Upload trainer_state.json with huggingface_hub
Browse files- trainer_state.json +349 -0
trainer_state.json
ADDED
|
@@ -0,0 +1,349 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": 3132,
|
| 3 |
+
"best_metric": 0.5881329726800972,
|
| 4 |
+
"best_model_checkpoint": "outputs/phobert-v1/checkpoint-3132",
|
| 5 |
+
"epoch": 18.0,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 3132,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.9942528735632183,
|
| 14 |
+
"grad_norm": 2.6583092212677,
|
| 15 |
+
"learning_rate": 6.88e-06,
|
| 16 |
+
"loss": 1.8469,
|
| 17 |
+
"step": 173
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 1.0,
|
| 21 |
+
"eval_accuracy": 0.3163265306122449,
|
| 22 |
+
"eval_loss": 1.7160369157791138,
|
| 23 |
+
"eval_macro_f1": 0.07588256112515117,
|
| 24 |
+
"eval_runtime": 2.3805,
|
| 25 |
+
"eval_samples_per_second": 288.175,
|
| 26 |
+
"eval_steps_per_second": 9.242,
|
| 27 |
+
"step": 174
|
| 28 |
+
},
|
| 29 |
+
{
|
| 30 |
+
"epoch": 1.9885057471264367,
|
| 31 |
+
"grad_norm": 9.339335441589355,
|
| 32 |
+
"learning_rate": 1.38e-05,
|
| 33 |
+
"loss": 1.6506,
|
| 34 |
+
"step": 346
|
| 35 |
+
},
|
| 36 |
+
{
|
| 37 |
+
"epoch": 2.0,
|
| 38 |
+
"eval_accuracy": 0.49271137026239065,
|
| 39 |
+
"eval_loss": 1.4279717206954956,
|
| 40 |
+
"eval_macro_f1": 0.26791037751403135,
|
| 41 |
+
"eval_runtime": 2.3818,
|
| 42 |
+
"eval_samples_per_second": 288.022,
|
| 43 |
+
"eval_steps_per_second": 9.237,
|
| 44 |
+
"step": 348
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 2.9827586206896552,
|
| 48 |
+
"grad_norm": 9.079231262207031,
|
| 49 |
+
"learning_rate": 1.999994401896218e-05,
|
| 50 |
+
"loss": 1.3315,
|
| 51 |
+
"step": 519
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 3.0,
|
| 55 |
+
"eval_accuracy": 0.5524781341107872,
|
| 56 |
+
"eval_loss": 1.207149624824524,
|
| 57 |
+
"eval_macro_f1": 0.401013356395679,
|
| 58 |
+
"eval_runtime": 2.383,
|
| 59 |
+
"eval_samples_per_second": 287.878,
|
| 60 |
+
"eval_steps_per_second": 9.232,
|
| 61 |
+
"step": 522
|
| 62 |
+
},
|
| 63 |
+
{
|
| 64 |
+
"epoch": 3.9770114942528734,
|
| 65 |
+
"grad_norm": 8.046290397644043,
|
| 66 |
+
"learning_rate": 1.9993697433304146e-05,
|
| 67 |
+
"loss": 1.0349,
|
| 68 |
+
"step": 692
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 4.0,
|
| 72 |
+
"eval_accuracy": 0.5845481049562682,
|
| 73 |
+
"eval_loss": 1.1730802059173584,
|
| 74 |
+
"eval_macro_f1": 0.5273093498298432,
|
| 75 |
+
"eval_runtime": 1.051,
|
| 76 |
+
"eval_samples_per_second": 652.7,
|
| 77 |
+
"eval_steps_per_second": 20.932,
|
| 78 |
+
"step": 696
|
| 79 |
+
},
|
| 80 |
+
{
|
| 81 |
+
"epoch": 4.971264367816092,
|
| 82 |
+
"grad_norm": 14.620841979980469,
|
| 83 |
+
"learning_rate": 1.9977115923137912e-05,
|
| 84 |
+
"loss": 0.8082,
|
| 85 |
+
"step": 865
|
| 86 |
+
},
|
| 87 |
+
{
|
| 88 |
+
"epoch": 5.0,
|
| 89 |
+
"eval_accuracy": 0.6064139941690962,
|
| 90 |
+
"eval_loss": 1.1103960275650024,
|
| 91 |
+
"eval_macro_f1": 0.5513612516051968,
|
| 92 |
+
"eval_runtime": 1.0497,
|
| 93 |
+
"eval_samples_per_second": 653.548,
|
| 94 |
+
"eval_steps_per_second": 20.959,
|
| 95 |
+
"step": 870
|
| 96 |
+
},
|
| 97 |
+
{
|
| 98 |
+
"epoch": 5.9655172413793105,
|
| 99 |
+
"grad_norm": 18.373315811157227,
|
| 100 |
+
"learning_rate": 1.995021663613649e-05,
|
| 101 |
+
"loss": 0.6432,
|
| 102 |
+
"step": 1038
|
| 103 |
+
},
|
| 104 |
+
{
|
| 105 |
+
"epoch": 6.0,
|
| 106 |
+
"eval_accuracy": 0.6180758017492711,
|
| 107 |
+
"eval_loss": 1.1242892742156982,
|
| 108 |
+
"eval_macro_f1": 0.5685682908450502,
|
| 109 |
+
"eval_runtime": 1.0561,
|
| 110 |
+
"eval_samples_per_second": 649.577,
|
| 111 |
+
"eval_steps_per_second": 20.832,
|
| 112 |
+
"step": 1044
|
| 113 |
+
},
|
| 114 |
+
{
|
| 115 |
+
"epoch": 6.959770114942529,
|
| 116 |
+
"grad_norm": 19.177682876586914,
|
| 117 |
+
"learning_rate": 1.991302739004225e-05,
|
| 118 |
+
"loss": 0.4856,
|
| 119 |
+
"step": 1211
|
| 120 |
+
},
|
| 121 |
+
{
|
| 122 |
+
"epoch": 7.0,
|
| 123 |
+
"eval_accuracy": 0.6180758017492711,
|
| 124 |
+
"eval_loss": 1.2190402746200562,
|
| 125 |
+
"eval_macro_f1": 0.5720115104548725,
|
| 126 |
+
"eval_runtime": 2.3657,
|
| 127 |
+
"eval_samples_per_second": 289.981,
|
| 128 |
+
"eval_steps_per_second": 9.3,
|
| 129 |
+
"step": 1218
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 7.954022988505747,
|
| 133 |
+
"grad_norm": 10.141107559204102,
|
| 134 |
+
"learning_rate": 1.9865586643899352e-05,
|
| 135 |
+
"loss": 0.3773,
|
| 136 |
+
"step": 1384
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 8.0,
|
| 140 |
+
"eval_accuracy": 0.6180758017492711,
|
| 141 |
+
"eval_loss": 1.332094430923462,
|
| 142 |
+
"eval_macro_f1": 0.5724900288812286,
|
| 143 |
+
"eval_runtime": 2.3855,
|
| 144 |
+
"eval_samples_per_second": 287.57,
|
| 145 |
+
"eval_steps_per_second": 9.222,
|
| 146 |
+
"step": 1392
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 8.948275862068966,
|
| 150 |
+
"grad_norm": 14.313020706176758,
|
| 151 |
+
"learning_rate": 1.9807943458281543e-05,
|
| 152 |
+
"loss": 0.2897,
|
| 153 |
+
"step": 1557
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"epoch": 9.0,
|
| 157 |
+
"eval_accuracy": 0.6064139941690962,
|
| 158 |
+
"eval_loss": 1.4038197994232178,
|
| 159 |
+
"eval_macro_f1": 0.5674191936637583,
|
| 160 |
+
"eval_runtime": 2.3637,
|
| 161 |
+
"eval_samples_per_second": 290.225,
|
| 162 |
+
"eval_steps_per_second": 9.307,
|
| 163 |
+
"step": 1566
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 9.942528735632184,
|
| 167 |
+
"grad_norm": 12.335651397705078,
|
| 168 |
+
"learning_rate": 1.974015744455647e-05,
|
| 169 |
+
"loss": 0.2137,
|
| 170 |
+
"step": 1730
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 10.0,
|
| 174 |
+
"eval_accuracy": 0.6209912536443148,
|
| 175 |
+
"eval_loss": 1.548543095588684,
|
| 176 |
+
"eval_macro_f1": 0.5761372608282923,
|
| 177 |
+
"eval_runtime": 2.3742,
|
| 178 |
+
"eval_samples_per_second": 288.935,
|
| 179 |
+
"eval_steps_per_second": 9.266,
|
| 180 |
+
"step": 1740
|
| 181 |
+
},
|
| 182 |
+
{
|
| 183 |
+
"epoch": 10.936781609195402,
|
| 184 |
+
"grad_norm": 16.48043441772461,
|
| 185 |
+
"learning_rate": 1.9662298703238904e-05,
|
| 186 |
+
"loss": 0.1764,
|
| 187 |
+
"step": 1903
|
| 188 |
+
},
|
| 189 |
+
{
|
| 190 |
+
"epoch": 11.0,
|
| 191 |
+
"eval_accuracy": 0.597667638483965,
|
| 192 |
+
"eval_loss": 1.697547435760498,
|
| 193 |
+
"eval_macro_f1": 0.5601682218476405,
|
| 194 |
+
"eval_runtime": 2.3856,
|
| 195 |
+
"eval_samples_per_second": 287.564,
|
| 196 |
+
"eval_steps_per_second": 9.222,
|
| 197 |
+
"step": 1914
|
| 198 |
+
},
|
| 199 |
+
{
|
| 200 |
+
"epoch": 11.931034482758621,
|
| 201 |
+
"grad_norm": 20.897550582885742,
|
| 202 |
+
"learning_rate": 1.9574447751496706e-05,
|
| 203 |
+
"loss": 0.128,
|
| 204 |
+
"step": 2076
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"epoch": 12.0,
|
| 208 |
+
"eval_accuracy": 0.6049562682215743,
|
| 209 |
+
"eval_loss": 1.832864761352539,
|
| 210 |
+
"eval_macro_f1": 0.5635033671342974,
|
| 211 |
+
"eval_runtime": 2.3698,
|
| 212 |
+
"eval_samples_per_second": 289.481,
|
| 213 |
+
"eval_steps_per_second": 9.284,
|
| 214 |
+
"step": 2088
|
| 215 |
+
},
|
| 216 |
+
{
|
| 217 |
+
"epoch": 12.925287356321839,
|
| 218 |
+
"grad_norm": 27.952417373657227,
|
| 219 |
+
"learning_rate": 1.9476695439884466e-05,
|
| 220 |
+
"loss": 0.12,
|
| 221 |
+
"step": 2249
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 13.0,
|
| 225 |
+
"eval_accuracy": 0.575801749271137,
|
| 226 |
+
"eval_loss": 2.1239633560180664,
|
| 227 |
+
"eval_macro_f1": 0.5353047636964179,
|
| 228 |
+
"eval_runtime": 2.3893,
|
| 229 |
+
"eval_samples_per_second": 287.113,
|
| 230 |
+
"eval_steps_per_second": 9.208,
|
| 231 |
+
"step": 2262
|
| 232 |
+
},
|
| 233 |
+
{
|
| 234 |
+
"epoch": 13.919540229885058,
|
| 235 |
+
"grad_norm": 7.873048782348633,
|
| 236 |
+
"learning_rate": 1.9369142858390936e-05,
|
| 237 |
+
"loss": 0.0868,
|
| 238 |
+
"step": 2422
|
| 239 |
+
},
|
| 240 |
+
{
|
| 241 |
+
"epoch": 14.0,
|
| 242 |
+
"eval_accuracy": 0.5947521865889213,
|
| 243 |
+
"eval_loss": 2.1011478900909424,
|
| 244 |
+
"eval_macro_f1": 0.5666855895738372,
|
| 245 |
+
"eval_runtime": 2.3541,
|
| 246 |
+
"eval_samples_per_second": 291.401,
|
| 247 |
+
"eval_steps_per_second": 9.345,
|
| 248 |
+
"step": 2436
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 14.913793103448276,
|
| 252 |
+
"grad_norm": 1.451875925064087,
|
| 253 |
+
"learning_rate": 1.9251901231897364e-05,
|
| 254 |
+
"loss": 0.0807,
|
| 255 |
+
"step": 2595
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"epoch": 15.0,
|
| 259 |
+
"eval_accuracy": 0.607871720116618,
|
| 260 |
+
"eval_loss": 2.229091167449951,
|
| 261 |
+
"eval_macro_f1": 0.5818994852139766,
|
| 262 |
+
"eval_runtime": 2.3965,
|
| 263 |
+
"eval_samples_per_second": 286.253,
|
| 264 |
+
"eval_steps_per_second": 9.18,
|
| 265 |
+
"step": 2610
|
| 266 |
+
},
|
| 267 |
+
{
|
| 268 |
+
"epoch": 15.908045977011493,
|
| 269 |
+
"grad_norm": 39.825321197509766,
|
| 270 |
+
"learning_rate": 1.9125091805154977e-05,
|
| 271 |
+
"loss": 0.0612,
|
| 272 |
+
"step": 2768
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 16.0,
|
| 276 |
+
"eval_accuracy": 0.6151603498542274,
|
| 277 |
+
"eval_loss": 2.392549753189087,
|
| 278 |
+
"eval_macro_f1": 0.5672571391448136,
|
| 279 |
+
"eval_runtime": 2.3779,
|
| 280 |
+
"eval_samples_per_second": 288.484,
|
| 281 |
+
"eval_steps_per_second": 9.252,
|
| 282 |
+
"step": 2784
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 16.902298850574713,
|
| 286 |
+
"grad_norm": 40.26604461669922,
|
| 287 |
+
"learning_rate": 1.8988845717400375e-05,
|
| 288 |
+
"loss": 0.0587,
|
| 289 |
+
"step": 2941
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 17.0,
|
| 293 |
+
"eval_accuracy": 0.6137026239067055,
|
| 294 |
+
"eval_loss": 2.4110476970672607,
|
| 295 |
+
"eval_macro_f1": 0.5690951143898209,
|
| 296 |
+
"eval_runtime": 2.3821,
|
| 297 |
+
"eval_samples_per_second": 287.975,
|
| 298 |
+
"eval_steps_per_second": 9.235,
|
| 299 |
+
"step": 2958
|
| 300 |
+
},
|
| 301 |
+
{
|
| 302 |
+
"epoch": 17.896551724137932,
|
| 303 |
+
"grad_norm": 2.2619950771331787,
|
| 304 |
+
"learning_rate": 1.884330386673869e-05,
|
| 305 |
+
"loss": 0.054,
|
| 306 |
+
"step": 3114
|
| 307 |
+
},
|
| 308 |
+
{
|
| 309 |
+
"epoch": 18.0,
|
| 310 |
+
"eval_accuracy": 0.6326530612244898,
|
| 311 |
+
"eval_loss": 2.3882195949554443,
|
| 312 |
+
"eval_macro_f1": 0.5881329726800972,
|
| 313 |
+
"eval_runtime": 2.3906,
|
| 314 |
+
"eval_samples_per_second": 286.963,
|
| 315 |
+
"eval_steps_per_second": 9.203,
|
| 316 |
+
"step": 3132
|
| 317 |
+
}
|
| 318 |
+
],
|
| 319 |
+
"logging_steps": 173,
|
| 320 |
+
"max_steps": 17400,
|
| 321 |
+
"num_input_tokens_seen": 0,
|
| 322 |
+
"num_train_epochs": 100,
|
| 323 |
+
"save_steps": 500,
|
| 324 |
+
"stateful_callbacks": {
|
| 325 |
+
"EarlyStoppingCallback": {
|
| 326 |
+
"args": {
|
| 327 |
+
"early_stopping_patience": 5,
|
| 328 |
+
"early_stopping_threshold": 0.0
|
| 329 |
+
},
|
| 330 |
+
"attributes": {
|
| 331 |
+
"early_stopping_patience_counter": 0
|
| 332 |
+
}
|
| 333 |
+
},
|
| 334 |
+
"TrainerControl": {
|
| 335 |
+
"args": {
|
| 336 |
+
"should_epoch_stop": false,
|
| 337 |
+
"should_evaluate": false,
|
| 338 |
+
"should_log": false,
|
| 339 |
+
"should_save": true,
|
| 340 |
+
"should_training_stop": false
|
| 341 |
+
},
|
| 342 |
+
"attributes": {}
|
| 343 |
+
}
|
| 344 |
+
},
|
| 345 |
+
"total_flos": 9289154030900400.0,
|
| 346 |
+
"train_batch_size": 32,
|
| 347 |
+
"trial_name": null,
|
| 348 |
+
"trial_params": null
|
| 349 |
+
}
|