schedule_llm / 0320-7B /trainer_state.json
kylinux's picture
Upload folder using huggingface_hub
88e80b2 verified
Raw
History Blame Contribute Delete
57.6 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 2.9952,
"eval_steps": 500,
"global_step": 468,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0064,
"learning_rate": 0.0,
"loss": 1.557,
"step": 1
},
{
"epoch": 0.0128,
"learning_rate": 7.525749891599529e-06,
"loss": 0.701,
"step": 2
},
{
"epoch": 0.0192,
"learning_rate": 1.192803136799156e-05,
"loss": 0.6793,
"step": 3
},
{
"epoch": 0.0256,
"learning_rate": 1.5051499783199057e-05,
"loss": 0.5178,
"step": 4
},
{
"epoch": 0.032,
"learning_rate": 1.7474250108400467e-05,
"loss": 0.3516,
"step": 5
},
{
"epoch": 0.0384,
"learning_rate": 1.945378125959109e-05,
"loss": 0.1764,
"step": 6
},
{
"epoch": 0.0448,
"learning_rate": 2.1127451000356418e-05,
"loss": 0.1193,
"step": 7
},
{
"epoch": 0.0512,
"learning_rate": 2.2577249674798584e-05,
"loss": 0.082,
"step": 8
},
{
"epoch": 0.0576,
"learning_rate": 2.385606273598312e-05,
"loss": 0.0655,
"step": 9
},
{
"epoch": 0.064,
"learning_rate": 2.4999999999999998e-05,
"loss": 0.0568,
"step": 10
},
{
"epoch": 0.0704,
"learning_rate": 2.6034817128955623e-05,
"loss": 0.0382,
"step": 11
},
{
"epoch": 0.0768,
"learning_rate": 2.6979531151190617e-05,
"loss": 0.0311,
"step": 12
},
{
"epoch": 0.0832,
"learning_rate": 2.7848583807670913e-05,
"loss": 0.0341,
"step": 13
},
{
"epoch": 0.0896,
"learning_rate": 2.8653200891955945e-05,
"loss": 0.0363,
"step": 14
},
{
"epoch": 0.096,
"learning_rate": 2.940228147639203e-05,
"loss": 0.0259,
"step": 15
},
{
"epoch": 0.1024,
"learning_rate": 3.0102999566398115e-05,
"loss": 0.019,
"step": 16
},
{
"epoch": 0.1088,
"learning_rate": 3.076122303445685e-05,
"loss": 0.0236,
"step": 17
},
{
"epoch": 0.1152,
"learning_rate": 3.1381812627582646e-05,
"loss": 0.0212,
"step": 18
},
{
"epoch": 0.1216,
"learning_rate": 3.1968840023820715e-05,
"loss": 0.0151,
"step": 19
},
{
"epoch": 0.128,
"learning_rate": 3.2525749891599525e-05,
"loss": 0.0147,
"step": 20
},
{
"epoch": 0.1344,
"learning_rate": 3.305548236834798e-05,
"loss": 0.0145,
"step": 21
},
{
"epoch": 0.1408,
"learning_rate": 3.3560567020555153e-05,
"loss": 0.0127,
"step": 22
},
{
"epoch": 0.1472,
"learning_rate": 3.404319590043982e-05,
"loss": 0.01,
"step": 23
},
{
"epoch": 0.1536,
"learning_rate": 3.450528104279015e-05,
"loss": 0.0147,
"step": 24
},
{
"epoch": 0.16,
"learning_rate": 3.4948500216800935e-05,
"loss": 0.0155,
"step": 25
},
{
"epoch": 0.1664,
"learning_rate": 3.537433369927044e-05,
"loss": 0.0121,
"step": 26
},
{
"epoch": 0.1728,
"learning_rate": 3.578409410397468e-05,
"loss": 0.0153,
"step": 27
},
{
"epoch": 0.1792,
"learning_rate": 3.6178950783555475e-05,
"loss": 0.0103,
"step": 28
},
{
"epoch": 0.1856,
"learning_rate": 3.65599499474739e-05,
"loss": 0.0099,
"step": 29
},
{
"epoch": 0.192,
"learning_rate": 3.6928031367991554e-05,
"loss": 0.011,
"step": 30
},
{
"epoch": 0.1984,
"learning_rate": 3.728404234585681e-05,
"loss": 0.013,
"step": 31
},
{
"epoch": 0.2048,
"learning_rate": 3.762874945799765e-05,
"loss": 0.0068,
"step": 32
},
{
"epoch": 0.2112,
"learning_rate": 3.796284849694718e-05,
"loss": 0.0068,
"step": 33
},
{
"epoch": 0.2176,
"learning_rate": 3.8286972926056376e-05,
"loss": 0.0109,
"step": 34
},
{
"epoch": 0.224,
"learning_rate": 3.8601701108756885e-05,
"loss": 0.0133,
"step": 35
},
{
"epoch": 0.2304,
"learning_rate": 3.890756251918218e-05,
"loss": 0.0046,
"step": 36
},
{
"epoch": 0.2368,
"learning_rate": 3.920504310167487e-05,
"loss": 0.0041,
"step": 37
},
{
"epoch": 0.2432,
"learning_rate": 3.949458991542025e-05,
"loss": 0.0045,
"step": 38
},
{
"epoch": 0.2496,
"learning_rate": 3.977661517566247e-05,
"loss": 0.0078,
"step": 39
},
{
"epoch": 0.256,
"learning_rate": 4.005149978319905e-05,
"loss": 0.0041,
"step": 40
},
{
"epoch": 0.2624,
"learning_rate": 4.031959641799338e-05,
"loss": 0.0031,
"step": 41
},
{
"epoch": 0.2688,
"learning_rate": 4.058123225994751e-05,
"loss": 0.0036,
"step": 42
},
{
"epoch": 0.2752,
"learning_rate": 4.0836711389489654e-05,
"loss": 0.0034,
"step": 43
},
{
"epoch": 0.2816,
"learning_rate": 4.108631691215468e-05,
"loss": 0.0024,
"step": 44
},
{
"epoch": 0.288,
"learning_rate": 4.133031284438358e-05,
"loss": 0.0042,
"step": 45
},
{
"epoch": 0.2944,
"learning_rate": 4.156894579203935e-05,
"loss": 0.0062,
"step": 46
},
{
"epoch": 0.3008,
"learning_rate": 4.180244644839293e-05,
"loss": 0.0065,
"step": 47
},
{
"epoch": 0.3072,
"learning_rate": 4.203103093438968e-05,
"loss": 0.0099,
"step": 48
},
{
"epoch": 0.3136,
"learning_rate": 4.2254902000712836e-05,
"loss": 0.0022,
"step": 49
},
{
"epoch": 0.32,
"learning_rate": 4.247425010840046e-05,
"loss": 0.0028,
"step": 50
},
{
"epoch": 0.3264,
"learning_rate": 4.2689254402448405e-05,
"loss": 0.005,
"step": 51
},
{
"epoch": 0.3328,
"learning_rate": 4.290008359086998e-05,
"loss": 0.0021,
"step": 52
},
{
"epoch": 0.3392,
"learning_rate": 4.310689674001973e-05,
"loss": 0.0059,
"step": 53
},
{
"epoch": 0.3456,
"learning_rate": 4.330984399557421e-05,
"loss": 0.004,
"step": 54
},
{
"epoch": 0.352,
"learning_rate": 4.350906723735609e-05,
"loss": 0.0035,
"step": 55
},
{
"epoch": 0.3584,
"learning_rate": 4.370470067515501e-05,
"loss": 0.0026,
"step": 56
},
{
"epoch": 0.3648,
"learning_rate": 4.3896871391812285e-05,
"loss": 0.0092,
"step": 57
},
{
"epoch": 0.3712,
"learning_rate": 4.408569983907343e-05,
"loss": 0.0037,
"step": 58
},
{
"epoch": 0.3776,
"learning_rate": 4.42713002910536e-05,
"loss": 0.0092,
"step": 59
},
{
"epoch": 0.384,
"learning_rate": 4.445378125959108e-05,
"loss": 0.005,
"step": 60
},
{
"epoch": 0.3904,
"learning_rate": 4.463324587526917e-05,
"loss": 0.0018,
"step": 61
},
{
"epoch": 0.3968,
"learning_rate": 4.4809792237456346e-05,
"loss": 0.007,
"step": 62
},
{
"epoch": 0.4032,
"learning_rate": 4.498351373633954e-05,
"loss": 0.0023,
"step": 63
},
{
"epoch": 0.4096,
"learning_rate": 4.515449934959717e-05,
"loss": 0.0045,
"step": 64
},
{
"epoch": 0.416,
"learning_rate": 4.532283391607138e-05,
"loss": 0.0049,
"step": 65
},
{
"epoch": 0.4224,
"learning_rate": 4.548859838854671e-05,
"loss": 0.0015,
"step": 66
},
{
"epoch": 0.4288,
"learning_rate": 4.565187006752065e-05,
"loss": 0.0169,
"step": 67
},
{
"epoch": 0.4352,
"learning_rate": 4.581272281765591e-05,
"loss": 0.0069,
"step": 68
},
{
"epoch": 0.4416,
"learning_rate": 4.597122726843138e-05,
"loss": 0.0187,
"step": 69
},
{
"epoch": 0.448,
"learning_rate": 4.612745100035642e-05,
"loss": 0.0076,
"step": 70
},
{
"epoch": 0.4544,
"learning_rate": 4.628145871797688e-05,
"loss": 0.0105,
"step": 71
},
{
"epoch": 0.4608,
"learning_rate": 4.643331241078171e-05,
"loss": 0.0135,
"step": 72
},
{
"epoch": 0.4672,
"learning_rate": 4.658307150301139e-05,
"loss": 0.006,
"step": 73
},
{
"epoch": 0.4736,
"learning_rate": 4.67307929932744e-05,
"loss": 0.0034,
"step": 74
},
{
"epoch": 0.48,
"learning_rate": 4.687653158479249e-05,
"loss": 0.0036,
"step": 75
},
{
"epoch": 0.4864,
"learning_rate": 4.702033980701978e-05,
"loss": 0.0089,
"step": 76
},
{
"epoch": 0.4928,
"learning_rate": 4.716226812931204e-05,
"loss": 0.0022,
"step": 77
},
{
"epoch": 0.4992,
"learning_rate": 4.7302365067262006e-05,
"loss": 0.0033,
"step": 78
},
{
"epoch": 0.5056,
"learning_rate": 4.744067728226103e-05,
"loss": 0.0014,
"step": 79
},
{
"epoch": 0.512,
"learning_rate": 4.757724967479858e-05,
"loss": 0.0042,
"step": 80
},
{
"epoch": 0.5184,
"learning_rate": 4.771212547196624e-05,
"loss": 0.0034,
"step": 81
},
{
"epoch": 0.5248,
"learning_rate": 4.7845346309592914e-05,
"loss": 0.0057,
"step": 82
},
{
"epoch": 0.5312,
"learning_rate": 4.7976952309401844e-05,
"loss": 0.0084,
"step": 83
},
{
"epoch": 0.5376,
"learning_rate": 4.810698215154703e-05,
"loss": 0.0018,
"step": 84
},
{
"epoch": 0.544,
"learning_rate": 4.823547314285732e-05,
"loss": 0.0008,
"step": 85
},
{
"epoch": 0.5504,
"learning_rate": 4.836246128108918e-05,
"loss": 0.0022,
"step": 86
},
{
"epoch": 0.5568,
"learning_rate": 4.8487981315465456e-05,
"loss": 0.01,
"step": 87
},
{
"epoch": 0.5632,
"learning_rate": 4.8612066803754214e-05,
"loss": 0.0016,
"step": 88
},
{
"epoch": 0.5696,
"learning_rate": 4.873475016612281e-05,
"loss": 0.0038,
"step": 89
},
{
"epoch": 0.576,
"learning_rate": 4.885606273598312e-05,
"loss": 0.0015,
"step": 90
},
{
"epoch": 0.5824,
"learning_rate": 4.897603480802733e-05,
"loss": 0.0015,
"step": 91
},
{
"epoch": 0.5888,
"learning_rate": 4.909469568363888e-05,
"loss": 0.0026,
"step": 92
},
{
"epoch": 0.5952,
"learning_rate": 4.9212073713848375e-05,
"loss": 0.0375,
"step": 93
},
{
"epoch": 0.6016,
"learning_rate": 4.932819633999246e-05,
"loss": 0.0036,
"step": 94
},
{
"epoch": 0.608,
"learning_rate": 4.9443090132221186e-05,
"loss": 0.0005,
"step": 95
},
{
"epoch": 0.6144,
"learning_rate": 4.9556780825989205e-05,
"loss": 0.0111,
"step": 96
},
{
"epoch": 0.6208,
"learning_rate": 4.9669293356656114e-05,
"loss": 0.0016,
"step": 97
},
{
"epoch": 0.6272,
"learning_rate": 4.978065189231237e-05,
"loss": 0.0018,
"step": 98
},
{
"epoch": 0.6336,
"learning_rate": 4.989087986493874e-05,
"loss": 0.0055,
"step": 99
},
{
"epoch": 0.64,
"learning_rate": 4.9999999999999996e-05,
"loss": 0.0068,
"step": 100
},
{
"epoch": 0.6464,
"learning_rate": 5e-05,
"loss": 0.0075,
"step": 101
},
{
"epoch": 0.6528,
"learning_rate": 4.986413043478261e-05,
"loss": 0.0066,
"step": 102
},
{
"epoch": 0.6592,
"learning_rate": 4.9728260869565216e-05,
"loss": 0.0079,
"step": 103
},
{
"epoch": 0.6656,
"learning_rate": 4.959239130434783e-05,
"loss": 0.0135,
"step": 104
},
{
"epoch": 0.672,
"learning_rate": 4.945652173913044e-05,
"loss": 0.0181,
"step": 105
},
{
"epoch": 0.6784,
"learning_rate": 4.932065217391305e-05,
"loss": 0.0042,
"step": 106
},
{
"epoch": 0.6848,
"learning_rate": 4.918478260869566e-05,
"loss": 0.0016,
"step": 107
},
{
"epoch": 0.6912,
"learning_rate": 4.904891304347826e-05,
"loss": 0.0011,
"step": 108
},
{
"epoch": 0.6976,
"learning_rate": 4.891304347826087e-05,
"loss": 0.0033,
"step": 109
},
{
"epoch": 0.704,
"learning_rate": 4.8777173913043476e-05,
"loss": 0.0029,
"step": 110
},
{
"epoch": 0.7104,
"learning_rate": 4.8641304347826086e-05,
"loss": 0.003,
"step": 111
},
{
"epoch": 0.7168,
"learning_rate": 4.8505434782608696e-05,
"loss": 0.0011,
"step": 112
},
{
"epoch": 0.7232,
"learning_rate": 4.836956521739131e-05,
"loss": 0.0007,
"step": 113
},
{
"epoch": 0.7296,
"learning_rate": 4.823369565217392e-05,
"loss": 0.0008,
"step": 114
},
{
"epoch": 0.736,
"learning_rate": 4.809782608695653e-05,
"loss": 0.0013,
"step": 115
},
{
"epoch": 0.7424,
"learning_rate": 4.796195652173913e-05,
"loss": 0.0021,
"step": 116
},
{
"epoch": 0.7488,
"learning_rate": 4.782608695652174e-05,
"loss": 0.0011,
"step": 117
},
{
"epoch": 0.7552,
"learning_rate": 4.7690217391304345e-05,
"loss": 0.0008,
"step": 118
},
{
"epoch": 0.7616,
"learning_rate": 4.7554347826086956e-05,
"loss": 0.001,
"step": 119
},
{
"epoch": 0.768,
"learning_rate": 4.741847826086957e-05,
"loss": 0.001,
"step": 120
},
{
"epoch": 0.7744,
"learning_rate": 4.7282608695652177e-05,
"loss": 0.008,
"step": 121
},
{
"epoch": 0.7808,
"learning_rate": 4.714673913043479e-05,
"loss": 0.0017,
"step": 122
},
{
"epoch": 0.7872,
"learning_rate": 4.701086956521739e-05,
"loss": 0.0006,
"step": 123
},
{
"epoch": 0.7936,
"learning_rate": 4.6875e-05,
"loss": 0.0013,
"step": 124
},
{
"epoch": 0.8,
"learning_rate": 4.673913043478261e-05,
"loss": 0.001,
"step": 125
},
{
"epoch": 0.8064,
"learning_rate": 4.660326086956522e-05,
"loss": 0.0014,
"step": 126
},
{
"epoch": 0.8128,
"learning_rate": 4.646739130434783e-05,
"loss": 0.0002,
"step": 127
},
{
"epoch": 0.8192,
"learning_rate": 4.6331521739130436e-05,
"loss": 0.0008,
"step": 128
},
{
"epoch": 0.8256,
"learning_rate": 4.6195652173913046e-05,
"loss": 0.0042,
"step": 129
},
{
"epoch": 0.832,
"learning_rate": 4.6059782608695657e-05,
"loss": 0.0039,
"step": 130
},
{
"epoch": 0.8384,
"learning_rate": 4.592391304347826e-05,
"loss": 0.0013,
"step": 131
},
{
"epoch": 0.8448,
"learning_rate": 4.578804347826087e-05,
"loss": 0.0028,
"step": 132
},
{
"epoch": 0.8512,
"learning_rate": 4.565217391304348e-05,
"loss": 0.0023,
"step": 133
},
{
"epoch": 0.8576,
"learning_rate": 4.551630434782609e-05,
"loss": 0.0016,
"step": 134
},
{
"epoch": 0.864,
"learning_rate": 4.53804347826087e-05,
"loss": 0.0013,
"step": 135
},
{
"epoch": 0.8704,
"learning_rate": 4.5244565217391305e-05,
"loss": 0.0018,
"step": 136
},
{
"epoch": 0.8768,
"learning_rate": 4.5108695652173916e-05,
"loss": 0.0017,
"step": 137
},
{
"epoch": 0.8832,
"learning_rate": 4.4972826086956526e-05,
"loss": 0.0006,
"step": 138
},
{
"epoch": 0.8896,
"learning_rate": 4.483695652173913e-05,
"loss": 0.0029,
"step": 139
},
{
"epoch": 0.896,
"learning_rate": 4.470108695652174e-05,
"loss": 0.0025,
"step": 140
},
{
"epoch": 0.9024,
"learning_rate": 4.456521739130435e-05,
"loss": 0.0012,
"step": 141
},
{
"epoch": 0.9088,
"learning_rate": 4.442934782608696e-05,
"loss": 0.0017,
"step": 142
},
{
"epoch": 0.9152,
"learning_rate": 4.429347826086957e-05,
"loss": 0.0014,
"step": 143
},
{
"epoch": 0.9216,
"learning_rate": 4.4157608695652175e-05,
"loss": 0.0016,
"step": 144
},
{
"epoch": 0.928,
"learning_rate": 4.4021739130434786e-05,
"loss": 0.0014,
"step": 145
},
{
"epoch": 0.9344,
"learning_rate": 4.388586956521739e-05,
"loss": 0.0016,
"step": 146
},
{
"epoch": 0.9408,
"learning_rate": 4.375e-05,
"loss": 0.0026,
"step": 147
},
{
"epoch": 0.9472,
"learning_rate": 4.361413043478261e-05,
"loss": 0.0008,
"step": 148
},
{
"epoch": 0.9536,
"learning_rate": 4.347826086956522e-05,
"loss": 0.0016,
"step": 149
},
{
"epoch": 0.96,
"learning_rate": 4.334239130434783e-05,
"loss": 0.0011,
"step": 150
},
{
"epoch": 0.9664,
"learning_rate": 4.3206521739130434e-05,
"loss": 0.0013,
"step": 151
},
{
"epoch": 0.9728,
"learning_rate": 4.3070652173913045e-05,
"loss": 0.001,
"step": 152
},
{
"epoch": 0.9792,
"learning_rate": 4.2934782608695655e-05,
"loss": 0.0014,
"step": 153
},
{
"epoch": 0.9856,
"learning_rate": 4.279891304347826e-05,
"loss": 0.0005,
"step": 154
},
{
"epoch": 0.992,
"learning_rate": 4.266304347826087e-05,
"loss": 0.003,
"step": 155
},
{
"epoch": 0.9984,
"learning_rate": 4.252717391304348e-05,
"loss": 0.0013,
"step": 156
},
{
"epoch": 1.0048,
"learning_rate": 4.239130434782609e-05,
"loss": 0.0001,
"step": 157
},
{
"epoch": 1.0112,
"learning_rate": 4.22554347826087e-05,
"loss": 0.0016,
"step": 158
},
{
"epoch": 1.0176,
"learning_rate": 4.2119565217391304e-05,
"loss": 0.0018,
"step": 159
},
{
"epoch": 1.024,
"learning_rate": 4.1983695652173914e-05,
"loss": 0.0008,
"step": 160
},
{
"epoch": 1.0304,
"learning_rate": 4.1847826086956525e-05,
"loss": 0.0073,
"step": 161
},
{
"epoch": 1.0368,
"learning_rate": 4.171195652173913e-05,
"loss": 0.0013,
"step": 162
},
{
"epoch": 1.0432,
"learning_rate": 4.1576086956521746e-05,
"loss": 0.0009,
"step": 163
},
{
"epoch": 1.0496,
"learning_rate": 4.144021739130435e-05,
"loss": 0.0017,
"step": 164
},
{
"epoch": 1.056,
"learning_rate": 4.130434782608696e-05,
"loss": 0.0015,
"step": 165
},
{
"epoch": 1.0624,
"learning_rate": 4.116847826086957e-05,
"loss": 0.0021,
"step": 166
},
{
"epoch": 1.0688,
"learning_rate": 4.1032608695652174e-05,
"loss": 0.0026,
"step": 167
},
{
"epoch": 1.0752,
"learning_rate": 4.0896739130434784e-05,
"loss": 0.0008,
"step": 168
},
{
"epoch": 1.0816,
"learning_rate": 4.076086956521739e-05,
"loss": 0.0003,
"step": 169
},
{
"epoch": 1.088,
"learning_rate": 4.0625000000000005e-05,
"loss": 0.0002,
"step": 170
},
{
"epoch": 1.0944,
"learning_rate": 4.0489130434782615e-05,
"loss": 0.0005,
"step": 171
},
{
"epoch": 1.1008,
"learning_rate": 4.035326086956522e-05,
"loss": 0.001,
"step": 172
},
{
"epoch": 1.1072,
"learning_rate": 4.021739130434783e-05,
"loss": 0.0009,
"step": 173
},
{
"epoch": 1.1136,
"learning_rate": 4.008152173913043e-05,
"loss": 0.0001,
"step": 174
},
{
"epoch": 1.12,
"learning_rate": 3.994565217391304e-05,
"loss": 0.0011,
"step": 175
},
{
"epoch": 1.1264,
"learning_rate": 3.9809782608695654e-05,
"loss": 0.001,
"step": 176
},
{
"epoch": 1.1328,
"learning_rate": 3.9673913043478264e-05,
"loss": 0.0003,
"step": 177
},
{
"epoch": 1.1392,
"learning_rate": 3.9538043478260875e-05,
"loss": 0.0012,
"step": 178
},
{
"epoch": 1.1456,
"learning_rate": 3.940217391304348e-05,
"loss": 0.0001,
"step": 179
},
{
"epoch": 1.152,
"learning_rate": 3.926630434782609e-05,
"loss": 0.0007,
"step": 180
},
{
"epoch": 1.1584,
"learning_rate": 3.91304347826087e-05,
"loss": 0.0006,
"step": 181
},
{
"epoch": 1.1648,
"learning_rate": 3.89945652173913e-05,
"loss": 0.0009,
"step": 182
},
{
"epoch": 1.1712,
"learning_rate": 3.885869565217391e-05,
"loss": 0.0015,
"step": 183
},
{
"epoch": 1.1776,
"learning_rate": 3.8722826086956523e-05,
"loss": 0.0004,
"step": 184
},
{
"epoch": 1.184,
"learning_rate": 3.8586956521739134e-05,
"loss": 0.001,
"step": 185
},
{
"epoch": 1.1904,
"learning_rate": 3.8451086956521744e-05,
"loss": 0.0008,
"step": 186
},
{
"epoch": 1.1968,
"learning_rate": 3.831521739130435e-05,
"loss": 0.0003,
"step": 187
},
{
"epoch": 1.2032,
"learning_rate": 3.817934782608696e-05,
"loss": 0.0003,
"step": 188
},
{
"epoch": 1.2096,
"learning_rate": 3.804347826086957e-05,
"loss": 0.0023,
"step": 189
},
{
"epoch": 1.216,
"learning_rate": 3.790760869565217e-05,
"loss": 0.0004,
"step": 190
},
{
"epoch": 1.2224,
"learning_rate": 3.777173913043478e-05,
"loss": 0.0006,
"step": 191
},
{
"epoch": 1.2288000000000001,
"learning_rate": 3.763586956521739e-05,
"loss": 0.0007,
"step": 192
},
{
"epoch": 1.2352,
"learning_rate": 3.7500000000000003e-05,
"loss": 0.0006,
"step": 193
},
{
"epoch": 1.2416,
"learning_rate": 3.7364130434782614e-05,
"loss": 0.0,
"step": 194
},
{
"epoch": 1.248,
"learning_rate": 3.722826086956522e-05,
"loss": 0.0005,
"step": 195
},
{
"epoch": 1.2544,
"learning_rate": 3.709239130434783e-05,
"loss": 0.0009,
"step": 196
},
{
"epoch": 1.2608,
"learning_rate": 3.695652173913043e-05,
"loss": 0.0011,
"step": 197
},
{
"epoch": 1.2671999999999999,
"learning_rate": 3.682065217391304e-05,
"loss": 0.0012,
"step": 198
},
{
"epoch": 1.2736,
"learning_rate": 3.668478260869566e-05,
"loss": 0.0001,
"step": 199
},
{
"epoch": 1.28,
"learning_rate": 3.654891304347826e-05,
"loss": 0.0001,
"step": 200
},
{
"epoch": 1.2864,
"learning_rate": 3.641304347826087e-05,
"loss": 0.0009,
"step": 201
},
{
"epoch": 1.2928,
"learning_rate": 3.627717391304348e-05,
"loss": 0.0004,
"step": 202
},
{
"epoch": 1.2992,
"learning_rate": 3.614130434782609e-05,
"loss": 0.001,
"step": 203
},
{
"epoch": 1.3056,
"learning_rate": 3.60054347826087e-05,
"loss": 0.0002,
"step": 204
},
{
"epoch": 1.312,
"learning_rate": 3.58695652173913e-05,
"loss": 0.0006,
"step": 205
},
{
"epoch": 1.3184,
"learning_rate": 3.573369565217392e-05,
"loss": 0.0004,
"step": 206
},
{
"epoch": 1.3248,
"learning_rate": 3.559782608695653e-05,
"loss": 0.0001,
"step": 207
},
{
"epoch": 1.3312,
"learning_rate": 3.546195652173913e-05,
"loss": 0.0003,
"step": 208
},
{
"epoch": 1.3376000000000001,
"learning_rate": 3.532608695652174e-05,
"loss": 0.0004,
"step": 209
},
{
"epoch": 1.3439999999999999,
"learning_rate": 3.5190217391304346e-05,
"loss": 0.0001,
"step": 210
},
{
"epoch": 1.3504,
"learning_rate": 3.505434782608696e-05,
"loss": 0.0008,
"step": 211
},
{
"epoch": 1.3568,
"learning_rate": 3.491847826086957e-05,
"loss": 0.0008,
"step": 212
},
{
"epoch": 1.3632,
"learning_rate": 3.478260869565218e-05,
"loss": 0.0002,
"step": 213
},
{
"epoch": 1.3696,
"learning_rate": 3.464673913043479e-05,
"loss": 0.0012,
"step": 214
},
{
"epoch": 1.376,
"learning_rate": 3.451086956521739e-05,
"loss": 0.0001,
"step": 215
},
{
"epoch": 1.3824,
"learning_rate": 3.4375e-05,
"loss": 0.0006,
"step": 216
},
{
"epoch": 1.3888,
"learning_rate": 3.423913043478261e-05,
"loss": 0.0004,
"step": 217
},
{
"epoch": 1.3952,
"learning_rate": 3.4103260869565216e-05,
"loss": 0.0027,
"step": 218
},
{
"epoch": 1.4016,
"learning_rate": 3.3967391304347826e-05,
"loss": 0.0006,
"step": 219
},
{
"epoch": 1.408,
"learning_rate": 3.383152173913044e-05,
"loss": 0.0001,
"step": 220
},
{
"epoch": 1.4144,
"learning_rate": 3.369565217391305e-05,
"loss": 0.0005,
"step": 221
},
{
"epoch": 1.4208,
"learning_rate": 3.355978260869566e-05,
"loss": 0.0001,
"step": 222
},
{
"epoch": 1.4272,
"learning_rate": 3.342391304347826e-05,
"loss": 0.0001,
"step": 223
},
{
"epoch": 1.4336,
"learning_rate": 3.328804347826087e-05,
"loss": 0.0003,
"step": 224
},
{
"epoch": 1.44,
"learning_rate": 3.3152173913043475e-05,
"loss": 0.0005,
"step": 225
},
{
"epoch": 1.4464000000000001,
"learning_rate": 3.3016304347826086e-05,
"loss": 0.0015,
"step": 226
},
{
"epoch": 1.4527999999999999,
"learning_rate": 3.2880434782608696e-05,
"loss": 0.0003,
"step": 227
},
{
"epoch": 1.4592,
"learning_rate": 3.2744565217391307e-05,
"loss": 0.0006,
"step": 228
},
{
"epoch": 1.4656,
"learning_rate": 3.260869565217392e-05,
"loss": 0.0007,
"step": 229
},
{
"epoch": 1.472,
"learning_rate": 3.247282608695653e-05,
"loss": 0.0007,
"step": 230
},
{
"epoch": 1.4784,
"learning_rate": 3.233695652173913e-05,
"loss": 0.0004,
"step": 231
},
{
"epoch": 1.4848,
"learning_rate": 3.220108695652174e-05,
"loss": 0.0007,
"step": 232
},
{
"epoch": 1.4912,
"learning_rate": 3.2065217391304345e-05,
"loss": 0.0,
"step": 233
},
{
"epoch": 1.4976,
"learning_rate": 3.1929347826086955e-05,
"loss": 0.001,
"step": 234
},
{
"epoch": 1.504,
"learning_rate": 3.1793478260869566e-05,
"loss": 0.0002,
"step": 235
},
{
"epoch": 1.5104,
"learning_rate": 3.1657608695652176e-05,
"loss": 0.0008,
"step": 236
},
{
"epoch": 1.5168,
"learning_rate": 3.152173913043479e-05,
"loss": 0.0008,
"step": 237
},
{
"epoch": 1.5232,
"learning_rate": 3.138586956521739e-05,
"loss": 0.0001,
"step": 238
},
{
"epoch": 1.5295999999999998,
"learning_rate": 3.125e-05,
"loss": 0.0007,
"step": 239
},
{
"epoch": 1.536,
"learning_rate": 3.111413043478261e-05,
"loss": 0.0004,
"step": 240
},
{
"epoch": 1.5424,
"learning_rate": 3.0978260869565215e-05,
"loss": 0.0002,
"step": 241
},
{
"epoch": 1.5488,
"learning_rate": 3.084239130434783e-05,
"loss": 0.0004,
"step": 242
},
{
"epoch": 1.5552000000000001,
"learning_rate": 3.0706521739130435e-05,
"loss": 0.0,
"step": 243
},
{
"epoch": 1.5615999999999999,
"learning_rate": 3.0570652173913046e-05,
"loss": 0.0016,
"step": 244
},
{
"epoch": 1.568,
"learning_rate": 3.0434782608695656e-05,
"loss": 0.0002,
"step": 245
},
{
"epoch": 1.5744,
"learning_rate": 3.029891304347826e-05,
"loss": 0.0015,
"step": 246
},
{
"epoch": 1.5808,
"learning_rate": 3.016304347826087e-05,
"loss": 0.0001,
"step": 247
},
{
"epoch": 1.5872000000000002,
"learning_rate": 3.0027173913043477e-05,
"loss": 0.0003,
"step": 248
},
{
"epoch": 1.5936,
"learning_rate": 2.9891304347826088e-05,
"loss": 0.0001,
"step": 249
},
{
"epoch": 1.6,
"learning_rate": 2.9755434782608698e-05,
"loss": 0.0013,
"step": 250
},
{
"epoch": 1.6064,
"learning_rate": 2.9619565217391305e-05,
"loss": 0.0007,
"step": 251
},
{
"epoch": 1.6128,
"learning_rate": 2.9483695652173916e-05,
"loss": 0.0003,
"step": 252
},
{
"epoch": 1.6192,
"learning_rate": 2.9347826086956526e-05,
"loss": 0.0,
"step": 253
},
{
"epoch": 1.6256,
"learning_rate": 2.921195652173913e-05,
"loss": 0.0002,
"step": 254
},
{
"epoch": 1.6320000000000001,
"learning_rate": 2.9076086956521743e-05,
"loss": 0.0001,
"step": 255
},
{
"epoch": 1.6383999999999999,
"learning_rate": 2.8940217391304347e-05,
"loss": 0.0001,
"step": 256
},
{
"epoch": 1.6448,
"learning_rate": 2.8804347826086957e-05,
"loss": 0.0001,
"step": 257
},
{
"epoch": 1.6512,
"learning_rate": 2.8668478260869568e-05,
"loss": 0.0002,
"step": 258
},
{
"epoch": 1.6576,
"learning_rate": 2.8532608695652175e-05,
"loss": 0.0002,
"step": 259
},
{
"epoch": 1.6640000000000001,
"learning_rate": 2.8396739130434785e-05,
"loss": 0.0002,
"step": 260
},
{
"epoch": 1.6703999999999999,
"learning_rate": 2.826086956521739e-05,
"loss": 0.0008,
"step": 261
},
{
"epoch": 1.6768,
"learning_rate": 2.8125000000000003e-05,
"loss": 0.0001,
"step": 262
},
{
"epoch": 1.6832,
"learning_rate": 2.7989130434782613e-05,
"loss": 0.001,
"step": 263
},
{
"epoch": 1.6896,
"learning_rate": 2.7853260869565217e-05,
"loss": 0.0005,
"step": 264
},
{
"epoch": 1.696,
"learning_rate": 2.7717391304347827e-05,
"loss": 0.0002,
"step": 265
},
{
"epoch": 1.7024,
"learning_rate": 2.7581521739130434e-05,
"loss": 0.0002,
"step": 266
},
{
"epoch": 1.7088,
"learning_rate": 2.7445652173913044e-05,
"loss": 0.0003,
"step": 267
},
{
"epoch": 1.7151999999999998,
"learning_rate": 2.7309782608695655e-05,
"loss": 0.0,
"step": 268
},
{
"epoch": 1.7216,
"learning_rate": 2.7173913043478262e-05,
"loss": 0.0007,
"step": 269
},
{
"epoch": 1.728,
"learning_rate": 2.7038043478260872e-05,
"loss": 0.0001,
"step": 270
},
{
"epoch": 1.7344,
"learning_rate": 2.6902173913043476e-05,
"loss": 0.0,
"step": 271
},
{
"epoch": 1.7408000000000001,
"learning_rate": 2.6766304347826086e-05,
"loss": 0.0001,
"step": 272
},
{
"epoch": 1.7471999999999999,
"learning_rate": 2.66304347826087e-05,
"loss": 0.0002,
"step": 273
},
{
"epoch": 1.7536,
"learning_rate": 2.6494565217391304e-05,
"loss": 0.0001,
"step": 274
},
{
"epoch": 1.76,
"learning_rate": 2.6358695652173914e-05,
"loss": 0.0001,
"step": 275
},
{
"epoch": 1.7664,
"learning_rate": 2.6222826086956525e-05,
"loss": 0.0002,
"step": 276
},
{
"epoch": 1.7728000000000002,
"learning_rate": 2.608695652173913e-05,
"loss": 0.0001,
"step": 277
},
{
"epoch": 1.7792,
"learning_rate": 2.5951086956521742e-05,
"loss": 0.0032,
"step": 278
},
{
"epoch": 1.7856,
"learning_rate": 2.5815217391304346e-05,
"loss": 0.0002,
"step": 279
},
{
"epoch": 1.792,
"learning_rate": 2.567934782608696e-05,
"loss": 0.0012,
"step": 280
},
{
"epoch": 1.7984,
"learning_rate": 2.554347826086957e-05,
"loss": 0.0002,
"step": 281
},
{
"epoch": 1.8048,
"learning_rate": 2.5407608695652173e-05,
"loss": 0.0007,
"step": 282
},
{
"epoch": 1.8112,
"learning_rate": 2.5271739130434784e-05,
"loss": 0.0001,
"step": 283
},
{
"epoch": 1.8176,
"learning_rate": 2.513586956521739e-05,
"loss": 0.0014,
"step": 284
},
{
"epoch": 1.8239999999999998,
"learning_rate": 2.5e-05,
"loss": 0.0004,
"step": 285
},
{
"epoch": 1.8304,
"learning_rate": 2.4864130434782608e-05,
"loss": 0.0005,
"step": 286
},
{
"epoch": 1.8368,
"learning_rate": 2.472826086956522e-05,
"loss": 0.0001,
"step": 287
},
{
"epoch": 1.8432,
"learning_rate": 2.459239130434783e-05,
"loss": 0.0002,
"step": 288
},
{
"epoch": 1.8496000000000001,
"learning_rate": 2.4456521739130436e-05,
"loss": 0.0,
"step": 289
},
{
"epoch": 1.8559999999999999,
"learning_rate": 2.4320652173913043e-05,
"loss": 0.0,
"step": 290
},
{
"epoch": 1.8624,
"learning_rate": 2.4184782608695653e-05,
"loss": 0.0005,
"step": 291
},
{
"epoch": 1.8688,
"learning_rate": 2.4048913043478264e-05,
"loss": 0.0015,
"step": 292
},
{
"epoch": 1.8752,
"learning_rate": 2.391304347826087e-05,
"loss": 0.0002,
"step": 293
},
{
"epoch": 1.8816000000000002,
"learning_rate": 2.3777173913043478e-05,
"loss": 0.0001,
"step": 294
},
{
"epoch": 1.888,
"learning_rate": 2.3641304347826088e-05,
"loss": 0.0003,
"step": 295
},
{
"epoch": 1.8944,
"learning_rate": 2.3505434782608695e-05,
"loss": 0.0002,
"step": 296
},
{
"epoch": 1.9008,
"learning_rate": 2.3369565217391306e-05,
"loss": 0.0003,
"step": 297
},
{
"epoch": 1.9072,
"learning_rate": 2.3233695652173916e-05,
"loss": 0.0001,
"step": 298
},
{
"epoch": 1.9136,
"learning_rate": 2.3097826086956523e-05,
"loss": 0.0001,
"step": 299
},
{
"epoch": 1.92,
"learning_rate": 2.296195652173913e-05,
"loss": 0.0003,
"step": 300
},
{
"epoch": 1.9264000000000001,
"learning_rate": 2.282608695652174e-05,
"loss": 0.0001,
"step": 301
},
{
"epoch": 1.9327999999999999,
"learning_rate": 2.269021739130435e-05,
"loss": 0.0002,
"step": 302
},
{
"epoch": 1.9392,
"learning_rate": 2.2554347826086958e-05,
"loss": 0.0002,
"step": 303
},
{
"epoch": 1.9456,
"learning_rate": 2.2418478260869565e-05,
"loss": 0.0001,
"step": 304
},
{
"epoch": 1.952,
"learning_rate": 2.2282608695652175e-05,
"loss": 0.0,
"step": 305
},
{
"epoch": 1.9584000000000001,
"learning_rate": 2.2146739130434786e-05,
"loss": 0.0001,
"step": 306
},
{
"epoch": 1.9647999999999999,
"learning_rate": 2.2010869565217393e-05,
"loss": 0.0001,
"step": 307
},
{
"epoch": 1.9712,
"learning_rate": 2.1875e-05,
"loss": 0.0001,
"step": 308
},
{
"epoch": 1.9776,
"learning_rate": 2.173913043478261e-05,
"loss": 0.0,
"step": 309
},
{
"epoch": 1.984,
"learning_rate": 2.1603260869565217e-05,
"loss": 0.0001,
"step": 310
},
{
"epoch": 1.9904,
"learning_rate": 2.1467391304347828e-05,
"loss": 0.0,
"step": 311
},
{
"epoch": 1.9968,
"learning_rate": 2.1331521739130435e-05,
"loss": 0.0,
"step": 312
},
{
"epoch": 2.0032,
"learning_rate": 2.1195652173913045e-05,
"loss": 0.0,
"step": 313
},
{
"epoch": 2.0096,
"learning_rate": 2.1059782608695652e-05,
"loss": 0.0003,
"step": 314
},
{
"epoch": 2.016,
"learning_rate": 2.0923913043478262e-05,
"loss": 0.0001,
"step": 315
},
{
"epoch": 2.0224,
"learning_rate": 2.0788043478260873e-05,
"loss": 0.0,
"step": 316
},
{
"epoch": 2.0288,
"learning_rate": 2.065217391304348e-05,
"loss": 0.0,
"step": 317
},
{
"epoch": 2.0352,
"learning_rate": 2.0516304347826087e-05,
"loss": 0.0,
"step": 318
},
{
"epoch": 2.0416,
"learning_rate": 2.0380434782608694e-05,
"loss": 0.0,
"step": 319
},
{
"epoch": 2.048,
"learning_rate": 2.0244565217391308e-05,
"loss": 0.0,
"step": 320
},
{
"epoch": 2.0544,
"learning_rate": 2.0108695652173915e-05,
"loss": 0.0,
"step": 321
},
{
"epoch": 2.0608,
"learning_rate": 1.997282608695652e-05,
"loss": 0.0,
"step": 322
},
{
"epoch": 2.0672,
"learning_rate": 1.9836956521739132e-05,
"loss": 0.0,
"step": 323
},
{
"epoch": 2.0736,
"learning_rate": 1.970108695652174e-05,
"loss": 0.0,
"step": 324
},
{
"epoch": 2.08,
"learning_rate": 1.956521739130435e-05,
"loss": 0.0,
"step": 325
},
{
"epoch": 2.0864,
"learning_rate": 1.9429347826086957e-05,
"loss": 0.0,
"step": 326
},
{
"epoch": 2.0928,
"learning_rate": 1.9293478260869567e-05,
"loss": 0.0,
"step": 327
},
{
"epoch": 2.0992,
"learning_rate": 1.9157608695652174e-05,
"loss": 0.0,
"step": 328
},
{
"epoch": 2.1056,
"learning_rate": 1.9021739130434784e-05,
"loss": 0.0001,
"step": 329
},
{
"epoch": 2.112,
"learning_rate": 1.888586956521739e-05,
"loss": 0.0002,
"step": 330
},
{
"epoch": 2.1184,
"learning_rate": 1.8750000000000002e-05,
"loss": 0.0,
"step": 331
},
{
"epoch": 2.1248,
"learning_rate": 1.861413043478261e-05,
"loss": 0.0,
"step": 332
},
{
"epoch": 2.1312,
"learning_rate": 1.8478260869565216e-05,
"loss": 0.0,
"step": 333
},
{
"epoch": 2.1376,
"learning_rate": 1.834239130434783e-05,
"loss": 0.0,
"step": 334
},
{
"epoch": 2.144,
"learning_rate": 1.8206521739130437e-05,
"loss": 0.0,
"step": 335
},
{
"epoch": 2.1504,
"learning_rate": 1.8070652173913044e-05,
"loss": 0.0,
"step": 336
},
{
"epoch": 2.1568,
"learning_rate": 1.793478260869565e-05,
"loss": 0.0,
"step": 337
},
{
"epoch": 2.1632,
"learning_rate": 1.7798913043478264e-05,
"loss": 0.0,
"step": 338
},
{
"epoch": 2.1696,
"learning_rate": 1.766304347826087e-05,
"loss": 0.0,
"step": 339
},
{
"epoch": 2.176,
"learning_rate": 1.752717391304348e-05,
"loss": 0.0,
"step": 340
},
{
"epoch": 2.1824,
"learning_rate": 1.739130434782609e-05,
"loss": 0.0,
"step": 341
},
{
"epoch": 2.1888,
"learning_rate": 1.7255434782608696e-05,
"loss": 0.0,
"step": 342
},
{
"epoch": 2.1952,
"learning_rate": 1.7119565217391306e-05,
"loss": 0.0,
"step": 343
},
{
"epoch": 2.2016,
"learning_rate": 1.6983695652173913e-05,
"loss": 0.0,
"step": 344
},
{
"epoch": 2.208,
"learning_rate": 1.6847826086956524e-05,
"loss": 0.0,
"step": 345
},
{
"epoch": 2.2144,
"learning_rate": 1.671195652173913e-05,
"loss": 0.0,
"step": 346
},
{
"epoch": 2.2208,
"learning_rate": 1.6576086956521738e-05,
"loss": 0.0,
"step": 347
},
{
"epoch": 2.2272,
"learning_rate": 1.6440217391304348e-05,
"loss": 0.0,
"step": 348
},
{
"epoch": 2.2336,
"learning_rate": 1.630434782608696e-05,
"loss": 0.0003,
"step": 349
},
{
"epoch": 2.24,
"learning_rate": 1.6168478260869565e-05,
"loss": 0.0,
"step": 350
},
{
"epoch": 2.2464,
"learning_rate": 1.6032608695652173e-05,
"loss": 0.0,
"step": 351
},
{
"epoch": 2.2528,
"learning_rate": 1.5896739130434783e-05,
"loss": 0.0002,
"step": 352
},
{
"epoch": 2.2592,
"learning_rate": 1.5760869565217393e-05,
"loss": 0.0,
"step": 353
},
{
"epoch": 2.2656,
"learning_rate": 1.5625e-05,
"loss": 0.0001,
"step": 354
},
{
"epoch": 2.2720000000000002,
"learning_rate": 1.5489130434782607e-05,
"loss": 0.0,
"step": 355
},
{
"epoch": 2.2784,
"learning_rate": 1.5353260869565218e-05,
"loss": 0.0,
"step": 356
},
{
"epoch": 2.2848,
"learning_rate": 1.5217391304347828e-05,
"loss": 0.0,
"step": 357
},
{
"epoch": 2.2912,
"learning_rate": 1.5081521739130435e-05,
"loss": 0.0,
"step": 358
},
{
"epoch": 2.2976,
"learning_rate": 1.4945652173913044e-05,
"loss": 0.0,
"step": 359
},
{
"epoch": 2.304,
"learning_rate": 1.4809782608695653e-05,
"loss": 0.0,
"step": 360
},
{
"epoch": 2.3104,
"learning_rate": 1.4673913043478263e-05,
"loss": 0.0,
"step": 361
},
{
"epoch": 2.3168,
"learning_rate": 1.4538043478260872e-05,
"loss": 0.0,
"step": 362
},
{
"epoch": 2.3232,
"learning_rate": 1.4402173913043479e-05,
"loss": 0.0,
"step": 363
},
{
"epoch": 2.3296,
"learning_rate": 1.4266304347826087e-05,
"loss": 0.0,
"step": 364
},
{
"epoch": 2.336,
"learning_rate": 1.4130434782608694e-05,
"loss": 0.0,
"step": 365
},
{
"epoch": 2.3424,
"learning_rate": 1.3994565217391307e-05,
"loss": 0.0,
"step": 366
},
{
"epoch": 2.3487999999999998,
"learning_rate": 1.3858695652173914e-05,
"loss": 0.0,
"step": 367
},
{
"epoch": 2.3552,
"learning_rate": 1.3722826086956522e-05,
"loss": 0.0,
"step": 368
},
{
"epoch": 2.3616,
"learning_rate": 1.3586956521739131e-05,
"loss": 0.0,
"step": 369
},
{
"epoch": 2.368,
"learning_rate": 1.3451086956521738e-05,
"loss": 0.0,
"step": 370
},
{
"epoch": 2.3744,
"learning_rate": 1.331521739130435e-05,
"loss": 0.0,
"step": 371
},
{
"epoch": 2.3808,
"learning_rate": 1.3179347826086957e-05,
"loss": 0.0,
"step": 372
},
{
"epoch": 2.3872,
"learning_rate": 1.3043478260869566e-05,
"loss": 0.0,
"step": 373
},
{
"epoch": 2.3936,
"learning_rate": 1.2907608695652173e-05,
"loss": 0.0,
"step": 374
},
{
"epoch": 2.4,
"learning_rate": 1.2771739130434785e-05,
"loss": 0.0,
"step": 375
},
{
"epoch": 2.4064,
"learning_rate": 1.2635869565217392e-05,
"loss": 0.0,
"step": 376
},
{
"epoch": 2.4128,
"learning_rate": 1.25e-05,
"loss": 0.0,
"step": 377
},
{
"epoch": 2.4192,
"learning_rate": 1.236413043478261e-05,
"loss": 0.0,
"step": 378
},
{
"epoch": 2.4256,
"learning_rate": 1.2228260869565218e-05,
"loss": 0.0013,
"step": 379
},
{
"epoch": 2.432,
"learning_rate": 1.2092391304347827e-05,
"loss": 0.0,
"step": 380
},
{
"epoch": 2.4384,
"learning_rate": 1.1956521739130435e-05,
"loss": 0.0,
"step": 381
},
{
"epoch": 2.4448,
"learning_rate": 1.1820652173913044e-05,
"loss": 0.0,
"step": 382
},
{
"epoch": 2.4512,
"learning_rate": 1.1684782608695653e-05,
"loss": 0.0,
"step": 383
},
{
"epoch": 2.4576000000000002,
"learning_rate": 1.1548913043478262e-05,
"loss": 0.0,
"step": 384
},
{
"epoch": 2.464,
"learning_rate": 1.141304347826087e-05,
"loss": 0.0,
"step": 385
},
{
"epoch": 2.4704,
"learning_rate": 1.1277173913043479e-05,
"loss": 0.0,
"step": 386
},
{
"epoch": 2.4768,
"learning_rate": 1.1141304347826088e-05,
"loss": 0.0,
"step": 387
},
{
"epoch": 2.4832,
"learning_rate": 1.1005434782608696e-05,
"loss": 0.0,
"step": 388
},
{
"epoch": 2.4896,
"learning_rate": 1.0869565217391305e-05,
"loss": 0.0,
"step": 389
},
{
"epoch": 2.496,
"learning_rate": 1.0733695652173914e-05,
"loss": 0.0,
"step": 390
},
{
"epoch": 2.5023999999999997,
"learning_rate": 1.0597826086956523e-05,
"loss": 0.0,
"step": 391
},
{
"epoch": 2.5088,
"learning_rate": 1.0461956521739131e-05,
"loss": 0.0,
"step": 392
},
{
"epoch": 2.5152,
"learning_rate": 1.032608695652174e-05,
"loss": 0.0,
"step": 393
},
{
"epoch": 2.5216,
"learning_rate": 1.0190217391304347e-05,
"loss": 0.0,
"step": 394
},
{
"epoch": 2.528,
"learning_rate": 1.0054347826086957e-05,
"loss": 0.0,
"step": 395
},
{
"epoch": 2.5343999999999998,
"learning_rate": 9.918478260869566e-06,
"loss": 0.0008,
"step": 396
},
{
"epoch": 2.5408,
"learning_rate": 9.782608695652175e-06,
"loss": 0.0,
"step": 397
},
{
"epoch": 2.5472,
"learning_rate": 9.646739130434783e-06,
"loss": 0.0,
"step": 398
},
{
"epoch": 2.5536,
"learning_rate": 9.510869565217392e-06,
"loss": 0.0,
"step": 399
},
{
"epoch": 2.56,
"learning_rate": 9.375000000000001e-06,
"loss": 0.0,
"step": 400
},
{
"epoch": 2.5664,
"learning_rate": 9.239130434782608e-06,
"loss": 0.0,
"step": 401
},
{
"epoch": 2.5728,
"learning_rate": 9.103260869565218e-06,
"loss": 0.0,
"step": 402
},
{
"epoch": 2.5792,
"learning_rate": 8.967391304347825e-06,
"loss": 0.0,
"step": 403
},
{
"epoch": 2.5856,
"learning_rate": 8.831521739130436e-06,
"loss": 0.0,
"step": 404
},
{
"epoch": 2.592,
"learning_rate": 8.695652173913044e-06,
"loss": 0.0,
"step": 405
},
{
"epoch": 2.5984,
"learning_rate": 8.559782608695653e-06,
"loss": 0.0,
"step": 406
},
{
"epoch": 2.6048,
"learning_rate": 8.423913043478262e-06,
"loss": 0.0,
"step": 407
},
{
"epoch": 2.6112,
"learning_rate": 8.288043478260869e-06,
"loss": 0.0,
"step": 408
},
{
"epoch": 2.6176,
"learning_rate": 8.15217391304348e-06,
"loss": 0.0,
"step": 409
},
{
"epoch": 2.624,
"learning_rate": 8.016304347826086e-06,
"loss": 0.0001,
"step": 410
},
{
"epoch": 2.6304,
"learning_rate": 7.880434782608697e-06,
"loss": 0.0,
"step": 411
},
{
"epoch": 2.6368,
"learning_rate": 7.744565217391304e-06,
"loss": 0.0,
"step": 412
},
{
"epoch": 2.6432,
"learning_rate": 7.608695652173914e-06,
"loss": 0.0,
"step": 413
},
{
"epoch": 2.6496,
"learning_rate": 7.472826086956522e-06,
"loss": 0.0,
"step": 414
},
{
"epoch": 2.656,
"learning_rate": 7.3369565217391315e-06,
"loss": 0.0,
"step": 415
},
{
"epoch": 2.6624,
"learning_rate": 7.201086956521739e-06,
"loss": 0.0,
"step": 416
},
{
"epoch": 2.6688,
"learning_rate": 7.065217391304347e-06,
"loss": 0.0,
"step": 417
},
{
"epoch": 2.6752000000000002,
"learning_rate": 6.929347826086957e-06,
"loss": 0.0002,
"step": 418
},
{
"epoch": 2.6816,
"learning_rate": 6.7934782608695655e-06,
"loss": 0.0,
"step": 419
},
{
"epoch": 2.6879999999999997,
"learning_rate": 6.657608695652175e-06,
"loss": 0.0,
"step": 420
},
{
"epoch": 2.6944,
"learning_rate": 6.521739130434783e-06,
"loss": 0.0,
"step": 421
},
{
"epoch": 2.7008,
"learning_rate": 6.3858695652173924e-06,
"loss": 0.0005,
"step": 422
},
{
"epoch": 2.7072000000000003,
"learning_rate": 6.25e-06,
"loss": 0.0,
"step": 423
},
{
"epoch": 2.7136,
"learning_rate": 6.114130434782609e-06,
"loss": 0.0,
"step": 424
},
{
"epoch": 2.7199999999999998,
"learning_rate": 5.978260869565218e-06,
"loss": 0.0001,
"step": 425
},
{
"epoch": 2.7264,
"learning_rate": 5.842391304347826e-06,
"loss": 0.0,
"step": 426
},
{
"epoch": 2.7328,
"learning_rate": 5.706521739130435e-06,
"loss": 0.0,
"step": 427
},
{
"epoch": 2.7392,
"learning_rate": 5.570652173913044e-06,
"loss": 0.0,
"step": 428
},
{
"epoch": 2.7456,
"learning_rate": 5.4347826086956525e-06,
"loss": 0.0,
"step": 429
},
{
"epoch": 2.752,
"learning_rate": 5.298913043478261e-06,
"loss": 0.0,
"step": 430
},
{
"epoch": 2.7584,
"learning_rate": 5.16304347826087e-06,
"loss": 0.0,
"step": 431
},
{
"epoch": 2.7648,
"learning_rate": 5.027173913043479e-06,
"loss": 0.0,
"step": 432
},
{
"epoch": 2.7712,
"learning_rate": 4.891304347826087e-06,
"loss": 0.0,
"step": 433
},
{
"epoch": 2.7776,
"learning_rate": 4.755434782608696e-06,
"loss": 0.0,
"step": 434
},
{
"epoch": 2.784,
"learning_rate": 4.619565217391304e-06,
"loss": 0.0,
"step": 435
},
{
"epoch": 2.7904,
"learning_rate": 4.483695652173913e-06,
"loss": 0.0,
"step": 436
},
{
"epoch": 2.7968,
"learning_rate": 4.347826086956522e-06,
"loss": 0.0,
"step": 437
},
{
"epoch": 2.8032,
"learning_rate": 4.211956521739131e-06,
"loss": 0.0,
"step": 438
},
{
"epoch": 2.8096,
"learning_rate": 4.07608695652174e-06,
"loss": 0.0,
"step": 439
},
{
"epoch": 2.816,
"learning_rate": 3.940217391304348e-06,
"loss": 0.0,
"step": 440
},
{
"epoch": 2.8224,
"learning_rate": 3.804347826086957e-06,
"loss": 0.0,
"step": 441
},
{
"epoch": 2.8288,
"learning_rate": 3.6684782608695657e-06,
"loss": 0.0,
"step": 442
},
{
"epoch": 2.8352,
"learning_rate": 3.5326086956521736e-06,
"loss": 0.0002,
"step": 443
},
{
"epoch": 2.8416,
"learning_rate": 3.3967391304347827e-06,
"loss": 0.0007,
"step": 444
},
{
"epoch": 2.848,
"learning_rate": 3.2608695652173914e-06,
"loss": 0.0,
"step": 445
},
{
"epoch": 2.8544,
"learning_rate": 3.125e-06,
"loss": 0.0002,
"step": 446
},
{
"epoch": 2.8608000000000002,
"learning_rate": 2.989130434782609e-06,
"loss": 0.0,
"step": 447
},
{
"epoch": 2.8672,
"learning_rate": 2.8532608695652176e-06,
"loss": 0.0001,
"step": 448
},
{
"epoch": 2.8736,
"learning_rate": 2.7173913043478263e-06,
"loss": 0.0,
"step": 449
},
{
"epoch": 2.88,
"learning_rate": 2.581521739130435e-06,
"loss": 0.0,
"step": 450
},
{
"epoch": 2.8864,
"learning_rate": 2.4456521739130437e-06,
"loss": 0.0,
"step": 451
},
{
"epoch": 2.8928000000000003,
"learning_rate": 2.309782608695652e-06,
"loss": 0.0,
"step": 452
},
{
"epoch": 2.8992,
"learning_rate": 2.173913043478261e-06,
"loss": 0.0,
"step": 453
},
{
"epoch": 2.9055999999999997,
"learning_rate": 2.03804347826087e-06,
"loss": 0.0,
"step": 454
},
{
"epoch": 2.912,
"learning_rate": 1.9021739130434785e-06,
"loss": 0.0,
"step": 455
},
{
"epoch": 2.9184,
"learning_rate": 1.7663043478260868e-06,
"loss": 0.0,
"step": 456
},
{
"epoch": 2.9248,
"learning_rate": 1.6304347826086957e-06,
"loss": 0.0,
"step": 457
},
{
"epoch": 2.9312,
"learning_rate": 1.4945652173913044e-06,
"loss": 0.0,
"step": 458
},
{
"epoch": 2.9375999999999998,
"learning_rate": 1.3586956521739131e-06,
"loss": 0.0001,
"step": 459
},
{
"epoch": 2.944,
"learning_rate": 1.2228260869565218e-06,
"loss": 0.0,
"step": 460
},
{
"epoch": 2.9504,
"learning_rate": 1.0869565217391306e-06,
"loss": 0.0,
"step": 461
},
{
"epoch": 2.9568,
"learning_rate": 9.510869565217393e-07,
"loss": 0.0,
"step": 462
},
{
"epoch": 2.9632,
"learning_rate": 8.152173913043479e-07,
"loss": 0.0,
"step": 463
},
{
"epoch": 2.9696,
"learning_rate": 6.793478260869566e-07,
"loss": 0.0,
"step": 464
},
{
"epoch": 2.976,
"learning_rate": 5.434782608695653e-07,
"loss": 0.0,
"step": 465
},
{
"epoch": 2.9824,
"learning_rate": 4.0760869565217393e-07,
"loss": 0.0,
"step": 466
},
{
"epoch": 2.9888,
"learning_rate": 2.7173913043478264e-07,
"loss": 0.0,
"step": 467
},
{
"epoch": 2.9952,
"learning_rate": 1.3586956521739132e-07,
"loss": 0.0,
"step": 468
},
{
"epoch": 2.9952,
"step": 468,
"total_flos": 21071443283968.0,
"train_loss": 0.011422658276770508,
"train_runtime": 42960.2523,
"train_samples_per_second": 0.349,
"train_steps_per_second": 0.011
}
],
"logging_steps": 1.0,
"max_steps": 468,
"num_input_tokens_seen": 0,
"num_train_epochs": 3,
"save_steps": 100,
"total_flos": 21071443283968.0,
"train_batch_size": 1,
"trial_name": null,
"trial_params": null
}