Commit
•
bed9e73
1
Parent(s):
fa194d4
Upload trainer_state.json with huggingface_hub
Browse files- trainer_state.json +253 -0
trainer_state.json
ADDED
|
@@ -0,0 +1,253 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 18.06451612903226,
|
| 5 |
+
"global_step": 140,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.9,
|
| 12 |
+
"eval_accuracy": 0.7506493506493507,
|
| 13 |
+
"eval_f1": 0.0,
|
| 14 |
+
"eval_loss": 1.2528471946716309,
|
| 15 |
+
"eval_precision": 0.0,
|
| 16 |
+
"eval_recall": 0.0,
|
| 17 |
+
"eval_runtime": 0.2089,
|
| 18 |
+
"eval_samples_per_second": 38.296,
|
| 19 |
+
"eval_steps_per_second": 4.787,
|
| 20 |
+
"step": 7
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 1.94,
|
| 24 |
+
"eval_accuracy": 0.7688311688311689,
|
| 25 |
+
"eval_f1": 0.04819277108433734,
|
| 26 |
+
"eval_loss": 0.8142604827880859,
|
| 27 |
+
"eval_precision": 0.18181818181818182,
|
| 28 |
+
"eval_recall": 0.027777777777777776,
|
| 29 |
+
"eval_runtime": 0.1224,
|
| 30 |
+
"eval_samples_per_second": 65.373,
|
| 31 |
+
"eval_steps_per_second": 8.172,
|
| 32 |
+
"step": 15
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 2.97,
|
| 36 |
+
"eval_accuracy": 0.8701298701298701,
|
| 37 |
+
"eval_f1": 0.4864864864864865,
|
| 38 |
+
"eval_loss": 0.5706822276115417,
|
| 39 |
+
"eval_precision": 0.47368421052631576,
|
| 40 |
+
"eval_recall": 0.5,
|
| 41 |
+
"eval_runtime": 0.1206,
|
| 42 |
+
"eval_samples_per_second": 66.316,
|
| 43 |
+
"eval_steps_per_second": 8.289,
|
| 44 |
+
"step": 23
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 4.0,
|
| 48 |
+
"eval_accuracy": 0.8909090909090909,
|
| 49 |
+
"eval_f1": 0.5379310344827586,
|
| 50 |
+
"eval_loss": 0.4967877268791199,
|
| 51 |
+
"eval_precision": 0.5342465753424658,
|
| 52 |
+
"eval_recall": 0.5416666666666666,
|
| 53 |
+
"eval_runtime": 0.121,
|
| 54 |
+
"eval_samples_per_second": 66.117,
|
| 55 |
+
"eval_steps_per_second": 8.265,
|
| 56 |
+
"step": 31
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 4.9,
|
| 60 |
+
"eval_accuracy": 0.9025974025974026,
|
| 61 |
+
"eval_f1": 0.5419354838709678,
|
| 62 |
+
"eval_loss": 0.4114609658718109,
|
| 63 |
+
"eval_precision": 0.5060240963855421,
|
| 64 |
+
"eval_recall": 0.5833333333333334,
|
| 65 |
+
"eval_runtime": 0.1266,
|
| 66 |
+
"eval_samples_per_second": 63.18,
|
| 67 |
+
"eval_steps_per_second": 7.897,
|
| 68 |
+
"step": 38
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 5.94,
|
| 72 |
+
"eval_accuracy": 0.9051948051948052,
|
| 73 |
+
"eval_f1": 0.5806451612903225,
|
| 74 |
+
"eval_loss": 0.37950557470321655,
|
| 75 |
+
"eval_precision": 0.5421686746987951,
|
| 76 |
+
"eval_recall": 0.625,
|
| 77 |
+
"eval_runtime": 0.1236,
|
| 78 |
+
"eval_samples_per_second": 64.708,
|
| 79 |
+
"eval_steps_per_second": 8.088,
|
| 80 |
+
"step": 46
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 6.97,
|
| 84 |
+
"eval_accuracy": 0.9077922077922078,
|
| 85 |
+
"eval_f1": 0.5359477124183006,
|
| 86 |
+
"eval_loss": 0.3877827525138855,
|
| 87 |
+
"eval_precision": 0.5061728395061729,
|
| 88 |
+
"eval_recall": 0.5694444444444444,
|
| 89 |
+
"eval_runtime": 0.1224,
|
| 90 |
+
"eval_samples_per_second": 65.385,
|
| 91 |
+
"eval_steps_per_second": 8.173,
|
| 92 |
+
"step": 54
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 8.0,
|
| 96 |
+
"eval_accuracy": 0.912987012987013,
|
| 97 |
+
"eval_f1": 0.6064516129032258,
|
| 98 |
+
"eval_loss": 0.4006221294403076,
|
| 99 |
+
"eval_precision": 0.5662650602409639,
|
| 100 |
+
"eval_recall": 0.6527777777777778,
|
| 101 |
+
"eval_runtime": 0.1196,
|
| 102 |
+
"eval_samples_per_second": 66.868,
|
| 103 |
+
"eval_steps_per_second": 8.358,
|
| 104 |
+
"step": 62
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 8.9,
|
| 108 |
+
"eval_accuracy": 0.9116883116883117,
|
| 109 |
+
"eval_f1": 0.6103896103896104,
|
| 110 |
+
"eval_loss": 0.3754499554634094,
|
| 111 |
+
"eval_precision": 0.573170731707317,
|
| 112 |
+
"eval_recall": 0.6527777777777778,
|
| 113 |
+
"eval_runtime": 0.124,
|
| 114 |
+
"eval_samples_per_second": 64.536,
|
| 115 |
+
"eval_steps_per_second": 8.067,
|
| 116 |
+
"step": 69
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 9.94,
|
| 120 |
+
"eval_accuracy": 0.9142857142857143,
|
| 121 |
+
"eval_f1": 0.5987261146496815,
|
| 122 |
+
"eval_loss": 0.3780848979949951,
|
| 123 |
+
"eval_precision": 0.5529411764705883,
|
| 124 |
+
"eval_recall": 0.6527777777777778,
|
| 125 |
+
"eval_runtime": 0.1879,
|
| 126 |
+
"eval_samples_per_second": 42.581,
|
| 127 |
+
"eval_steps_per_second": 5.323,
|
| 128 |
+
"step": 77
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 10.97,
|
| 132 |
+
"eval_accuracy": 0.912987012987013,
|
| 133 |
+
"eval_f1": 0.6075949367088608,
|
| 134 |
+
"eval_loss": 0.395553857088089,
|
| 135 |
+
"eval_precision": 0.5581395348837209,
|
| 136 |
+
"eval_recall": 0.6666666666666666,
|
| 137 |
+
"eval_runtime": 0.1224,
|
| 138 |
+
"eval_samples_per_second": 65.381,
|
| 139 |
+
"eval_steps_per_second": 8.173,
|
| 140 |
+
"step": 85
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 12.0,
|
| 144 |
+
"eval_accuracy": 0.9155844155844156,
|
| 145 |
+
"eval_f1": 0.6075949367088608,
|
| 146 |
+
"eval_loss": 0.387939453125,
|
| 147 |
+
"eval_precision": 0.5581395348837209,
|
| 148 |
+
"eval_recall": 0.6666666666666666,
|
| 149 |
+
"eval_runtime": 0.3171,
|
| 150 |
+
"eval_samples_per_second": 25.232,
|
| 151 |
+
"eval_steps_per_second": 3.154,
|
| 152 |
+
"step": 93
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 12.9,
|
| 156 |
+
"eval_accuracy": 0.9142857142857143,
|
| 157 |
+
"eval_f1": 0.6134969325153374,
|
| 158 |
+
"eval_loss": 0.3951228857040405,
|
| 159 |
+
"eval_precision": 0.5494505494505495,
|
| 160 |
+
"eval_recall": 0.6944444444444444,
|
| 161 |
+
"eval_runtime": 0.452,
|
| 162 |
+
"eval_samples_per_second": 17.699,
|
| 163 |
+
"eval_steps_per_second": 2.212,
|
| 164 |
+
"step": 100
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 13.94,
|
| 168 |
+
"eval_accuracy": 0.9168831168831169,
|
| 169 |
+
"eval_f1": 0.6211180124223603,
|
| 170 |
+
"eval_loss": 0.39693745970726013,
|
| 171 |
+
"eval_precision": 0.5617977528089888,
|
| 172 |
+
"eval_recall": 0.6944444444444444,
|
| 173 |
+
"eval_runtime": 0.431,
|
| 174 |
+
"eval_samples_per_second": 18.56,
|
| 175 |
+
"eval_steps_per_second": 2.32,
|
| 176 |
+
"step": 108
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 14.97,
|
| 180 |
+
"eval_accuracy": 0.912987012987013,
|
| 181 |
+
"eval_f1": 0.5962732919254659,
|
| 182 |
+
"eval_loss": 0.3942334055900574,
|
| 183 |
+
"eval_precision": 0.5393258426966292,
|
| 184 |
+
"eval_recall": 0.6666666666666666,
|
| 185 |
+
"eval_runtime": 0.1222,
|
| 186 |
+
"eval_samples_per_second": 65.44,
|
| 187 |
+
"eval_steps_per_second": 8.18,
|
| 188 |
+
"step": 116
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 16.0,
|
| 192 |
+
"eval_accuracy": 0.9155844155844156,
|
| 193 |
+
"eval_f1": 0.608695652173913,
|
| 194 |
+
"eval_loss": 0.3981378674507141,
|
| 195 |
+
"eval_precision": 0.550561797752809,
|
| 196 |
+
"eval_recall": 0.6805555555555556,
|
| 197 |
+
"eval_runtime": 0.5596,
|
| 198 |
+
"eval_samples_per_second": 14.297,
|
| 199 |
+
"eval_steps_per_second": 1.787,
|
| 200 |
+
"step": 124
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 16.9,
|
| 204 |
+
"eval_accuracy": 0.9168831168831169,
|
| 205 |
+
"eval_f1": 0.625,
|
| 206 |
+
"eval_loss": 0.39964303374290466,
|
| 207 |
+
"eval_precision": 0.5681818181818182,
|
| 208 |
+
"eval_recall": 0.6944444444444444,
|
| 209 |
+
"eval_runtime": 0.1224,
|
| 210 |
+
"eval_samples_per_second": 65.348,
|
| 211 |
+
"eval_steps_per_second": 8.169,
|
| 212 |
+
"step": 131
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 17.94,
|
| 216 |
+
"eval_accuracy": 0.9168831168831169,
|
| 217 |
+
"eval_f1": 0.625,
|
| 218 |
+
"eval_loss": 0.39898383617401123,
|
| 219 |
+
"eval_precision": 0.5681818181818182,
|
| 220 |
+
"eval_recall": 0.6944444444444444,
|
| 221 |
+
"eval_runtime": 0.1264,
|
| 222 |
+
"eval_samples_per_second": 63.289,
|
| 223 |
+
"eval_steps_per_second": 7.911,
|
| 224 |
+
"step": 139
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 18.06,
|
| 228 |
+
"eval_accuracy": 0.9168831168831169,
|
| 229 |
+
"eval_f1": 0.625,
|
| 230 |
+
"eval_loss": 0.39904916286468506,
|
| 231 |
+
"eval_precision": 0.5681818181818182,
|
| 232 |
+
"eval_recall": 0.6944444444444444,
|
| 233 |
+
"eval_runtime": 0.1186,
|
| 234 |
+
"eval_samples_per_second": 67.465,
|
| 235 |
+
"eval_steps_per_second": 8.433,
|
| 236 |
+
"step": 140
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 18.06,
|
| 240 |
+
"step": 140,
|
| 241 |
+
"total_flos": 507367899021024.0,
|
| 242 |
+
"train_loss": 0.25862868172781805,
|
| 243 |
+
"train_runtime": 1414.9426,
|
| 244 |
+
"train_samples_per_second": 1.71,
|
| 245 |
+
"train_steps_per_second": 0.099
|
| 246 |
+
}
|
| 247 |
+
],
|
| 248 |
+
"max_steps": 140,
|
| 249 |
+
"num_train_epochs": 20,
|
| 250 |
+
"total_flos": 507367899021024.0,
|
| 251 |
+
"trial_name": null,
|
| 252 |
+
"trial_params": null
|
| 253 |
+
}
|