End of training
Browse files- all_results.json +7 -7
- test_results.json +7 -7
- trainer_state.json +1114 -1005
all_results.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
-
"accuracy": 0.
|
| 3 |
-
"f1": 0.
|
| 4 |
-
"precision": 0.
|
| 5 |
-
"recall": 0.
|
| 6 |
-
"top_10_accuracy": 0.
|
| 7 |
-
"top_1_accuracy": 0.
|
| 8 |
-
"top_5_accuracy": 0.
|
| 9 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"accuracy": 0.5697674418604651,
|
| 3 |
+
"f1": 0.5385868653310515,
|
| 4 |
+
"precision": 0.5749307862679955,
|
| 5 |
+
"recall": 0.5697674418604651,
|
| 6 |
+
"top_10_accuracy": 0.875968992248062,
|
| 7 |
+
"top_1_accuracy": 0.5697674418604651,
|
| 8 |
+
"top_5_accuracy": 0.8217054263565892
|
| 9 |
}
|
test_results.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
-
"accuracy": 0.
|
| 3 |
-
"f1": 0.
|
| 4 |
-
"precision": 0.
|
| 5 |
-
"recall": 0.
|
| 6 |
-
"top_10_accuracy": 0.
|
| 7 |
-
"top_1_accuracy": 0.
|
| 8 |
-
"top_5_accuracy": 0.
|
| 9 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"accuracy": 0.5697674418604651,
|
| 3 |
+
"f1": 0.5385868653310515,
|
| 4 |
+
"precision": 0.5749307862679955,
|
| 5 |
+
"recall": 0.5697674418604651,
|
| 6 |
+
"top_10_accuracy": 0.875968992248062,
|
| 7 |
+
"top_1_accuracy": 0.5697674418604651,
|
| 8 |
+
"top_5_accuracy": 0.8217054263565892
|
| 9 |
}
|
trainer_state.json
CHANGED
|
@@ -1,1539 +1,1648 @@
|
|
| 1 |
{
|
| 2 |
-
"best_metric": 0.
|
| 3 |
-
"best_model_checkpoint": "/media/cse/HDD/Shawon/shawon/MY DATA/VideoMAE_Base_WLASL_100_200_epochs_p20_SR_8/checkpoint-
|
| 4 |
-
"epoch":
|
| 5 |
"eval_steps": 500,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
| 10 |
"log_history": [
|
| 11 |
{
|
| 12 |
"epoch": 0.002777777777777778,
|
| 13 |
-
"grad_norm":
|
| 14 |
"learning_rate": 1.3333333333333334e-06,
|
| 15 |
-
"loss": 18.
|
| 16 |
"step": 100
|
| 17 |
},
|
| 18 |
{
|
| 19 |
"epoch": 0.005,
|
| 20 |
"eval_accuracy": 0.008875739644970414,
|
| 21 |
-
"eval_f1": 0.
|
| 22 |
-
"eval_loss": 4.
|
| 23 |
-
"eval_precision": 0.
|
| 24 |
"eval_recall": 0.008875739644970414,
|
| 25 |
-
"eval_runtime":
|
| 26 |
-
"eval_samples_per_second":
|
| 27 |
-
"eval_steps_per_second":
|
| 28 |
-
"eval_top_10_accuracy": 0.
|
| 29 |
"eval_top_1_accuracy": 0.008875739644970414,
|
| 30 |
-
"eval_top_5_accuracy": 0.
|
| 31 |
"step": 180
|
| 32 |
},
|
| 33 |
{
|
| 34 |
-
"epoch": 1.
|
| 35 |
-
"grad_norm":
|
| 36 |
"learning_rate": 2.7222222222222224e-06,
|
| 37 |
-
"loss": 18.
|
| 38 |
"step": 200
|
| 39 |
},
|
| 40 |
{
|
| 41 |
-
"epoch": 1.
|
| 42 |
-
"grad_norm":
|
| 43 |
"learning_rate": 4.111111111111112e-06,
|
| 44 |
-
"loss": 18.
|
| 45 |
"step": 300
|
| 46 |
},
|
| 47 |
{
|
| 48 |
-
"epoch": 1.
|
| 49 |
-
"eval_accuracy": 0.
|
| 50 |
-
"eval_f1": 0.
|
| 51 |
-
"eval_loss": 4.
|
| 52 |
-
"eval_precision": 0.
|
| 53 |
-
"eval_recall": 0.
|
| 54 |
-
"eval_runtime":
|
| 55 |
-
"eval_samples_per_second": 5.
|
| 56 |
-
"eval_steps_per_second": 2.
|
| 57 |
-
"eval_top_10_accuracy": 0.
|
| 58 |
-
"eval_top_1_accuracy": 0.
|
| 59 |
"eval_top_5_accuracy": 0.0621301775147929,
|
| 60 |
"step": 360
|
| 61 |
},
|
| 62 |
{
|
| 63 |
-
"epoch": 2.
|
| 64 |
-
"grad_norm":
|
| 65 |
"learning_rate": 5.500000000000001e-06,
|
| 66 |
-
"loss": 18.
|
| 67 |
"step": 400
|
| 68 |
},
|
| 69 |
{
|
| 70 |
-
"epoch": 2.
|
| 71 |
-
"grad_norm":
|
| 72 |
"learning_rate": 6.888888888888889e-06,
|
| 73 |
-
"loss": 18.
|
| 74 |
"step": 500
|
| 75 |
},
|
| 76 |
{
|
| 77 |
-
"epoch": 2.
|
| 78 |
-
"eval_accuracy": 0.
|
| 79 |
-
"eval_f1": 0.
|
| 80 |
-
"eval_loss": 4.
|
| 81 |
-
"eval_precision": 0.
|
| 82 |
-
"eval_recall": 0.
|
| 83 |
-
"eval_runtime":
|
| 84 |
-
"eval_samples_per_second": 5.
|
| 85 |
-
"eval_steps_per_second": 2.
|
| 86 |
-
"eval_top_10_accuracy": 0.
|
| 87 |
-
"eval_top_1_accuracy": 0.
|
| 88 |
-
"eval_top_5_accuracy": 0.
|
| 89 |
-
"step":
|
| 90 |
},
|
| 91 |
{
|
| 92 |
-
"epoch": 3.
|
| 93 |
-
"grad_norm":
|
| 94 |
"learning_rate": 8.27777777777778e-06,
|
| 95 |
-
"loss": 18.
|
| 96 |
"step": 600
|
| 97 |
},
|
| 98 |
{
|
| 99 |
-
"epoch": 3.
|
| 100 |
-
"grad_norm":
|
| 101 |
"learning_rate": 9.666666666666667e-06,
|
| 102 |
-
"loss": 18.
|
| 103 |
"step": 700
|
| 104 |
},
|
| 105 |
{
|
| 106 |
-
"epoch": 3.
|
| 107 |
-
"eval_accuracy": 0.
|
| 108 |
-
"eval_f1": 0.
|
| 109 |
-
"eval_loss": 4.
|
| 110 |
-
"eval_precision": 0.
|
| 111 |
-
"eval_recall": 0.
|
| 112 |
-
"eval_runtime": 58.
|
| 113 |
-
"eval_samples_per_second": 5.
|
| 114 |
-
"eval_steps_per_second": 2.
|
| 115 |
-
"eval_top_10_accuracy": 0.
|
| 116 |
-
"eval_top_1_accuracy": 0.
|
| 117 |
-
"eval_top_5_accuracy": 0.
|
| 118 |
"step": 721
|
| 119 |
},
|
| 120 |
{
|
| 121 |
-
"epoch": 4.
|
| 122 |
-
"grad_norm":
|
| 123 |
"learning_rate": 1.1055555555555556e-05,
|
| 124 |
-
"loss": 18.
|
| 125 |
"step": 800
|
| 126 |
},
|
| 127 |
{
|
| 128 |
-
"epoch": 4.
|
| 129 |
-
"grad_norm":
|
| 130 |
"learning_rate": 1.2444444444444445e-05,
|
| 131 |
-
"loss": 18.
|
| 132 |
"step": 900
|
| 133 |
},
|
| 134 |
{
|
| 135 |
-
"epoch": 4.
|
| 136 |
-
"eval_accuracy": 0.
|
| 137 |
-
"eval_f1": 0.
|
| 138 |
-
"eval_loss": 4.
|
| 139 |
-
"eval_precision": 0.
|
| 140 |
-
"eval_recall": 0.
|
| 141 |
-
"eval_runtime":
|
| 142 |
-
"eval_samples_per_second":
|
| 143 |
-
"eval_steps_per_second": 2.
|
| 144 |
-
"eval_top_10_accuracy": 0.
|
| 145 |
-
"eval_top_1_accuracy": 0.
|
| 146 |
-
"eval_top_5_accuracy": 0.
|
| 147 |
"step": 901
|
| 148 |
},
|
| 149 |
{
|
| 150 |
-
"epoch": 5.
|
| 151 |
-
"grad_norm":
|
| 152 |
"learning_rate": 1.3833333333333334e-05,
|
| 153 |
-
"loss": 18.
|
| 154 |
"step": 1000
|
| 155 |
},
|
| 156 |
{
|
| 157 |
-
"epoch": 5.
|
| 158 |
-
"eval_accuracy": 0.
|
| 159 |
-
"eval_f1": 0.
|
| 160 |
-
"eval_loss": 4.
|
| 161 |
-
"eval_precision": 0.
|
| 162 |
-
"eval_recall": 0.
|
| 163 |
-
"eval_runtime":
|
| 164 |
-
"eval_samples_per_second":
|
| 165 |
-
"eval_steps_per_second":
|
| 166 |
-
"eval_top_10_accuracy": 0.
|
| 167 |
-
"eval_top_1_accuracy": 0.
|
| 168 |
-
"eval_top_5_accuracy": 0.
|
| 169 |
"step": 1081
|
| 170 |
},
|
| 171 |
{
|
| 172 |
-
"epoch": 6.
|
| 173 |
-
"grad_norm":
|
| 174 |
"learning_rate": 1.5222222222222224e-05,
|
| 175 |
-
"loss": 18.
|
| 176 |
"step": 1100
|
| 177 |
},
|
| 178 |
{
|
| 179 |
-
"epoch": 6.
|
| 180 |
-
"grad_norm":
|
| 181 |
"learning_rate": 1.661111111111111e-05,
|
| 182 |
-
"loss": 18.
|
| 183 |
"step": 1200
|
| 184 |
},
|
| 185 |
{
|
| 186 |
-
"epoch": 6.
|
| 187 |
-
"eval_accuracy": 0.
|
| 188 |
-
"eval_f1": 0.
|
| 189 |
-
"eval_loss": 4.
|
| 190 |
-
"eval_precision": 0.
|
| 191 |
-
"eval_recall": 0.
|
| 192 |
-
"eval_runtime":
|
| 193 |
-
"eval_samples_per_second":
|
| 194 |
-
"eval_steps_per_second": 2.
|
| 195 |
-
"eval_top_10_accuracy": 0.
|
| 196 |
-
"eval_top_1_accuracy": 0.
|
| 197 |
-
"eval_top_5_accuracy": 0.
|
| 198 |
-
"step":
|
| 199 |
},
|
| 200 |
{
|
| 201 |
-
"epoch": 7.
|
| 202 |
-
"grad_norm":
|
| 203 |
"learning_rate": 1.8e-05,
|
| 204 |
-
"loss": 18.
|
| 205 |
"step": 1300
|
| 206 |
},
|
| 207 |
{
|
| 208 |
-
"epoch": 7.
|
| 209 |
-
"grad_norm":
|
| 210 |
"learning_rate": 1.938888888888889e-05,
|
| 211 |
-
"loss": 18.
|
| 212 |
"step": 1400
|
| 213 |
},
|
| 214 |
{
|
| 215 |
-
"epoch": 7.
|
| 216 |
-
"eval_accuracy": 0.
|
| 217 |
-
"eval_f1": 0.
|
| 218 |
-
"eval_loss": 4.
|
| 219 |
-
"eval_precision": 0.
|
| 220 |
-
"eval_recall": 0.
|
| 221 |
-
"eval_runtime":
|
| 222 |
-
"eval_samples_per_second":
|
| 223 |
-
"eval_steps_per_second": 2.
|
| 224 |
-
"eval_top_10_accuracy": 0.
|
| 225 |
-
"eval_top_1_accuracy": 0.
|
| 226 |
-
"eval_top_5_accuracy": 0.
|
| 227 |
"step": 1442
|
| 228 |
},
|
| 229 |
{
|
| 230 |
-
"epoch": 8.
|
| 231 |
-
"grad_norm":
|
| 232 |
"learning_rate": 2.077777777777778e-05,
|
| 233 |
-
"loss": 18.
|
| 234 |
"step": 1500
|
| 235 |
},
|
| 236 |
{
|
| 237 |
-
"epoch": 8.
|
| 238 |
-
"grad_norm":
|
| 239 |
"learning_rate": 2.216666666666667e-05,
|
| 240 |
-
"loss":
|
| 241 |
"step": 1600
|
| 242 |
},
|
| 243 |
{
|
| 244 |
-
"epoch": 8.
|
| 245 |
-
"eval_accuracy": 0.
|
| 246 |
-
"eval_f1": 0.
|
| 247 |
-
"eval_loss": 4.
|
| 248 |
-
"eval_precision": 0.
|
| 249 |
-
"eval_recall": 0.
|
| 250 |
-
"eval_runtime":
|
| 251 |
-
"eval_samples_per_second": 5.
|
| 252 |
-
"eval_steps_per_second": 2.
|
| 253 |
-
"eval_top_10_accuracy": 0.
|
| 254 |
-
"eval_top_1_accuracy": 0.
|
| 255 |
-
"eval_top_5_accuracy": 0.
|
| 256 |
"step": 1622
|
| 257 |
},
|
| 258 |
{
|
| 259 |
-
"epoch": 9.
|
| 260 |
-
"grad_norm":
|
| 261 |
"learning_rate": 2.3555555555555556e-05,
|
| 262 |
-
"loss": 17.
|
| 263 |
"step": 1700
|
| 264 |
},
|
| 265 |
{
|
| 266 |
-
"epoch": 9.
|
| 267 |
-
"grad_norm":
|
| 268 |
"learning_rate": 2.4944444444444447e-05,
|
| 269 |
-
"loss": 17.
|
| 270 |
"step": 1800
|
| 271 |
},
|
| 272 |
{
|
| 273 |
-
"epoch": 9.
|
| 274 |
-
"eval_accuracy": 0.
|
| 275 |
-
"eval_f1": 0.
|
| 276 |
-
"eval_loss": 4.
|
| 277 |
-
"eval_precision": 0.
|
| 278 |
-
"eval_recall": 0.
|
| 279 |
-
"eval_runtime":
|
| 280 |
-
"eval_samples_per_second": 5.
|
| 281 |
-
"eval_steps_per_second": 2.
|
| 282 |
-
"eval_top_10_accuracy": 0.
|
| 283 |
-
"eval_top_1_accuracy": 0.
|
| 284 |
-
"eval_top_5_accuracy": 0.
|
| 285 |
"step": 1802
|
| 286 |
},
|
| 287 |
{
|
| 288 |
-
"epoch": 10.
|
| 289 |
-
"grad_norm":
|
| 290 |
"learning_rate": 2.633333333333333e-05,
|
| 291 |
-
"loss":
|
| 292 |
"step": 1900
|
| 293 |
},
|
| 294 |
{
|
| 295 |
-
"epoch": 10.
|
| 296 |
-
"eval_accuracy": 0.
|
| 297 |
-
"eval_f1": 0.
|
| 298 |
-
"eval_loss": 4.
|
| 299 |
-
"eval_precision": 0.
|
| 300 |
-
"eval_recall": 0.
|
| 301 |
-
"eval_runtime":
|
| 302 |
-
"eval_samples_per_second": 5.
|
| 303 |
-
"eval_steps_per_second": 2.
|
| 304 |
-
"eval_top_10_accuracy": 0.
|
| 305 |
-
"eval_top_1_accuracy": 0.
|
| 306 |
-
"eval_top_5_accuracy": 0.
|
| 307 |
-
"step":
|
| 308 |
},
|
| 309 |
{
|
| 310 |
-
"epoch": 11.
|
| 311 |
-
"grad_norm":
|
| 312 |
"learning_rate": 2.772222222222222e-05,
|
| 313 |
-
"loss":
|
| 314 |
"step": 2000
|
| 315 |
},
|
| 316 |
{
|
| 317 |
-
"epoch": 11.
|
| 318 |
-
"grad_norm":
|
| 319 |
"learning_rate": 2.9111111111111112e-05,
|
| 320 |
-
"loss":
|
| 321 |
"step": 2100
|
| 322 |
},
|
| 323 |
{
|
| 324 |
-
"epoch": 11.
|
| 325 |
-
"eval_accuracy": 0.
|
| 326 |
-
"eval_f1": 0.
|
| 327 |
-
"eval_loss":
|
| 328 |
-
"eval_precision": 0.
|
| 329 |
-
"eval_recall": 0.
|
| 330 |
-
"eval_runtime":
|
| 331 |
-
"eval_samples_per_second":
|
| 332 |
-
"eval_steps_per_second": 2.
|
| 333 |
-
"eval_top_10_accuracy": 0.
|
| 334 |
-
"eval_top_1_accuracy": 0.
|
| 335 |
-
"eval_top_5_accuracy": 0.
|
| 336 |
"step": 2163
|
| 337 |
},
|
| 338 |
{
|
| 339 |
-
"epoch": 12.
|
| 340 |
-
"grad_norm":
|
| 341 |
"learning_rate": 3.05e-05,
|
| 342 |
-
"loss":
|
| 343 |
"step": 2200
|
| 344 |
},
|
| 345 |
{
|
| 346 |
-
"epoch": 12.
|
| 347 |
-
"grad_norm":
|
| 348 |
"learning_rate": 3.188888888888889e-05,
|
| 349 |
-
"loss":
|
| 350 |
"step": 2300
|
| 351 |
},
|
| 352 |
{
|
| 353 |
-
"epoch": 12.
|
| 354 |
-
"eval_accuracy": 0.
|
| 355 |
-
"eval_f1": 0.
|
| 356 |
-
"eval_loss": 3.
|
| 357 |
-
"eval_precision": 0.
|
| 358 |
-
"eval_recall": 0.
|
| 359 |
-
"eval_runtime":
|
| 360 |
-
"eval_samples_per_second": 5.
|
| 361 |
-
"eval_steps_per_second": 2.
|
| 362 |
-
"eval_top_10_accuracy": 0.
|
| 363 |
-
"eval_top_1_accuracy": 0.
|
| 364 |
-
"eval_top_5_accuracy": 0.
|
| 365 |
"step": 2343
|
| 366 |
},
|
| 367 |
{
|
| 368 |
-
"epoch": 13.
|
| 369 |
-
"grad_norm":
|
| 370 |
"learning_rate": 3.327777777777778e-05,
|
| 371 |
-
"loss":
|
| 372 |
"step": 2400
|
| 373 |
},
|
| 374 |
{
|
| 375 |
-
"epoch": 13.
|
| 376 |
-
"grad_norm":
|
| 377 |
"learning_rate": 3.466666666666667e-05,
|
| 378 |
-
"loss":
|
| 379 |
"step": 2500
|
| 380 |
},
|
| 381 |
{
|
| 382 |
-
"epoch": 13.
|
| 383 |
-
"eval_accuracy": 0.
|
| 384 |
-
"eval_f1": 0.
|
| 385 |
-
"eval_loss": 3.
|
| 386 |
-
"eval_precision": 0.
|
| 387 |
-
"eval_recall": 0.
|
| 388 |
-
"eval_runtime":
|
| 389 |
-
"eval_samples_per_second": 5.
|
| 390 |
-
"eval_steps_per_second": 2.
|
| 391 |
-
"eval_top_10_accuracy": 0.
|
| 392 |
-
"eval_top_1_accuracy": 0.
|
| 393 |
-
"eval_top_5_accuracy": 0.
|
| 394 |
"step": 2523
|
| 395 |
},
|
| 396 |
{
|
| 397 |
-
"epoch": 14.
|
| 398 |
-
"grad_norm":
|
| 399 |
"learning_rate": 3.605555555555556e-05,
|
| 400 |
-
"loss":
|
| 401 |
"step": 2600
|
| 402 |
},
|
| 403 |
{
|
| 404 |
-
"epoch": 14.
|
| 405 |
-
"grad_norm":
|
| 406 |
"learning_rate": 3.7444444444444446e-05,
|
| 407 |
-
"loss":
|
| 408 |
"step": 2700
|
| 409 |
},
|
| 410 |
{
|
| 411 |
-
"epoch": 14.
|
| 412 |
-
"eval_accuracy": 0.
|
| 413 |
-
"eval_f1": 0.
|
| 414 |
-
"eval_loss":
|
| 415 |
-
"eval_precision": 0.
|
| 416 |
-
"eval_recall": 0.
|
| 417 |
-
"eval_runtime":
|
| 418 |
-
"eval_samples_per_second":
|
| 419 |
-
"eval_steps_per_second": 2.
|
| 420 |
-
"eval_top_10_accuracy": 0.
|
| 421 |
-
"eval_top_1_accuracy": 0.
|
| 422 |
-
"eval_top_5_accuracy": 0.
|
| 423 |
-
"step":
|
| 424 |
},
|
| 425 |
{
|
| 426 |
-
"epoch": 15.
|
| 427 |
-
"grad_norm":
|
| 428 |
"learning_rate": 3.883333333333333e-05,
|
| 429 |
-
"loss":
|
| 430 |
"step": 2800
|
| 431 |
},
|
| 432 |
{
|
| 433 |
-
"epoch": 15.
|
| 434 |
-
"eval_accuracy": 0.
|
| 435 |
-
"eval_f1": 0.
|
| 436 |
-
"eval_loss": 2.
|
| 437 |
-
"eval_precision": 0.
|
| 438 |
-
"eval_recall": 0.
|
| 439 |
-
"eval_runtime":
|
| 440 |
-
"eval_samples_per_second":
|
| 441 |
-
"eval_steps_per_second":
|
| 442 |
-
"eval_top_10_accuracy": 0.
|
| 443 |
-
"eval_top_1_accuracy": 0.
|
| 444 |
-
"eval_top_5_accuracy": 0.
|
| 445 |
"step": 2884
|
| 446 |
},
|
| 447 |
{
|
| 448 |
-
"epoch": 16.
|
| 449 |
-
"grad_norm":
|
| 450 |
"learning_rate": 4.022222222222222e-05,
|
| 451 |
-
"loss":
|
| 452 |
"step": 2900
|
| 453 |
},
|
| 454 |
{
|
| 455 |
-
"epoch": 16.
|
| 456 |
-
"grad_norm":
|
| 457 |
"learning_rate": 4.1611111111111114e-05,
|
| 458 |
-
"loss":
|
| 459 |
"step": 3000
|
| 460 |
},
|
| 461 |
{
|
| 462 |
-
"epoch": 16.
|
| 463 |
-
"eval_accuracy": 0.
|
| 464 |
-
"eval_f1": 0.
|
| 465 |
-
"eval_loss": 2.
|
| 466 |
-
"eval_precision": 0.
|
| 467 |
-
"eval_recall": 0.
|
| 468 |
-
"eval_runtime":
|
| 469 |
-
"eval_samples_per_second":
|
| 470 |
-
"eval_steps_per_second":
|
| 471 |
-
"eval_top_10_accuracy": 0.
|
| 472 |
-
"eval_top_1_accuracy": 0.
|
| 473 |
-
"eval_top_5_accuracy": 0.
|
| 474 |
"step": 3064
|
| 475 |
},
|
| 476 |
{
|
| 477 |
-
"epoch": 17.
|
| 478 |
-
"grad_norm":
|
| 479 |
"learning_rate": 4.3e-05,
|
| 480 |
-
"loss":
|
| 481 |
"step": 3100
|
| 482 |
},
|
| 483 |
{
|
| 484 |
-
"epoch": 17.
|
| 485 |
-
"grad_norm":
|
| 486 |
"learning_rate": 4.438888888888889e-05,
|
| 487 |
-
"loss": 6.
|
| 488 |
"step": 3200
|
| 489 |
},
|
| 490 |
{
|
| 491 |
-
"epoch": 17.
|
| 492 |
-
"eval_accuracy": 0.
|
| 493 |
-
"eval_f1": 0.
|
| 494 |
-
"eval_loss": 2.
|
| 495 |
-
"eval_precision": 0.
|
| 496 |
-
"eval_recall": 0.
|
| 497 |
-
"eval_runtime":
|
| 498 |
-
"eval_samples_per_second":
|
| 499 |
-
"eval_steps_per_second":
|
| 500 |
-
"eval_top_10_accuracy": 0.
|
| 501 |
-
"eval_top_1_accuracy": 0.
|
| 502 |
-
"eval_top_5_accuracy": 0.
|
| 503 |
"step": 3244
|
| 504 |
},
|
| 505 |
{
|
| 506 |
-
"epoch": 18.
|
| 507 |
-
"grad_norm":
|
| 508 |
"learning_rate": 4.577777777777778e-05,
|
| 509 |
-
"loss": 5.
|
| 510 |
"step": 3300
|
| 511 |
},
|
| 512 |
{
|
| 513 |
-
"epoch": 18.
|
| 514 |
-
"grad_norm":
|
| 515 |
"learning_rate": 4.716666666666667e-05,
|
| 516 |
-
"loss":
|
| 517 |
"step": 3400
|
| 518 |
},
|
| 519 |
{
|
| 520 |
-
"epoch": 18.
|
| 521 |
-
"eval_accuracy": 0.
|
| 522 |
-
"eval_f1": 0.
|
| 523 |
-
"eval_loss": 2.
|
| 524 |
-
"eval_precision": 0.
|
| 525 |
-
"eval_recall": 0.
|
| 526 |
-
"eval_runtime":
|
| 527 |
-
"eval_samples_per_second":
|
| 528 |
-
"eval_steps_per_second":
|
| 529 |
-
"eval_top_10_accuracy": 0.
|
| 530 |
-
"eval_top_1_accuracy": 0.
|
| 531 |
-
"eval_top_5_accuracy": 0.
|
| 532 |
-
"step":
|
| 533 |
},
|
| 534 |
{
|
| 535 |
-
"epoch": 19.
|
| 536 |
-
"grad_norm":
|
| 537 |
"learning_rate": 4.855555555555556e-05,
|
| 538 |
-
"loss": 4.
|
| 539 |
"step": 3500
|
| 540 |
},
|
| 541 |
{
|
| 542 |
-
"epoch": 19.
|
| 543 |
-
"grad_norm":
|
| 544 |
"learning_rate": 4.994444444444445e-05,
|
| 545 |
-
"loss":
|
| 546 |
"step": 3600
|
| 547 |
},
|
| 548 |
{
|
| 549 |
-
"epoch": 19.
|
| 550 |
-
"eval_accuracy": 0.
|
| 551 |
-
"eval_f1": 0.
|
| 552 |
-
"eval_loss": 2.
|
| 553 |
-
"eval_precision": 0.
|
| 554 |
-
"eval_recall": 0.
|
| 555 |
-
"eval_runtime":
|
| 556 |
-
"eval_samples_per_second":
|
| 557 |
-
"eval_steps_per_second":
|
| 558 |
-
"eval_top_10_accuracy": 0.
|
| 559 |
-
"eval_top_1_accuracy": 0.
|
| 560 |
"eval_top_5_accuracy": 0.7840236686390533,
|
| 561 |
"step": 3605
|
| 562 |
},
|
| 563 |
{
|
| 564 |
-
"epoch": 20.
|
| 565 |
-
"grad_norm":
|
| 566 |
"learning_rate": 4.9851851851851855e-05,
|
| 567 |
-
"loss":
|
| 568 |
"step": 3700
|
| 569 |
},
|
| 570 |
{
|
| 571 |
-
"epoch": 20.
|
| 572 |
-
"eval_accuracy": 0.
|
| 573 |
-
"eval_f1": 0.
|
| 574 |
-
"eval_loss":
|
| 575 |
-
"eval_precision": 0.
|
| 576 |
-
"eval_recall": 0.
|
| 577 |
-
"eval_runtime":
|
| 578 |
-
"eval_samples_per_second":
|
| 579 |
-
"eval_steps_per_second":
|
| 580 |
-
"eval_top_10_accuracy": 0.
|
| 581 |
-
"eval_top_1_accuracy": 0.
|
| 582 |
-
"eval_top_5_accuracy": 0.
|
| 583 |
"step": 3785
|
| 584 |
},
|
| 585 |
{
|
| 586 |
-
"epoch": 21.
|
| 587 |
-
"grad_norm":
|
| 588 |
"learning_rate": 4.969753086419753e-05,
|
| 589 |
-
"loss":
|
| 590 |
"step": 3800
|
| 591 |
},
|
| 592 |
{
|
| 593 |
-
"epoch": 21.
|
| 594 |
-
"grad_norm":
|
| 595 |
"learning_rate": 4.954320987654321e-05,
|
| 596 |
-
"loss": 2.
|
| 597 |
"step": 3900
|
| 598 |
},
|
| 599 |
{
|
| 600 |
-
"epoch": 21.
|
| 601 |
-
"eval_accuracy": 0.
|
| 602 |
-
"eval_f1": 0.
|
| 603 |
-
"eval_loss":
|
| 604 |
-
"eval_precision": 0.
|
| 605 |
-
"eval_recall": 0.
|
| 606 |
-
"eval_runtime":
|
| 607 |
-
"eval_samples_per_second":
|
| 608 |
-
"eval_steps_per_second":
|
| 609 |
-
"eval_top_10_accuracy": 0.
|
| 610 |
-
"eval_top_1_accuracy": 0.
|
| 611 |
-
"eval_top_5_accuracy": 0.
|
| 612 |
"step": 3965
|
| 613 |
},
|
| 614 |
{
|
| 615 |
-
"epoch": 22.
|
| 616 |
-
"grad_norm":
|
| 617 |
"learning_rate": 4.938888888888889e-05,
|
| 618 |
-
"loss":
|
| 619 |
"step": 4000
|
| 620 |
},
|
| 621 |
{
|
| 622 |
-
"epoch": 22.
|
| 623 |
-
"grad_norm":
|
| 624 |
-
"learning_rate": 4.
|
| 625 |
-
"loss": 1.
|
| 626 |
"step": 4100
|
| 627 |
},
|
| 628 |
{
|
| 629 |
-
"epoch": 22.
|
| 630 |
-
"eval_accuracy": 0.
|
| 631 |
-
"eval_f1": 0.
|
| 632 |
-
"eval_loss":
|
| 633 |
-
"eval_precision": 0.
|
| 634 |
-
"eval_recall": 0.
|
| 635 |
-
"eval_runtime":
|
| 636 |
-
"eval_samples_per_second":
|
| 637 |
-
"eval_steps_per_second":
|
| 638 |
-
"eval_top_10_accuracy": 0.
|
| 639 |
-
"eval_top_1_accuracy": 0.
|
| 640 |
-
"eval_top_5_accuracy": 0.
|
| 641 |
-
"step":
|
| 642 |
},
|
| 643 |
{
|
| 644 |
-
"epoch": 23.
|
| 645 |
-
"grad_norm":
|
| 646 |
"learning_rate": 4.9081790123456794e-05,
|
| 647 |
-
"loss": 1.
|
| 648 |
"step": 4200
|
| 649 |
},
|
| 650 |
{
|
| 651 |
-
"epoch": 23.
|
| 652 |
-
"grad_norm":
|
| 653 |
"learning_rate": 4.892746913580247e-05,
|
| 654 |
-
"loss": 1.
|
| 655 |
"step": 4300
|
| 656 |
},
|
| 657 |
{
|
| 658 |
-
"epoch": 23.
|
| 659 |
-
"eval_accuracy": 0.
|
| 660 |
-
"eval_f1": 0.
|
| 661 |
-
"eval_loss": 1.
|
| 662 |
-
"eval_precision": 0.
|
| 663 |
-
"eval_recall": 0.
|
| 664 |
-
"eval_runtime":
|
| 665 |
-
"eval_samples_per_second":
|
| 666 |
-
"eval_steps_per_second":
|
| 667 |
-
"eval_top_10_accuracy": 0.
|
| 668 |
-
"eval_top_1_accuracy": 0.
|
| 669 |
-
"eval_top_5_accuracy": 0.
|
| 670 |
"step": 4326
|
| 671 |
},
|
| 672 |
{
|
| 673 |
-
"epoch": 24.
|
| 674 |
-
"grad_norm":
|
| 675 |
"learning_rate": 4.877314814814815e-05,
|
| 676 |
-
"loss": 1.
|
| 677 |
"step": 4400
|
| 678 |
},
|
| 679 |
{
|
| 680 |
-
"epoch": 24.
|
| 681 |
-
"grad_norm":
|
| 682 |
"learning_rate": 4.861882716049383e-05,
|
| 683 |
-
"loss": 1.
|
| 684 |
"step": 4500
|
| 685 |
},
|
| 686 |
{
|
| 687 |
-
"epoch": 24.
|
| 688 |
-
"eval_accuracy": 0.
|
| 689 |
-
"eval_f1": 0.
|
| 690 |
-
"eval_loss": 1.
|
| 691 |
-
"eval_precision": 0.
|
| 692 |
-
"eval_recall": 0.
|
| 693 |
-
"eval_runtime":
|
| 694 |
-
"eval_samples_per_second":
|
| 695 |
-
"eval_steps_per_second":
|
| 696 |
-
"eval_top_10_accuracy": 0.
|
| 697 |
-
"eval_top_1_accuracy": 0.
|
| 698 |
-
"eval_top_5_accuracy": 0.
|
| 699 |
"step": 4506
|
| 700 |
},
|
| 701 |
{
|
| 702 |
-
"epoch": 25.
|
| 703 |
-
"grad_norm":
|
| 704 |
"learning_rate": 4.8464506172839505e-05,
|
| 705 |
-
"loss": 0.
|
| 706 |
"step": 4600
|
| 707 |
},
|
| 708 |
{
|
| 709 |
-
"epoch": 25.
|
| 710 |
-
"eval_accuracy": 0.
|
| 711 |
-
"eval_f1": 0.
|
| 712 |
-
"eval_loss": 1.
|
| 713 |
-
"eval_precision": 0.
|
| 714 |
-
"eval_recall": 0.
|
| 715 |
-
"eval_runtime":
|
| 716 |
-
"eval_samples_per_second":
|
| 717 |
-
"eval_steps_per_second": 2.
|
| 718 |
-
"eval_top_10_accuracy": 0.
|
| 719 |
-
"eval_top_1_accuracy": 0.
|
| 720 |
-
"eval_top_5_accuracy": 0.
|
| 721 |
"step": 4686
|
| 722 |
},
|
| 723 |
{
|
| 724 |
-
"epoch": 26.
|
| 725 |
-
"grad_norm":
|
| 726 |
"learning_rate": 4.831018518518518e-05,
|
| 727 |
-
"loss": 0.
|
| 728 |
"step": 4700
|
| 729 |
},
|
| 730 |
{
|
| 731 |
-
"epoch": 26.
|
| 732 |
-
"grad_norm":
|
| 733 |
"learning_rate": 4.815586419753087e-05,
|
| 734 |
-
"loss": 0.
|
| 735 |
"step": 4800
|
| 736 |
},
|
| 737 |
{
|
| 738 |
-
"epoch": 26.
|
| 739 |
"eval_accuracy": 0.5384615384615384,
|
| 740 |
-
"eval_f1": 0.
|
| 741 |
-
"eval_loss": 1.
|
| 742 |
-
"eval_precision": 0.
|
| 743 |
"eval_recall": 0.5384615384615384,
|
| 744 |
-
"eval_runtime": 58.
|
| 745 |
-
"eval_samples_per_second": 5.
|
| 746 |
-
"eval_steps_per_second": 2.
|
| 747 |
-
"eval_top_10_accuracy": 0.
|
| 748 |
-
"eval_top_1_accuracy": 0.
|
| 749 |
-
"eval_top_5_accuracy": 0.
|
| 750 |
-
"step":
|
| 751 |
},
|
| 752 |
{
|
| 753 |
-
"epoch": 27.
|
| 754 |
-
"grad_norm": 11.
|
| 755 |
"learning_rate": 4.8001543209876545e-05,
|
| 756 |
-
"loss": 0.
|
| 757 |
"step": 4900
|
| 758 |
},
|
| 759 |
{
|
| 760 |
-
"epoch": 27.
|
| 761 |
-
"grad_norm":
|
| 762 |
"learning_rate": 4.784722222222223e-05,
|
| 763 |
-
"loss": 0.
|
| 764 |
"step": 5000
|
| 765 |
},
|
| 766 |
{
|
| 767 |
-
"epoch": 27.
|
| 768 |
-
"eval_accuracy": 0.
|
| 769 |
-
"eval_f1": 0.
|
| 770 |
-
"eval_loss":
|
| 771 |
-
"eval_precision": 0.
|
| 772 |
-
"eval_recall": 0.
|
| 773 |
-
"eval_runtime":
|
| 774 |
-
"eval_samples_per_second":
|
| 775 |
-
"eval_steps_per_second":
|
| 776 |
-
"eval_top_10_accuracy": 0.
|
| 777 |
-
"eval_top_1_accuracy": 0.
|
| 778 |
-
"eval_top_5_accuracy": 0.
|
| 779 |
"step": 5047
|
| 780 |
},
|
| 781 |
{
|
| 782 |
-
"epoch": 28.
|
| 783 |
-
"grad_norm":
|
| 784 |
"learning_rate": 4.769290123456791e-05,
|
| 785 |
-
"loss": 0.
|
| 786 |
"step": 5100
|
| 787 |
},
|
| 788 |
{
|
| 789 |
-
"epoch": 28.
|
| 790 |
-
"grad_norm":
|
| 791 |
"learning_rate": 4.7538580246913585e-05,
|
| 792 |
-
"loss": 0.
|
| 793 |
"step": 5200
|
| 794 |
},
|
| 795 |
{
|
| 796 |
-
"epoch": 28.
|
| 797 |
-
"eval_accuracy": 0.
|
| 798 |
-
"eval_f1": 0.
|
| 799 |
-
"eval_loss": 1.
|
| 800 |
-
"eval_precision": 0.
|
| 801 |
-
"eval_recall": 0.
|
| 802 |
-
"eval_runtime": 60.
|
| 803 |
-
"eval_samples_per_second": 5.
|
| 804 |
-
"eval_steps_per_second": 2.
|
| 805 |
-
"eval_top_10_accuracy": 0.
|
| 806 |
-
"eval_top_1_accuracy": 0.
|
| 807 |
-
"eval_top_5_accuracy": 0.
|
| 808 |
"step": 5227
|
| 809 |
},
|
| 810 |
{
|
| 811 |
-
"epoch": 29.
|
| 812 |
-
"grad_norm":
|
| 813 |
"learning_rate": 4.738425925925926e-05,
|
| 814 |
-
"loss": 0.
|
| 815 |
"step": 5300
|
| 816 |
},
|
| 817 |
{
|
| 818 |
-
"epoch": 29.
|
| 819 |
-
"grad_norm":
|
| 820 |
"learning_rate": 4.722993827160494e-05,
|
| 821 |
-
"loss": 0.
|
| 822 |
"step": 5400
|
| 823 |
},
|
| 824 |
{
|
| 825 |
-
"epoch": 29.
|
| 826 |
-
"eval_accuracy": 0.
|
| 827 |
-
"eval_f1": 0.
|
| 828 |
-
"eval_loss":
|
| 829 |
-
"eval_precision": 0.
|
| 830 |
-
"eval_recall": 0.
|
| 831 |
-
"eval_runtime":
|
| 832 |
-
"eval_samples_per_second": 5.
|
| 833 |
-
"eval_steps_per_second": 2.
|
| 834 |
-
"eval_top_10_accuracy": 0.
|
| 835 |
-
"eval_top_1_accuracy": 0.
|
| 836 |
-
"eval_top_5_accuracy": 0.
|
| 837 |
"step": 5407
|
| 838 |
},
|
| 839 |
{
|
| 840 |
-
"epoch": 30.
|
| 841 |
-
"grad_norm":
|
| 842 |
"learning_rate": 4.707561728395062e-05,
|
| 843 |
-
"loss": 0.
|
| 844 |
"step": 5500
|
| 845 |
},
|
| 846 |
{
|
| 847 |
-
"epoch": 30.
|
| 848 |
-
"eval_accuracy": 0.
|
| 849 |
-
"eval_f1": 0.
|
| 850 |
-
"eval_loss":
|
| 851 |
-
"eval_precision": 0.
|
| 852 |
-
"eval_recall": 0.
|
| 853 |
-
"eval_runtime":
|
| 854 |
-
"eval_samples_per_second": 5.
|
| 855 |
-
"eval_steps_per_second": 2.
|
| 856 |
-
"eval_top_10_accuracy": 0.
|
| 857 |
-
"eval_top_1_accuracy": 0.
|
| 858 |
-
"eval_top_5_accuracy": 0.
|
| 859 |
-
"step":
|
| 860 |
-
},
|
| 861 |
-
{
|
| 862 |
-
"epoch": 31.
|
| 863 |
-
"grad_norm":
|
| 864 |
-
"learning_rate": 4.
|
| 865 |
-
"loss": 0.
|
| 866 |
"step": 5600
|
| 867 |
},
|
| 868 |
{
|
| 869 |
-
"epoch": 31.
|
| 870 |
-
"grad_norm":
|
| 871 |
-
"learning_rate": 4.
|
| 872 |
-
"loss": 0.
|
| 873 |
"step": 5700
|
| 874 |
},
|
| 875 |
{
|
| 876 |
-
"epoch": 31.
|
| 877 |
-
"eval_accuracy": 0.
|
| 878 |
-
"eval_f1": 0.
|
| 879 |
-
"eval_loss": 2.
|
| 880 |
-
"eval_precision": 0.
|
| 881 |
-
"eval_recall": 0.
|
| 882 |
-
"eval_runtime":
|
| 883 |
-
"eval_samples_per_second":
|
| 884 |
-
"eval_steps_per_second":
|
| 885 |
-
"eval_top_10_accuracy": 0.
|
| 886 |
-
"eval_top_1_accuracy": 0.
|
| 887 |
-
"eval_top_5_accuracy": 0.
|
| 888 |
"step": 5768
|
| 889 |
},
|
| 890 |
{
|
| 891 |
-
"epoch": 32.
|
| 892 |
-
"grad_norm": 1.
|
| 893 |
-
"learning_rate": 4.
|
| 894 |
-
"loss": 0.
|
| 895 |
"step": 5800
|
| 896 |
},
|
| 897 |
{
|
| 898 |
-
"epoch": 32.
|
| 899 |
-
"grad_norm":
|
| 900 |
-
"learning_rate": 4.
|
| 901 |
-
"loss": 0.
|
| 902 |
"step": 5900
|
| 903 |
},
|
| 904 |
{
|
| 905 |
-
"epoch": 32.
|
| 906 |
-
"eval_accuracy": 0.
|
| 907 |
-
"eval_f1": 0.
|
| 908 |
-
"eval_loss":
|
| 909 |
-
"eval_precision": 0.
|
| 910 |
-
"eval_recall": 0.
|
| 911 |
-
"eval_runtime":
|
| 912 |
-
"eval_samples_per_second": 5.
|
| 913 |
-
"eval_steps_per_second": 2.
|
| 914 |
-
"eval_top_10_accuracy": 0.
|
| 915 |
-
"eval_top_1_accuracy": 0.
|
| 916 |
"eval_top_5_accuracy": 0.8136094674556213,
|
| 917 |
"step": 5948
|
| 918 |
},
|
| 919 |
{
|
| 920 |
-
"epoch": 33.
|
| 921 |
-
"grad_norm":
|
| 922 |
-
"learning_rate": 4.
|
| 923 |
-
"loss": 0.
|
| 924 |
"step": 6000
|
| 925 |
},
|
| 926 |
{
|
| 927 |
-
"epoch": 33.
|
| 928 |
-
"grad_norm":
|
| 929 |
-
"learning_rate": 4.
|
| 930 |
-
"loss": 0.
|
| 931 |
"step": 6100
|
| 932 |
},
|
| 933 |
{
|
| 934 |
-
"epoch": 33.
|
| 935 |
-
"eval_accuracy": 0.
|
| 936 |
-
"eval_f1": 0.
|
| 937 |
-
"eval_loss":
|
| 938 |
-
"eval_precision": 0.
|
| 939 |
-
"eval_recall": 0.
|
| 940 |
-
"eval_runtime":
|
| 941 |
-
"eval_samples_per_second": 5.
|
| 942 |
-
"eval_steps_per_second": 2.
|
| 943 |
-
"eval_top_10_accuracy": 0.
|
| 944 |
-
"eval_top_1_accuracy": 0.
|
| 945 |
-
"eval_top_5_accuracy": 0.
|
| 946 |
"step": 6128
|
| 947 |
},
|
| 948 |
{
|
| 949 |
-
"epoch": 34.
|
| 950 |
-
"grad_norm":
|
| 951 |
"learning_rate": 4.599691358024691e-05,
|
| 952 |
-
"loss": 0.
|
| 953 |
"step": 6200
|
| 954 |
},
|
| 955 |
{
|
| 956 |
-
"epoch": 34.
|
| 957 |
-
"grad_norm":
|
| 958 |
"learning_rate": 4.584259259259259e-05,
|
| 959 |
-
"loss": 0.
|
| 960 |
"step": 6300
|
| 961 |
},
|
| 962 |
{
|
| 963 |
-
"epoch": 34.
|
| 964 |
-
"eval_accuracy": 0.
|
| 965 |
-
"eval_f1": 0.
|
| 966 |
-
"eval_loss":
|
| 967 |
-
"eval_precision": 0.
|
| 968 |
-
"eval_recall": 0.
|
| 969 |
-
"eval_runtime":
|
| 970 |
-
"eval_samples_per_second":
|
| 971 |
-
"eval_steps_per_second": 2.
|
| 972 |
-
"eval_top_10_accuracy": 0.
|
| 973 |
-
"eval_top_1_accuracy": 0.
|
| 974 |
-
"eval_top_5_accuracy": 0.
|
| 975 |
-
"step":
|
| 976 |
},
|
| 977 |
{
|
| 978 |
-
"epoch": 35.
|
| 979 |
-
"grad_norm":
|
| 980 |
"learning_rate": 4.5688271604938275e-05,
|
| 981 |
-
"loss": 0.
|
| 982 |
"step": 6400
|
| 983 |
},
|
| 984 |
{
|
| 985 |
-
"epoch": 35.
|
| 986 |
-
"eval_accuracy": 0.
|
| 987 |
-
"eval_f1": 0.
|
| 988 |
-
"eval_loss":
|
| 989 |
-
"eval_precision": 0.
|
| 990 |
-
"eval_recall": 0.
|
| 991 |
-
"eval_runtime":
|
| 992 |
-
"eval_samples_per_second": 5.
|
| 993 |
-
"eval_steps_per_second": 2.
|
| 994 |
-
"eval_top_10_accuracy": 0.
|
| 995 |
-
"eval_top_1_accuracy": 0.
|
| 996 |
-
"eval_top_5_accuracy": 0.
|
| 997 |
"step": 6489
|
| 998 |
},
|
| 999 |
{
|
| 1000 |
-
"epoch": 36.
|
| 1001 |
-
"grad_norm": 0.
|
| 1002 |
-
"learning_rate": 4.
|
| 1003 |
-
"loss": 0.
|
| 1004 |
"step": 6500
|
| 1005 |
},
|
| 1006 |
{
|
| 1007 |
-
"epoch": 36.
|
| 1008 |
-
"grad_norm":
|
| 1009 |
-
"learning_rate": 4.
|
| 1010 |
-
"loss": 0.
|
| 1011 |
"step": 6600
|
| 1012 |
},
|
| 1013 |
{
|
| 1014 |
-
"epoch": 36.
|
| 1015 |
-
"eval_accuracy": 0.
|
| 1016 |
-
"eval_f1": 0.
|
| 1017 |
-
"eval_loss": 2.
|
| 1018 |
-
"eval_precision": 0.
|
| 1019 |
-
"eval_recall": 0.
|
| 1020 |
-
"eval_runtime":
|
| 1021 |
-
"eval_samples_per_second": 5.
|
| 1022 |
-
"eval_steps_per_second": 2.
|
| 1023 |
-
"eval_top_10_accuracy": 0.
|
| 1024 |
-
"eval_top_1_accuracy": 0.
|
| 1025 |
-
"eval_top_5_accuracy": 0.
|
| 1026 |
"step": 6669
|
| 1027 |
},
|
| 1028 |
{
|
| 1029 |
-
"epoch": 37.
|
| 1030 |
-
"grad_norm":
|
| 1031 |
-
"learning_rate": 4.
|
| 1032 |
-
"loss": 0.
|
| 1033 |
"step": 6700
|
| 1034 |
},
|
| 1035 |
{
|
| 1036 |
-
"epoch": 37.
|
| 1037 |
-
"grad_norm": 0.
|
| 1038 |
-
"learning_rate": 4.
|
| 1039 |
-
"loss": 0.
|
| 1040 |
"step": 6800
|
| 1041 |
},
|
| 1042 |
{
|
| 1043 |
-
"epoch": 37.
|
| 1044 |
-
"eval_accuracy": 0.
|
| 1045 |
-
"eval_f1": 0.
|
| 1046 |
-
"eval_loss":
|
| 1047 |
-
"eval_precision": 0.
|
| 1048 |
-
"eval_recall": 0.
|
| 1049 |
-
"eval_runtime":
|
| 1050 |
-
"eval_samples_per_second": 5.
|
| 1051 |
-
"eval_steps_per_second": 2.
|
| 1052 |
-
"eval_top_10_accuracy": 0.
|
| 1053 |
-
"eval_top_1_accuracy": 0.
|
| 1054 |
-
"eval_top_5_accuracy": 0.
|
| 1055 |
"step": 6849
|
| 1056 |
},
|
| 1057 |
{
|
| 1058 |
-
"epoch": 38.
|
| 1059 |
-
"grad_norm":
|
| 1060 |
-
"learning_rate": 4.
|
| 1061 |
-
"loss": 0.
|
| 1062 |
"step": 6900
|
| 1063 |
},
|
| 1064 |
{
|
| 1065 |
-
"epoch": 38.
|
| 1066 |
-
"grad_norm":
|
| 1067 |
-
"learning_rate": 4.
|
| 1068 |
-
"loss": 0.
|
| 1069 |
"step": 7000
|
| 1070 |
},
|
| 1071 |
{
|
| 1072 |
-
"epoch": 38.
|
| 1073 |
-
"eval_accuracy": 0.
|
| 1074 |
-
"eval_f1": 0.
|
| 1075 |
-
"eval_loss":
|
| 1076 |
-
"eval_precision": 0.
|
| 1077 |
-
"eval_recall": 0.
|
| 1078 |
-
"eval_runtime":
|
| 1079 |
-
"eval_samples_per_second":
|
| 1080 |
-
"eval_steps_per_second":
|
| 1081 |
-
"eval_top_10_accuracy": 0.
|
| 1082 |
-
"eval_top_1_accuracy": 0.
|
| 1083 |
-
"eval_top_5_accuracy": 0.
|
| 1084 |
-
"step":
|
| 1085 |
-
},
|
| 1086 |
-
{
|
| 1087 |
-
"epoch": 39.
|
| 1088 |
-
"grad_norm":
|
| 1089 |
-
"learning_rate": 4.
|
| 1090 |
-
"loss": 0.
|
| 1091 |
"step": 7100
|
| 1092 |
},
|
| 1093 |
{
|
| 1094 |
-
"epoch": 39.
|
| 1095 |
-
"grad_norm":
|
| 1096 |
-
"learning_rate": 4.
|
| 1097 |
-
"loss": 0.
|
| 1098 |
"step": 7200
|
| 1099 |
},
|
| 1100 |
{
|
| 1101 |
-
"epoch": 39.
|
| 1102 |
-
"eval_accuracy": 0.
|
| 1103 |
-
"eval_f1": 0.
|
| 1104 |
-
"eval_loss":
|
| 1105 |
-
"eval_precision": 0.
|
| 1106 |
-
"eval_recall": 0.
|
| 1107 |
-
"eval_runtime":
|
| 1108 |
-
"eval_samples_per_second": 5.
|
| 1109 |
-
"eval_steps_per_second": 2.
|
| 1110 |
-
"eval_top_10_accuracy": 0.
|
| 1111 |
-
"eval_top_1_accuracy": 0.
|
| 1112 |
-
"eval_top_5_accuracy": 0.
|
| 1113 |
"step": 7210
|
| 1114 |
},
|
| 1115 |
{
|
| 1116 |
-
"epoch": 40.
|
| 1117 |
-
"grad_norm": 0.
|
| 1118 |
-
"learning_rate": 4.
|
| 1119 |
-
"loss": 0.
|
| 1120 |
"step": 7300
|
| 1121 |
},
|
| 1122 |
{
|
| 1123 |
-
"epoch": 40.
|
| 1124 |
-
"eval_accuracy": 0.
|
| 1125 |
-
"eval_f1": 0.
|
| 1126 |
-
"eval_loss": 2.
|
| 1127 |
-
"eval_precision": 0.
|
| 1128 |
-
"eval_recall": 0.
|
| 1129 |
-
"eval_runtime":
|
| 1130 |
-
"eval_samples_per_second": 5.
|
| 1131 |
-
"eval_steps_per_second": 2.
|
| 1132 |
-
"eval_top_10_accuracy": 0.
|
| 1133 |
-
"eval_top_1_accuracy": 0.
|
| 1134 |
-
"eval_top_5_accuracy": 0.
|
| 1135 |
"step": 7390
|
| 1136 |
},
|
| 1137 |
{
|
| 1138 |
-
"epoch": 41.
|
| 1139 |
-
"grad_norm": 0.
|
| 1140 |
-
"learning_rate": 4.
|
| 1141 |
-
"loss": 0.
|
| 1142 |
"step": 7400
|
| 1143 |
},
|
| 1144 |
{
|
| 1145 |
-
"epoch": 41.
|
| 1146 |
-
"grad_norm":
|
| 1147 |
-
"learning_rate": 4.
|
| 1148 |
-
"loss": 0.
|
| 1149 |
"step": 7500
|
| 1150 |
},
|
| 1151 |
{
|
| 1152 |
-
"epoch": 41.
|
| 1153 |
-
"eval_accuracy": 0.
|
| 1154 |
-
"eval_f1": 0.
|
| 1155 |
-
"eval_loss": 2.
|
| 1156 |
-
"eval_precision": 0.
|
| 1157 |
-
"eval_recall": 0.
|
| 1158 |
-
"eval_runtime":
|
| 1159 |
-
"eval_samples_per_second":
|
| 1160 |
-
"eval_steps_per_second":
|
| 1161 |
-
"eval_top_10_accuracy": 0.
|
| 1162 |
-
"eval_top_1_accuracy": 0.
|
| 1163 |
-
"eval_top_5_accuracy": 0.
|
| 1164 |
"step": 7570
|
| 1165 |
},
|
| 1166 |
{
|
| 1167 |
-
"epoch": 42.
|
| 1168 |
-
"grad_norm":
|
| 1169 |
-
"learning_rate": 4.
|
| 1170 |
-
"loss": 0.
|
| 1171 |
"step": 7600
|
| 1172 |
},
|
| 1173 |
{
|
| 1174 |
-
"epoch": 42.
|
| 1175 |
-
"grad_norm": 0.
|
| 1176 |
-
"learning_rate": 4.
|
| 1177 |
-
"loss": 0.
|
| 1178 |
"step": 7700
|
| 1179 |
},
|
| 1180 |
{
|
| 1181 |
-
"epoch": 42.
|
| 1182 |
-
"eval_accuracy": 0.
|
| 1183 |
-
"eval_f1": 0.
|
| 1184 |
-
"eval_loss": 2.
|
| 1185 |
-
"eval_precision": 0.
|
| 1186 |
-
"eval_recall": 0.
|
| 1187 |
-
"eval_runtime":
|
| 1188 |
-
"eval_samples_per_second": 5.
|
| 1189 |
-
"eval_steps_per_second": 2.
|
| 1190 |
-
"eval_top_10_accuracy": 0.
|
| 1191 |
-
"eval_top_1_accuracy": 0.
|
| 1192 |
-
"eval_top_5_accuracy": 0.
|
| 1193 |
-
"step":
|
| 1194 |
-
},
|
| 1195 |
-
{
|
| 1196 |
-
"epoch": 43.
|
| 1197 |
-
"grad_norm": 0.
|
| 1198 |
-
"learning_rate": 4.
|
| 1199 |
-
"loss": 0.
|
| 1200 |
"step": 7800
|
| 1201 |
},
|
| 1202 |
{
|
| 1203 |
-
"epoch": 43.
|
| 1204 |
-
"grad_norm": 0.
|
| 1205 |
-
"learning_rate": 4.
|
| 1206 |
-
"loss": 0.
|
| 1207 |
"step": 7900
|
| 1208 |
},
|
| 1209 |
{
|
| 1210 |
-
"epoch": 43.
|
| 1211 |
-
"eval_accuracy": 0.
|
| 1212 |
-
"eval_f1": 0.
|
| 1213 |
-
"eval_loss": 2.
|
| 1214 |
-
"eval_precision": 0.
|
| 1215 |
-
"eval_recall": 0.
|
| 1216 |
-
"eval_runtime":
|
| 1217 |
-
"eval_samples_per_second": 4.
|
| 1218 |
-
"eval_steps_per_second": 2.
|
| 1219 |
-
"eval_top_10_accuracy": 0.
|
| 1220 |
-
"eval_top_1_accuracy": 0.
|
| 1221 |
-
"eval_top_5_accuracy": 0.
|
| 1222 |
"step": 7931
|
| 1223 |
},
|
| 1224 |
{
|
| 1225 |
-
"epoch": 44.
|
| 1226 |
-
"grad_norm":
|
| 1227 |
"learning_rate": 4.322067901234568e-05,
|
| 1228 |
-
"loss": 0.
|
| 1229 |
"step": 8000
|
| 1230 |
},
|
| 1231 |
{
|
| 1232 |
-
"epoch": 44.
|
| 1233 |
-
"grad_norm":
|
| 1234 |
"learning_rate": 4.306635802469136e-05,
|
| 1235 |
-
"loss": 0.
|
| 1236 |
"step": 8100
|
| 1237 |
},
|
| 1238 |
{
|
| 1239 |
-
"epoch": 44.
|
| 1240 |
-
"eval_accuracy": 0.
|
| 1241 |
-
"eval_f1": 0.
|
| 1242 |
-
"eval_loss": 2.
|
| 1243 |
-
"eval_precision": 0.
|
| 1244 |
-
"eval_recall": 0.
|
| 1245 |
-
"eval_runtime":
|
| 1246 |
-
"eval_samples_per_second":
|
| 1247 |
-
"eval_steps_per_second": 2.
|
| 1248 |
-
"eval_top_10_accuracy": 0.
|
| 1249 |
-
"eval_top_1_accuracy": 0.
|
| 1250 |
-
"eval_top_5_accuracy": 0.
|
| 1251 |
"step": 8111
|
| 1252 |
},
|
| 1253 |
{
|
| 1254 |
-
"epoch": 45.
|
| 1255 |
-
"grad_norm":
|
| 1256 |
"learning_rate": 4.291203703703704e-05,
|
| 1257 |
-
"loss": 0.
|
| 1258 |
"step": 8200
|
| 1259 |
},
|
| 1260 |
{
|
| 1261 |
-
"epoch": 45.
|
| 1262 |
-
"eval_accuracy": 0.
|
| 1263 |
-
"eval_f1": 0.
|
| 1264 |
-
"eval_loss": 2.
|
| 1265 |
-
"eval_precision": 0.
|
| 1266 |
-
"eval_recall": 0.
|
| 1267 |
-
"eval_runtime":
|
| 1268 |
-
"eval_samples_per_second": 5.
|
| 1269 |
-
"eval_steps_per_second": 2.
|
| 1270 |
-
"eval_top_10_accuracy": 0.
|
| 1271 |
-
"eval_top_1_accuracy": 0.
|
| 1272 |
-
"eval_top_5_accuracy": 0.
|
| 1273 |
"step": 8291
|
| 1274 |
},
|
| 1275 |
{
|
| 1276 |
-
"epoch": 46.
|
| 1277 |
-
"grad_norm":
|
| 1278 |
"learning_rate": 4.2757716049382715e-05,
|
| 1279 |
-
"loss": 0.
|
| 1280 |
"step": 8300
|
| 1281 |
},
|
| 1282 |
{
|
| 1283 |
-
"epoch": 46.
|
| 1284 |
-
"grad_norm": 0.
|
| 1285 |
-
"learning_rate": 4.
|
| 1286 |
-
"loss": 0.
|
| 1287 |
"step": 8400
|
| 1288 |
},
|
| 1289 |
{
|
| 1290 |
-
"epoch": 46.
|
| 1291 |
-
"eval_accuracy": 0.
|
| 1292 |
-
"eval_f1": 0.
|
| 1293 |
-
"eval_loss": 2.
|
| 1294 |
-
"eval_precision": 0.
|
| 1295 |
-
"eval_recall": 0.
|
| 1296 |
-
"eval_runtime":
|
| 1297 |
-
"eval_samples_per_second": 5.
|
| 1298 |
-
"eval_steps_per_second": 2.
|
| 1299 |
-
"eval_top_10_accuracy": 0.
|
| 1300 |
-
"eval_top_1_accuracy": 0.
|
| 1301 |
-
"eval_top_5_accuracy": 0.
|
| 1302 |
-
"step":
|
| 1303 |
},
|
| 1304 |
{
|
| 1305 |
-
"epoch": 47.
|
| 1306 |
-
"grad_norm":
|
| 1307 |
-
"learning_rate": 4.
|
| 1308 |
-
"loss": 0.
|
| 1309 |
"step": 8500
|
| 1310 |
},
|
| 1311 |
{
|
| 1312 |
-
"epoch": 47.
|
| 1313 |
-
"grad_norm":
|
| 1314 |
-
"learning_rate": 4.
|
| 1315 |
-
"loss": 0.
|
| 1316 |
"step": 8600
|
| 1317 |
},
|
| 1318 |
{
|
| 1319 |
-
"epoch": 47.
|
| 1320 |
-
"eval_accuracy": 0.
|
| 1321 |
-
"eval_f1": 0.
|
| 1322 |
-
"eval_loss": 2.
|
| 1323 |
-
"eval_precision": 0.
|
| 1324 |
-
"eval_recall": 0.
|
| 1325 |
-
"eval_runtime":
|
| 1326 |
-
"eval_samples_per_second":
|
| 1327 |
-
"eval_steps_per_second":
|
| 1328 |
-
"eval_top_10_accuracy": 0.
|
| 1329 |
-
"eval_top_1_accuracy": 0.
|
| 1330 |
-
"eval_top_5_accuracy": 0.
|
| 1331 |
"step": 8652
|
| 1332 |
},
|
| 1333 |
{
|
| 1334 |
-
"epoch": 48.
|
| 1335 |
-
"grad_norm": 0.
|
| 1336 |
-
"learning_rate": 4.
|
| 1337 |
-
"loss": 0.
|
| 1338 |
"step": 8700
|
| 1339 |
},
|
| 1340 |
{
|
| 1341 |
-
"epoch": 48.
|
| 1342 |
-
"grad_norm":
|
| 1343 |
-
"learning_rate": 4.
|
| 1344 |
-
"loss": 0.
|
| 1345 |
"step": 8800
|
| 1346 |
},
|
| 1347 |
{
|
| 1348 |
-
"epoch": 48.
|
| 1349 |
-
"eval_accuracy": 0.
|
| 1350 |
-
"eval_f1": 0.
|
| 1351 |
-
"eval_loss": 2.
|
| 1352 |
-
"eval_precision": 0.
|
| 1353 |
-
"eval_recall": 0.
|
| 1354 |
-
"eval_runtime":
|
| 1355 |
-
"eval_samples_per_second": 5.
|
| 1356 |
-
"eval_steps_per_second": 2.
|
| 1357 |
-
"eval_top_10_accuracy": 0.
|
| 1358 |
-
"eval_top_1_accuracy": 0.
|
| 1359 |
-
"eval_top_5_accuracy": 0.
|
| 1360 |
"step": 8832
|
| 1361 |
},
|
| 1362 |
{
|
| 1363 |
-
"epoch": 49.
|
| 1364 |
-
"grad_norm":
|
| 1365 |
"learning_rate": 4.183333333333334e-05,
|
| 1366 |
-
"loss": 0.
|
| 1367 |
"step": 8900
|
| 1368 |
},
|
| 1369 |
{
|
| 1370 |
-
"epoch": 49.
|
| 1371 |
-
"grad_norm":
|
| 1372 |
"learning_rate": 4.1679012345679016e-05,
|
| 1373 |
-
"loss": 0.
|
| 1374 |
"step": 9000
|
| 1375 |
},
|
| 1376 |
{
|
| 1377 |
-
"epoch": 49.
|
| 1378 |
-
"eval_accuracy": 0.
|
| 1379 |
-
"eval_f1": 0.
|
| 1380 |
-
"eval_loss": 2.
|
| 1381 |
-
"eval_precision": 0.
|
| 1382 |
-
"eval_recall": 0.
|
| 1383 |
-
"eval_runtime":
|
| 1384 |
-
"eval_samples_per_second": 5.
|
| 1385 |
-
"eval_steps_per_second": 2.
|
| 1386 |
-
"eval_top_10_accuracy": 0.
|
| 1387 |
-
"eval_top_1_accuracy": 0.
|
| 1388 |
-
"eval_top_5_accuracy": 0.
|
| 1389 |
"step": 9012
|
| 1390 |
},
|
| 1391 |
{
|
| 1392 |
-
"epoch": 50.
|
| 1393 |
-
"grad_norm":
|
| 1394 |
"learning_rate": 4.1524691358024694e-05,
|
| 1395 |
-
"loss": 0.
|
| 1396 |
"step": 9100
|
| 1397 |
},
|
| 1398 |
{
|
| 1399 |
-
"epoch": 50.
|
| 1400 |
-
"eval_accuracy": 0.
|
| 1401 |
-
"eval_f1": 0.
|
| 1402 |
-
"eval_loss": 2.
|
| 1403 |
-
"eval_precision": 0.
|
| 1404 |
-
"eval_recall": 0.
|
| 1405 |
-
"eval_runtime":
|
| 1406 |
-
"eval_samples_per_second":
|
| 1407 |
-
"eval_steps_per_second":
|
| 1408 |
-
"eval_top_10_accuracy": 0.
|
| 1409 |
-
"eval_top_1_accuracy": 0.
|
| 1410 |
-
"eval_top_5_accuracy": 0.
|
| 1411 |
-
"step":
|
| 1412 |
},
|
| 1413 |
{
|
| 1414 |
-
"epoch": 51.
|
| 1415 |
-
"grad_norm":
|
| 1416 |
"learning_rate": 4.137037037037037e-05,
|
| 1417 |
-
"loss": 0.
|
| 1418 |
"step": 9200
|
| 1419 |
},
|
| 1420 |
{
|
| 1421 |
-
"epoch": 51.
|
| 1422 |
-
"grad_norm": 0.
|
| 1423 |
"learning_rate": 4.121604938271605e-05,
|
| 1424 |
-
"loss": 0.
|
| 1425 |
"step": 9300
|
| 1426 |
},
|
| 1427 |
{
|
| 1428 |
-
"epoch": 51.
|
| 1429 |
-
"eval_accuracy": 0.
|
| 1430 |
-
"eval_f1": 0.
|
| 1431 |
-
"eval_loss":
|
| 1432 |
-
"eval_precision": 0.
|
| 1433 |
-
"eval_recall": 0.
|
| 1434 |
-
"eval_runtime":
|
| 1435 |
-
"eval_samples_per_second": 5.
|
| 1436 |
-
"eval_steps_per_second": 2.
|
| 1437 |
-
"eval_top_10_accuracy": 0.
|
| 1438 |
-
"eval_top_1_accuracy": 0.
|
| 1439 |
-
"eval_top_5_accuracy": 0.
|
| 1440 |
"step": 9373
|
| 1441 |
},
|
| 1442 |
{
|
| 1443 |
-
"epoch": 52.
|
| 1444 |
-
"grad_norm":
|
| 1445 |
"learning_rate": 4.1061728395061733e-05,
|
| 1446 |
-
"loss": 0.
|
| 1447 |
"step": 9400
|
| 1448 |
},
|
| 1449 |
{
|
| 1450 |
-
"epoch": 52.
|
| 1451 |
-
"grad_norm":
|
| 1452 |
"learning_rate": 4.090740740740741e-05,
|
| 1453 |
-
"loss": 0.
|
| 1454 |
"step": 9500
|
| 1455 |
},
|
| 1456 |
{
|
| 1457 |
-
"epoch": 52.
|
| 1458 |
-
"eval_accuracy": 0.
|
| 1459 |
-
"eval_f1": 0.
|
| 1460 |
-
"eval_loss": 2.
|
| 1461 |
-
"eval_precision": 0.
|
| 1462 |
-
"eval_recall": 0.
|
| 1463 |
-
"eval_runtime":
|
| 1464 |
-
"eval_samples_per_second":
|
| 1465 |
-
"eval_steps_per_second": 2.
|
| 1466 |
-
"eval_top_10_accuracy": 0.
|
| 1467 |
-
"eval_top_1_accuracy": 0.
|
| 1468 |
-
"eval_top_5_accuracy": 0.
|
| 1469 |
"step": 9553
|
| 1470 |
},
|
| 1471 |
{
|
| 1472 |
-
"epoch": 53.
|
| 1473 |
-
"grad_norm": 0.
|
| 1474 |
"learning_rate": 4.075308641975309e-05,
|
| 1475 |
-
"loss": 0.
|
| 1476 |
"step": 9600
|
| 1477 |
},
|
| 1478 |
{
|
| 1479 |
-
"epoch": 53.
|
| 1480 |
-
"grad_norm": 0.
|
| 1481 |
"learning_rate": 4.0598765432098767e-05,
|
| 1482 |
-
"loss": 0.
|
| 1483 |
"step": 9700
|
| 1484 |
},
|
| 1485 |
{
|
| 1486 |
-
"epoch": 53.
|
| 1487 |
-
"eval_accuracy": 0.
|
| 1488 |
-
"eval_f1": 0.
|
| 1489 |
-
"eval_loss":
|
| 1490 |
-
"eval_precision": 0.
|
| 1491 |
-
"eval_recall": 0.
|
| 1492 |
-
"eval_runtime":
|
| 1493 |
-
"eval_samples_per_second": 4.
|
| 1494 |
-
"eval_steps_per_second": 2.
|
| 1495 |
-
"eval_top_10_accuracy": 0.
|
| 1496 |
-
"eval_top_1_accuracy": 0.
|
| 1497 |
-
"eval_top_5_accuracy": 0.
|
| 1498 |
"step": 9733
|
| 1499 |
},
|
| 1500 |
{
|
| 1501 |
-
"epoch": 54.
|
| 1502 |
-
"grad_norm":
|
| 1503 |
"learning_rate": 4.0444444444444444e-05,
|
| 1504 |
-
"loss": 0.
|
| 1505 |
"step": 9800
|
| 1506 |
},
|
| 1507 |
{
|
| 1508 |
-
"epoch": 54.
|
| 1509 |
-
"grad_norm":
|
| 1510 |
"learning_rate": 4.029012345679012e-05,
|
| 1511 |
-
"loss": 0.
|
| 1512 |
"step": 9900
|
| 1513 |
},
|
| 1514 |
{
|
| 1515 |
-
"epoch": 54.
|
| 1516 |
-
"eval_accuracy": 0.
|
| 1517 |
-
"eval_f1": 0.
|
| 1518 |
-
"eval_loss": 2.
|
| 1519 |
-
"eval_precision": 0.
|
| 1520 |
-
"eval_recall": 0.
|
| 1521 |
-
"eval_runtime":
|
| 1522 |
-
"eval_samples_per_second": 5.
|
| 1523 |
-
"eval_steps_per_second": 2.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1524 |
"eval_top_10_accuracy": 0.8698224852071006,
|
| 1525 |
-
"eval_top_1_accuracy": 0.
|
| 1526 |
-
"eval_top_5_accuracy": 0.
|
| 1527 |
-
"step":
|
| 1528 |
},
|
| 1529 |
{
|
| 1530 |
-
"epoch":
|
| 1531 |
-
"step":
|
| 1532 |
-
"total_flos":
|
| 1533 |
-
"train_loss": 5.
|
| 1534 |
-
"train_runtime":
|
| 1535 |
-
"train_samples_per_second":
|
| 1536 |
-
"train_steps_per_second":
|
| 1537 |
}
|
| 1538 |
],
|
| 1539 |
"logging_steps": 100,
|
|
@@ -1562,7 +1671,7 @@
|
|
| 1562 |
"attributes": {}
|
| 1563 |
}
|
| 1564 |
},
|
| 1565 |
-
"total_flos":
|
| 1566 |
"train_batch_size": 2,
|
| 1567 |
"trial_name": null,
|
| 1568 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_metric": 0.6301775147928994,
|
| 3 |
+
"best_model_checkpoint": "/media/cse/HDD/Shawon/shawon/MY DATA/VideoMAE_Base_WLASL_100_200_epochs_p20_SR_8/checkpoint-7030",
|
| 4 |
+
"epoch": 58.005006944444446,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 10635,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
| 10 |
"log_history": [
|
| 11 |
{
|
| 12 |
"epoch": 0.002777777777777778,
|
| 13 |
+
"grad_norm": 52.335811614990234,
|
| 14 |
"learning_rate": 1.3333333333333334e-06,
|
| 15 |
+
"loss": 18.5755,
|
| 16 |
"step": 100
|
| 17 |
},
|
| 18 |
{
|
| 19 |
"epoch": 0.005,
|
| 20 |
"eval_accuracy": 0.008875739644970414,
|
| 21 |
+
"eval_f1": 0.0009576475529349738,
|
| 22 |
+
"eval_loss": 4.641324043273926,
|
| 23 |
+
"eval_precision": 0.0005193342938593639,
|
| 24 |
"eval_recall": 0.008875739644970414,
|
| 25 |
+
"eval_runtime": 181.5799,
|
| 26 |
+
"eval_samples_per_second": 1.861,
|
| 27 |
+
"eval_steps_per_second": 0.931,
|
| 28 |
+
"eval_top_10_accuracy": 0.10059171597633136,
|
| 29 |
"eval_top_1_accuracy": 0.008875739644970414,
|
| 30 |
+
"eval_top_5_accuracy": 0.04437869822485207,
|
| 31 |
"step": 180
|
| 32 |
},
|
| 33 |
{
|
| 34 |
+
"epoch": 1.0005416666666667,
|
| 35 |
+
"grad_norm": 59.252769470214844,
|
| 36 |
"learning_rate": 2.7222222222222224e-06,
|
| 37 |
+
"loss": 18.6055,
|
| 38 |
"step": 200
|
| 39 |
},
|
| 40 |
{
|
| 41 |
+
"epoch": 1.0033194444444444,
|
| 42 |
+
"grad_norm": 52.176109313964844,
|
| 43 |
"learning_rate": 4.111111111111112e-06,
|
| 44 |
+
"loss": 18.4843,
|
| 45 |
"step": 300
|
| 46 |
},
|
| 47 |
{
|
| 48 |
+
"epoch": 1.0049861111111111,
|
| 49 |
+
"eval_accuracy": 0.008875739644970414,
|
| 50 |
+
"eval_f1": 0.0011768857922704076,
|
| 51 |
+
"eval_loss": 4.630766868591309,
|
| 52 |
+
"eval_precision": 0.0006326211365999165,
|
| 53 |
+
"eval_recall": 0.008875739644970414,
|
| 54 |
+
"eval_runtime": 60.2416,
|
| 55 |
+
"eval_samples_per_second": 5.611,
|
| 56 |
+
"eval_steps_per_second": 2.805,
|
| 57 |
+
"eval_top_10_accuracy": 0.10059171597633136,
|
| 58 |
+
"eval_top_1_accuracy": 0.008875739644970414,
|
| 59 |
"eval_top_5_accuracy": 0.0621301775147929,
|
| 60 |
"step": 360
|
| 61 |
},
|
| 62 |
{
|
| 63 |
+
"epoch": 2.0010902777777777,
|
| 64 |
+
"grad_norm": 37.923606872558594,
|
| 65 |
"learning_rate": 5.500000000000001e-06,
|
| 66 |
+
"loss": 18.5315,
|
| 67 |
"step": 400
|
| 68 |
},
|
| 69 |
{
|
| 70 |
+
"epoch": 2.0038680555555555,
|
| 71 |
+
"grad_norm": 38.569297790527344,
|
| 72 |
"learning_rate": 6.888888888888889e-06,
|
| 73 |
+
"loss": 18.4592,
|
| 74 |
"step": 500
|
| 75 |
},
|
| 76 |
{
|
| 77 |
+
"epoch": 2.0050069444444443,
|
| 78 |
+
"eval_accuracy": 0.020710059171597635,
|
| 79 |
+
"eval_f1": 0.002684633160978603,
|
| 80 |
+
"eval_loss": 4.6208624839782715,
|
| 81 |
+
"eval_precision": 0.0016963969626691518,
|
| 82 |
+
"eval_recall": 0.020710059171597635,
|
| 83 |
+
"eval_runtime": 59.58,
|
| 84 |
+
"eval_samples_per_second": 5.673,
|
| 85 |
+
"eval_steps_per_second": 2.837,
|
| 86 |
+
"eval_top_10_accuracy": 0.10650887573964497,
|
| 87 |
+
"eval_top_1_accuracy": 0.020710059171597635,
|
| 88 |
+
"eval_top_5_accuracy": 0.07100591715976332,
|
| 89 |
+
"step": 541
|
| 90 |
},
|
| 91 |
{
|
| 92 |
+
"epoch": 3.001638888888889,
|
| 93 |
+
"grad_norm": 41.79819107055664,
|
| 94 |
"learning_rate": 8.27777777777778e-06,
|
| 95 |
+
"loss": 18.3856,
|
| 96 |
"step": 600
|
| 97 |
},
|
| 98 |
{
|
| 99 |
+
"epoch": 3.0044166666666667,
|
| 100 |
+
"grad_norm": 31.180952072143555,
|
| 101 |
"learning_rate": 9.666666666666667e-06,
|
| 102 |
+
"loss": 18.4604,
|
| 103 |
"step": 700
|
| 104 |
},
|
| 105 |
{
|
| 106 |
+
"epoch": 3.005,
|
| 107 |
+
"eval_accuracy": 0.01775147928994083,
|
| 108 |
+
"eval_f1": 0.0010059742201274729,
|
| 109 |
+
"eval_loss": 4.617129802703857,
|
| 110 |
+
"eval_precision": 0.0005190329473889052,
|
| 111 |
+
"eval_recall": 0.01775147928994083,
|
| 112 |
+
"eval_runtime": 58.1107,
|
| 113 |
+
"eval_samples_per_second": 5.816,
|
| 114 |
+
"eval_steps_per_second": 2.908,
|
| 115 |
+
"eval_top_10_accuracy": 0.12130177514792899,
|
| 116 |
+
"eval_top_1_accuracy": 0.01775147928994083,
|
| 117 |
+
"eval_top_5_accuracy": 0.06804733727810651,
|
| 118 |
"step": 721
|
| 119 |
},
|
| 120 |
{
|
| 121 |
+
"epoch": 4.0021875,
|
| 122 |
+
"grad_norm": 31.53841781616211,
|
| 123 |
"learning_rate": 1.1055555555555556e-05,
|
| 124 |
+
"loss": 18.3655,
|
| 125 |
"step": 800
|
| 126 |
},
|
| 127 |
{
|
| 128 |
+
"epoch": 4.004965277777778,
|
| 129 |
+
"grad_norm": 29.355567932128906,
|
| 130 |
"learning_rate": 1.2444444444444445e-05,
|
| 131 |
+
"loss": 18.4543,
|
| 132 |
"step": 900
|
| 133 |
},
|
| 134 |
{
|
| 135 |
+
"epoch": 4.0049930555555555,
|
| 136 |
+
"eval_accuracy": 0.023668639053254437,
|
| 137 |
+
"eval_f1": 0.0015770532334260577,
|
| 138 |
+
"eval_loss": 4.6106462478637695,
|
| 139 |
+
"eval_precision": 0.0008262008711267436,
|
| 140 |
+
"eval_recall": 0.023668639053254437,
|
| 141 |
+
"eval_runtime": 60.8324,
|
| 142 |
+
"eval_samples_per_second": 5.556,
|
| 143 |
+
"eval_steps_per_second": 2.778,
|
| 144 |
+
"eval_top_10_accuracy": 0.14497041420118342,
|
| 145 |
+
"eval_top_1_accuracy": 0.023668639053254437,
|
| 146 |
+
"eval_top_5_accuracy": 0.08579881656804733,
|
| 147 |
"step": 901
|
| 148 |
},
|
| 149 |
{
|
| 150 |
+
"epoch": 5.002736111111111,
|
| 151 |
+
"grad_norm": 29.539840698242188,
|
| 152 |
"learning_rate": 1.3833333333333334e-05,
|
| 153 |
+
"loss": 18.3384,
|
| 154 |
"step": 1000
|
| 155 |
},
|
| 156 |
{
|
| 157 |
+
"epoch": 5.004986111111111,
|
| 158 |
+
"eval_accuracy": 0.01775147928994083,
|
| 159 |
+
"eval_f1": 0.000667767246016583,
|
| 160 |
+
"eval_loss": 4.619048118591309,
|
| 161 |
+
"eval_precision": 0.00034028394804998393,
|
| 162 |
+
"eval_recall": 0.01775147928994083,
|
| 163 |
+
"eval_runtime": 55.7817,
|
| 164 |
+
"eval_samples_per_second": 6.059,
|
| 165 |
+
"eval_steps_per_second": 3.03,
|
| 166 |
+
"eval_top_10_accuracy": 0.10946745562130178,
|
| 167 |
+
"eval_top_1_accuracy": 0.01775147928994083,
|
| 168 |
+
"eval_top_5_accuracy": 0.06804733727810651,
|
| 169 |
"step": 1081
|
| 170 |
},
|
| 171 |
{
|
| 172 |
+
"epoch": 6.000506944444444,
|
| 173 |
+
"grad_norm": 31.13288116455078,
|
| 174 |
"learning_rate": 1.5222222222222224e-05,
|
| 175 |
+
"loss": 18.3755,
|
| 176 |
"step": 1100
|
| 177 |
},
|
| 178 |
{
|
| 179 |
+
"epoch": 6.003284722222222,
|
| 180 |
+
"grad_norm": 37.07182312011719,
|
| 181 |
"learning_rate": 1.661111111111111e-05,
|
| 182 |
+
"loss": 18.1947,
|
| 183 |
"step": 1200
|
| 184 |
},
|
| 185 |
{
|
| 186 |
+
"epoch": 6.005006944444444,
|
| 187 |
+
"eval_accuracy": 0.029585798816568046,
|
| 188 |
+
"eval_f1": 0.005653493576969065,
|
| 189 |
+
"eval_loss": 4.60960054397583,
|
| 190 |
+
"eval_precision": 0.0037760128257865815,
|
| 191 |
+
"eval_recall": 0.029585798816568046,
|
| 192 |
+
"eval_runtime": 73.4506,
|
| 193 |
+
"eval_samples_per_second": 4.602,
|
| 194 |
+
"eval_steps_per_second": 2.301,
|
| 195 |
+
"eval_top_10_accuracy": 0.13609467455621302,
|
| 196 |
+
"eval_top_1_accuracy": 0.029585798816568046,
|
| 197 |
+
"eval_top_5_accuracy": 0.09171597633136094,
|
| 198 |
+
"step": 1262
|
| 199 |
},
|
| 200 |
{
|
| 201 |
+
"epoch": 7.001055555555555,
|
| 202 |
+
"grad_norm": 32.610511779785156,
|
| 203 |
"learning_rate": 1.8e-05,
|
| 204 |
+
"loss": 18.3084,
|
| 205 |
"step": 1300
|
| 206 |
},
|
| 207 |
{
|
| 208 |
+
"epoch": 7.003833333333334,
|
| 209 |
+
"grad_norm": 31.119285583496094,
|
| 210 |
"learning_rate": 1.938888888888889e-05,
|
| 211 |
+
"loss": 18.1104,
|
| 212 |
"step": 1400
|
| 213 |
},
|
| 214 |
{
|
| 215 |
+
"epoch": 7.005,
|
| 216 |
+
"eval_accuracy": 0.03550295857988166,
|
| 217 |
+
"eval_f1": 0.008167727362359187,
|
| 218 |
+
"eval_loss": 4.599233627319336,
|
| 219 |
+
"eval_precision": 0.006001785543098896,
|
| 220 |
+
"eval_recall": 0.03550295857988166,
|
| 221 |
+
"eval_runtime": 68.5452,
|
| 222 |
+
"eval_samples_per_second": 4.931,
|
| 223 |
+
"eval_steps_per_second": 2.466,
|
| 224 |
+
"eval_top_10_accuracy": 0.14497041420118342,
|
| 225 |
+
"eval_top_1_accuracy": 0.03550295857988166,
|
| 226 |
+
"eval_top_5_accuracy": 0.09467455621301775,
|
| 227 |
"step": 1442
|
| 228 |
},
|
| 229 |
{
|
| 230 |
+
"epoch": 8.001604166666667,
|
| 231 |
+
"grad_norm": 29.667177200317383,
|
| 232 |
"learning_rate": 2.077777777777778e-05,
|
| 233 |
+
"loss": 18.0862,
|
| 234 |
"step": 1500
|
| 235 |
},
|
| 236 |
{
|
| 237 |
+
"epoch": 8.004381944444445,
|
| 238 |
+
"grad_norm": 31.353796005249023,
|
| 239 |
"learning_rate": 2.216666666666667e-05,
|
| 240 |
+
"loss": 18.0825,
|
| 241 |
"step": 1600
|
| 242 |
},
|
| 243 |
{
|
| 244 |
+
"epoch": 8.004993055555556,
|
| 245 |
+
"eval_accuracy": 0.01775147928994083,
|
| 246 |
+
"eval_f1": 0.008027781632833297,
|
| 247 |
+
"eval_loss": 4.593963623046875,
|
| 248 |
+
"eval_precision": 0.005914603105136552,
|
| 249 |
+
"eval_recall": 0.01775147928994083,
|
| 250 |
+
"eval_runtime": 63.1593,
|
| 251 |
+
"eval_samples_per_second": 5.352,
|
| 252 |
+
"eval_steps_per_second": 2.676,
|
| 253 |
+
"eval_top_10_accuracy": 0.15680473372781065,
|
| 254 |
+
"eval_top_1_accuracy": 0.01775147928994083,
|
| 255 |
+
"eval_top_5_accuracy": 0.09763313609467456,
|
| 256 |
"step": 1622
|
| 257 |
},
|
| 258 |
{
|
| 259 |
+
"epoch": 9.002152777777777,
|
| 260 |
+
"grad_norm": 29.327001571655273,
|
| 261 |
"learning_rate": 2.3555555555555556e-05,
|
| 262 |
+
"loss": 17.8731,
|
| 263 |
"step": 1700
|
| 264 |
},
|
| 265 |
{
|
| 266 |
+
"epoch": 9.004930555555555,
|
| 267 |
+
"grad_norm": 32.24522399902344,
|
| 268 |
"learning_rate": 2.4944444444444447e-05,
|
| 269 |
+
"loss": 17.9081,
|
| 270 |
"step": 1800
|
| 271 |
},
|
| 272 |
{
|
| 273 |
+
"epoch": 9.00498611111111,
|
| 274 |
+
"eval_accuracy": 0.03254437869822485,
|
| 275 |
+
"eval_f1": 0.010324695547367611,
|
| 276 |
+
"eval_loss": 4.562719345092773,
|
| 277 |
+
"eval_precision": 0.007627220692831552,
|
| 278 |
+
"eval_recall": 0.03254437869822485,
|
| 279 |
+
"eval_runtime": 60.367,
|
| 280 |
+
"eval_samples_per_second": 5.599,
|
| 281 |
+
"eval_steps_per_second": 2.8,
|
| 282 |
+
"eval_top_10_accuracy": 0.16863905325443787,
|
| 283 |
+
"eval_top_1_accuracy": 0.03254437869822485,
|
| 284 |
+
"eval_top_5_accuracy": 0.10650887573964497,
|
| 285 |
"step": 1802
|
| 286 |
},
|
| 287 |
{
|
| 288 |
+
"epoch": 10.00270138888889,
|
| 289 |
+
"grad_norm": 35.42505645751953,
|
| 290 |
"learning_rate": 2.633333333333333e-05,
|
| 291 |
+
"loss": 17.4986,
|
| 292 |
"step": 1900
|
| 293 |
},
|
| 294 |
{
|
| 295 |
+
"epoch": 10.005006944444444,
|
| 296 |
+
"eval_accuracy": 0.03550295857988166,
|
| 297 |
+
"eval_f1": 0.012635094540038232,
|
| 298 |
+
"eval_loss": 4.428702354431152,
|
| 299 |
+
"eval_precision": 0.008686492137909141,
|
| 300 |
+
"eval_recall": 0.03550295857988166,
|
| 301 |
+
"eval_runtime": 65.5677,
|
| 302 |
+
"eval_samples_per_second": 5.155,
|
| 303 |
+
"eval_steps_per_second": 2.577,
|
| 304 |
+
"eval_top_10_accuracy": 0.20118343195266272,
|
| 305 |
+
"eval_top_1_accuracy": 0.03550295857988166,
|
| 306 |
+
"eval_top_5_accuracy": 0.12130177514792899,
|
| 307 |
+
"step": 1983
|
| 308 |
},
|
| 309 |
{
|
| 310 |
+
"epoch": 11.000472222222223,
|
| 311 |
+
"grad_norm": 33.5329475402832,
|
| 312 |
"learning_rate": 2.772222222222222e-05,
|
| 313 |
+
"loss": 17.2661,
|
| 314 |
"step": 2000
|
| 315 |
},
|
| 316 |
{
|
| 317 |
+
"epoch": 11.00325,
|
| 318 |
+
"grad_norm": 34.00961685180664,
|
| 319 |
"learning_rate": 2.9111111111111112e-05,
|
| 320 |
+
"loss": 16.3869,
|
| 321 |
"step": 2100
|
| 322 |
},
|
| 323 |
{
|
| 324 |
+
"epoch": 11.005,
|
| 325 |
+
"eval_accuracy": 0.0650887573964497,
|
| 326 |
+
"eval_f1": 0.028937908468581415,
|
| 327 |
+
"eval_loss": 4.136640548706055,
|
| 328 |
+
"eval_precision": 0.022820963480770305,
|
| 329 |
+
"eval_recall": 0.0650887573964497,
|
| 330 |
+
"eval_runtime": 69.729,
|
| 331 |
+
"eval_samples_per_second": 4.847,
|
| 332 |
+
"eval_steps_per_second": 2.424,
|
| 333 |
+
"eval_top_10_accuracy": 0.3757396449704142,
|
| 334 |
+
"eval_top_1_accuracy": 0.0650887573964497,
|
| 335 |
+
"eval_top_5_accuracy": 0.23372781065088757,
|
| 336 |
"step": 2163
|
| 337 |
},
|
| 338 |
{
|
| 339 |
+
"epoch": 12.001020833333333,
|
| 340 |
+
"grad_norm": 31.025253295898438,
|
| 341 |
"learning_rate": 3.05e-05,
|
| 342 |
+
"loss": 15.7548,
|
| 343 |
"step": 2200
|
| 344 |
},
|
| 345 |
{
|
| 346 |
+
"epoch": 12.003798611111112,
|
| 347 |
+
"grad_norm": 38.748043060302734,
|
| 348 |
"learning_rate": 3.188888888888889e-05,
|
| 349 |
+
"loss": 15.0278,
|
| 350 |
"step": 2300
|
| 351 |
},
|
| 352 |
{
|
| 353 |
+
"epoch": 12.004993055555556,
|
| 354 |
+
"eval_accuracy": 0.10650887573964497,
|
| 355 |
+
"eval_f1": 0.0571553229183928,
|
| 356 |
+
"eval_loss": 3.7821457386016846,
|
| 357 |
+
"eval_precision": 0.060074719736818764,
|
| 358 |
+
"eval_recall": 0.10650887573964497,
|
| 359 |
+
"eval_runtime": 57.4783,
|
| 360 |
+
"eval_samples_per_second": 5.88,
|
| 361 |
+
"eval_steps_per_second": 2.94,
|
| 362 |
+
"eval_top_10_accuracy": 0.5177514792899408,
|
| 363 |
+
"eval_top_1_accuracy": 0.10650887573964497,
|
| 364 |
+
"eval_top_5_accuracy": 0.35798816568047337,
|
| 365 |
"step": 2343
|
| 366 |
},
|
| 367 |
{
|
| 368 |
+
"epoch": 13.001569444444444,
|
| 369 |
+
"grad_norm": 42.372962951660156,
|
| 370 |
"learning_rate": 3.327777777777778e-05,
|
| 371 |
+
"loss": 14.1209,
|
| 372 |
"step": 2400
|
| 373 |
},
|
| 374 |
{
|
| 375 |
+
"epoch": 13.004347222222222,
|
| 376 |
+
"grad_norm": 53.95140075683594,
|
| 377 |
"learning_rate": 3.466666666666667e-05,
|
| 378 |
+
"loss": 13.4974,
|
| 379 |
"step": 2500
|
| 380 |
},
|
| 381 |
{
|
| 382 |
+
"epoch": 13.00498611111111,
|
| 383 |
+
"eval_accuracy": 0.20118343195266272,
|
| 384 |
+
"eval_f1": 0.1544137887289645,
|
| 385 |
+
"eval_loss": 3.441917657852173,
|
| 386 |
+
"eval_precision": 0.17607648509523535,
|
| 387 |
+
"eval_recall": 0.20118343195266272,
|
| 388 |
+
"eval_runtime": 61.9222,
|
| 389 |
+
"eval_samples_per_second": 5.458,
|
| 390 |
+
"eval_steps_per_second": 2.729,
|
| 391 |
+
"eval_top_10_accuracy": 0.6686390532544378,
|
| 392 |
+
"eval_top_1_accuracy": 0.20118343195266272,
|
| 393 |
+
"eval_top_5_accuracy": 0.5029585798816568,
|
| 394 |
"step": 2523
|
| 395 |
},
|
| 396 |
{
|
| 397 |
+
"epoch": 14.002118055555556,
|
| 398 |
+
"grad_norm": 47.6016731262207,
|
| 399 |
"learning_rate": 3.605555555555556e-05,
|
| 400 |
+
"loss": 12.1829,
|
| 401 |
"step": 2600
|
| 402 |
},
|
| 403 |
{
|
| 404 |
+
"epoch": 14.004895833333334,
|
| 405 |
+
"grad_norm": 35.87943649291992,
|
| 406 |
"learning_rate": 3.7444444444444446e-05,
|
| 407 |
+
"loss": 11.5774,
|
| 408 |
"step": 2700
|
| 409 |
},
|
| 410 |
{
|
| 411 |
+
"epoch": 14.005006944444444,
|
| 412 |
+
"eval_accuracy": 0.21893491124260356,
|
| 413 |
+
"eval_f1": 0.17250136680337552,
|
| 414 |
+
"eval_loss": 3.2604691982269287,
|
| 415 |
+
"eval_precision": 0.20685830348019885,
|
| 416 |
+
"eval_recall": 0.21893491124260356,
|
| 417 |
+
"eval_runtime": 58.8349,
|
| 418 |
+
"eval_samples_per_second": 5.745,
|
| 419 |
+
"eval_steps_per_second": 2.872,
|
| 420 |
+
"eval_top_10_accuracy": 0.6982248520710059,
|
| 421 |
+
"eval_top_1_accuracy": 0.21893491124260356,
|
| 422 |
+
"eval_top_5_accuracy": 0.5355029585798816,
|
| 423 |
+
"step": 2704
|
| 424 |
},
|
| 425 |
{
|
| 426 |
+
"epoch": 15.002666666666666,
|
| 427 |
+
"grad_norm": 50.407718658447266,
|
| 428 |
"learning_rate": 3.883333333333333e-05,
|
| 429 |
+
"loss": 10.263,
|
| 430 |
"step": 2800
|
| 431 |
},
|
| 432 |
{
|
| 433 |
+
"epoch": 15.005,
|
| 434 |
+
"eval_accuracy": 0.31952662721893493,
|
| 435 |
+
"eval_f1": 0.29516694216805966,
|
| 436 |
+
"eval_loss": 2.8508408069610596,
|
| 437 |
+
"eval_precision": 0.3730181143642682,
|
| 438 |
+
"eval_recall": 0.31952662721893493,
|
| 439 |
+
"eval_runtime": 228.2279,
|
| 440 |
+
"eval_samples_per_second": 1.481,
|
| 441 |
+
"eval_steps_per_second": 0.74,
|
| 442 |
+
"eval_top_10_accuracy": 0.8076923076923077,
|
| 443 |
+
"eval_top_1_accuracy": 0.31952662721893493,
|
| 444 |
+
"eval_top_5_accuracy": 0.6627218934911243,
|
| 445 |
"step": 2884
|
| 446 |
},
|
| 447 |
{
|
| 448 |
+
"epoch": 16.0004375,
|
| 449 |
+
"grad_norm": 51.00344467163086,
|
| 450 |
"learning_rate": 4.022222222222222e-05,
|
| 451 |
+
"loss": 9.6519,
|
| 452 |
"step": 2900
|
| 453 |
},
|
| 454 |
{
|
| 455 |
+
"epoch": 16.003215277777777,
|
| 456 |
+
"grad_norm": 55.5887451171875,
|
| 457 |
"learning_rate": 4.1611111111111114e-05,
|
| 458 |
+
"loss": 8.1589,
|
| 459 |
"step": 3000
|
| 460 |
},
|
| 461 |
{
|
| 462 |
+
"epoch": 16.004993055555556,
|
| 463 |
+
"eval_accuracy": 0.3905325443786982,
|
| 464 |
+
"eval_f1": 0.3592464386756068,
|
| 465 |
+
"eval_loss": 2.594480514526367,
|
| 466 |
+
"eval_precision": 0.41684405338251496,
|
| 467 |
+
"eval_recall": 0.3905325443786982,
|
| 468 |
+
"eval_runtime": 190.8374,
|
| 469 |
+
"eval_samples_per_second": 1.771,
|
| 470 |
+
"eval_steps_per_second": 0.886,
|
| 471 |
+
"eval_top_10_accuracy": 0.8461538461538461,
|
| 472 |
+
"eval_top_1_accuracy": 0.3905325443786982,
|
| 473 |
+
"eval_top_5_accuracy": 0.7307692307692307,
|
| 474 |
"step": 3064
|
| 475 |
},
|
| 476 |
{
|
| 477 |
+
"epoch": 17.00098611111111,
|
| 478 |
+
"grad_norm": 54.826148986816406,
|
| 479 |
"learning_rate": 4.3e-05,
|
| 480 |
+
"loss": 7.9513,
|
| 481 |
"step": 3100
|
| 482 |
},
|
| 483 |
{
|
| 484 |
+
"epoch": 17.003763888888887,
|
| 485 |
+
"grad_norm": 76.26374053955078,
|
| 486 |
"learning_rate": 4.438888888888889e-05,
|
| 487 |
+
"loss": 6.8221,
|
| 488 |
"step": 3200
|
| 489 |
},
|
| 490 |
{
|
| 491 |
+
"epoch": 17.004986111111112,
|
| 492 |
+
"eval_accuracy": 0.3994082840236686,
|
| 493 |
+
"eval_f1": 0.3639671675936601,
|
| 494 |
+
"eval_loss": 2.4310550689697266,
|
| 495 |
+
"eval_precision": 0.3999949948836589,
|
| 496 |
+
"eval_recall": 0.3994082840236686,
|
| 497 |
+
"eval_runtime": 187.0322,
|
| 498 |
+
"eval_samples_per_second": 1.807,
|
| 499 |
+
"eval_steps_per_second": 0.904,
|
| 500 |
+
"eval_top_10_accuracy": 0.8727810650887574,
|
| 501 |
+
"eval_top_1_accuracy": 0.3994082840236686,
|
| 502 |
+
"eval_top_5_accuracy": 0.7337278106508875,
|
| 503 |
"step": 3244
|
| 504 |
},
|
| 505 |
{
|
| 506 |
+
"epoch": 18.00153472222222,
|
| 507 |
+
"grad_norm": 55.72072219848633,
|
| 508 |
"learning_rate": 4.577777777777778e-05,
|
| 509 |
+
"loss": 5.8926,
|
| 510 |
"step": 3300
|
| 511 |
},
|
| 512 |
{
|
| 513 |
+
"epoch": 18.0043125,
|
| 514 |
+
"grad_norm": 42.606868743896484,
|
| 515 |
"learning_rate": 4.716666666666667e-05,
|
| 516 |
+
"loss": 5.4923,
|
| 517 |
"step": 3400
|
| 518 |
},
|
| 519 |
{
|
| 520 |
+
"epoch": 18.005006944444446,
|
| 521 |
+
"eval_accuracy": 0.46449704142011833,
|
| 522 |
+
"eval_f1": 0.43497012092329823,
|
| 523 |
+
"eval_loss": 2.213894844055176,
|
| 524 |
+
"eval_precision": 0.48856049681841535,
|
| 525 |
+
"eval_recall": 0.46449704142011833,
|
| 526 |
+
"eval_runtime": 171.9348,
|
| 527 |
+
"eval_samples_per_second": 1.966,
|
| 528 |
+
"eval_steps_per_second": 0.983,
|
| 529 |
+
"eval_top_10_accuracy": 0.8668639053254438,
|
| 530 |
+
"eval_top_1_accuracy": 0.46153846153846156,
|
| 531 |
+
"eval_top_5_accuracy": 0.7928994082840237,
|
| 532 |
+
"step": 3425
|
| 533 |
},
|
| 534 |
{
|
| 535 |
+
"epoch": 19.002083333333335,
|
| 536 |
+
"grad_norm": 66.9966812133789,
|
| 537 |
"learning_rate": 4.855555555555556e-05,
|
| 538 |
+
"loss": 4.6739,
|
| 539 |
"step": 3500
|
| 540 |
},
|
| 541 |
{
|
| 542 |
+
"epoch": 19.00486111111111,
|
| 543 |
+
"grad_norm": 29.572689056396484,
|
| 544 |
"learning_rate": 4.994444444444445e-05,
|
| 545 |
+
"loss": 4.1619,
|
| 546 |
"step": 3600
|
| 547 |
},
|
| 548 |
{
|
| 549 |
+
"epoch": 19.005,
|
| 550 |
+
"eval_accuracy": 0.47337278106508873,
|
| 551 |
+
"eval_f1": 0.4502898182762436,
|
| 552 |
+
"eval_loss": 2.1384310722351074,
|
| 553 |
+
"eval_precision": 0.5276356764827005,
|
| 554 |
+
"eval_recall": 0.47337278106508873,
|
| 555 |
+
"eval_runtime": 180.7053,
|
| 556 |
+
"eval_samples_per_second": 1.87,
|
| 557 |
+
"eval_steps_per_second": 0.935,
|
| 558 |
+
"eval_top_10_accuracy": 0.8816568047337278,
|
| 559 |
+
"eval_top_1_accuracy": 0.47337278106508873,
|
| 560 |
"eval_top_5_accuracy": 0.7840236686390533,
|
| 561 |
"step": 3605
|
| 562 |
},
|
| 563 |
{
|
| 564 |
+
"epoch": 20.002631944444445,
|
| 565 |
+
"grad_norm": 62.397911071777344,
|
| 566 |
"learning_rate": 4.9851851851851855e-05,
|
| 567 |
+
"loss": 3.3413,
|
| 568 |
"step": 3700
|
| 569 |
},
|
| 570 |
{
|
| 571 |
+
"epoch": 20.004993055555556,
|
| 572 |
+
"eval_accuracy": 0.5118343195266272,
|
| 573 |
+
"eval_f1": 0.49080667210330176,
|
| 574 |
+
"eval_loss": 1.958280324935913,
|
| 575 |
+
"eval_precision": 0.5484620698689525,
|
| 576 |
+
"eval_recall": 0.5118343195266272,
|
| 577 |
+
"eval_runtime": 173.8049,
|
| 578 |
+
"eval_samples_per_second": 1.945,
|
| 579 |
+
"eval_steps_per_second": 0.972,
|
| 580 |
+
"eval_top_10_accuracy": 0.9053254437869822,
|
| 581 |
+
"eval_top_1_accuracy": 0.5118343195266272,
|
| 582 |
+
"eval_top_5_accuracy": 0.8106508875739645,
|
| 583 |
"step": 3785
|
| 584 |
},
|
| 585 |
{
|
| 586 |
+
"epoch": 21.00040277777778,
|
| 587 |
+
"grad_norm": 40.85105514526367,
|
| 588 |
"learning_rate": 4.969753086419753e-05,
|
| 589 |
+
"loss": 3.4134,
|
| 590 |
"step": 3800
|
| 591 |
},
|
| 592 |
{
|
| 593 |
+
"epoch": 21.003180555555556,
|
| 594 |
+
"grad_norm": 32.64407730102539,
|
| 595 |
"learning_rate": 4.954320987654321e-05,
|
| 596 |
+
"loss": 2.5832,
|
| 597 |
"step": 3900
|
| 598 |
},
|
| 599 |
{
|
| 600 |
+
"epoch": 21.004986111111112,
|
| 601 |
+
"eval_accuracy": 0.5,
|
| 602 |
+
"eval_f1": 0.47126995292675755,
|
| 603 |
+
"eval_loss": 1.86040198802948,
|
| 604 |
+
"eval_precision": 0.49999156826079894,
|
| 605 |
+
"eval_recall": 0.5,
|
| 606 |
+
"eval_runtime": 182.2543,
|
| 607 |
+
"eval_samples_per_second": 1.855,
|
| 608 |
+
"eval_steps_per_second": 0.927,
|
| 609 |
+
"eval_top_10_accuracy": 0.9201183431952663,
|
| 610 |
+
"eval_top_1_accuracy": 0.5,
|
| 611 |
+
"eval_top_5_accuracy": 0.8284023668639053,
|
| 612 |
"step": 3965
|
| 613 |
},
|
| 614 |
{
|
| 615 |
+
"epoch": 22.00095138888889,
|
| 616 |
+
"grad_norm": 15.257333755493164,
|
| 617 |
"learning_rate": 4.938888888888889e-05,
|
| 618 |
+
"loss": 2.2189,
|
| 619 |
"step": 4000
|
| 620 |
},
|
| 621 |
{
|
| 622 |
+
"epoch": 22.003729166666666,
|
| 623 |
+
"grad_norm": 26.528512954711914,
|
| 624 |
+
"learning_rate": 4.923611111111112e-05,
|
| 625 |
+
"loss": 1.9003,
|
| 626 |
"step": 4100
|
| 627 |
},
|
| 628 |
{
|
| 629 |
+
"epoch": 22.005006944444446,
|
| 630 |
+
"eval_accuracy": 0.5295857988165681,
|
| 631 |
+
"eval_f1": 0.5139563428024967,
|
| 632 |
+
"eval_loss": 1.9389936923980713,
|
| 633 |
+
"eval_precision": 0.5860205028757064,
|
| 634 |
+
"eval_recall": 0.5295857988165681,
|
| 635 |
+
"eval_runtime": 176.2308,
|
| 636 |
+
"eval_samples_per_second": 1.918,
|
| 637 |
+
"eval_steps_per_second": 0.959,
|
| 638 |
+
"eval_top_10_accuracy": 0.8905325443786982,
|
| 639 |
+
"eval_top_1_accuracy": 0.5295857988165681,
|
| 640 |
+
"eval_top_5_accuracy": 0.8195266272189349,
|
| 641 |
+
"step": 4146
|
| 642 |
},
|
| 643 |
{
|
| 644 |
+
"epoch": 23.0015,
|
| 645 |
+
"grad_norm": 42.9820556640625,
|
| 646 |
"learning_rate": 4.9081790123456794e-05,
|
| 647 |
+
"loss": 1.5994,
|
| 648 |
"step": 4200
|
| 649 |
},
|
| 650 |
{
|
| 651 |
+
"epoch": 23.004277777777776,
|
| 652 |
+
"grad_norm": 44.47490692138672,
|
| 653 |
"learning_rate": 4.892746913580247e-05,
|
| 654 |
+
"loss": 1.4226,
|
| 655 |
"step": 4300
|
| 656 |
},
|
| 657 |
{
|
| 658 |
+
"epoch": 23.005,
|
| 659 |
+
"eval_accuracy": 0.5266272189349113,
|
| 660 |
+
"eval_f1": 0.5303364916400071,
|
| 661 |
+
"eval_loss": 1.9288358688354492,
|
| 662 |
+
"eval_precision": 0.6340754166937599,
|
| 663 |
+
"eval_recall": 0.5266272189349113,
|
| 664 |
+
"eval_runtime": 175.095,
|
| 665 |
+
"eval_samples_per_second": 1.93,
|
| 666 |
+
"eval_steps_per_second": 0.965,
|
| 667 |
+
"eval_top_10_accuracy": 0.908284023668639,
|
| 668 |
+
"eval_top_1_accuracy": 0.5266272189349113,
|
| 669 |
+
"eval_top_5_accuracy": 0.8076923076923077,
|
| 670 |
"step": 4326
|
| 671 |
},
|
| 672 |
{
|
| 673 |
+
"epoch": 24.00204861111111,
|
| 674 |
+
"grad_norm": 43.410343170166016,
|
| 675 |
"learning_rate": 4.877314814814815e-05,
|
| 676 |
+
"loss": 1.1426,
|
| 677 |
"step": 4400
|
| 678 |
},
|
| 679 |
{
|
| 680 |
+
"epoch": 24.00482638888889,
|
| 681 |
+
"grad_norm": 89.68692016601562,
|
| 682 |
"learning_rate": 4.861882716049383e-05,
|
| 683 |
+
"loss": 1.1341,
|
| 684 |
"step": 4500
|
| 685 |
},
|
| 686 |
{
|
| 687 |
+
"epoch": 24.004993055555556,
|
| 688 |
+
"eval_accuracy": 0.5266272189349113,
|
| 689 |
+
"eval_f1": 0.5078898808033033,
|
| 690 |
+
"eval_loss": 1.8854230642318726,
|
| 691 |
+
"eval_precision": 0.5719492128515797,
|
| 692 |
+
"eval_recall": 0.5266272189349113,
|
| 693 |
+
"eval_runtime": 184.1675,
|
| 694 |
+
"eval_samples_per_second": 1.835,
|
| 695 |
+
"eval_steps_per_second": 0.918,
|
| 696 |
+
"eval_top_10_accuracy": 0.8964497041420119,
|
| 697 |
+
"eval_top_1_accuracy": 0.5266272189349113,
|
| 698 |
+
"eval_top_5_accuracy": 0.7899408284023669,
|
| 699 |
"step": 4506
|
| 700 |
},
|
| 701 |
{
|
| 702 |
+
"epoch": 25.00259722222222,
|
| 703 |
+
"grad_norm": 7.714802265167236,
|
| 704 |
"learning_rate": 4.8464506172839505e-05,
|
| 705 |
+
"loss": 0.9315,
|
| 706 |
"step": 4600
|
| 707 |
},
|
| 708 |
{
|
| 709 |
+
"epoch": 25.004986111111112,
|
| 710 |
+
"eval_accuracy": 0.5769230769230769,
|
| 711 |
+
"eval_f1": 0.5564708795478026,
|
| 712 |
+
"eval_loss": 1.7328224182128906,
|
| 713 |
+
"eval_precision": 0.6260812905043673,
|
| 714 |
+
"eval_recall": 0.5769230769230769,
|
| 715 |
+
"eval_runtime": 67.8286,
|
| 716 |
+
"eval_samples_per_second": 4.983,
|
| 717 |
+
"eval_steps_per_second": 2.492,
|
| 718 |
+
"eval_top_10_accuracy": 0.8994082840236687,
|
| 719 |
+
"eval_top_1_accuracy": 0.5769230769230769,
|
| 720 |
+
"eval_top_5_accuracy": 0.8461538461538461,
|
| 721 |
"step": 4686
|
| 722 |
},
|
| 723 |
{
|
| 724 |
+
"epoch": 26.000368055555555,
|
| 725 |
+
"grad_norm": 11.666605949401855,
|
| 726 |
"learning_rate": 4.831018518518518e-05,
|
| 727 |
+
"loss": 0.7589,
|
| 728 |
"step": 4700
|
| 729 |
},
|
| 730 |
{
|
| 731 |
+
"epoch": 26.003145833333335,
|
| 732 |
+
"grad_norm": 80.17562866210938,
|
| 733 |
"learning_rate": 4.815586419753087e-05,
|
| 734 |
+
"loss": 0.7536,
|
| 735 |
"step": 4800
|
| 736 |
},
|
| 737 |
{
|
| 738 |
+
"epoch": 26.005006944444446,
|
| 739 |
"eval_accuracy": 0.5384615384615384,
|
| 740 |
+
"eval_f1": 0.5296188525478466,
|
| 741 |
+
"eval_loss": 1.8349210023880005,
|
| 742 |
+
"eval_precision": 0.6288461538461538,
|
| 743 |
"eval_recall": 0.5384615384615384,
|
| 744 |
+
"eval_runtime": 58.1307,
|
| 745 |
+
"eval_samples_per_second": 5.814,
|
| 746 |
+
"eval_steps_per_second": 2.907,
|
| 747 |
+
"eval_top_10_accuracy": 0.8964497041420119,
|
| 748 |
+
"eval_top_1_accuracy": 0.5384615384615384,
|
| 749 |
+
"eval_top_5_accuracy": 0.8195266272189349,
|
| 750 |
+
"step": 4867
|
| 751 |
},
|
| 752 |
{
|
| 753 |
+
"epoch": 27.000916666666665,
|
| 754 |
+
"grad_norm": 11.256963729858398,
|
| 755 |
"learning_rate": 4.8001543209876545e-05,
|
| 756 |
+
"loss": 0.6656,
|
| 757 |
"step": 4900
|
| 758 |
},
|
| 759 |
{
|
| 760 |
+
"epoch": 27.003694444444445,
|
| 761 |
+
"grad_norm": 13.579202651977539,
|
| 762 |
"learning_rate": 4.784722222222223e-05,
|
| 763 |
+
"loss": 0.4518,
|
| 764 |
"step": 5000
|
| 765 |
},
|
| 766 |
{
|
| 767 |
+
"epoch": 27.005,
|
| 768 |
+
"eval_accuracy": 0.5532544378698225,
|
| 769 |
+
"eval_f1": 0.5436848818505623,
|
| 770 |
+
"eval_loss": 1.7999422550201416,
|
| 771 |
+
"eval_precision": 0.6275922795153565,
|
| 772 |
+
"eval_recall": 0.5532544378698225,
|
| 773 |
+
"eval_runtime": 56.1604,
|
| 774 |
+
"eval_samples_per_second": 6.018,
|
| 775 |
+
"eval_steps_per_second": 3.009,
|
| 776 |
+
"eval_top_10_accuracy": 0.9171597633136095,
|
| 777 |
+
"eval_top_1_accuracy": 0.5532544378698225,
|
| 778 |
+
"eval_top_5_accuracy": 0.8550295857988166,
|
| 779 |
"step": 5047
|
| 780 |
},
|
| 781 |
{
|
| 782 |
+
"epoch": 28.00146527777778,
|
| 783 |
+
"grad_norm": 18.927217483520508,
|
| 784 |
"learning_rate": 4.769290123456791e-05,
|
| 785 |
+
"loss": 0.3555,
|
| 786 |
"step": 5100
|
| 787 |
},
|
| 788 |
{
|
| 789 |
+
"epoch": 28.004243055555555,
|
| 790 |
+
"grad_norm": 7.268669605255127,
|
| 791 |
"learning_rate": 4.7538580246913585e-05,
|
| 792 |
+
"loss": 0.3322,
|
| 793 |
"step": 5200
|
| 794 |
},
|
| 795 |
{
|
| 796 |
+
"epoch": 28.004993055555556,
|
| 797 |
+
"eval_accuracy": 0.6005917159763313,
|
| 798 |
+
"eval_f1": 0.5895704722627799,
|
| 799 |
+
"eval_loss": 1.693082571029663,
|
| 800 |
+
"eval_precision": 0.6577521837137221,
|
| 801 |
+
"eval_recall": 0.6005917159763313,
|
| 802 |
+
"eval_runtime": 60.4433,
|
| 803 |
+
"eval_samples_per_second": 5.592,
|
| 804 |
+
"eval_steps_per_second": 2.796,
|
| 805 |
+
"eval_top_10_accuracy": 0.9260355029585798,
|
| 806 |
+
"eval_top_1_accuracy": 0.6005917159763313,
|
| 807 |
+
"eval_top_5_accuracy": 0.849112426035503,
|
| 808 |
"step": 5227
|
| 809 |
},
|
| 810 |
{
|
| 811 |
+
"epoch": 29.00201388888889,
|
| 812 |
+
"grad_norm": 17.36412239074707,
|
| 813 |
"learning_rate": 4.738425925925926e-05,
|
| 814 |
+
"loss": 0.2749,
|
| 815 |
"step": 5300
|
| 816 |
},
|
| 817 |
{
|
| 818 |
+
"epoch": 29.004791666666666,
|
| 819 |
+
"grad_norm": 1.4117032289505005,
|
| 820 |
"learning_rate": 4.722993827160494e-05,
|
| 821 |
+
"loss": 0.403,
|
| 822 |
"step": 5400
|
| 823 |
},
|
| 824 |
{
|
| 825 |
+
"epoch": 29.004986111111112,
|
| 826 |
+
"eval_accuracy": 0.5739644970414202,
|
| 827 |
+
"eval_f1": 0.5584189150757197,
|
| 828 |
+
"eval_loss": 1.7999780178070068,
|
| 829 |
+
"eval_precision": 0.6239610652332546,
|
| 830 |
+
"eval_recall": 0.5739644970414202,
|
| 831 |
+
"eval_runtime": 57.9837,
|
| 832 |
+
"eval_samples_per_second": 5.829,
|
| 833 |
+
"eval_steps_per_second": 2.915,
|
| 834 |
+
"eval_top_10_accuracy": 0.9142011834319527,
|
| 835 |
+
"eval_top_1_accuracy": 0.5739644970414202,
|
| 836 |
+
"eval_top_5_accuracy": 0.8461538461538461,
|
| 837 |
"step": 5407
|
| 838 |
},
|
| 839 |
{
|
| 840 |
+
"epoch": 30.0025625,
|
| 841 |
+
"grad_norm": 0.5923660397529602,
|
| 842 |
"learning_rate": 4.707561728395062e-05,
|
| 843 |
+
"loss": 0.1837,
|
| 844 |
"step": 5500
|
| 845 |
},
|
| 846 |
{
|
| 847 |
+
"epoch": 30.005006944444446,
|
| 848 |
+
"eval_accuracy": 0.5769230769230769,
|
| 849 |
+
"eval_f1": 0.5664270128680152,
|
| 850 |
+
"eval_loss": 1.876481056213379,
|
| 851 |
+
"eval_precision": 0.6391236968160044,
|
| 852 |
+
"eval_recall": 0.5769230769230769,
|
| 853 |
+
"eval_runtime": 60.7728,
|
| 854 |
+
"eval_samples_per_second": 5.562,
|
| 855 |
+
"eval_steps_per_second": 2.781,
|
| 856 |
+
"eval_top_10_accuracy": 0.893491124260355,
|
| 857 |
+
"eval_top_1_accuracy": 0.5769230769230769,
|
| 858 |
+
"eval_top_5_accuracy": 0.8372781065088757,
|
| 859 |
+
"step": 5588
|
| 860 |
+
},
|
| 861 |
+
{
|
| 862 |
+
"epoch": 31.000333333333334,
|
| 863 |
+
"grad_norm": 20.429960250854492,
|
| 864 |
+
"learning_rate": 4.6922839506172846e-05,
|
| 865 |
+
"loss": 0.3429,
|
| 866 |
"step": 5600
|
| 867 |
},
|
| 868 |
{
|
| 869 |
+
"epoch": 31.00311111111111,
|
| 870 |
+
"grad_norm": 11.760103225708008,
|
| 871 |
+
"learning_rate": 4.6768518518518524e-05,
|
| 872 |
+
"loss": 0.1579,
|
| 873 |
"step": 5700
|
| 874 |
},
|
| 875 |
{
|
| 876 |
+
"epoch": 31.005,
|
| 877 |
+
"eval_accuracy": 0.5473372781065089,
|
| 878 |
+
"eval_f1": 0.5426257667381927,
|
| 879 |
+
"eval_loss": 2.075249195098877,
|
| 880 |
+
"eval_precision": 0.6336519815026602,
|
| 881 |
+
"eval_recall": 0.5473372781065089,
|
| 882 |
+
"eval_runtime": 60.9135,
|
| 883 |
+
"eval_samples_per_second": 5.549,
|
| 884 |
+
"eval_steps_per_second": 2.774,
|
| 885 |
+
"eval_top_10_accuracy": 0.8994082840236687,
|
| 886 |
+
"eval_top_1_accuracy": 0.5473372781065089,
|
| 887 |
+
"eval_top_5_accuracy": 0.8431952662721893,
|
| 888 |
"step": 5768
|
| 889 |
},
|
| 890 |
{
|
| 891 |
+
"epoch": 32.000881944444444,
|
| 892 |
+
"grad_norm": 1.1906695365905762,
|
| 893 |
+
"learning_rate": 4.66141975308642e-05,
|
| 894 |
+
"loss": 0.3029,
|
| 895 |
"step": 5800
|
| 896 |
},
|
| 897 |
{
|
| 898 |
+
"epoch": 32.003659722222224,
|
| 899 |
+
"grad_norm": 75.9498519897461,
|
| 900 |
+
"learning_rate": 4.645987654320988e-05,
|
| 901 |
+
"loss": 0.2079,
|
| 902 |
"step": 5900
|
| 903 |
},
|
| 904 |
{
|
| 905 |
+
"epoch": 32.00499305555556,
|
| 906 |
+
"eval_accuracy": 0.5946745562130178,
|
| 907 |
+
"eval_f1": 0.5791488754506505,
|
| 908 |
+
"eval_loss": 1.9234323501586914,
|
| 909 |
+
"eval_precision": 0.6524024052870208,
|
| 910 |
+
"eval_recall": 0.5946745562130178,
|
| 911 |
+
"eval_runtime": 57.5153,
|
| 912 |
+
"eval_samples_per_second": 5.877,
|
| 913 |
+
"eval_steps_per_second": 2.938,
|
| 914 |
+
"eval_top_10_accuracy": 0.8964497041420119,
|
| 915 |
+
"eval_top_1_accuracy": 0.5946745562130178,
|
| 916 |
"eval_top_5_accuracy": 0.8136094674556213,
|
| 917 |
"step": 5948
|
| 918 |
},
|
| 919 |
{
|
| 920 |
+
"epoch": 33.00143055555556,
|
| 921 |
+
"grad_norm": 0.5180000066757202,
|
| 922 |
+
"learning_rate": 4.630555555555556e-05,
|
| 923 |
+
"loss": 0.1503,
|
| 924 |
"step": 6000
|
| 925 |
},
|
| 926 |
{
|
| 927 |
+
"epoch": 33.00420833333333,
|
| 928 |
+
"grad_norm": 0.7615421414375305,
|
| 929 |
+
"learning_rate": 4.6151234567901235e-05,
|
| 930 |
+
"loss": 0.2738,
|
| 931 |
"step": 6100
|
| 932 |
},
|
| 933 |
{
|
| 934 |
+
"epoch": 33.00498611111111,
|
| 935 |
+
"eval_accuracy": 0.606508875739645,
|
| 936 |
+
"eval_f1": 0.5892127021564892,
|
| 937 |
+
"eval_loss": 1.8528887033462524,
|
| 938 |
+
"eval_precision": 0.6621978202599504,
|
| 939 |
+
"eval_recall": 0.606508875739645,
|
| 940 |
+
"eval_runtime": 61.6919,
|
| 941 |
+
"eval_samples_per_second": 5.479,
|
| 942 |
+
"eval_steps_per_second": 2.739,
|
| 943 |
+
"eval_top_10_accuracy": 0.9023668639053254,
|
| 944 |
+
"eval_top_1_accuracy": 0.6035502958579881,
|
| 945 |
+
"eval_top_5_accuracy": 0.849112426035503,
|
| 946 |
"step": 6128
|
| 947 |
},
|
| 948 |
{
|
| 949 |
+
"epoch": 34.001979166666665,
|
| 950 |
+
"grad_norm": 0.34027883410453796,
|
| 951 |
"learning_rate": 4.599691358024691e-05,
|
| 952 |
+
"loss": 0.0955,
|
| 953 |
"step": 6200
|
| 954 |
},
|
| 955 |
{
|
| 956 |
+
"epoch": 34.004756944444445,
|
| 957 |
+
"grad_norm": 0.2066214680671692,
|
| 958 |
"learning_rate": 4.584259259259259e-05,
|
| 959 |
+
"loss": 0.2621,
|
| 960 |
"step": 6300
|
| 961 |
},
|
| 962 |
{
|
| 963 |
+
"epoch": 34.005006944444446,
|
| 964 |
+
"eval_accuracy": 0.5739644970414202,
|
| 965 |
+
"eval_f1": 0.5639503658734427,
|
| 966 |
+
"eval_loss": 1.9905767440795898,
|
| 967 |
+
"eval_precision": 0.6191074950690336,
|
| 968 |
+
"eval_recall": 0.5739644970414202,
|
| 969 |
+
"eval_runtime": 75.9577,
|
| 970 |
+
"eval_samples_per_second": 4.45,
|
| 971 |
+
"eval_steps_per_second": 2.225,
|
| 972 |
+
"eval_top_10_accuracy": 0.9171597633136095,
|
| 973 |
+
"eval_top_1_accuracy": 0.5739644970414202,
|
| 974 |
+
"eval_top_5_accuracy": 0.8284023668639053,
|
| 975 |
+
"step": 6309
|
| 976 |
},
|
| 977 |
{
|
| 978 |
+
"epoch": 35.00252777777778,
|
| 979 |
+
"grad_norm": 0.1237727478146553,
|
| 980 |
"learning_rate": 4.5688271604938275e-05,
|
| 981 |
+
"loss": 0.2024,
|
| 982 |
"step": 6400
|
| 983 |
},
|
| 984 |
{
|
| 985 |
+
"epoch": 35.005,
|
| 986 |
+
"eval_accuracy": 0.5976331360946746,
|
| 987 |
+
"eval_f1": 0.5885829981983829,
|
| 988 |
+
"eval_loss": 1.8942253589630127,
|
| 989 |
+
"eval_precision": 0.6614584133814904,
|
| 990 |
+
"eval_recall": 0.5976331360946746,
|
| 991 |
+
"eval_runtime": 60.0376,
|
| 992 |
+
"eval_samples_per_second": 5.63,
|
| 993 |
+
"eval_steps_per_second": 2.815,
|
| 994 |
+
"eval_top_10_accuracy": 0.9260355029585798,
|
| 995 |
+
"eval_top_1_accuracy": 0.5976331360946746,
|
| 996 |
+
"eval_top_5_accuracy": 0.863905325443787,
|
| 997 |
"step": 6489
|
| 998 |
},
|
| 999 |
{
|
| 1000 |
+
"epoch": 36.00029861111111,
|
| 1001 |
+
"grad_norm": 0.21603639423847198,
|
| 1002 |
+
"learning_rate": 4.553395061728395e-05,
|
| 1003 |
+
"loss": 0.1411,
|
| 1004 |
"step": 6500
|
| 1005 |
},
|
| 1006 |
{
|
| 1007 |
+
"epoch": 36.003076388888886,
|
| 1008 |
+
"grad_norm": 0.11866825819015503,
|
| 1009 |
+
"learning_rate": 4.537962962962963e-05,
|
| 1010 |
+
"loss": 0.0983,
|
| 1011 |
"step": 6600
|
| 1012 |
},
|
| 1013 |
{
|
| 1014 |
+
"epoch": 36.00499305555556,
|
| 1015 |
+
"eval_accuracy": 0.5857988165680473,
|
| 1016 |
+
"eval_f1": 0.5728611953464025,
|
| 1017 |
+
"eval_loss": 2.0340006351470947,
|
| 1018 |
+
"eval_precision": 0.6499939163400703,
|
| 1019 |
+
"eval_recall": 0.5857988165680473,
|
| 1020 |
+
"eval_runtime": 58.3576,
|
| 1021 |
+
"eval_samples_per_second": 5.792,
|
| 1022 |
+
"eval_steps_per_second": 2.896,
|
| 1023 |
+
"eval_top_10_accuracy": 0.8846153846153846,
|
| 1024 |
+
"eval_top_1_accuracy": 0.5857988165680473,
|
| 1025 |
+
"eval_top_5_accuracy": 0.8254437869822485,
|
| 1026 |
"step": 6669
|
| 1027 |
},
|
| 1028 |
{
|
| 1029 |
+
"epoch": 37.00084722222222,
|
| 1030 |
+
"grad_norm": 8.090852737426758,
|
| 1031 |
+
"learning_rate": 4.5225308641975314e-05,
|
| 1032 |
+
"loss": 0.13,
|
| 1033 |
"step": 6700
|
| 1034 |
},
|
| 1035 |
{
|
| 1036 |
+
"epoch": 37.003625,
|
| 1037 |
+
"grad_norm": 0.11927696317434311,
|
| 1038 |
+
"learning_rate": 4.507098765432099e-05,
|
| 1039 |
+
"loss": 0.0592,
|
| 1040 |
"step": 6800
|
| 1041 |
},
|
| 1042 |
{
|
| 1043 |
+
"epoch": 37.00498611111111,
|
| 1044 |
+
"eval_accuracy": 0.6094674556213018,
|
| 1045 |
+
"eval_f1": 0.5985522367179171,
|
| 1046 |
+
"eval_loss": 1.8492796421051025,
|
| 1047 |
+
"eval_precision": 0.6775016650016649,
|
| 1048 |
+
"eval_recall": 0.6094674556213018,
|
| 1049 |
+
"eval_runtime": 58.2549,
|
| 1050 |
+
"eval_samples_per_second": 5.802,
|
| 1051 |
+
"eval_steps_per_second": 2.901,
|
| 1052 |
+
"eval_top_10_accuracy": 0.9230769230769231,
|
| 1053 |
+
"eval_top_1_accuracy": 0.6094674556213018,
|
| 1054 |
+
"eval_top_5_accuracy": 0.8609467455621301,
|
| 1055 |
"step": 6849
|
| 1056 |
},
|
| 1057 |
{
|
| 1058 |
+
"epoch": 38.00139583333333,
|
| 1059 |
+
"grad_norm": 0.092709481716156,
|
| 1060 |
+
"learning_rate": 4.491666666666667e-05,
|
| 1061 |
+
"loss": 0.1859,
|
| 1062 |
"step": 6900
|
| 1063 |
},
|
| 1064 |
{
|
| 1065 |
+
"epoch": 38.00417361111111,
|
| 1066 |
+
"grad_norm": 1.1074143648147583,
|
| 1067 |
+
"learning_rate": 4.476234567901235e-05,
|
| 1068 |
+
"loss": 0.0922,
|
| 1069 |
"step": 7000
|
| 1070 |
},
|
| 1071 |
{
|
| 1072 |
+
"epoch": 38.005006944444446,
|
| 1073 |
+
"eval_accuracy": 0.6301775147928994,
|
| 1074 |
+
"eval_f1": 0.6124992036234641,
|
| 1075 |
+
"eval_loss": 1.9035998582839966,
|
| 1076 |
+
"eval_precision": 0.6824995517303208,
|
| 1077 |
+
"eval_recall": 0.6301775147928994,
|
| 1078 |
+
"eval_runtime": 55.9719,
|
| 1079 |
+
"eval_samples_per_second": 6.039,
|
| 1080 |
+
"eval_steps_per_second": 3.019,
|
| 1081 |
+
"eval_top_10_accuracy": 0.9260355029585798,
|
| 1082 |
+
"eval_top_1_accuracy": 0.6301775147928994,
|
| 1083 |
+
"eval_top_5_accuracy": 0.8668639053254438,
|
| 1084 |
+
"step": 7030
|
| 1085 |
+
},
|
| 1086 |
+
{
|
| 1087 |
+
"epoch": 39.00194444444445,
|
| 1088 |
+
"grad_norm": 25.988325119018555,
|
| 1089 |
+
"learning_rate": 4.4608024691358025e-05,
|
| 1090 |
+
"loss": 0.1847,
|
| 1091 |
"step": 7100
|
| 1092 |
},
|
| 1093 |
{
|
| 1094 |
+
"epoch": 39.00472222222222,
|
| 1095 |
+
"grad_norm": 22.964984893798828,
|
| 1096 |
+
"learning_rate": 4.445370370370371e-05,
|
| 1097 |
+
"loss": 0.1547,
|
| 1098 |
"step": 7200
|
| 1099 |
},
|
| 1100 |
{
|
| 1101 |
+
"epoch": 39.005,
|
| 1102 |
+
"eval_accuracy": 0.6035502958579881,
|
| 1103 |
+
"eval_f1": 0.5948209446121037,
|
| 1104 |
+
"eval_loss": 1.989665150642395,
|
| 1105 |
+
"eval_precision": 0.6725563391243864,
|
| 1106 |
+
"eval_recall": 0.6035502958579881,
|
| 1107 |
+
"eval_runtime": 61.9464,
|
| 1108 |
+
"eval_samples_per_second": 5.456,
|
| 1109 |
+
"eval_steps_per_second": 2.728,
|
| 1110 |
+
"eval_top_10_accuracy": 0.9053254437869822,
|
| 1111 |
+
"eval_top_1_accuracy": 0.6035502958579881,
|
| 1112 |
+
"eval_top_5_accuracy": 0.8431952662721893,
|
| 1113 |
"step": 7210
|
| 1114 |
},
|
| 1115 |
{
|
| 1116 |
+
"epoch": 40.002493055555554,
|
| 1117 |
+
"grad_norm": 0.05825158208608627,
|
| 1118 |
+
"learning_rate": 4.429938271604939e-05,
|
| 1119 |
+
"loss": 0.1162,
|
| 1120 |
"step": 7300
|
| 1121 |
},
|
| 1122 |
{
|
| 1123 |
+
"epoch": 40.00499305555556,
|
| 1124 |
+
"eval_accuracy": 0.5828402366863905,
|
| 1125 |
+
"eval_f1": 0.5674965566977401,
|
| 1126 |
+
"eval_loss": 2.305551767349243,
|
| 1127 |
+
"eval_precision": 0.651775147928994,
|
| 1128 |
+
"eval_recall": 0.5828402366863905,
|
| 1129 |
+
"eval_runtime": 63.1766,
|
| 1130 |
+
"eval_samples_per_second": 5.35,
|
| 1131 |
+
"eval_steps_per_second": 2.675,
|
| 1132 |
+
"eval_top_10_accuracy": 0.8875739644970414,
|
| 1133 |
+
"eval_top_1_accuracy": 0.5828402366863905,
|
| 1134 |
+
"eval_top_5_accuracy": 0.8284023668639053,
|
| 1135 |
"step": 7390
|
| 1136 |
},
|
| 1137 |
{
|
| 1138 |
+
"epoch": 41.00026388888889,
|
| 1139 |
+
"grad_norm": 0.08664705604314804,
|
| 1140 |
+
"learning_rate": 4.4145061728395065e-05,
|
| 1141 |
+
"loss": 0.1941,
|
| 1142 |
"step": 7400
|
| 1143 |
},
|
| 1144 |
{
|
| 1145 |
+
"epoch": 41.00304166666667,
|
| 1146 |
+
"grad_norm": 1.6665265560150146,
|
| 1147 |
+
"learning_rate": 4.399074074074074e-05,
|
| 1148 |
+
"loss": 0.0514,
|
| 1149 |
"step": 7500
|
| 1150 |
},
|
| 1151 |
{
|
| 1152 |
+
"epoch": 41.00498611111111,
|
| 1153 |
+
"eval_accuracy": 0.5887573964497042,
|
| 1154 |
+
"eval_f1": 0.5767313882698497,
|
| 1155 |
+
"eval_loss": 2.3211424350738525,
|
| 1156 |
+
"eval_precision": 0.6510214144829529,
|
| 1157 |
+
"eval_recall": 0.5887573964497042,
|
| 1158 |
+
"eval_runtime": 55.5478,
|
| 1159 |
+
"eval_samples_per_second": 6.085,
|
| 1160 |
+
"eval_steps_per_second": 3.042,
|
| 1161 |
+
"eval_top_10_accuracy": 0.8816568047337278,
|
| 1162 |
+
"eval_top_1_accuracy": 0.5887573964497042,
|
| 1163 |
+
"eval_top_5_accuracy": 0.7988165680473372,
|
| 1164 |
"step": 7570
|
| 1165 |
},
|
| 1166 |
{
|
| 1167 |
+
"epoch": 42.0008125,
|
| 1168 |
+
"grad_norm": 0.07863179594278336,
|
| 1169 |
+
"learning_rate": 4.383641975308642e-05,
|
| 1170 |
+
"loss": 0.1709,
|
| 1171 |
"step": 7600
|
| 1172 |
},
|
| 1173 |
{
|
| 1174 |
+
"epoch": 42.003590277777775,
|
| 1175 |
+
"grad_norm": 0.14477179944515228,
|
| 1176 |
+
"learning_rate": 4.36820987654321e-05,
|
| 1177 |
+
"loss": 0.1138,
|
| 1178 |
"step": 7700
|
| 1179 |
},
|
| 1180 |
{
|
| 1181 |
+
"epoch": 42.005006944444446,
|
| 1182 |
+
"eval_accuracy": 0.5739644970414202,
|
| 1183 |
+
"eval_f1": 0.562847029406375,
|
| 1184 |
+
"eval_loss": 2.3148906230926514,
|
| 1185 |
+
"eval_precision": 0.6284772919388303,
|
| 1186 |
+
"eval_recall": 0.5739644970414202,
|
| 1187 |
+
"eval_runtime": 59.6045,
|
| 1188 |
+
"eval_samples_per_second": 5.671,
|
| 1189 |
+
"eval_steps_per_second": 2.835,
|
| 1190 |
+
"eval_top_10_accuracy": 0.9053254437869822,
|
| 1191 |
+
"eval_top_1_accuracy": 0.5739644970414202,
|
| 1192 |
+
"eval_top_5_accuracy": 0.849112426035503,
|
| 1193 |
+
"step": 7751
|
| 1194 |
+
},
|
| 1195 |
+
{
|
| 1196 |
+
"epoch": 43.00136111111111,
|
| 1197 |
+
"grad_norm": 0.37179991602897644,
|
| 1198 |
+
"learning_rate": 4.3527777777777776e-05,
|
| 1199 |
+
"loss": 0.1083,
|
| 1200 |
"step": 7800
|
| 1201 |
},
|
| 1202 |
{
|
| 1203 |
+
"epoch": 43.00413888888889,
|
| 1204 |
+
"grad_norm": 0.11481554806232452,
|
| 1205 |
+
"learning_rate": 4.337345679012345e-05,
|
| 1206 |
+
"loss": 0.1197,
|
| 1207 |
"step": 7900
|
| 1208 |
},
|
| 1209 |
{
|
| 1210 |
+
"epoch": 43.005,
|
| 1211 |
+
"eval_accuracy": 0.6124260355029586,
|
| 1212 |
+
"eval_f1": 0.6072961965713446,
|
| 1213 |
+
"eval_loss": 2.1155619621276855,
|
| 1214 |
+
"eval_precision": 0.6907422459197607,
|
| 1215 |
+
"eval_recall": 0.6124260355029586,
|
| 1216 |
+
"eval_runtime": 73.3826,
|
| 1217 |
+
"eval_samples_per_second": 4.606,
|
| 1218 |
+
"eval_steps_per_second": 2.303,
|
| 1219 |
+
"eval_top_10_accuracy": 0.9171597633136095,
|
| 1220 |
+
"eval_top_1_accuracy": 0.6124260355029586,
|
| 1221 |
+
"eval_top_5_accuracy": 0.8668639053254438,
|
| 1222 |
"step": 7931
|
| 1223 |
},
|
| 1224 |
{
|
| 1225 |
+
"epoch": 44.00190972222222,
|
| 1226 |
+
"grad_norm": 0.03723059222102165,
|
| 1227 |
"learning_rate": 4.322067901234568e-05,
|
| 1228 |
+
"loss": 0.1633,
|
| 1229 |
"step": 8000
|
| 1230 |
},
|
| 1231 |
{
|
| 1232 |
+
"epoch": 44.0046875,
|
| 1233 |
+
"grad_norm": 2.34633207321167,
|
| 1234 |
"learning_rate": 4.306635802469136e-05,
|
| 1235 |
+
"loss": 0.0673,
|
| 1236 |
"step": 8100
|
| 1237 |
},
|
| 1238 |
{
|
| 1239 |
+
"epoch": 44.00499305555556,
|
| 1240 |
+
"eval_accuracy": 0.5828402366863905,
|
| 1241 |
+
"eval_f1": 0.5640534640360605,
|
| 1242 |
+
"eval_loss": 2.2834787368774414,
|
| 1243 |
+
"eval_precision": 0.6274122984182156,
|
| 1244 |
+
"eval_recall": 0.5828402366863905,
|
| 1245 |
+
"eval_runtime": 72.2307,
|
| 1246 |
+
"eval_samples_per_second": 4.679,
|
| 1247 |
+
"eval_steps_per_second": 2.34,
|
| 1248 |
+
"eval_top_10_accuracy": 0.908284023668639,
|
| 1249 |
+
"eval_top_1_accuracy": 0.5828402366863905,
|
| 1250 |
+
"eval_top_5_accuracy": 0.8579881656804734,
|
| 1251 |
"step": 8111
|
| 1252 |
},
|
| 1253 |
{
|
| 1254 |
+
"epoch": 45.00245833333334,
|
| 1255 |
+
"grad_norm": 0.08984989672899246,
|
| 1256 |
"learning_rate": 4.291203703703704e-05,
|
| 1257 |
+
"loss": 0.1501,
|
| 1258 |
"step": 8200
|
| 1259 |
},
|
| 1260 |
{
|
| 1261 |
+
"epoch": 45.00498611111111,
|
| 1262 |
+
"eval_accuracy": 0.591715976331361,
|
| 1263 |
+
"eval_f1": 0.5756663372787633,
|
| 1264 |
+
"eval_loss": 2.271914482116699,
|
| 1265 |
+
"eval_precision": 0.6418798903266358,
|
| 1266 |
+
"eval_recall": 0.591715976331361,
|
| 1267 |
+
"eval_runtime": 57.8957,
|
| 1268 |
+
"eval_samples_per_second": 5.838,
|
| 1269 |
+
"eval_steps_per_second": 2.919,
|
| 1270 |
+
"eval_top_10_accuracy": 0.8905325443786982,
|
| 1271 |
+
"eval_top_1_accuracy": 0.591715976331361,
|
| 1272 |
+
"eval_top_5_accuracy": 0.8520710059171598,
|
| 1273 |
"step": 8291
|
| 1274 |
},
|
| 1275 |
{
|
| 1276 |
+
"epoch": 46.000229166666664,
|
| 1277 |
+
"grad_norm": 0.03222246095538139,
|
| 1278 |
"learning_rate": 4.2757716049382715e-05,
|
| 1279 |
+
"loss": 0.059,
|
| 1280 |
"step": 8300
|
| 1281 |
},
|
| 1282 |
{
|
| 1283 |
+
"epoch": 46.003006944444444,
|
| 1284 |
+
"grad_norm": 0.024875333532691002,
|
| 1285 |
+
"learning_rate": 4.2604938271604936e-05,
|
| 1286 |
+
"loss": 0.2022,
|
| 1287 |
"step": 8400
|
| 1288 |
},
|
| 1289 |
{
|
| 1290 |
+
"epoch": 46.005006944444446,
|
| 1291 |
+
"eval_accuracy": 0.5562130177514792,
|
| 1292 |
+
"eval_f1": 0.5402277209969517,
|
| 1293 |
+
"eval_loss": 2.3422443866729736,
|
| 1294 |
+
"eval_precision": 0.6034023668639052,
|
| 1295 |
+
"eval_recall": 0.5562130177514792,
|
| 1296 |
+
"eval_runtime": 58.5821,
|
| 1297 |
+
"eval_samples_per_second": 5.77,
|
| 1298 |
+
"eval_steps_per_second": 2.885,
|
| 1299 |
+
"eval_top_10_accuracy": 0.9053254437869822,
|
| 1300 |
+
"eval_top_1_accuracy": 0.5562130177514792,
|
| 1301 |
+
"eval_top_5_accuracy": 0.849112426035503,
|
| 1302 |
+
"step": 8472
|
| 1303 |
},
|
| 1304 |
{
|
| 1305 |
+
"epoch": 47.00077777777778,
|
| 1306 |
+
"grad_norm": 1.308508038520813,
|
| 1307 |
+
"learning_rate": 4.2450617283950614e-05,
|
| 1308 |
+
"loss": 0.1197,
|
| 1309 |
"step": 8500
|
| 1310 |
},
|
| 1311 |
{
|
| 1312 |
+
"epoch": 47.00355555555556,
|
| 1313 |
+
"grad_norm": 4.812650203704834,
|
| 1314 |
+
"learning_rate": 4.22962962962963e-05,
|
| 1315 |
+
"loss": 0.2185,
|
| 1316 |
"step": 8600
|
| 1317 |
},
|
| 1318 |
{
|
| 1319 |
+
"epoch": 47.005,
|
| 1320 |
+
"eval_accuracy": 0.5236686390532544,
|
| 1321 |
+
"eval_f1": 0.5125209685120928,
|
| 1322 |
+
"eval_loss": 2.64306378364563,
|
| 1323 |
+
"eval_precision": 0.580839847007616,
|
| 1324 |
+
"eval_recall": 0.5236686390532544,
|
| 1325 |
+
"eval_runtime": 57.4714,
|
| 1326 |
+
"eval_samples_per_second": 5.881,
|
| 1327 |
+
"eval_steps_per_second": 2.941,
|
| 1328 |
+
"eval_top_10_accuracy": 0.8816568047337278,
|
| 1329 |
+
"eval_top_1_accuracy": 0.5236686390532544,
|
| 1330 |
+
"eval_top_5_accuracy": 0.8284023668639053,
|
| 1331 |
"step": 8652
|
| 1332 |
},
|
| 1333 |
{
|
| 1334 |
+
"epoch": 48.00132638888889,
|
| 1335 |
+
"grad_norm": 0.9780556559562683,
|
| 1336 |
+
"learning_rate": 4.2141975308641976e-05,
|
| 1337 |
+
"loss": 0.1822,
|
| 1338 |
"step": 8700
|
| 1339 |
},
|
| 1340 |
{
|
| 1341 |
+
"epoch": 48.004104166666664,
|
| 1342 |
+
"grad_norm": 0.10755691677331924,
|
| 1343 |
+
"learning_rate": 4.1987654320987654e-05,
|
| 1344 |
+
"loss": 0.2385,
|
| 1345 |
"step": 8800
|
| 1346 |
},
|
| 1347 |
{
|
| 1348 |
+
"epoch": 48.00499305555556,
|
| 1349 |
+
"eval_accuracy": 0.5798816568047337,
|
| 1350 |
+
"eval_f1": 0.5622720967395524,
|
| 1351 |
+
"eval_loss": 2.3147056102752686,
|
| 1352 |
+
"eval_precision": 0.6323823612285151,
|
| 1353 |
+
"eval_recall": 0.5798816568047337,
|
| 1354 |
+
"eval_runtime": 61.6966,
|
| 1355 |
+
"eval_samples_per_second": 5.478,
|
| 1356 |
+
"eval_steps_per_second": 2.739,
|
| 1357 |
+
"eval_top_10_accuracy": 0.9053254437869822,
|
| 1358 |
+
"eval_top_1_accuracy": 0.5798816568047337,
|
| 1359 |
+
"eval_top_5_accuracy": 0.8520710059171598,
|
| 1360 |
"step": 8832
|
| 1361 |
},
|
| 1362 |
{
|
| 1363 |
+
"epoch": 49.001875,
|
| 1364 |
+
"grad_norm": 0.048398274928331375,
|
| 1365 |
"learning_rate": 4.183333333333334e-05,
|
| 1366 |
+
"loss": 0.1209,
|
| 1367 |
"step": 8900
|
| 1368 |
},
|
| 1369 |
{
|
| 1370 |
+
"epoch": 49.00465277777778,
|
| 1371 |
+
"grad_norm": 116.13093566894531,
|
| 1372 |
"learning_rate": 4.1679012345679016e-05,
|
| 1373 |
+
"loss": 0.1769,
|
| 1374 |
"step": 9000
|
| 1375 |
},
|
| 1376 |
{
|
| 1377 |
+
"epoch": 49.00498611111111,
|
| 1378 |
+
"eval_accuracy": 0.5769230769230769,
|
| 1379 |
+
"eval_f1": 0.5621830092983939,
|
| 1380 |
+
"eval_loss": 2.345111131668091,
|
| 1381 |
+
"eval_precision": 0.6246256701878005,
|
| 1382 |
+
"eval_recall": 0.5769230769230769,
|
| 1383 |
+
"eval_runtime": 56.4804,
|
| 1384 |
+
"eval_samples_per_second": 5.984,
|
| 1385 |
+
"eval_steps_per_second": 2.992,
|
| 1386 |
+
"eval_top_10_accuracy": 0.8875739644970414,
|
| 1387 |
+
"eval_top_1_accuracy": 0.5769230769230769,
|
| 1388 |
+
"eval_top_5_accuracy": 0.8372781065088757,
|
| 1389 |
"step": 9012
|
| 1390 |
},
|
| 1391 |
{
|
| 1392 |
+
"epoch": 50.00242361111111,
|
| 1393 |
+
"grad_norm": 6.833434104919434,
|
| 1394 |
"learning_rate": 4.1524691358024694e-05,
|
| 1395 |
+
"loss": 0.1927,
|
| 1396 |
"step": 9100
|
| 1397 |
},
|
| 1398 |
{
|
| 1399 |
+
"epoch": 50.005006944444446,
|
| 1400 |
+
"eval_accuracy": 0.5562130177514792,
|
| 1401 |
+
"eval_f1": 0.5347152354548804,
|
| 1402 |
+
"eval_loss": 2.7140073776245117,
|
| 1403 |
+
"eval_precision": 0.6024619611158072,
|
| 1404 |
+
"eval_recall": 0.5562130177514792,
|
| 1405 |
+
"eval_runtime": 57.0126,
|
| 1406 |
+
"eval_samples_per_second": 5.929,
|
| 1407 |
+
"eval_steps_per_second": 2.964,
|
| 1408 |
+
"eval_top_10_accuracy": 0.8727810650887574,
|
| 1409 |
+
"eval_top_1_accuracy": 0.5562130177514792,
|
| 1410 |
+
"eval_top_5_accuracy": 0.8017751479289941,
|
| 1411 |
+
"step": 9193
|
| 1412 |
},
|
| 1413 |
{
|
| 1414 |
+
"epoch": 51.000194444444446,
|
| 1415 |
+
"grad_norm": 169.2981719970703,
|
| 1416 |
"learning_rate": 4.137037037037037e-05,
|
| 1417 |
+
"loss": 0.302,
|
| 1418 |
"step": 9200
|
| 1419 |
},
|
| 1420 |
{
|
| 1421 |
+
"epoch": 51.00297222222222,
|
| 1422 |
+
"grad_norm": 0.07411950081586838,
|
| 1423 |
"learning_rate": 4.121604938271605e-05,
|
| 1424 |
+
"loss": 0.2048,
|
| 1425 |
"step": 9300
|
| 1426 |
},
|
| 1427 |
{
|
| 1428 |
+
"epoch": 51.005,
|
| 1429 |
+
"eval_accuracy": 0.591715976331361,
|
| 1430 |
+
"eval_f1": 0.5748348298644157,
|
| 1431 |
+
"eval_loss": 2.3875668048858643,
|
| 1432 |
+
"eval_precision": 0.6367497886728656,
|
| 1433 |
+
"eval_recall": 0.591715976331361,
|
| 1434 |
+
"eval_runtime": 57.5583,
|
| 1435 |
+
"eval_samples_per_second": 5.872,
|
| 1436 |
+
"eval_steps_per_second": 2.936,
|
| 1437 |
+
"eval_top_10_accuracy": 0.893491124260355,
|
| 1438 |
+
"eval_top_1_accuracy": 0.591715976331361,
|
| 1439 |
+
"eval_top_5_accuracy": 0.8224852071005917,
|
| 1440 |
"step": 9373
|
| 1441 |
},
|
| 1442 |
{
|
| 1443 |
+
"epoch": 52.00074305555555,
|
| 1444 |
+
"grad_norm": 80.62044525146484,
|
| 1445 |
"learning_rate": 4.1061728395061733e-05,
|
| 1446 |
+
"loss": 0.2414,
|
| 1447 |
"step": 9400
|
| 1448 |
},
|
| 1449 |
{
|
| 1450 |
+
"epoch": 52.00352083333333,
|
| 1451 |
+
"grad_norm": 2.4975171089172363,
|
| 1452 |
"learning_rate": 4.090740740740741e-05,
|
| 1453 |
+
"loss": 0.1608,
|
| 1454 |
"step": 9500
|
| 1455 |
},
|
| 1456 |
{
|
| 1457 |
+
"epoch": 52.00499305555556,
|
| 1458 |
+
"eval_accuracy": 0.5266272189349113,
|
| 1459 |
+
"eval_f1": 0.5013487498694599,
|
| 1460 |
+
"eval_loss": 2.6982638835906982,
|
| 1461 |
+
"eval_precision": 0.5645437895437895,
|
| 1462 |
+
"eval_recall": 0.5266272189349113,
|
| 1463 |
+
"eval_runtime": 59.8704,
|
| 1464 |
+
"eval_samples_per_second": 5.646,
|
| 1465 |
+
"eval_steps_per_second": 2.823,
|
| 1466 |
+
"eval_top_10_accuracy": 0.8579881656804734,
|
| 1467 |
+
"eval_top_1_accuracy": 0.5266272189349113,
|
| 1468 |
+
"eval_top_5_accuracy": 0.8076923076923077,
|
| 1469 |
"step": 9553
|
| 1470 |
},
|
| 1471 |
{
|
| 1472 |
+
"epoch": 53.00129166666667,
|
| 1473 |
+
"grad_norm": 0.140806183218956,
|
| 1474 |
"learning_rate": 4.075308641975309e-05,
|
| 1475 |
+
"loss": 0.1078,
|
| 1476 |
"step": 9600
|
| 1477 |
},
|
| 1478 |
{
|
| 1479 |
+
"epoch": 53.00406944444445,
|
| 1480 |
+
"grad_norm": 0.03946617618203163,
|
| 1481 |
"learning_rate": 4.0598765432098767e-05,
|
| 1482 |
+
"loss": 0.1256,
|
| 1483 |
"step": 9700
|
| 1484 |
},
|
| 1485 |
{
|
| 1486 |
+
"epoch": 53.00498611111111,
|
| 1487 |
+
"eval_accuracy": 0.5384615384615384,
|
| 1488 |
+
"eval_f1": 0.5257081839922076,
|
| 1489 |
+
"eval_loss": 2.7464194297790527,
|
| 1490 |
+
"eval_precision": 0.5773428494582341,
|
| 1491 |
+
"eval_recall": 0.5384615384615384,
|
| 1492 |
+
"eval_runtime": 79.9762,
|
| 1493 |
+
"eval_samples_per_second": 4.226,
|
| 1494 |
+
"eval_steps_per_second": 2.113,
|
| 1495 |
+
"eval_top_10_accuracy": 0.8905325443786982,
|
| 1496 |
+
"eval_top_1_accuracy": 0.5384615384615384,
|
| 1497 |
+
"eval_top_5_accuracy": 0.8017751479289941,
|
| 1498 |
"step": 9733
|
| 1499 |
},
|
| 1500 |
{
|
| 1501 |
+
"epoch": 54.00184027777778,
|
| 1502 |
+
"grad_norm": 0.018318645656108856,
|
| 1503 |
"learning_rate": 4.0444444444444444e-05,
|
| 1504 |
+
"loss": 0.1146,
|
| 1505 |
"step": 9800
|
| 1506 |
},
|
| 1507 |
{
|
| 1508 |
+
"epoch": 54.004618055555554,
|
| 1509 |
+
"grad_norm": 107.91517639160156,
|
| 1510 |
"learning_rate": 4.029012345679012e-05,
|
| 1511 |
+
"loss": 0.1327,
|
| 1512 |
"step": 9900
|
| 1513 |
},
|
| 1514 |
{
|
| 1515 |
+
"epoch": 54.005006944444446,
|
| 1516 |
+
"eval_accuracy": 0.5650887573964497,
|
| 1517 |
+
"eval_f1": 0.5342032063925555,
|
| 1518 |
+
"eval_loss": 2.513326644897461,
|
| 1519 |
+
"eval_precision": 0.5812130177514792,
|
| 1520 |
+
"eval_recall": 0.5650887573964497,
|
| 1521 |
+
"eval_runtime": 57.9139,
|
| 1522 |
+
"eval_samples_per_second": 5.836,
|
| 1523 |
+
"eval_steps_per_second": 2.918,
|
| 1524 |
+
"eval_top_10_accuracy": 0.8846153846153846,
|
| 1525 |
+
"eval_top_1_accuracy": 0.5650887573964497,
|
| 1526 |
+
"eval_top_5_accuracy": 0.7988165680473372,
|
| 1527 |
+
"step": 9914
|
| 1528 |
+
},
|
| 1529 |
+
{
|
| 1530 |
+
"epoch": 55.00238888888889,
|
| 1531 |
+
"grad_norm": 0.045430343598127365,
|
| 1532 |
+
"learning_rate": 4.01358024691358e-05,
|
| 1533 |
+
"loss": 0.0503,
|
| 1534 |
+
"step": 10000
|
| 1535 |
+
},
|
| 1536 |
+
{
|
| 1537 |
+
"epoch": 55.005,
|
| 1538 |
+
"eval_accuracy": 0.5769230769230769,
|
| 1539 |
+
"eval_f1": 0.5635061355475556,
|
| 1540 |
+
"eval_loss": 2.5686514377593994,
|
| 1541 |
+
"eval_precision": 0.6335951441720672,
|
| 1542 |
+
"eval_recall": 0.5769230769230769,
|
| 1543 |
+
"eval_runtime": 62.2713,
|
| 1544 |
+
"eval_samples_per_second": 5.428,
|
| 1545 |
+
"eval_steps_per_second": 2.714,
|
| 1546 |
+
"eval_top_10_accuracy": 0.8964497041420119,
|
| 1547 |
+
"eval_top_1_accuracy": 0.5769230769230769,
|
| 1548 |
+
"eval_top_5_accuracy": 0.8313609467455622,
|
| 1549 |
+
"step": 10094
|
| 1550 |
+
},
|
| 1551 |
+
{
|
| 1552 |
+
"epoch": 56.00015972222222,
|
| 1553 |
+
"grad_norm": 0.011242308653891087,
|
| 1554 |
+
"learning_rate": 3.9981481481481484e-05,
|
| 1555 |
+
"loss": 0.1127,
|
| 1556 |
+
"step": 10100
|
| 1557 |
+
},
|
| 1558 |
+
{
|
| 1559 |
+
"epoch": 56.0029375,
|
| 1560 |
+
"grad_norm": 0.01931876875460148,
|
| 1561 |
+
"learning_rate": 3.982716049382716e-05,
|
| 1562 |
+
"loss": 0.0841,
|
| 1563 |
+
"step": 10200
|
| 1564 |
+
},
|
| 1565 |
+
{
|
| 1566 |
+
"epoch": 56.00499305555556,
|
| 1567 |
+
"eval_accuracy": 0.5502958579881657,
|
| 1568 |
+
"eval_f1": 0.53286526097177,
|
| 1569 |
+
"eval_loss": 2.731083869934082,
|
| 1570 |
+
"eval_precision": 0.6024830938292477,
|
| 1571 |
+
"eval_recall": 0.5502958579881657,
|
| 1572 |
+
"eval_runtime": 64.4351,
|
| 1573 |
+
"eval_samples_per_second": 5.246,
|
| 1574 |
+
"eval_steps_per_second": 2.623,
|
| 1575 |
+
"eval_top_10_accuracy": 0.8905325443786982,
|
| 1576 |
+
"eval_top_1_accuracy": 0.5502958579881657,
|
| 1577 |
+
"eval_top_5_accuracy": 0.8284023668639053,
|
| 1578 |
+
"step": 10274
|
| 1579 |
+
},
|
| 1580 |
+
{
|
| 1581 |
+
"epoch": 57.000708333333336,
|
| 1582 |
+
"grad_norm": 0.7857475280761719,
|
| 1583 |
+
"learning_rate": 3.9672839506172846e-05,
|
| 1584 |
+
"loss": 0.2066,
|
| 1585 |
+
"step": 10300
|
| 1586 |
+
},
|
| 1587 |
+
{
|
| 1588 |
+
"epoch": 57.00348611111111,
|
| 1589 |
+
"grad_norm": 0.0037257361691445112,
|
| 1590 |
+
"learning_rate": 3.9518518518518524e-05,
|
| 1591 |
+
"loss": 0.0888,
|
| 1592 |
+
"step": 10400
|
| 1593 |
+
},
|
| 1594 |
+
{
|
| 1595 |
+
"epoch": 57.00498611111111,
|
| 1596 |
+
"eval_accuracy": 0.5591715976331361,
|
| 1597 |
+
"eval_f1": 0.5473162081759367,
|
| 1598 |
+
"eval_loss": 2.6771368980407715,
|
| 1599 |
+
"eval_precision": 0.6141800507185123,
|
| 1600 |
+
"eval_recall": 0.5591715976331361,
|
| 1601 |
+
"eval_runtime": 66.5741,
|
| 1602 |
+
"eval_samples_per_second": 5.077,
|
| 1603 |
+
"eval_steps_per_second": 2.539,
|
| 1604 |
+
"eval_top_10_accuracy": 0.893491124260355,
|
| 1605 |
+
"eval_top_1_accuracy": 0.5591715976331361,
|
| 1606 |
+
"eval_top_5_accuracy": 0.8195266272189349,
|
| 1607 |
+
"step": 10454
|
| 1608 |
+
},
|
| 1609 |
+
{
|
| 1610 |
+
"epoch": 58.00125694444444,
|
| 1611 |
+
"grad_norm": 217.6994171142578,
|
| 1612 |
+
"learning_rate": 3.93641975308642e-05,
|
| 1613 |
+
"loss": 0.0478,
|
| 1614 |
+
"step": 10500
|
| 1615 |
+
},
|
| 1616 |
+
{
|
| 1617 |
+
"epoch": 58.00403472222222,
|
| 1618 |
+
"grad_norm": 0.19375790655612946,
|
| 1619 |
+
"learning_rate": 3.920987654320988e-05,
|
| 1620 |
+
"loss": 0.0629,
|
| 1621 |
+
"step": 10600
|
| 1622 |
+
},
|
| 1623 |
+
{
|
| 1624 |
+
"epoch": 58.005006944444446,
|
| 1625 |
+
"eval_accuracy": 0.5295857988165681,
|
| 1626 |
+
"eval_f1": 0.5132937382067211,
|
| 1627 |
+
"eval_loss": 2.8300716876983643,
|
| 1628 |
+
"eval_precision": 0.5767704517704518,
|
| 1629 |
+
"eval_recall": 0.5295857988165681,
|
| 1630 |
+
"eval_runtime": 66.7729,
|
| 1631 |
+
"eval_samples_per_second": 5.062,
|
| 1632 |
+
"eval_steps_per_second": 2.531,
|
| 1633 |
"eval_top_10_accuracy": 0.8698224852071006,
|
| 1634 |
+
"eval_top_1_accuracy": 0.5295857988165681,
|
| 1635 |
+
"eval_top_5_accuracy": 0.7988165680473372,
|
| 1636 |
+
"step": 10635
|
| 1637 |
},
|
| 1638 |
{
|
| 1639 |
+
"epoch": 58.005006944444446,
|
| 1640 |
+
"step": 10635,
|
| 1641 |
+
"total_flos": 1.0610658302492344e+20,
|
| 1642 |
+
"train_loss": 5.244687236773334,
|
| 1643 |
+
"train_runtime": 25999.5358,
|
| 1644 |
+
"train_samples_per_second": 11.077,
|
| 1645 |
+
"train_steps_per_second": 1.385
|
| 1646 |
}
|
| 1647 |
],
|
| 1648 |
"logging_steps": 100,
|
|
|
|
| 1671 |
"attributes": {}
|
| 1672 |
}
|
| 1673 |
},
|
| 1674 |
+
"total_flos": 1.0610658302492344e+20,
|
| 1675 |
"train_batch_size": 2,
|
| 1676 |
"trial_name": null,
|
| 1677 |
"trial_params": null
|