End of training
Browse files
all_results.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch":
|
| 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": 10.0,
|
| 3 |
+
"eval_accuracy": 0.8324022346368715,
|
| 4 |
+
"eval_loss": 0.8124023079872131,
|
| 5 |
+
"eval_runtime": 56.7655,
|
| 6 |
+
"eval_samples_per_second": 3.153,
|
| 7 |
+
"eval_steps_per_second": 3.153
|
| 8 |
}
|
runs/Mar30_09-46-52_dd03d1f94ea0/events.out.tfevents.1680173285.dd03d1f94ea0.138.2
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:adcb4a78fe9f63f189b4674b87cd7a3d056f3d89bc8b63d9c630f5fd6322cc0c
|
| 3 |
+
size 686
|
test_results.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch":
|
| 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": 10.0,
|
| 3 |
+
"eval_accuracy": 0.8324022346368715,
|
| 4 |
+
"eval_loss": 0.8124023079872131,
|
| 5 |
+
"eval_runtime": 56.7655,
|
| 6 |
+
"eval_samples_per_second": 3.153,
|
| 7 |
+
"eval_steps_per_second": 3.153
|
| 8 |
}
|
trainer_state.json
CHANGED
|
@@ -1,2323 +1,2536 @@
|
|
| 1 |
{
|
| 2 |
-
"best_metric": 0.
|
| 3 |
-
"best_model_checkpoint": "videomae-base-finetuned-ucf101-subset/checkpoint-
|
| 4 |
-
"epoch":
|
| 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": 0.0,
|
| 12 |
-
"learning_rate": 1.
|
| 13 |
-
"loss":
|
| 14 |
"step": 10
|
| 15 |
},
|
| 16 |
{
|
| 17 |
"epoch": 0.01,
|
| 18 |
-
"learning_rate": 2.
|
| 19 |
-
"loss":
|
| 20 |
"step": 20
|
| 21 |
},
|
| 22 |
{
|
| 23 |
"epoch": 0.01,
|
| 24 |
-
"learning_rate":
|
| 25 |
-
"loss":
|
| 26 |
"step": 30
|
| 27 |
},
|
| 28 |
{
|
| 29 |
"epoch": 0.01,
|
| 30 |
-
"learning_rate": 5.
|
| 31 |
-
"loss":
|
| 32 |
"step": 40
|
| 33 |
},
|
| 34 |
{
|
| 35 |
"epoch": 0.01,
|
| 36 |
-
"learning_rate": 6.
|
| 37 |
-
"loss":
|
| 38 |
"step": 50
|
| 39 |
},
|
| 40 |
{
|
| 41 |
"epoch": 0.02,
|
| 42 |
-
"learning_rate":
|
| 43 |
-
"loss":
|
| 44 |
"step": 60
|
| 45 |
},
|
| 46 |
{
|
| 47 |
"epoch": 0.02,
|
| 48 |
-
"learning_rate":
|
| 49 |
-
"loss":
|
| 50 |
"step": 70
|
| 51 |
},
|
| 52 |
{
|
| 53 |
"epoch": 0.02,
|
| 54 |
-
"learning_rate": 1.
|
| 55 |
-
"loss":
|
| 56 |
"step": 80
|
| 57 |
},
|
| 58 |
{
|
| 59 |
"epoch": 0.02,
|
| 60 |
-
"learning_rate": 1.
|
| 61 |
-
"loss":
|
| 62 |
"step": 90
|
| 63 |
},
|
| 64 |
{
|
| 65 |
"epoch": 0.03,
|
| 66 |
-
"learning_rate": 1.
|
| 67 |
-
"loss":
|
| 68 |
"step": 100
|
| 69 |
},
|
| 70 |
{
|
| 71 |
"epoch": 0.03,
|
| 72 |
-
"learning_rate": 1.
|
| 73 |
-
"loss":
|
| 74 |
"step": 110
|
| 75 |
},
|
| 76 |
{
|
| 77 |
"epoch": 0.03,
|
| 78 |
-
"learning_rate": 1.
|
| 79 |
-
"loss":
|
| 80 |
"step": 120
|
| 81 |
},
|
| 82 |
{
|
| 83 |
-
"epoch": 0.
|
| 84 |
-
"learning_rate": 1.
|
| 85 |
-
"loss":
|
| 86 |
"step": 130
|
| 87 |
},
|
| 88 |
{
|
| 89 |
"epoch": 0.04,
|
| 90 |
-
"learning_rate": 1.
|
| 91 |
-
"loss":
|
| 92 |
"step": 140
|
| 93 |
},
|
| 94 |
{
|
| 95 |
"epoch": 0.04,
|
| 96 |
-
"learning_rate":
|
| 97 |
-
"loss":
|
| 98 |
"step": 150
|
| 99 |
},
|
| 100 |
{
|
| 101 |
"epoch": 0.04,
|
| 102 |
-
"learning_rate": 2.
|
| 103 |
-
"loss":
|
| 104 |
"step": 160
|
| 105 |
},
|
| 106 |
{
|
| 107 |
-
"epoch": 0.
|
| 108 |
-
"learning_rate": 2.
|
| 109 |
-
"loss":
|
| 110 |
"step": 170
|
| 111 |
},
|
| 112 |
{
|
| 113 |
"epoch": 0.05,
|
| 114 |
-
"learning_rate": 2.
|
| 115 |
-
"loss":
|
| 116 |
"step": 180
|
| 117 |
},
|
| 118 |
{
|
| 119 |
"epoch": 0.05,
|
| 120 |
-
"learning_rate": 2.
|
| 121 |
-
"loss":
|
| 122 |
"step": 190
|
| 123 |
},
|
| 124 |
{
|
| 125 |
"epoch": 0.05,
|
| 126 |
-
"learning_rate": 2.
|
| 127 |
-
"loss":
|
| 128 |
"step": 200
|
| 129 |
},
|
| 130 |
{
|
| 131 |
-
"epoch": 0.
|
| 132 |
-
"learning_rate": 2.
|
| 133 |
-
"loss":
|
| 134 |
"step": 210
|
| 135 |
},
|
| 136 |
{
|
| 137 |
"epoch": 0.06,
|
| 138 |
-
"learning_rate":
|
| 139 |
-
"loss":
|
| 140 |
"step": 220
|
| 141 |
},
|
| 142 |
{
|
| 143 |
"epoch": 0.06,
|
| 144 |
-
"learning_rate":
|
| 145 |
-
"loss":
|
| 146 |
"step": 230
|
| 147 |
},
|
| 148 |
{
|
| 149 |
-
"epoch": 0.
|
| 150 |
-
"learning_rate": 3.
|
| 151 |
-
"loss":
|
| 152 |
"step": 240
|
| 153 |
},
|
| 154 |
{
|
| 155 |
-
"epoch": 0.
|
| 156 |
-
"learning_rate": 3.
|
| 157 |
-
"loss":
|
| 158 |
"step": 250
|
| 159 |
},
|
| 160 |
{
|
| 161 |
"epoch": 0.07,
|
| 162 |
-
"learning_rate": 3.
|
| 163 |
-
"loss":
|
| 164 |
"step": 260
|
| 165 |
},
|
| 166 |
{
|
| 167 |
"epoch": 0.07,
|
| 168 |
-
"learning_rate": 3.
|
| 169 |
-
"loss":
|
| 170 |
"step": 270
|
| 171 |
},
|
| 172 |
{
|
| 173 |
-
"epoch": 0.
|
| 174 |
-
"learning_rate": 3.
|
| 175 |
-
"loss":
|
| 176 |
"step": 280
|
| 177 |
},
|
| 178 |
{
|
| 179 |
-
"epoch": 0.
|
| 180 |
-
"learning_rate": 3.
|
| 181 |
-
"loss":
|
| 182 |
"step": 290
|
| 183 |
},
|
| 184 |
{
|
| 185 |
"epoch": 0.08,
|
| 186 |
-
"learning_rate":
|
| 187 |
-
"loss":
|
| 188 |
"step": 300
|
| 189 |
},
|
| 190 |
{
|
| 191 |
"epoch": 0.08,
|
| 192 |
-
"learning_rate":
|
| 193 |
-
"loss": 1.
|
| 194 |
"step": 310
|
| 195 |
},
|
| 196 |
{
|
| 197 |
-
"epoch": 0.
|
| 198 |
-
"learning_rate": 4.
|
| 199 |
-
"loss":
|
| 200 |
"step": 320
|
| 201 |
},
|
| 202 |
{
|
| 203 |
-
"epoch": 0.
|
| 204 |
-
"learning_rate": 4.
|
| 205 |
-
"loss":
|
| 206 |
"step": 330
|
| 207 |
},
|
| 208 |
{
|
| 209 |
"epoch": 0.09,
|
| 210 |
-
"learning_rate": 4.
|
| 211 |
-
"loss":
|
| 212 |
"step": 340
|
| 213 |
},
|
| 214 |
{
|
| 215 |
-
"epoch": 0.
|
| 216 |
-
"learning_rate": 4.
|
| 217 |
-
"loss":
|
| 218 |
"step": 350
|
| 219 |
},
|
| 220 |
{
|
| 221 |
-
"epoch": 0.
|
| 222 |
-
"learning_rate": 4.
|
| 223 |
-
"loss":
|
| 224 |
"step": 360
|
| 225 |
},
|
| 226 |
{
|
| 227 |
-
"epoch": 0.
|
| 228 |
-
"
|
| 229 |
-
"
|
| 230 |
-
"eval_runtime": 26.1399,
|
| 231 |
-
"eval_samples_per_second": 3.711,
|
| 232 |
-
"eval_steps_per_second": 3.711,
|
| 233 |
-
"step": 365
|
| 234 |
-
},
|
| 235 |
-
{
|
| 236 |
-
"epoch": 1.0,
|
| 237 |
-
"learning_rate": 4.992389649923896e-05,
|
| 238 |
-
"loss": 1.3704,
|
| 239 |
"step": 370
|
| 240 |
},
|
| 241 |
{
|
| 242 |
-
"epoch": 1
|
| 243 |
-
"learning_rate": 4.
|
| 244 |
-
"loss":
|
| 245 |
"step": 380
|
| 246 |
},
|
| 247 |
{
|
| 248 |
-
"epoch": 1
|
| 249 |
-
"learning_rate": 4.
|
| 250 |
-
"loss":
|
| 251 |
"step": 390
|
| 252 |
},
|
| 253 |
{
|
| 254 |
-
"epoch": 1
|
| 255 |
-
"
|
| 256 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 257 |
"step": 400
|
| 258 |
},
|
| 259 |
{
|
| 260 |
-
"epoch": 1.
|
| 261 |
-
"learning_rate": 4.
|
| 262 |
-
"loss":
|
| 263 |
"step": 410
|
| 264 |
},
|
| 265 |
{
|
| 266 |
-
"epoch": 1.
|
| 267 |
-
"learning_rate": 4.
|
| 268 |
-
"loss":
|
| 269 |
"step": 420
|
| 270 |
},
|
| 271 |
{
|
| 272 |
-
"epoch": 1.
|
| 273 |
-
"learning_rate": 4.
|
| 274 |
-
"loss":
|
| 275 |
"step": 430
|
| 276 |
},
|
| 277 |
{
|
| 278 |
-
"epoch": 1.
|
| 279 |
-
"learning_rate": 4.
|
| 280 |
-
"loss":
|
| 281 |
"step": 440
|
| 282 |
},
|
| 283 |
{
|
| 284 |
-
"epoch": 1.
|
| 285 |
-
"learning_rate": 4.
|
| 286 |
-
"loss": 1.
|
| 287 |
"step": 450
|
| 288 |
},
|
| 289 |
{
|
| 290 |
-
"epoch": 1.
|
| 291 |
-
"learning_rate": 4.
|
| 292 |
-
"loss": 1.
|
| 293 |
"step": 460
|
| 294 |
},
|
| 295 |
{
|
| 296 |
-
"epoch": 1.
|
| 297 |
-
"learning_rate": 4.
|
| 298 |
-
"loss": 1.
|
| 299 |
"step": 470
|
| 300 |
},
|
| 301 |
{
|
| 302 |
-
"epoch": 1.
|
| 303 |
-
"learning_rate": 4.
|
| 304 |
-
"loss":
|
| 305 |
"step": 480
|
| 306 |
},
|
| 307 |
{
|
| 308 |
-
"epoch": 1.
|
| 309 |
-
"learning_rate": 4.
|
| 310 |
-
"loss": 1.
|
| 311 |
"step": 490
|
| 312 |
},
|
| 313 |
{
|
| 314 |
-
"epoch": 1.
|
| 315 |
-
"learning_rate": 4.
|
| 316 |
-
"loss":
|
| 317 |
"step": 500
|
| 318 |
},
|
| 319 |
{
|
| 320 |
-
"epoch": 1.
|
| 321 |
-
"learning_rate": 4.
|
| 322 |
-
"loss":
|
| 323 |
"step": 510
|
| 324 |
},
|
| 325 |
{
|
| 326 |
-
"epoch": 1.
|
| 327 |
-
"learning_rate": 4.
|
| 328 |
-
"loss":
|
| 329 |
"step": 520
|
| 330 |
},
|
| 331 |
{
|
| 332 |
-
"epoch": 1.
|
| 333 |
-
"learning_rate": 4.
|
| 334 |
-
"loss":
|
| 335 |
"step": 530
|
| 336 |
},
|
| 337 |
{
|
| 338 |
-
"epoch": 1.
|
| 339 |
-
"learning_rate": 4.
|
| 340 |
-
"loss":
|
| 341 |
"step": 540
|
| 342 |
},
|
| 343 |
{
|
| 344 |
-
"epoch": 1.
|
| 345 |
-
"learning_rate": 4.
|
| 346 |
-
"loss":
|
| 347 |
"step": 550
|
| 348 |
},
|
| 349 |
{
|
| 350 |
-
"epoch": 1.
|
| 351 |
-
"learning_rate": 4.
|
| 352 |
-
"loss":
|
| 353 |
"step": 560
|
| 354 |
},
|
| 355 |
{
|
| 356 |
-
"epoch": 1.
|
| 357 |
-
"learning_rate": 4.
|
| 358 |
-
"loss":
|
| 359 |
"step": 570
|
| 360 |
},
|
| 361 |
{
|
| 362 |
-
"epoch": 1.
|
| 363 |
-
"learning_rate": 4.
|
| 364 |
-
"loss": 1.
|
| 365 |
"step": 580
|
| 366 |
},
|
| 367 |
{
|
| 368 |
-
"epoch": 1.
|
| 369 |
-
"learning_rate": 4.
|
| 370 |
-
"loss": 1.
|
| 371 |
"step": 590
|
| 372 |
},
|
| 373 |
{
|
| 374 |
-
"epoch": 1.
|
| 375 |
-
"learning_rate": 4.
|
| 376 |
-
"loss":
|
| 377 |
"step": 600
|
| 378 |
},
|
| 379 |
{
|
| 380 |
-
"epoch": 1.
|
| 381 |
-
"learning_rate": 4.
|
| 382 |
-
"loss":
|
| 383 |
"step": 610
|
| 384 |
},
|
| 385 |
{
|
| 386 |
-
"epoch": 1.
|
| 387 |
-
"learning_rate": 4.
|
| 388 |
-
"loss":
|
| 389 |
"step": 620
|
| 390 |
},
|
| 391 |
{
|
| 392 |
-
"epoch": 1.
|
| 393 |
-
"learning_rate": 4.
|
| 394 |
-
"loss":
|
| 395 |
"step": 630
|
| 396 |
},
|
| 397 |
{
|
| 398 |
-
"epoch": 1.
|
| 399 |
-
"learning_rate": 4.
|
| 400 |
-
"loss":
|
| 401 |
"step": 640
|
| 402 |
},
|
| 403 |
{
|
| 404 |
-
"epoch": 1.
|
| 405 |
-
"learning_rate": 4.
|
| 406 |
-
"loss": 1.
|
| 407 |
"step": 650
|
| 408 |
},
|
| 409 |
{
|
| 410 |
-
"epoch": 1.
|
| 411 |
-
"learning_rate": 4.
|
| 412 |
-
"loss":
|
| 413 |
"step": 660
|
| 414 |
},
|
| 415 |
{
|
| 416 |
-
"epoch": 1.
|
| 417 |
-
"learning_rate": 4.
|
| 418 |
-
"loss": 1.
|
| 419 |
"step": 670
|
| 420 |
},
|
| 421 |
{
|
| 422 |
-
"epoch": 1.
|
| 423 |
-
"learning_rate": 4.
|
| 424 |
-
"loss": 2.
|
| 425 |
"step": 680
|
| 426 |
},
|
| 427 |
{
|
| 428 |
-
"epoch": 1.
|
| 429 |
-
"learning_rate": 4.
|
| 430 |
-
"loss":
|
| 431 |
"step": 690
|
| 432 |
},
|
| 433 |
{
|
| 434 |
-
"epoch": 1.
|
| 435 |
-
"learning_rate": 4.
|
| 436 |
-
"loss":
|
| 437 |
"step": 700
|
| 438 |
},
|
| 439 |
{
|
| 440 |
-
"epoch": 1.
|
| 441 |
-
"learning_rate": 4.
|
| 442 |
-
"loss": 0.
|
| 443 |
"step": 710
|
| 444 |
},
|
| 445 |
{
|
| 446 |
-
"epoch": 1.
|
| 447 |
-
"learning_rate": 4.
|
| 448 |
-
"loss":
|
| 449 |
"step": 720
|
| 450 |
},
|
| 451 |
{
|
| 452 |
-
"epoch": 1.
|
| 453 |
-
"learning_rate": 4.
|
| 454 |
-
"loss": 1.
|
| 455 |
-
"step": 730
|
| 456 |
-
},
|
| 457 |
-
{
|
| 458 |
-
"epoch": 1.1,
|
| 459 |
-
"eval_accuracy": 0.8041237113402062,
|
| 460 |
-
"eval_loss": 0.6252075433731079,
|
| 461 |
-
"eval_runtime": 26.3776,
|
| 462 |
-
"eval_samples_per_second": 3.677,
|
| 463 |
-
"eval_steps_per_second": 3.677,
|
| 464 |
"step": 730
|
| 465 |
},
|
| 466 |
{
|
| 467 |
-
"epoch":
|
| 468 |
-
"learning_rate": 4.
|
| 469 |
-
"loss":
|
| 470 |
"step": 740
|
| 471 |
},
|
| 472 |
{
|
| 473 |
-
"epoch":
|
| 474 |
-
"learning_rate": 4.
|
| 475 |
-
"loss": 0.
|
| 476 |
"step": 750
|
| 477 |
},
|
| 478 |
{
|
| 479 |
-
"epoch":
|
| 480 |
-
"learning_rate": 4.
|
| 481 |
-
"loss":
|
| 482 |
"step": 760
|
| 483 |
},
|
| 484 |
{
|
| 485 |
-
"epoch":
|
| 486 |
-
"learning_rate": 4.
|
| 487 |
-
"loss":
|
| 488 |
"step": 770
|
| 489 |
},
|
| 490 |
{
|
| 491 |
-
"epoch":
|
| 492 |
-
"learning_rate": 4.
|
| 493 |
-
"loss": 0.
|
| 494 |
"step": 780
|
| 495 |
},
|
| 496 |
{
|
| 497 |
-
"epoch":
|
| 498 |
-
"learning_rate": 4.
|
| 499 |
-
"loss":
|
| 500 |
"step": 790
|
| 501 |
},
|
| 502 |
{
|
| 503 |
-
"epoch":
|
| 504 |
-
"
|
| 505 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 506 |
"step": 800
|
| 507 |
},
|
| 508 |
{
|
| 509 |
-
"epoch": 2.
|
| 510 |
-
"learning_rate": 4.
|
| 511 |
-
"loss":
|
| 512 |
"step": 810
|
| 513 |
},
|
| 514 |
{
|
| 515 |
-
"epoch": 2.
|
| 516 |
-
"learning_rate": 4.
|
| 517 |
-
"loss": 1.
|
| 518 |
"step": 820
|
| 519 |
},
|
| 520 |
{
|
| 521 |
-
"epoch": 2.
|
| 522 |
-
"learning_rate": 4.
|
| 523 |
-
"loss": 0.
|
| 524 |
"step": 830
|
| 525 |
},
|
| 526 |
{
|
| 527 |
-
"epoch": 2.
|
| 528 |
-
"learning_rate": 4.
|
| 529 |
-
"loss": 0.
|
| 530 |
"step": 840
|
| 531 |
},
|
| 532 |
{
|
| 533 |
-
"epoch": 2.
|
| 534 |
-
"learning_rate": 4.
|
| 535 |
-
"loss":
|
| 536 |
"step": 850
|
| 537 |
},
|
| 538 |
{
|
| 539 |
-
"epoch": 2.
|
| 540 |
-
"learning_rate": 4.
|
| 541 |
-
"loss":
|
| 542 |
"step": 860
|
| 543 |
},
|
| 544 |
{
|
| 545 |
-
"epoch": 2.
|
| 546 |
-
"learning_rate": 4.
|
| 547 |
-
"loss":
|
| 548 |
"step": 870
|
| 549 |
},
|
| 550 |
{
|
| 551 |
-
"epoch": 2.
|
| 552 |
-
"learning_rate": 4.
|
| 553 |
-
"loss":
|
| 554 |
"step": 880
|
| 555 |
},
|
| 556 |
{
|
| 557 |
-
"epoch": 2.
|
| 558 |
-
"learning_rate": 4.
|
| 559 |
-
"loss": 0.
|
| 560 |
"step": 890
|
| 561 |
},
|
| 562 |
{
|
| 563 |
-
"epoch": 2.
|
| 564 |
-
"learning_rate": 4.
|
| 565 |
-
"loss":
|
| 566 |
"step": 900
|
| 567 |
},
|
| 568 |
{
|
| 569 |
-
"epoch": 2.
|
| 570 |
-
"learning_rate": 4.
|
| 571 |
-
"loss": 0.
|
| 572 |
"step": 910
|
| 573 |
},
|
| 574 |
{
|
| 575 |
-
"epoch": 2.
|
| 576 |
-
"learning_rate": 4.
|
| 577 |
-
"loss": 0.
|
| 578 |
"step": 920
|
| 579 |
},
|
| 580 |
{
|
| 581 |
-
"epoch": 2.
|
| 582 |
-
"learning_rate": 4.
|
| 583 |
-
"loss": 0.
|
| 584 |
"step": 930
|
| 585 |
},
|
| 586 |
{
|
| 587 |
-
"epoch": 2.
|
| 588 |
-
"learning_rate": 4.
|
| 589 |
-
"loss": 0.
|
| 590 |
"step": 940
|
| 591 |
},
|
| 592 |
{
|
| 593 |
-
"epoch": 2.
|
| 594 |
-
"learning_rate": 4.
|
| 595 |
-
"loss": 1.
|
| 596 |
"step": 950
|
| 597 |
},
|
| 598 |
{
|
| 599 |
-
"epoch": 2.
|
| 600 |
-
"learning_rate": 4.
|
| 601 |
-
"loss": 0.
|
| 602 |
"step": 960
|
| 603 |
},
|
| 604 |
{
|
| 605 |
-
"epoch": 2.
|
| 606 |
-
"learning_rate": 4.
|
| 607 |
-
"loss":
|
| 608 |
"step": 970
|
| 609 |
},
|
| 610 |
{
|
| 611 |
-
"epoch": 2.
|
| 612 |
-
"learning_rate": 4.
|
| 613 |
-
"loss":
|
| 614 |
"step": 980
|
| 615 |
},
|
| 616 |
{
|
| 617 |
-
"epoch": 2.
|
| 618 |
-
"learning_rate": 4.
|
| 619 |
-
"loss": 0.
|
| 620 |
"step": 990
|
| 621 |
},
|
| 622 |
{
|
| 623 |
-
"epoch": 2.
|
| 624 |
-
"learning_rate": 4.
|
| 625 |
-
"loss": 0.
|
| 626 |
"step": 1000
|
| 627 |
},
|
| 628 |
{
|
| 629 |
-
"epoch": 2.
|
| 630 |
-
"learning_rate": 4.
|
| 631 |
-
"loss":
|
| 632 |
"step": 1010
|
| 633 |
},
|
| 634 |
{
|
| 635 |
-
"epoch": 2.
|
| 636 |
-
"learning_rate": 4.
|
| 637 |
-
"loss":
|
| 638 |
"step": 1020
|
| 639 |
},
|
| 640 |
{
|
| 641 |
-
"epoch": 2.
|
| 642 |
-
"learning_rate":
|
| 643 |
-
"loss": 0.
|
| 644 |
"step": 1030
|
| 645 |
},
|
| 646 |
{
|
| 647 |
-
"epoch": 2.
|
| 648 |
-
"learning_rate":
|
| 649 |
-
"loss": 0.
|
| 650 |
"step": 1040
|
| 651 |
},
|
| 652 |
{
|
| 653 |
-
"epoch": 2.
|
| 654 |
-
"learning_rate":
|
| 655 |
-
"loss":
|
| 656 |
"step": 1050
|
| 657 |
},
|
| 658 |
{
|
| 659 |
-
"epoch": 2.
|
| 660 |
-
"learning_rate":
|
| 661 |
-
"loss": 0.
|
| 662 |
"step": 1060
|
| 663 |
},
|
| 664 |
{
|
| 665 |
-
"epoch": 2.
|
| 666 |
-
"learning_rate":
|
| 667 |
-
"loss": 1.
|
| 668 |
"step": 1070
|
| 669 |
},
|
| 670 |
{
|
| 671 |
-
"epoch": 2.
|
| 672 |
-
"learning_rate":
|
| 673 |
-
"loss": 0.
|
| 674 |
"step": 1080
|
| 675 |
},
|
| 676 |
{
|
| 677 |
-
"epoch": 2.
|
| 678 |
-
"learning_rate":
|
| 679 |
-
"loss": 0.
|
| 680 |
"step": 1090
|
| 681 |
},
|
| 682 |
{
|
| 683 |
-
"epoch": 2.
|
| 684 |
-
"
|
| 685 |
-
"
|
| 686 |
-
"eval_runtime": 26.1764,
|
| 687 |
-
"eval_samples_per_second": 3.706,
|
| 688 |
-
"eval_steps_per_second": 3.706,
|
| 689 |
-
"step": 1095
|
| 690 |
-
},
|
| 691 |
-
{
|
| 692 |
-
"epoch": 3.0,
|
| 693 |
-
"learning_rate": 3.881278538812785e-05,
|
| 694 |
-
"loss": 0.3452,
|
| 695 |
"step": 1100
|
| 696 |
},
|
| 697 |
{
|
| 698 |
-
"epoch":
|
| 699 |
-
"learning_rate":
|
| 700 |
-
"loss": 0.
|
| 701 |
"step": 1110
|
| 702 |
},
|
| 703 |
{
|
| 704 |
-
"epoch":
|
| 705 |
-
"learning_rate": 3.
|
| 706 |
-
"loss":
|
| 707 |
"step": 1120
|
| 708 |
},
|
| 709 |
{
|
| 710 |
-
"epoch":
|
| 711 |
-
"learning_rate": 3.
|
| 712 |
-
"loss": 0.
|
| 713 |
"step": 1130
|
| 714 |
},
|
| 715 |
{
|
| 716 |
-
"epoch":
|
| 717 |
-
"learning_rate": 3.
|
| 718 |
-
"loss":
|
| 719 |
"step": 1140
|
| 720 |
},
|
| 721 |
{
|
| 722 |
-
"epoch":
|
| 723 |
-
"learning_rate": 3.
|
| 724 |
-
"loss":
|
| 725 |
"step": 1150
|
| 726 |
},
|
| 727 |
{
|
| 728 |
-
"epoch":
|
| 729 |
-
"learning_rate": 3.
|
| 730 |
-
"loss":
|
| 731 |
"step": 1160
|
| 732 |
},
|
| 733 |
{
|
| 734 |
-
"epoch":
|
| 735 |
-
"learning_rate": 3.
|
| 736 |
-
"loss": 0.
|
| 737 |
"step": 1170
|
| 738 |
},
|
| 739 |
{
|
| 740 |
-
"epoch":
|
| 741 |
-
"learning_rate": 3.
|
| 742 |
-
"loss":
|
| 743 |
"step": 1180
|
| 744 |
},
|
| 745 |
{
|
| 746 |
-
"epoch":
|
| 747 |
-
"learning_rate": 3.
|
| 748 |
-
"loss": 0.
|
| 749 |
"step": 1190
|
| 750 |
},
|
| 751 |
{
|
| 752 |
-
"epoch":
|
| 753 |
-
"
|
| 754 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 755 |
"step": 1200
|
| 756 |
},
|
| 757 |
{
|
| 758 |
-
"epoch": 3.
|
| 759 |
-
"learning_rate": 3.
|
| 760 |
-
"loss":
|
| 761 |
"step": 1210
|
| 762 |
},
|
| 763 |
{
|
| 764 |
-
"epoch": 3.
|
| 765 |
-
"learning_rate": 3.
|
| 766 |
-
"loss": 0.
|
| 767 |
"step": 1220
|
| 768 |
},
|
| 769 |
{
|
| 770 |
-
"epoch": 3.
|
| 771 |
-
"learning_rate": 3.
|
| 772 |
-
"loss": 0.
|
| 773 |
"step": 1230
|
| 774 |
},
|
| 775 |
{
|
| 776 |
-
"epoch": 3.
|
| 777 |
-
"learning_rate": 3.
|
| 778 |
-
"loss": 0.
|
| 779 |
"step": 1240
|
| 780 |
},
|
| 781 |
{
|
| 782 |
-
"epoch": 3.
|
| 783 |
-
"learning_rate": 3.
|
| 784 |
-
"loss": 0.
|
| 785 |
"step": 1250
|
| 786 |
},
|
| 787 |
{
|
| 788 |
-
"epoch": 3.
|
| 789 |
-
"learning_rate": 3.
|
| 790 |
-
"loss": 0.
|
| 791 |
"step": 1260
|
| 792 |
},
|
| 793 |
{
|
| 794 |
-
"epoch": 3.
|
| 795 |
-
"learning_rate": 3.
|
| 796 |
-
"loss": 1.
|
| 797 |
"step": 1270
|
| 798 |
},
|
| 799 |
{
|
| 800 |
-
"epoch": 3.
|
| 801 |
-
"learning_rate": 3.
|
| 802 |
-
"loss":
|
| 803 |
"step": 1280
|
| 804 |
},
|
| 805 |
{
|
| 806 |
-
"epoch": 3.
|
| 807 |
-
"learning_rate": 3.
|
| 808 |
-
"loss": 0.
|
| 809 |
"step": 1290
|
| 810 |
},
|
| 811 |
{
|
| 812 |
-
"epoch": 3.
|
| 813 |
-
"learning_rate": 3.
|
| 814 |
-
"loss": 0.
|
| 815 |
"step": 1300
|
| 816 |
},
|
| 817 |
{
|
| 818 |
-
"epoch": 3.
|
| 819 |
-
"learning_rate": 3.
|
| 820 |
-
"loss":
|
| 821 |
"step": 1310
|
| 822 |
},
|
| 823 |
{
|
| 824 |
-
"epoch": 3.
|
| 825 |
-
"learning_rate": 3.
|
| 826 |
-
"loss": 0.
|
| 827 |
"step": 1320
|
| 828 |
},
|
| 829 |
{
|
| 830 |
-
"epoch": 3.
|
| 831 |
-
"learning_rate": 3.
|
| 832 |
-
"loss": 0.
|
| 833 |
"step": 1330
|
| 834 |
},
|
| 835 |
{
|
| 836 |
-
"epoch": 3.
|
| 837 |
-
"learning_rate": 3.
|
| 838 |
-
"loss": 0.
|
| 839 |
"step": 1340
|
| 840 |
},
|
| 841 |
{
|
| 842 |
-
"epoch": 3.
|
| 843 |
-
"learning_rate": 3.
|
| 844 |
-
"loss": 0.
|
| 845 |
"step": 1350
|
| 846 |
},
|
| 847 |
{
|
| 848 |
-
"epoch": 3.
|
| 849 |
-
"learning_rate": 3.
|
| 850 |
-
"loss": 0.
|
| 851 |
"step": 1360
|
| 852 |
},
|
| 853 |
{
|
| 854 |
-
"epoch": 3.
|
| 855 |
-
"learning_rate": 3.
|
| 856 |
-
"loss": 0.
|
| 857 |
"step": 1370
|
| 858 |
},
|
| 859 |
{
|
| 860 |
-
"epoch": 3.
|
| 861 |
-
"learning_rate": 3.
|
| 862 |
-
"loss": 0.
|
| 863 |
"step": 1380
|
| 864 |
},
|
| 865 |
{
|
| 866 |
-
"epoch": 3.
|
| 867 |
-
"learning_rate": 3.
|
| 868 |
-
"loss": 0.
|
| 869 |
"step": 1390
|
| 870 |
},
|
| 871 |
{
|
| 872 |
-
"epoch": 3.
|
| 873 |
-
"learning_rate": 3.
|
| 874 |
-
"loss": 0.
|
| 875 |
"step": 1400
|
| 876 |
},
|
| 877 |
{
|
| 878 |
-
"epoch": 3.
|
| 879 |
-
"learning_rate": 3.
|
| 880 |
-
"loss": 0.
|
| 881 |
"step": 1410
|
| 882 |
},
|
| 883 |
{
|
| 884 |
-
"epoch": 3.
|
| 885 |
-
"learning_rate": 3.
|
| 886 |
-
"loss": 0.
|
| 887 |
"step": 1420
|
| 888 |
},
|
| 889 |
{
|
| 890 |
-
"epoch": 3.
|
| 891 |
-
"learning_rate": 3.
|
| 892 |
-
"loss":
|
| 893 |
"step": 1430
|
| 894 |
},
|
| 895 |
{
|
| 896 |
-
"epoch": 3.
|
| 897 |
-
"learning_rate": 3.
|
| 898 |
-
"loss":
|
| 899 |
"step": 1440
|
| 900 |
},
|
| 901 |
{
|
| 902 |
-
"epoch": 3.
|
| 903 |
-
"learning_rate": 3.
|
| 904 |
-
"loss": 0.
|
| 905 |
"step": 1450
|
| 906 |
},
|
| 907 |
{
|
| 908 |
-
"epoch": 3.
|
| 909 |
-
"learning_rate": 3.
|
| 910 |
-
"loss":
|
| 911 |
-
"step": 1460
|
| 912 |
-
},
|
| 913 |
-
{
|
| 914 |
-
"epoch": 3.1,
|
| 915 |
-
"eval_accuracy": 0.7731958762886598,
|
| 916 |
-
"eval_loss": 0.9237529039382935,
|
| 917 |
-
"eval_runtime": 26.0231,
|
| 918 |
-
"eval_samples_per_second": 3.727,
|
| 919 |
-
"eval_steps_per_second": 3.727,
|
| 920 |
"step": 1460
|
| 921 |
},
|
| 922 |
{
|
| 923 |
-
"epoch":
|
| 924 |
-
"learning_rate": 3.
|
| 925 |
-
"loss":
|
| 926 |
"step": 1470
|
| 927 |
},
|
| 928 |
{
|
| 929 |
-
"epoch":
|
| 930 |
-
"learning_rate": 3.
|
| 931 |
-
"loss": 0.
|
| 932 |
"step": 1480
|
| 933 |
},
|
| 934 |
{
|
| 935 |
-
"epoch":
|
| 936 |
-
"learning_rate": 3.
|
| 937 |
-
"loss":
|
| 938 |
"step": 1490
|
| 939 |
},
|
| 940 |
{
|
| 941 |
-
"epoch":
|
| 942 |
-
"learning_rate": 3.
|
| 943 |
-
"loss":
|
| 944 |
"step": 1500
|
| 945 |
},
|
| 946 |
{
|
| 947 |
-
"epoch":
|
| 948 |
-
"learning_rate": 3.
|
| 949 |
-
"loss": 0.
|
| 950 |
"step": 1510
|
| 951 |
},
|
| 952 |
{
|
| 953 |
-
"epoch":
|
| 954 |
-
"learning_rate": 3.
|
| 955 |
-
"loss": 0.
|
| 956 |
"step": 1520
|
| 957 |
},
|
| 958 |
{
|
| 959 |
-
"epoch":
|
| 960 |
-
"learning_rate": 3.
|
| 961 |
-
"loss": 0.
|
| 962 |
"step": 1530
|
| 963 |
},
|
| 964 |
{
|
| 965 |
-
"epoch":
|
| 966 |
-
"learning_rate": 3.
|
| 967 |
-
"loss": 0.
|
| 968 |
"step": 1540
|
| 969 |
},
|
| 970 |
{
|
| 971 |
-
"epoch":
|
| 972 |
-
"learning_rate": 3.
|
| 973 |
-
"loss": 0.
|
| 974 |
"step": 1550
|
| 975 |
},
|
| 976 |
{
|
| 977 |
-
"epoch":
|
| 978 |
-
"learning_rate": 3.
|
| 979 |
-
"loss": 0.
|
| 980 |
"step": 1560
|
| 981 |
},
|
| 982 |
{
|
| 983 |
-
"epoch":
|
| 984 |
-
"learning_rate": 3.
|
| 985 |
-
"loss":
|
| 986 |
"step": 1570
|
| 987 |
},
|
| 988 |
{
|
| 989 |
-
"epoch":
|
| 990 |
-
"learning_rate": 3.
|
| 991 |
-
"loss": 0.
|
| 992 |
"step": 1580
|
| 993 |
},
|
| 994 |
{
|
| 995 |
-
"epoch":
|
| 996 |
-
"learning_rate": 3.
|
| 997 |
-
"loss": 0.
|
| 998 |
"step": 1590
|
| 999 |
},
|
| 1000 |
{
|
| 1001 |
-
"epoch":
|
| 1002 |
-
"
|
| 1003 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1004 |
"step": 1600
|
| 1005 |
},
|
| 1006 |
{
|
| 1007 |
-
"epoch": 4.
|
| 1008 |
-
"learning_rate": 3.
|
| 1009 |
-
"loss":
|
| 1010 |
"step": 1610
|
| 1011 |
},
|
| 1012 |
{
|
| 1013 |
-
"epoch": 4.
|
| 1014 |
-
"learning_rate": 3.
|
| 1015 |
-
"loss": 0.
|
| 1016 |
"step": 1620
|
| 1017 |
},
|
| 1018 |
{
|
| 1019 |
-
"epoch": 4.
|
| 1020 |
-
"learning_rate": 3.
|
| 1021 |
-
"loss": 0.
|
| 1022 |
"step": 1630
|
| 1023 |
},
|
| 1024 |
{
|
| 1025 |
-
"epoch": 4.
|
| 1026 |
-
"learning_rate": 3.
|
| 1027 |
-
"loss":
|
| 1028 |
"step": 1640
|
| 1029 |
},
|
| 1030 |
{
|
| 1031 |
-
"epoch": 4.
|
| 1032 |
-
"learning_rate": 3.
|
| 1033 |
-
"loss": 0.
|
| 1034 |
"step": 1650
|
| 1035 |
},
|
| 1036 |
{
|
| 1037 |
-
"epoch": 4.
|
| 1038 |
-
"learning_rate": 3.
|
| 1039 |
-
"loss": 0.
|
| 1040 |
"step": 1660
|
| 1041 |
},
|
| 1042 |
{
|
| 1043 |
-
"epoch": 4.
|
| 1044 |
-
"learning_rate": 3.
|
| 1045 |
-
"loss": 0.
|
| 1046 |
"step": 1670
|
| 1047 |
},
|
| 1048 |
{
|
| 1049 |
-
"epoch": 4.
|
| 1050 |
-
"learning_rate":
|
| 1051 |
-
"loss": 0.
|
| 1052 |
"step": 1680
|
| 1053 |
},
|
| 1054 |
{
|
| 1055 |
-
"epoch": 4.
|
| 1056 |
-
"learning_rate":
|
| 1057 |
-
"loss":
|
| 1058 |
"step": 1690
|
| 1059 |
},
|
| 1060 |
{
|
| 1061 |
-
"epoch": 4.
|
| 1062 |
-
"learning_rate":
|
| 1063 |
-
"loss": 0.
|
| 1064 |
"step": 1700
|
| 1065 |
},
|
| 1066 |
{
|
| 1067 |
-
"epoch": 4.
|
| 1068 |
-
"learning_rate":
|
| 1069 |
-
"loss": 0.
|
| 1070 |
"step": 1710
|
| 1071 |
},
|
| 1072 |
{
|
| 1073 |
-
"epoch": 4.
|
| 1074 |
-
"learning_rate":
|
| 1075 |
-
"loss": 0.
|
| 1076 |
"step": 1720
|
| 1077 |
},
|
| 1078 |
{
|
| 1079 |
-
"epoch": 4.
|
| 1080 |
-
"learning_rate":
|
| 1081 |
-
"loss": 0.
|
| 1082 |
"step": 1730
|
| 1083 |
},
|
| 1084 |
{
|
| 1085 |
-
"epoch": 4.
|
| 1086 |
-
"learning_rate":
|
| 1087 |
-
"loss":
|
| 1088 |
"step": 1740
|
| 1089 |
},
|
| 1090 |
{
|
| 1091 |
-
"epoch": 4.
|
| 1092 |
-
"learning_rate":
|
| 1093 |
-
"loss": 0.
|
| 1094 |
"step": 1750
|
| 1095 |
},
|
| 1096 |
{
|
| 1097 |
-
"epoch": 4.
|
| 1098 |
-
"learning_rate":
|
| 1099 |
-
"loss": 0.
|
| 1100 |
"step": 1760
|
| 1101 |
},
|
| 1102 |
{
|
| 1103 |
-
"epoch": 4.
|
| 1104 |
-
"learning_rate":
|
| 1105 |
-
"loss": 0.
|
| 1106 |
"step": 1770
|
| 1107 |
},
|
| 1108 |
{
|
| 1109 |
-
"epoch": 4.
|
| 1110 |
-
"learning_rate":
|
| 1111 |
-
"loss": 0.
|
| 1112 |
"step": 1780
|
| 1113 |
},
|
| 1114 |
{
|
| 1115 |
-
"epoch": 4.
|
| 1116 |
-
"learning_rate":
|
| 1117 |
-
"loss": 0.
|
| 1118 |
"step": 1790
|
| 1119 |
},
|
| 1120 |
{
|
| 1121 |
-
"epoch": 4.
|
| 1122 |
-
"learning_rate":
|
| 1123 |
-
"loss": 0.
|
| 1124 |
"step": 1800
|
| 1125 |
},
|
| 1126 |
{
|
| 1127 |
-
"epoch": 4.
|
| 1128 |
-
"learning_rate":
|
| 1129 |
-
"loss": 0.
|
| 1130 |
"step": 1810
|
| 1131 |
},
|
| 1132 |
{
|
| 1133 |
-
"epoch": 4.
|
| 1134 |
-
"learning_rate":
|
| 1135 |
-
"loss": 0.
|
| 1136 |
"step": 1820
|
| 1137 |
},
|
| 1138 |
{
|
| 1139 |
-
"epoch": 4.
|
| 1140 |
-
"
|
| 1141 |
-
"
|
| 1142 |
-
"eval_runtime": 26.3617,
|
| 1143 |
-
"eval_samples_per_second": 3.68,
|
| 1144 |
-
"eval_steps_per_second": 3.68,
|
| 1145 |
-
"step": 1825
|
| 1146 |
-
},
|
| 1147 |
-
{
|
| 1148 |
-
"epoch": 5.0,
|
| 1149 |
-
"learning_rate": 2.7701674277016743e-05,
|
| 1150 |
-
"loss": 1.0092,
|
| 1151 |
"step": 1830
|
| 1152 |
},
|
| 1153 |
{
|
| 1154 |
-
"epoch":
|
| 1155 |
-
"learning_rate": 2.
|
| 1156 |
-
"loss": 0.
|
| 1157 |
"step": 1840
|
| 1158 |
},
|
| 1159 |
{
|
| 1160 |
-
"epoch":
|
| 1161 |
-
"learning_rate": 2.
|
| 1162 |
-
"loss": 0.
|
| 1163 |
"step": 1850
|
| 1164 |
},
|
| 1165 |
{
|
| 1166 |
-
"epoch":
|
| 1167 |
-
"learning_rate": 2.
|
| 1168 |
-
"loss": 0.
|
| 1169 |
"step": 1860
|
| 1170 |
},
|
| 1171 |
{
|
| 1172 |
-
"epoch":
|
| 1173 |
-
"learning_rate": 2.
|
| 1174 |
-
"loss": 0.
|
| 1175 |
"step": 1870
|
| 1176 |
},
|
| 1177 |
{
|
| 1178 |
-
"epoch":
|
| 1179 |
-
"learning_rate": 2.
|
| 1180 |
-
"loss": 0.
|
| 1181 |
"step": 1880
|
| 1182 |
},
|
| 1183 |
{
|
| 1184 |
-
"epoch":
|
| 1185 |
-
"learning_rate": 2.
|
| 1186 |
-
"loss": 0.
|
| 1187 |
"step": 1890
|
| 1188 |
},
|
| 1189 |
{
|
| 1190 |
-
"epoch":
|
| 1191 |
-
"learning_rate": 2.
|
| 1192 |
-
"loss":
|
| 1193 |
"step": 1900
|
| 1194 |
},
|
| 1195 |
{
|
| 1196 |
-
"epoch":
|
| 1197 |
-
"learning_rate": 2.
|
| 1198 |
-
"loss": 0.
|
| 1199 |
"step": 1910
|
| 1200 |
},
|
| 1201 |
{
|
| 1202 |
-
"epoch":
|
| 1203 |
-
"learning_rate": 2.
|
| 1204 |
-
"loss": 0.
|
| 1205 |
"step": 1920
|
| 1206 |
},
|
| 1207 |
{
|
| 1208 |
-
"epoch":
|
| 1209 |
-
"learning_rate": 2.
|
| 1210 |
-
"loss": 0.
|
| 1211 |
"step": 1930
|
| 1212 |
},
|
| 1213 |
{
|
| 1214 |
-
"epoch":
|
| 1215 |
-
"learning_rate": 2.
|
| 1216 |
-
"loss":
|
| 1217 |
"step": 1940
|
| 1218 |
},
|
| 1219 |
{
|
| 1220 |
-
"epoch":
|
| 1221 |
-
"learning_rate": 2.
|
| 1222 |
-
"loss": 0.
|
| 1223 |
"step": 1950
|
| 1224 |
},
|
| 1225 |
{
|
| 1226 |
-
"epoch":
|
| 1227 |
-
"learning_rate": 2.
|
| 1228 |
-
"loss": 0.
|
| 1229 |
"step": 1960
|
| 1230 |
},
|
| 1231 |
{
|
| 1232 |
-
"epoch":
|
| 1233 |
-
"learning_rate": 2.
|
| 1234 |
-
"loss":
|
| 1235 |
"step": 1970
|
| 1236 |
},
|
| 1237 |
{
|
| 1238 |
-
"epoch":
|
| 1239 |
-
"learning_rate": 2.
|
| 1240 |
-
"loss": 0.
|
| 1241 |
"step": 1980
|
| 1242 |
},
|
| 1243 |
{
|
| 1244 |
-
"epoch":
|
| 1245 |
-
"learning_rate": 2.
|
| 1246 |
-
"loss":
|
| 1247 |
"step": 1990
|
| 1248 |
},
|
| 1249 |
{
|
| 1250 |
-
"epoch":
|
| 1251 |
-
"
|
| 1252 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1253 |
"step": 2000
|
| 1254 |
},
|
| 1255 |
{
|
| 1256 |
-
"epoch": 5.
|
| 1257 |
-
"learning_rate": 2.
|
| 1258 |
-
"loss": 0.
|
| 1259 |
"step": 2010
|
| 1260 |
},
|
| 1261 |
{
|
| 1262 |
-
"epoch": 5.
|
| 1263 |
-
"learning_rate": 2.
|
| 1264 |
-
"loss": 0.
|
| 1265 |
"step": 2020
|
| 1266 |
},
|
| 1267 |
{
|
| 1268 |
-
"epoch": 5.
|
| 1269 |
-
"learning_rate": 2.
|
| 1270 |
-
"loss": 0.
|
| 1271 |
"step": 2030
|
| 1272 |
},
|
| 1273 |
{
|
| 1274 |
-
"epoch": 5.
|
| 1275 |
-
"learning_rate": 2.
|
| 1276 |
-
"loss": 0.
|
| 1277 |
"step": 2040
|
| 1278 |
},
|
| 1279 |
{
|
| 1280 |
-
"epoch": 5.
|
| 1281 |
-
"learning_rate": 2.
|
| 1282 |
-
"loss": 0.
|
| 1283 |
"step": 2050
|
| 1284 |
},
|
| 1285 |
{
|
| 1286 |
-
"epoch": 5.
|
| 1287 |
-
"learning_rate": 2.
|
| 1288 |
-
"loss": 0.
|
| 1289 |
"step": 2060
|
| 1290 |
},
|
| 1291 |
{
|
| 1292 |
-
"epoch": 5.
|
| 1293 |
-
"learning_rate": 2.
|
| 1294 |
-
"loss": 0.
|
| 1295 |
"step": 2070
|
| 1296 |
},
|
| 1297 |
{
|
| 1298 |
-
"epoch": 5.
|
| 1299 |
-
"learning_rate": 2.
|
| 1300 |
-
"loss": 0.
|
| 1301 |
"step": 2080
|
| 1302 |
},
|
| 1303 |
{
|
| 1304 |
-
"epoch": 5.
|
| 1305 |
-
"learning_rate": 2.
|
| 1306 |
-
"loss": 0.
|
| 1307 |
"step": 2090
|
| 1308 |
},
|
| 1309 |
{
|
| 1310 |
-
"epoch": 5.
|
| 1311 |
-
"learning_rate": 2.
|
| 1312 |
-
"loss": 0.
|
| 1313 |
"step": 2100
|
| 1314 |
},
|
| 1315 |
{
|
| 1316 |
-
"epoch": 5.
|
| 1317 |
-
"learning_rate": 2.
|
| 1318 |
-
"loss": 0.
|
| 1319 |
"step": 2110
|
| 1320 |
},
|
| 1321 |
{
|
| 1322 |
-
"epoch": 5.
|
| 1323 |
-
"learning_rate": 2.
|
| 1324 |
-
"loss": 0.
|
| 1325 |
"step": 2120
|
| 1326 |
},
|
| 1327 |
{
|
| 1328 |
-
"epoch": 5.
|
| 1329 |
-
"learning_rate": 2.
|
| 1330 |
-
"loss": 0.
|
| 1331 |
"step": 2130
|
| 1332 |
},
|
| 1333 |
{
|
| 1334 |
-
"epoch": 5.
|
| 1335 |
-
"learning_rate": 2.
|
| 1336 |
-
"loss": 0.
|
| 1337 |
"step": 2140
|
| 1338 |
},
|
| 1339 |
{
|
| 1340 |
-
"epoch": 5.
|
| 1341 |
-
"learning_rate": 2.
|
| 1342 |
-
"loss": 0.
|
| 1343 |
"step": 2150
|
| 1344 |
},
|
| 1345 |
{
|
| 1346 |
-
"epoch": 5.
|
| 1347 |
-
"learning_rate": 2.
|
| 1348 |
-
"loss": 0.
|
| 1349 |
"step": 2160
|
| 1350 |
},
|
| 1351 |
{
|
| 1352 |
-
"epoch": 5.
|
| 1353 |
-
"learning_rate": 2.
|
| 1354 |
-
"loss": 0.
|
| 1355 |
"step": 2170
|
| 1356 |
},
|
| 1357 |
{
|
| 1358 |
-
"epoch": 5.
|
| 1359 |
-
"learning_rate": 2.
|
| 1360 |
-
"loss": 0.
|
| 1361 |
"step": 2180
|
| 1362 |
},
|
| 1363 |
{
|
| 1364 |
-
"epoch": 5.
|
| 1365 |
-
"learning_rate": 2.
|
| 1366 |
-
"loss": 0.
|
| 1367 |
-
"step": 2190
|
| 1368 |
-
},
|
| 1369 |
-
{
|
| 1370 |
-
"epoch": 5.1,
|
| 1371 |
-
"eval_accuracy": 0.7628865979381443,
|
| 1372 |
-
"eval_loss": 1.474695086479187,
|
| 1373 |
-
"eval_runtime": 27.4624,
|
| 1374 |
-
"eval_samples_per_second": 3.532,
|
| 1375 |
-
"eval_steps_per_second": 3.532,
|
| 1376 |
"step": 2190
|
| 1377 |
},
|
| 1378 |
{
|
| 1379 |
-
"epoch":
|
| 1380 |
-
"learning_rate": 2.
|
| 1381 |
-
"loss": 0.
|
| 1382 |
"step": 2200
|
| 1383 |
},
|
| 1384 |
{
|
| 1385 |
-
"epoch":
|
| 1386 |
-
"learning_rate": 2.
|
| 1387 |
-
"loss": 0.
|
| 1388 |
"step": 2210
|
| 1389 |
},
|
| 1390 |
{
|
| 1391 |
-
"epoch":
|
| 1392 |
-
"learning_rate": 2.
|
| 1393 |
-
"loss": 0.
|
| 1394 |
"step": 2220
|
| 1395 |
},
|
| 1396 |
{
|
| 1397 |
-
"epoch":
|
| 1398 |
-
"learning_rate": 2.
|
| 1399 |
-
"loss": 0.
|
| 1400 |
"step": 2230
|
| 1401 |
},
|
| 1402 |
{
|
| 1403 |
-
"epoch":
|
| 1404 |
-
"learning_rate": 2.
|
| 1405 |
-
"loss": 0.
|
| 1406 |
"step": 2240
|
| 1407 |
},
|
| 1408 |
{
|
| 1409 |
-
"epoch":
|
| 1410 |
-
"learning_rate": 2.
|
| 1411 |
-
"loss": 0.
|
| 1412 |
"step": 2250
|
| 1413 |
},
|
| 1414 |
{
|
| 1415 |
-
"epoch":
|
| 1416 |
-
"learning_rate": 2.
|
| 1417 |
-
"loss": 0.
|
| 1418 |
"step": 2260
|
| 1419 |
},
|
| 1420 |
{
|
| 1421 |
-
"epoch":
|
| 1422 |
-
"learning_rate": 2.
|
| 1423 |
-
"loss": 0.
|
| 1424 |
"step": 2270
|
| 1425 |
},
|
| 1426 |
{
|
| 1427 |
-
"epoch":
|
| 1428 |
-
"learning_rate": 2.
|
| 1429 |
-
"loss": 0.
|
| 1430 |
"step": 2280
|
| 1431 |
},
|
| 1432 |
{
|
| 1433 |
-
"epoch":
|
| 1434 |
-
"learning_rate": 2.
|
| 1435 |
-
"loss": 0.
|
| 1436 |
"step": 2290
|
| 1437 |
},
|
| 1438 |
{
|
| 1439 |
-
"epoch":
|
| 1440 |
-
"learning_rate": 2.
|
| 1441 |
-
"loss": 0.
|
| 1442 |
"step": 2300
|
| 1443 |
},
|
| 1444 |
{
|
| 1445 |
-
"epoch":
|
| 1446 |
-
"learning_rate": 2.
|
| 1447 |
-
"loss": 0.
|
| 1448 |
"step": 2310
|
| 1449 |
},
|
| 1450 |
{
|
| 1451 |
-
"epoch":
|
| 1452 |
-
"learning_rate": 2.
|
| 1453 |
-
"loss": 0.
|
| 1454 |
"step": 2320
|
| 1455 |
},
|
| 1456 |
{
|
| 1457 |
-
"epoch":
|
| 1458 |
-
"learning_rate": 2.
|
| 1459 |
-
"loss": 0.
|
| 1460 |
"step": 2330
|
| 1461 |
},
|
| 1462 |
{
|
| 1463 |
-
"epoch":
|
| 1464 |
-
"learning_rate":
|
| 1465 |
-
"loss": 0.
|
| 1466 |
"step": 2340
|
| 1467 |
},
|
| 1468 |
{
|
| 1469 |
-
"epoch":
|
| 1470 |
-
"learning_rate":
|
| 1471 |
-
"loss": 0.
|
| 1472 |
"step": 2350
|
| 1473 |
},
|
| 1474 |
{
|
| 1475 |
-
"epoch":
|
| 1476 |
-
"learning_rate":
|
| 1477 |
-
"loss": 0.
|
| 1478 |
"step": 2360
|
| 1479 |
},
|
| 1480 |
{
|
| 1481 |
-
"epoch":
|
| 1482 |
-
"learning_rate":
|
| 1483 |
-
"loss": 0.
|
| 1484 |
"step": 2370
|
| 1485 |
},
|
| 1486 |
{
|
| 1487 |
-
"epoch":
|
| 1488 |
-
"learning_rate":
|
| 1489 |
-
"loss": 0.
|
| 1490 |
"step": 2380
|
| 1491 |
},
|
| 1492 |
{
|
| 1493 |
-
"epoch":
|
| 1494 |
-
"
|
| 1495 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1496 |
"step": 2390
|
| 1497 |
},
|
| 1498 |
{
|
| 1499 |
-
"epoch": 6.
|
| 1500 |
-
"learning_rate":
|
| 1501 |
-
"loss": 0.
|
| 1502 |
"step": 2400
|
| 1503 |
},
|
| 1504 |
{
|
| 1505 |
-
"epoch": 6.
|
| 1506 |
-
"learning_rate":
|
| 1507 |
-
"loss": 0.
|
| 1508 |
"step": 2410
|
| 1509 |
},
|
| 1510 |
{
|
| 1511 |
-
"epoch": 6.
|
| 1512 |
-
"learning_rate":
|
| 1513 |
-
"loss": 0.
|
| 1514 |
"step": 2420
|
| 1515 |
},
|
| 1516 |
{
|
| 1517 |
-
"epoch": 6.
|
| 1518 |
-
"learning_rate":
|
| 1519 |
-
"loss": 0.
|
| 1520 |
"step": 2430
|
| 1521 |
},
|
| 1522 |
{
|
| 1523 |
-
"epoch": 6.
|
| 1524 |
-
"learning_rate":
|
| 1525 |
-
"loss": 0.
|
| 1526 |
"step": 2440
|
| 1527 |
},
|
| 1528 |
{
|
| 1529 |
-
"epoch": 6.
|
| 1530 |
-
"learning_rate":
|
| 1531 |
-
"loss": 0.
|
| 1532 |
"step": 2450
|
| 1533 |
},
|
| 1534 |
{
|
| 1535 |
-
"epoch": 6.
|
| 1536 |
-
"learning_rate":
|
| 1537 |
-
"loss": 0.
|
| 1538 |
"step": 2460
|
| 1539 |
},
|
| 1540 |
{
|
| 1541 |
-
"epoch": 6.
|
| 1542 |
-
"learning_rate":
|
| 1543 |
-
"loss": 0.
|
| 1544 |
"step": 2470
|
| 1545 |
},
|
| 1546 |
{
|
| 1547 |
-
"epoch": 6.
|
| 1548 |
-
"learning_rate":
|
| 1549 |
-
"loss": 0.
|
| 1550 |
"step": 2480
|
| 1551 |
},
|
| 1552 |
{
|
| 1553 |
-
"epoch": 6.
|
| 1554 |
-
"learning_rate":
|
| 1555 |
-
"loss": 0.
|
| 1556 |
"step": 2490
|
| 1557 |
},
|
| 1558 |
{
|
| 1559 |
-
"epoch": 6.
|
| 1560 |
-
"learning_rate":
|
| 1561 |
-
"loss": 0.
|
| 1562 |
"step": 2500
|
| 1563 |
},
|
| 1564 |
{
|
| 1565 |
-
"epoch": 6.
|
| 1566 |
-
"learning_rate":
|
| 1567 |
-
"loss": 0.
|
| 1568 |
"step": 2510
|
| 1569 |
},
|
| 1570 |
{
|
| 1571 |
-
"epoch": 6.
|
| 1572 |
-
"learning_rate":
|
| 1573 |
-
"loss": 0.
|
| 1574 |
"step": 2520
|
| 1575 |
},
|
| 1576 |
{
|
| 1577 |
-
"epoch": 6.
|
| 1578 |
-
"learning_rate":
|
| 1579 |
-
"loss": 0.
|
| 1580 |
"step": 2530
|
| 1581 |
},
|
| 1582 |
{
|
| 1583 |
-
"epoch": 6.
|
| 1584 |
-
"learning_rate":
|
| 1585 |
-
"loss": 0.
|
| 1586 |
"step": 2540
|
| 1587 |
},
|
| 1588 |
{
|
| 1589 |
-
"epoch": 6.
|
| 1590 |
-
"learning_rate":
|
| 1591 |
-
"loss": 0.
|
| 1592 |
"step": 2550
|
| 1593 |
},
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1594 |
{
|
| 1595 |
"epoch": 6.1,
|
| 1596 |
-
"eval_accuracy": 0.
|
| 1597 |
-
"eval_loss": 1.
|
| 1598 |
-
"eval_runtime":
|
| 1599 |
-
"eval_samples_per_second": 3.
|
| 1600 |
-
"eval_steps_per_second": 3.
|
| 1601 |
-
"step":
|
| 1602 |
},
|
| 1603 |
{
|
| 1604 |
"epoch": 7.0,
|
| 1605 |
-
"learning_rate": 1.
|
| 1606 |
-
"loss": 0.
|
| 1607 |
-
"step":
|
| 1608 |
},
|
| 1609 |
{
|
| 1610 |
"epoch": 7.0,
|
| 1611 |
-
"learning_rate": 1.
|
| 1612 |
-
"loss": 0.
|
| 1613 |
-
"step":
|
| 1614 |
},
|
| 1615 |
{
|
| 1616 |
"epoch": 7.01,
|
| 1617 |
-
"learning_rate": 1.
|
| 1618 |
-
"loss": 0.
|
| 1619 |
-
"step":
|
| 1620 |
},
|
| 1621 |
{
|
| 1622 |
"epoch": 7.01,
|
| 1623 |
-
"learning_rate": 1.
|
| 1624 |
-
"loss": 0.
|
| 1625 |
-
"step":
|
| 1626 |
},
|
| 1627 |
{
|
| 1628 |
"epoch": 7.01,
|
| 1629 |
-
"learning_rate": 1.
|
| 1630 |
-
"loss": 0.
|
| 1631 |
-
"step":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1632 |
},
|
| 1633 |
{
|
| 1634 |
"epoch": 7.02,
|
| 1635 |
-
"learning_rate": 1.
|
| 1636 |
-
"loss": 0.
|
| 1637 |
-
"step":
|
| 1638 |
},
|
| 1639 |
{
|
| 1640 |
"epoch": 7.02,
|
| 1641 |
-
"learning_rate": 1.
|
| 1642 |
-
"loss": 0.
|
| 1643 |
-
"step":
|
| 1644 |
},
|
| 1645 |
{
|
| 1646 |
"epoch": 7.02,
|
| 1647 |
-
"learning_rate": 1.
|
| 1648 |
-
"loss": 0.
|
| 1649 |
-
"step":
|
| 1650 |
},
|
| 1651 |
{
|
| 1652 |
"epoch": 7.02,
|
| 1653 |
-
"learning_rate": 1.
|
| 1654 |
-
"loss": 0.
|
| 1655 |
-
"step":
|
| 1656 |
},
|
| 1657 |
{
|
| 1658 |
"epoch": 7.03,
|
| 1659 |
-
"learning_rate": 1.
|
| 1660 |
-
"loss": 0.
|
| 1661 |
-
"step":
|
| 1662 |
},
|
| 1663 |
{
|
| 1664 |
"epoch": 7.03,
|
| 1665 |
-
"learning_rate": 1.
|
| 1666 |
-
"loss": 0.
|
| 1667 |
-
"step":
|
| 1668 |
},
|
| 1669 |
{
|
| 1670 |
"epoch": 7.03,
|
| 1671 |
-
"learning_rate": 1.
|
| 1672 |
-
"loss": 0.
|
| 1673 |
-
"step":
|
| 1674 |
},
|
| 1675 |
{
|
| 1676 |
"epoch": 7.03,
|
| 1677 |
-
"learning_rate": 1.
|
| 1678 |
-
"loss": 0.
|
| 1679 |
-
"step":
|
| 1680 |
},
|
| 1681 |
{
|
| 1682 |
"epoch": 7.04,
|
| 1683 |
-
"learning_rate": 1.
|
| 1684 |
-
"loss": 0.
|
| 1685 |
-
"step":
|
| 1686 |
},
|
| 1687 |
{
|
| 1688 |
"epoch": 7.04,
|
| 1689 |
-
"learning_rate": 1.
|
| 1690 |
-
"loss": 0.
|
| 1691 |
-
"step":
|
| 1692 |
},
|
| 1693 |
{
|
| 1694 |
"epoch": 7.04,
|
| 1695 |
-
"learning_rate": 1.
|
| 1696 |
-
"loss": 0.
|
| 1697 |
-
"step":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1698 |
},
|
| 1699 |
{
|
| 1700 |
"epoch": 7.05,
|
| 1701 |
-
"learning_rate": 1.
|
| 1702 |
-
"loss": 0.
|
| 1703 |
-
"step":
|
| 1704 |
},
|
| 1705 |
{
|
| 1706 |
"epoch": 7.05,
|
| 1707 |
-
"learning_rate": 1.
|
| 1708 |
-
"loss": 0.
|
| 1709 |
-
"step":
|
| 1710 |
},
|
| 1711 |
{
|
| 1712 |
"epoch": 7.05,
|
| 1713 |
-
"learning_rate": 1.
|
| 1714 |
-
"loss": 0.
|
| 1715 |
-
"step":
|
| 1716 |
},
|
| 1717 |
{
|
| 1718 |
"epoch": 7.05,
|
| 1719 |
-
"learning_rate": 1.
|
| 1720 |
-
"loss": 0.
|
| 1721 |
-
"step":
|
| 1722 |
},
|
| 1723 |
{
|
| 1724 |
"epoch": 7.06,
|
| 1725 |
-
"learning_rate": 1.
|
| 1726 |
-
"loss": 0.
|
| 1727 |
-
"step":
|
| 1728 |
},
|
| 1729 |
{
|
| 1730 |
"epoch": 7.06,
|
| 1731 |
-
"learning_rate": 1.
|
| 1732 |
-
"loss": 0.
|
| 1733 |
-
"step":
|
| 1734 |
},
|
| 1735 |
{
|
| 1736 |
"epoch": 7.06,
|
| 1737 |
-
"learning_rate": 1.
|
| 1738 |
-
"loss": 0.
|
| 1739 |
-
"step":
|
| 1740 |
},
|
| 1741 |
{
|
| 1742 |
"epoch": 7.06,
|
| 1743 |
-
"learning_rate": 1.
|
| 1744 |
-
"loss": 0.
|
| 1745 |
-
"step":
|
| 1746 |
},
|
| 1747 |
{
|
| 1748 |
"epoch": 7.07,
|
| 1749 |
-
"learning_rate": 1.
|
| 1750 |
-
"loss": 0.
|
| 1751 |
-
"step":
|
| 1752 |
},
|
| 1753 |
{
|
| 1754 |
"epoch": 7.07,
|
| 1755 |
-
"learning_rate": 1.
|
| 1756 |
-
"loss": 0.
|
| 1757 |
-
"step":
|
| 1758 |
},
|
| 1759 |
{
|
| 1760 |
"epoch": 7.07,
|
| 1761 |
-
"learning_rate": 1.
|
| 1762 |
-
"loss": 0.
|
| 1763 |
-
"step":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1764 |
},
|
| 1765 |
{
|
| 1766 |
"epoch": 7.08,
|
| 1767 |
-
"learning_rate": 1.
|
| 1768 |
-
"loss": 0.
|
| 1769 |
-
"step":
|
| 1770 |
},
|
| 1771 |
{
|
| 1772 |
"epoch": 7.08,
|
| 1773 |
-
"learning_rate": 1.
|
| 1774 |
-
"loss": 0.
|
| 1775 |
-
"step":
|
| 1776 |
},
|
| 1777 |
{
|
| 1778 |
"epoch": 7.08,
|
| 1779 |
-
"learning_rate": 1.
|
| 1780 |
-
"loss": 0.
|
| 1781 |
-
"step":
|
| 1782 |
},
|
| 1783 |
{
|
| 1784 |
"epoch": 7.08,
|
| 1785 |
-
"learning_rate": 1.
|
| 1786 |
-
"loss": 0.
|
| 1787 |
-
"step":
|
| 1788 |
},
|
| 1789 |
{
|
| 1790 |
"epoch": 7.09,
|
| 1791 |
-
"learning_rate": 1.
|
| 1792 |
-
"loss": 0.
|
| 1793 |
-
"step":
|
| 1794 |
},
|
| 1795 |
{
|
| 1796 |
"epoch": 7.09,
|
| 1797 |
-
"learning_rate": 1.
|
| 1798 |
-
"loss": 0.
|
| 1799 |
-
"step":
|
| 1800 |
},
|
| 1801 |
{
|
| 1802 |
"epoch": 7.09,
|
| 1803 |
-
"learning_rate": 1.
|
| 1804 |
-
"loss": 0.
|
| 1805 |
-
"step":
|
| 1806 |
},
|
| 1807 |
{
|
| 1808 |
"epoch": 7.09,
|
| 1809 |
-
"learning_rate": 1.
|
| 1810 |
-
"loss": 0.
|
| 1811 |
-
"step":
|
| 1812 |
},
|
| 1813 |
{
|
| 1814 |
"epoch": 7.1,
|
| 1815 |
-
"learning_rate": 1.
|
| 1816 |
-
"loss": 0.
|
| 1817 |
-
"step":
|
| 1818 |
},
|
| 1819 |
{
|
| 1820 |
"epoch": 7.1,
|
| 1821 |
-
"learning_rate": 1.
|
| 1822 |
-
"loss": 0.
|
| 1823 |
-
"step":
|
| 1824 |
},
|
| 1825 |
{
|
| 1826 |
"epoch": 7.1,
|
| 1827 |
-
"eval_accuracy": 0.
|
| 1828 |
-
"eval_loss": 1.
|
| 1829 |
-
"eval_runtime": 27.
|
| 1830 |
-
"eval_samples_per_second": 3.
|
| 1831 |
-
"eval_steps_per_second": 3.
|
| 1832 |
-
"step":
|
| 1833 |
},
|
| 1834 |
{
|
| 1835 |
"epoch": 8.0,
|
| 1836 |
-
"learning_rate": 1.
|
| 1837 |
-
"loss": 0.
|
| 1838 |
-
"step":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1839 |
},
|
| 1840 |
{
|
| 1841 |
"epoch": 8.01,
|
| 1842 |
-
"learning_rate": 1.
|
| 1843 |
"loss": 0.0016,
|
| 1844 |
-
"step":
|
| 1845 |
},
|
| 1846 |
{
|
| 1847 |
"epoch": 8.01,
|
| 1848 |
-
"learning_rate": 1.
|
| 1849 |
-
"loss": 0.
|
| 1850 |
-
"step":
|
| 1851 |
},
|
| 1852 |
{
|
| 1853 |
"epoch": 8.01,
|
| 1854 |
-
"learning_rate": 1.
|
| 1855 |
-
"loss": 0.
|
| 1856 |
-
"step":
|
| 1857 |
},
|
| 1858 |
{
|
| 1859 |
"epoch": 8.01,
|
| 1860 |
-
"learning_rate": 1.
|
| 1861 |
-
"loss": 0.
|
| 1862 |
-
"step":
|
| 1863 |
},
|
| 1864 |
{
|
| 1865 |
"epoch": 8.02,
|
| 1866 |
-
"learning_rate": 1.
|
| 1867 |
-
"loss": 0.
|
| 1868 |
-
"step":
|
| 1869 |
},
|
| 1870 |
{
|
| 1871 |
"epoch": 8.02,
|
| 1872 |
-
"learning_rate": 1.
|
| 1873 |
-
"loss": 0.
|
| 1874 |
-
"step":
|
| 1875 |
},
|
| 1876 |
{
|
| 1877 |
"epoch": 8.02,
|
| 1878 |
-
"learning_rate":
|
| 1879 |
-
"loss": 0.
|
| 1880 |
-
"step":
|
| 1881 |
},
|
| 1882 |
{
|
| 1883 |
"epoch": 8.02,
|
| 1884 |
-
"learning_rate": 9.
|
| 1885 |
-
"loss": 0.
|
| 1886 |
-
"step":
|
| 1887 |
},
|
| 1888 |
{
|
| 1889 |
"epoch": 8.03,
|
| 1890 |
-
"learning_rate": 9.
|
| 1891 |
-
"loss": 0.
|
| 1892 |
-
"step":
|
| 1893 |
},
|
| 1894 |
{
|
| 1895 |
"epoch": 8.03,
|
| 1896 |
-
"learning_rate": 9.
|
| 1897 |
-
"loss": 0.
|
| 1898 |
-
"step":
|
| 1899 |
},
|
| 1900 |
{
|
| 1901 |
"epoch": 8.03,
|
| 1902 |
-
"learning_rate": 9.
|
| 1903 |
-
"loss": 0.
|
| 1904 |
-
"step":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1905 |
},
|
| 1906 |
{
|
| 1907 |
"epoch": 8.04,
|
| 1908 |
-
"learning_rate": 9.
|
| 1909 |
-
"loss": 0.
|
| 1910 |
-
"step":
|
| 1911 |
},
|
| 1912 |
{
|
| 1913 |
"epoch": 8.04,
|
| 1914 |
-
"learning_rate":
|
| 1915 |
-
"loss": 0.
|
| 1916 |
-
"step":
|
| 1917 |
},
|
| 1918 |
{
|
| 1919 |
"epoch": 8.04,
|
| 1920 |
-
"learning_rate": 8.
|
| 1921 |
-
"loss": 0.
|
| 1922 |
-
"step":
|
| 1923 |
},
|
| 1924 |
{
|
| 1925 |
"epoch": 8.04,
|
| 1926 |
-
"learning_rate": 8.
|
| 1927 |
-
"loss": 0.
|
| 1928 |
-
"step":
|
| 1929 |
},
|
| 1930 |
{
|
| 1931 |
"epoch": 8.05,
|
| 1932 |
-
"learning_rate": 8.
|
| 1933 |
-
"loss": 0.
|
| 1934 |
-
"step":
|
| 1935 |
},
|
| 1936 |
{
|
| 1937 |
"epoch": 8.05,
|
| 1938 |
-
"learning_rate": 8.
|
| 1939 |
-
"loss": 0.
|
| 1940 |
-
"step":
|
| 1941 |
},
|
| 1942 |
{
|
| 1943 |
"epoch": 8.05,
|
| 1944 |
-
"learning_rate": 8.
|
| 1945 |
-
"loss": 0.
|
| 1946 |
-
"step":
|
| 1947 |
},
|
| 1948 |
{
|
| 1949 |
"epoch": 8.05,
|
| 1950 |
-
"learning_rate": 8.
|
| 1951 |
-
"loss": 0.
|
| 1952 |
-
"step":
|
| 1953 |
},
|
| 1954 |
{
|
| 1955 |
"epoch": 8.06,
|
| 1956 |
-
"learning_rate":
|
| 1957 |
-
"loss": 0.
|
| 1958 |
-
"step":
|
| 1959 |
},
|
| 1960 |
{
|
| 1961 |
"epoch": 8.06,
|
| 1962 |
-
"learning_rate": 7.
|
| 1963 |
-
"loss": 0.
|
| 1964 |
-
"step":
|
| 1965 |
},
|
| 1966 |
{
|
| 1967 |
"epoch": 8.06,
|
| 1968 |
-
"learning_rate": 7.
|
| 1969 |
-
"loss": 0.
|
| 1970 |
-
"step":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1971 |
},
|
| 1972 |
{
|
| 1973 |
"epoch": 8.07,
|
| 1974 |
-
"learning_rate": 7.
|
| 1975 |
-
"loss":
|
| 1976 |
-
"step":
|
| 1977 |
},
|
| 1978 |
{
|
| 1979 |
"epoch": 8.07,
|
| 1980 |
-
"learning_rate": 7.
|
| 1981 |
-
"loss": 0.
|
| 1982 |
-
"step":
|
| 1983 |
},
|
| 1984 |
{
|
| 1985 |
"epoch": 8.07,
|
| 1986 |
-
"learning_rate": 7.
|
| 1987 |
-
"loss": 0.
|
| 1988 |
-
"step":
|
| 1989 |
},
|
| 1990 |
{
|
| 1991 |
"epoch": 8.07,
|
| 1992 |
-
"learning_rate": 7.
|
| 1993 |
-
"loss": 0.
|
| 1994 |
-
"step":
|
| 1995 |
},
|
| 1996 |
{
|
| 1997 |
"epoch": 8.08,
|
| 1998 |
-
"learning_rate": 6.
|
| 1999 |
-
"loss": 0.
|
| 2000 |
-
"step":
|
| 2001 |
},
|
| 2002 |
{
|
| 2003 |
"epoch": 8.08,
|
| 2004 |
-
"learning_rate": 6.
|
| 2005 |
-
"loss": 0.
|
| 2006 |
-
"step":
|
| 2007 |
},
|
| 2008 |
{
|
| 2009 |
"epoch": 8.08,
|
| 2010 |
-
"learning_rate": 6.
|
| 2011 |
-
"loss": 0.
|
| 2012 |
-
"step":
|
| 2013 |
},
|
| 2014 |
{
|
| 2015 |
"epoch": 8.08,
|
| 2016 |
-
"learning_rate": 6.
|
| 2017 |
-
"loss": 0.
|
| 2018 |
-
"step":
|
| 2019 |
},
|
| 2020 |
{
|
| 2021 |
"epoch": 8.09,
|
| 2022 |
-
"learning_rate": 6.
|
| 2023 |
-
"loss": 0.
|
| 2024 |
-
"step":
|
| 2025 |
},
|
| 2026 |
{
|
| 2027 |
"epoch": 8.09,
|
| 2028 |
-
"learning_rate": 6.
|
| 2029 |
-
"loss": 0.
|
| 2030 |
-
"step":
|
| 2031 |
},
|
| 2032 |
{
|
| 2033 |
"epoch": 8.09,
|
| 2034 |
-
"learning_rate":
|
| 2035 |
-
"loss": 0.
|
| 2036 |
-
"step":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 2037 |
},
|
| 2038 |
{
|
| 2039 |
"epoch": 8.1,
|
| 2040 |
-
"learning_rate": 5.
|
| 2041 |
-
"loss": 0.
|
| 2042 |
-
"step":
|
| 2043 |
},
|
| 2044 |
{
|
| 2045 |
"epoch": 8.1,
|
| 2046 |
-
"learning_rate": 5.
|
| 2047 |
-
"loss": 0.
|
| 2048 |
-
"step":
|
| 2049 |
},
|
| 2050 |
{
|
| 2051 |
"epoch": 8.1,
|
| 2052 |
-
"eval_accuracy": 0.
|
| 2053 |
-
"eval_loss": 1.
|
| 2054 |
-
"eval_runtime": 28.
|
| 2055 |
-
"eval_samples_per_second": 3.
|
| 2056 |
-
"eval_steps_per_second": 3.
|
| 2057 |
-
"step":
|
| 2058 |
},
|
| 2059 |
{
|
| 2060 |
"epoch": 9.0,
|
| 2061 |
-
"learning_rate": 5.
|
| 2062 |
-
"loss": 0.
|
| 2063 |
-
"step":
|
| 2064 |
},
|
| 2065 |
{
|
| 2066 |
"epoch": 9.0,
|
| 2067 |
-
"learning_rate": 5.
|
| 2068 |
-
"loss": 0.
|
| 2069 |
-
"step":
|
| 2070 |
},
|
| 2071 |
{
|
| 2072 |
"epoch": 9.01,
|
| 2073 |
-
"learning_rate": 5.
|
| 2074 |
-
"loss": 0.
|
| 2075 |
-
"step":
|
| 2076 |
},
|
| 2077 |
{
|
| 2078 |
"epoch": 9.01,
|
| 2079 |
-
"learning_rate": 5.
|
| 2080 |
-
"loss": 0.
|
| 2081 |
-
"step":
|
| 2082 |
},
|
| 2083 |
{
|
| 2084 |
"epoch": 9.01,
|
| 2085 |
-
"learning_rate":
|
| 2086 |
-
"loss": 0.
|
| 2087 |
-
"step":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 2088 |
},
|
| 2089 |
{
|
| 2090 |
"epoch": 9.02,
|
| 2091 |
-
"learning_rate": 4.
|
| 2092 |
-
"loss": 0.
|
| 2093 |
-
"step":
|
| 2094 |
},
|
| 2095 |
{
|
| 2096 |
"epoch": 9.02,
|
| 2097 |
-
"learning_rate": 4.
|
| 2098 |
-
"loss": 0.
|
| 2099 |
-
"step":
|
| 2100 |
},
|
| 2101 |
{
|
| 2102 |
"epoch": 9.02,
|
| 2103 |
-
"learning_rate": 4.
|
| 2104 |
-
"loss": 0.
|
| 2105 |
-
"step":
|
| 2106 |
},
|
| 2107 |
{
|
| 2108 |
"epoch": 9.02,
|
| 2109 |
-
"learning_rate": 4.
|
| 2110 |
-
"loss": 0.
|
| 2111 |
-
"step":
|
| 2112 |
},
|
| 2113 |
{
|
| 2114 |
"epoch": 9.03,
|
| 2115 |
-
"learning_rate": 4.
|
| 2116 |
-
"loss": 0.
|
| 2117 |
-
"step":
|
| 2118 |
},
|
| 2119 |
{
|
| 2120 |
"epoch": 9.03,
|
| 2121 |
-
"learning_rate":
|
| 2122 |
-
"loss": 0.
|
| 2123 |
-
"step":
|
| 2124 |
},
|
| 2125 |
{
|
| 2126 |
"epoch": 9.03,
|
| 2127 |
-
"learning_rate": 3.
|
| 2128 |
-
"loss": 0.
|
| 2129 |
-
"step":
|
| 2130 |
},
|
| 2131 |
{
|
| 2132 |
"epoch": 9.03,
|
| 2133 |
-
"learning_rate": 3.
|
| 2134 |
-
"loss": 0.
|
| 2135 |
-
"step":
|
| 2136 |
},
|
| 2137 |
{
|
| 2138 |
"epoch": 9.04,
|
| 2139 |
-
"learning_rate": 3.
|
| 2140 |
-
"loss": 0.
|
| 2141 |
-
"step":
|
| 2142 |
},
|
| 2143 |
{
|
| 2144 |
"epoch": 9.04,
|
| 2145 |
-
"learning_rate": 3.
|
| 2146 |
-
"loss": 0.
|
| 2147 |
-
"step":
|
| 2148 |
},
|
| 2149 |
{
|
| 2150 |
"epoch": 9.04,
|
| 2151 |
-
"learning_rate": 3.
|
| 2152 |
-
"loss": 0.
|
| 2153 |
-
"step":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 2154 |
},
|
| 2155 |
{
|
| 2156 |
"epoch": 9.05,
|
| 2157 |
-
"learning_rate": 3.
|
| 2158 |
-
"loss": 0.
|
| 2159 |
-
"step":
|
| 2160 |
},
|
| 2161 |
{
|
| 2162 |
"epoch": 9.05,
|
| 2163 |
-
"learning_rate": 2.
|
| 2164 |
-
"loss": 0.
|
| 2165 |
-
"step":
|
| 2166 |
},
|
| 2167 |
{
|
| 2168 |
"epoch": 9.05,
|
| 2169 |
-
"learning_rate": 2.
|
| 2170 |
-
"loss": 0.
|
| 2171 |
-
"step":
|
| 2172 |
},
|
| 2173 |
{
|
| 2174 |
"epoch": 9.05,
|
| 2175 |
-
"learning_rate": 2.
|
| 2176 |
-
"loss": 0.
|
| 2177 |
-
"step":
|
| 2178 |
},
|
| 2179 |
{
|
| 2180 |
"epoch": 9.06,
|
| 2181 |
-
"learning_rate": 2.
|
| 2182 |
-
"loss": 0.
|
| 2183 |
-
"step":
|
| 2184 |
},
|
| 2185 |
{
|
| 2186 |
"epoch": 9.06,
|
| 2187 |
-
"learning_rate": 2.
|
| 2188 |
-
"loss": 0.
|
| 2189 |
-
"step":
|
| 2190 |
},
|
| 2191 |
{
|
| 2192 |
"epoch": 9.06,
|
| 2193 |
-
"learning_rate": 2.
|
| 2194 |
-
"loss": 0.
|
| 2195 |
-
"step":
|
| 2196 |
},
|
| 2197 |
{
|
| 2198 |
"epoch": 9.06,
|
| 2199 |
-
"learning_rate":
|
| 2200 |
-
"loss": 0.
|
| 2201 |
-
"step":
|
| 2202 |
},
|
| 2203 |
{
|
| 2204 |
"epoch": 9.07,
|
| 2205 |
-
"learning_rate": 1.
|
| 2206 |
-
"loss": 0.
|
| 2207 |
-
"step":
|
| 2208 |
},
|
| 2209 |
{
|
| 2210 |
"epoch": 9.07,
|
| 2211 |
-
"learning_rate": 1.
|
| 2212 |
-
"loss": 0.
|
| 2213 |
-
"step":
|
| 2214 |
},
|
| 2215 |
{
|
| 2216 |
"epoch": 9.07,
|
| 2217 |
-
"learning_rate": 1.
|
| 2218 |
-
"loss": 0.
|
| 2219 |
-
"step":
|
| 2220 |
},
|
| 2221 |
{
|
| 2222 |
-
"epoch": 9.
|
| 2223 |
-
"learning_rate": 1.
|
| 2224 |
-
"loss": 0.
|
| 2225 |
-
"step":
|
| 2226 |
},
|
| 2227 |
{
|
| 2228 |
"epoch": 9.08,
|
| 2229 |
-
"learning_rate": 1.
|
| 2230 |
-
"loss": 0.
|
| 2231 |
-
"step":
|
| 2232 |
},
|
| 2233 |
{
|
| 2234 |
"epoch": 9.08,
|
| 2235 |
-
"learning_rate": 1.
|
| 2236 |
-
"loss": 0.
|
| 2237 |
-
"step":
|
| 2238 |
},
|
| 2239 |
{
|
| 2240 |
"epoch": 9.08,
|
| 2241 |
-
"learning_rate":
|
| 2242 |
-
"loss": 0.
|
| 2243 |
-
"step":
|
| 2244 |
},
|
| 2245 |
{
|
| 2246 |
-
"epoch": 9.
|
| 2247 |
-
"learning_rate":
|
| 2248 |
-
"loss": 0.
|
| 2249 |
-
"step":
|
| 2250 |
},
|
| 2251 |
{
|
| 2252 |
"epoch": 9.09,
|
| 2253 |
-
"learning_rate":
|
| 2254 |
-
"loss": 0.
|
| 2255 |
-
"step":
|
| 2256 |
},
|
| 2257 |
{
|
| 2258 |
"epoch": 9.09,
|
| 2259 |
-
"learning_rate":
|
| 2260 |
-
"loss": 0.
|
| 2261 |
-
"step":
|
| 2262 |
},
|
| 2263 |
{
|
| 2264 |
"epoch": 9.09,
|
| 2265 |
-
"learning_rate":
|
| 2266 |
-
"loss": 0.
|
| 2267 |
-
"step":
|
| 2268 |
-
},
|
| 2269 |
-
{
|
| 2270 |
-
"epoch": 9.1,
|
| 2271 |
-
"learning_rate": 1.5220700152207e-07,
|
| 2272 |
-
"loss": 0.0004,
|
| 2273 |
-
"step": 3640
|
| 2274 |
},
|
| 2275 |
{
|
| 2276 |
-
"epoch": 9.
|
| 2277 |
-
"learning_rate":
|
| 2278 |
-
"loss": 0.
|
| 2279 |
-
"step":
|
| 2280 |
},
|
| 2281 |
{
|
| 2282 |
"epoch": 9.1,
|
| 2283 |
-
"
|
| 2284 |
-
"
|
| 2285 |
-
"
|
| 2286 |
-
"eval_samples_per_second": 3.407,
|
| 2287 |
-
"eval_steps_per_second": 3.407,
|
| 2288 |
-
"step": 3650
|
| 2289 |
},
|
| 2290 |
{
|
| 2291 |
"epoch": 9.1,
|
| 2292 |
-
"
|
| 2293 |
-
"
|
| 2294 |
-
"
|
| 2295 |
-
"train_runtime": 2930.7636,
|
| 2296 |
-
"train_samples_per_second": 1.245,
|
| 2297 |
-
"train_steps_per_second": 1.245
|
| 2298 |
},
|
| 2299 |
{
|
| 2300 |
"epoch": 9.1,
|
| 2301 |
-
"eval_accuracy": 0.
|
| 2302 |
-
"eval_loss":
|
| 2303 |
-
"eval_runtime":
|
| 2304 |
-
"eval_samples_per_second": 3.
|
| 2305 |
-
"eval_steps_per_second": 3.
|
| 2306 |
-
"step":
|
| 2307 |
},
|
| 2308 |
{
|
| 2309 |
-
"epoch":
|
| 2310 |
-
"
|
| 2311 |
-
"
|
| 2312 |
-
"
|
| 2313 |
-
|
| 2314 |
-
|
| 2315 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 2316 |
}
|
| 2317 |
],
|
| 2318 |
-
"max_steps":
|
| 2319 |
"num_train_epochs": 9223372036854775807,
|
| 2320 |
-
"total_flos": 4.
|
| 2321 |
"trial_name": null,
|
| 2322 |
"trial_params": null
|
| 2323 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_metric": 0.78,
|
| 3 |
+
"best_model_checkpoint": "videomae-base-finetuned-ucf101-subset/checkpoint-3184",
|
| 4 |
+
"epoch": 10.002506265664161,
|
| 5 |
+
"global_step": 3990,
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
| 9 |
"log_history": [
|
| 10 |
{
|
| 11 |
"epoch": 0.0,
|
| 12 |
+
"learning_rate": 1.2531328320802005e-06,
|
| 13 |
+
"loss": 2.6652,
|
| 14 |
"step": 10
|
| 15 |
},
|
| 16 |
{
|
| 17 |
"epoch": 0.01,
|
| 18 |
+
"learning_rate": 2.506265664160401e-06,
|
| 19 |
+
"loss": 2.6755,
|
| 20 |
"step": 20
|
| 21 |
},
|
| 22 |
{
|
| 23 |
"epoch": 0.01,
|
| 24 |
+
"learning_rate": 3.7593984962406014e-06,
|
| 25 |
+
"loss": 2.5836,
|
| 26 |
"step": 30
|
| 27 |
},
|
| 28 |
{
|
| 29 |
"epoch": 0.01,
|
| 30 |
+
"learning_rate": 5.012531328320802e-06,
|
| 31 |
+
"loss": 2.661,
|
| 32 |
"step": 40
|
| 33 |
},
|
| 34 |
{
|
| 35 |
"epoch": 0.01,
|
| 36 |
+
"learning_rate": 6.265664160401003e-06,
|
| 37 |
+
"loss": 2.3916,
|
| 38 |
"step": 50
|
| 39 |
},
|
| 40 |
{
|
| 41 |
"epoch": 0.02,
|
| 42 |
+
"learning_rate": 7.518796992481203e-06,
|
| 43 |
+
"loss": 2.5649,
|
| 44 |
"step": 60
|
| 45 |
},
|
| 46 |
{
|
| 47 |
"epoch": 0.02,
|
| 48 |
+
"learning_rate": 8.771929824561403e-06,
|
| 49 |
+
"loss": 2.6852,
|
| 50 |
"step": 70
|
| 51 |
},
|
| 52 |
{
|
| 53 |
"epoch": 0.02,
|
| 54 |
+
"learning_rate": 1.0025062656641604e-05,
|
| 55 |
+
"loss": 2.6172,
|
| 56 |
"step": 80
|
| 57 |
},
|
| 58 |
{
|
| 59 |
"epoch": 0.02,
|
| 60 |
+
"learning_rate": 1.1278195488721805e-05,
|
| 61 |
+
"loss": 2.5301,
|
| 62 |
"step": 90
|
| 63 |
},
|
| 64 |
{
|
| 65 |
"epoch": 0.03,
|
| 66 |
+
"learning_rate": 1.2531328320802006e-05,
|
| 67 |
+
"loss": 2.6607,
|
| 68 |
"step": 100
|
| 69 |
},
|
| 70 |
{
|
| 71 |
"epoch": 0.03,
|
| 72 |
+
"learning_rate": 1.3784461152882205e-05,
|
| 73 |
+
"loss": 2.4796,
|
| 74 |
"step": 110
|
| 75 |
},
|
| 76 |
{
|
| 77 |
"epoch": 0.03,
|
| 78 |
+
"learning_rate": 1.5037593984962406e-05,
|
| 79 |
+
"loss": 2.4986,
|
| 80 |
"step": 120
|
| 81 |
},
|
| 82 |
{
|
| 83 |
+
"epoch": 0.03,
|
| 84 |
+
"learning_rate": 1.6290726817042605e-05,
|
| 85 |
+
"loss": 2.5192,
|
| 86 |
"step": 130
|
| 87 |
},
|
| 88 |
{
|
| 89 |
"epoch": 0.04,
|
| 90 |
+
"learning_rate": 1.7543859649122806e-05,
|
| 91 |
+
"loss": 2.615,
|
| 92 |
"step": 140
|
| 93 |
},
|
| 94 |
{
|
| 95 |
"epoch": 0.04,
|
| 96 |
+
"learning_rate": 1.8796992481203007e-05,
|
| 97 |
+
"loss": 2.5708,
|
| 98 |
"step": 150
|
| 99 |
},
|
| 100 |
{
|
| 101 |
"epoch": 0.04,
|
| 102 |
+
"learning_rate": 2.0050125313283208e-05,
|
| 103 |
+
"loss": 2.3817,
|
| 104 |
"step": 160
|
| 105 |
},
|
| 106 |
{
|
| 107 |
+
"epoch": 0.04,
|
| 108 |
+
"learning_rate": 2.130325814536341e-05,
|
| 109 |
+
"loss": 2.4658,
|
| 110 |
"step": 170
|
| 111 |
},
|
| 112 |
{
|
| 113 |
"epoch": 0.05,
|
| 114 |
+
"learning_rate": 2.255639097744361e-05,
|
| 115 |
+
"loss": 2.5414,
|
| 116 |
"step": 180
|
| 117 |
},
|
| 118 |
{
|
| 119 |
"epoch": 0.05,
|
| 120 |
+
"learning_rate": 2.380952380952381e-05,
|
| 121 |
+
"loss": 2.6049,
|
| 122 |
"step": 190
|
| 123 |
},
|
| 124 |
{
|
| 125 |
"epoch": 0.05,
|
| 126 |
+
"learning_rate": 2.506265664160401e-05,
|
| 127 |
+
"loss": 2.5113,
|
| 128 |
"step": 200
|
| 129 |
},
|
| 130 |
{
|
| 131 |
+
"epoch": 0.05,
|
| 132 |
+
"learning_rate": 2.6315789473684212e-05,
|
| 133 |
+
"loss": 2.6226,
|
| 134 |
"step": 210
|
| 135 |
},
|
| 136 |
{
|
| 137 |
"epoch": 0.06,
|
| 138 |
+
"learning_rate": 2.756892230576441e-05,
|
| 139 |
+
"loss": 2.4038,
|
| 140 |
"step": 220
|
| 141 |
},
|
| 142 |
{
|
| 143 |
"epoch": 0.06,
|
| 144 |
+
"learning_rate": 2.882205513784461e-05,
|
| 145 |
+
"loss": 2.2845,
|
| 146 |
"step": 230
|
| 147 |
},
|
| 148 |
{
|
| 149 |
+
"epoch": 0.06,
|
| 150 |
+
"learning_rate": 3.007518796992481e-05,
|
| 151 |
+
"loss": 2.3279,
|
| 152 |
"step": 240
|
| 153 |
},
|
| 154 |
{
|
| 155 |
+
"epoch": 0.06,
|
| 156 |
+
"learning_rate": 3.132832080200501e-05,
|
| 157 |
+
"loss": 2.3244,
|
| 158 |
"step": 250
|
| 159 |
},
|
| 160 |
{
|
| 161 |
"epoch": 0.07,
|
| 162 |
+
"learning_rate": 3.258145363408521e-05,
|
| 163 |
+
"loss": 2.2399,
|
| 164 |
"step": 260
|
| 165 |
},
|
| 166 |
{
|
| 167 |
"epoch": 0.07,
|
| 168 |
+
"learning_rate": 3.3834586466165414e-05,
|
| 169 |
+
"loss": 2.5944,
|
| 170 |
"step": 270
|
| 171 |
},
|
| 172 |
{
|
| 173 |
+
"epoch": 0.07,
|
| 174 |
+
"learning_rate": 3.508771929824561e-05,
|
| 175 |
+
"loss": 2.3696,
|
| 176 |
"step": 280
|
| 177 |
},
|
| 178 |
{
|
| 179 |
+
"epoch": 0.07,
|
| 180 |
+
"learning_rate": 3.6340852130325816e-05,
|
| 181 |
+
"loss": 2.3556,
|
| 182 |
"step": 290
|
| 183 |
},
|
| 184 |
{
|
| 185 |
"epoch": 0.08,
|
| 186 |
+
"learning_rate": 3.759398496240601e-05,
|
| 187 |
+
"loss": 2.1208,
|
| 188 |
"step": 300
|
| 189 |
},
|
| 190 |
{
|
| 191 |
"epoch": 0.08,
|
| 192 |
+
"learning_rate": 3.884711779448622e-05,
|
| 193 |
+
"loss": 1.9662,
|
| 194 |
"step": 310
|
| 195 |
},
|
| 196 |
{
|
| 197 |
+
"epoch": 0.08,
|
| 198 |
+
"learning_rate": 4.0100250626566415e-05,
|
| 199 |
+
"loss": 2.3956,
|
| 200 |
"step": 320
|
| 201 |
},
|
| 202 |
{
|
| 203 |
+
"epoch": 0.08,
|
| 204 |
+
"learning_rate": 4.135338345864662e-05,
|
| 205 |
+
"loss": 2.3429,
|
| 206 |
"step": 330
|
| 207 |
},
|
| 208 |
{
|
| 209 |
"epoch": 0.09,
|
| 210 |
+
"learning_rate": 4.260651629072682e-05,
|
| 211 |
+
"loss": 2.3078,
|
| 212 |
"step": 340
|
| 213 |
},
|
| 214 |
{
|
| 215 |
+
"epoch": 0.09,
|
| 216 |
+
"learning_rate": 4.3859649122807014e-05,
|
| 217 |
+
"loss": 2.1788,
|
| 218 |
"step": 350
|
| 219 |
},
|
| 220 |
{
|
| 221 |
+
"epoch": 0.09,
|
| 222 |
+
"learning_rate": 4.511278195488722e-05,
|
| 223 |
+
"loss": 2.2873,
|
| 224 |
"step": 360
|
| 225 |
},
|
| 226 |
{
|
| 227 |
+
"epoch": 0.09,
|
| 228 |
+
"learning_rate": 4.6365914786967416e-05,
|
| 229 |
+
"loss": 2.4288,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 230 |
"step": 370
|
| 231 |
},
|
| 232 |
{
|
| 233 |
+
"epoch": 0.1,
|
| 234 |
+
"learning_rate": 4.761904761904762e-05,
|
| 235 |
+
"loss": 1.7902,
|
| 236 |
"step": 380
|
| 237 |
},
|
| 238 |
{
|
| 239 |
+
"epoch": 0.1,
|
| 240 |
+
"learning_rate": 4.887218045112782e-05,
|
| 241 |
+
"loss": 2.1012,
|
| 242 |
"step": 390
|
| 243 |
},
|
| 244 |
{
|
| 245 |
+
"epoch": 0.1,
|
| 246 |
+
"eval_accuracy": 0.38,
|
| 247 |
+
"eval_loss": 1.9809015989303589,
|
| 248 |
+
"eval_runtime": 27.6095,
|
| 249 |
+
"eval_samples_per_second": 3.622,
|
| 250 |
+
"eval_steps_per_second": 3.622,
|
| 251 |
+
"step": 398
|
| 252 |
+
},
|
| 253 |
+
{
|
| 254 |
+
"epoch": 1.0,
|
| 255 |
+
"learning_rate": 4.998607630186578e-05,
|
| 256 |
+
"loss": 1.6446,
|
| 257 |
"step": 400
|
| 258 |
},
|
| 259 |
{
|
| 260 |
+
"epoch": 1.0,
|
| 261 |
+
"learning_rate": 4.9846839320523534e-05,
|
| 262 |
+
"loss": 2.495,
|
| 263 |
"step": 410
|
| 264 |
},
|
| 265 |
{
|
| 266 |
+
"epoch": 1.01,
|
| 267 |
+
"learning_rate": 4.970760233918128e-05,
|
| 268 |
+
"loss": 2.1706,
|
| 269 |
"step": 420
|
| 270 |
},
|
| 271 |
{
|
| 272 |
+
"epoch": 1.01,
|
| 273 |
+
"learning_rate": 4.956836535783904e-05,
|
| 274 |
+
"loss": 1.6511,
|
| 275 |
"step": 430
|
| 276 |
},
|
| 277 |
{
|
| 278 |
+
"epoch": 1.01,
|
| 279 |
+
"learning_rate": 4.9429128376496795e-05,
|
| 280 |
+
"loss": 2.0323,
|
| 281 |
"step": 440
|
| 282 |
},
|
| 283 |
{
|
| 284 |
+
"epoch": 1.01,
|
| 285 |
+
"learning_rate": 4.928989139515455e-05,
|
| 286 |
+
"loss": 1.8594,
|
| 287 |
"step": 450
|
| 288 |
},
|
| 289 |
{
|
| 290 |
+
"epoch": 1.02,
|
| 291 |
+
"learning_rate": 4.9150654413812306e-05,
|
| 292 |
+
"loss": 1.6434,
|
| 293 |
"step": 460
|
| 294 |
},
|
| 295 |
{
|
| 296 |
+
"epoch": 1.02,
|
| 297 |
+
"learning_rate": 4.901141743247006e-05,
|
| 298 |
+
"loss": 1.5981,
|
| 299 |
"step": 470
|
| 300 |
},
|
| 301 |
{
|
| 302 |
+
"epoch": 1.02,
|
| 303 |
+
"learning_rate": 4.887218045112782e-05,
|
| 304 |
+
"loss": 2.3124,
|
| 305 |
"step": 480
|
| 306 |
},
|
| 307 |
{
|
| 308 |
+
"epoch": 1.02,
|
| 309 |
+
"learning_rate": 4.8732943469785574e-05,
|
| 310 |
+
"loss": 1.3398,
|
| 311 |
"step": 490
|
| 312 |
},
|
| 313 |
{
|
| 314 |
+
"epoch": 1.03,
|
| 315 |
+
"learning_rate": 4.859370648844333e-05,
|
| 316 |
+
"loss": 2.075,
|
| 317 |
"step": 500
|
| 318 |
},
|
| 319 |
{
|
| 320 |
+
"epoch": 1.03,
|
| 321 |
+
"learning_rate": 4.8454469507101085e-05,
|
| 322 |
+
"loss": 2.1273,
|
| 323 |
"step": 510
|
| 324 |
},
|
| 325 |
{
|
| 326 |
+
"epoch": 1.03,
|
| 327 |
+
"learning_rate": 4.831523252575884e-05,
|
| 328 |
+
"loss": 1.4408,
|
| 329 |
"step": 520
|
| 330 |
},
|
| 331 |
{
|
| 332 |
+
"epoch": 1.03,
|
| 333 |
+
"learning_rate": 4.81759955444166e-05,
|
| 334 |
+
"loss": 1.8349,
|
| 335 |
"step": 530
|
| 336 |
},
|
| 337 |
{
|
| 338 |
+
"epoch": 1.04,
|
| 339 |
+
"learning_rate": 4.803675856307435e-05,
|
| 340 |
+
"loss": 1.4592,
|
| 341 |
"step": 540
|
| 342 |
},
|
| 343 |
{
|
| 344 |
+
"epoch": 1.04,
|
| 345 |
+
"learning_rate": 4.789752158173211e-05,
|
| 346 |
+
"loss": 2.0761,
|
| 347 |
"step": 550
|
| 348 |
},
|
| 349 |
{
|
| 350 |
+
"epoch": 1.04,
|
| 351 |
+
"learning_rate": 4.7758284600389865e-05,
|
| 352 |
+
"loss": 1.6495,
|
| 353 |
"step": 560
|
| 354 |
},
|
| 355 |
{
|
| 356 |
+
"epoch": 1.04,
|
| 357 |
+
"learning_rate": 4.761904761904762e-05,
|
| 358 |
+
"loss": 1.0583,
|
| 359 |
"step": 570
|
| 360 |
},
|
| 361 |
{
|
| 362 |
+
"epoch": 1.05,
|
| 363 |
+
"learning_rate": 4.7479810637705376e-05,
|
| 364 |
+
"loss": 1.4962,
|
| 365 |
"step": 580
|
| 366 |
},
|
| 367 |
{
|
| 368 |
+
"epoch": 1.05,
|
| 369 |
+
"learning_rate": 4.734057365636313e-05,
|
| 370 |
+
"loss": 1.065,
|
| 371 |
"step": 590
|
| 372 |
},
|
| 373 |
{
|
| 374 |
+
"epoch": 1.05,
|
| 375 |
+
"learning_rate": 4.720133667502089e-05,
|
| 376 |
+
"loss": 2.0059,
|
| 377 |
"step": 600
|
| 378 |
},
|
| 379 |
{
|
| 380 |
+
"epoch": 1.05,
|
| 381 |
+
"learning_rate": 4.7062099693678644e-05,
|
| 382 |
+
"loss": 1.8712,
|
| 383 |
"step": 610
|
| 384 |
},
|
| 385 |
{
|
| 386 |
+
"epoch": 1.06,
|
| 387 |
+
"learning_rate": 4.69228627123364e-05,
|
| 388 |
+
"loss": 1.9663,
|
| 389 |
"step": 620
|
| 390 |
},
|
| 391 |
{
|
| 392 |
+
"epoch": 1.06,
|
| 393 |
+
"learning_rate": 4.678362573099415e-05,
|
| 394 |
+
"loss": 2.1448,
|
| 395 |
"step": 630
|
| 396 |
},
|
| 397 |
{
|
| 398 |
+
"epoch": 1.06,
|
| 399 |
+
"learning_rate": 4.6644388749651904e-05,
|
| 400 |
+
"loss": 1.5725,
|
| 401 |
"step": 640
|
| 402 |
},
|
| 403 |
{
|
| 404 |
+
"epoch": 1.06,
|
| 405 |
+
"learning_rate": 4.650515176830966e-05,
|
| 406 |
+
"loss": 1.6991,
|
| 407 |
"step": 650
|
| 408 |
},
|
| 409 |
{
|
| 410 |
+
"epoch": 1.07,
|
| 411 |
+
"learning_rate": 4.6365914786967416e-05,
|
| 412 |
+
"loss": 1.291,
|
| 413 |
"step": 660
|
| 414 |
},
|
| 415 |
{
|
| 416 |
+
"epoch": 1.07,
|
| 417 |
+
"learning_rate": 4.622667780562517e-05,
|
| 418 |
+
"loss": 1.8409,
|
| 419 |
"step": 670
|
| 420 |
},
|
| 421 |
{
|
| 422 |
+
"epoch": 1.07,
|
| 423 |
+
"learning_rate": 4.608744082428293e-05,
|
| 424 |
+
"loss": 2.2183,
|
| 425 |
"step": 680
|
| 426 |
},
|
| 427 |
{
|
| 428 |
+
"epoch": 1.07,
|
| 429 |
+
"learning_rate": 4.5948203842940684e-05,
|
| 430 |
+
"loss": 1.6034,
|
| 431 |
"step": 690
|
| 432 |
},
|
| 433 |
{
|
| 434 |
+
"epoch": 1.08,
|
| 435 |
+
"learning_rate": 4.580896686159844e-05,
|
| 436 |
+
"loss": 1.4481,
|
| 437 |
"step": 700
|
| 438 |
},
|
| 439 |
{
|
| 440 |
+
"epoch": 1.08,
|
| 441 |
+
"learning_rate": 4.5669729880256195e-05,
|
| 442 |
+
"loss": 0.9722,
|
| 443 |
"step": 710
|
| 444 |
},
|
| 445 |
{
|
| 446 |
+
"epoch": 1.08,
|
| 447 |
+
"learning_rate": 4.553049289891395e-05,
|
| 448 |
+
"loss": 1.447,
|
| 449 |
"step": 720
|
| 450 |
},
|
| 451 |
{
|
| 452 |
+
"epoch": 1.08,
|
| 453 |
+
"learning_rate": 4.539125591757171e-05,
|
| 454 |
+
"loss": 1.6108,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 455 |
"step": 730
|
| 456 |
},
|
| 457 |
{
|
| 458 |
+
"epoch": 1.09,
|
| 459 |
+
"learning_rate": 4.525201893622946e-05,
|
| 460 |
+
"loss": 2.3506,
|
| 461 |
"step": 740
|
| 462 |
},
|
| 463 |
{
|
| 464 |
+
"epoch": 1.09,
|
| 465 |
+
"learning_rate": 4.511278195488722e-05,
|
| 466 |
+
"loss": 0.9662,
|
| 467 |
"step": 750
|
| 468 |
},
|
| 469 |
{
|
| 470 |
+
"epoch": 1.09,
|
| 471 |
+
"learning_rate": 4.4973544973544974e-05,
|
| 472 |
+
"loss": 1.3062,
|
| 473 |
"step": 760
|
| 474 |
},
|
| 475 |
{
|
| 476 |
+
"epoch": 1.09,
|
| 477 |
+
"learning_rate": 4.483430799220273e-05,
|
| 478 |
+
"loss": 1.3071,
|
| 479 |
"step": 770
|
| 480 |
},
|
| 481 |
{
|
| 482 |
+
"epoch": 1.1,
|
| 483 |
+
"learning_rate": 4.4695071010860486e-05,
|
| 484 |
+
"loss": 0.9615,
|
| 485 |
"step": 780
|
| 486 |
},
|
| 487 |
{
|
| 488 |
+
"epoch": 1.1,
|
| 489 |
+
"learning_rate": 4.455583402951824e-05,
|
| 490 |
+
"loss": 1.0416,
|
| 491 |
"step": 790
|
| 492 |
},
|
| 493 |
{
|
| 494 |
+
"epoch": 1.1,
|
| 495 |
+
"eval_accuracy": 0.56,
|
| 496 |
+
"eval_loss": 1.6140196323394775,
|
| 497 |
+
"eval_runtime": 27.3792,
|
| 498 |
+
"eval_samples_per_second": 3.652,
|
| 499 |
+
"eval_steps_per_second": 3.652,
|
| 500 |
+
"step": 796
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 2.0,
|
| 504 |
+
"learning_rate": 4.4416597048176e-05,
|
| 505 |
+
"loss": 0.567,
|
| 506 |
"step": 800
|
| 507 |
},
|
| 508 |
{
|
| 509 |
+
"epoch": 2.0,
|
| 510 |
+
"learning_rate": 4.4277360066833754e-05,
|
| 511 |
+
"loss": 1.2888,
|
| 512 |
"step": 810
|
| 513 |
},
|
| 514 |
{
|
| 515 |
+
"epoch": 2.01,
|
| 516 |
+
"learning_rate": 4.413812308549151e-05,
|
| 517 |
+
"loss": 1.2286,
|
| 518 |
"step": 820
|
| 519 |
},
|
| 520 |
{
|
| 521 |
+
"epoch": 2.01,
|
| 522 |
+
"learning_rate": 4.399888610414926e-05,
|
| 523 |
+
"loss": 0.7935,
|
| 524 |
"step": 830
|
| 525 |
},
|
| 526 |
{
|
| 527 |
+
"epoch": 2.01,
|
| 528 |
+
"learning_rate": 4.3859649122807014e-05,
|
| 529 |
+
"loss": 0.9961,
|
| 530 |
"step": 840
|
| 531 |
},
|
| 532 |
{
|
| 533 |
+
"epoch": 2.01,
|
| 534 |
+
"learning_rate": 4.372041214146477e-05,
|
| 535 |
+
"loss": 1.3997,
|
| 536 |
"step": 850
|
| 537 |
},
|
| 538 |
{
|
| 539 |
+
"epoch": 2.02,
|
| 540 |
+
"learning_rate": 4.3581175160122526e-05,
|
| 541 |
+
"loss": 1.112,
|
| 542 |
"step": 860
|
| 543 |
},
|
| 544 |
{
|
| 545 |
+
"epoch": 2.02,
|
| 546 |
+
"learning_rate": 4.344193817878028e-05,
|
| 547 |
+
"loss": 0.3837,
|
| 548 |
"step": 870
|
| 549 |
},
|
| 550 |
{
|
| 551 |
+
"epoch": 2.02,
|
| 552 |
+
"learning_rate": 4.330270119743804e-05,
|
| 553 |
+
"loss": 2.3791,
|
| 554 |
"step": 880
|
| 555 |
},
|
| 556 |
{
|
| 557 |
+
"epoch": 2.02,
|
| 558 |
+
"learning_rate": 4.3163464216095793e-05,
|
| 559 |
+
"loss": 0.7274,
|
| 560 |
"step": 890
|
| 561 |
},
|
| 562 |
{
|
| 563 |
+
"epoch": 2.03,
|
| 564 |
+
"learning_rate": 4.302422723475355e-05,
|
| 565 |
+
"loss": 1.2903,
|
| 566 |
"step": 900
|
| 567 |
},
|
| 568 |
{
|
| 569 |
+
"epoch": 2.03,
|
| 570 |
+
"learning_rate": 4.2884990253411305e-05,
|
| 571 |
+
"loss": 0.3864,
|
| 572 |
"step": 910
|
| 573 |
},
|
| 574 |
{
|
| 575 |
+
"epoch": 2.03,
|
| 576 |
+
"learning_rate": 4.274575327206906e-05,
|
| 577 |
+
"loss": 0.8519,
|
| 578 |
"step": 920
|
| 579 |
},
|
| 580 |
{
|
| 581 |
+
"epoch": 2.03,
|
| 582 |
+
"learning_rate": 4.260651629072682e-05,
|
| 583 |
+
"loss": 0.2129,
|
| 584 |
"step": 930
|
| 585 |
},
|
| 586 |
{
|
| 587 |
+
"epoch": 2.04,
|
| 588 |
+
"learning_rate": 4.246727930938457e-05,
|
| 589 |
+
"loss": 0.9185,
|
| 590 |
"step": 940
|
| 591 |
},
|
| 592 |
{
|
| 593 |
+
"epoch": 2.04,
|
| 594 |
+
"learning_rate": 4.232804232804233e-05,
|
| 595 |
+
"loss": 1.5575,
|
| 596 |
"step": 950
|
| 597 |
},
|
| 598 |
{
|
| 599 |
+
"epoch": 2.04,
|
| 600 |
+
"learning_rate": 4.2188805346700084e-05,
|
| 601 |
+
"loss": 0.3068,
|
| 602 |
"step": 960
|
| 603 |
},
|
| 604 |
{
|
| 605 |
+
"epoch": 2.04,
|
| 606 |
+
"learning_rate": 4.204956836535784e-05,
|
| 607 |
+
"loss": 0.8577,
|
| 608 |
"step": 970
|
| 609 |
},
|
| 610 |
{
|
| 611 |
+
"epoch": 2.05,
|
| 612 |
+
"learning_rate": 4.1910331384015596e-05,
|
| 613 |
+
"loss": 1.2818,
|
| 614 |
"step": 980
|
| 615 |
},
|
| 616 |
{
|
| 617 |
+
"epoch": 2.05,
|
| 618 |
+
"learning_rate": 4.177109440267335e-05,
|
| 619 |
+
"loss": 0.2316,
|
| 620 |
"step": 990
|
| 621 |
},
|
| 622 |
{
|
| 623 |
+
"epoch": 2.05,
|
| 624 |
+
"learning_rate": 4.163185742133111e-05,
|
| 625 |
+
"loss": 0.9911,
|
| 626 |
"step": 1000
|
| 627 |
},
|
| 628 |
{
|
| 629 |
+
"epoch": 2.05,
|
| 630 |
+
"learning_rate": 4.1492620439988863e-05,
|
| 631 |
+
"loss": 1.0393,
|
| 632 |
"step": 1010
|
| 633 |
},
|
| 634 |
{
|
| 635 |
+
"epoch": 2.06,
|
| 636 |
+
"learning_rate": 4.135338345864662e-05,
|
| 637 |
+
"loss": 1.7585,
|
| 638 |
"step": 1020
|
| 639 |
},
|
| 640 |
{
|
| 641 |
+
"epoch": 2.06,
|
| 642 |
+
"learning_rate": 4.121414647730437e-05,
|
| 643 |
+
"loss": 0.6806,
|
| 644 |
"step": 1030
|
| 645 |
},
|
| 646 |
{
|
| 647 |
+
"epoch": 2.06,
|
| 648 |
+
"learning_rate": 4.1074909495962124e-05,
|
| 649 |
+
"loss": 0.3257,
|
| 650 |
"step": 1040
|
| 651 |
},
|
| 652 |
{
|
| 653 |
+
"epoch": 2.06,
|
| 654 |
+
"learning_rate": 4.093567251461988e-05,
|
| 655 |
+
"loss": 1.249,
|
| 656 |
"step": 1050
|
| 657 |
},
|
| 658 |
{
|
| 659 |
+
"epoch": 2.07,
|
| 660 |
+
"learning_rate": 4.0796435533277636e-05,
|
| 661 |
+
"loss": 0.5324,
|
| 662 |
"step": 1060
|
| 663 |
},
|
| 664 |
{
|
| 665 |
+
"epoch": 2.07,
|
| 666 |
+
"learning_rate": 4.065719855193539e-05,
|
| 667 |
+
"loss": 1.3123,
|
| 668 |
"step": 1070
|
| 669 |
},
|
| 670 |
{
|
| 671 |
+
"epoch": 2.07,
|
| 672 |
+
"learning_rate": 4.051796157059315e-05,
|
| 673 |
+
"loss": 0.285,
|
| 674 |
"step": 1080
|
| 675 |
},
|
| 676 |
{
|
| 677 |
+
"epoch": 2.07,
|
| 678 |
+
"learning_rate": 4.03787245892509e-05,
|
| 679 |
+
"loss": 0.5411,
|
| 680 |
"step": 1090
|
| 681 |
},
|
| 682 |
{
|
| 683 |
+
"epoch": 2.08,
|
| 684 |
+
"learning_rate": 4.023948760790866e-05,
|
| 685 |
+
"loss": 1.7209,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 686 |
"step": 1100
|
| 687 |
},
|
| 688 |
{
|
| 689 |
+
"epoch": 2.08,
|
| 690 |
+
"learning_rate": 4.0100250626566415e-05,
|
| 691 |
+
"loss": 0.5554,
|
| 692 |
"step": 1110
|
| 693 |
},
|
| 694 |
{
|
| 695 |
+
"epoch": 2.08,
|
| 696 |
+
"learning_rate": 3.996101364522417e-05,
|
| 697 |
+
"loss": 1.8234,
|
| 698 |
"step": 1120
|
| 699 |
},
|
| 700 |
{
|
| 701 |
+
"epoch": 2.08,
|
| 702 |
+
"learning_rate": 3.982177666388193e-05,
|
| 703 |
+
"loss": 0.9143,
|
| 704 |
"step": 1130
|
| 705 |
},
|
| 706 |
{
|
| 707 |
+
"epoch": 2.09,
|
| 708 |
+
"learning_rate": 3.968253968253968e-05,
|
| 709 |
+
"loss": 1.6595,
|
| 710 |
"step": 1140
|
| 711 |
},
|
| 712 |
{
|
| 713 |
+
"epoch": 2.09,
|
| 714 |
+
"learning_rate": 3.954330270119744e-05,
|
| 715 |
+
"loss": 1.0291,
|
| 716 |
"step": 1150
|
| 717 |
},
|
| 718 |
{
|
| 719 |
+
"epoch": 2.09,
|
| 720 |
+
"learning_rate": 3.9404065719855194e-05,
|
| 721 |
+
"loss": 1.24,
|
| 722 |
"step": 1160
|
| 723 |
},
|
| 724 |
{
|
| 725 |
+
"epoch": 2.09,
|
| 726 |
+
"learning_rate": 3.926482873851295e-05,
|
| 727 |
+
"loss": 0.849,
|
| 728 |
"step": 1170
|
| 729 |
},
|
| 730 |
{
|
| 731 |
+
"epoch": 2.1,
|
| 732 |
+
"learning_rate": 3.9125591757170706e-05,
|
| 733 |
+
"loss": 1.5756,
|
| 734 |
"step": 1180
|
| 735 |
},
|
| 736 |
{
|
| 737 |
+
"epoch": 2.1,
|
| 738 |
+
"learning_rate": 3.898635477582846e-05,
|
| 739 |
+
"loss": 0.2096,
|
| 740 |
"step": 1190
|
| 741 |
},
|
| 742 |
{
|
| 743 |
+
"epoch": 2.1,
|
| 744 |
+
"eval_accuracy": 0.66,
|
| 745 |
+
"eval_loss": 1.5775662660598755,
|
| 746 |
+
"eval_runtime": 28.2107,
|
| 747 |
+
"eval_samples_per_second": 3.545,
|
| 748 |
+
"eval_steps_per_second": 3.545,
|
| 749 |
+
"step": 1194
|
| 750 |
+
},
|
| 751 |
+
{
|
| 752 |
+
"epoch": 3.0,
|
| 753 |
+
"learning_rate": 3.884711779448622e-05,
|
| 754 |
+
"loss": 1.3493,
|
| 755 |
"step": 1200
|
| 756 |
},
|
| 757 |
{
|
| 758 |
+
"epoch": 3.0,
|
| 759 |
+
"learning_rate": 3.870788081314397e-05,
|
| 760 |
+
"loss": 1.3687,
|
| 761 |
"step": 1210
|
| 762 |
},
|
| 763 |
{
|
| 764 |
+
"epoch": 3.01,
|
| 765 |
+
"learning_rate": 3.856864383180173e-05,
|
| 766 |
+
"loss": 0.8319,
|
| 767 |
"step": 1220
|
| 768 |
},
|
| 769 |
{
|
| 770 |
+
"epoch": 3.01,
|
| 771 |
+
"learning_rate": 3.8429406850459485e-05,
|
| 772 |
+
"loss": 0.9219,
|
| 773 |
"step": 1230
|
| 774 |
},
|
| 775 |
{
|
| 776 |
+
"epoch": 3.01,
|
| 777 |
+
"learning_rate": 3.8290169869117234e-05,
|
| 778 |
+
"loss": 0.3956,
|
| 779 |
"step": 1240
|
| 780 |
},
|
| 781 |
{
|
| 782 |
+
"epoch": 3.01,
|
| 783 |
+
"learning_rate": 3.815093288777499e-05,
|
| 784 |
+
"loss": 0.6158,
|
| 785 |
"step": 1250
|
| 786 |
},
|
| 787 |
{
|
| 788 |
+
"epoch": 3.02,
|
| 789 |
+
"learning_rate": 3.8011695906432746e-05,
|
| 790 |
+
"loss": 0.8011,
|
| 791 |
"step": 1260
|
| 792 |
},
|
| 793 |
{
|
| 794 |
+
"epoch": 3.02,
|
| 795 |
+
"learning_rate": 3.78724589250905e-05,
|
| 796 |
+
"loss": 1.3677,
|
| 797 |
"step": 1270
|
| 798 |
},
|
| 799 |
{
|
| 800 |
+
"epoch": 3.02,
|
| 801 |
+
"learning_rate": 3.773322194374826e-05,
|
| 802 |
+
"loss": 1.5107,
|
| 803 |
"step": 1280
|
| 804 |
},
|
| 805 |
{
|
| 806 |
+
"epoch": 3.02,
|
| 807 |
+
"learning_rate": 3.759398496240601e-05,
|
| 808 |
+
"loss": 0.1896,
|
| 809 |
"step": 1290
|
| 810 |
},
|
| 811 |
{
|
| 812 |
+
"epoch": 3.03,
|
| 813 |
+
"learning_rate": 3.745474798106377e-05,
|
| 814 |
+
"loss": 0.6759,
|
| 815 |
"step": 1300
|
| 816 |
},
|
| 817 |
{
|
| 818 |
+
"epoch": 3.03,
|
| 819 |
+
"learning_rate": 3.7315510999721525e-05,
|
| 820 |
+
"loss": 0.0941,
|
| 821 |
"step": 1310
|
| 822 |
},
|
| 823 |
{
|
| 824 |
+
"epoch": 3.03,
|
| 825 |
+
"learning_rate": 3.717627401837928e-05,
|
| 826 |
+
"loss": 0.7546,
|
| 827 |
"step": 1320
|
| 828 |
},
|
| 829 |
{
|
| 830 |
+
"epoch": 3.03,
|
| 831 |
+
"learning_rate": 3.7037037037037037e-05,
|
| 832 |
+
"loss": 0.6543,
|
| 833 |
"step": 1330
|
| 834 |
},
|
| 835 |
{
|
| 836 |
+
"epoch": 3.04,
|
| 837 |
+
"learning_rate": 3.689780005569479e-05,
|
| 838 |
+
"loss": 0.5026,
|
| 839 |
"step": 1340
|
| 840 |
},
|
| 841 |
{
|
| 842 |
+
"epoch": 3.04,
|
| 843 |
+
"learning_rate": 3.675856307435255e-05,
|
| 844 |
+
"loss": 0.376,
|
| 845 |
"step": 1350
|
| 846 |
},
|
| 847 |
{
|
| 848 |
+
"epoch": 3.04,
|
| 849 |
+
"learning_rate": 3.6619326093010304e-05,
|
| 850 |
+
"loss": 0.9097,
|
| 851 |
"step": 1360
|
| 852 |
},
|
| 853 |
{
|
| 854 |
+
"epoch": 3.04,
|
| 855 |
+
"learning_rate": 3.648008911166806e-05,
|
| 856 |
+
"loss": 0.5431,
|
| 857 |
"step": 1370
|
| 858 |
},
|
| 859 |
{
|
| 860 |
+
"epoch": 3.05,
|
| 861 |
+
"learning_rate": 3.6340852130325816e-05,
|
| 862 |
+
"loss": 0.6503,
|
| 863 |
"step": 1380
|
| 864 |
},
|
| 865 |
{
|
| 866 |
+
"epoch": 3.05,
|
| 867 |
+
"learning_rate": 3.620161514898357e-05,
|
| 868 |
+
"loss": 0.6169,
|
| 869 |
"step": 1390
|
| 870 |
},
|
| 871 |
{
|
| 872 |
+
"epoch": 3.05,
|
| 873 |
+
"learning_rate": 3.606237816764133e-05,
|
| 874 |
+
"loss": 0.3653,
|
| 875 |
"step": 1400
|
| 876 |
},
|
| 877 |
{
|
| 878 |
+
"epoch": 3.05,
|
| 879 |
+
"learning_rate": 3.592314118629908e-05,
|
| 880 |
+
"loss": 0.1474,
|
| 881 |
"step": 1410
|
| 882 |
},
|
| 883 |
{
|
| 884 |
+
"epoch": 3.06,
|
| 885 |
+
"learning_rate": 3.578390420495684e-05,
|
| 886 |
+
"loss": 0.9832,
|
| 887 |
"step": 1420
|
| 888 |
},
|
| 889 |
{
|
| 890 |
+
"epoch": 3.06,
|
| 891 |
+
"learning_rate": 3.5644667223614595e-05,
|
| 892 |
+
"loss": 0.3245,
|
| 893 |
"step": 1430
|
| 894 |
},
|
| 895 |
{
|
| 896 |
+
"epoch": 3.06,
|
| 897 |
+
"learning_rate": 3.5505430242272344e-05,
|
| 898 |
+
"loss": 1.3136,
|
| 899 |
"step": 1440
|
| 900 |
},
|
| 901 |
{
|
| 902 |
+
"epoch": 3.06,
|
| 903 |
+
"learning_rate": 3.53661932609301e-05,
|
| 904 |
+
"loss": 0.8991,
|
| 905 |
"step": 1450
|
| 906 |
},
|
| 907 |
{
|
| 908 |
+
"epoch": 3.07,
|
| 909 |
+
"learning_rate": 3.5226956279587856e-05,
|
| 910 |
+
"loss": 1.8971,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 911 |
"step": 1460
|
| 912 |
},
|
| 913 |
{
|
| 914 |
+
"epoch": 3.07,
|
| 915 |
+
"learning_rate": 3.508771929824561e-05,
|
| 916 |
+
"loss": 1.1886,
|
| 917 |
"step": 1470
|
| 918 |
},
|
| 919 |
{
|
| 920 |
+
"epoch": 3.07,
|
| 921 |
+
"learning_rate": 3.494848231690337e-05,
|
| 922 |
+
"loss": 0.6224,
|
| 923 |
"step": 1480
|
| 924 |
},
|
| 925 |
{
|
| 926 |
+
"epoch": 3.07,
|
| 927 |
+
"learning_rate": 3.480924533556112e-05,
|
| 928 |
+
"loss": 0.1107,
|
| 929 |
"step": 1490
|
| 930 |
},
|
| 931 |
{
|
| 932 |
+
"epoch": 3.08,
|
| 933 |
+
"learning_rate": 3.467000835421888e-05,
|
| 934 |
+
"loss": 2.0252,
|
| 935 |
"step": 1500
|
| 936 |
},
|
| 937 |
{
|
| 938 |
+
"epoch": 3.08,
|
| 939 |
+
"learning_rate": 3.4530771372876635e-05,
|
| 940 |
+
"loss": 0.5866,
|
| 941 |
"step": 1510
|
| 942 |
},
|
| 943 |
{
|
| 944 |
+
"epoch": 3.08,
|
| 945 |
+
"learning_rate": 3.439153439153439e-05,
|
| 946 |
+
"loss": 0.5512,
|
| 947 |
"step": 1520
|
| 948 |
},
|
| 949 |
{
|
| 950 |
+
"epoch": 3.08,
|
| 951 |
+
"learning_rate": 3.4252297410192146e-05,
|
| 952 |
+
"loss": 0.5085,
|
| 953 |
"step": 1530
|
| 954 |
},
|
| 955 |
{
|
| 956 |
+
"epoch": 3.09,
|
| 957 |
+
"learning_rate": 3.41130604288499e-05,
|
| 958 |
+
"loss": 0.0272,
|
| 959 |
"step": 1540
|
| 960 |
},
|
| 961 |
{
|
| 962 |
+
"epoch": 3.09,
|
| 963 |
+
"learning_rate": 3.397382344750766e-05,
|
| 964 |
+
"loss": 0.0063,
|
| 965 |
"step": 1550
|
| 966 |
},
|
| 967 |
{
|
| 968 |
+
"epoch": 3.09,
|
| 969 |
+
"learning_rate": 3.3834586466165414e-05,
|
| 970 |
+
"loss": 0.1569,
|
| 971 |
"step": 1560
|
| 972 |
},
|
| 973 |
{
|
| 974 |
+
"epoch": 3.09,
|
| 975 |
+
"learning_rate": 3.369534948482317e-05,
|
| 976 |
+
"loss": 1.3151,
|
| 977 |
"step": 1570
|
| 978 |
},
|
| 979 |
{
|
| 980 |
+
"epoch": 3.1,
|
| 981 |
+
"learning_rate": 3.3556112503480926e-05,
|
| 982 |
+
"loss": 0.0256,
|
| 983 |
"step": 1580
|
| 984 |
},
|
| 985 |
{
|
| 986 |
+
"epoch": 3.1,
|
| 987 |
+
"learning_rate": 3.341687552213868e-05,
|
| 988 |
+
"loss": 0.7101,
|
| 989 |
"step": 1590
|
| 990 |
},
|
| 991 |
{
|
| 992 |
+
"epoch": 3.1,
|
| 993 |
+
"eval_accuracy": 0.74,
|
| 994 |
+
"eval_loss": 1.2003577947616577,
|
| 995 |
+
"eval_runtime": 28.5623,
|
| 996 |
+
"eval_samples_per_second": 3.501,
|
| 997 |
+
"eval_steps_per_second": 3.501,
|
| 998 |
+
"step": 1592
|
| 999 |
+
},
|
| 1000 |
+
{
|
| 1001 |
+
"epoch": 4.0,
|
| 1002 |
+
"learning_rate": 3.327763854079644e-05,
|
| 1003 |
+
"loss": 0.0158,
|
| 1004 |
"step": 1600
|
| 1005 |
},
|
| 1006 |
{
|
| 1007 |
+
"epoch": 4.0,
|
| 1008 |
+
"learning_rate": 3.313840155945419e-05,
|
| 1009 |
+
"loss": 1.1735,
|
| 1010 |
"step": 1610
|
| 1011 |
},
|
| 1012 |
{
|
| 1013 |
+
"epoch": 4.01,
|
| 1014 |
+
"learning_rate": 3.299916457811195e-05,
|
| 1015 |
+
"loss": 0.0063,
|
| 1016 |
"step": 1620
|
| 1017 |
},
|
| 1018 |
{
|
| 1019 |
+
"epoch": 4.01,
|
| 1020 |
+
"learning_rate": 3.2859927596769705e-05,
|
| 1021 |
+
"loss": 0.2802,
|
| 1022 |
"step": 1630
|
| 1023 |
},
|
| 1024 |
{
|
| 1025 |
+
"epoch": 4.01,
|
| 1026 |
+
"learning_rate": 3.272069061542746e-05,
|
| 1027 |
+
"loss": 1.0417,
|
| 1028 |
"step": 1640
|
| 1029 |
},
|
| 1030 |
{
|
| 1031 |
+
"epoch": 4.01,
|
| 1032 |
+
"learning_rate": 3.258145363408521e-05,
|
| 1033 |
+
"loss": 0.2255,
|
| 1034 |
"step": 1650
|
| 1035 |
},
|
| 1036 |
{
|
| 1037 |
+
"epoch": 4.02,
|
| 1038 |
+
"learning_rate": 3.2442216652742965e-05,
|
| 1039 |
+
"loss": 0.2863,
|
| 1040 |
"step": 1660
|
| 1041 |
},
|
| 1042 |
{
|
| 1043 |
+
"epoch": 4.02,
|
| 1044 |
+
"learning_rate": 3.230297967140072e-05,
|
| 1045 |
+
"loss": 0.0177,
|
| 1046 |
"step": 1670
|
| 1047 |
},
|
| 1048 |
{
|
| 1049 |
+
"epoch": 4.02,
|
| 1050 |
+
"learning_rate": 3.216374269005848e-05,
|
| 1051 |
+
"loss": 0.0132,
|
| 1052 |
"step": 1680
|
| 1053 |
},
|
| 1054 |
{
|
| 1055 |
+
"epoch": 4.02,
|
| 1056 |
+
"learning_rate": 3.202450570871623e-05,
|
| 1057 |
+
"loss": 1.1547,
|
| 1058 |
"step": 1690
|
| 1059 |
},
|
| 1060 |
{
|
| 1061 |
+
"epoch": 4.03,
|
| 1062 |
+
"learning_rate": 3.188526872737399e-05,
|
| 1063 |
+
"loss": 0.4432,
|
| 1064 |
"step": 1700
|
| 1065 |
},
|
| 1066 |
{
|
| 1067 |
+
"epoch": 4.03,
|
| 1068 |
+
"learning_rate": 3.1746031746031745e-05,
|
| 1069 |
+
"loss": 0.4861,
|
| 1070 |
"step": 1710
|
| 1071 |
},
|
| 1072 |
{
|
| 1073 |
+
"epoch": 4.03,
|
| 1074 |
+
"learning_rate": 3.16067947646895e-05,
|
| 1075 |
+
"loss": 0.5689,
|
| 1076 |
"step": 1720
|
| 1077 |
},
|
| 1078 |
{
|
| 1079 |
+
"epoch": 4.03,
|
| 1080 |
+
"learning_rate": 3.1467557783347256e-05,
|
| 1081 |
+
"loss": 0.0335,
|
| 1082 |
"step": 1730
|
| 1083 |
},
|
| 1084 |
{
|
| 1085 |
+
"epoch": 4.04,
|
| 1086 |
+
"learning_rate": 3.132832080200501e-05,
|
| 1087 |
+
"loss": 0.6061,
|
| 1088 |
"step": 1740
|
| 1089 |
},
|
| 1090 |
{
|
| 1091 |
+
"epoch": 4.04,
|
| 1092 |
+
"learning_rate": 3.118908382066277e-05,
|
| 1093 |
+
"loss": 0.0297,
|
| 1094 |
"step": 1750
|
| 1095 |
},
|
| 1096 |
{
|
| 1097 |
+
"epoch": 4.04,
|
| 1098 |
+
"learning_rate": 3.1049846839320524e-05,
|
| 1099 |
+
"loss": 0.6656,
|
| 1100 |
"step": 1760
|
| 1101 |
},
|
| 1102 |
{
|
| 1103 |
+
"epoch": 4.04,
|
| 1104 |
+
"learning_rate": 3.091060985797828e-05,
|
| 1105 |
+
"loss": 0.0602,
|
| 1106 |
"step": 1770
|
| 1107 |
},
|
| 1108 |
{
|
| 1109 |
+
"epoch": 4.05,
|
| 1110 |
+
"learning_rate": 3.0771372876636035e-05,
|
| 1111 |
+
"loss": 0.7032,
|
| 1112 |
"step": 1780
|
| 1113 |
},
|
| 1114 |
{
|
| 1115 |
+
"epoch": 4.05,
|
| 1116 |
+
"learning_rate": 3.063213589529379e-05,
|
| 1117 |
+
"loss": 0.4036,
|
| 1118 |
"step": 1790
|
| 1119 |
},
|
| 1120 |
{
|
| 1121 |
+
"epoch": 4.05,
|
| 1122 |
+
"learning_rate": 3.0492898913951544e-05,
|
| 1123 |
+
"loss": 0.0042,
|
| 1124 |
"step": 1800
|
| 1125 |
},
|
| 1126 |
{
|
| 1127 |
+
"epoch": 4.05,
|
| 1128 |
+
"learning_rate": 3.03536619326093e-05,
|
| 1129 |
+
"loss": 0.3816,
|
| 1130 |
"step": 1810
|
| 1131 |
},
|
| 1132 |
{
|
| 1133 |
+
"epoch": 4.06,
|
| 1134 |
+
"learning_rate": 3.0214424951267055e-05,
|
| 1135 |
+
"loss": 0.1152,
|
| 1136 |
"step": 1820
|
| 1137 |
},
|
| 1138 |
{
|
| 1139 |
+
"epoch": 4.06,
|
| 1140 |
+
"learning_rate": 3.007518796992481e-05,
|
| 1141 |
+
"loss": 0.0035,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1142 |
"step": 1830
|
| 1143 |
},
|
| 1144 |
{
|
| 1145 |
+
"epoch": 4.06,
|
| 1146 |
+
"learning_rate": 2.9935950988582567e-05,
|
| 1147 |
+
"loss": 0.0045,
|
| 1148 |
"step": 1840
|
| 1149 |
},
|
| 1150 |
{
|
| 1151 |
+
"epoch": 4.06,
|
| 1152 |
+
"learning_rate": 2.9796714007240323e-05,
|
| 1153 |
+
"loss": 0.5133,
|
| 1154 |
"step": 1850
|
| 1155 |
},
|
| 1156 |
{
|
| 1157 |
+
"epoch": 4.07,
|
| 1158 |
+
"learning_rate": 2.965747702589808e-05,
|
| 1159 |
+
"loss": 0.0034,
|
| 1160 |
"step": 1860
|
| 1161 |
},
|
| 1162 |
{
|
| 1163 |
+
"epoch": 4.07,
|
| 1164 |
+
"learning_rate": 2.9518240044555835e-05,
|
| 1165 |
+
"loss": 0.3994,
|
| 1166 |
"step": 1870
|
| 1167 |
},
|
| 1168 |
{
|
| 1169 |
+
"epoch": 4.07,
|
| 1170 |
+
"learning_rate": 2.937900306321359e-05,
|
| 1171 |
+
"loss": 0.727,
|
| 1172 |
"step": 1880
|
| 1173 |
},
|
| 1174 |
{
|
| 1175 |
+
"epoch": 4.07,
|
| 1176 |
+
"learning_rate": 2.9239766081871346e-05,
|
| 1177 |
+
"loss": 0.5794,
|
| 1178 |
"step": 1890
|
| 1179 |
},
|
| 1180 |
{
|
| 1181 |
+
"epoch": 4.08,
|
| 1182 |
+
"learning_rate": 2.91005291005291e-05,
|
| 1183 |
+
"loss": 1.3733,
|
| 1184 |
"step": 1900
|
| 1185 |
},
|
| 1186 |
{
|
| 1187 |
+
"epoch": 4.08,
|
| 1188 |
+
"learning_rate": 2.8961292119186855e-05,
|
| 1189 |
+
"loss": 0.1292,
|
| 1190 |
"step": 1910
|
| 1191 |
},
|
| 1192 |
{
|
| 1193 |
+
"epoch": 4.08,
|
| 1194 |
+
"learning_rate": 2.882205513784461e-05,
|
| 1195 |
+
"loss": 0.6721,
|
| 1196 |
"step": 1920
|
| 1197 |
},
|
| 1198 |
{
|
| 1199 |
+
"epoch": 4.08,
|
| 1200 |
+
"learning_rate": 2.8682818156502366e-05,
|
| 1201 |
+
"loss": 0.6031,
|
| 1202 |
"step": 1930
|
| 1203 |
},
|
| 1204 |
{
|
| 1205 |
+
"epoch": 4.09,
|
| 1206 |
+
"learning_rate": 2.8543581175160122e-05,
|
| 1207 |
+
"loss": 0.0148,
|
| 1208 |
"step": 1940
|
| 1209 |
},
|
| 1210 |
{
|
| 1211 |
+
"epoch": 4.09,
|
| 1212 |
+
"learning_rate": 2.8404344193817878e-05,
|
| 1213 |
+
"loss": 0.7074,
|
| 1214 |
"step": 1950
|
| 1215 |
},
|
| 1216 |
{
|
| 1217 |
+
"epoch": 4.09,
|
| 1218 |
+
"learning_rate": 2.8265107212475634e-05,
|
| 1219 |
+
"loss": 0.004,
|
| 1220 |
"step": 1960
|
| 1221 |
},
|
| 1222 |
{
|
| 1223 |
+
"epoch": 4.09,
|
| 1224 |
+
"learning_rate": 2.812587023113339e-05,
|
| 1225 |
+
"loss": 1.1926,
|
| 1226 |
"step": 1970
|
| 1227 |
},
|
| 1228 |
{
|
| 1229 |
+
"epoch": 4.1,
|
| 1230 |
+
"learning_rate": 2.7986633249791145e-05,
|
| 1231 |
+
"loss": 0.7448,
|
| 1232 |
"step": 1980
|
| 1233 |
},
|
| 1234 |
{
|
| 1235 |
+
"epoch": 4.1,
|
| 1236 |
+
"learning_rate": 2.78473962684489e-05,
|
| 1237 |
+
"loss": 1.2344,
|
| 1238 |
"step": 1990
|
| 1239 |
},
|
| 1240 |
{
|
| 1241 |
+
"epoch": 4.1,
|
| 1242 |
+
"eval_accuracy": 0.58,
|
| 1243 |
+
"eval_loss": 1.962124228477478,
|
| 1244 |
+
"eval_runtime": 28.4167,
|
| 1245 |
+
"eval_samples_per_second": 3.519,
|
| 1246 |
+
"eval_steps_per_second": 3.519,
|
| 1247 |
+
"step": 1990
|
| 1248 |
+
},
|
| 1249 |
+
{
|
| 1250 |
+
"epoch": 5.0,
|
| 1251 |
+
"learning_rate": 2.7708159287106654e-05,
|
| 1252 |
+
"loss": 0.8195,
|
| 1253 |
"step": 2000
|
| 1254 |
},
|
| 1255 |
{
|
| 1256 |
+
"epoch": 5.01,
|
| 1257 |
+
"learning_rate": 2.756892230576441e-05,
|
| 1258 |
+
"loss": 0.4829,
|
| 1259 |
"step": 2010
|
| 1260 |
},
|
| 1261 |
{
|
| 1262 |
+
"epoch": 5.01,
|
| 1263 |
+
"learning_rate": 2.7429685324422165e-05,
|
| 1264 |
+
"loss": 0.0041,
|
| 1265 |
"step": 2020
|
| 1266 |
},
|
| 1267 |
{
|
| 1268 |
+
"epoch": 5.01,
|
| 1269 |
+
"learning_rate": 2.729044834307992e-05,
|
| 1270 |
+
"loss": 0.2123,
|
| 1271 |
"step": 2030
|
| 1272 |
},
|
| 1273 |
{
|
| 1274 |
+
"epoch": 5.01,
|
| 1275 |
+
"learning_rate": 2.7151211361737677e-05,
|
| 1276 |
+
"loss": 0.5932,
|
| 1277 |
"step": 2040
|
| 1278 |
},
|
| 1279 |
{
|
| 1280 |
+
"epoch": 5.02,
|
| 1281 |
+
"learning_rate": 2.7011974380395433e-05,
|
| 1282 |
+
"loss": 0.0063,
|
| 1283 |
"step": 2050
|
| 1284 |
},
|
| 1285 |
{
|
| 1286 |
+
"epoch": 5.02,
|
| 1287 |
+
"learning_rate": 2.687273739905319e-05,
|
| 1288 |
+
"loss": 0.3498,
|
| 1289 |
"step": 2060
|
| 1290 |
},
|
| 1291 |
{
|
| 1292 |
+
"epoch": 5.02,
|
| 1293 |
+
"learning_rate": 2.6733500417710944e-05,
|
| 1294 |
+
"loss": 0.9616,
|
| 1295 |
"step": 2070
|
| 1296 |
},
|
| 1297 |
{
|
| 1298 |
+
"epoch": 5.02,
|
| 1299 |
+
"learning_rate": 2.65942634363687e-05,
|
| 1300 |
+
"loss": 0.0021,
|
| 1301 |
"step": 2080
|
| 1302 |
},
|
| 1303 |
{
|
| 1304 |
+
"epoch": 5.03,
|
| 1305 |
+
"learning_rate": 2.6455026455026456e-05,
|
| 1306 |
+
"loss": 0.4937,
|
| 1307 |
"step": 2090
|
| 1308 |
},
|
| 1309 |
{
|
| 1310 |
+
"epoch": 5.03,
|
| 1311 |
+
"learning_rate": 2.6315789473684212e-05,
|
| 1312 |
+
"loss": 0.2354,
|
| 1313 |
"step": 2100
|
| 1314 |
},
|
| 1315 |
{
|
| 1316 |
+
"epoch": 5.03,
|
| 1317 |
+
"learning_rate": 2.6176552492341964e-05,
|
| 1318 |
+
"loss": 0.3209,
|
| 1319 |
"step": 2110
|
| 1320 |
},
|
| 1321 |
{
|
| 1322 |
+
"epoch": 5.03,
|
| 1323 |
+
"learning_rate": 2.603731551099972e-05,
|
| 1324 |
+
"loss": 0.0217,
|
| 1325 |
"step": 2120
|
| 1326 |
},
|
| 1327 |
{
|
| 1328 |
+
"epoch": 5.04,
|
| 1329 |
+
"learning_rate": 2.5898078529657476e-05,
|
| 1330 |
+
"loss": 0.5236,
|
| 1331 |
"step": 2130
|
| 1332 |
},
|
| 1333 |
{
|
| 1334 |
+
"epoch": 5.04,
|
| 1335 |
+
"learning_rate": 2.5758841548315232e-05,
|
| 1336 |
+
"loss": 0.0044,
|
| 1337 |
"step": 2140
|
| 1338 |
},
|
| 1339 |
{
|
| 1340 |
+
"epoch": 5.04,
|
| 1341 |
+
"learning_rate": 2.5619604566972988e-05,
|
| 1342 |
+
"loss": 0.0032,
|
| 1343 |
"step": 2150
|
| 1344 |
},
|
| 1345 |
{
|
| 1346 |
+
"epoch": 5.04,
|
| 1347 |
+
"learning_rate": 2.5480367585630744e-05,
|
| 1348 |
+
"loss": 0.0117,
|
| 1349 |
"step": 2160
|
| 1350 |
},
|
| 1351 |
{
|
| 1352 |
+
"epoch": 5.05,
|
| 1353 |
+
"learning_rate": 2.53411306042885e-05,
|
| 1354 |
+
"loss": 0.0071,
|
| 1355 |
"step": 2170
|
| 1356 |
},
|
| 1357 |
{
|
| 1358 |
+
"epoch": 5.05,
|
| 1359 |
+
"learning_rate": 2.5201893622946255e-05,
|
| 1360 |
+
"loss": 0.0029,
|
| 1361 |
"step": 2180
|
| 1362 |
},
|
| 1363 |
{
|
| 1364 |
+
"epoch": 5.05,
|
| 1365 |
+
"learning_rate": 2.506265664160401e-05,
|
| 1366 |
+
"loss": 0.5591,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1367 |
"step": 2190
|
| 1368 |
},
|
| 1369 |
{
|
| 1370 |
+
"epoch": 5.05,
|
| 1371 |
+
"learning_rate": 2.4923419660261767e-05,
|
| 1372 |
+
"loss": 0.0042,
|
| 1373 |
"step": 2200
|
| 1374 |
},
|
| 1375 |
{
|
| 1376 |
+
"epoch": 5.06,
|
| 1377 |
+
"learning_rate": 2.478418267891952e-05,
|
| 1378 |
+
"loss": 0.0027,
|
| 1379 |
"step": 2210
|
| 1380 |
},
|
| 1381 |
{
|
| 1382 |
+
"epoch": 5.06,
|
| 1383 |
+
"learning_rate": 2.4644945697577275e-05,
|
| 1384 |
+
"loss": 0.2002,
|
| 1385 |
"step": 2220
|
| 1386 |
},
|
| 1387 |
{
|
| 1388 |
+
"epoch": 5.06,
|
| 1389 |
+
"learning_rate": 2.450570871623503e-05,
|
| 1390 |
+
"loss": 0.4917,
|
| 1391 |
"step": 2230
|
| 1392 |
},
|
| 1393 |
{
|
| 1394 |
+
"epoch": 5.06,
|
| 1395 |
+
"learning_rate": 2.4366471734892787e-05,
|
| 1396 |
+
"loss": 0.5791,
|
| 1397 |
"step": 2240
|
| 1398 |
},
|
| 1399 |
{
|
| 1400 |
+
"epoch": 5.07,
|
| 1401 |
+
"learning_rate": 2.4227234753550543e-05,
|
| 1402 |
+
"loss": 0.6229,
|
| 1403 |
"step": 2250
|
| 1404 |
},
|
| 1405 |
{
|
| 1406 |
+
"epoch": 5.07,
|
| 1407 |
+
"learning_rate": 2.40879977722083e-05,
|
| 1408 |
+
"loss": 0.4335,
|
| 1409 |
"step": 2260
|
| 1410 |
},
|
| 1411 |
{
|
| 1412 |
+
"epoch": 5.07,
|
| 1413 |
+
"learning_rate": 2.3948760790866054e-05,
|
| 1414 |
+
"loss": 0.7612,
|
| 1415 |
"step": 2270
|
| 1416 |
},
|
| 1417 |
{
|
| 1418 |
+
"epoch": 5.07,
|
| 1419 |
+
"learning_rate": 2.380952380952381e-05,
|
| 1420 |
+
"loss": 0.8699,
|
| 1421 |
"step": 2280
|
| 1422 |
},
|
| 1423 |
{
|
| 1424 |
+
"epoch": 5.08,
|
| 1425 |
+
"learning_rate": 2.3670286828181566e-05,
|
| 1426 |
+
"loss": 0.7213,
|
| 1427 |
"step": 2290
|
| 1428 |
},
|
| 1429 |
{
|
| 1430 |
+
"epoch": 5.08,
|
| 1431 |
+
"learning_rate": 2.3531049846839322e-05,
|
| 1432 |
+
"loss": 0.0036,
|
| 1433 |
"step": 2300
|
| 1434 |
},
|
| 1435 |
{
|
| 1436 |
+
"epoch": 5.08,
|
| 1437 |
+
"learning_rate": 2.3391812865497074e-05,
|
| 1438 |
+
"loss": 0.5293,
|
| 1439 |
"step": 2310
|
| 1440 |
},
|
| 1441 |
{
|
| 1442 |
+
"epoch": 5.08,
|
| 1443 |
+
"learning_rate": 2.325257588415483e-05,
|
| 1444 |
+
"loss": 0.002,
|
| 1445 |
"step": 2320
|
| 1446 |
},
|
| 1447 |
{
|
| 1448 |
+
"epoch": 5.09,
|
| 1449 |
+
"learning_rate": 2.3113338902812586e-05,
|
| 1450 |
+
"loss": 0.0098,
|
| 1451 |
"step": 2330
|
| 1452 |
},
|
| 1453 |
{
|
| 1454 |
+
"epoch": 5.09,
|
| 1455 |
+
"learning_rate": 2.2974101921470342e-05,
|
| 1456 |
+
"loss": 0.0034,
|
| 1457 |
"step": 2340
|
| 1458 |
},
|
| 1459 |
{
|
| 1460 |
+
"epoch": 5.09,
|
| 1461 |
+
"learning_rate": 2.2834864940128098e-05,
|
| 1462 |
+
"loss": 0.0018,
|
| 1463 |
"step": 2350
|
| 1464 |
},
|
| 1465 |
{
|
| 1466 |
+
"epoch": 5.09,
|
| 1467 |
+
"learning_rate": 2.2695627958785853e-05,
|
| 1468 |
+
"loss": 0.0022,
|
| 1469 |
"step": 2360
|
| 1470 |
},
|
| 1471 |
{
|
| 1472 |
+
"epoch": 5.1,
|
| 1473 |
+
"learning_rate": 2.255639097744361e-05,
|
| 1474 |
+
"loss": 0.9986,
|
| 1475 |
"step": 2370
|
| 1476 |
},
|
| 1477 |
{
|
| 1478 |
+
"epoch": 5.1,
|
| 1479 |
+
"learning_rate": 2.2417153996101365e-05,
|
| 1480 |
+
"loss": 0.1809,
|
| 1481 |
"step": 2380
|
| 1482 |
},
|
| 1483 |
{
|
| 1484 |
+
"epoch": 5.1,
|
| 1485 |
+
"eval_accuracy": 0.71,
|
| 1486 |
+
"eval_loss": 1.6321651935577393,
|
| 1487 |
+
"eval_runtime": 26.5939,
|
| 1488 |
+
"eval_samples_per_second": 3.76,
|
| 1489 |
+
"eval_steps_per_second": 3.76,
|
| 1490 |
+
"step": 2388
|
| 1491 |
+
},
|
| 1492 |
+
{
|
| 1493 |
+
"epoch": 6.0,
|
| 1494 |
+
"learning_rate": 2.227791701475912e-05,
|
| 1495 |
+
"loss": 0.0023,
|
| 1496 |
"step": 2390
|
| 1497 |
},
|
| 1498 |
{
|
| 1499 |
+
"epoch": 6.0,
|
| 1500 |
+
"learning_rate": 2.2138680033416877e-05,
|
| 1501 |
+
"loss": 0.2565,
|
| 1502 |
"step": 2400
|
| 1503 |
},
|
| 1504 |
{
|
| 1505 |
+
"epoch": 6.01,
|
| 1506 |
+
"learning_rate": 2.199944305207463e-05,
|
| 1507 |
+
"loss": 0.2508,
|
| 1508 |
"step": 2410
|
| 1509 |
},
|
| 1510 |
{
|
| 1511 |
+
"epoch": 6.01,
|
| 1512 |
+
"learning_rate": 2.1860206070732385e-05,
|
| 1513 |
+
"loss": 0.0077,
|
| 1514 |
"step": 2420
|
| 1515 |
},
|
| 1516 |
{
|
| 1517 |
+
"epoch": 6.01,
|
| 1518 |
+
"learning_rate": 2.172096908939014e-05,
|
| 1519 |
+
"loss": 0.0019,
|
| 1520 |
"step": 2430
|
| 1521 |
},
|
| 1522 |
{
|
| 1523 |
+
"epoch": 6.01,
|
| 1524 |
+
"learning_rate": 2.1581732108047897e-05,
|
| 1525 |
+
"loss": 0.1219,
|
| 1526 |
"step": 2440
|
| 1527 |
},
|
| 1528 |
{
|
| 1529 |
+
"epoch": 6.02,
|
| 1530 |
+
"learning_rate": 2.1442495126705653e-05,
|
| 1531 |
+
"loss": 0.004,
|
| 1532 |
"step": 2450
|
| 1533 |
},
|
| 1534 |
{
|
| 1535 |
+
"epoch": 6.02,
|
| 1536 |
+
"learning_rate": 2.130325814536341e-05,
|
| 1537 |
+
"loss": 0.002,
|
| 1538 |
"step": 2460
|
| 1539 |
},
|
| 1540 |
{
|
| 1541 |
+
"epoch": 6.02,
|
| 1542 |
+
"learning_rate": 2.1164021164021164e-05,
|
| 1543 |
+
"loss": 0.0189,
|
| 1544 |
"step": 2470
|
| 1545 |
},
|
| 1546 |
{
|
| 1547 |
+
"epoch": 6.02,
|
| 1548 |
+
"learning_rate": 2.102478418267892e-05,
|
| 1549 |
+
"loss": 0.0526,
|
| 1550 |
"step": 2480
|
| 1551 |
},
|
| 1552 |
{
|
| 1553 |
+
"epoch": 6.03,
|
| 1554 |
+
"learning_rate": 2.0885547201336676e-05,
|
| 1555 |
+
"loss": 0.0016,
|
| 1556 |
"step": 2490
|
| 1557 |
},
|
| 1558 |
{
|
| 1559 |
+
"epoch": 6.03,
|
| 1560 |
+
"learning_rate": 2.0746310219994432e-05,
|
| 1561 |
+
"loss": 0.0014,
|
| 1562 |
"step": 2500
|
| 1563 |
},
|
| 1564 |
{
|
| 1565 |
+
"epoch": 6.03,
|
| 1566 |
+
"learning_rate": 2.0607073238652184e-05,
|
| 1567 |
+
"loss": 0.002,
|
| 1568 |
"step": 2510
|
| 1569 |
},
|
| 1570 |
{
|
| 1571 |
+
"epoch": 6.03,
|
| 1572 |
+
"learning_rate": 2.046783625730994e-05,
|
| 1573 |
+
"loss": 0.0288,
|
| 1574 |
"step": 2520
|
| 1575 |
},
|
| 1576 |
{
|
| 1577 |
+
"epoch": 6.04,
|
| 1578 |
+
"learning_rate": 2.0328599275967696e-05,
|
| 1579 |
+
"loss": 0.0026,
|
| 1580 |
"step": 2530
|
| 1581 |
},
|
| 1582 |
{
|
| 1583 |
+
"epoch": 6.04,
|
| 1584 |
+
"learning_rate": 2.018936229462545e-05,
|
| 1585 |
+
"loss": 0.0537,
|
| 1586 |
"step": 2540
|
| 1587 |
},
|
| 1588 |
{
|
| 1589 |
+
"epoch": 6.04,
|
| 1590 |
+
"learning_rate": 2.0050125313283208e-05,
|
| 1591 |
+
"loss": 0.0024,
|
| 1592 |
"step": 2550
|
| 1593 |
},
|
| 1594 |
+
{
|
| 1595 |
+
"epoch": 6.04,
|
| 1596 |
+
"learning_rate": 1.9910888331940963e-05,
|
| 1597 |
+
"loss": 0.0017,
|
| 1598 |
+
"step": 2560
|
| 1599 |
+
},
|
| 1600 |
+
{
|
| 1601 |
+
"epoch": 6.05,
|
| 1602 |
+
"learning_rate": 1.977165135059872e-05,
|
| 1603 |
+
"loss": 0.005,
|
| 1604 |
+
"step": 2570
|
| 1605 |
+
},
|
| 1606 |
+
{
|
| 1607 |
+
"epoch": 6.05,
|
| 1608 |
+
"learning_rate": 1.9632414369256475e-05,
|
| 1609 |
+
"loss": 0.0019,
|
| 1610 |
+
"step": 2580
|
| 1611 |
+
},
|
| 1612 |
+
{
|
| 1613 |
+
"epoch": 6.05,
|
| 1614 |
+
"learning_rate": 1.949317738791423e-05,
|
| 1615 |
+
"loss": 0.4555,
|
| 1616 |
+
"step": 2590
|
| 1617 |
+
},
|
| 1618 |
+
{
|
| 1619 |
+
"epoch": 6.05,
|
| 1620 |
+
"learning_rate": 1.9353940406571987e-05,
|
| 1621 |
+
"loss": 0.5544,
|
| 1622 |
+
"step": 2600
|
| 1623 |
+
},
|
| 1624 |
+
{
|
| 1625 |
+
"epoch": 6.06,
|
| 1626 |
+
"learning_rate": 1.9214703425229743e-05,
|
| 1627 |
+
"loss": 0.0016,
|
| 1628 |
+
"step": 2610
|
| 1629 |
+
},
|
| 1630 |
+
{
|
| 1631 |
+
"epoch": 6.06,
|
| 1632 |
+
"learning_rate": 1.9075466443887495e-05,
|
| 1633 |
+
"loss": 0.0032,
|
| 1634 |
+
"step": 2620
|
| 1635 |
+
},
|
| 1636 |
+
{
|
| 1637 |
+
"epoch": 6.06,
|
| 1638 |
+
"learning_rate": 1.893622946254525e-05,
|
| 1639 |
+
"loss": 0.3974,
|
| 1640 |
+
"step": 2630
|
| 1641 |
+
},
|
| 1642 |
+
{
|
| 1643 |
+
"epoch": 6.06,
|
| 1644 |
+
"learning_rate": 1.8796992481203007e-05,
|
| 1645 |
+
"loss": 0.1622,
|
| 1646 |
+
"step": 2640
|
| 1647 |
+
},
|
| 1648 |
+
{
|
| 1649 |
+
"epoch": 6.07,
|
| 1650 |
+
"learning_rate": 1.8657755499860762e-05,
|
| 1651 |
+
"loss": 0.5769,
|
| 1652 |
+
"step": 2650
|
| 1653 |
+
},
|
| 1654 |
+
{
|
| 1655 |
+
"epoch": 6.07,
|
| 1656 |
+
"learning_rate": 1.8518518518518518e-05,
|
| 1657 |
+
"loss": 0.0036,
|
| 1658 |
+
"step": 2660
|
| 1659 |
+
},
|
| 1660 |
+
{
|
| 1661 |
+
"epoch": 6.07,
|
| 1662 |
+
"learning_rate": 1.8379281537176274e-05,
|
| 1663 |
+
"loss": 0.0026,
|
| 1664 |
+
"step": 2670
|
| 1665 |
+
},
|
| 1666 |
+
{
|
| 1667 |
+
"epoch": 6.07,
|
| 1668 |
+
"learning_rate": 1.824004455583403e-05,
|
| 1669 |
+
"loss": 0.0015,
|
| 1670 |
+
"step": 2680
|
| 1671 |
+
},
|
| 1672 |
+
{
|
| 1673 |
+
"epoch": 6.08,
|
| 1674 |
+
"learning_rate": 1.8100807574491786e-05,
|
| 1675 |
+
"loss": 0.2167,
|
| 1676 |
+
"step": 2690
|
| 1677 |
+
},
|
| 1678 |
+
{
|
| 1679 |
+
"epoch": 6.08,
|
| 1680 |
+
"learning_rate": 1.796157059314954e-05,
|
| 1681 |
+
"loss": 0.5677,
|
| 1682 |
+
"step": 2700
|
| 1683 |
+
},
|
| 1684 |
+
{
|
| 1685 |
+
"epoch": 6.08,
|
| 1686 |
+
"learning_rate": 1.7822333611807297e-05,
|
| 1687 |
+
"loss": 0.0023,
|
| 1688 |
+
"step": 2710
|
| 1689 |
+
},
|
| 1690 |
+
{
|
| 1691 |
+
"epoch": 6.08,
|
| 1692 |
+
"learning_rate": 1.768309663046505e-05,
|
| 1693 |
+
"loss": 0.0018,
|
| 1694 |
+
"step": 2720
|
| 1695 |
+
},
|
| 1696 |
+
{
|
| 1697 |
+
"epoch": 6.09,
|
| 1698 |
+
"learning_rate": 1.7543859649122806e-05,
|
| 1699 |
+
"loss": 0.0217,
|
| 1700 |
+
"step": 2730
|
| 1701 |
+
},
|
| 1702 |
+
{
|
| 1703 |
+
"epoch": 6.09,
|
| 1704 |
+
"learning_rate": 1.740462266778056e-05,
|
| 1705 |
+
"loss": 0.0017,
|
| 1706 |
+
"step": 2740
|
| 1707 |
+
},
|
| 1708 |
+
{
|
| 1709 |
+
"epoch": 6.09,
|
| 1710 |
+
"learning_rate": 1.7265385686438317e-05,
|
| 1711 |
+
"loss": 0.0015,
|
| 1712 |
+
"step": 2750
|
| 1713 |
+
},
|
| 1714 |
+
{
|
| 1715 |
+
"epoch": 6.09,
|
| 1716 |
+
"learning_rate": 1.7126148705096073e-05,
|
| 1717 |
+
"loss": 0.265,
|
| 1718 |
+
"step": 2760
|
| 1719 |
+
},
|
| 1720 |
+
{
|
| 1721 |
+
"epoch": 6.1,
|
| 1722 |
+
"learning_rate": 1.698691172375383e-05,
|
| 1723 |
+
"loss": 0.0016,
|
| 1724 |
+
"step": 2770
|
| 1725 |
+
},
|
| 1726 |
+
{
|
| 1727 |
+
"epoch": 6.1,
|
| 1728 |
+
"learning_rate": 1.6847674742411585e-05,
|
| 1729 |
+
"loss": 0.0011,
|
| 1730 |
+
"step": 2780
|
| 1731 |
+
},
|
| 1732 |
{
|
| 1733 |
"epoch": 6.1,
|
| 1734 |
+
"eval_accuracy": 0.71,
|
| 1735 |
+
"eval_loss": 1.8265665769577026,
|
| 1736 |
+
"eval_runtime": 26.2083,
|
| 1737 |
+
"eval_samples_per_second": 3.816,
|
| 1738 |
+
"eval_steps_per_second": 3.816,
|
| 1739 |
+
"step": 2786
|
| 1740 |
},
|
| 1741 |
{
|
| 1742 |
"epoch": 7.0,
|
| 1743 |
+
"learning_rate": 1.670843776106934e-05,
|
| 1744 |
+
"loss": 0.0023,
|
| 1745 |
+
"step": 2790
|
| 1746 |
},
|
| 1747 |
{
|
| 1748 |
"epoch": 7.0,
|
| 1749 |
+
"learning_rate": 1.6569200779727097e-05,
|
| 1750 |
+
"loss": 0.0036,
|
| 1751 |
+
"step": 2800
|
| 1752 |
},
|
| 1753 |
{
|
| 1754 |
"epoch": 7.01,
|
| 1755 |
+
"learning_rate": 1.6429963798384852e-05,
|
| 1756 |
+
"loss": 0.0022,
|
| 1757 |
+
"step": 2810
|
| 1758 |
},
|
| 1759 |
{
|
| 1760 |
"epoch": 7.01,
|
| 1761 |
+
"learning_rate": 1.6290726817042605e-05,
|
| 1762 |
+
"loss": 0.0016,
|
| 1763 |
+
"step": 2820
|
| 1764 |
},
|
| 1765 |
{
|
| 1766 |
"epoch": 7.01,
|
| 1767 |
+
"learning_rate": 1.615148983570036e-05,
|
| 1768 |
+
"loss": 0.0029,
|
| 1769 |
+
"step": 2830
|
| 1770 |
+
},
|
| 1771 |
+
{
|
| 1772 |
+
"epoch": 7.01,
|
| 1773 |
+
"learning_rate": 1.6012252854358116e-05,
|
| 1774 |
+
"loss": 0.7778,
|
| 1775 |
+
"step": 2840
|
| 1776 |
},
|
| 1777 |
{
|
| 1778 |
"epoch": 7.02,
|
| 1779 |
+
"learning_rate": 1.5873015873015872e-05,
|
| 1780 |
+
"loss": 0.0014,
|
| 1781 |
+
"step": 2850
|
| 1782 |
},
|
| 1783 |
{
|
| 1784 |
"epoch": 7.02,
|
| 1785 |
+
"learning_rate": 1.5733778891673628e-05,
|
| 1786 |
+
"loss": 0.0317,
|
| 1787 |
+
"step": 2860
|
| 1788 |
},
|
| 1789 |
{
|
| 1790 |
"epoch": 7.02,
|
| 1791 |
+
"learning_rate": 1.5594541910331384e-05,
|
| 1792 |
+
"loss": 0.0015,
|
| 1793 |
+
"step": 2870
|
| 1794 |
},
|
| 1795 |
{
|
| 1796 |
"epoch": 7.02,
|
| 1797 |
+
"learning_rate": 1.545530492898914e-05,
|
| 1798 |
+
"loss": 0.3381,
|
| 1799 |
+
"step": 2880
|
| 1800 |
},
|
| 1801 |
{
|
| 1802 |
"epoch": 7.03,
|
| 1803 |
+
"learning_rate": 1.5316067947646896e-05,
|
| 1804 |
+
"loss": 0.0014,
|
| 1805 |
+
"step": 2890
|
| 1806 |
},
|
| 1807 |
{
|
| 1808 |
"epoch": 7.03,
|
| 1809 |
+
"learning_rate": 1.517683096630465e-05,
|
| 1810 |
+
"loss": 0.0014,
|
| 1811 |
+
"step": 2900
|
| 1812 |
},
|
| 1813 |
{
|
| 1814 |
"epoch": 7.03,
|
| 1815 |
+
"learning_rate": 1.5037593984962406e-05,
|
| 1816 |
+
"loss": 0.0017,
|
| 1817 |
+
"step": 2910
|
| 1818 |
},
|
| 1819 |
{
|
| 1820 |
"epoch": 7.03,
|
| 1821 |
+
"learning_rate": 1.4898357003620161e-05,
|
| 1822 |
+
"loss": 0.7069,
|
| 1823 |
+
"step": 2920
|
| 1824 |
},
|
| 1825 |
{
|
| 1826 |
"epoch": 7.04,
|
| 1827 |
+
"learning_rate": 1.4759120022277917e-05,
|
| 1828 |
+
"loss": 0.0016,
|
| 1829 |
+
"step": 2930
|
| 1830 |
},
|
| 1831 |
{
|
| 1832 |
"epoch": 7.04,
|
| 1833 |
+
"learning_rate": 1.4619883040935673e-05,
|
| 1834 |
+
"loss": 0.0069,
|
| 1835 |
+
"step": 2940
|
| 1836 |
},
|
| 1837 |
{
|
| 1838 |
"epoch": 7.04,
|
| 1839 |
+
"learning_rate": 1.4480646059593427e-05,
|
| 1840 |
+
"loss": 0.0411,
|
| 1841 |
+
"step": 2950
|
| 1842 |
+
},
|
| 1843 |
+
{
|
| 1844 |
+
"epoch": 7.04,
|
| 1845 |
+
"learning_rate": 1.4341409078251183e-05,
|
| 1846 |
+
"loss": 0.0076,
|
| 1847 |
+
"step": 2960
|
| 1848 |
},
|
| 1849 |
{
|
| 1850 |
"epoch": 7.05,
|
| 1851 |
+
"learning_rate": 1.4202172096908939e-05,
|
| 1852 |
+
"loss": 0.0174,
|
| 1853 |
+
"step": 2970
|
| 1854 |
},
|
| 1855 |
{
|
| 1856 |
"epoch": 7.05,
|
| 1857 |
+
"learning_rate": 1.4062935115566695e-05,
|
| 1858 |
+
"loss": 0.007,
|
| 1859 |
+
"step": 2980
|
| 1860 |
},
|
| 1861 |
{
|
| 1862 |
"epoch": 7.05,
|
| 1863 |
+
"learning_rate": 1.392369813422445e-05,
|
| 1864 |
+
"loss": 0.7089,
|
| 1865 |
+
"step": 2990
|
| 1866 |
},
|
| 1867 |
{
|
| 1868 |
"epoch": 7.05,
|
| 1869 |
+
"learning_rate": 1.3784461152882205e-05,
|
| 1870 |
+
"loss": 0.2224,
|
| 1871 |
+
"step": 3000
|
| 1872 |
},
|
| 1873 |
{
|
| 1874 |
"epoch": 7.06,
|
| 1875 |
+
"learning_rate": 1.364522417153996e-05,
|
| 1876 |
+
"loss": 0.0039,
|
| 1877 |
+
"step": 3010
|
| 1878 |
},
|
| 1879 |
{
|
| 1880 |
"epoch": 7.06,
|
| 1881 |
+
"learning_rate": 1.3505987190197716e-05,
|
| 1882 |
+
"loss": 0.0011,
|
| 1883 |
+
"step": 3020
|
| 1884 |
},
|
| 1885 |
{
|
| 1886 |
"epoch": 7.06,
|
| 1887 |
+
"learning_rate": 1.3366750208855472e-05,
|
| 1888 |
+
"loss": 0.0014,
|
| 1889 |
+
"step": 3030
|
| 1890 |
},
|
| 1891 |
{
|
| 1892 |
"epoch": 7.06,
|
| 1893 |
+
"learning_rate": 1.3227513227513228e-05,
|
| 1894 |
+
"loss": 0.0749,
|
| 1895 |
+
"step": 3040
|
| 1896 |
},
|
| 1897 |
{
|
| 1898 |
"epoch": 7.07,
|
| 1899 |
+
"learning_rate": 1.3088276246170982e-05,
|
| 1900 |
+
"loss": 0.7371,
|
| 1901 |
+
"step": 3050
|
| 1902 |
},
|
| 1903 |
{
|
| 1904 |
"epoch": 7.07,
|
| 1905 |
+
"learning_rate": 1.2949039264828738e-05,
|
| 1906 |
+
"loss": 0.0012,
|
| 1907 |
+
"step": 3060
|
| 1908 |
},
|
| 1909 |
{
|
| 1910 |
"epoch": 7.07,
|
| 1911 |
+
"learning_rate": 1.2809802283486494e-05,
|
| 1912 |
+
"loss": 0.6728,
|
| 1913 |
+
"step": 3070
|
| 1914 |
+
},
|
| 1915 |
+
{
|
| 1916 |
+
"epoch": 7.07,
|
| 1917 |
+
"learning_rate": 1.267056530214425e-05,
|
| 1918 |
+
"loss": 0.0017,
|
| 1919 |
+
"step": 3080
|
| 1920 |
},
|
| 1921 |
{
|
| 1922 |
"epoch": 7.08,
|
| 1923 |
+
"learning_rate": 1.2531328320802006e-05,
|
| 1924 |
+
"loss": 0.0018,
|
| 1925 |
+
"step": 3090
|
| 1926 |
},
|
| 1927 |
{
|
| 1928 |
"epoch": 7.08,
|
| 1929 |
+
"learning_rate": 1.239209133945976e-05,
|
| 1930 |
+
"loss": 0.0038,
|
| 1931 |
+
"step": 3100
|
| 1932 |
},
|
| 1933 |
{
|
| 1934 |
"epoch": 7.08,
|
| 1935 |
+
"learning_rate": 1.2252854358117516e-05,
|
| 1936 |
+
"loss": 0.0014,
|
| 1937 |
+
"step": 3110
|
| 1938 |
},
|
| 1939 |
{
|
| 1940 |
"epoch": 7.08,
|
| 1941 |
+
"learning_rate": 1.2113617376775271e-05,
|
| 1942 |
+
"loss": 0.0026,
|
| 1943 |
+
"step": 3120
|
| 1944 |
},
|
| 1945 |
{
|
| 1946 |
"epoch": 7.09,
|
| 1947 |
+
"learning_rate": 1.1974380395433027e-05,
|
| 1948 |
+
"loss": 0.0014,
|
| 1949 |
+
"step": 3130
|
| 1950 |
},
|
| 1951 |
{
|
| 1952 |
"epoch": 7.09,
|
| 1953 |
+
"learning_rate": 1.1835143414090783e-05,
|
| 1954 |
+
"loss": 0.0013,
|
| 1955 |
+
"step": 3140
|
| 1956 |
},
|
| 1957 |
{
|
| 1958 |
"epoch": 7.09,
|
| 1959 |
+
"learning_rate": 1.1695906432748537e-05,
|
| 1960 |
+
"loss": 0.003,
|
| 1961 |
+
"step": 3150
|
| 1962 |
},
|
| 1963 |
{
|
| 1964 |
"epoch": 7.09,
|
| 1965 |
+
"learning_rate": 1.1556669451406293e-05,
|
| 1966 |
+
"loss": 0.0014,
|
| 1967 |
+
"step": 3160
|
| 1968 |
},
|
| 1969 |
{
|
| 1970 |
"epoch": 7.1,
|
| 1971 |
+
"learning_rate": 1.1417432470064049e-05,
|
| 1972 |
+
"loss": 0.0018,
|
| 1973 |
+
"step": 3170
|
| 1974 |
},
|
| 1975 |
{
|
| 1976 |
"epoch": 7.1,
|
| 1977 |
+
"learning_rate": 1.1278195488721805e-05,
|
| 1978 |
+
"loss": 0.0951,
|
| 1979 |
+
"step": 3180
|
| 1980 |
},
|
| 1981 |
{
|
| 1982 |
"epoch": 7.1,
|
| 1983 |
+
"eval_accuracy": 0.78,
|
| 1984 |
+
"eval_loss": 1.5909953117370605,
|
| 1985 |
+
"eval_runtime": 27.0785,
|
| 1986 |
+
"eval_samples_per_second": 3.693,
|
| 1987 |
+
"eval_steps_per_second": 3.693,
|
| 1988 |
+
"step": 3184
|
| 1989 |
},
|
| 1990 |
{
|
| 1991 |
"epoch": 8.0,
|
| 1992 |
+
"learning_rate": 1.113895850737956e-05,
|
| 1993 |
+
"loss": 0.0011,
|
| 1994 |
+
"step": 3190
|
| 1995 |
+
},
|
| 1996 |
+
{
|
| 1997 |
+
"epoch": 8.0,
|
| 1998 |
+
"learning_rate": 1.0999721526037315e-05,
|
| 1999 |
+
"loss": 0.0011,
|
| 2000 |
+
"step": 3200
|
| 2001 |
},
|
| 2002 |
{
|
| 2003 |
"epoch": 8.01,
|
| 2004 |
+
"learning_rate": 1.086048454469507e-05,
|
| 2005 |
"loss": 0.0016,
|
| 2006 |
+
"step": 3210
|
| 2007 |
},
|
| 2008 |
{
|
| 2009 |
"epoch": 8.01,
|
| 2010 |
+
"learning_rate": 1.0721247563352826e-05,
|
| 2011 |
+
"loss": 0.0014,
|
| 2012 |
+
"step": 3220
|
| 2013 |
},
|
| 2014 |
{
|
| 2015 |
"epoch": 8.01,
|
| 2016 |
+
"learning_rate": 1.0582010582010582e-05,
|
| 2017 |
+
"loss": 0.0011,
|
| 2018 |
+
"step": 3230
|
| 2019 |
},
|
| 2020 |
{
|
| 2021 |
"epoch": 8.01,
|
| 2022 |
+
"learning_rate": 1.0442773600668338e-05,
|
| 2023 |
+
"loss": 0.0015,
|
| 2024 |
+
"step": 3240
|
| 2025 |
},
|
| 2026 |
{
|
| 2027 |
"epoch": 8.02,
|
| 2028 |
+
"learning_rate": 1.0303536619326092e-05,
|
| 2029 |
+
"loss": 0.0023,
|
| 2030 |
+
"step": 3250
|
| 2031 |
},
|
| 2032 |
{
|
| 2033 |
"epoch": 8.02,
|
| 2034 |
+
"learning_rate": 1.0164299637983848e-05,
|
| 2035 |
+
"loss": 0.0021,
|
| 2036 |
+
"step": 3260
|
| 2037 |
},
|
| 2038 |
{
|
| 2039 |
"epoch": 8.02,
|
| 2040 |
+
"learning_rate": 1.0025062656641604e-05,
|
| 2041 |
+
"loss": 0.0013,
|
| 2042 |
+
"step": 3270
|
| 2043 |
},
|
| 2044 |
{
|
| 2045 |
"epoch": 8.02,
|
| 2046 |
+
"learning_rate": 9.88582567529936e-06,
|
| 2047 |
+
"loss": 0.0011,
|
| 2048 |
+
"step": 3280
|
| 2049 |
},
|
| 2050 |
{
|
| 2051 |
"epoch": 8.03,
|
| 2052 |
+
"learning_rate": 9.746588693957115e-06,
|
| 2053 |
+
"loss": 0.0011,
|
| 2054 |
+
"step": 3290
|
| 2055 |
},
|
| 2056 |
{
|
| 2057 |
"epoch": 8.03,
|
| 2058 |
+
"learning_rate": 9.607351712614871e-06,
|
| 2059 |
+
"loss": 0.0013,
|
| 2060 |
+
"step": 3300
|
| 2061 |
},
|
| 2062 |
{
|
| 2063 |
"epoch": 8.03,
|
| 2064 |
+
"learning_rate": 9.468114731272625e-06,
|
| 2065 |
+
"loss": 0.0011,
|
| 2066 |
+
"step": 3310
|
| 2067 |
+
},
|
| 2068 |
+
{
|
| 2069 |
+
"epoch": 8.03,
|
| 2070 |
+
"learning_rate": 9.328877749930381e-06,
|
| 2071 |
+
"loss": 0.0563,
|
| 2072 |
+
"step": 3320
|
| 2073 |
},
|
| 2074 |
{
|
| 2075 |
"epoch": 8.04,
|
| 2076 |
+
"learning_rate": 9.189640768588137e-06,
|
| 2077 |
+
"loss": 0.0018,
|
| 2078 |
+
"step": 3330
|
| 2079 |
},
|
| 2080 |
{
|
| 2081 |
"epoch": 8.04,
|
| 2082 |
+
"learning_rate": 9.050403787245893e-06,
|
| 2083 |
+
"loss": 0.002,
|
| 2084 |
+
"step": 3340
|
| 2085 |
},
|
| 2086 |
{
|
| 2087 |
"epoch": 8.04,
|
| 2088 |
+
"learning_rate": 8.911166805903649e-06,
|
| 2089 |
+
"loss": 0.0016,
|
| 2090 |
+
"step": 3350
|
| 2091 |
},
|
| 2092 |
{
|
| 2093 |
"epoch": 8.04,
|
| 2094 |
+
"learning_rate": 8.771929824561403e-06,
|
| 2095 |
+
"loss": 0.0017,
|
| 2096 |
+
"step": 3360
|
| 2097 |
},
|
| 2098 |
{
|
| 2099 |
"epoch": 8.05,
|
| 2100 |
+
"learning_rate": 8.632692843219159e-06,
|
| 2101 |
+
"loss": 0.0013,
|
| 2102 |
+
"step": 3370
|
| 2103 |
},
|
| 2104 |
{
|
| 2105 |
"epoch": 8.05,
|
| 2106 |
+
"learning_rate": 8.493455861876915e-06,
|
| 2107 |
+
"loss": 0.0015,
|
| 2108 |
+
"step": 3380
|
| 2109 |
},
|
| 2110 |
{
|
| 2111 |
"epoch": 8.05,
|
| 2112 |
+
"learning_rate": 8.35421888053467e-06,
|
| 2113 |
+
"loss": 0.0013,
|
| 2114 |
+
"step": 3390
|
| 2115 |
},
|
| 2116 |
{
|
| 2117 |
"epoch": 8.05,
|
| 2118 |
+
"learning_rate": 8.214981899192426e-06,
|
| 2119 |
+
"loss": 0.0011,
|
| 2120 |
+
"step": 3400
|
| 2121 |
},
|
| 2122 |
{
|
| 2123 |
"epoch": 8.06,
|
| 2124 |
+
"learning_rate": 8.07574491785018e-06,
|
| 2125 |
+
"loss": 0.0012,
|
| 2126 |
+
"step": 3410
|
| 2127 |
},
|
| 2128 |
{
|
| 2129 |
"epoch": 8.06,
|
| 2130 |
+
"learning_rate": 7.936507936507936e-06,
|
| 2131 |
+
"loss": 0.0015,
|
| 2132 |
+
"step": 3420
|
| 2133 |
},
|
| 2134 |
{
|
| 2135 |
"epoch": 8.06,
|
| 2136 |
+
"learning_rate": 7.797270955165692e-06,
|
| 2137 |
+
"loss": 0.0015,
|
| 2138 |
+
"step": 3430
|
| 2139 |
+
},
|
| 2140 |
+
{
|
| 2141 |
+
"epoch": 8.06,
|
| 2142 |
+
"learning_rate": 7.658033973823448e-06,
|
| 2143 |
+
"loss": 0.0015,
|
| 2144 |
+
"step": 3440
|
| 2145 |
},
|
| 2146 |
{
|
| 2147 |
"epoch": 8.07,
|
| 2148 |
+
"learning_rate": 7.518796992481203e-06,
|
| 2149 |
+
"loss": 1.1364,
|
| 2150 |
+
"step": 3450
|
| 2151 |
},
|
| 2152 |
{
|
| 2153 |
"epoch": 8.07,
|
| 2154 |
+
"learning_rate": 7.379560011138959e-06,
|
| 2155 |
+
"loss": 0.0014,
|
| 2156 |
+
"step": 3460
|
| 2157 |
},
|
| 2158 |
{
|
| 2159 |
"epoch": 8.07,
|
| 2160 |
+
"learning_rate": 7.240323029796714e-06,
|
| 2161 |
+
"loss": 0.0016,
|
| 2162 |
+
"step": 3470
|
| 2163 |
},
|
| 2164 |
{
|
| 2165 |
"epoch": 8.07,
|
| 2166 |
+
"learning_rate": 7.1010860484544695e-06,
|
| 2167 |
+
"loss": 0.0013,
|
| 2168 |
+
"step": 3480
|
| 2169 |
},
|
| 2170 |
{
|
| 2171 |
"epoch": 8.08,
|
| 2172 |
+
"learning_rate": 6.961849067112225e-06,
|
| 2173 |
+
"loss": 0.0015,
|
| 2174 |
+
"step": 3490
|
| 2175 |
},
|
| 2176 |
{
|
| 2177 |
"epoch": 8.08,
|
| 2178 |
+
"learning_rate": 6.82261208576998e-06,
|
| 2179 |
+
"loss": 0.0012,
|
| 2180 |
+
"step": 3500
|
| 2181 |
},
|
| 2182 |
{
|
| 2183 |
"epoch": 8.08,
|
| 2184 |
+
"learning_rate": 6.683375104427736e-06,
|
| 2185 |
+
"loss": 0.0014,
|
| 2186 |
+
"step": 3510
|
| 2187 |
},
|
| 2188 |
{
|
| 2189 |
"epoch": 8.08,
|
| 2190 |
+
"learning_rate": 6.544138123085491e-06,
|
| 2191 |
+
"loss": 0.001,
|
| 2192 |
+
"step": 3520
|
| 2193 |
},
|
| 2194 |
{
|
| 2195 |
"epoch": 8.09,
|
| 2196 |
+
"learning_rate": 6.404901141743247e-06,
|
| 2197 |
+
"loss": 0.0009,
|
| 2198 |
+
"step": 3530
|
| 2199 |
},
|
| 2200 |
{
|
| 2201 |
"epoch": 8.09,
|
| 2202 |
+
"learning_rate": 6.265664160401003e-06,
|
| 2203 |
+
"loss": 0.0014,
|
| 2204 |
+
"step": 3540
|
| 2205 |
},
|
| 2206 |
{
|
| 2207 |
"epoch": 8.09,
|
| 2208 |
+
"learning_rate": 6.126427179058758e-06,
|
| 2209 |
+
"loss": 0.0014,
|
| 2210 |
+
"step": 3550
|
| 2211 |
+
},
|
| 2212 |
+
{
|
| 2213 |
+
"epoch": 8.09,
|
| 2214 |
+
"learning_rate": 5.987190197716514e-06,
|
| 2215 |
+
"loss": 0.0016,
|
| 2216 |
+
"step": 3560
|
| 2217 |
},
|
| 2218 |
{
|
| 2219 |
"epoch": 8.1,
|
| 2220 |
+
"learning_rate": 5.8479532163742686e-06,
|
| 2221 |
+
"loss": 0.0014,
|
| 2222 |
+
"step": 3570
|
| 2223 |
},
|
| 2224 |
{
|
| 2225 |
"epoch": 8.1,
|
| 2226 |
+
"learning_rate": 5.708716235032024e-06,
|
| 2227 |
+
"loss": 0.4047,
|
| 2228 |
+
"step": 3580
|
| 2229 |
},
|
| 2230 |
{
|
| 2231 |
"epoch": 8.1,
|
| 2232 |
+
"eval_accuracy": 0.7,
|
| 2233 |
+
"eval_loss": 1.9999462366104126,
|
| 2234 |
+
"eval_runtime": 28.4623,
|
| 2235 |
+
"eval_samples_per_second": 3.513,
|
| 2236 |
+
"eval_steps_per_second": 3.513,
|
| 2237 |
+
"step": 3582
|
| 2238 |
},
|
| 2239 |
{
|
| 2240 |
"epoch": 9.0,
|
| 2241 |
+
"learning_rate": 5.56947925368978e-06,
|
| 2242 |
+
"loss": 0.0014,
|
| 2243 |
+
"step": 3590
|
| 2244 |
},
|
| 2245 |
{
|
| 2246 |
"epoch": 9.0,
|
| 2247 |
+
"learning_rate": 5.430242272347535e-06,
|
| 2248 |
+
"loss": 0.0013,
|
| 2249 |
+
"step": 3600
|
| 2250 |
},
|
| 2251 |
{
|
| 2252 |
"epoch": 9.01,
|
| 2253 |
+
"learning_rate": 5.291005291005291e-06,
|
| 2254 |
+
"loss": 0.0354,
|
| 2255 |
+
"step": 3610
|
| 2256 |
},
|
| 2257 |
{
|
| 2258 |
"epoch": 9.01,
|
| 2259 |
+
"learning_rate": 5.151768309663046e-06,
|
| 2260 |
+
"loss": 0.0011,
|
| 2261 |
+
"step": 3620
|
| 2262 |
},
|
| 2263 |
{
|
| 2264 |
"epoch": 9.01,
|
| 2265 |
+
"learning_rate": 5.012531328320802e-06,
|
| 2266 |
+
"loss": 0.576,
|
| 2267 |
+
"step": 3630
|
| 2268 |
+
},
|
| 2269 |
+
{
|
| 2270 |
+
"epoch": 9.01,
|
| 2271 |
+
"learning_rate": 4.873294346978558e-06,
|
| 2272 |
+
"loss": 0.002,
|
| 2273 |
+
"step": 3640
|
| 2274 |
},
|
| 2275 |
{
|
| 2276 |
"epoch": 9.02,
|
| 2277 |
+
"learning_rate": 4.734057365636313e-06,
|
| 2278 |
+
"loss": 0.0009,
|
| 2279 |
+
"step": 3650
|
| 2280 |
},
|
| 2281 |
{
|
| 2282 |
"epoch": 9.02,
|
| 2283 |
+
"learning_rate": 4.5948203842940685e-06,
|
| 2284 |
+
"loss": 0.048,
|
| 2285 |
+
"step": 3660
|
| 2286 |
},
|
| 2287 |
{
|
| 2288 |
"epoch": 9.02,
|
| 2289 |
+
"learning_rate": 4.455583402951824e-06,
|
| 2290 |
+
"loss": 0.0017,
|
| 2291 |
+
"step": 3670
|
| 2292 |
},
|
| 2293 |
{
|
| 2294 |
"epoch": 9.02,
|
| 2295 |
+
"learning_rate": 4.316346421609579e-06,
|
| 2296 |
+
"loss": 0.0011,
|
| 2297 |
+
"step": 3680
|
| 2298 |
},
|
| 2299 |
{
|
| 2300 |
"epoch": 9.03,
|
| 2301 |
+
"learning_rate": 4.177109440267335e-06,
|
| 2302 |
+
"loss": 0.7831,
|
| 2303 |
+
"step": 3690
|
| 2304 |
},
|
| 2305 |
{
|
| 2306 |
"epoch": 9.03,
|
| 2307 |
+
"learning_rate": 4.03787245892509e-06,
|
| 2308 |
+
"loss": 0.001,
|
| 2309 |
+
"step": 3700
|
| 2310 |
},
|
| 2311 |
{
|
| 2312 |
"epoch": 9.03,
|
| 2313 |
+
"learning_rate": 3.898635477582846e-06,
|
| 2314 |
+
"loss": 0.0011,
|
| 2315 |
+
"step": 3710
|
| 2316 |
},
|
| 2317 |
{
|
| 2318 |
"epoch": 9.03,
|
| 2319 |
+
"learning_rate": 3.7593984962406014e-06,
|
| 2320 |
+
"loss": 0.4288,
|
| 2321 |
+
"step": 3720
|
| 2322 |
},
|
| 2323 |
{
|
| 2324 |
"epoch": 9.04,
|
| 2325 |
+
"learning_rate": 3.620161514898357e-06,
|
| 2326 |
+
"loss": 0.001,
|
| 2327 |
+
"step": 3730
|
| 2328 |
},
|
| 2329 |
{
|
| 2330 |
"epoch": 9.04,
|
| 2331 |
+
"learning_rate": 3.4809245335561126e-06,
|
| 2332 |
+
"loss": 0.001,
|
| 2333 |
+
"step": 3740
|
| 2334 |
},
|
| 2335 |
{
|
| 2336 |
"epoch": 9.04,
|
| 2337 |
+
"learning_rate": 3.341687552213868e-06,
|
| 2338 |
+
"loss": 0.0012,
|
| 2339 |
+
"step": 3750
|
| 2340 |
+
},
|
| 2341 |
+
{
|
| 2342 |
+
"epoch": 9.04,
|
| 2343 |
+
"learning_rate": 3.2024505708716235e-06,
|
| 2344 |
+
"loss": 0.0012,
|
| 2345 |
+
"step": 3760
|
| 2346 |
},
|
| 2347 |
{
|
| 2348 |
"epoch": 9.05,
|
| 2349 |
+
"learning_rate": 3.063213589529379e-06,
|
| 2350 |
+
"loss": 0.0227,
|
| 2351 |
+
"step": 3770
|
| 2352 |
},
|
| 2353 |
{
|
| 2354 |
"epoch": 9.05,
|
| 2355 |
+
"learning_rate": 2.9239766081871343e-06,
|
| 2356 |
+
"loss": 0.0012,
|
| 2357 |
+
"step": 3780
|
| 2358 |
},
|
| 2359 |
{
|
| 2360 |
"epoch": 9.05,
|
| 2361 |
+
"learning_rate": 2.78473962684489e-06,
|
| 2362 |
+
"loss": 0.001,
|
| 2363 |
+
"step": 3790
|
| 2364 |
},
|
| 2365 |
{
|
| 2366 |
"epoch": 9.05,
|
| 2367 |
+
"learning_rate": 2.6455026455026455e-06,
|
| 2368 |
+
"loss": 0.001,
|
| 2369 |
+
"step": 3800
|
| 2370 |
},
|
| 2371 |
{
|
| 2372 |
"epoch": 9.06,
|
| 2373 |
+
"learning_rate": 2.506265664160401e-06,
|
| 2374 |
+
"loss": 0.0014,
|
| 2375 |
+
"step": 3810
|
| 2376 |
},
|
| 2377 |
{
|
| 2378 |
"epoch": 9.06,
|
| 2379 |
+
"learning_rate": 2.3670286828181563e-06,
|
| 2380 |
+
"loss": 0.0012,
|
| 2381 |
+
"step": 3820
|
| 2382 |
},
|
| 2383 |
{
|
| 2384 |
"epoch": 9.06,
|
| 2385 |
+
"learning_rate": 2.227791701475912e-06,
|
| 2386 |
+
"loss": 0.0016,
|
| 2387 |
+
"step": 3830
|
| 2388 |
},
|
| 2389 |
{
|
| 2390 |
"epoch": 9.06,
|
| 2391 |
+
"learning_rate": 2.0885547201336676e-06,
|
| 2392 |
+
"loss": 0.0011,
|
| 2393 |
+
"step": 3840
|
| 2394 |
},
|
| 2395 |
{
|
| 2396 |
"epoch": 9.07,
|
| 2397 |
+
"learning_rate": 1.949317738791423e-06,
|
| 2398 |
+
"loss": 0.2977,
|
| 2399 |
+
"step": 3850
|
| 2400 |
},
|
| 2401 |
{
|
| 2402 |
"epoch": 9.07,
|
| 2403 |
+
"learning_rate": 1.8100807574491784e-06,
|
| 2404 |
+
"loss": 0.0011,
|
| 2405 |
+
"step": 3860
|
| 2406 |
},
|
| 2407 |
{
|
| 2408 |
"epoch": 9.07,
|
| 2409 |
+
"learning_rate": 1.670843776106934e-06,
|
| 2410 |
+
"loss": 0.0012,
|
| 2411 |
+
"step": 3870
|
| 2412 |
},
|
| 2413 |
{
|
| 2414 |
+
"epoch": 9.07,
|
| 2415 |
+
"learning_rate": 1.5316067947646894e-06,
|
| 2416 |
+
"loss": 0.0011,
|
| 2417 |
+
"step": 3880
|
| 2418 |
},
|
| 2419 |
{
|
| 2420 |
"epoch": 9.08,
|
| 2421 |
+
"learning_rate": 1.392369813422445e-06,
|
| 2422 |
+
"loss": 0.8924,
|
| 2423 |
+
"step": 3890
|
| 2424 |
},
|
| 2425 |
{
|
| 2426 |
"epoch": 9.08,
|
| 2427 |
+
"learning_rate": 1.2531328320802005e-06,
|
| 2428 |
+
"loss": 0.0011,
|
| 2429 |
+
"step": 3900
|
| 2430 |
},
|
| 2431 |
{
|
| 2432 |
"epoch": 9.08,
|
| 2433 |
+
"learning_rate": 1.113895850737956e-06,
|
| 2434 |
+
"loss": 0.0269,
|
| 2435 |
+
"step": 3910
|
| 2436 |
},
|
| 2437 |
{
|
| 2438 |
+
"epoch": 9.08,
|
| 2439 |
+
"learning_rate": 9.746588693957115e-07,
|
| 2440 |
+
"loss": 0.0018,
|
| 2441 |
+
"step": 3920
|
| 2442 |
},
|
| 2443 |
{
|
| 2444 |
"epoch": 9.09,
|
| 2445 |
+
"learning_rate": 8.35421888053467e-07,
|
| 2446 |
+
"loss": 0.001,
|
| 2447 |
+
"step": 3930
|
| 2448 |
},
|
| 2449 |
{
|
| 2450 |
"epoch": 9.09,
|
| 2451 |
+
"learning_rate": 6.961849067112225e-07,
|
| 2452 |
+
"loss": 0.0015,
|
| 2453 |
+
"step": 3940
|
| 2454 |
},
|
| 2455 |
{
|
| 2456 |
"epoch": 9.09,
|
| 2457 |
+
"learning_rate": 5.56947925368978e-07,
|
| 2458 |
+
"loss": 0.0011,
|
| 2459 |
+
"step": 3950
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 2460 |
},
|
| 2461 |
{
|
| 2462 |
+
"epoch": 9.09,
|
| 2463 |
+
"learning_rate": 4.177109440267335e-07,
|
| 2464 |
+
"loss": 0.0012,
|
| 2465 |
+
"step": 3960
|
| 2466 |
},
|
| 2467 |
{
|
| 2468 |
"epoch": 9.1,
|
| 2469 |
+
"learning_rate": 2.78473962684489e-07,
|
| 2470 |
+
"loss": 0.0013,
|
| 2471 |
+
"step": 3970
|
|
|
|
|
|
|
|
|
|
| 2472 |
},
|
| 2473 |
{
|
| 2474 |
"epoch": 9.1,
|
| 2475 |
+
"learning_rate": 1.392369813422445e-07,
|
| 2476 |
+
"loss": 0.0011,
|
| 2477 |
+
"step": 3980
|
|
|
|
|
|
|
|
|
|
| 2478 |
},
|
| 2479 |
{
|
| 2480 |
"epoch": 9.1,
|
| 2481 |
+
"eval_accuracy": 0.78,
|
| 2482 |
+
"eval_loss": 1.5902831554412842,
|
| 2483 |
+
"eval_runtime": 28.5194,
|
| 2484 |
+
"eval_samples_per_second": 3.506,
|
| 2485 |
+
"eval_steps_per_second": 3.506,
|
| 2486 |
+
"step": 3980
|
| 2487 |
},
|
| 2488 |
{
|
| 2489 |
+
"epoch": 10.0,
|
| 2490 |
+
"learning_rate": 0.0,
|
| 2491 |
+
"loss": 0.001,
|
| 2492 |
+
"step": 3990
|
| 2493 |
+
},
|
| 2494 |
+
{
|
| 2495 |
+
"epoch": 10.0,
|
| 2496 |
+
"eval_accuracy": 0.78,
|
| 2497 |
+
"eval_loss": 1.5902682542800903,
|
| 2498 |
+
"eval_runtime": 30.2846,
|
| 2499 |
+
"eval_samples_per_second": 3.302,
|
| 2500 |
+
"eval_steps_per_second": 3.302,
|
| 2501 |
+
"step": 3990
|
| 2502 |
+
},
|
| 2503 |
+
{
|
| 2504 |
+
"epoch": 10.0,
|
| 2505 |
+
"step": 3990,
|
| 2506 |
+
"total_flos": 4.972285443087729e+18,
|
| 2507 |
+
"train_loss": 0.6824237227138776,
|
| 2508 |
+
"train_runtime": 3124.1401,
|
| 2509 |
+
"train_samples_per_second": 1.277,
|
| 2510 |
+
"train_steps_per_second": 1.277
|
| 2511 |
+
},
|
| 2512 |
+
{
|
| 2513 |
+
"epoch": 10.0,
|
| 2514 |
+
"eval_accuracy": 0.8324022346368715,
|
| 2515 |
+
"eval_loss": 0.8124022483825684,
|
| 2516 |
+
"eval_runtime": 57.24,
|
| 2517 |
+
"eval_samples_per_second": 3.127,
|
| 2518 |
+
"eval_steps_per_second": 3.127,
|
| 2519 |
+
"step": 3990
|
| 2520 |
+
},
|
| 2521 |
+
{
|
| 2522 |
+
"epoch": 10.0,
|
| 2523 |
+
"eval_accuracy": 0.8324022346368715,
|
| 2524 |
+
"eval_loss": 0.8124023079872131,
|
| 2525 |
+
"eval_runtime": 56.7655,
|
| 2526 |
+
"eval_samples_per_second": 3.153,
|
| 2527 |
+
"eval_steps_per_second": 3.153,
|
| 2528 |
+
"step": 3990
|
| 2529 |
}
|
| 2530 |
],
|
| 2531 |
+
"max_steps": 3990,
|
| 2532 |
"num_train_epochs": 9223372036854775807,
|
| 2533 |
+
"total_flos": 4.972285443087729e+18,
|
| 2534 |
"trial_name": null,
|
| 2535 |
"trial_params": null
|
| 2536 |
}
|