End of training
Browse files
all_results.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
"epoch": 3.25,
|
| 3 |
-
"eval_accuracy": 0.
|
| 4 |
-
"eval_loss": 0.
|
| 5 |
-
"eval_runtime":
|
| 6 |
-
"eval_samples_per_second": 3.
|
| 7 |
-
"eval_steps_per_second": 3.
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"epoch": 3.25,
|
| 3 |
+
"eval_accuracy": 0.7716049382716049,
|
| 4 |
+
"eval_loss": 0.7548928260803223,
|
| 5 |
+
"eval_runtime": 46.747,
|
| 6 |
+
"eval_samples_per_second": 3.465,
|
| 7 |
+
"eval_steps_per_second": 3.465
|
| 8 |
}
|
runs/Mar29_08-28-57_ebf3b38c8ac4/events.out.tfevents.1680079787.ebf3b38c8ac4.131.2
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8ee7039f3d298e1dbb9a42c7d9aa40b2e729b5a328691bc1e1108b881ec34fe6
|
| 3 |
+
size 686
|
test_results.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
"epoch": 3.25,
|
| 3 |
-
"eval_accuracy": 0.
|
| 4 |
-
"eval_loss": 0.
|
| 5 |
-
"eval_runtime":
|
| 6 |
-
"eval_samples_per_second": 3.
|
| 7 |
-
"eval_steps_per_second": 3.
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"epoch": 3.25,
|
| 3 |
+
"eval_accuracy": 0.7716049382716049,
|
| 4 |
+
"eval_loss": 0.7548928260803223,
|
| 5 |
+
"eval_runtime": 46.747,
|
| 6 |
+
"eval_samples_per_second": 3.465,
|
| 7 |
+
"eval_steps_per_second": 3.465
|
| 8 |
}
|
trainer_state.json
CHANGED
|
@@ -1,799 +1,883 @@
|
|
| 1 |
{
|
| 2 |
-
"best_metric": 0.
|
| 3 |
-
"best_model_checkpoint": "videomae-base-finetuned-ucf101-subset/checkpoint-
|
| 4 |
"epoch": 3.25,
|
| 5 |
-
"global_step":
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
| 9 |
"log_history": [
|
| 10 |
{
|
| 11 |
"epoch": 0.01,
|
| 12 |
-
"learning_rate":
|
| 13 |
-
"loss": 2.
|
| 14 |
"step": 10
|
| 15 |
},
|
| 16 |
{
|
| 17 |
-
"epoch": 0.
|
| 18 |
-
"learning_rate":
|
| 19 |
-
"loss": 2.
|
| 20 |
"step": 20
|
| 21 |
},
|
| 22 |
{
|
| 23 |
-
"epoch": 0.
|
| 24 |
-
"learning_rate": 1.
|
| 25 |
-
"loss": 2.
|
| 26 |
"step": 30
|
| 27 |
},
|
| 28 |
{
|
| 29 |
"epoch": 0.03,
|
| 30 |
-
"learning_rate": 1.
|
| 31 |
-
"loss": 2.
|
| 32 |
"step": 40
|
| 33 |
},
|
| 34 |
{
|
| 35 |
"epoch": 0.04,
|
| 36 |
-
"learning_rate":
|
| 37 |
-
"loss": 2.
|
| 38 |
"step": 50
|
| 39 |
},
|
| 40 |
{
|
| 41 |
-
"epoch": 0.
|
| 42 |
-
"learning_rate": 2.
|
| 43 |
-
"loss": 2.
|
| 44 |
"step": 60
|
| 45 |
},
|
| 46 |
{
|
| 47 |
-
"epoch": 0.
|
| 48 |
-
"learning_rate": 2.
|
| 49 |
-
"loss": 2.
|
| 50 |
"step": 70
|
| 51 |
},
|
| 52 |
{
|
| 53 |
-
"epoch": 0.
|
| 54 |
-
"learning_rate":
|
| 55 |
-
"loss": 2.
|
| 56 |
"step": 80
|
| 57 |
},
|
| 58 |
{
|
| 59 |
"epoch": 0.07,
|
| 60 |
-
"learning_rate": 3.
|
| 61 |
-
"loss": 2.
|
| 62 |
"step": 90
|
| 63 |
},
|
| 64 |
{
|
| 65 |
-
"epoch": 0.
|
| 66 |
-
"learning_rate":
|
| 67 |
-
"loss": 2.
|
| 68 |
"step": 100
|
| 69 |
},
|
| 70 |
{
|
| 71 |
-
"epoch": 0.
|
| 72 |
-
"learning_rate": 4.
|
| 73 |
-
"loss": 2.
|
| 74 |
"step": 110
|
| 75 |
},
|
| 76 |
{
|
| 77 |
-
"epoch": 0.
|
| 78 |
-
"learning_rate":
|
| 79 |
-
"loss": 2.
|
| 80 |
"step": 120
|
| 81 |
},
|
| 82 |
{
|
| 83 |
-
"epoch": 0.
|
| 84 |
-
"learning_rate": 4.
|
| 85 |
-
"loss": 2.
|
| 86 |
"step": 130
|
| 87 |
},
|
| 88 |
{
|
| 89 |
-
"epoch": 0.
|
| 90 |
-
"learning_rate": 4.
|
| 91 |
-
"loss": 2.
|
| 92 |
"step": 140
|
| 93 |
},
|
| 94 |
{
|
| 95 |
-
"epoch": 0.
|
| 96 |
-
"learning_rate": 4.
|
| 97 |
-
"loss":
|
| 98 |
"step": 150
|
| 99 |
},
|
| 100 |
{
|
| 101 |
-
"epoch": 0.
|
| 102 |
-
"learning_rate": 4.
|
| 103 |
-
"loss":
|
| 104 |
"step": 160
|
| 105 |
},
|
| 106 |
{
|
| 107 |
-
"epoch": 0.
|
| 108 |
-
"learning_rate": 4.
|
| 109 |
-
"loss": 1.
|
| 110 |
"step": 170
|
| 111 |
},
|
| 112 |
{
|
| 113 |
-
"epoch": 0.
|
| 114 |
-
"learning_rate": 4.
|
| 115 |
-
"loss": 1.
|
| 116 |
"step": 180
|
| 117 |
},
|
| 118 |
{
|
| 119 |
-
"epoch": 0.
|
| 120 |
-
"learning_rate": 4.
|
| 121 |
-
"loss": 1.
|
| 122 |
"step": 190
|
| 123 |
},
|
| 124 |
{
|
| 125 |
-
"epoch": 0.
|
| 126 |
-
"learning_rate": 4.
|
| 127 |
-
"loss":
|
| 128 |
"step": 200
|
| 129 |
},
|
| 130 |
{
|
| 131 |
-
"epoch": 0.
|
| 132 |
-
"learning_rate": 4.
|
| 133 |
-
"loss": 2.
|
| 134 |
"step": 210
|
| 135 |
},
|
| 136 |
{
|
| 137 |
-
"epoch": 0.
|
| 138 |
-
"learning_rate": 4.
|
| 139 |
-
"loss":
|
| 140 |
"step": 220
|
| 141 |
},
|
| 142 |
{
|
| 143 |
-
"epoch": 0.
|
| 144 |
-
"learning_rate": 4.
|
| 145 |
-
"loss":
|
| 146 |
"step": 230
|
| 147 |
},
|
| 148 |
{
|
| 149 |
-
"epoch": 0.
|
| 150 |
-
"learning_rate": 4.
|
| 151 |
-
"loss": 2.
|
| 152 |
"step": 240
|
| 153 |
},
|
| 154 |
{
|
| 155 |
-
"epoch": 0.
|
| 156 |
-
"learning_rate": 4.
|
| 157 |
-
"loss":
|
| 158 |
"step": 250
|
| 159 |
},
|
| 160 |
{
|
| 161 |
-
"epoch": 0.
|
| 162 |
-
"learning_rate": 4.
|
| 163 |
-
"loss": 1.
|
| 164 |
"step": 260
|
| 165 |
},
|
| 166 |
{
|
| 167 |
-
"epoch": 0.
|
| 168 |
-
"learning_rate": 4.
|
| 169 |
-
"loss":
|
| 170 |
"step": 270
|
| 171 |
},
|
| 172 |
{
|
| 173 |
-
"epoch": 0.
|
| 174 |
-
"learning_rate": 4.
|
| 175 |
-
"loss": 1.
|
| 176 |
"step": 280
|
| 177 |
},
|
| 178 |
{
|
| 179 |
-
"epoch": 0.
|
| 180 |
-
"learning_rate": 4.
|
| 181 |
-
"loss":
|
| 182 |
"step": 290
|
| 183 |
},
|
| 184 |
{
|
| 185 |
-
"epoch": 0.
|
| 186 |
-
"learning_rate": 4.
|
| 187 |
-
"loss":
|
| 188 |
-
"step": 300
|
| 189 |
-
},
|
| 190 |
-
{
|
| 191 |
-
"epoch": 0.25,
|
| 192 |
-
"eval_accuracy": 0.5,
|
| 193 |
-
"eval_loss": 1.184158444404602,
|
| 194 |
-
"eval_runtime": 18.7511,
|
| 195 |
-
"eval_samples_per_second": 3.733,
|
| 196 |
-
"eval_steps_per_second": 3.733,
|
| 197 |
"step": 300
|
| 198 |
},
|
| 199 |
{
|
| 200 |
-
"epoch":
|
| 201 |
-
"learning_rate": 4.
|
| 202 |
-
"loss":
|
| 203 |
"step": 310
|
| 204 |
},
|
| 205 |
{
|
| 206 |
-
"epoch":
|
| 207 |
-
"learning_rate": 4.
|
| 208 |
-
"loss": 1.
|
| 209 |
"step": 320
|
| 210 |
},
|
| 211 |
{
|
| 212 |
-
"epoch":
|
| 213 |
-
"learning_rate": 4.
|
| 214 |
-
"loss": 1.
|
| 215 |
"step": 330
|
| 216 |
},
|
| 217 |
{
|
| 218 |
-
"epoch":
|
| 219 |
-
"
|
| 220 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 221 |
"step": 340
|
| 222 |
},
|
| 223 |
{
|
| 224 |
-
"epoch": 1.
|
| 225 |
-
"learning_rate":
|
| 226 |
-
"loss":
|
| 227 |
"step": 350
|
| 228 |
},
|
| 229 |
{
|
| 230 |
-
"epoch": 1.
|
| 231 |
-
"learning_rate":
|
| 232 |
-
"loss": 1.
|
| 233 |
"step": 360
|
| 234 |
},
|
| 235 |
{
|
| 236 |
-
"epoch": 1.
|
| 237 |
-
"learning_rate":
|
| 238 |
-
"loss": 1.
|
| 239 |
"step": 370
|
| 240 |
},
|
| 241 |
{
|
| 242 |
-
"epoch": 1.
|
| 243 |
-
"learning_rate": 3.
|
| 244 |
-
"loss":
|
| 245 |
"step": 380
|
| 246 |
},
|
| 247 |
{
|
| 248 |
-
"epoch": 1.
|
| 249 |
-
"learning_rate": 3.
|
| 250 |
-
"loss":
|
| 251 |
"step": 390
|
| 252 |
},
|
| 253 |
{
|
| 254 |
-
"epoch": 1.
|
| 255 |
-
"learning_rate": 3.
|
| 256 |
-
"loss": 1.
|
| 257 |
"step": 400
|
| 258 |
},
|
| 259 |
{
|
| 260 |
-
"epoch": 1.
|
| 261 |
-
"learning_rate": 3.
|
| 262 |
-
"loss":
|
| 263 |
"step": 410
|
| 264 |
},
|
| 265 |
{
|
| 266 |
-
"epoch": 1.
|
| 267 |
-
"learning_rate": 3.
|
| 268 |
-
"loss": 1.
|
| 269 |
"step": 420
|
| 270 |
},
|
| 271 |
{
|
| 272 |
-
"epoch": 1.
|
| 273 |
-
"learning_rate": 3.
|
| 274 |
-
"loss": 1.
|
| 275 |
"step": 430
|
| 276 |
},
|
| 277 |
{
|
| 278 |
-
"epoch": 1.
|
| 279 |
-
"learning_rate": 3.
|
| 280 |
-
"loss":
|
| 281 |
"step": 440
|
| 282 |
},
|
| 283 |
{
|
| 284 |
-
"epoch": 1.
|
| 285 |
-
"learning_rate": 3.
|
| 286 |
-
"loss":
|
| 287 |
"step": 450
|
| 288 |
},
|
| 289 |
{
|
| 290 |
-
"epoch": 1.
|
| 291 |
-
"learning_rate": 3.
|
| 292 |
-
"loss": 0.
|
| 293 |
"step": 460
|
| 294 |
},
|
| 295 |
{
|
| 296 |
-
"epoch": 1.
|
| 297 |
-
"learning_rate": 3.
|
| 298 |
-
"loss":
|
| 299 |
"step": 470
|
| 300 |
},
|
| 301 |
{
|
| 302 |
-
"epoch": 1.
|
| 303 |
-
"learning_rate": 3.
|
| 304 |
-
"loss":
|
| 305 |
"step": 480
|
| 306 |
},
|
| 307 |
{
|
| 308 |
-
"epoch": 1.
|
| 309 |
-
"learning_rate": 3.
|
| 310 |
-
"loss":
|
| 311 |
"step": 490
|
| 312 |
},
|
| 313 |
{
|
| 314 |
-
"epoch": 1.
|
| 315 |
-
"learning_rate": 3.
|
| 316 |
-
"loss":
|
| 317 |
"step": 500
|
| 318 |
},
|
| 319 |
{
|
| 320 |
-
"epoch": 1.
|
| 321 |
-
"learning_rate": 3.
|
| 322 |
-
"loss":
|
| 323 |
"step": 510
|
| 324 |
},
|
| 325 |
{
|
| 326 |
-
"epoch": 1.
|
| 327 |
-
"learning_rate": 3.
|
| 328 |
-
"loss": 0.
|
| 329 |
"step": 520
|
| 330 |
},
|
| 331 |
{
|
| 332 |
-
"epoch": 1.
|
| 333 |
-
"learning_rate": 3.
|
| 334 |
-
"loss": 0.
|
| 335 |
"step": 530
|
| 336 |
},
|
| 337 |
{
|
| 338 |
-
"epoch": 1.
|
| 339 |
-
"learning_rate": 3.
|
| 340 |
-
"loss": 1.
|
| 341 |
"step": 540
|
| 342 |
},
|
| 343 |
{
|
| 344 |
-
"epoch": 1.
|
| 345 |
-
"learning_rate": 3.
|
| 346 |
-
"loss":
|
| 347 |
"step": 550
|
| 348 |
},
|
| 349 |
{
|
| 350 |
-
"epoch": 1.
|
| 351 |
-
"learning_rate":
|
| 352 |
-
"loss": 0.
|
| 353 |
"step": 560
|
| 354 |
},
|
| 355 |
{
|
| 356 |
-
"epoch": 1.
|
| 357 |
-
"learning_rate":
|
| 358 |
-
"loss": 1.
|
| 359 |
"step": 570
|
| 360 |
},
|
| 361 |
{
|
| 362 |
-
"epoch": 1.
|
| 363 |
-
"learning_rate":
|
| 364 |
-
"loss": 1.
|
| 365 |
"step": 580
|
| 366 |
},
|
| 367 |
{
|
| 368 |
-
"epoch": 1.
|
| 369 |
-
"learning_rate":
|
| 370 |
-
"loss": 1.
|
| 371 |
"step": 590
|
| 372 |
},
|
| 373 |
{
|
| 374 |
-
"epoch": 1.
|
| 375 |
-
"learning_rate":
|
| 376 |
-
"loss":
|
| 377 |
-
"step": 600
|
| 378 |
-
},
|
| 379 |
-
{
|
| 380 |
-
"epoch": 1.25,
|
| 381 |
-
"eval_accuracy": 0.6142857142857143,
|
| 382 |
-
"eval_loss": 1.4308925867080688,
|
| 383 |
-
"eval_runtime": 18.6034,
|
| 384 |
-
"eval_samples_per_second": 3.763,
|
| 385 |
-
"eval_steps_per_second": 3.763,
|
| 386 |
"step": 600
|
| 387 |
},
|
| 388 |
{
|
| 389 |
-
"epoch":
|
| 390 |
-
"learning_rate":
|
| 391 |
-
"loss":
|
| 392 |
"step": 610
|
| 393 |
},
|
| 394 |
{
|
| 395 |
-
"epoch":
|
| 396 |
-
"learning_rate": 2.
|
| 397 |
-
"loss":
|
| 398 |
"step": 620
|
| 399 |
},
|
| 400 |
{
|
| 401 |
-
"epoch":
|
| 402 |
-
"learning_rate": 2.
|
| 403 |
-
"loss":
|
| 404 |
"step": 630
|
| 405 |
},
|
| 406 |
{
|
| 407 |
-
"epoch":
|
| 408 |
-
"learning_rate": 2.
|
| 409 |
-
"loss":
|
| 410 |
"step": 640
|
| 411 |
},
|
| 412 |
{
|
| 413 |
-
"epoch":
|
| 414 |
-
"learning_rate": 2.
|
| 415 |
-
"loss":
|
| 416 |
"step": 650
|
| 417 |
},
|
| 418 |
{
|
| 419 |
-
"epoch":
|
| 420 |
-
"learning_rate": 2.
|
| 421 |
-
"loss": 0.
|
| 422 |
"step": 660
|
| 423 |
},
|
| 424 |
{
|
| 425 |
-
"epoch":
|
| 426 |
-
"learning_rate": 2.
|
| 427 |
-
"loss": 0.
|
| 428 |
"step": 670
|
| 429 |
},
|
| 430 |
{
|
| 431 |
-
"epoch":
|
| 432 |
-
"
|
| 433 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 434 |
"step": 680
|
| 435 |
},
|
| 436 |
{
|
| 437 |
-
"epoch": 2.
|
| 438 |
-
"learning_rate": 2.
|
| 439 |
-
"loss": 0.
|
| 440 |
"step": 690
|
| 441 |
},
|
| 442 |
{
|
| 443 |
-
"epoch": 2.
|
| 444 |
-
"learning_rate": 2.
|
| 445 |
-
"loss": 0.
|
| 446 |
"step": 700
|
| 447 |
},
|
| 448 |
{
|
| 449 |
-
"epoch": 2.
|
| 450 |
-
"learning_rate": 2.
|
| 451 |
-
"loss": 0.
|
| 452 |
"step": 710
|
| 453 |
},
|
| 454 |
{
|
| 455 |
-
"epoch": 2.
|
| 456 |
-
"learning_rate": 2.
|
| 457 |
-
"loss": 0.
|
| 458 |
"step": 720
|
| 459 |
},
|
| 460 |
{
|
| 461 |
-
"epoch": 2.
|
| 462 |
-
"learning_rate": 2.
|
| 463 |
-
"loss": 0.
|
| 464 |
"step": 730
|
| 465 |
},
|
| 466 |
{
|
| 467 |
-
"epoch": 2.
|
| 468 |
-
"learning_rate": 2.
|
| 469 |
-
"loss":
|
| 470 |
"step": 740
|
| 471 |
},
|
| 472 |
{
|
| 473 |
-
"epoch": 2.
|
| 474 |
-
"learning_rate": 2.
|
| 475 |
-
"loss": 0.
|
| 476 |
"step": 750
|
| 477 |
},
|
| 478 |
{
|
| 479 |
-
"epoch": 2.
|
| 480 |
-
"learning_rate": 2.
|
| 481 |
-
"loss":
|
| 482 |
"step": 760
|
| 483 |
},
|
| 484 |
{
|
| 485 |
-
"epoch": 2.
|
| 486 |
-
"learning_rate":
|
| 487 |
-
"loss": 0.
|
| 488 |
"step": 770
|
| 489 |
},
|
| 490 |
{
|
| 491 |
-
"epoch": 2.
|
| 492 |
-
"learning_rate":
|
| 493 |
-
"loss":
|
| 494 |
"step": 780
|
| 495 |
},
|
| 496 |
{
|
| 497 |
-
"epoch": 2.
|
| 498 |
-
"learning_rate":
|
| 499 |
-
"loss": 0.
|
| 500 |
"step": 790
|
| 501 |
},
|
| 502 |
{
|
| 503 |
-
"epoch": 2.
|
| 504 |
-
"learning_rate":
|
| 505 |
-
"loss": 0.
|
| 506 |
"step": 800
|
| 507 |
},
|
| 508 |
{
|
| 509 |
-
"epoch": 2.
|
| 510 |
-
"learning_rate":
|
| 511 |
-
"loss": 0.
|
| 512 |
"step": 810
|
| 513 |
},
|
| 514 |
{
|
| 515 |
-
"epoch": 2.
|
| 516 |
-
"learning_rate":
|
| 517 |
-
"loss": 0.
|
| 518 |
"step": 820
|
| 519 |
},
|
| 520 |
{
|
| 521 |
-
"epoch": 2.
|
| 522 |
-
"learning_rate":
|
| 523 |
-
"loss":
|
| 524 |
"step": 830
|
| 525 |
},
|
| 526 |
{
|
| 527 |
-
"epoch": 2.
|
| 528 |
-
"learning_rate":
|
| 529 |
-
"loss": 0.
|
| 530 |
"step": 840
|
| 531 |
},
|
| 532 |
{
|
| 533 |
-
"epoch": 2.
|
| 534 |
-
"learning_rate":
|
| 535 |
-
"loss": 0.
|
| 536 |
"step": 850
|
| 537 |
},
|
| 538 |
{
|
| 539 |
-
"epoch": 2.
|
| 540 |
-
"learning_rate": 1.
|
| 541 |
-
"loss": 0.
|
| 542 |
"step": 860
|
| 543 |
},
|
| 544 |
{
|
| 545 |
-
"epoch": 2.
|
| 546 |
-
"learning_rate": 1.
|
| 547 |
-
"loss": 0.
|
| 548 |
"step": 870
|
| 549 |
},
|
| 550 |
{
|
| 551 |
-
"epoch": 2.
|
| 552 |
-
"learning_rate": 1.
|
| 553 |
-
"loss":
|
| 554 |
"step": 880
|
| 555 |
},
|
| 556 |
{
|
| 557 |
-
"epoch": 2.
|
| 558 |
-
"learning_rate": 1.
|
| 559 |
-
"loss": 0.
|
| 560 |
"step": 890
|
| 561 |
},
|
| 562 |
{
|
| 563 |
-
"epoch": 2.
|
| 564 |
-
"learning_rate": 1.
|
| 565 |
-
"loss": 0.
|
| 566 |
-
"step": 900
|
| 567 |
-
},
|
| 568 |
-
{
|
| 569 |
-
"epoch": 2.25,
|
| 570 |
-
"eval_accuracy": 0.8,
|
| 571 |
-
"eval_loss": 1.079487919807434,
|
| 572 |
-
"eval_runtime": 18.4652,
|
| 573 |
-
"eval_samples_per_second": 3.791,
|
| 574 |
-
"eval_steps_per_second": 3.791,
|
| 575 |
"step": 900
|
| 576 |
},
|
| 577 |
{
|
| 578 |
-
"epoch":
|
| 579 |
-
"learning_rate": 1.
|
| 580 |
-
"loss": 0.
|
| 581 |
"step": 910
|
| 582 |
},
|
| 583 |
{
|
| 584 |
-
"epoch":
|
| 585 |
-
"learning_rate": 1.
|
| 586 |
-
"loss": 0.
|
| 587 |
"step": 920
|
| 588 |
},
|
| 589 |
{
|
| 590 |
-
"epoch":
|
| 591 |
-
"learning_rate": 1.
|
| 592 |
-
"loss": 0.
|
| 593 |
"step": 930
|
| 594 |
},
|
| 595 |
{
|
| 596 |
-
"epoch":
|
| 597 |
-
"learning_rate": 1.
|
| 598 |
-
"loss": 0.
|
| 599 |
"step": 940
|
| 600 |
},
|
| 601 |
{
|
| 602 |
-
"epoch":
|
| 603 |
-
"learning_rate": 1.
|
| 604 |
-
"loss": 0.
|
| 605 |
"step": 950
|
| 606 |
},
|
| 607 |
{
|
| 608 |
-
"epoch":
|
| 609 |
-
"learning_rate": 1.
|
| 610 |
-
"loss": 0.
|
| 611 |
"step": 960
|
| 612 |
},
|
| 613 |
{
|
| 614 |
-
"epoch":
|
| 615 |
-
"learning_rate": 1.
|
| 616 |
-
"loss": 0.
|
| 617 |
"step": 970
|
| 618 |
},
|
| 619 |
{
|
| 620 |
-
"epoch":
|
| 621 |
-
"learning_rate": 1.
|
| 622 |
-
"loss": 0.
|
| 623 |
"step": 980
|
| 624 |
},
|
| 625 |
{
|
| 626 |
-
"epoch":
|
| 627 |
-
"learning_rate":
|
| 628 |
-
"loss": 0.
|
| 629 |
"step": 990
|
| 630 |
},
|
| 631 |
{
|
| 632 |
-
"epoch":
|
| 633 |
-
"learning_rate":
|
| 634 |
-
"loss": 0.
|
| 635 |
"step": 1000
|
| 636 |
},
|
| 637 |
{
|
| 638 |
-
"epoch":
|
| 639 |
-
"
|
| 640 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 641 |
"step": 1010
|
| 642 |
},
|
| 643 |
{
|
| 644 |
-
"epoch": 3.
|
| 645 |
-
"learning_rate":
|
| 646 |
-
"loss": 0.
|
| 647 |
"step": 1020
|
| 648 |
},
|
| 649 |
{
|
| 650 |
-
"epoch": 3.
|
| 651 |
-
"learning_rate":
|
| 652 |
-
"loss": 0.
|
| 653 |
"step": 1030
|
| 654 |
},
|
| 655 |
{
|
| 656 |
-
"epoch": 3.
|
| 657 |
-
"learning_rate":
|
| 658 |
-
"loss": 0.
|
| 659 |
"step": 1040
|
| 660 |
},
|
| 661 |
{
|
| 662 |
-
"epoch": 3.
|
| 663 |
-
"learning_rate":
|
| 664 |
-
"loss": 0.
|
| 665 |
"step": 1050
|
| 666 |
},
|
| 667 |
{
|
| 668 |
-
"epoch": 3.
|
| 669 |
-
"learning_rate":
|
| 670 |
-
"loss": 0.
|
| 671 |
"step": 1060
|
| 672 |
},
|
| 673 |
{
|
| 674 |
-
"epoch": 3.
|
| 675 |
-
"learning_rate":
|
| 676 |
-
"loss": 0.
|
| 677 |
"step": 1070
|
| 678 |
},
|
| 679 |
{
|
| 680 |
-
"epoch": 3.
|
| 681 |
-
"learning_rate":
|
| 682 |
-
"loss":
|
| 683 |
"step": 1080
|
| 684 |
},
|
| 685 |
{
|
| 686 |
-
"epoch": 3.
|
| 687 |
-
"learning_rate":
|
| 688 |
-
"loss": 0.
|
| 689 |
"step": 1090
|
| 690 |
},
|
| 691 |
{
|
| 692 |
-
"epoch": 3.
|
| 693 |
-
"learning_rate":
|
| 694 |
-
"loss": 0.
|
| 695 |
"step": 1100
|
| 696 |
},
|
| 697 |
{
|
| 698 |
-
"epoch": 3.
|
| 699 |
-
"learning_rate":
|
| 700 |
-
"loss": 0.
|
| 701 |
"step": 1110
|
| 702 |
},
|
| 703 |
{
|
| 704 |
-
"epoch": 3.
|
| 705 |
-
"learning_rate":
|
| 706 |
-
"loss": 0.
|
| 707 |
"step": 1120
|
| 708 |
},
|
| 709 |
{
|
| 710 |
-
"epoch": 3.
|
| 711 |
-
"learning_rate":
|
| 712 |
-
"loss": 0.
|
| 713 |
"step": 1130
|
| 714 |
},
|
| 715 |
{
|
| 716 |
-
"epoch": 3.
|
| 717 |
-
"learning_rate":
|
| 718 |
-
"loss": 0.
|
| 719 |
"step": 1140
|
| 720 |
},
|
| 721 |
{
|
| 722 |
-
"epoch": 3.
|
| 723 |
-
"learning_rate":
|
| 724 |
-
"loss": 0.
|
| 725 |
"step": 1150
|
| 726 |
},
|
| 727 |
{
|
| 728 |
-
"epoch": 3.
|
| 729 |
-
"learning_rate":
|
| 730 |
-
"loss": 0.
|
| 731 |
"step": 1160
|
| 732 |
},
|
| 733 |
{
|
| 734 |
-
"epoch": 3.
|
| 735 |
-
"learning_rate":
|
| 736 |
-
"loss": 0.
|
| 737 |
"step": 1170
|
| 738 |
},
|
| 739 |
{
|
| 740 |
-
"epoch": 3.
|
| 741 |
-
"learning_rate":
|
| 742 |
-
"loss": 0.
|
| 743 |
"step": 1180
|
| 744 |
},
|
| 745 |
{
|
| 746 |
-
"epoch": 3.
|
| 747 |
-
"learning_rate":
|
| 748 |
-
"loss": 0.
|
| 749 |
"step": 1190
|
| 750 |
},
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 751 |
{
|
| 752 |
"epoch": 3.25,
|
| 753 |
"learning_rate": 0.0,
|
| 754 |
-
"loss":
|
| 755 |
-
"step":
|
| 756 |
},
|
| 757 |
{
|
| 758 |
"epoch": 3.25,
|
| 759 |
-
"eval_accuracy": 0.
|
| 760 |
-
"eval_loss": 0.
|
| 761 |
-
"eval_runtime":
|
| 762 |
-
"eval_samples_per_second": 3.
|
| 763 |
-
"eval_steps_per_second": 3.
|
| 764 |
-
"step":
|
| 765 |
},
|
| 766 |
{
|
| 767 |
"epoch": 3.25,
|
| 768 |
-
"step":
|
| 769 |
-
"total_flos": 1.
|
| 770 |
-
"train_loss":
|
| 771 |
-
"train_runtime":
|
| 772 |
-
"train_samples_per_second": 1.
|
| 773 |
-
"train_steps_per_second": 1.
|
| 774 |
},
|
| 775 |
{
|
| 776 |
"epoch": 3.25,
|
| 777 |
-
"eval_accuracy": 0.
|
| 778 |
-
"eval_loss": 0.
|
| 779 |
-
"eval_runtime":
|
| 780 |
-
"eval_samples_per_second": 3.
|
| 781 |
-
"eval_steps_per_second": 3.
|
| 782 |
-
"step":
|
| 783 |
},
|
| 784 |
{
|
| 785 |
"epoch": 3.25,
|
| 786 |
-
"eval_accuracy": 0.
|
| 787 |
-
"eval_loss": 0.
|
| 788 |
-
"eval_runtime":
|
| 789 |
-
"eval_samples_per_second": 3.
|
| 790 |
-
"eval_steps_per_second": 3.
|
| 791 |
-
"step":
|
| 792 |
}
|
| 793 |
],
|
| 794 |
-
"max_steps":
|
| 795 |
"num_train_epochs": 9223372036854775807,
|
| 796 |
-
"total_flos": 1.
|
| 797 |
"trial_name": null,
|
| 798 |
"trial_params": null
|
| 799 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_metric": 0.7297297297297297,
|
| 3 |
+
"best_model_checkpoint": "videomae-base-finetuned-ucf101-subset/checkpoint-1340",
|
| 4 |
"epoch": 3.25,
|
| 5 |
+
"global_step": 1340,
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
| 9 |
"log_history": [
|
| 10 |
{
|
| 11 |
"epoch": 0.01,
|
| 12 |
+
"learning_rate": 3.7313432835820893e-06,
|
| 13 |
+
"loss": 2.4537,
|
| 14 |
"step": 10
|
| 15 |
},
|
| 16 |
{
|
| 17 |
+
"epoch": 0.01,
|
| 18 |
+
"learning_rate": 7.4626865671641785e-06,
|
| 19 |
+
"loss": 2.4565,
|
| 20 |
"step": 20
|
| 21 |
},
|
| 22 |
{
|
| 23 |
+
"epoch": 0.02,
|
| 24 |
+
"learning_rate": 1.119402985074627e-05,
|
| 25 |
+
"loss": 2.3771,
|
| 26 |
"step": 30
|
| 27 |
},
|
| 28 |
{
|
| 29 |
"epoch": 0.03,
|
| 30 |
+
"learning_rate": 1.4925373134328357e-05,
|
| 31 |
+
"loss": 2.3446,
|
| 32 |
"step": 40
|
| 33 |
},
|
| 34 |
{
|
| 35 |
"epoch": 0.04,
|
| 36 |
+
"learning_rate": 1.865671641791045e-05,
|
| 37 |
+
"loss": 2.3192,
|
| 38 |
"step": 50
|
| 39 |
},
|
| 40 |
{
|
| 41 |
+
"epoch": 0.04,
|
| 42 |
+
"learning_rate": 2.238805970149254e-05,
|
| 43 |
+
"loss": 2.3792,
|
| 44 |
"step": 60
|
| 45 |
},
|
| 46 |
{
|
| 47 |
+
"epoch": 0.05,
|
| 48 |
+
"learning_rate": 2.6119402985074626e-05,
|
| 49 |
+
"loss": 2.4095,
|
| 50 |
"step": 70
|
| 51 |
},
|
| 52 |
{
|
| 53 |
+
"epoch": 0.06,
|
| 54 |
+
"learning_rate": 2.9850746268656714e-05,
|
| 55 |
+
"loss": 2.4641,
|
| 56 |
"step": 80
|
| 57 |
},
|
| 58 |
{
|
| 59 |
"epoch": 0.07,
|
| 60 |
+
"learning_rate": 3.358208955223881e-05,
|
| 61 |
+
"loss": 2.2262,
|
| 62 |
"step": 90
|
| 63 |
},
|
| 64 |
{
|
| 65 |
+
"epoch": 0.07,
|
| 66 |
+
"learning_rate": 3.73134328358209e-05,
|
| 67 |
+
"loss": 2.3417,
|
| 68 |
"step": 100
|
| 69 |
},
|
| 70 |
{
|
| 71 |
+
"epoch": 0.08,
|
| 72 |
+
"learning_rate": 4.104477611940299e-05,
|
| 73 |
+
"loss": 2.2397,
|
| 74 |
"step": 110
|
| 75 |
},
|
| 76 |
{
|
| 77 |
+
"epoch": 0.09,
|
| 78 |
+
"learning_rate": 4.477611940298508e-05,
|
| 79 |
+
"loss": 2.0114,
|
| 80 |
"step": 120
|
| 81 |
},
|
| 82 |
{
|
| 83 |
+
"epoch": 0.1,
|
| 84 |
+
"learning_rate": 4.850746268656717e-05,
|
| 85 |
+
"loss": 2.349,
|
| 86 |
"step": 130
|
| 87 |
},
|
| 88 |
{
|
| 89 |
+
"epoch": 0.1,
|
| 90 |
+
"learning_rate": 4.975124378109453e-05,
|
| 91 |
+
"loss": 2.058,
|
| 92 |
"step": 140
|
| 93 |
},
|
| 94 |
{
|
| 95 |
+
"epoch": 0.11,
|
| 96 |
+
"learning_rate": 4.933665008291874e-05,
|
| 97 |
+
"loss": 2.0433,
|
| 98 |
"step": 150
|
| 99 |
},
|
| 100 |
{
|
| 101 |
+
"epoch": 0.12,
|
| 102 |
+
"learning_rate": 4.8922056384742954e-05,
|
| 103 |
+
"loss": 2.3097,
|
| 104 |
"step": 160
|
| 105 |
},
|
| 106 |
{
|
| 107 |
+
"epoch": 0.13,
|
| 108 |
+
"learning_rate": 4.850746268656717e-05,
|
| 109 |
+
"loss": 1.9613,
|
| 110 |
"step": 170
|
| 111 |
},
|
| 112 |
{
|
| 113 |
+
"epoch": 0.13,
|
| 114 |
+
"learning_rate": 4.8092868988391376e-05,
|
| 115 |
+
"loss": 1.715,
|
| 116 |
"step": 180
|
| 117 |
},
|
| 118 |
{
|
| 119 |
+
"epoch": 0.14,
|
| 120 |
+
"learning_rate": 4.767827529021559e-05,
|
| 121 |
+
"loss": 1.9524,
|
| 122 |
"step": 190
|
| 123 |
},
|
| 124 |
{
|
| 125 |
+
"epoch": 0.15,
|
| 126 |
+
"learning_rate": 4.7263681592039805e-05,
|
| 127 |
+
"loss": 2.4086,
|
| 128 |
"step": 200
|
| 129 |
},
|
| 130 |
{
|
| 131 |
+
"epoch": 0.16,
|
| 132 |
+
"learning_rate": 4.684908789386401e-05,
|
| 133 |
+
"loss": 2.7159,
|
| 134 |
"step": 210
|
| 135 |
},
|
| 136 |
{
|
| 137 |
+
"epoch": 0.16,
|
| 138 |
+
"learning_rate": 4.643449419568823e-05,
|
| 139 |
+
"loss": 2.1701,
|
| 140 |
"step": 220
|
| 141 |
},
|
| 142 |
{
|
| 143 |
+
"epoch": 0.17,
|
| 144 |
+
"learning_rate": 4.601990049751244e-05,
|
| 145 |
+
"loss": 2.3041,
|
| 146 |
"step": 230
|
| 147 |
},
|
| 148 |
{
|
| 149 |
+
"epoch": 0.18,
|
| 150 |
+
"learning_rate": 4.560530679933665e-05,
|
| 151 |
+
"loss": 2.2435,
|
| 152 |
"step": 240
|
| 153 |
},
|
| 154 |
{
|
| 155 |
+
"epoch": 0.19,
|
| 156 |
+
"learning_rate": 4.5190713101160863e-05,
|
| 157 |
+
"loss": 2.1136,
|
| 158 |
"step": 250
|
| 159 |
},
|
| 160 |
{
|
| 161 |
+
"epoch": 0.19,
|
| 162 |
+
"learning_rate": 4.477611940298508e-05,
|
| 163 |
+
"loss": 1.3329,
|
| 164 |
"step": 260
|
| 165 |
},
|
| 166 |
{
|
| 167 |
+
"epoch": 0.2,
|
| 168 |
+
"learning_rate": 4.4361525704809285e-05,
|
| 169 |
+
"loss": 2.1726,
|
| 170 |
"step": 270
|
| 171 |
},
|
| 172 |
{
|
| 173 |
+
"epoch": 0.21,
|
| 174 |
+
"learning_rate": 4.39469320066335e-05,
|
| 175 |
+
"loss": 1.7117,
|
| 176 |
"step": 280
|
| 177 |
},
|
| 178 |
{
|
| 179 |
+
"epoch": 0.22,
|
| 180 |
+
"learning_rate": 4.3532338308457714e-05,
|
| 181 |
+
"loss": 2.1894,
|
| 182 |
"step": 290
|
| 183 |
},
|
| 184 |
{
|
| 185 |
+
"epoch": 0.22,
|
| 186 |
+
"learning_rate": 4.311774461028192e-05,
|
| 187 |
+
"loss": 2.308,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 188 |
"step": 300
|
| 189 |
},
|
| 190 |
{
|
| 191 |
+
"epoch": 0.23,
|
| 192 |
+
"learning_rate": 4.2703150912106136e-05,
|
| 193 |
+
"loss": 2.2462,
|
| 194 |
"step": 310
|
| 195 |
},
|
| 196 |
{
|
| 197 |
+
"epoch": 0.24,
|
| 198 |
+
"learning_rate": 4.228855721393035e-05,
|
| 199 |
+
"loss": 1.6541,
|
| 200 |
"step": 320
|
| 201 |
},
|
| 202 |
{
|
| 203 |
+
"epoch": 0.25,
|
| 204 |
+
"learning_rate": 4.187396351575456e-05,
|
| 205 |
+
"loss": 1.2417,
|
| 206 |
"step": 330
|
| 207 |
},
|
| 208 |
{
|
| 209 |
+
"epoch": 0.25,
|
| 210 |
+
"eval_accuracy": 0.35135135135135137,
|
| 211 |
+
"eval_loss": 1.5927890539169312,
|
| 212 |
+
"eval_runtime": 21.5381,
|
| 213 |
+
"eval_samples_per_second": 3.436,
|
| 214 |
+
"eval_steps_per_second": 3.436,
|
| 215 |
+
"step": 335
|
| 216 |
+
},
|
| 217 |
+
{
|
| 218 |
+
"epoch": 1.0,
|
| 219 |
+
"learning_rate": 4.145936981757877e-05,
|
| 220 |
+
"loss": 2.0567,
|
| 221 |
"step": 340
|
| 222 |
},
|
| 223 |
{
|
| 224 |
+
"epoch": 1.01,
|
| 225 |
+
"learning_rate": 4.104477611940299e-05,
|
| 226 |
+
"loss": 1.7841,
|
| 227 |
"step": 350
|
| 228 |
},
|
| 229 |
{
|
| 230 |
+
"epoch": 1.02,
|
| 231 |
+
"learning_rate": 4.0630182421227194e-05,
|
| 232 |
+
"loss": 1.2214,
|
| 233 |
"step": 360
|
| 234 |
},
|
| 235 |
{
|
| 236 |
+
"epoch": 1.03,
|
| 237 |
+
"learning_rate": 4.021558872305141e-05,
|
| 238 |
+
"loss": 1.3657,
|
| 239 |
"step": 370
|
| 240 |
},
|
| 241 |
{
|
| 242 |
+
"epoch": 1.03,
|
| 243 |
+
"learning_rate": 3.980099502487562e-05,
|
| 244 |
+
"loss": 1.3927,
|
| 245 |
"step": 380
|
| 246 |
},
|
| 247 |
{
|
| 248 |
+
"epoch": 1.04,
|
| 249 |
+
"learning_rate": 3.938640132669984e-05,
|
| 250 |
+
"loss": 1.5291,
|
| 251 |
"step": 390
|
| 252 |
},
|
| 253 |
{
|
| 254 |
+
"epoch": 1.05,
|
| 255 |
+
"learning_rate": 3.8971807628524045e-05,
|
| 256 |
+
"loss": 1.4722,
|
| 257 |
"step": 400
|
| 258 |
},
|
| 259 |
{
|
| 260 |
+
"epoch": 1.06,
|
| 261 |
+
"learning_rate": 3.855721393034826e-05,
|
| 262 |
+
"loss": 1.024,
|
| 263 |
"step": 410
|
| 264 |
},
|
| 265 |
{
|
| 266 |
+
"epoch": 1.06,
|
| 267 |
+
"learning_rate": 3.8142620232172474e-05,
|
| 268 |
+
"loss": 1.7965,
|
| 269 |
"step": 420
|
| 270 |
},
|
| 271 |
{
|
| 272 |
+
"epoch": 1.07,
|
| 273 |
+
"learning_rate": 3.772802653399669e-05,
|
| 274 |
+
"loss": 1.5893,
|
| 275 |
"step": 430
|
| 276 |
},
|
| 277 |
{
|
| 278 |
+
"epoch": 1.08,
|
| 279 |
+
"learning_rate": 3.73134328358209e-05,
|
| 280 |
+
"loss": 0.9593,
|
| 281 |
"step": 440
|
| 282 |
},
|
| 283 |
{
|
| 284 |
+
"epoch": 1.09,
|
| 285 |
+
"learning_rate": 3.689883913764511e-05,
|
| 286 |
+
"loss": 1.4799,
|
| 287 |
"step": 450
|
| 288 |
},
|
| 289 |
{
|
| 290 |
+
"epoch": 1.09,
|
| 291 |
+
"learning_rate": 3.6484245439469325e-05,
|
| 292 |
+
"loss": 0.7833,
|
| 293 |
"step": 460
|
| 294 |
},
|
| 295 |
{
|
| 296 |
+
"epoch": 1.1,
|
| 297 |
+
"learning_rate": 3.606965174129354e-05,
|
| 298 |
+
"loss": 2.0512,
|
| 299 |
"step": 470
|
| 300 |
},
|
| 301 |
{
|
| 302 |
+
"epoch": 1.11,
|
| 303 |
+
"learning_rate": 3.565505804311775e-05,
|
| 304 |
+
"loss": 1.063,
|
| 305 |
"step": 480
|
| 306 |
},
|
| 307 |
{
|
| 308 |
+
"epoch": 1.12,
|
| 309 |
+
"learning_rate": 3.524046434494196e-05,
|
| 310 |
+
"loss": 1.4163,
|
| 311 |
"step": 490
|
| 312 |
},
|
| 313 |
{
|
| 314 |
+
"epoch": 1.12,
|
| 315 |
+
"learning_rate": 3.4825870646766175e-05,
|
| 316 |
+
"loss": 1.9212,
|
| 317 |
"step": 500
|
| 318 |
},
|
| 319 |
{
|
| 320 |
+
"epoch": 1.13,
|
| 321 |
+
"learning_rate": 3.441127694859038e-05,
|
| 322 |
+
"loss": 1.4935,
|
| 323 |
"step": 510
|
| 324 |
},
|
| 325 |
{
|
| 326 |
+
"epoch": 1.14,
|
| 327 |
+
"learning_rate": 3.39966832504146e-05,
|
| 328 |
+
"loss": 0.5912,
|
| 329 |
"step": 520
|
| 330 |
},
|
| 331 |
{
|
| 332 |
+
"epoch": 1.15,
|
| 333 |
+
"learning_rate": 3.358208955223881e-05,
|
| 334 |
+
"loss": 0.4764,
|
| 335 |
"step": 530
|
| 336 |
},
|
| 337 |
{
|
| 338 |
+
"epoch": 1.15,
|
| 339 |
+
"learning_rate": 3.316749585406302e-05,
|
| 340 |
+
"loss": 1.0399,
|
| 341 |
"step": 540
|
| 342 |
},
|
| 343 |
{
|
| 344 |
+
"epoch": 1.16,
|
| 345 |
+
"learning_rate": 3.2752902155887234e-05,
|
| 346 |
+
"loss": 1.4846,
|
| 347 |
"step": 550
|
| 348 |
},
|
| 349 |
{
|
| 350 |
+
"epoch": 1.17,
|
| 351 |
+
"learning_rate": 3.233830845771145e-05,
|
| 352 |
+
"loss": 0.8735,
|
| 353 |
"step": 560
|
| 354 |
},
|
| 355 |
{
|
| 356 |
+
"epoch": 1.18,
|
| 357 |
+
"learning_rate": 3.1923714759535656e-05,
|
| 358 |
+
"loss": 1.193,
|
| 359 |
"step": 570
|
| 360 |
},
|
| 361 |
{
|
| 362 |
+
"epoch": 1.18,
|
| 363 |
+
"learning_rate": 3.150912106135987e-05,
|
| 364 |
+
"loss": 1.5802,
|
| 365 |
"step": 580
|
| 366 |
},
|
| 367 |
{
|
| 368 |
+
"epoch": 1.19,
|
| 369 |
+
"learning_rate": 3.109452736318408e-05,
|
| 370 |
+
"loss": 1.5537,
|
| 371 |
"step": 590
|
| 372 |
},
|
| 373 |
{
|
| 374 |
+
"epoch": 1.2,
|
| 375 |
+
"learning_rate": 3.067993366500829e-05,
|
| 376 |
+
"loss": 0.5706,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 377 |
"step": 600
|
| 378 |
},
|
| 379 |
{
|
| 380 |
+
"epoch": 1.21,
|
| 381 |
+
"learning_rate": 3.0265339966832506e-05,
|
| 382 |
+
"loss": 1.7501,
|
| 383 |
"step": 610
|
| 384 |
},
|
| 385 |
{
|
| 386 |
+
"epoch": 1.21,
|
| 387 |
+
"learning_rate": 2.9850746268656714e-05,
|
| 388 |
+
"loss": 1.4373,
|
| 389 |
"step": 620
|
| 390 |
},
|
| 391 |
{
|
| 392 |
+
"epoch": 1.22,
|
| 393 |
+
"learning_rate": 2.943615257048093e-05,
|
| 394 |
+
"loss": 1.1142,
|
| 395 |
"step": 630
|
| 396 |
},
|
| 397 |
{
|
| 398 |
+
"epoch": 1.23,
|
| 399 |
+
"learning_rate": 2.9021558872305143e-05,
|
| 400 |
+
"loss": 1.596,
|
| 401 |
"step": 640
|
| 402 |
},
|
| 403 |
{
|
| 404 |
+
"epoch": 1.24,
|
| 405 |
+
"learning_rate": 2.8606965174129354e-05,
|
| 406 |
+
"loss": 1.2776,
|
| 407 |
"step": 650
|
| 408 |
},
|
| 409 |
{
|
| 410 |
+
"epoch": 1.24,
|
| 411 |
+
"learning_rate": 2.8192371475953565e-05,
|
| 412 |
+
"loss": 0.8291,
|
| 413 |
"step": 660
|
| 414 |
},
|
| 415 |
{
|
| 416 |
+
"epoch": 1.25,
|
| 417 |
+
"learning_rate": 2.777777777777778e-05,
|
| 418 |
+
"loss": 0.7873,
|
| 419 |
"step": 670
|
| 420 |
},
|
| 421 |
{
|
| 422 |
+
"epoch": 1.25,
|
| 423 |
+
"eval_accuracy": 0.5,
|
| 424 |
+
"eval_loss": 1.0956870317459106,
|
| 425 |
+
"eval_runtime": 21.8938,
|
| 426 |
+
"eval_samples_per_second": 3.38,
|
| 427 |
+
"eval_steps_per_second": 3.38,
|
| 428 |
+
"step": 670
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 2.01,
|
| 432 |
+
"learning_rate": 2.736318407960199e-05,
|
| 433 |
+
"loss": 0.4087,
|
| 434 |
"step": 680
|
| 435 |
},
|
| 436 |
{
|
| 437 |
+
"epoch": 2.01,
|
| 438 |
+
"learning_rate": 2.6948590381426204e-05,
|
| 439 |
+
"loss": 0.6448,
|
| 440 |
"step": 690
|
| 441 |
},
|
| 442 |
{
|
| 443 |
+
"epoch": 2.02,
|
| 444 |
+
"learning_rate": 2.653399668325042e-05,
|
| 445 |
+
"loss": 0.5034,
|
| 446 |
"step": 700
|
| 447 |
},
|
| 448 |
{
|
| 449 |
+
"epoch": 2.03,
|
| 450 |
+
"learning_rate": 2.6119402985074626e-05,
|
| 451 |
+
"loss": 0.2098,
|
| 452 |
"step": 710
|
| 453 |
},
|
| 454 |
{
|
| 455 |
+
"epoch": 2.04,
|
| 456 |
+
"learning_rate": 2.570480928689884e-05,
|
| 457 |
+
"loss": 0.792,
|
| 458 |
"step": 720
|
| 459 |
},
|
| 460 |
{
|
| 461 |
+
"epoch": 2.04,
|
| 462 |
+
"learning_rate": 2.5290215588723055e-05,
|
| 463 |
+
"loss": 0.0555,
|
| 464 |
"step": 730
|
| 465 |
},
|
| 466 |
{
|
| 467 |
+
"epoch": 2.05,
|
| 468 |
+
"learning_rate": 2.4875621890547266e-05,
|
| 469 |
+
"loss": 1.0835,
|
| 470 |
"step": 740
|
| 471 |
},
|
| 472 |
{
|
| 473 |
+
"epoch": 2.06,
|
| 474 |
+
"learning_rate": 2.4461028192371477e-05,
|
| 475 |
+
"loss": 0.6056,
|
| 476 |
"step": 750
|
| 477 |
},
|
| 478 |
{
|
| 479 |
+
"epoch": 2.07,
|
| 480 |
+
"learning_rate": 2.4046434494195688e-05,
|
| 481 |
+
"loss": 1.2307,
|
| 482 |
"step": 760
|
| 483 |
},
|
| 484 |
{
|
| 485 |
+
"epoch": 2.07,
|
| 486 |
+
"learning_rate": 2.3631840796019903e-05,
|
| 487 |
+
"loss": 0.0849,
|
| 488 |
"step": 770
|
| 489 |
},
|
| 490 |
{
|
| 491 |
+
"epoch": 2.08,
|
| 492 |
+
"learning_rate": 2.3217247097844114e-05,
|
| 493 |
+
"loss": 1.4995,
|
| 494 |
"step": 780
|
| 495 |
},
|
| 496 |
{
|
| 497 |
+
"epoch": 2.09,
|
| 498 |
+
"learning_rate": 2.2802653399668325e-05,
|
| 499 |
+
"loss": 0.8404,
|
| 500 |
"step": 790
|
| 501 |
},
|
| 502 |
{
|
| 503 |
+
"epoch": 2.1,
|
| 504 |
+
"learning_rate": 2.238805970149254e-05,
|
| 505 |
+
"loss": 0.7694,
|
| 506 |
"step": 800
|
| 507 |
},
|
| 508 |
{
|
| 509 |
+
"epoch": 2.1,
|
| 510 |
+
"learning_rate": 2.197346600331675e-05,
|
| 511 |
+
"loss": 0.2,
|
| 512 |
"step": 810
|
| 513 |
},
|
| 514 |
{
|
| 515 |
+
"epoch": 2.11,
|
| 516 |
+
"learning_rate": 2.155887230514096e-05,
|
| 517 |
+
"loss": 0.8996,
|
| 518 |
"step": 820
|
| 519 |
},
|
| 520 |
{
|
| 521 |
+
"epoch": 2.12,
|
| 522 |
+
"learning_rate": 2.1144278606965175e-05,
|
| 523 |
+
"loss": 1.4479,
|
| 524 |
"step": 830
|
| 525 |
},
|
| 526 |
{
|
| 527 |
+
"epoch": 2.13,
|
| 528 |
+
"learning_rate": 2.0729684908789386e-05,
|
| 529 |
+
"loss": 0.9172,
|
| 530 |
"step": 840
|
| 531 |
},
|
| 532 |
{
|
| 533 |
+
"epoch": 2.13,
|
| 534 |
+
"learning_rate": 2.0315091210613597e-05,
|
| 535 |
+
"loss": 0.3513,
|
| 536 |
"step": 850
|
| 537 |
},
|
| 538 |
{
|
| 539 |
+
"epoch": 2.14,
|
| 540 |
+
"learning_rate": 1.990049751243781e-05,
|
| 541 |
+
"loss": 0.8308,
|
| 542 |
"step": 860
|
| 543 |
},
|
| 544 |
{
|
| 545 |
+
"epoch": 2.15,
|
| 546 |
+
"learning_rate": 1.9485903814262023e-05,
|
| 547 |
+
"loss": 0.177,
|
| 548 |
"step": 870
|
| 549 |
},
|
| 550 |
{
|
| 551 |
+
"epoch": 2.16,
|
| 552 |
+
"learning_rate": 1.9071310116086237e-05,
|
| 553 |
+
"loss": 1.0408,
|
| 554 |
"step": 880
|
| 555 |
},
|
| 556 |
{
|
| 557 |
+
"epoch": 2.16,
|
| 558 |
+
"learning_rate": 1.865671641791045e-05,
|
| 559 |
+
"loss": 0.7758,
|
| 560 |
"step": 890
|
| 561 |
},
|
| 562 |
{
|
| 563 |
+
"epoch": 2.17,
|
| 564 |
+
"learning_rate": 1.8242122719734662e-05,
|
| 565 |
+
"loss": 0.5689,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 566 |
"step": 900
|
| 567 |
},
|
| 568 |
{
|
| 569 |
+
"epoch": 2.18,
|
| 570 |
+
"learning_rate": 1.7827529021558873e-05,
|
| 571 |
+
"loss": 0.176,
|
| 572 |
"step": 910
|
| 573 |
},
|
| 574 |
{
|
| 575 |
+
"epoch": 2.19,
|
| 576 |
+
"learning_rate": 1.7412935323383088e-05,
|
| 577 |
+
"loss": 0.7186,
|
| 578 |
"step": 920
|
| 579 |
},
|
| 580 |
{
|
| 581 |
+
"epoch": 2.19,
|
| 582 |
+
"learning_rate": 1.69983416252073e-05,
|
| 583 |
+
"loss": 0.8592,
|
| 584 |
"step": 930
|
| 585 |
},
|
| 586 |
{
|
| 587 |
+
"epoch": 2.2,
|
| 588 |
+
"learning_rate": 1.658374792703151e-05,
|
| 589 |
+
"loss": 0.6164,
|
| 590 |
"step": 940
|
| 591 |
},
|
| 592 |
{
|
| 593 |
+
"epoch": 2.21,
|
| 594 |
+
"learning_rate": 1.6169154228855724e-05,
|
| 595 |
+
"loss": 0.6385,
|
| 596 |
"step": 950
|
| 597 |
},
|
| 598 |
{
|
| 599 |
+
"epoch": 2.22,
|
| 600 |
+
"learning_rate": 1.5754560530679935e-05,
|
| 601 |
+
"loss": 0.5576,
|
| 602 |
"step": 960
|
| 603 |
},
|
| 604 |
{
|
| 605 |
+
"epoch": 2.22,
|
| 606 |
+
"learning_rate": 1.5339966832504146e-05,
|
| 607 |
+
"loss": 0.2168,
|
| 608 |
"step": 970
|
| 609 |
},
|
| 610 |
{
|
| 611 |
+
"epoch": 2.23,
|
| 612 |
+
"learning_rate": 1.4925373134328357e-05,
|
| 613 |
+
"loss": 0.7471,
|
| 614 |
"step": 980
|
| 615 |
},
|
| 616 |
{
|
| 617 |
+
"epoch": 2.24,
|
| 618 |
+
"learning_rate": 1.4510779436152571e-05,
|
| 619 |
+
"loss": 0.0901,
|
| 620 |
"step": 990
|
| 621 |
},
|
| 622 |
{
|
| 623 |
+
"epoch": 2.25,
|
| 624 |
+
"learning_rate": 1.4096185737976782e-05,
|
| 625 |
+
"loss": 0.7945,
|
| 626 |
"step": 1000
|
| 627 |
},
|
| 628 |
{
|
| 629 |
+
"epoch": 2.25,
|
| 630 |
+
"eval_accuracy": 0.7027027027027027,
|
| 631 |
+
"eval_loss": 1.0747544765472412,
|
| 632 |
+
"eval_runtime": 23.049,
|
| 633 |
+
"eval_samples_per_second": 3.211,
|
| 634 |
+
"eval_steps_per_second": 3.211,
|
| 635 |
+
"step": 1005
|
| 636 |
+
},
|
| 637 |
+
{
|
| 638 |
+
"epoch": 3.0,
|
| 639 |
+
"learning_rate": 1.3681592039800995e-05,
|
| 640 |
+
"loss": 0.1508,
|
| 641 |
"step": 1010
|
| 642 |
},
|
| 643 |
{
|
| 644 |
+
"epoch": 3.01,
|
| 645 |
+
"learning_rate": 1.326699834162521e-05,
|
| 646 |
+
"loss": 0.6429,
|
| 647 |
"step": 1020
|
| 648 |
},
|
| 649 |
{
|
| 650 |
+
"epoch": 3.02,
|
| 651 |
+
"learning_rate": 1.285240464344942e-05,
|
| 652 |
+
"loss": 0.6257,
|
| 653 |
"step": 1030
|
| 654 |
},
|
| 655 |
{
|
| 656 |
+
"epoch": 3.03,
|
| 657 |
+
"learning_rate": 1.2437810945273633e-05,
|
| 658 |
+
"loss": 0.2605,
|
| 659 |
"step": 1040
|
| 660 |
},
|
| 661 |
{
|
| 662 |
+
"epoch": 3.03,
|
| 663 |
+
"learning_rate": 1.2023217247097844e-05,
|
| 664 |
+
"loss": 0.5422,
|
| 665 |
"step": 1050
|
| 666 |
},
|
| 667 |
{
|
| 668 |
+
"epoch": 3.04,
|
| 669 |
+
"learning_rate": 1.1608623548922057e-05,
|
| 670 |
+
"loss": 0.1763,
|
| 671 |
"step": 1060
|
| 672 |
},
|
| 673 |
{
|
| 674 |
+
"epoch": 3.05,
|
| 675 |
+
"learning_rate": 1.119402985074627e-05,
|
| 676 |
+
"loss": 0.2139,
|
| 677 |
"step": 1070
|
| 678 |
},
|
| 679 |
{
|
| 680 |
+
"epoch": 3.06,
|
| 681 |
+
"learning_rate": 1.077943615257048e-05,
|
| 682 |
+
"loss": 1.1283,
|
| 683 |
"step": 1080
|
| 684 |
},
|
| 685 |
{
|
| 686 |
+
"epoch": 3.06,
|
| 687 |
+
"learning_rate": 1.0364842454394693e-05,
|
| 688 |
+
"loss": 0.2503,
|
| 689 |
"step": 1090
|
| 690 |
},
|
| 691 |
{
|
| 692 |
+
"epoch": 3.07,
|
| 693 |
+
"learning_rate": 9.950248756218906e-06,
|
| 694 |
+
"loss": 0.0209,
|
| 695 |
"step": 1100
|
| 696 |
},
|
| 697 |
{
|
| 698 |
+
"epoch": 3.08,
|
| 699 |
+
"learning_rate": 9.535655058043118e-06,
|
| 700 |
+
"loss": 0.286,
|
| 701 |
"step": 1110
|
| 702 |
},
|
| 703 |
{
|
| 704 |
+
"epoch": 3.09,
|
| 705 |
+
"learning_rate": 9.121061359867331e-06,
|
| 706 |
+
"loss": 0.4992,
|
| 707 |
"step": 1120
|
| 708 |
},
|
| 709 |
{
|
| 710 |
+
"epoch": 3.09,
|
| 711 |
+
"learning_rate": 8.706467661691544e-06,
|
| 712 |
+
"loss": 0.082,
|
| 713 |
"step": 1130
|
| 714 |
},
|
| 715 |
{
|
| 716 |
+
"epoch": 3.1,
|
| 717 |
+
"learning_rate": 8.291873963515755e-06,
|
| 718 |
+
"loss": 0.7033,
|
| 719 |
"step": 1140
|
| 720 |
},
|
| 721 |
{
|
| 722 |
+
"epoch": 3.11,
|
| 723 |
+
"learning_rate": 7.877280265339968e-06,
|
| 724 |
+
"loss": 0.4956,
|
| 725 |
"step": 1150
|
| 726 |
},
|
| 727 |
{
|
| 728 |
+
"epoch": 3.12,
|
| 729 |
+
"learning_rate": 7.4626865671641785e-06,
|
| 730 |
+
"loss": 0.02,
|
| 731 |
"step": 1160
|
| 732 |
},
|
| 733 |
{
|
| 734 |
+
"epoch": 3.12,
|
| 735 |
+
"learning_rate": 7.048092868988391e-06,
|
| 736 |
+
"loss": 0.0545,
|
| 737 |
"step": 1170
|
| 738 |
},
|
| 739 |
{
|
| 740 |
+
"epoch": 3.13,
|
| 741 |
+
"learning_rate": 6.633499170812605e-06,
|
| 742 |
+
"loss": 0.0159,
|
| 743 |
"step": 1180
|
| 744 |
},
|
| 745 |
{
|
| 746 |
+
"epoch": 3.14,
|
| 747 |
+
"learning_rate": 6.2189054726368165e-06,
|
| 748 |
+
"loss": 0.2031,
|
| 749 |
"step": 1190
|
| 750 |
},
|
| 751 |
+
{
|
| 752 |
+
"epoch": 3.15,
|
| 753 |
+
"learning_rate": 5.804311774461028e-06,
|
| 754 |
+
"loss": 0.208,
|
| 755 |
+
"step": 1200
|
| 756 |
+
},
|
| 757 |
+
{
|
| 758 |
+
"epoch": 3.15,
|
| 759 |
+
"learning_rate": 5.38971807628524e-06,
|
| 760 |
+
"loss": 0.4455,
|
| 761 |
+
"step": 1210
|
| 762 |
+
},
|
| 763 |
+
{
|
| 764 |
+
"epoch": 3.16,
|
| 765 |
+
"learning_rate": 4.975124378109453e-06,
|
| 766 |
+
"loss": 0.0516,
|
| 767 |
+
"step": 1220
|
| 768 |
+
},
|
| 769 |
+
{
|
| 770 |
+
"epoch": 3.17,
|
| 771 |
+
"learning_rate": 4.560530679933666e-06,
|
| 772 |
+
"loss": 0.4953,
|
| 773 |
+
"step": 1230
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 3.18,
|
| 777 |
+
"learning_rate": 4.145936981757877e-06,
|
| 778 |
+
"loss": 0.0238,
|
| 779 |
+
"step": 1240
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 3.18,
|
| 783 |
+
"learning_rate": 3.7313432835820893e-06,
|
| 784 |
+
"loss": 0.3327,
|
| 785 |
+
"step": 1250
|
| 786 |
+
},
|
| 787 |
+
{
|
| 788 |
+
"epoch": 3.19,
|
| 789 |
+
"learning_rate": 3.3167495854063024e-06,
|
| 790 |
+
"loss": 0.7936,
|
| 791 |
+
"step": 1260
|
| 792 |
+
},
|
| 793 |
+
{
|
| 794 |
+
"epoch": 3.2,
|
| 795 |
+
"learning_rate": 2.902155887230514e-06,
|
| 796 |
+
"loss": 1.226,
|
| 797 |
+
"step": 1270
|
| 798 |
+
},
|
| 799 |
+
{
|
| 800 |
+
"epoch": 3.21,
|
| 801 |
+
"learning_rate": 2.4875621890547264e-06,
|
| 802 |
+
"loss": 0.0363,
|
| 803 |
+
"step": 1280
|
| 804 |
+
},
|
| 805 |
+
{
|
| 806 |
+
"epoch": 3.21,
|
| 807 |
+
"learning_rate": 2.0729684908789387e-06,
|
| 808 |
+
"loss": 0.2792,
|
| 809 |
+
"step": 1290
|
| 810 |
+
},
|
| 811 |
+
{
|
| 812 |
+
"epoch": 3.22,
|
| 813 |
+
"learning_rate": 1.6583747927031512e-06,
|
| 814 |
+
"loss": 0.061,
|
| 815 |
+
"step": 1300
|
| 816 |
+
},
|
| 817 |
+
{
|
| 818 |
+
"epoch": 3.23,
|
| 819 |
+
"learning_rate": 1.2437810945273632e-06,
|
| 820 |
+
"loss": 0.1353,
|
| 821 |
+
"step": 1310
|
| 822 |
+
},
|
| 823 |
+
{
|
| 824 |
+
"epoch": 3.24,
|
| 825 |
+
"learning_rate": 8.291873963515756e-07,
|
| 826 |
+
"loss": 0.3527,
|
| 827 |
+
"step": 1320
|
| 828 |
+
},
|
| 829 |
+
{
|
| 830 |
+
"epoch": 3.24,
|
| 831 |
+
"learning_rate": 4.145936981757878e-07,
|
| 832 |
+
"loss": 0.0164,
|
| 833 |
+
"step": 1330
|
| 834 |
+
},
|
| 835 |
{
|
| 836 |
"epoch": 3.25,
|
| 837 |
"learning_rate": 0.0,
|
| 838 |
+
"loss": 0.0604,
|
| 839 |
+
"step": 1340
|
| 840 |
},
|
| 841 |
{
|
| 842 |
"epoch": 3.25,
|
| 843 |
+
"eval_accuracy": 0.7297297297297297,
|
| 844 |
+
"eval_loss": 0.7424588799476624,
|
| 845 |
+
"eval_runtime": 22.0218,
|
| 846 |
+
"eval_samples_per_second": 3.36,
|
| 847 |
+
"eval_steps_per_second": 3.36,
|
| 848 |
+
"step": 1340
|
| 849 |
},
|
| 850 |
{
|
| 851 |
"epoch": 3.25,
|
| 852 |
+
"step": 1340,
|
| 853 |
+
"total_flos": 1.6698605675957453e+18,
|
| 854 |
+
"train_loss": 1.1046247358308798,
|
| 855 |
+
"train_runtime": 1053.1884,
|
| 856 |
+
"train_samples_per_second": 1.272,
|
| 857 |
+
"train_steps_per_second": 1.272
|
| 858 |
},
|
| 859 |
{
|
| 860 |
"epoch": 3.25,
|
| 861 |
+
"eval_accuracy": 0.7716049382716049,
|
| 862 |
+
"eval_loss": 0.7548928260803223,
|
| 863 |
+
"eval_runtime": 52.2864,
|
| 864 |
+
"eval_samples_per_second": 3.098,
|
| 865 |
+
"eval_steps_per_second": 3.098,
|
| 866 |
+
"step": 1340
|
| 867 |
},
|
| 868 |
{
|
| 869 |
"epoch": 3.25,
|
| 870 |
+
"eval_accuracy": 0.7716049382716049,
|
| 871 |
+
"eval_loss": 0.7548928260803223,
|
| 872 |
+
"eval_runtime": 46.747,
|
| 873 |
+
"eval_samples_per_second": 3.465,
|
| 874 |
+
"eval_steps_per_second": 3.465,
|
| 875 |
+
"step": 1340
|
| 876 |
}
|
| 877 |
],
|
| 878 |
+
"max_steps": 1340,
|
| 879 |
"num_train_epochs": 9223372036854775807,
|
| 880 |
+
"total_flos": 1.6698605675957453e+18,
|
| 881 |
"trial_name": null,
|
| 882 |
"trial_params": null
|
| 883 |
}
|