning_rate": 2.9400000000000002e-06,
"loss": 4.7007,
"step": 50
},
{
"epoch": 0.17,
"learning_rate": 3e-06,
"loss": 4.4742,
"step": 51
},
{
"epoch": 0.17,
"learning_rate": 3.06e-06,
"loss": 4.3621,
"step": 52
},
{
"epoch": 0.18,
"learning_rate": 3.1199999999999998e-06,
"loss": 4.313,
"step": 53
},
{
"epoch": 0.18,
"learning_rate": 3.18e-06,
"loss": 4.3308,
"step": 54
},
{
"epoch": 0.18,
"learning_rate": 3.24e-06,
"loss": 4.2678,
"step": 55
},
{
"epoch": 0.19,
"learning_rate": 3.3e-06,
"loss": 4.2976,
"step": 56
},
{
"epoch": 0.19,
"learning_rate": 3.36e-06,
"loss": 4.4298,
"step": 57
},
{
"epoch": 0.2,
"learning_rate": 3.4200000000000003e-06,
"loss": 4.4307,
"step": 58
},
{
"epoch": 0.2,
"learning_rate": 3.48e-06,
"loss": 4.3931,
"step": 59
},
{
"epoch": 0.2,
"learning_rate": 3.54e-06,
"loss": 4.3221,
"step": 60
},
{
"epoch": 0.21,
"learning_rate": 3.6e-06,
"loss": 4.3944,
"step": 61
},
{
"epoch": 0.21,
"learning_rate": 3.66e-06,
"loss": 4.2647,
"step": 62
},
{
"epoch": 0.21,
"learning_rate": 3.72e-06,
"loss": 4.4111,
"step": 63
},
{
"epoch": 0.22,
"learning_rate": 3.7800000000000002e-06,
"loss": 4.3514,
"step": 64
},
{
"epoch": 0.22,
"learning_rate": 3.8400000000000005e-06,
"loss": 4.4005,
"step": 65
},
{
"epoch": 0.22,
"learning_rate": 3.9e-06,
"loss": 4.4042,
"step": 66
},
{
"epoch": 0.23,
"learning_rate": 3.96e-06,
"loss": 4.3515,
"step": 67
},
{
"epoch": 0.23,
"learning_rate": 4.0200000000000005e-06,
"loss": 4.392,
"step": 68
},
{
"epoch": 0.23,
"learning_rate": 4.080000000000001e-06,
"loss": 4.3347,
"step": 69
},
{
"epoch": 0.24,
"learning_rate": 4.14e-06,
"loss": 4.2648,
"step": 70
},
{
"epoch": 0.24,
"learning_rate": 4.2000000000000004e-06,
"loss": 4.2698,
"step": 71
},
{
"epoch": 0.24,
"learning_rate": 4.26e-06,
"loss": 4.3961,
"step": 72
},
{
"epoch": 0.25,
"learning_rate": 4.32e-06,
"loss": 4.3111,
"step": 73
},
{
"epoch": 0.25,
"learning_rate": 4.3799999999999996e-06,
"loss": 4.305,
"step": 74
},
{
"epoch": 0.25,
"learning_rate": 4.44e-06,
"loss": 4.4161,
"step": 75
},
{
"epoch": 0.26,
"learning_rate": 4.5e-06,
"loss": 4.2863,
"step": 76
},
{
"epoch": 0.26,
"learning_rate": 4.56e-06,
"loss": 4.2497,
"step": 77
},
{
"epoch": 0.26,
"learning_rate": 4.62e-06,
"loss": 4.309,
"step": 78
},
{
"epoch": 0.27,
"learning_rate": 4.68e-06,
"loss": 4.1875,
"step": 79
},
{
"epoch": 0.27,
"learning_rate": 4.74e-06,
"loss": 4.3753,
"step": 80
},
{
"epoch": 0.27,
"learning_rate": 4.800000000000001e-06,
"loss": 4.3082,
"step": 81
},
{
"epoch": 0.28,
"learning_rate": 4.86e-06,
"loss": 4.3039,
"step": 82
},
{
"epoch": 0.28,
"learning_rate": 4.92e-06,
"loss": 4.2782,
"step": 83
},
{
"epoch": 0.28,
"learning_rate": 4.980000000000001e-06,
"loss": 4.3015,
"step": 84
},
{
"epoch": 0.29,
"learning_rate": 5.04e-06,
"loss": 4.4091,
"step": 85
},
{
"epoch": 0.29,
"learning_rate": 5.1e-06,
"loss": 4.2884,
"step": 86
},
{
"epoch": 0.29,
"learning_rate": 5.16e-06,
"loss": 4.3427,
"step": 87
},
{
"epoch": 0.3,
"learning_rate": 5.22e-06,
"loss": 4.3656,
"step": 88
},
{
"epoch": 0.3,
"learning_rate": 5.279999999999999e-06,
"loss": 4.3874,
"step": 89
},
{
"epoch": 0.3,
"learning_rate": 5.34e-06,
"loss": 4.409,
"step": 90
},
{
"epoch": 0.31,
"learning_rate": 5.4e-06,
"loss": 4.4265,
"step": 91
},
{
"epoch": 0.31,
"learning_rate": 5.46e-06,
"loss": 4.4879,
"step": 92
},
{
"epoch": 0.31,
"learning_rate": 5.52e-06,
"loss": 4.4224,
"step": 93
},
{
"epoch": 0.32,
"learning_rate": 5.58e-06,
"loss": 4.3628,
"step": 94
},
{
"epoch": 0.32,
"learning_rate": 5.64e-06,
"loss": 4.3252,
"step": 95
},
{
"epoch": 0.32,
"learning_rate": 5.7000000000000005e-06,
"loss": 4.4884,
"step": 96
},
{
"epoch": 0.33,
"learning_rate": 5.76e-06,
"loss": 4.6793,
"step": 97
},
{
"epoch": 0.33,
"learning_rate": 5.82e-06,
"loss": 4.6355,
"step": 98
},
{
"epoch": 0.33,
"learning_rate": 5.8800000000000005e-06,
"loss": 4.6138,
"step": 99
},
{
"epoch": 0.34,
"learning_rate": 5.940000000000001e-06,
"loss": 4.6567,
"step": 100
},
{
"epoch": 0.34,
"learning_rate": 6e-06,
"loss": 4.2214,
"step": 101
},
{
"epoch": 0.34,
"learning_rate": 6.0600000000000004e-06,
"loss": 4.2734,
"step": 102
},
{
"epoch": 0.35,
"learning_rate": 6.12e-06,
"loss": 4.2677,
"step": 103
},
{
"epoch": 0.35,
"learning_rate": 6.18e-06,
"loss": 4.1772,
"step": 104
},
{
"epoch": 0.35,
"learning_rate": 6.2399999999999995e-06,
"loss": 4.2855,
"step": 105
},
{
"epoch": 0.36,
"learning_rate": 6.3e-06,
"loss": 4.2154,
"step": 106
},
{
"epoch": 0.36,
"learning_rate": 6.36e-06,
"loss": 4.1592,
"step": 107
},
{
"epoch": 0.36,
"learning_rate": 6.42e-06,
"loss": 4.2708,
"step": 108
},
{
"epoch": 0.37,
"learning_rate": 6.48e-06,
"loss": 4.2779,
"step": 109
},
{
"epoch": 0.37,
"learning_rate": 6.54e-06,
"loss": 4.2569,
"step": 110
},
{
"epoch": 0.37,
"learning_rate": 6.6e-06,
"loss": 4.3222,
"step": 111
},
{
"epoch": 0.38,
"learning_rate": 6.660000000000001e-06,
"loss": 4.2514,
"step": 112
},
{
"epoch": 0.38,
"learning_rate": 6.72e-06,
"loss": 4.1859,
"step": 113
},
{
"epoch": 0.38,
"learning_rate": 6.78e-06,
"loss": 4.252,
"step": 114
},
{
"epoch": 0.39,
"learning_rate": 6.840000000000001e-06,
"loss": 4.2145,
"step": 115
},
{
"epoch": 0.39,
"learning_rate": 6.900000000000001e-06,
"loss": 4.227,
"step": 116
},
{
"epoch": 0.39,
"learning_rate": 6.96e-06,
"loss": 4.1717,
"step": 117
},
{
"epoch": 0.4,
"learning_rate": 7.0200000000000006e-06,
"loss": 4.2234,
"step": 118
},
{
"epoch": 0.4,
"learning_rate": 7.08e-06,
"loss": 4.2586,
"step": 119
},
{
"epoch": 0.4,
"learning_rate": 7.14e-06,
"loss": 4.2685,
"step": 120
},
{
"epoch": 0.41,
"learning_rate": 7.2e-06,
"loss": 4.2873,
"step": 121
},
{
"epoch": 0.41,
"learning_rate": 7.26e-06,
"loss": 4.3439,
"step": 122
},
{
"epoch": 0.41,
"learning_rate": 7.32e-06,
"loss": 4.1526,
"step": 123
},
{
"epoch": 0.42,
"learning_rate": 7.3800000000000005e-06,
"loss": 4.2956,
"step": 124
},
{
"epoch": 0.42,
"learning_rate": 7.44e-06,
"loss": 4.346,
"step": 125
},
{
"epoch": 0.42,
"learning_rate": 7.5e-06,
"loss": 4.2628,
"step": 126
},
{
"epoch": 0.43,
"learning_rate": 7.5600000000000005e-06,
"loss": 4.3075,
"step": 127
},
{
"epoch": 0.43,
"learning_rate": 7.62e-06,
"loss": 4.1708,
"step": 128
},
{
"epoch": 0.43,
"learning_rate": 7.680000000000001e-06,
"loss": 4.3129,
"step": 129
},
{
"epoch": 0.44,
"learning_rate": 7.74e-06,
"loss": 4.4039,
"step": 130
},
{
"epoch": 0.44,
"learning_rate": 7.8e-06,
"loss": 4.3201,
"step": 131
},
{
"epoch": 0.44,
"learning_rate": 7.860000000000001e-06,
"loss": 4.1504,
"step": 132
},
{
"epoch": 0.45,
"learning_rate": 7.92e-06,
"loss": 4.313,
"step": 133
},
{
"epoch": 0.45,
"learning_rate": 7.98e-06,
"loss": 4.1888,
"step": 134
},
{
"epoch": 0.45,
"learning_rate": 8.040000000000001e-06,
"loss": 4.3874,
"step": 135
},
{
"epoch": 0.46,
"learning_rate": 8.1e-06,
"loss": 4.2967,
"step": 136
},
{
"epoch": 0.46,
"learning_rate": 8.160000000000001e-06,
"loss": 4.2278,
"step": 137
},
{
"epoch": 0.46,
"learning_rate": 8.220000000000001e-06,
"loss": 4.1554,
"step": 138
},
{
"epoch": 0.47,
"learning_rate": 8.28e-06,
"loss": 4.1977,
"step": 139
},
{
"epoch": 0.47,
"learning_rate": 8.340000000000001e-06,
"loss": 4.2253,
"step": 140
},
{
"epoch": 0.47,
"learning_rate": 8.400000000000001e-06,
"loss": 4.4326,
"step": 141
},
{
"epoch": 0.48,
"learning_rate": 8.459999999999999e-06,
"loss": 4.1968,
"step": 142
},
{
"epoch": 0.48,
"learning_rate": 8.52e-06,
"loss": 4.3182,
"step": 143
},
{
"epoch": 0.48,
"learning_rate": 8.58e-06,
"loss": 4.332,
"step": 144
},
{
"epoch": 0.49,
"learning_rate": 8.64e-06,
"loss": 4.4191,
"step": 145
},
{
"epoch": 0.49,
"learning_rate": 8.7e-06,
"loss": 4.2627,
"step": 146
},
{
"epoch": 0.49,
"learning_rate": 8.759999999999999e-06,
"loss": 4.3813,
"step": 147
},
{
"epoch": 0.5,
"learning_rate": 8.82e-06,
"loss": 4.5375,
"step": 148
},
{
"epoch": 0.5,
"learning_rate": 8.88e-06,
"loss": 4.5315,
"step": 149
},
{
"epoch": 0.5,
"learning_rate": 8.939999999999999e-06,
"loss": 4.6419,
"step": 150
},
{
"epoch": 0.51,
"learning_rate": 9e-06,
"loss": 4.2596,
"step": 151
},
{
"epoch": 0.51,
"learning_rate": 9.06e-06,
"loss": 4.2331,
"step": 152
},
{
"epoch": 0.51,
"learning_rate": 9.12e-06,
"loss": 4.2649,
"step": 153
},
{
"epoch": 0.52,
"learning_rate": 9.18e-06,
"loss": 4.197,
"step": 154
},
{
"epoch": 0.52,
"learning_rate": 9.24e-06,
"loss": 4.2249,
"step": 155
},
{
"epoch": 0.52,
"learning_rate": 9.3e-06,
"loss": 4.2527,
"step": 156
},
{
"epoch": 0.53,
"learning_rate": 9.36e-06,
"loss": 4.162,
"step": 157
},
{
"epoch": 0.53,
"learning_rate": 9.42e-06,
"loss": 4.2314,
"step": 158
},
{
"epoch": 0.53,
"learning_rate": 9.48e-06,
"loss": 4.2237,
"step": 159
},
{
"epoch": 0.54,
"learning_rate": 9.54e-06,
"loss": 4.2041,
"step": 160
},
{
"epoch": 0.54,
"learning_rate": 9.600000000000001e-06,
"loss": 4.2173,
"step": 161
},
{
"epoch": 0.54,
"learning_rate": 9.66e-06,
"loss": 4.1877,
"step": 162
},
{
"epoch": 0.55,
"learning_rate": 9.72e-06,
"loss": 4.2562,
"step": 163
},
{
"epoch": 0.55,
"learning_rate": 9.780000000000001e-06,
"loss": 4.1768,
"step": 164
},
{
"epoch": 0.55,
"learning_rate": 9.84e-06,
"loss": 4.24,
"step": 165
},
{
"epoch": 0.56,
"learning_rate": 9.9e-06,
"loss": 4.2073,
"step": 166
},
{
"epoch": 0.56,
"learning_rate": 9.960000000000001e-06,
"loss": 4.2227,
"step": 167
},
{
"epoch": 0.56,
"learning_rate": 1.002e-05,
"loss": 4.2148,
"step": 168
},
{
"epoch": 0.57,
"learning_rate": 1.008e-05,
"loss": 4.188,
"step": 169
},
{
"epoch": 0.57,
"learning_rate": 1.0140000000000001e-05,
"loss": 4.1319,
"step": 170
},
{
"epoch": 0.58,
"learning_rate": 1.02e-05,
"loss": 4.1213,
"step": 171
},
{
"epoch": 0.58,
"learning_rate": 1.0260000000000002e-05,
"loss": 4.2283,
"step": 172
},
{
"epoch": 0.58,
"learning_rate": 1.032e-05,
"loss": 4.0896,
"step": 173
},
{
"epoch": 0.59,
"learning_rate": 1.0379999999999999e-05,
"loss": 4.1497,
"step": 174
},
{
"epoch": 0.59,
"learning_rate": 1.044e-05,
"loss": 4.3003,
"step": 175
},
{
"epoch": 0.59,
"learning_rate": 1.05e-05,
"loss": 4.1535,
"step": 176
},
{
"epoch": 0.6,
"learning_rate": 1.0559999999999999e-05,
"loss": 4.2832,
"step": 177
},
{
"epoch": 0.6,
"learning_rate": 1.062e-05,
"loss": 4.1339,
"step": 178
},
{
"epoch": 0.6,
"learning_rate": 1.068e-05,
"loss": 4.1825,
"step": 179
},
{
"epoch": 0.61,
"learning_rate": 1.074e-05,
"loss": 4.3281,
"step": 180
},
{
"epoch": 0.61,
"learning_rate": 1.08e-05,
"loss": 4.1971,
"step": 181
},
{
"epoch": 0.61,
"learning_rate": 1.086e-05,
"loss": 4.1407,
"step": 182
},
{
"epoch": 0.62,
"learning_rate": 1.092e-05,
"loss": 4.1232,
"step": 183
},
{
"epoch": 0.62,
"learning_rate": 1.098e-05,
"loss": 4.2273,
"step": 184
},
{
"epoch": 0.62,
"learning_rate": 1.104e-05,
"loss": 4.2611,
"step": 185
},
{
"epoch": 0.63,
"learning_rate": 1.11e-05,
"loss": 4.2122,
"step": 186
},
{
"epoch": 0.63,
"learning_rate": 1.116e-05,
"loss": 4.2406,
"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": 4.2793,
"step": 190
},
{
"epoch": 0.64,
"learning_rate": 1.1400000000000001e-05,
"loss": 4.3105,
"step": 191
},
{
"epoch": 0.65,
"learning_rate": 1.146e-05,
"loss": 4.3703,
"step": 192
},
{
"epoch": 0.65,
"learning_rate": 1.152e-05,
"loss": 4.4363,
"step": 193
},
{
"epoch": 0.65,
"learning_rate": 1.1580000000000001e-05,
"loss": 4.3557,
"step": 194
},
{
"epoch": 0.66,
"learning_rate": 1.164e-05,
"loss": 4.3534,
"step": 195
},
{
"epoch": 0.66,
"learning_rate": 1.1700000000000001e-05,
"loss": 4.301,
"step": 196
},
{
"epoch": 0.66,
"learning_rate": 1.1760000000000001e-05,
"loss": 4.1916,
"step": 197
},
{
"epoch": 0.67,
"learning_rate": 1.182e-05,
"loss": 4.6746,
"step": 198
},
{
"epoch": 0.67,
"learning_rate": 1.1880000000000001e-05,
"loss": 4.4123,
"step": 199
},
{
"epoch": 0.67,
"learning_rate": 1.1940000000000001e-05,
"loss": 4.6418,
"step": 200
},
{
"epoch": 0.68,
"learning_rate": 1.2e-05,
"loss": 4.2584,
"step": 201
},
{
"epoch": 0.68,
"learning_rate": 1.2060000000000001e-05,
"loss": 4.1954,
"step": 202
},
{
"epoch": 0.68,
"learning_rate": 1.2120000000000001e-05,
"loss": 4.1895,
"step": 203
},
{
"epoch": 0.69,
"learning_rate": 1.2180000000000002e-05,
"loss": 4.1144,
"step": 204
},
{
"epoch": 0.69,
"learning_rate": 1.224e-05,
"loss": 4.1703,
"step": 205
},
{
"epoch": 0.69,
"learning_rate": 1.2299999999999999e-05,
"loss": 4.1669,
"step": 206
},
{
"epoch": 0.7,
"learning_rate": 1.236e-05,
"loss": 4.2149,
"step": 207
},
{
"epoch": 0.7,
"learning_rate": 1.242e-05,
"loss": 4.1921,
"step": 208
},
{
"epoch": 0.7,
"learning_rate": 1.2479999999999999e-05,
"loss": 4.1564,
"step": 209
},
{
"epoch": 0.71,
"learning_rate": 1.254e-05,
"loss": 4.2629,
"step": 210
},
{
"epoch": 0.71,
"learning_rate": 1.26e-05,
"loss": 4.0521,
"step": 211
},
{
"epoch": 0.71,
"learning_rate": 1.2659999999999999e-05,
"loss": 4.1862,
"step": 212
},
{
"epoch": 0.72,
"learning_rate": 1.272e-05,
"loss": 4.2156,
"step": 213
},
{
"epoch": 0.72,
"learning_rate": 1.278e-05,
"loss": 4.2049,
"step": 214
},
{
"epoch": 0.72,
"learning_rate": 1.284e-05,
"loss": 4.2142,
"step": 215
},
{
"epoch": 0.73,
"learning_rate": 1.29e-05,
"loss": 4.1973,
"step": 216
},
{
"epoch": 0.73,
"learning_rate": 1.296e-05,
"loss": 4.1582,
"step": 217
},
{
"epoch": 0.73,
"learning_rate": 1.302e-05,
"loss": 4.1307,
"step": 218
},
{
"epoch": 0.74,
"learning_rate": 1.308e-05,
"loss": 4.1369,
"step": 219
},
{
"epoch": 0.74,
"learning_rate": 1.314e-05,
"loss": 4.1963,
"step": 220
},
{
"epoch": 0.74,
"learning_rate": 1.32e-05,
"loss": 4.1158,
"step": 221
},
{
"epoch": 0.75,
"learning_rate": 1.326e-05,
"loss": 4.1061,
"step": 222
},
{
"epoch": 0.75,
"learning_rate": 1.3320000000000001e-05,
"loss": 4.1329,
"step": 223
},
{
"epoch": 0.75,
"learning_rate": 1.338e-05,
"loss": 4.1788,
"step": 224
},
{
"epoch": 0.76,
"learning_rate": 1.344e-05,
"loss": 4.1626,
"step": 225
},
{
"epoch": 0.76,
"learning_rate": 1.3500000000000001e-05,
"loss": 4.1729,
"step": 226
},
{
"epoch": 0.76,
"learning_rate": 1.356e-05,
"loss": 4.1994,
"step": 227
},
{
"epoch": 0.77,
"learning_rate": 1.362e-05,
"loss": 4.123,
"step": 228
},
{
"epoch": 0.77,
"learning_rate": 1.3680000000000001e-05,
"loss": 4.1524,
"step": 229
},
{
"epoch": 0.77,
"learning_rate": 1.374e-05,
"loss": 4.1629,
"step": 230
},
{
"epoch": 0.78,
"learning_rate": 1.3800000000000002e-05,
"loss": 4.2594,
"step": 231
},
{
"epoch": 0.78,
"learning_rate": 1.3860000000000001e-05,
"loss": 4.1334,
"step": 232
},
{
"epoch": 0.78,
"learning_rate": 1.392e-05,
"loss": 4.2471,
"step": 233
},
{
"epoch": 0.79,
"learning_rate": 1.3980000000000002e-05,
"loss": 4.2207,
"step": 234
},
{
"epoch": 0.79,
"learning_rate": 1.4040000000000001e-05,
"loss": 4.2613,
"step": 235
},
{
"epoch": 0.79,
"learning_rate": 1.4099999999999999e-05,
"loss": 4.0942,
"step": 236
},
{
"epoch": 0.8,
"learning_rate": 1.416e-05,
"loss": 4.1965,
"step": 237
},
{
"epoch": 0.8,
"learning_rate": 1.422e-05,
"loss": 4.3305,
"step": 238
},
{
"epoch": 0.8,
"learning_rate": 1.428e-05,
"loss": 4.1596,
"step": 239
},
{
"epoch": 0.81,
"learning_rate": 1.434e-05,
"loss": 4.2163,
"step": 240
},
{
"epoch": 0.81,
"learning_rate": 1.44e-05,
"loss": 4.3178,
"step": 241
},
{
"epoch": 0.81,
"learning_rate": 1.446e-05,
"loss": 4.2505,
"step": 242
},
{
"epoch": 0.82,
"learning_rate": 1.452e-05,
"loss": 4.2029,
"step": 243
},
{
"epoch": 0.82,
"learning_rate": 1.458e-05,
"loss": 4.2315,
"step": 244
},
{
"epoch": 0.82,
"learning_rate": 1.464e-05,
"loss": 4.2068,
"step": 245
},
{
"epoch": 0.83,
"learning_rate": 1.47e-05,
"loss": 4.3899,
"step": 246
},
{
"epoch": 0.83,
"learning_rate": 1.4760000000000001e-05,
"loss": 4.3341,
"step": 247
},
{
"epoch": 0.83,
"learning_rate": 1.482e-05,
"loss": 4.3292,
"step": 248
},
{
"epoch": 0.84,
"learning_rate": 1.488e-05,
"loss": 4.4397,
"step": 249
},
{
"epoch": 0.84,
"learning_rate": 1.4940000000000001e-05,
"loss": 4.4744,
"step": 250
},
{
"epoch": 0.84,
"learning_rate": 1.5e-05,
"loss": 4.2392,
"step": 251
},
{
"epoch": 0.85,
"learning_rate": 1.506e-05,
"loss": 4.1471,
"step": 252
},
{
"epoch": 0.85,
"learning_rate": 1.5120000000000001e-05,
"loss": 4.1138,
"step": 253
},
{
"epoch": 0.85,
"learning_rate": 1.518e-05,
"loss": 4.1184,
"step": 254
},
{
"epoch": 0.86,
"learning_rate": 1.524e-05,
"loss": 4.1417,
"step": 255
},
{
"epoch": 0.86,
"learning_rate": 1.53e-05,
"loss": 4.1246,
"step": 256
},
{
"epoch": 0.86,
"learning_rate": 1.5360000000000002e-05,
"loss": 4.2689,
"step": 257
},
{
"epoch": 0.87,
"learning_rate": 1.542e-05,
"loss": 4.1376,
"step": 258
},
{
"epoch": 0.87,
"learning_rate": 1.548e-05,
"loss": 4.1749,
"step": 259
},
{
"epoch": 0.87,
"learning_rate": 1.554e-05,
"loss": 4.0988,
"step": 260
},
{
"epoch": 0.88,
"learning_rate": 1.56e-05,
"loss": 4.1985,
"step": 261
},
{
"epoch": 0.88,
"learning_rate": 1.5660000000000003e-05,
"loss": 4.2305,
"step": 262
},
{
"epoch": 0.88,
"learning_rate": 1.5720000000000002e-05,
"loss": 4.095,
"step": 263
},
{
"epoch": 0.89,
"learning_rate": 1.578e-05,
"loss": 4.133,
"step": 264
},
{
"epoch": 0.89,
"learning_rate": 1.584e-05,
"loss": 4.1428,
"step": 265
},
{
"epoch": 0.89,
"learning_rate": 1.59e-05,
"loss": 4.2381,
"step": 266
},
{
"epoch": 0.9,
"learning_rate": 1.596e-05,
"loss": 4.1467,
"step": 267
},
{
"epoch": 0.9,
"learning_rate": 1.6020000000000002e-05,
"loss": 4.1078,
"step": 268
},
{
"epoch": 0.9,
"learning_rate": 1.6080000000000002e-05,
"loss": 4.0748,
"step": 269
},
{
"epoch": 0.91,
"learning_rate": 1.614e-05,
"loss": 4.21,
"step": 270
},
{
"epoch": 0.91,
"learning_rate": 1.62e-05,
"loss": 4.0815,
"step": 271
},
{
"epoch": 0.91,
"learning_rate": 1.626e-05,
"loss": 4.0578,
"step": 272
},
{
"epoch": 0.92,
"learning_rate": 1.6320000000000003e-05,
"loss": 4.1182,
"step": 273
},
{
"epoch": 0.92,
"learning_rate": 1.6380000000000002e-05,
"loss": 4.0982,
"step": 274
},
{
"epoch": 0.92,
"learning_rate": 1.6440000000000002e-05,
"loss": 4.2321,
"step": 275
},
{
"epoch": 0.93,
"learning_rate": 1.65e-05,
"loss": 4.1565,
"step": 276
},
{
"epoch": 0.93,
"learning_rate": 1.656e-05,
"loss": 4.1676,
"step": 277
},
{
"epoch": 0.93,
"learning_rate": 1.6620000000000004e-05,
"loss": 4.0858,
"step": 278
},
{
"epoch": 0.94,
"learning_rate": 1.6680000000000003e-05,
"loss": 4.1377,
"step": 279
},
{
"epoch": 0.94,
"learning_rate": 1.6740000000000002e-05,
"loss": 4.1011,
"step": 280
},
{
"epoch": 0.94,
"learning_rate": 1.6800000000000002e-05,
"loss": 4.1503,
"step": 281
},
{
"epoch": 0.95,
"learning_rate": 1.686e-05,
"loss": 4.1427,
"step": 282
},
{
"epoch": 0.95,
"learning_rate": 1.6919999999999997e-05,
"loss": 4.3036,
"step": 283
},
{
"epoch": 0.96,
"learning_rate": 1.698e-05,
"loss": 4.0708,
"step": 284
},
{
"epoch": 0.96,
"learning_rate": 1.704e-05,
"loss": 4.1898,
"step": 285
},
{
"epoch": 0.96,
"learning_rate": 1.71e-05,
"loss": 4.2033,
"step": 286
},
{
"epoch": 0.97,
"learning_rate": 1.716e-05,
"loss": 4.1129,
"step": 287
},
{
"epoch": 0.97,
"learning_rate": 1.7219999999999998e-05,
"loss": 4.1294,
"step": 288
},
{
"epoch": 0.97,
"learning_rate": 1.728e-05,
"loss": 4.138,
"step": 289
},
{
"epoch": 0.98,
"learning_rate": 1.734e-05,
"loss": 4.2347,
"step": 290
},
{
"epoch": 0.98,
"learning_rate": 1.74e-05,
"loss": 4.2166,
"step": 291
},
{
"epoch": 0.98,
"learning_rate": 1.746e-05,
"loss": 4.3042,
"step": 292
},
{
"epoch": 0.99,
"learning_rate": 1.7519999999999998e-05,
"loss": 4.2677,
"step": 293
},
{
"epoch": 0.99,
"learning_rate": 1.758e-05,
"loss": 4.4732,
"step": 294
},
{
"epoch": 0.99,
"learning_rate": 1.764e-05,
"loss": 4.2505,
"step": 295
},
{
"epoch": 1.0,
"learning_rate": 1.77e-05,
"loss": 4.6178,
"step": 296
},
{
"epoch": 1.0,
"learning_rate": 1.776e-05,
"loss": 4.5025,
"step": 297
},
{
"epoch": 1.0,
"step": 297,
"total_flos": 0.0,
"train_loss": 4.34206679051974,
"train_runtime": 4631.5065,
"train_samples_per_second": 6.162,
"train_steps_per_second": 0.064
}
],
"max_steps": 297,
"num_train_epochs": 1,
"total_flos": 0.0,
"trial_name": null,
"trial_params": null
}