End of training
Browse files- all_results.json +6 -6
- train_results.json +6 -6
- trainer_state.json +538 -700
all_results.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch": 14.
|
| 3 |
-
"total_flos":
|
| 4 |
-
"train_loss": 0.
|
| 5 |
-
"train_runtime":
|
| 6 |
-
"train_samples_per_second":
|
| 7 |
-
"train_steps_per_second": 0.
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 14.96,
|
| 3 |
+
"total_flos": 6.546875329145733e+18,
|
| 4 |
+
"train_loss": 0.44751356618874,
|
| 5 |
+
"train_runtime": 3414.1758,
|
| 6 |
+
"train_samples_per_second": 24.814,
|
| 7 |
+
"train_steps_per_second": 0.387
|
| 8 |
}
|
train_results.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch": 14.
|
| 3 |
-
"total_flos":
|
| 4 |
-
"train_loss": 0.
|
| 5 |
-
"train_runtime":
|
| 6 |
-
"train_samples_per_second":
|
| 7 |
-
"train_steps_per_second": 0.
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 14.96,
|
| 3 |
+
"total_flos": 6.546875329145733e+18,
|
| 4 |
+
"train_loss": 0.44751356618874,
|
| 5 |
+
"train_runtime": 3414.1758,
|
| 6 |
+
"train_samples_per_second": 24.814,
|
| 7 |
+
"train_steps_per_second": 0.387
|
| 8 |
}
|
trainer_state.json
CHANGED
|
@@ -1,1117 +1,955 @@
|
|
| 1 |
{
|
| 2 |
-
"best_metric": 0.
|
| 3 |
-
"best_model_checkpoint": "vit-base-patch16-224-brand/checkpoint-
|
| 4 |
-
"epoch": 14.
|
| 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.
|
| 13 |
-
"learning_rate": 3.
|
| 14 |
-
"loss": 2.
|
| 15 |
"step": 10
|
| 16 |
},
|
| 17 |
{
|
| 18 |
-
"epoch": 0.
|
| 19 |
-
"learning_rate":
|
| 20 |
-
"loss": 2.
|
| 21 |
"step": 20
|
| 22 |
},
|
| 23 |
{
|
| 24 |
-
"epoch": 0.
|
| 25 |
-
"learning_rate":
|
| 26 |
-
"loss": 2.
|
| 27 |
"step": 30
|
| 28 |
},
|
| 29 |
{
|
| 30 |
-
"epoch": 0.
|
| 31 |
-
"learning_rate": 1.
|
| 32 |
-
"loss": 2.
|
| 33 |
"step": 40
|
| 34 |
},
|
| 35 |
{
|
| 36 |
-
"epoch": 0.
|
| 37 |
-
"learning_rate": 1.
|
| 38 |
-
"loss":
|
| 39 |
"step": 50
|
| 40 |
},
|
| 41 |
{
|
| 42 |
-
"epoch": 0.
|
| 43 |
-
"learning_rate":
|
| 44 |
-
"loss":
|
| 45 |
"step": 60
|
| 46 |
},
|
| 47 |
{
|
| 48 |
-
"epoch": 0.
|
| 49 |
-
"learning_rate": 2.
|
| 50 |
-
"loss": 1.
|
| 51 |
"step": 70
|
| 52 |
},
|
| 53 |
{
|
| 54 |
-
"epoch": 0.
|
| 55 |
-
"learning_rate":
|
| 56 |
-
"loss": 1.
|
| 57 |
"step": 80
|
| 58 |
},
|
| 59 |
{
|
| 60 |
-
"epoch":
|
| 61 |
-
"
|
| 62 |
-
"
|
| 63 |
-
"
|
|
|
|
|
|
|
|
|
|
| 64 |
},
|
| 65 |
{
|
| 66 |
-
"epoch":
|
| 67 |
-
"learning_rate": 3.
|
| 68 |
-
"loss": 1.
|
| 69 |
-
"step":
|
| 70 |
},
|
| 71 |
{
|
| 72 |
-
"epoch":
|
| 73 |
-
"
|
| 74 |
-
"
|
| 75 |
-
"
|
| 76 |
-
"eval_samples_per_second": 76.304,
|
| 77 |
-
"eval_steps_per_second": 4.782,
|
| 78 |
-
"step": 106
|
| 79 |
},
|
| 80 |
{
|
| 81 |
-
"epoch": 1.
|
| 82 |
-
"learning_rate":
|
| 83 |
-
"loss": 1.
|
| 84 |
"step": 110
|
| 85 |
},
|
| 86 |
{
|
| 87 |
-
"epoch": 1.
|
| 88 |
-
"learning_rate":
|
| 89 |
-
"loss": 1.
|
| 90 |
"step": 120
|
| 91 |
},
|
| 92 |
{
|
| 93 |
-
"epoch": 1.
|
| 94 |
-
"learning_rate": 4.
|
| 95 |
-
"loss": 1.
|
| 96 |
"step": 130
|
| 97 |
},
|
| 98 |
{
|
| 99 |
-
"epoch": 1.
|
| 100 |
-
"learning_rate": 4.
|
| 101 |
-
"loss": 1.
|
| 102 |
"step": 140
|
| 103 |
},
|
| 104 |
{
|
| 105 |
-
"epoch": 1.
|
| 106 |
-
"learning_rate": 4.
|
| 107 |
-
"loss":
|
| 108 |
"step": 150
|
| 109 |
},
|
| 110 |
{
|
| 111 |
-
"epoch": 1.
|
| 112 |
-
"learning_rate": 4.
|
| 113 |
-
"loss":
|
| 114 |
"step": 160
|
| 115 |
},
|
| 116 |
{
|
| 117 |
-
"epoch": 1.
|
| 118 |
-
"learning_rate": 4.
|
| 119 |
-
"loss":
|
| 120 |
"step": 170
|
| 121 |
},
|
| 122 |
{
|
| 123 |
-
"epoch": 1.
|
| 124 |
-
"
|
| 125 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 126 |
"step": 180
|
| 127 |
},
|
| 128 |
{
|
| 129 |
-
"epoch":
|
| 130 |
-
"learning_rate": 4.
|
| 131 |
-
"loss":
|
| 132 |
"step": 190
|
| 133 |
},
|
| 134 |
{
|
| 135 |
-
"epoch":
|
| 136 |
-
"learning_rate": 4.
|
| 137 |
-
"loss":
|
| 138 |
"step": 200
|
| 139 |
},
|
| 140 |
{
|
| 141 |
-
"epoch":
|
| 142 |
-
"learning_rate": 4.
|
| 143 |
-
"loss":
|
| 144 |
"step": 210
|
| 145 |
},
|
| 146 |
{
|
| 147 |
-
"epoch": 2.
|
| 148 |
-
"
|
| 149 |
-
"
|
| 150 |
-
"eval_runtime": 29.9218,
|
| 151 |
-
"eval_samples_per_second": 75.731,
|
| 152 |
-
"eval_steps_per_second": 4.746,
|
| 153 |
-
"step": 213
|
| 154 |
-
},
|
| 155 |
-
{
|
| 156 |
-
"epoch": 2.06,
|
| 157 |
-
"learning_rate": 4.7868623340321454e-05,
|
| 158 |
-
"loss": 0.9655,
|
| 159 |
"step": 220
|
| 160 |
},
|
| 161 |
{
|
| 162 |
-
"epoch": 2.
|
| 163 |
-
"learning_rate": 4.
|
| 164 |
-
"loss": 0.
|
| 165 |
"step": 230
|
| 166 |
},
|
| 167 |
{
|
| 168 |
-
"epoch": 2.
|
| 169 |
-
"learning_rate": 4.
|
| 170 |
-
"loss": 0.
|
| 171 |
"step": 240
|
| 172 |
},
|
| 173 |
{
|
| 174 |
-
"epoch": 2.
|
| 175 |
-
"learning_rate": 4.
|
| 176 |
-
"loss": 0.
|
| 177 |
"step": 250
|
| 178 |
},
|
| 179 |
{
|
| 180 |
-
"epoch": 2.
|
| 181 |
-
"learning_rate": 4.
|
| 182 |
-
"loss": 0.
|
| 183 |
"step": 260
|
| 184 |
},
|
| 185 |
{
|
| 186 |
-
"epoch": 2.
|
| 187 |
-
"
|
| 188 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 189 |
"step": 270
|
| 190 |
},
|
| 191 |
{
|
| 192 |
-
"epoch":
|
| 193 |
-
"learning_rate": 4.
|
| 194 |
-
"loss": 0.
|
| 195 |
"step": 280
|
| 196 |
},
|
| 197 |
{
|
| 198 |
-
"epoch":
|
| 199 |
-
"learning_rate": 4.
|
| 200 |
-
"loss": 0.
|
| 201 |
"step": 290
|
| 202 |
},
|
| 203 |
{
|
| 204 |
-
"epoch":
|
| 205 |
-
"learning_rate": 4.
|
| 206 |
-
"loss": 0.
|
| 207 |
"step": 300
|
| 208 |
},
|
| 209 |
{
|
| 210 |
-
"epoch":
|
| 211 |
-
"learning_rate": 4.
|
| 212 |
-
"loss": 0.
|
| 213 |
"step": 310
|
| 214 |
},
|
| 215 |
{
|
| 216 |
-
"epoch": 3.
|
| 217 |
-
"learning_rate": 4.
|
| 218 |
-
"loss": 0.
|
| 219 |
"step": 320
|
| 220 |
},
|
| 221 |
{
|
| 222 |
-
"epoch": 3.
|
| 223 |
-
"
|
| 224 |
-
"
|
| 225 |
-
"eval_runtime": 29.9179,
|
| 226 |
-
"eval_samples_per_second": 75.741,
|
| 227 |
-
"eval_steps_per_second": 4.746,
|
| 228 |
-
"step": 320
|
| 229 |
-
},
|
| 230 |
-
{
|
| 231 |
-
"epoch": 3.09,
|
| 232 |
-
"learning_rate": 4.402515723270441e-05,
|
| 233 |
-
"loss": 0.7598,
|
| 234 |
"step": 330
|
| 235 |
},
|
| 236 |
{
|
| 237 |
-
"epoch": 3.
|
| 238 |
-
"learning_rate": 4.
|
| 239 |
-
"loss": 0.
|
| 240 |
"step": 340
|
| 241 |
},
|
| 242 |
{
|
| 243 |
-
"epoch": 3.
|
| 244 |
-
"learning_rate": 4.
|
| 245 |
-
"loss": 0.
|
| 246 |
"step": 350
|
| 247 |
},
|
| 248 |
{
|
| 249 |
-
"epoch":
|
| 250 |
-
"
|
| 251 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 252 |
"step": 360
|
| 253 |
},
|
| 254 |
{
|
| 255 |
-
"epoch":
|
| 256 |
-
"learning_rate":
|
| 257 |
-
"loss": 0.
|
| 258 |
"step": 370
|
| 259 |
},
|
| 260 |
{
|
| 261 |
-
"epoch":
|
| 262 |
-
"learning_rate":
|
| 263 |
-
"loss": 0.
|
| 264 |
"step": 380
|
| 265 |
},
|
| 266 |
{
|
| 267 |
-
"epoch":
|
| 268 |
-
"learning_rate":
|
| 269 |
-
"loss": 0.
|
| 270 |
"step": 390
|
| 271 |
},
|
| 272 |
{
|
| 273 |
-
"epoch":
|
| 274 |
-
"learning_rate":
|
| 275 |
-
"loss": 0.
|
| 276 |
"step": 400
|
| 277 |
},
|
| 278 |
{
|
| 279 |
-
"epoch":
|
| 280 |
-
"learning_rate":
|
| 281 |
-
"loss": 0.
|
| 282 |
"step": 410
|
| 283 |
},
|
| 284 |
{
|
| 285 |
-
"epoch":
|
| 286 |
-
"learning_rate":
|
| 287 |
-
"loss": 0.
|
| 288 |
"step": 420
|
| 289 |
},
|
| 290 |
{
|
| 291 |
-
"epoch": 4.
|
| 292 |
-
"
|
| 293 |
-
"
|
| 294 |
-
"eval_runtime": 29.5461,
|
| 295 |
-
"eval_samples_per_second": 76.694,
|
| 296 |
-
"eval_steps_per_second": 4.806,
|
| 297 |
-
"step": 427
|
| 298 |
-
},
|
| 299 |
-
{
|
| 300 |
-
"epoch": 4.03,
|
| 301 |
-
"learning_rate": 4.053109713487072e-05,
|
| 302 |
-
"loss": 0.5489,
|
| 303 |
"step": 430
|
| 304 |
},
|
| 305 |
{
|
| 306 |
-
"epoch": 4.
|
| 307 |
-
"learning_rate":
|
| 308 |
-
"loss": 0.
|
| 309 |
"step": 440
|
| 310 |
},
|
| 311 |
{
|
| 312 |
-
"epoch":
|
| 313 |
-
"
|
| 314 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 315 |
"step": 450
|
| 316 |
},
|
| 317 |
{
|
| 318 |
-
"epoch":
|
| 319 |
-
"learning_rate": 3.
|
| 320 |
-
"loss": 0.
|
| 321 |
"step": 460
|
| 322 |
},
|
| 323 |
{
|
| 324 |
-
"epoch":
|
| 325 |
-
"learning_rate": 3.
|
| 326 |
-
"loss": 0.
|
| 327 |
"step": 470
|
| 328 |
},
|
| 329 |
{
|
| 330 |
-
"epoch":
|
| 331 |
-
"learning_rate": 3.
|
| 332 |
-
"loss": 0.
|
| 333 |
"step": 480
|
| 334 |
},
|
| 335 |
{
|
| 336 |
-
"epoch":
|
| 337 |
-
"learning_rate": 3.
|
| 338 |
-
"loss": 0.
|
| 339 |
"step": 490
|
| 340 |
},
|
| 341 |
{
|
| 342 |
-
"epoch":
|
| 343 |
-
"learning_rate": 3.
|
| 344 |
-
"loss": 0.
|
| 345 |
"step": 500
|
| 346 |
},
|
| 347 |
{
|
| 348 |
-
"epoch":
|
| 349 |
-
"learning_rate": 3.
|
| 350 |
-
"loss": 0.
|
| 351 |
"step": 510
|
| 352 |
},
|
| 353 |
{
|
| 354 |
-
"epoch":
|
| 355 |
-
"learning_rate": 3.
|
| 356 |
-
"loss": 0.
|
| 357 |
"step": 520
|
| 358 |
},
|
| 359 |
{
|
| 360 |
-
"epoch":
|
| 361 |
-
"
|
| 362 |
-
"
|
| 363 |
-
"
|
|
|
|
|
|
|
|
|
|
| 364 |
},
|
| 365 |
{
|
| 366 |
-
"epoch":
|
| 367 |
-
"
|
| 368 |
-
"
|
| 369 |
-
"
|
| 370 |
-
"eval_samples_per_second": 75.806,
|
| 371 |
-
"eval_steps_per_second": 4.75,
|
| 372 |
-
"step": 533
|
| 373 |
},
|
| 374 |
{
|
| 375 |
-
"epoch":
|
| 376 |
-
"learning_rate": 3.
|
| 377 |
-
"loss": 0.
|
| 378 |
"step": 540
|
| 379 |
},
|
| 380 |
{
|
| 381 |
-
"epoch":
|
| 382 |
-
"learning_rate": 3.
|
| 383 |
-
"loss": 0.
|
| 384 |
"step": 550
|
| 385 |
},
|
| 386 |
{
|
| 387 |
-
"epoch":
|
| 388 |
-
"learning_rate": 3.
|
| 389 |
-
"loss": 0.
|
| 390 |
"step": 560
|
| 391 |
},
|
| 392 |
{
|
| 393 |
-
"epoch":
|
| 394 |
-
"learning_rate": 3.
|
| 395 |
-
"loss": 0.
|
| 396 |
"step": 570
|
| 397 |
},
|
| 398 |
{
|
| 399 |
-
"epoch":
|
| 400 |
-
"learning_rate": 3.
|
| 401 |
-
"loss": 0.
|
| 402 |
"step": 580
|
| 403 |
},
|
| 404 |
{
|
| 405 |
-
"epoch":
|
| 406 |
-
"learning_rate": 3.
|
| 407 |
-
"loss": 0.
|
| 408 |
"step": 590
|
| 409 |
},
|
| 410 |
{
|
| 411 |
-
"epoch":
|
| 412 |
-
"learning_rate": 3.
|
| 413 |
-
"loss": 0.
|
| 414 |
"step": 600
|
| 415 |
},
|
| 416 |
{
|
| 417 |
-
"epoch":
|
| 418 |
-
"learning_rate":
|
| 419 |
-
"loss": 0.
|
| 420 |
"step": 610
|
| 421 |
},
|
| 422 |
{
|
| 423 |
-
"epoch":
|
| 424 |
-
"
|
| 425 |
-
"
|
| 426 |
-
"
|
|
|
|
|
|
|
|
|
|
| 427 |
},
|
| 428 |
{
|
| 429 |
-
"epoch":
|
| 430 |
-
"learning_rate":
|
| 431 |
-
"loss": 0.
|
| 432 |
-
"step":
|
| 433 |
},
|
| 434 |
{
|
| 435 |
-
"epoch":
|
| 436 |
-
"learning_rate":
|
| 437 |
-
"loss": 0.
|
| 438 |
-
"step":
|
| 439 |
},
|
| 440 |
{
|
| 441 |
-
"epoch":
|
| 442 |
-
"
|
| 443 |
-
"
|
| 444 |
-
"eval_runtime": 30.1158,
|
| 445 |
-
"eval_samples_per_second": 75.243,
|
| 446 |
-
"eval_steps_per_second": 4.715,
|
| 447 |
"step": 640
|
| 448 |
},
|
| 449 |
{
|
| 450 |
-
"epoch":
|
| 451 |
-
"learning_rate":
|
| 452 |
-
"loss": 0.
|
| 453 |
"step": 650
|
| 454 |
},
|
| 455 |
{
|
| 456 |
-
"epoch":
|
| 457 |
-
"learning_rate":
|
| 458 |
-
"loss": 0.
|
| 459 |
"step": 660
|
| 460 |
},
|
| 461 |
{
|
| 462 |
-
"epoch":
|
| 463 |
-
"learning_rate":
|
| 464 |
-
"loss": 0.
|
| 465 |
"step": 670
|
| 466 |
},
|
| 467 |
{
|
| 468 |
-
"epoch":
|
| 469 |
-
"learning_rate":
|
| 470 |
-
"loss": 0.
|
| 471 |
"step": 680
|
| 472 |
},
|
| 473 |
{
|
| 474 |
-
"epoch":
|
| 475 |
-
"learning_rate":
|
| 476 |
-
"loss": 0.
|
| 477 |
"step": 690
|
| 478 |
},
|
| 479 |
{
|
| 480 |
-
"epoch":
|
| 481 |
-
"learning_rate":
|
| 482 |
-
"loss": 0.
|
| 483 |
"step": 700
|
| 484 |
},
|
| 485 |
{
|
| 486 |
-
"epoch":
|
| 487 |
-
"
|
| 488 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 489 |
"step": 710
|
| 490 |
},
|
| 491 |
{
|
| 492 |
-
"epoch":
|
| 493 |
-
"learning_rate":
|
| 494 |
-
"loss": 0.
|
| 495 |
"step": 720
|
| 496 |
},
|
| 497 |
{
|
| 498 |
-
"epoch":
|
| 499 |
-
"learning_rate":
|
| 500 |
-
"loss": 0.
|
| 501 |
"step": 730
|
| 502 |
},
|
| 503 |
{
|
| 504 |
-
"epoch":
|
| 505 |
-
"learning_rate": 2.
|
| 506 |
-
"loss": 0.
|
| 507 |
"step": 740
|
| 508 |
},
|
| 509 |
{
|
| 510 |
-
"epoch":
|
| 511 |
-
"
|
| 512 |
-
"
|
| 513 |
-
"eval_runtime": 29.6593,
|
| 514 |
-
"eval_samples_per_second": 76.401,
|
| 515 |
-
"eval_steps_per_second": 4.788,
|
| 516 |
-
"step": 747
|
| 517 |
-
},
|
| 518 |
-
{
|
| 519 |
-
"epoch": 7.03,
|
| 520 |
-
"learning_rate": 2.935010482180294e-05,
|
| 521 |
-
"loss": 0.373,
|
| 522 |
"step": 750
|
| 523 |
},
|
| 524 |
{
|
| 525 |
-
"epoch":
|
| 526 |
-
"learning_rate": 2.
|
| 527 |
-
"loss": 0.
|
| 528 |
"step": 760
|
| 529 |
},
|
| 530 |
{
|
| 531 |
-
"epoch":
|
| 532 |
-
"learning_rate": 2.
|
| 533 |
-
"loss": 0.
|
| 534 |
"step": 770
|
| 535 |
},
|
| 536 |
{
|
| 537 |
-
"epoch":
|
| 538 |
-
"learning_rate": 2.
|
| 539 |
-
"loss": 0.
|
| 540 |
"step": 780
|
| 541 |
},
|
| 542 |
{
|
| 543 |
-
"epoch":
|
| 544 |
-
"learning_rate": 2.
|
| 545 |
-
"loss": 0.
|
| 546 |
"step": 790
|
| 547 |
},
|
| 548 |
{
|
| 549 |
-
"epoch":
|
| 550 |
-
"
|
| 551 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 552 |
"step": 800
|
| 553 |
},
|
| 554 |
{
|
| 555 |
-
"epoch":
|
| 556 |
-
"learning_rate": 2.
|
| 557 |
-
"loss": 0.
|
| 558 |
"step": 810
|
| 559 |
},
|
| 560 |
{
|
| 561 |
-
"epoch":
|
| 562 |
-
"learning_rate": 2.
|
| 563 |
-
"loss": 0.
|
| 564 |
"step": 820
|
| 565 |
},
|
| 566 |
{
|
| 567 |
-
"epoch":
|
| 568 |
-
"learning_rate": 2.
|
| 569 |
-
"loss": 0.
|
| 570 |
"step": 830
|
| 571 |
},
|
| 572 |
{
|
| 573 |
-
"epoch":
|
| 574 |
-
"learning_rate": 2.
|
| 575 |
-
"loss": 0.
|
| 576 |
"step": 840
|
| 577 |
},
|
| 578 |
{
|
| 579 |
-
"epoch":
|
| 580 |
-
"learning_rate":
|
| 581 |
-
"loss": 0.
|
| 582 |
"step": 850
|
| 583 |
},
|
| 584 |
{
|
| 585 |
-
"epoch":
|
| 586 |
-
"
|
| 587 |
-
"
|
| 588 |
-
"eval_runtime": 29.7498,
|
| 589 |
-
"eval_samples_per_second": 76.169,
|
| 590 |
-
"eval_steps_per_second": 4.773,
|
| 591 |
-
"step": 854
|
| 592 |
-
},
|
| 593 |
-
{
|
| 594 |
-
"epoch": 8.06,
|
| 595 |
-
"learning_rate": 2.5506638714185887e-05,
|
| 596 |
-
"loss": 0.2842,
|
| 597 |
"step": 860
|
| 598 |
},
|
| 599 |
{
|
| 600 |
-
"epoch":
|
| 601 |
-
"learning_rate":
|
| 602 |
-
"loss": 0.
|
| 603 |
"step": 870
|
| 604 |
},
|
| 605 |
{
|
| 606 |
-
"epoch":
|
| 607 |
-
"learning_rate":
|
| 608 |
-
"loss": 0.
|
| 609 |
"step": 880
|
| 610 |
},
|
| 611 |
{
|
| 612 |
-
"epoch":
|
| 613 |
-
"
|
| 614 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 615 |
"step": 890
|
| 616 |
},
|
| 617 |
{
|
| 618 |
-
"epoch":
|
| 619 |
-
"learning_rate":
|
| 620 |
-
"loss": 0.
|
| 621 |
"step": 900
|
| 622 |
},
|
| 623 |
{
|
| 624 |
-
"epoch":
|
| 625 |
-
"learning_rate":
|
| 626 |
-
"loss": 0.
|
| 627 |
"step": 910
|
| 628 |
},
|
| 629 |
{
|
| 630 |
-
"epoch":
|
| 631 |
-
"learning_rate":
|
| 632 |
-
"loss": 0.
|
| 633 |
"step": 920
|
| 634 |
},
|
| 635 |
{
|
| 636 |
-
"epoch":
|
| 637 |
-
"learning_rate":
|
| 638 |
-
"loss": 0.
|
| 639 |
"step": 930
|
| 640 |
},
|
| 641 |
{
|
| 642 |
-
"epoch":
|
| 643 |
-
"learning_rate":
|
| 644 |
-
"loss": 0.
|
| 645 |
"step": 940
|
| 646 |
},
|
| 647 |
{
|
| 648 |
-
"epoch":
|
| 649 |
-
"learning_rate":
|
| 650 |
-
"loss": 0.
|
| 651 |
"step": 950
|
| 652 |
},
|
| 653 |
{
|
| 654 |
-
"epoch":
|
| 655 |
-
"learning_rate":
|
| 656 |
-
"loss": 0.
|
| 657 |
"step": 960
|
| 658 |
},
|
| 659 |
{
|
| 660 |
-
"epoch":
|
| 661 |
-
"
|
| 662 |
-
"
|
| 663 |
-
"
|
| 664 |
-
"eval_samples_per_second": 74.719,
|
| 665 |
-
"eval_steps_per_second": 4.682,
|
| 666 |
-
"step": 960
|
| 667 |
},
|
| 668 |
{
|
| 669 |
-
"epoch":
|
| 670 |
-
"
|
| 671 |
-
"
|
|
|
|
|
|
|
|
|
|
| 672 |
"step": 970
|
| 673 |
},
|
| 674 |
{
|
| 675 |
-
"epoch":
|
| 676 |
-
"learning_rate":
|
| 677 |
-
"loss": 0.
|
| 678 |
"step": 980
|
| 679 |
},
|
| 680 |
{
|
| 681 |
-
"epoch":
|
| 682 |
-
"learning_rate":
|
| 683 |
-
"loss": 0.
|
| 684 |
"step": 990
|
| 685 |
},
|
| 686 |
{
|
| 687 |
-
"epoch":
|
| 688 |
-
"learning_rate":
|
| 689 |
-
"loss": 0.
|
| 690 |
"step": 1000
|
| 691 |
},
|
| 692 |
{
|
| 693 |
-
"epoch":
|
| 694 |
-
"learning_rate":
|
| 695 |
-
"loss": 0.
|
| 696 |
"step": 1010
|
| 697 |
},
|
| 698 |
{
|
| 699 |
-
"epoch":
|
| 700 |
-
"learning_rate": 1.
|
| 701 |
-
"loss": 0.
|
| 702 |
"step": 1020
|
| 703 |
},
|
| 704 |
{
|
| 705 |
-
"epoch":
|
| 706 |
-
"learning_rate": 1.
|
| 707 |
-
"loss": 0.
|
| 708 |
"step": 1030
|
| 709 |
},
|
| 710 |
{
|
| 711 |
-
"epoch":
|
| 712 |
-
"learning_rate": 1.
|
| 713 |
-
"loss": 0.
|
| 714 |
"step": 1040
|
| 715 |
},
|
| 716 |
{
|
| 717 |
-
"epoch":
|
| 718 |
-
"learning_rate": 1.
|
| 719 |
-
"loss": 0.
|
| 720 |
"step": 1050
|
| 721 |
},
|
| 722 |
{
|
| 723 |
-
"epoch":
|
| 724 |
-
"
|
| 725 |
-
"
|
| 726 |
-
"
|
|
|
|
|
|
|
|
|
|
| 727 |
},
|
| 728 |
{
|
| 729 |
-
"epoch":
|
| 730 |
-
"
|
| 731 |
-
"
|
| 732 |
-
"
|
| 733 |
-
"eval_samples_per_second": 76.4,
|
| 734 |
-
"eval_steps_per_second": 4.788,
|
| 735 |
-
"step": 1067
|
| 736 |
},
|
| 737 |
{
|
| 738 |
-
"epoch":
|
| 739 |
-
"learning_rate": 1.
|
| 740 |
-
"loss": 0.
|
| 741 |
"step": 1070
|
| 742 |
},
|
| 743 |
{
|
| 744 |
-
"epoch":
|
| 745 |
-
"learning_rate": 1.
|
| 746 |
-
"loss": 0.
|
| 747 |
"step": 1080
|
| 748 |
},
|
| 749 |
{
|
| 750 |
-
"epoch":
|
| 751 |
-
"learning_rate":
|
| 752 |
-
"loss": 0.
|
| 753 |
"step": 1090
|
| 754 |
},
|
| 755 |
{
|
| 756 |
-
"epoch":
|
| 757 |
-
"learning_rate":
|
| 758 |
-
"loss": 0.
|
| 759 |
"step": 1100
|
| 760 |
},
|
| 761 |
{
|
| 762 |
-
"epoch":
|
| 763 |
-
"learning_rate":
|
| 764 |
-
"loss": 0.
|
| 765 |
"step": 1110
|
| 766 |
},
|
| 767 |
{
|
| 768 |
-
"epoch":
|
| 769 |
-
"learning_rate":
|
| 770 |
-
"loss": 0.
|
| 771 |
"step": 1120
|
| 772 |
},
|
| 773 |
{
|
| 774 |
-
"epoch":
|
| 775 |
-
"learning_rate":
|
| 776 |
-
"loss": 0.
|
| 777 |
"step": 1130
|
| 778 |
},
|
| 779 |
{
|
| 780 |
-
"epoch":
|
| 781 |
-
"learning_rate":
|
| 782 |
-
"loss": 0.
|
| 783 |
"step": 1140
|
| 784 |
},
|
| 785 |
{
|
| 786 |
-
"epoch":
|
| 787 |
-
"
|
| 788 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 789 |
"step": 1150
|
| 790 |
},
|
| 791 |
{
|
| 792 |
-
"epoch":
|
| 793 |
-
"learning_rate":
|
| 794 |
-
"loss": 0.
|
| 795 |
"step": 1160
|
| 796 |
},
|
| 797 |
{
|
| 798 |
-
"epoch":
|
| 799 |
-
"learning_rate":
|
| 800 |
-
"loss": 0.
|
| 801 |
"step": 1170
|
| 802 |
},
|
| 803 |
{
|
| 804 |
-
"epoch":
|
| 805 |
-
"
|
| 806 |
-
"
|
| 807 |
-
"eval_runtime": 29.5637,
|
| 808 |
-
"eval_samples_per_second": 76.648,
|
| 809 |
-
"eval_steps_per_second": 4.803,
|
| 810 |
-
"step": 1174
|
| 811 |
-
},
|
| 812 |
-
{
|
| 813 |
-
"epoch": 11.05,
|
| 814 |
-
"learning_rate": 1.4325646401118102e-05,
|
| 815 |
-
"loss": 0.1944,
|
| 816 |
"step": 1180
|
| 817 |
},
|
| 818 |
{
|
| 819 |
-
"epoch":
|
| 820 |
-
"learning_rate":
|
| 821 |
-
"loss": 0.
|
| 822 |
"step": 1190
|
| 823 |
},
|
| 824 |
{
|
| 825 |
-
"epoch":
|
| 826 |
-
"learning_rate":
|
| 827 |
-
"loss": 0.
|
| 828 |
"step": 1200
|
| 829 |
},
|
| 830 |
{
|
| 831 |
-
"epoch":
|
| 832 |
-
"learning_rate":
|
| 833 |
-
"loss": 0.
|
| 834 |
"step": 1210
|
| 835 |
},
|
| 836 |
{
|
| 837 |
-
"epoch":
|
| 838 |
-
"learning_rate":
|
| 839 |
-
"loss": 0.
|
| 840 |
"step": 1220
|
| 841 |
},
|
| 842 |
{
|
| 843 |
-
"epoch":
|
| 844 |
-
"learning_rate":
|
| 845 |
-
"loss": 0.
|
| 846 |
"step": 1230
|
| 847 |
},
|
| 848 |
{
|
| 849 |
-
"epoch":
|
| 850 |
-
"
|
| 851 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 852 |
"step": 1240
|
| 853 |
},
|
| 854 |
{
|
| 855 |
-
"epoch":
|
| 856 |
-
"learning_rate":
|
| 857 |
-
"loss": 0.
|
| 858 |
"step": 1250
|
| 859 |
},
|
| 860 |
{
|
| 861 |
-
"epoch":
|
| 862 |
-
"learning_rate":
|
| 863 |
-
"loss": 0.
|
| 864 |
"step": 1260
|
| 865 |
},
|
| 866 |
{
|
| 867 |
-
"epoch":
|
| 868 |
-
"learning_rate":
|
| 869 |
-
"loss": 0.
|
| 870 |
"step": 1270
|
| 871 |
},
|
| 872 |
{
|
| 873 |
-
"epoch":
|
| 874 |
-
"learning_rate": 1.
|
| 875 |
-
"loss": 0.
|
| 876 |
"step": 1280
|
| 877 |
},
|
| 878 |
{
|
| 879 |
-
"epoch":
|
| 880 |
-
"
|
| 881 |
-
"
|
| 882 |
-
"eval_runtime": 29.7573,
|
| 883 |
-
"eval_samples_per_second": 76.149,
|
| 884 |
-
"eval_steps_per_second": 4.772,
|
| 885 |
-
"step": 1281
|
| 886 |
-
},
|
| 887 |
-
{
|
| 888 |
-
"epoch": 12.08,
|
| 889 |
-
"learning_rate": 1.0482180293501048e-05,
|
| 890 |
-
"loss": 0.1265,
|
| 891 |
"step": 1290
|
| 892 |
},
|
| 893 |
{
|
| 894 |
-
"epoch":
|
| 895 |
-
"learning_rate":
|
| 896 |
-
"loss": 0.
|
| 897 |
"step": 1300
|
| 898 |
},
|
| 899 |
{
|
| 900 |
-
"epoch":
|
| 901 |
-
"learning_rate":
|
| 902 |
-
"loss": 0.
|
| 903 |
"step": 1310
|
| 904 |
},
|
| 905 |
{
|
| 906 |
-
"epoch":
|
| 907 |
-
"learning_rate":
|
| 908 |
-
"loss": 0.
|
| 909 |
"step": 1320
|
| 910 |
},
|
| 911 |
{
|
| 912 |
-
"epoch":
|
| 913 |
-
"
|
| 914 |
-
"
|
| 915 |
-
"
|
| 916 |
-
|
| 917 |
-
|
| 918 |
-
"
|
| 919 |
-
"learning_rate": 8.735150244584208e-06,
|
| 920 |
-
"loss": 0.1563,
|
| 921 |
-
"step": 1340
|
| 922 |
-
},
|
| 923 |
-
{
|
| 924 |
-
"epoch": 12.65,
|
| 925 |
-
"learning_rate": 8.38574423480084e-06,
|
| 926 |
-
"loss": 0.1583,
|
| 927 |
-
"step": 1350
|
| 928 |
-
},
|
| 929 |
-
{
|
| 930 |
-
"epoch": 12.74,
|
| 931 |
-
"learning_rate": 8.03633822501747e-06,
|
| 932 |
-
"loss": 0.1185,
|
| 933 |
-
"step": 1360
|
| 934 |
-
},
|
| 935 |
-
{
|
| 936 |
-
"epoch": 12.83,
|
| 937 |
-
"learning_rate": 7.686932215234103e-06,
|
| 938 |
-
"loss": 0.1229,
|
| 939 |
-
"step": 1370
|
| 940 |
-
},
|
| 941 |
-
{
|
| 942 |
-
"epoch": 12.93,
|
| 943 |
-
"learning_rate": 7.337526205450735e-06,
|
| 944 |
-
"loss": 0.1551,
|
| 945 |
-
"step": 1380
|
| 946 |
-
},
|
| 947 |
-
{
|
| 948 |
-
"epoch": 12.99,
|
| 949 |
-
"eval_accuracy": 0.8375992939099736,
|
| 950 |
-
"eval_loss": 0.5339873433113098,
|
| 951 |
-
"eval_runtime": 29.8516,
|
| 952 |
-
"eval_samples_per_second": 75.909,
|
| 953 |
-
"eval_steps_per_second": 4.757,
|
| 954 |
-
"step": 1387
|
| 955 |
-
},
|
| 956 |
-
{
|
| 957 |
-
"epoch": 13.02,
|
| 958 |
-
"learning_rate": 6.988120195667367e-06,
|
| 959 |
-
"loss": 0.1566,
|
| 960 |
-
"step": 1390
|
| 961 |
-
},
|
| 962 |
-
{
|
| 963 |
-
"epoch": 13.11,
|
| 964 |
-
"learning_rate": 6.638714185883997e-06,
|
| 965 |
-
"loss": 0.1394,
|
| 966 |
-
"step": 1400
|
| 967 |
-
},
|
| 968 |
-
{
|
| 969 |
-
"epoch": 13.21,
|
| 970 |
-
"learning_rate": 6.289308176100629e-06,
|
| 971 |
-
"loss": 0.1642,
|
| 972 |
-
"step": 1410
|
| 973 |
-
},
|
| 974 |
-
{
|
| 975 |
-
"epoch": 13.3,
|
| 976 |
-
"learning_rate": 5.9399021663172605e-06,
|
| 977 |
-
"loss": 0.1334,
|
| 978 |
-
"step": 1420
|
| 979 |
-
},
|
| 980 |
-
{
|
| 981 |
-
"epoch": 13.4,
|
| 982 |
-
"learning_rate": 5.590496156533893e-06,
|
| 983 |
-
"loss": 0.157,
|
| 984 |
-
"step": 1430
|
| 985 |
-
},
|
| 986 |
-
{
|
| 987 |
-
"epoch": 13.49,
|
| 988 |
-
"learning_rate": 5.241090146750524e-06,
|
| 989 |
-
"loss": 0.1658,
|
| 990 |
-
"step": 1440
|
| 991 |
-
},
|
| 992 |
-
{
|
| 993 |
-
"epoch": 13.58,
|
| 994 |
-
"learning_rate": 4.891684136967156e-06,
|
| 995 |
-
"loss": 0.1692,
|
| 996 |
-
"step": 1450
|
| 997 |
-
},
|
| 998 |
-
{
|
| 999 |
-
"epoch": 13.68,
|
| 1000 |
-
"learning_rate": 4.542278127183788e-06,
|
| 1001 |
-
"loss": 0.1393,
|
| 1002 |
-
"step": 1460
|
| 1003 |
-
},
|
| 1004 |
-
{
|
| 1005 |
-
"epoch": 13.77,
|
| 1006 |
-
"learning_rate": 4.19287211740042e-06,
|
| 1007 |
-
"loss": 0.1492,
|
| 1008 |
-
"step": 1470
|
| 1009 |
-
},
|
| 1010 |
-
{
|
| 1011 |
-
"epoch": 13.86,
|
| 1012 |
-
"learning_rate": 3.843466107617051e-06,
|
| 1013 |
-
"loss": 0.1503,
|
| 1014 |
-
"step": 1480
|
| 1015 |
-
},
|
| 1016 |
-
{
|
| 1017 |
-
"epoch": 13.96,
|
| 1018 |
-
"learning_rate": 3.4940600978336834e-06,
|
| 1019 |
-
"loss": 0.1548,
|
| 1020 |
-
"step": 1490
|
| 1021 |
-
},
|
| 1022 |
-
{
|
| 1023 |
-
"epoch": 14.0,
|
| 1024 |
-
"eval_accuracy": 0.8367166813768756,
|
| 1025 |
-
"eval_loss": 0.5312865376472473,
|
| 1026 |
-
"eval_runtime": 29.7732,
|
| 1027 |
-
"eval_samples_per_second": 76.109,
|
| 1028 |
-
"eval_steps_per_second": 4.769,
|
| 1029 |
-
"step": 1494
|
| 1030 |
-
},
|
| 1031 |
-
{
|
| 1032 |
-
"epoch": 14.05,
|
| 1033 |
-
"learning_rate": 3.1446540880503146e-06,
|
| 1034 |
-
"loss": 0.1186,
|
| 1035 |
-
"step": 1500
|
| 1036 |
-
},
|
| 1037 |
-
{
|
| 1038 |
-
"epoch": 14.15,
|
| 1039 |
-
"learning_rate": 2.7952480782669463e-06,
|
| 1040 |
-
"loss": 0.1213,
|
| 1041 |
-
"step": 1510
|
| 1042 |
-
},
|
| 1043 |
-
{
|
| 1044 |
-
"epoch": 14.24,
|
| 1045 |
-
"learning_rate": 2.445842068483578e-06,
|
| 1046 |
-
"loss": 0.1601,
|
| 1047 |
-
"step": 1520
|
| 1048 |
-
},
|
| 1049 |
-
{
|
| 1050 |
-
"epoch": 14.33,
|
| 1051 |
-
"learning_rate": 2.09643605870021e-06,
|
| 1052 |
-
"loss": 0.1916,
|
| 1053 |
-
"step": 1530
|
| 1054 |
-
},
|
| 1055 |
-
{
|
| 1056 |
-
"epoch": 14.43,
|
| 1057 |
-
"learning_rate": 1.7470300489168417e-06,
|
| 1058 |
-
"loss": 0.1243,
|
| 1059 |
-
"step": 1540
|
| 1060 |
-
},
|
| 1061 |
-
{
|
| 1062 |
-
"epoch": 14.52,
|
| 1063 |
-
"learning_rate": 1.3976240391334732e-06,
|
| 1064 |
-
"loss": 0.1323,
|
| 1065 |
-
"step": 1550
|
| 1066 |
-
},
|
| 1067 |
-
{
|
| 1068 |
-
"epoch": 14.61,
|
| 1069 |
-
"learning_rate": 1.048218029350105e-06,
|
| 1070 |
-
"loss": 0.1535,
|
| 1071 |
-
"step": 1560
|
| 1072 |
-
},
|
| 1073 |
-
{
|
| 1074 |
-
"epoch": 14.71,
|
| 1075 |
-
"learning_rate": 6.988120195667366e-07,
|
| 1076 |
-
"loss": 0.1385,
|
| 1077 |
-
"step": 1570
|
| 1078 |
-
},
|
| 1079 |
-
{
|
| 1080 |
-
"epoch": 14.8,
|
| 1081 |
-
"learning_rate": 3.494060097833683e-07,
|
| 1082 |
-
"loss": 0.1466,
|
| 1083 |
-
"step": 1580
|
| 1084 |
},
|
| 1085 |
{
|
| 1086 |
-
"epoch": 14.
|
| 1087 |
-
"
|
| 1088 |
-
"
|
| 1089 |
-
"
|
| 1090 |
-
|
| 1091 |
-
|
| 1092 |
-
"
|
| 1093 |
-
"eval_accuracy": 0.8398058252427184,
|
| 1094 |
-
"eval_loss": 0.5298482775688171,
|
| 1095 |
-
"eval_runtime": 29.7814,
|
| 1096 |
-
"eval_samples_per_second": 76.088,
|
| 1097 |
-
"eval_steps_per_second": 4.768,
|
| 1098 |
-
"step": 1590
|
| 1099 |
-
},
|
| 1100 |
-
{
|
| 1101 |
-
"epoch": 14.89,
|
| 1102 |
-
"step": 1590,
|
| 1103 |
-
"total_flos": 7.877470546062803e+18,
|
| 1104 |
-
"train_loss": 0.5140632056590146,
|
| 1105 |
-
"train_runtime": 3982.2883,
|
| 1106 |
-
"train_samples_per_second": 25.704,
|
| 1107 |
-
"train_steps_per_second": 0.399
|
| 1108 |
}
|
| 1109 |
],
|
| 1110 |
"logging_steps": 10,
|
| 1111 |
-
"max_steps":
|
| 1112 |
"num_train_epochs": 15,
|
| 1113 |
"save_steps": 500,
|
| 1114 |
-
"total_flos":
|
| 1115 |
"trial_name": null,
|
| 1116 |
"trial_params": null
|
| 1117 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_metric": 0.8528925619834711,
|
| 3 |
+
"best_model_checkpoint": "vit-base-patch16-224-brand/checkpoint-970",
|
| 4 |
+
"epoch": 14.957507082152974,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 1320,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
| 10 |
"log_history": [
|
| 11 |
{
|
| 12 |
+
"epoch": 0.11,
|
| 13 |
+
"learning_rate": 3.7878787878787882e-06,
|
| 14 |
+
"loss": 2.2055,
|
| 15 |
"step": 10
|
| 16 |
},
|
| 17 |
{
|
| 18 |
+
"epoch": 0.23,
|
| 19 |
+
"learning_rate": 7.5757575757575764e-06,
|
| 20 |
+
"loss": 2.162,
|
| 21 |
"step": 20
|
| 22 |
},
|
| 23 |
{
|
| 24 |
+
"epoch": 0.34,
|
| 25 |
+
"learning_rate": 1.1363636363636365e-05,
|
| 26 |
+
"loss": 2.0983,
|
| 27 |
"step": 30
|
| 28 |
},
|
| 29 |
{
|
| 30 |
+
"epoch": 0.45,
|
| 31 |
+
"learning_rate": 1.5151515151515153e-05,
|
| 32 |
+
"loss": 2.0166,
|
| 33 |
"step": 40
|
| 34 |
},
|
| 35 |
{
|
| 36 |
+
"epoch": 0.57,
|
| 37 |
+
"learning_rate": 1.893939393939394e-05,
|
| 38 |
+
"loss": 1.9044,
|
| 39 |
"step": 50
|
| 40 |
},
|
| 41 |
{
|
| 42 |
+
"epoch": 0.68,
|
| 43 |
+
"learning_rate": 2.272727272727273e-05,
|
| 44 |
+
"loss": 1.7381,
|
| 45 |
"step": 60
|
| 46 |
},
|
| 47 |
{
|
| 48 |
+
"epoch": 0.79,
|
| 49 |
+
"learning_rate": 2.6515151515151516e-05,
|
| 50 |
+
"loss": 1.6089,
|
| 51 |
"step": 70
|
| 52 |
},
|
| 53 |
{
|
| 54 |
+
"epoch": 0.91,
|
| 55 |
+
"learning_rate": 3.0303030303030306e-05,
|
| 56 |
+
"loss": 1.4669,
|
| 57 |
"step": 80
|
| 58 |
},
|
| 59 |
{
|
| 60 |
+
"epoch": 1.0,
|
| 61 |
+
"eval_accuracy": 0.5611570247933885,
|
| 62 |
+
"eval_loss": 1.3067070245742798,
|
| 63 |
+
"eval_runtime": 16.7756,
|
| 64 |
+
"eval_samples_per_second": 72.129,
|
| 65 |
+
"eval_steps_per_second": 4.53,
|
| 66 |
+
"step": 88
|
| 67 |
},
|
| 68 |
{
|
| 69 |
+
"epoch": 1.02,
|
| 70 |
+
"learning_rate": 3.409090909090909e-05,
|
| 71 |
+
"loss": 1.4484,
|
| 72 |
+
"step": 90
|
| 73 |
},
|
| 74 |
{
|
| 75 |
+
"epoch": 1.13,
|
| 76 |
+
"learning_rate": 3.787878787878788e-05,
|
| 77 |
+
"loss": 1.2513,
|
| 78 |
+
"step": 100
|
|
|
|
|
|
|
|
|
|
| 79 |
},
|
| 80 |
{
|
| 81 |
+
"epoch": 1.25,
|
| 82 |
+
"learning_rate": 4.166666666666667e-05,
|
| 83 |
+
"loss": 1.1741,
|
| 84 |
"step": 110
|
| 85 |
},
|
| 86 |
{
|
| 87 |
+
"epoch": 1.36,
|
| 88 |
+
"learning_rate": 4.545454545454546e-05,
|
| 89 |
+
"loss": 1.0883,
|
| 90 |
"step": 120
|
| 91 |
},
|
| 92 |
{
|
| 93 |
+
"epoch": 1.47,
|
| 94 |
+
"learning_rate": 4.9242424242424245e-05,
|
| 95 |
+
"loss": 1.0522,
|
| 96 |
"step": 130
|
| 97 |
},
|
| 98 |
{
|
| 99 |
+
"epoch": 1.59,
|
| 100 |
+
"learning_rate": 4.966329966329967e-05,
|
| 101 |
+
"loss": 1.0214,
|
| 102 |
"step": 140
|
| 103 |
},
|
| 104 |
{
|
| 105 |
+
"epoch": 1.7,
|
| 106 |
+
"learning_rate": 4.9242424242424245e-05,
|
| 107 |
+
"loss": 0.9231,
|
| 108 |
"step": 150
|
| 109 |
},
|
| 110 |
{
|
| 111 |
+
"epoch": 1.81,
|
| 112 |
+
"learning_rate": 4.882154882154882e-05,
|
| 113 |
+
"loss": 0.8961,
|
| 114 |
"step": 160
|
| 115 |
},
|
| 116 |
{
|
| 117 |
+
"epoch": 1.93,
|
| 118 |
+
"learning_rate": 4.84006734006734e-05,
|
| 119 |
+
"loss": 0.8898,
|
| 120 |
"step": 170
|
| 121 |
},
|
| 122 |
{
|
| 123 |
+
"epoch": 1.99,
|
| 124 |
+
"eval_accuracy": 0.7140495867768595,
|
| 125 |
+
"eval_loss": 0.8380156755447388,
|
| 126 |
+
"eval_runtime": 16.8759,
|
| 127 |
+
"eval_samples_per_second": 71.7,
|
| 128 |
+
"eval_steps_per_second": 4.503,
|
| 129 |
+
"step": 176
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 2.04,
|
| 133 |
+
"learning_rate": 4.797979797979798e-05,
|
| 134 |
+
"loss": 0.8856,
|
| 135 |
"step": 180
|
| 136 |
},
|
| 137 |
{
|
| 138 |
+
"epoch": 2.15,
|
| 139 |
+
"learning_rate": 4.755892255892256e-05,
|
| 140 |
+
"loss": 0.7916,
|
| 141 |
"step": 190
|
| 142 |
},
|
| 143 |
{
|
| 144 |
+
"epoch": 2.27,
|
| 145 |
+
"learning_rate": 4.713804713804714e-05,
|
| 146 |
+
"loss": 0.7614,
|
| 147 |
"step": 200
|
| 148 |
},
|
| 149 |
{
|
| 150 |
+
"epoch": 2.38,
|
| 151 |
+
"learning_rate": 4.671717171717172e-05,
|
| 152 |
+
"loss": 0.7696,
|
| 153 |
"step": 210
|
| 154 |
},
|
| 155 |
{
|
| 156 |
+
"epoch": 2.49,
|
| 157 |
+
"learning_rate": 4.62962962962963e-05,
|
| 158 |
+
"loss": 0.7874,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 159 |
"step": 220
|
| 160 |
},
|
| 161 |
{
|
| 162 |
+
"epoch": 2.61,
|
| 163 |
+
"learning_rate": 4.5875420875420876e-05,
|
| 164 |
+
"loss": 0.7126,
|
| 165 |
"step": 230
|
| 166 |
},
|
| 167 |
{
|
| 168 |
+
"epoch": 2.72,
|
| 169 |
+
"learning_rate": 4.545454545454546e-05,
|
| 170 |
+
"loss": 0.7119,
|
| 171 |
"step": 240
|
| 172 |
},
|
| 173 |
{
|
| 174 |
+
"epoch": 2.83,
|
| 175 |
+
"learning_rate": 4.5033670033670034e-05,
|
| 176 |
+
"loss": 0.7095,
|
| 177 |
"step": 250
|
| 178 |
},
|
| 179 |
{
|
| 180 |
+
"epoch": 2.95,
|
| 181 |
+
"learning_rate": 4.4612794612794616e-05,
|
| 182 |
+
"loss": 0.7243,
|
| 183 |
"step": 260
|
| 184 |
},
|
| 185 |
{
|
| 186 |
+
"epoch": 2.99,
|
| 187 |
+
"eval_accuracy": 0.7694214876033058,
|
| 188 |
+
"eval_loss": 0.6559494137763977,
|
| 189 |
+
"eval_runtime": 16.9139,
|
| 190 |
+
"eval_samples_per_second": 71.539,
|
| 191 |
+
"eval_steps_per_second": 4.493,
|
| 192 |
+
"step": 264
|
| 193 |
+
},
|
| 194 |
+
{
|
| 195 |
+
"epoch": 3.06,
|
| 196 |
+
"learning_rate": 4.41919191919192e-05,
|
| 197 |
+
"loss": 0.7109,
|
| 198 |
"step": 270
|
| 199 |
},
|
| 200 |
{
|
| 201 |
+
"epoch": 3.17,
|
| 202 |
+
"learning_rate": 4.3771043771043774e-05,
|
| 203 |
+
"loss": 0.5651,
|
| 204 |
"step": 280
|
| 205 |
},
|
| 206 |
{
|
| 207 |
+
"epoch": 3.29,
|
| 208 |
+
"learning_rate": 4.335016835016835e-05,
|
| 209 |
+
"loss": 0.5838,
|
| 210 |
"step": 290
|
| 211 |
},
|
| 212 |
{
|
| 213 |
+
"epoch": 3.4,
|
| 214 |
+
"learning_rate": 4.292929292929293e-05,
|
| 215 |
+
"loss": 0.5879,
|
| 216 |
"step": 300
|
| 217 |
},
|
| 218 |
{
|
| 219 |
+
"epoch": 3.51,
|
| 220 |
+
"learning_rate": 4.250841750841751e-05,
|
| 221 |
+
"loss": 0.5541,
|
| 222 |
"step": 310
|
| 223 |
},
|
| 224 |
{
|
| 225 |
+
"epoch": 3.63,
|
| 226 |
+
"learning_rate": 4.208754208754209e-05,
|
| 227 |
+
"loss": 0.5522,
|
| 228 |
"step": 320
|
| 229 |
},
|
| 230 |
{
|
| 231 |
+
"epoch": 3.74,
|
| 232 |
+
"learning_rate": 4.166666666666667e-05,
|
| 233 |
+
"loss": 0.5105,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 234 |
"step": 330
|
| 235 |
},
|
| 236 |
{
|
| 237 |
+
"epoch": 3.85,
|
| 238 |
+
"learning_rate": 4.124579124579125e-05,
|
| 239 |
+
"loss": 0.5289,
|
| 240 |
"step": 340
|
| 241 |
},
|
| 242 |
{
|
| 243 |
+
"epoch": 3.97,
|
| 244 |
+
"learning_rate": 4.082491582491583e-05,
|
| 245 |
+
"loss": 0.5158,
|
| 246 |
"step": 350
|
| 247 |
},
|
| 248 |
{
|
| 249 |
+
"epoch": 4.0,
|
| 250 |
+
"eval_accuracy": 0.7950413223140496,
|
| 251 |
+
"eval_loss": 0.5982227921485901,
|
| 252 |
+
"eval_runtime": 16.8053,
|
| 253 |
+
"eval_samples_per_second": 72.001,
|
| 254 |
+
"eval_steps_per_second": 4.522,
|
| 255 |
+
"step": 353
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"epoch": 4.08,
|
| 259 |
+
"learning_rate": 4.0404040404040405e-05,
|
| 260 |
+
"loss": 0.4296,
|
| 261 |
"step": 360
|
| 262 |
},
|
| 263 |
{
|
| 264 |
+
"epoch": 4.19,
|
| 265 |
+
"learning_rate": 3.998316498316498e-05,
|
| 266 |
+
"loss": 0.4105,
|
| 267 |
"step": 370
|
| 268 |
},
|
| 269 |
{
|
| 270 |
+
"epoch": 4.31,
|
| 271 |
+
"learning_rate": 3.956228956228956e-05,
|
| 272 |
+
"loss": 0.4658,
|
| 273 |
"step": 380
|
| 274 |
},
|
| 275 |
{
|
| 276 |
+
"epoch": 4.42,
|
| 277 |
+
"learning_rate": 3.9141414141414145e-05,
|
| 278 |
+
"loss": 0.4464,
|
| 279 |
"step": 390
|
| 280 |
},
|
| 281 |
{
|
| 282 |
+
"epoch": 4.53,
|
| 283 |
+
"learning_rate": 3.872053872053872e-05,
|
| 284 |
+
"loss": 0.431,
|
| 285 |
"step": 400
|
| 286 |
},
|
| 287 |
{
|
| 288 |
+
"epoch": 4.65,
|
| 289 |
+
"learning_rate": 3.82996632996633e-05,
|
| 290 |
+
"loss": 0.4499,
|
| 291 |
"step": 410
|
| 292 |
},
|
| 293 |
{
|
| 294 |
+
"epoch": 4.76,
|
| 295 |
+
"learning_rate": 3.787878787878788e-05,
|
| 296 |
+
"loss": 0.4568,
|
| 297 |
"step": 420
|
| 298 |
},
|
| 299 |
{
|
| 300 |
+
"epoch": 4.87,
|
| 301 |
+
"learning_rate": 3.745791245791246e-05,
|
| 302 |
+
"loss": 0.4397,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 303 |
"step": 430
|
| 304 |
},
|
| 305 |
{
|
| 306 |
+
"epoch": 4.99,
|
| 307 |
+
"learning_rate": 3.7037037037037037e-05,
|
| 308 |
+
"loss": 0.4605,
|
| 309 |
"step": 440
|
| 310 |
},
|
| 311 |
{
|
| 312 |
+
"epoch": 5.0,
|
| 313 |
+
"eval_accuracy": 0.8082644628099174,
|
| 314 |
+
"eval_loss": 0.5856056809425354,
|
| 315 |
+
"eval_runtime": 16.7221,
|
| 316 |
+
"eval_samples_per_second": 72.359,
|
| 317 |
+
"eval_steps_per_second": 4.545,
|
| 318 |
+
"step": 441
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 5.1,
|
| 322 |
+
"learning_rate": 3.661616161616162e-05,
|
| 323 |
+
"loss": 0.3757,
|
| 324 |
"step": 450
|
| 325 |
},
|
| 326 |
{
|
| 327 |
+
"epoch": 5.21,
|
| 328 |
+
"learning_rate": 3.61952861952862e-05,
|
| 329 |
+
"loss": 0.369,
|
| 330 |
"step": 460
|
| 331 |
},
|
| 332 |
{
|
| 333 |
+
"epoch": 5.33,
|
| 334 |
+
"learning_rate": 3.577441077441078e-05,
|
| 335 |
+
"loss": 0.3952,
|
| 336 |
"step": 470
|
| 337 |
},
|
| 338 |
{
|
| 339 |
+
"epoch": 5.44,
|
| 340 |
+
"learning_rate": 3.535353535353535e-05,
|
| 341 |
+
"loss": 0.3841,
|
| 342 |
"step": 480
|
| 343 |
},
|
| 344 |
{
|
| 345 |
+
"epoch": 5.55,
|
| 346 |
+
"learning_rate": 3.4932659932659934e-05,
|
| 347 |
+
"loss": 0.3707,
|
| 348 |
"step": 490
|
| 349 |
},
|
| 350 |
{
|
| 351 |
+
"epoch": 5.67,
|
| 352 |
+
"learning_rate": 3.451178451178451e-05,
|
| 353 |
+
"loss": 0.3874,
|
| 354 |
"step": 500
|
| 355 |
},
|
| 356 |
{
|
| 357 |
+
"epoch": 5.78,
|
| 358 |
+
"learning_rate": 3.409090909090909e-05,
|
| 359 |
+
"loss": 0.347,
|
| 360 |
"step": 510
|
| 361 |
},
|
| 362 |
{
|
| 363 |
+
"epoch": 5.89,
|
| 364 |
+
"learning_rate": 3.3670033670033675e-05,
|
| 365 |
+
"loss": 0.332,
|
| 366 |
"step": 520
|
| 367 |
},
|
| 368 |
{
|
| 369 |
+
"epoch": 5.99,
|
| 370 |
+
"eval_accuracy": 0.8355371900826446,
|
| 371 |
+
"eval_loss": 0.5138491988182068,
|
| 372 |
+
"eval_runtime": 16.6373,
|
| 373 |
+
"eval_samples_per_second": 72.728,
|
| 374 |
+
"eval_steps_per_second": 4.568,
|
| 375 |
+
"step": 529
|
| 376 |
},
|
| 377 |
{
|
| 378 |
+
"epoch": 6.01,
|
| 379 |
+
"learning_rate": 3.324915824915825e-05,
|
| 380 |
+
"loss": 0.3767,
|
| 381 |
+
"step": 530
|
|
|
|
|
|
|
|
|
|
| 382 |
},
|
| 383 |
{
|
| 384 |
+
"epoch": 6.12,
|
| 385 |
+
"learning_rate": 3.282828282828283e-05,
|
| 386 |
+
"loss": 0.2646,
|
| 387 |
"step": 540
|
| 388 |
},
|
| 389 |
{
|
| 390 |
+
"epoch": 6.23,
|
| 391 |
+
"learning_rate": 3.240740740740741e-05,
|
| 392 |
+
"loss": 0.3142,
|
| 393 |
"step": 550
|
| 394 |
},
|
| 395 |
{
|
| 396 |
+
"epoch": 6.35,
|
| 397 |
+
"learning_rate": 3.198653198653199e-05,
|
| 398 |
+
"loss": 0.3171,
|
| 399 |
"step": 560
|
| 400 |
},
|
| 401 |
{
|
| 402 |
+
"epoch": 6.46,
|
| 403 |
+
"learning_rate": 3.1565656565656566e-05,
|
| 404 |
+
"loss": 0.3318,
|
| 405 |
"step": 570
|
| 406 |
},
|
| 407 |
{
|
| 408 |
+
"epoch": 6.57,
|
| 409 |
+
"learning_rate": 3.114478114478115e-05,
|
| 410 |
+
"loss": 0.309,
|
| 411 |
"step": 580
|
| 412 |
},
|
| 413 |
{
|
| 414 |
+
"epoch": 6.69,
|
| 415 |
+
"learning_rate": 3.072390572390573e-05,
|
| 416 |
+
"loss": 0.2928,
|
| 417 |
"step": 590
|
| 418 |
},
|
| 419 |
{
|
| 420 |
+
"epoch": 6.8,
|
| 421 |
+
"learning_rate": 3.0303030303030306e-05,
|
| 422 |
+
"loss": 0.3622,
|
| 423 |
"step": 600
|
| 424 |
},
|
| 425 |
{
|
| 426 |
+
"epoch": 6.91,
|
| 427 |
+
"learning_rate": 2.9882154882154885e-05,
|
| 428 |
+
"loss": 0.3375,
|
| 429 |
"step": 610
|
| 430 |
},
|
| 431 |
{
|
| 432 |
+
"epoch": 6.99,
|
| 433 |
+
"eval_accuracy": 0.8264462809917356,
|
| 434 |
+
"eval_loss": 0.5094661116600037,
|
| 435 |
+
"eval_runtime": 16.7644,
|
| 436 |
+
"eval_samples_per_second": 72.177,
|
| 437 |
+
"eval_steps_per_second": 4.533,
|
| 438 |
+
"step": 617
|
| 439 |
},
|
| 440 |
{
|
| 441 |
+
"epoch": 7.03,
|
| 442 |
+
"learning_rate": 2.946127946127946e-05,
|
| 443 |
+
"loss": 0.2897,
|
| 444 |
+
"step": 620
|
| 445 |
},
|
| 446 |
{
|
| 447 |
+
"epoch": 7.14,
|
| 448 |
+
"learning_rate": 2.904040404040404e-05,
|
| 449 |
+
"loss": 0.2764,
|
| 450 |
+
"step": 630
|
| 451 |
},
|
| 452 |
{
|
| 453 |
+
"epoch": 7.25,
|
| 454 |
+
"learning_rate": 2.8619528619528618e-05,
|
| 455 |
+
"loss": 0.2627,
|
|
|
|
|
|
|
|
|
|
| 456 |
"step": 640
|
| 457 |
},
|
| 458 |
{
|
| 459 |
+
"epoch": 7.37,
|
| 460 |
+
"learning_rate": 2.8198653198653204e-05,
|
| 461 |
+
"loss": 0.2899,
|
| 462 |
"step": 650
|
| 463 |
},
|
| 464 |
{
|
| 465 |
+
"epoch": 7.48,
|
| 466 |
+
"learning_rate": 2.777777777777778e-05,
|
| 467 |
+
"loss": 0.2316,
|
| 468 |
"step": 660
|
| 469 |
},
|
| 470 |
{
|
| 471 |
+
"epoch": 7.59,
|
| 472 |
+
"learning_rate": 2.7356902356902358e-05,
|
| 473 |
+
"loss": 0.2729,
|
| 474 |
"step": 670
|
| 475 |
},
|
| 476 |
{
|
| 477 |
+
"epoch": 7.71,
|
| 478 |
+
"learning_rate": 2.6936026936026937e-05,
|
| 479 |
+
"loss": 0.2598,
|
| 480 |
"step": 680
|
| 481 |
},
|
| 482 |
{
|
| 483 |
+
"epoch": 7.82,
|
| 484 |
+
"learning_rate": 2.6515151515151516e-05,
|
| 485 |
+
"loss": 0.2331,
|
| 486 |
"step": 690
|
| 487 |
},
|
| 488 |
{
|
| 489 |
+
"epoch": 7.93,
|
| 490 |
+
"learning_rate": 2.6094276094276095e-05,
|
| 491 |
+
"loss": 0.2188,
|
| 492 |
"step": 700
|
| 493 |
},
|
| 494 |
{
|
| 495 |
+
"epoch": 8.0,
|
| 496 |
+
"eval_accuracy": 0.8322314049586776,
|
| 497 |
+
"eval_loss": 0.5088740587234497,
|
| 498 |
+
"eval_runtime": 16.6796,
|
| 499 |
+
"eval_samples_per_second": 72.544,
|
| 500 |
+
"eval_steps_per_second": 4.556,
|
| 501 |
+
"step": 706
|
| 502 |
+
},
|
| 503 |
+
{
|
| 504 |
+
"epoch": 8.05,
|
| 505 |
+
"learning_rate": 2.5673400673400677e-05,
|
| 506 |
+
"loss": 0.2445,
|
| 507 |
"step": 710
|
| 508 |
},
|
| 509 |
{
|
| 510 |
+
"epoch": 8.16,
|
| 511 |
+
"learning_rate": 2.5252525252525256e-05,
|
| 512 |
+
"loss": 0.2144,
|
| 513 |
"step": 720
|
| 514 |
},
|
| 515 |
{
|
| 516 |
+
"epoch": 8.27,
|
| 517 |
+
"learning_rate": 2.4831649831649835e-05,
|
| 518 |
+
"loss": 0.1956,
|
| 519 |
"step": 730
|
| 520 |
},
|
| 521 |
{
|
| 522 |
+
"epoch": 8.39,
|
| 523 |
+
"learning_rate": 2.441077441077441e-05,
|
| 524 |
+
"loss": 0.2103,
|
| 525 |
"step": 740
|
| 526 |
},
|
| 527 |
{
|
| 528 |
+
"epoch": 8.5,
|
| 529 |
+
"learning_rate": 2.398989898989899e-05,
|
| 530 |
+
"loss": 0.2539,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 531 |
"step": 750
|
| 532 |
},
|
| 533 |
{
|
| 534 |
+
"epoch": 8.61,
|
| 535 |
+
"learning_rate": 2.356902356902357e-05,
|
| 536 |
+
"loss": 0.2078,
|
| 537 |
"step": 760
|
| 538 |
},
|
| 539 |
{
|
| 540 |
+
"epoch": 8.73,
|
| 541 |
+
"learning_rate": 2.314814814814815e-05,
|
| 542 |
+
"loss": 0.2112,
|
| 543 |
"step": 770
|
| 544 |
},
|
| 545 |
{
|
| 546 |
+
"epoch": 8.84,
|
| 547 |
+
"learning_rate": 2.272727272727273e-05,
|
| 548 |
+
"loss": 0.2418,
|
| 549 |
"step": 780
|
| 550 |
},
|
| 551 |
{
|
| 552 |
+
"epoch": 8.95,
|
| 553 |
+
"learning_rate": 2.2306397306397308e-05,
|
| 554 |
+
"loss": 0.2112,
|
| 555 |
"step": 790
|
| 556 |
},
|
| 557 |
{
|
| 558 |
+
"epoch": 9.0,
|
| 559 |
+
"eval_accuracy": 0.8380165289256198,
|
| 560 |
+
"eval_loss": 0.5125700235366821,
|
| 561 |
+
"eval_runtime": 16.6395,
|
| 562 |
+
"eval_samples_per_second": 72.718,
|
| 563 |
+
"eval_steps_per_second": 4.567,
|
| 564 |
+
"step": 794
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"epoch": 9.07,
|
| 568 |
+
"learning_rate": 2.1885521885521887e-05,
|
| 569 |
+
"loss": 0.207,
|
| 570 |
"step": 800
|
| 571 |
},
|
| 572 |
{
|
| 573 |
+
"epoch": 9.18,
|
| 574 |
+
"learning_rate": 2.1464646464646466e-05,
|
| 575 |
+
"loss": 0.1844,
|
| 576 |
"step": 810
|
| 577 |
},
|
| 578 |
{
|
| 579 |
+
"epoch": 9.29,
|
| 580 |
+
"learning_rate": 2.1043771043771045e-05,
|
| 581 |
+
"loss": 0.1742,
|
| 582 |
"step": 820
|
| 583 |
},
|
| 584 |
{
|
| 585 |
+
"epoch": 9.41,
|
| 586 |
+
"learning_rate": 2.0622895622895624e-05,
|
| 587 |
+
"loss": 0.178,
|
| 588 |
"step": 830
|
| 589 |
},
|
| 590 |
{
|
| 591 |
+
"epoch": 9.52,
|
| 592 |
+
"learning_rate": 2.0202020202020203e-05,
|
| 593 |
+
"loss": 0.1957,
|
| 594 |
"step": 840
|
| 595 |
},
|
| 596 |
{
|
| 597 |
+
"epoch": 9.63,
|
| 598 |
+
"learning_rate": 1.978114478114478e-05,
|
| 599 |
+
"loss": 0.1833,
|
| 600 |
"step": 850
|
| 601 |
},
|
| 602 |
{
|
| 603 |
+
"epoch": 9.75,
|
| 604 |
+
"learning_rate": 1.936026936026936e-05,
|
| 605 |
+
"loss": 0.1955,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 606 |
"step": 860
|
| 607 |
},
|
| 608 |
{
|
| 609 |
+
"epoch": 9.86,
|
| 610 |
+
"learning_rate": 1.893939393939394e-05,
|
| 611 |
+
"loss": 0.2161,
|
| 612 |
"step": 870
|
| 613 |
},
|
| 614 |
{
|
| 615 |
+
"epoch": 9.97,
|
| 616 |
+
"learning_rate": 1.8518518518518518e-05,
|
| 617 |
+
"loss": 0.1895,
|
| 618 |
"step": 880
|
| 619 |
},
|
| 620 |
{
|
| 621 |
+
"epoch": 9.99,
|
| 622 |
+
"eval_accuracy": 0.8363636363636363,
|
| 623 |
+
"eval_loss": 0.5057435631752014,
|
| 624 |
+
"eval_runtime": 16.6718,
|
| 625 |
+
"eval_samples_per_second": 72.578,
|
| 626 |
+
"eval_steps_per_second": 4.559,
|
| 627 |
+
"step": 882
|
| 628 |
+
},
|
| 629 |
+
{
|
| 630 |
+
"epoch": 10.08,
|
| 631 |
+
"learning_rate": 1.80976430976431e-05,
|
| 632 |
+
"loss": 0.1848,
|
| 633 |
"step": 890
|
| 634 |
},
|
| 635 |
{
|
| 636 |
+
"epoch": 10.2,
|
| 637 |
+
"learning_rate": 1.7676767676767676e-05,
|
| 638 |
+
"loss": 0.1328,
|
| 639 |
"step": 900
|
| 640 |
},
|
| 641 |
{
|
| 642 |
+
"epoch": 10.31,
|
| 643 |
+
"learning_rate": 1.7255892255892255e-05,
|
| 644 |
+
"loss": 0.1383,
|
| 645 |
"step": 910
|
| 646 |
},
|
| 647 |
{
|
| 648 |
+
"epoch": 10.42,
|
| 649 |
+
"learning_rate": 1.6835016835016837e-05,
|
| 650 |
+
"loss": 0.1773,
|
| 651 |
"step": 920
|
| 652 |
},
|
| 653 |
{
|
| 654 |
+
"epoch": 10.54,
|
| 655 |
+
"learning_rate": 1.6414141414141416e-05,
|
| 656 |
+
"loss": 0.1363,
|
| 657 |
"step": 930
|
| 658 |
},
|
| 659 |
{
|
| 660 |
+
"epoch": 10.65,
|
| 661 |
+
"learning_rate": 1.5993265993265995e-05,
|
| 662 |
+
"loss": 0.1516,
|
| 663 |
"step": 940
|
| 664 |
},
|
| 665 |
{
|
| 666 |
+
"epoch": 10.76,
|
| 667 |
+
"learning_rate": 1.5572390572390574e-05,
|
| 668 |
+
"loss": 0.1917,
|
| 669 |
"step": 950
|
| 670 |
},
|
| 671 |
{
|
| 672 |
+
"epoch": 10.88,
|
| 673 |
+
"learning_rate": 1.5151515151515153e-05,
|
| 674 |
+
"loss": 0.1928,
|
| 675 |
"step": 960
|
| 676 |
},
|
| 677 |
{
|
| 678 |
+
"epoch": 10.99,
|
| 679 |
+
"learning_rate": 1.473063973063973e-05,
|
| 680 |
+
"loss": 0.1593,
|
| 681 |
+
"step": 970
|
|
|
|
|
|
|
|
|
|
| 682 |
},
|
| 683 |
{
|
| 684 |
+
"epoch": 10.99,
|
| 685 |
+
"eval_accuracy": 0.8528925619834711,
|
| 686 |
+
"eval_loss": 0.4851677417755127,
|
| 687 |
+
"eval_runtime": 16.8845,
|
| 688 |
+
"eval_samples_per_second": 71.664,
|
| 689 |
+
"eval_steps_per_second": 4.501,
|
| 690 |
"step": 970
|
| 691 |
},
|
| 692 |
{
|
| 693 |
+
"epoch": 11.1,
|
| 694 |
+
"learning_rate": 1.4309764309764309e-05,
|
| 695 |
+
"loss": 0.1578,
|
| 696 |
"step": 980
|
| 697 |
},
|
| 698 |
{
|
| 699 |
+
"epoch": 11.22,
|
| 700 |
+
"learning_rate": 1.388888888888889e-05,
|
| 701 |
+
"loss": 0.1757,
|
| 702 |
"step": 990
|
| 703 |
},
|
| 704 |
{
|
| 705 |
+
"epoch": 11.33,
|
| 706 |
+
"learning_rate": 1.3468013468013468e-05,
|
| 707 |
+
"loss": 0.125,
|
| 708 |
"step": 1000
|
| 709 |
},
|
| 710 |
{
|
| 711 |
+
"epoch": 11.44,
|
| 712 |
+
"learning_rate": 1.3047138047138047e-05,
|
| 713 |
+
"loss": 0.1566,
|
| 714 |
"step": 1010
|
| 715 |
},
|
| 716 |
{
|
| 717 |
+
"epoch": 11.56,
|
| 718 |
+
"learning_rate": 1.2626262626262628e-05,
|
| 719 |
+
"loss": 0.1841,
|
| 720 |
"step": 1020
|
| 721 |
},
|
| 722 |
{
|
| 723 |
+
"epoch": 11.67,
|
| 724 |
+
"learning_rate": 1.2205387205387205e-05,
|
| 725 |
+
"loss": 0.1832,
|
| 726 |
"step": 1030
|
| 727 |
},
|
| 728 |
{
|
| 729 |
+
"epoch": 11.78,
|
| 730 |
+
"learning_rate": 1.1784511784511786e-05,
|
| 731 |
+
"loss": 0.1604,
|
| 732 |
"step": 1040
|
| 733 |
},
|
| 734 |
{
|
| 735 |
+
"epoch": 11.9,
|
| 736 |
+
"learning_rate": 1.1363636363636365e-05,
|
| 737 |
+
"loss": 0.1463,
|
| 738 |
"step": 1050
|
| 739 |
},
|
| 740 |
{
|
| 741 |
+
"epoch": 12.0,
|
| 742 |
+
"eval_accuracy": 0.8429752066115702,
|
| 743 |
+
"eval_loss": 0.49344199895858765,
|
| 744 |
+
"eval_runtime": 16.7817,
|
| 745 |
+
"eval_samples_per_second": 72.102,
|
| 746 |
+
"eval_steps_per_second": 4.529,
|
| 747 |
+
"step": 1059
|
| 748 |
},
|
| 749 |
{
|
| 750 |
+
"epoch": 12.01,
|
| 751 |
+
"learning_rate": 1.0942760942760944e-05,
|
| 752 |
+
"loss": 0.133,
|
| 753 |
+
"step": 1060
|
|
|
|
|
|
|
|
|
|
| 754 |
},
|
| 755 |
{
|
| 756 |
+
"epoch": 12.12,
|
| 757 |
+
"learning_rate": 1.0521885521885522e-05,
|
| 758 |
+
"loss": 0.1293,
|
| 759 |
"step": 1070
|
| 760 |
},
|
| 761 |
{
|
| 762 |
+
"epoch": 12.24,
|
| 763 |
+
"learning_rate": 1.0101010101010101e-05,
|
| 764 |
+
"loss": 0.1232,
|
| 765 |
"step": 1080
|
| 766 |
},
|
| 767 |
{
|
| 768 |
+
"epoch": 12.35,
|
| 769 |
+
"learning_rate": 9.68013468013468e-06,
|
| 770 |
+
"loss": 0.1415,
|
| 771 |
"step": 1090
|
| 772 |
},
|
| 773 |
{
|
| 774 |
+
"epoch": 12.46,
|
| 775 |
+
"learning_rate": 9.259259259259259e-06,
|
| 776 |
+
"loss": 0.1425,
|
| 777 |
"step": 1100
|
| 778 |
},
|
| 779 |
{
|
| 780 |
+
"epoch": 12.58,
|
| 781 |
+
"learning_rate": 8.838383838383838e-06,
|
| 782 |
+
"loss": 0.1062,
|
| 783 |
"step": 1110
|
| 784 |
},
|
| 785 |
{
|
| 786 |
+
"epoch": 12.69,
|
| 787 |
+
"learning_rate": 8.417508417508419e-06,
|
| 788 |
+
"loss": 0.1193,
|
| 789 |
"step": 1120
|
| 790 |
},
|
| 791 |
{
|
| 792 |
+
"epoch": 12.8,
|
| 793 |
+
"learning_rate": 7.996632996632998e-06,
|
| 794 |
+
"loss": 0.1539,
|
| 795 |
"step": 1130
|
| 796 |
},
|
| 797 |
{
|
| 798 |
+
"epoch": 12.92,
|
| 799 |
+
"learning_rate": 7.5757575757575764e-06,
|
| 800 |
+
"loss": 0.1565,
|
| 801 |
"step": 1140
|
| 802 |
},
|
| 803 |
{
|
| 804 |
+
"epoch": 13.0,
|
| 805 |
+
"eval_accuracy": 0.8495867768595041,
|
| 806 |
+
"eval_loss": 0.47939661145210266,
|
| 807 |
+
"eval_runtime": 16.5708,
|
| 808 |
+
"eval_samples_per_second": 73.02,
|
| 809 |
+
"eval_steps_per_second": 4.586,
|
| 810 |
+
"step": 1147
|
| 811 |
+
},
|
| 812 |
+
{
|
| 813 |
+
"epoch": 13.03,
|
| 814 |
+
"learning_rate": 7.1548821548821545e-06,
|
| 815 |
+
"loss": 0.1193,
|
| 816 |
"step": 1150
|
| 817 |
},
|
| 818 |
{
|
| 819 |
+
"epoch": 13.14,
|
| 820 |
+
"learning_rate": 6.734006734006734e-06,
|
| 821 |
+
"loss": 0.1287,
|
| 822 |
"step": 1160
|
| 823 |
},
|
| 824 |
{
|
| 825 |
+
"epoch": 13.26,
|
| 826 |
+
"learning_rate": 6.313131313131314e-06,
|
| 827 |
+
"loss": 0.1186,
|
| 828 |
"step": 1170
|
| 829 |
},
|
| 830 |
{
|
| 831 |
+
"epoch": 13.37,
|
| 832 |
+
"learning_rate": 5.892255892255893e-06,
|
| 833 |
+
"loss": 0.1562,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 834 |
"step": 1180
|
| 835 |
},
|
| 836 |
{
|
| 837 |
+
"epoch": 13.48,
|
| 838 |
+
"learning_rate": 5.471380471380472e-06,
|
| 839 |
+
"loss": 0.1297,
|
| 840 |
"step": 1190
|
| 841 |
},
|
| 842 |
{
|
| 843 |
+
"epoch": 13.6,
|
| 844 |
+
"learning_rate": 5.050505050505051e-06,
|
| 845 |
+
"loss": 0.158,
|
| 846 |
"step": 1200
|
| 847 |
},
|
| 848 |
{
|
| 849 |
+
"epoch": 13.71,
|
| 850 |
+
"learning_rate": 4.6296296296296296e-06,
|
| 851 |
+
"loss": 0.1188,
|
| 852 |
"step": 1210
|
| 853 |
},
|
| 854 |
{
|
| 855 |
+
"epoch": 13.82,
|
| 856 |
+
"learning_rate": 4.208754208754209e-06,
|
| 857 |
+
"loss": 0.1237,
|
| 858 |
"step": 1220
|
| 859 |
},
|
| 860 |
{
|
| 861 |
+
"epoch": 13.94,
|
| 862 |
+
"learning_rate": 3.7878787878787882e-06,
|
| 863 |
+
"loss": 0.1236,
|
| 864 |
"step": 1230
|
| 865 |
},
|
| 866 |
{
|
| 867 |
+
"epoch": 13.99,
|
| 868 |
+
"eval_accuracy": 0.8462809917355372,
|
| 869 |
+
"eval_loss": 0.48630189895629883,
|
| 870 |
+
"eval_runtime": 16.6832,
|
| 871 |
+
"eval_samples_per_second": 72.528,
|
| 872 |
+
"eval_steps_per_second": 4.555,
|
| 873 |
+
"step": 1235
|
| 874 |
+
},
|
| 875 |
+
{
|
| 876 |
+
"epoch": 14.05,
|
| 877 |
+
"learning_rate": 3.367003367003367e-06,
|
| 878 |
+
"loss": 0.103,
|
| 879 |
"step": 1240
|
| 880 |
},
|
| 881 |
{
|
| 882 |
+
"epoch": 14.16,
|
| 883 |
+
"learning_rate": 2.9461279461279464e-06,
|
| 884 |
+
"loss": 0.1352,
|
| 885 |
"step": 1250
|
| 886 |
},
|
| 887 |
{
|
| 888 |
+
"epoch": 14.28,
|
| 889 |
+
"learning_rate": 2.5252525252525253e-06,
|
| 890 |
+
"loss": 0.1372,
|
| 891 |
"step": 1260
|
| 892 |
},
|
| 893 |
{
|
| 894 |
+
"epoch": 14.39,
|
| 895 |
+
"learning_rate": 2.1043771043771047e-06,
|
| 896 |
+
"loss": 0.1215,
|
| 897 |
"step": 1270
|
| 898 |
},
|
| 899 |
{
|
| 900 |
+
"epoch": 14.5,
|
| 901 |
+
"learning_rate": 1.6835016835016836e-06,
|
| 902 |
+
"loss": 0.1334,
|
| 903 |
"step": 1280
|
| 904 |
},
|
| 905 |
{
|
| 906 |
+
"epoch": 14.62,
|
| 907 |
+
"learning_rate": 1.2626262626262627e-06,
|
| 908 |
+
"loss": 0.1326,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 909 |
"step": 1290
|
| 910 |
},
|
| 911 |
{
|
| 912 |
+
"epoch": 14.73,
|
| 913 |
+
"learning_rate": 8.417508417508418e-07,
|
| 914 |
+
"loss": 0.0859,
|
| 915 |
"step": 1300
|
| 916 |
},
|
| 917 |
{
|
| 918 |
+
"epoch": 14.84,
|
| 919 |
+
"learning_rate": 4.208754208754209e-07,
|
| 920 |
+
"loss": 0.1313,
|
| 921 |
"step": 1310
|
| 922 |
},
|
| 923 |
{
|
| 924 |
+
"epoch": 14.96,
|
| 925 |
+
"learning_rate": 0.0,
|
| 926 |
+
"loss": 0.1407,
|
| 927 |
"step": 1320
|
| 928 |
},
|
| 929 |
{
|
| 930 |
+
"epoch": 14.96,
|
| 931 |
+
"eval_accuracy": 0.8495867768595041,
|
| 932 |
+
"eval_loss": 0.48121175169944763,
|
| 933 |
+
"eval_runtime": 16.5675,
|
| 934 |
+
"eval_samples_per_second": 73.034,
|
| 935 |
+
"eval_steps_per_second": 4.587,
|
| 936 |
+
"step": 1320
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 937 |
},
|
| 938 |
{
|
| 939 |
+
"epoch": 14.96,
|
| 940 |
+
"step": 1320,
|
| 941 |
+
"total_flos": 6.546875329145733e+18,
|
| 942 |
+
"train_loss": 0.44751356618874,
|
| 943 |
+
"train_runtime": 3414.1758,
|
| 944 |
+
"train_samples_per_second": 24.814,
|
| 945 |
+
"train_steps_per_second": 0.387
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 946 |
}
|
| 947 |
],
|
| 948 |
"logging_steps": 10,
|
| 949 |
+
"max_steps": 1320,
|
| 950 |
"num_train_epochs": 15,
|
| 951 |
"save_steps": 500,
|
| 952 |
+
"total_flos": 6.546875329145733e+18,
|
| 953 |
"trial_name": null,
|
| 954 |
"trial_params": null
|
| 955 |
}
|