Training in progress, step 500
Browse files- all_results.json +6 -6
- model.safetensors +1 -1
- test_results.json +6 -6
- trainer_state.json +802 -549
- training_args.bin +1 -1
all_results.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch":
|
| 3 |
-
"eval_accuracy": 0.
|
| 4 |
-
"eval_loss": 0.
|
| 5 |
"eval_model_preparation_time": 0.004,
|
| 6 |
-
"eval_runtime": 9.
|
| 7 |
-
"eval_samples_per_second":
|
| 8 |
-
"eval_steps_per_second":
|
| 9 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 177.004375,
|
| 3 |
+
"eval_accuracy": 0.7916666666666666,
|
| 4 |
+
"eval_loss": 0.8197056651115417,
|
| 5 |
"eval_model_preparation_time": 0.004,
|
| 6 |
+
"eval_runtime": 9.9982,
|
| 7 |
+
"eval_samples_per_second": 4.801,
|
| 8 |
+
"eval_steps_per_second": 1.6
|
| 9 |
}
|
model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 344946604
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e4d34433973a95432865c7fef94154165beefdec8adb87bd94d05ed44cf3e26f
|
| 3 |
size 344946604
|
test_results.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch":
|
| 3 |
-
"eval_accuracy": 0.
|
| 4 |
-
"eval_loss": 0.
|
| 5 |
-
"eval_runtime": 9.
|
| 6 |
-
"eval_samples_per_second":
|
| 7 |
-
"eval_steps_per_second":
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 177.004375,
|
| 3 |
+
"eval_accuracy": 0.7916666666666666,
|
| 4 |
+
"eval_loss": 0.8197056651115417,
|
| 5 |
+
"eval_runtime": 9.9982,
|
| 6 |
+
"eval_samples_per_second": 4.801,
|
| 7 |
+
"eval_steps_per_second": 1.6
|
| 8 |
}
|
trainer_state.json
CHANGED
|
@@ -1,931 +1,1184 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": 0.8529411764705882,
|
| 3 |
-
"best_model_checkpoint": "videomae-base-finetuned-ucf101-subset-hand\\checkpoint-
|
| 4 |
-
"epoch":
|
| 5 |
-
"eval_steps":
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
| 10 |
"log_history": [
|
| 11 |
{
|
| 12 |
-
"epoch":
|
| 13 |
-
"grad_norm":
|
| 14 |
-
"learning_rate":
|
| 15 |
-
"loss": 1.
|
| 16 |
"step": 10
|
| 17 |
},
|
| 18 |
{
|
| 19 |
-
"epoch":
|
| 20 |
-
"grad_norm":
|
| 21 |
-
"learning_rate":
|
| 22 |
-
"loss": 1.
|
| 23 |
"step": 20
|
| 24 |
},
|
| 25 |
{
|
| 26 |
-
"epoch":
|
| 27 |
-
"grad_norm": 21.
|
| 28 |
-
"learning_rate":
|
| 29 |
-
"loss": 1.
|
| 30 |
"step": 30
|
| 31 |
},
|
| 32 |
{
|
| 33 |
-
"epoch":
|
| 34 |
-
"grad_norm":
|
| 35 |
-
"learning_rate": 1.
|
| 36 |
-
"loss": 1.
|
| 37 |
"step": 40
|
| 38 |
},
|
| 39 |
{
|
| 40 |
-
"epoch":
|
| 41 |
-
"grad_norm": 15.
|
| 42 |
-
"learning_rate":
|
| 43 |
-
"loss": 1.
|
| 44 |
"step": 50
|
| 45 |
},
|
| 46 |
{
|
| 47 |
-
"epoch":
|
| 48 |
-
"grad_norm":
|
| 49 |
-
"learning_rate":
|
| 50 |
-
"loss": 1.
|
| 51 |
"step": 60
|
| 52 |
},
|
| 53 |
{
|
| 54 |
-
"epoch":
|
| 55 |
-
"grad_norm":
|
| 56 |
-
"learning_rate": 2.
|
| 57 |
-
"loss": 1.
|
| 58 |
"step": 70
|
| 59 |
},
|
| 60 |
{
|
| 61 |
-
"epoch":
|
| 62 |
-
"grad_norm": 10.
|
| 63 |
-
"learning_rate":
|
| 64 |
-
"loss": 1.
|
| 65 |
"step": 80
|
| 66 |
},
|
| 67 |
{
|
| 68 |
-
"epoch":
|
| 69 |
-
"grad_norm": 20.
|
| 70 |
-
"learning_rate":
|
| 71 |
-
"loss": 1.
|
| 72 |
"step": 90
|
| 73 |
},
|
| 74 |
{
|
| 75 |
-
"epoch":
|
| 76 |
-
"grad_norm":
|
| 77 |
-
"learning_rate":
|
| 78 |
-
"loss": 1.
|
| 79 |
"step": 100
|
| 80 |
},
|
| 81 |
{
|
| 82 |
-
"epoch":
|
| 83 |
-
"grad_norm":
|
| 84 |
-
"learning_rate":
|
| 85 |
-
"loss": 1.
|
| 86 |
"step": 110
|
| 87 |
},
|
| 88 |
{
|
| 89 |
-
"epoch":
|
| 90 |
-
"grad_norm":
|
| 91 |
-
"learning_rate":
|
| 92 |
-
"loss": 1.
|
| 93 |
"step": 120
|
| 94 |
},
|
| 95 |
{
|
| 96 |
-
"epoch":
|
| 97 |
-
"grad_norm":
|
| 98 |
-
"learning_rate": 4.
|
| 99 |
-
"loss": 1.
|
| 100 |
"step": 130
|
| 101 |
},
|
| 102 |
{
|
| 103 |
-
"epoch":
|
| 104 |
-
"grad_norm": 10.
|
| 105 |
-
"learning_rate": 4.
|
| 106 |
-
"loss": 1.
|
| 107 |
"step": 140
|
| 108 |
},
|
| 109 |
{
|
| 110 |
-
"epoch":
|
| 111 |
-
"grad_norm":
|
| 112 |
-
"learning_rate": 4.
|
| 113 |
-
"loss": 1.
|
| 114 |
"step": 150
|
| 115 |
},
|
| 116 |
{
|
| 117 |
-
"epoch":
|
| 118 |
-
"grad_norm":
|
| 119 |
-
"learning_rate":
|
| 120 |
-
"loss": 1.
|
| 121 |
"step": 160
|
| 122 |
},
|
| 123 |
{
|
| 124 |
-
"epoch":
|
| 125 |
-
"grad_norm":
|
| 126 |
-
"learning_rate": 4.
|
| 127 |
-
"loss": 1.
|
| 128 |
"step": 170
|
| 129 |
},
|
| 130 |
{
|
| 131 |
-
"epoch":
|
| 132 |
-
"grad_norm":
|
| 133 |
-
"learning_rate": 4.
|
| 134 |
-
"loss": 1.
|
| 135 |
"step": 180
|
| 136 |
},
|
| 137 |
{
|
| 138 |
-
"epoch":
|
| 139 |
-
"grad_norm":
|
| 140 |
-
"learning_rate": 4.
|
| 141 |
-
"loss": 1.
|
| 142 |
"step": 190
|
| 143 |
},
|
| 144 |
{
|
| 145 |
-
"epoch":
|
| 146 |
-
"grad_norm":
|
| 147 |
-
"learning_rate": 4.
|
| 148 |
-
"loss": 1.
|
| 149 |
-
"step": 200
|
| 150 |
-
},
|
| 151 |
-
{
|
| 152 |
-
"epoch": 15.004166666666666,
|
| 153 |
-
"eval_accuracy": 0.17647058823529413,
|
| 154 |
-
"eval_loss": 1.6326980590820312,
|
| 155 |
-
"eval_runtime": 6.293,
|
| 156 |
-
"eval_samples_per_second": 5.403,
|
| 157 |
-
"eval_steps_per_second": 2.701,
|
| 158 |
"step": 200
|
| 159 |
},
|
| 160 |
{
|
| 161 |
-
"epoch":
|
| 162 |
-
"grad_norm":
|
| 163 |
-
"learning_rate": 4.
|
| 164 |
-
"loss": 1.
|
| 165 |
"step": 210
|
| 166 |
},
|
| 167 |
{
|
| 168 |
-
"epoch":
|
| 169 |
-
"grad_norm":
|
| 170 |
-
"learning_rate": 4.
|
| 171 |
-
"loss": 1.
|
| 172 |
"step": 220
|
| 173 |
},
|
| 174 |
{
|
| 175 |
-
"epoch":
|
| 176 |
-
"grad_norm":
|
| 177 |
-
"learning_rate": 4.
|
| 178 |
-
"loss": 1.
|
| 179 |
"step": 230
|
| 180 |
},
|
| 181 |
{
|
| 182 |
-
"epoch":
|
| 183 |
-
"grad_norm":
|
| 184 |
-
"learning_rate": 4.
|
| 185 |
-
"loss": 1.
|
| 186 |
"step": 240
|
| 187 |
},
|
| 188 |
{
|
| 189 |
-
"epoch":
|
| 190 |
-
"grad_norm": 8.
|
| 191 |
-
"learning_rate": 4.
|
| 192 |
-
"loss": 1.
|
| 193 |
"step": 250
|
| 194 |
},
|
| 195 |
{
|
| 196 |
-
"epoch":
|
| 197 |
-
"grad_norm":
|
| 198 |
-
"learning_rate": 4.
|
| 199 |
-
"loss": 1.
|
| 200 |
"step": 260
|
| 201 |
},
|
| 202 |
{
|
| 203 |
-
"epoch":
|
| 204 |
-
"grad_norm":
|
| 205 |
-
"learning_rate": 4.
|
| 206 |
-
"loss": 1.
|
| 207 |
"step": 270
|
| 208 |
},
|
| 209 |
{
|
| 210 |
-
"epoch":
|
| 211 |
-
"grad_norm":
|
| 212 |
-
"learning_rate": 4.
|
| 213 |
-
"loss": 1.
|
| 214 |
"step": 280
|
| 215 |
},
|
| 216 |
{
|
| 217 |
-
"epoch":
|
| 218 |
-
"grad_norm":
|
| 219 |
-
"learning_rate": 4.
|
| 220 |
-
"loss": 1.
|
| 221 |
"step": 290
|
| 222 |
},
|
| 223 |
{
|
| 224 |
-
"epoch":
|
| 225 |
-
"grad_norm":
|
| 226 |
-
"learning_rate": 4.
|
| 227 |
-
"loss": 1.
|
| 228 |
"step": 300
|
| 229 |
},
|
| 230 |
{
|
| 231 |
-
"epoch":
|
| 232 |
-
"grad_norm": 9.
|
| 233 |
-
"learning_rate": 4.
|
| 234 |
-
"loss": 1.
|
| 235 |
"step": 310
|
| 236 |
},
|
| 237 |
{
|
| 238 |
-
"epoch":
|
| 239 |
-
"grad_norm":
|
| 240 |
-
"learning_rate": 4.
|
| 241 |
-
"loss": 1.
|
| 242 |
"step": 320
|
| 243 |
},
|
| 244 |
{
|
| 245 |
-
"epoch":
|
| 246 |
-
"grad_norm":
|
| 247 |
-
"learning_rate": 4.
|
| 248 |
-
"loss": 1.
|
| 249 |
"step": 330
|
| 250 |
},
|
| 251 |
{
|
| 252 |
-
"epoch":
|
| 253 |
-
"grad_norm":
|
| 254 |
-
"learning_rate":
|
| 255 |
-
"loss": 1.
|
| 256 |
"step": 340
|
| 257 |
},
|
| 258 |
{
|
| 259 |
-
"epoch":
|
| 260 |
-
"grad_norm": 9.
|
| 261 |
-
"learning_rate":
|
| 262 |
-
"loss": 1.
|
| 263 |
"step": 350
|
| 264 |
},
|
| 265 |
{
|
| 266 |
-
"epoch":
|
| 267 |
-
"grad_norm":
|
| 268 |
-
"learning_rate":
|
| 269 |
-
"loss": 1.
|
| 270 |
"step": 360
|
| 271 |
},
|
| 272 |
{
|
| 273 |
-
"epoch":
|
| 274 |
-
"grad_norm": 8.
|
| 275 |
-
"learning_rate":
|
| 276 |
-
"loss": 1.
|
| 277 |
"step": 370
|
| 278 |
},
|
| 279 |
{
|
| 280 |
-
"epoch":
|
| 281 |
-
"grad_norm":
|
| 282 |
-
"learning_rate":
|
| 283 |
-
"loss": 1.
|
| 284 |
"step": 380
|
| 285 |
},
|
| 286 |
{
|
| 287 |
-
"epoch":
|
| 288 |
-
"grad_norm":
|
| 289 |
-
"learning_rate":
|
| 290 |
-
"loss": 1.
|
| 291 |
"step": 390
|
| 292 |
},
|
| 293 |
{
|
| 294 |
-
"epoch":
|
| 295 |
-
"grad_norm":
|
| 296 |
-
"learning_rate":
|
| 297 |
-
"loss": 1.
|
| 298 |
-
"step": 400
|
| 299 |
-
},
|
| 300 |
-
{
|
| 301 |
-
"epoch": 30.008333333333333,
|
| 302 |
-
"eval_accuracy": 0.23529411764705882,
|
| 303 |
-
"eval_loss": 1.6021785736083984,
|
| 304 |
-
"eval_runtime": 6.8312,
|
| 305 |
-
"eval_samples_per_second": 4.977,
|
| 306 |
-
"eval_steps_per_second": 2.489,
|
| 307 |
"step": 400
|
| 308 |
},
|
| 309 |
{
|
| 310 |
-
"epoch":
|
| 311 |
-
"grad_norm":
|
| 312 |
-
"learning_rate":
|
| 313 |
-
"loss": 1.
|
| 314 |
"step": 410
|
| 315 |
},
|
| 316 |
{
|
| 317 |
-
"epoch":
|
| 318 |
-
"grad_norm":
|
| 319 |
-
"learning_rate":
|
| 320 |
-
"loss": 1.
|
| 321 |
"step": 420
|
| 322 |
},
|
| 323 |
{
|
| 324 |
-
"epoch":
|
| 325 |
-
"grad_norm":
|
| 326 |
-
"learning_rate":
|
| 327 |
-
"loss": 1.
|
| 328 |
"step": 430
|
| 329 |
},
|
| 330 |
{
|
| 331 |
-
"epoch":
|
| 332 |
-
"grad_norm":
|
| 333 |
-
"learning_rate":
|
| 334 |
-
"loss": 1.
|
| 335 |
"step": 440
|
| 336 |
},
|
| 337 |
{
|
| 338 |
-
"epoch":
|
| 339 |
-
"grad_norm":
|
| 340 |
-
"learning_rate": 3.
|
| 341 |
-
"loss": 1.
|
| 342 |
"step": 450
|
| 343 |
},
|
| 344 |
{
|
| 345 |
-
"epoch":
|
| 346 |
-
"grad_norm":
|
| 347 |
-
"learning_rate": 3.
|
| 348 |
-
"loss": 1.
|
| 349 |
"step": 460
|
| 350 |
},
|
| 351 |
{
|
| 352 |
-
"epoch":
|
| 353 |
-
"grad_norm":
|
| 354 |
-
"learning_rate": 3.
|
| 355 |
-
"loss": 1.
|
| 356 |
"step": 470
|
| 357 |
},
|
| 358 |
{
|
| 359 |
-
"epoch":
|
| 360 |
-
"grad_norm":
|
| 361 |
-
"learning_rate": 3.
|
| 362 |
-
"loss": 1.
|
| 363 |
"step": 480
|
| 364 |
},
|
| 365 |
{
|
| 366 |
-
"epoch":
|
| 367 |
-
"grad_norm":
|
| 368 |
-
"learning_rate": 3.
|
| 369 |
-
"loss": 1.
|
| 370 |
"step": 490
|
| 371 |
},
|
| 372 |
{
|
| 373 |
-
"epoch":
|
| 374 |
-
"grad_norm":
|
| 375 |
-
"learning_rate": 3.
|
| 376 |
-
"loss":
|
| 377 |
"step": 500
|
| 378 |
},
|
| 379 |
{
|
| 380 |
-
"epoch":
|
| 381 |
-
"
|
| 382 |
-
"
|
| 383 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 384 |
"step": 510
|
| 385 |
},
|
| 386 |
{
|
| 387 |
-
"epoch":
|
| 388 |
-
"grad_norm":
|
| 389 |
-
"learning_rate": 3.
|
| 390 |
-
"loss": 1.
|
| 391 |
"step": 520
|
| 392 |
},
|
| 393 |
{
|
| 394 |
-
"epoch":
|
| 395 |
-
"grad_norm":
|
| 396 |
-
"learning_rate": 3.
|
| 397 |
-
"loss": 1.
|
| 398 |
"step": 530
|
| 399 |
},
|
| 400 |
{
|
| 401 |
-
"epoch":
|
| 402 |
-
"grad_norm":
|
| 403 |
-
"learning_rate": 3.
|
| 404 |
-
"loss": 1.
|
| 405 |
"step": 540
|
| 406 |
},
|
| 407 |
{
|
| 408 |
-
"epoch":
|
| 409 |
-
"grad_norm":
|
| 410 |
-
"learning_rate": 3.
|
| 411 |
-
"loss": 1.
|
| 412 |
"step": 550
|
| 413 |
},
|
| 414 |
{
|
| 415 |
-
"epoch":
|
| 416 |
-
"grad_norm":
|
| 417 |
-
"learning_rate":
|
| 418 |
-
"loss":
|
| 419 |
"step": 560
|
| 420 |
},
|
| 421 |
{
|
| 422 |
-
"epoch":
|
| 423 |
-
"grad_norm":
|
| 424 |
-
"learning_rate":
|
| 425 |
-
"loss":
|
| 426 |
"step": 570
|
| 427 |
},
|
| 428 |
{
|
| 429 |
-
"epoch":
|
| 430 |
-
"grad_norm":
|
| 431 |
-
"learning_rate":
|
| 432 |
-
"loss":
|
| 433 |
"step": 580
|
| 434 |
},
|
| 435 |
{
|
| 436 |
-
"epoch":
|
| 437 |
-
"grad_norm":
|
| 438 |
-
"learning_rate":
|
| 439 |
-
"loss":
|
| 440 |
"step": 590
|
| 441 |
},
|
| 442 |
{
|
| 443 |
-
"epoch":
|
| 444 |
-
"grad_norm":
|
| 445 |
-
"learning_rate":
|
| 446 |
-
"loss":
|
| 447 |
-
"step": 600
|
| 448 |
-
},
|
| 449 |
-
{
|
| 450 |
-
"epoch": 46.001666666666665,
|
| 451 |
-
"eval_accuracy": 0.38235294117647056,
|
| 452 |
-
"eval_loss": 1.3752310276031494,
|
| 453 |
-
"eval_runtime": 6.3667,
|
| 454 |
-
"eval_samples_per_second": 5.34,
|
| 455 |
-
"eval_steps_per_second": 2.67,
|
| 456 |
"step": 600
|
| 457 |
},
|
| 458 |
{
|
| 459 |
-
"epoch":
|
| 460 |
-
"grad_norm":
|
| 461 |
-
"learning_rate":
|
| 462 |
-
"loss": 1.
|
| 463 |
"step": 610
|
| 464 |
},
|
| 465 |
{
|
| 466 |
-
"epoch":
|
| 467 |
-
"grad_norm":
|
| 468 |
-
"learning_rate":
|
| 469 |
-
"loss":
|
| 470 |
"step": 620
|
| 471 |
},
|
| 472 |
{
|
| 473 |
-
"epoch":
|
| 474 |
-
"grad_norm":
|
| 475 |
-
"learning_rate":
|
| 476 |
-
"loss":
|
| 477 |
"step": 630
|
| 478 |
},
|
| 479 |
{
|
| 480 |
-
"epoch":
|
| 481 |
-
"grad_norm":
|
| 482 |
-
"learning_rate":
|
| 483 |
-
"loss":
|
| 484 |
"step": 640
|
| 485 |
},
|
| 486 |
{
|
| 487 |
-
"epoch":
|
| 488 |
-
"grad_norm":
|
| 489 |
-
"learning_rate":
|
| 490 |
-
"loss": 1.
|
| 491 |
"step": 650
|
| 492 |
},
|
| 493 |
{
|
| 494 |
-
"epoch":
|
| 495 |
-
"grad_norm":
|
| 496 |
-
"learning_rate":
|
| 497 |
-
"loss":
|
| 498 |
"step": 660
|
| 499 |
},
|
| 500 |
{
|
| 501 |
-
"epoch":
|
| 502 |
-
"grad_norm":
|
| 503 |
-
"learning_rate":
|
| 504 |
-
"loss": 1.
|
| 505 |
"step": 670
|
| 506 |
},
|
| 507 |
{
|
| 508 |
-
"epoch":
|
| 509 |
-
"grad_norm":
|
| 510 |
-
"learning_rate":
|
| 511 |
-
"loss":
|
| 512 |
"step": 680
|
| 513 |
},
|
| 514 |
{
|
| 515 |
-
"epoch":
|
| 516 |
-
"grad_norm":
|
| 517 |
-
"learning_rate":
|
| 518 |
-
"loss":
|
| 519 |
"step": 690
|
| 520 |
},
|
| 521 |
{
|
| 522 |
-
"epoch":
|
| 523 |
-
"grad_norm":
|
| 524 |
-
"learning_rate":
|
| 525 |
-
"loss":
|
| 526 |
"step": 700
|
| 527 |
},
|
| 528 |
{
|
| 529 |
-
"epoch":
|
| 530 |
-
"grad_norm":
|
| 531 |
-
"learning_rate":
|
| 532 |
-
"loss":
|
| 533 |
"step": 710
|
| 534 |
},
|
| 535 |
{
|
| 536 |
-
"epoch":
|
| 537 |
-
"grad_norm":
|
| 538 |
-
"learning_rate":
|
| 539 |
-
"loss":
|
| 540 |
"step": 720
|
| 541 |
},
|
| 542 |
{
|
| 543 |
-
"epoch":
|
| 544 |
-
"grad_norm":
|
| 545 |
-
"learning_rate":
|
| 546 |
-
"loss":
|
| 547 |
"step": 730
|
| 548 |
},
|
| 549 |
{
|
| 550 |
-
"epoch":
|
| 551 |
-
"grad_norm":
|
| 552 |
-
"learning_rate": 2.
|
| 553 |
-
"loss":
|
| 554 |
"step": 740
|
| 555 |
},
|
| 556 |
{
|
| 557 |
-
"epoch":
|
| 558 |
-
"grad_norm":
|
| 559 |
-
"learning_rate": 2.
|
| 560 |
-
"loss":
|
| 561 |
"step": 750
|
| 562 |
},
|
| 563 |
{
|
| 564 |
-
"epoch":
|
| 565 |
-
"grad_norm":
|
| 566 |
-
"learning_rate": 2.
|
| 567 |
-
"loss": 0.
|
| 568 |
"step": 760
|
| 569 |
},
|
| 570 |
{
|
| 571 |
-
"epoch":
|
| 572 |
-
"grad_norm":
|
| 573 |
-
"learning_rate":
|
| 574 |
-
"loss":
|
| 575 |
"step": 770
|
| 576 |
},
|
| 577 |
{
|
| 578 |
-
"epoch":
|
| 579 |
-
"grad_norm":
|
| 580 |
-
"learning_rate":
|
| 581 |
-
"loss": 0.
|
| 582 |
"step": 780
|
| 583 |
},
|
| 584 |
{
|
| 585 |
-
"epoch":
|
| 586 |
-
"grad_norm":
|
| 587 |
-
"learning_rate":
|
| 588 |
-
"loss": 0.
|
| 589 |
"step": 790
|
| 590 |
},
|
| 591 |
{
|
| 592 |
-
"epoch":
|
| 593 |
-
"grad_norm":
|
| 594 |
-
"learning_rate":
|
| 595 |
-
"loss": 0.
|
| 596 |
-
"step": 800
|
| 597 |
-
},
|
| 598 |
-
{
|
| 599 |
-
"epoch": 61.005833333333335,
|
| 600 |
-
"eval_accuracy": 0.35294117647058826,
|
| 601 |
-
"eval_loss": 1.4181767702102661,
|
| 602 |
-
"eval_runtime": 6.3183,
|
| 603 |
-
"eval_samples_per_second": 5.381,
|
| 604 |
-
"eval_steps_per_second": 2.691,
|
| 605 |
"step": 800
|
| 606 |
},
|
| 607 |
{
|
| 608 |
-
"epoch":
|
| 609 |
-
"grad_norm":
|
| 610 |
-
"learning_rate":
|
| 611 |
-
"loss":
|
| 612 |
"step": 810
|
| 613 |
},
|
| 614 |
{
|
| 615 |
-
"epoch":
|
| 616 |
-
"grad_norm":
|
| 617 |
-
"learning_rate":
|
| 618 |
-
"loss":
|
| 619 |
"step": 820
|
| 620 |
},
|
| 621 |
{
|
| 622 |
-
"epoch":
|
| 623 |
-
"grad_norm":
|
| 624 |
-
"learning_rate":
|
| 625 |
-
"loss": 0.
|
| 626 |
"step": 830
|
| 627 |
},
|
| 628 |
{
|
| 629 |
-
"epoch":
|
| 630 |
-
"grad_norm":
|
| 631 |
-
"learning_rate":
|
| 632 |
-
"loss": 0.
|
| 633 |
"step": 840
|
| 634 |
},
|
| 635 |
{
|
| 636 |
-
"epoch":
|
| 637 |
-
"grad_norm":
|
| 638 |
-
"learning_rate":
|
| 639 |
-
"loss": 0.
|
| 640 |
"step": 850
|
| 641 |
},
|
| 642 |
{
|
| 643 |
-
"epoch":
|
| 644 |
-
"grad_norm":
|
| 645 |
-
"learning_rate":
|
| 646 |
-
"loss":
|
| 647 |
"step": 860
|
| 648 |
},
|
| 649 |
{
|
| 650 |
-
"epoch":
|
| 651 |
-
"grad_norm":
|
| 652 |
-
"learning_rate":
|
| 653 |
-
"loss":
|
| 654 |
"step": 870
|
| 655 |
},
|
| 656 |
{
|
| 657 |
-
"epoch":
|
| 658 |
-
"grad_norm":
|
| 659 |
-
"learning_rate":
|
| 660 |
-
"loss": 0.
|
| 661 |
"step": 880
|
| 662 |
},
|
| 663 |
{
|
| 664 |
-
"epoch":
|
| 665 |
-
"grad_norm":
|
| 666 |
-
"learning_rate":
|
| 667 |
-
"loss": 0.
|
| 668 |
"step": 890
|
| 669 |
},
|
| 670 |
{
|
| 671 |
-
"epoch":
|
| 672 |
-
"grad_norm":
|
| 673 |
-
"learning_rate":
|
| 674 |
-
"loss": 0.
|
| 675 |
"step": 900
|
| 676 |
},
|
| 677 |
{
|
| 678 |
-
"epoch":
|
| 679 |
-
"grad_norm":
|
| 680 |
-
"learning_rate":
|
| 681 |
-
"loss":
|
| 682 |
"step": 910
|
| 683 |
},
|
| 684 |
{
|
| 685 |
-
"epoch":
|
| 686 |
-
"grad_norm":
|
| 687 |
-
"learning_rate":
|
| 688 |
-
"loss": 0.
|
| 689 |
"step": 920
|
| 690 |
},
|
| 691 |
{
|
| 692 |
-
"epoch":
|
| 693 |
-
"grad_norm":
|
| 694 |
-
"learning_rate":
|
| 695 |
-
"loss": 0.
|
| 696 |
"step": 930
|
| 697 |
},
|
| 698 |
{
|
| 699 |
-
"epoch":
|
| 700 |
-
"grad_norm":
|
| 701 |
-
"learning_rate":
|
| 702 |
-
"loss": 0.
|
| 703 |
"step": 940
|
| 704 |
},
|
| 705 |
{
|
| 706 |
-
"epoch":
|
| 707 |
-
"grad_norm": 0.
|
| 708 |
-
"learning_rate":
|
| 709 |
-
"loss": 0.
|
| 710 |
"step": 950
|
| 711 |
},
|
| 712 |
{
|
| 713 |
-
"epoch":
|
| 714 |
-
"grad_norm":
|
| 715 |
-
"learning_rate":
|
| 716 |
-
"loss": 0.
|
| 717 |
"step": 960
|
| 718 |
},
|
| 719 |
{
|
| 720 |
-
"epoch":
|
| 721 |
-
"grad_norm":
|
| 722 |
-
"learning_rate":
|
| 723 |
-
"loss": 0.
|
| 724 |
"step": 970
|
| 725 |
},
|
| 726 |
{
|
| 727 |
-
"epoch":
|
| 728 |
-
"grad_norm":
|
| 729 |
-
"learning_rate":
|
| 730 |
-
"loss": 0.
|
| 731 |
"step": 980
|
| 732 |
},
|
| 733 |
{
|
| 734 |
-
"epoch":
|
| 735 |
-
"grad_norm": 0.
|
| 736 |
-
"learning_rate":
|
| 737 |
-
"loss": 0.
|
| 738 |
"step": 990
|
| 739 |
},
|
| 740 |
{
|
| 741 |
-
"epoch":
|
| 742 |
-
"grad_norm":
|
| 743 |
-
"learning_rate":
|
| 744 |
-
"loss": 0.
|
| 745 |
"step": 1000
|
| 746 |
},
|
| 747 |
{
|
| 748 |
-
"epoch":
|
| 749 |
"eval_accuracy": 0.8235294117647058,
|
| 750 |
-
"eval_loss": 0.
|
| 751 |
-
"eval_runtime": 6.
|
| 752 |
-
"eval_samples_per_second": 5.
|
| 753 |
-
"eval_steps_per_second":
|
| 754 |
"step": 1000
|
| 755 |
},
|
| 756 |
{
|
| 757 |
-
"epoch":
|
| 758 |
-
"grad_norm":
|
| 759 |
-
"learning_rate":
|
| 760 |
-
"loss": 0.
|
| 761 |
"step": 1010
|
| 762 |
},
|
| 763 |
{
|
| 764 |
-
"epoch":
|
| 765 |
-
"grad_norm": 0.
|
| 766 |
-
"learning_rate":
|
| 767 |
-
"loss": 0.
|
| 768 |
"step": 1020
|
| 769 |
},
|
| 770 |
{
|
| 771 |
-
"epoch":
|
| 772 |
-
"grad_norm":
|
| 773 |
-
"learning_rate":
|
| 774 |
-
"loss": 0.
|
| 775 |
"step": 1030
|
| 776 |
},
|
| 777 |
{
|
| 778 |
-
"epoch":
|
| 779 |
-
"grad_norm": 0.
|
| 780 |
-
"learning_rate":
|
| 781 |
-
"loss": 0.
|
| 782 |
"step": 1040
|
| 783 |
},
|
| 784 |
{
|
| 785 |
-
"epoch":
|
| 786 |
-
"grad_norm":
|
| 787 |
-
"learning_rate":
|
| 788 |
-
"loss": 0.
|
| 789 |
"step": 1050
|
| 790 |
},
|
| 791 |
{
|
| 792 |
-
"epoch":
|
| 793 |
-
"grad_norm":
|
| 794 |
-
"learning_rate":
|
| 795 |
-
"loss": 0.
|
| 796 |
"step": 1060
|
| 797 |
},
|
| 798 |
{
|
| 799 |
-
"epoch":
|
| 800 |
-
"grad_norm": 0.
|
| 801 |
-
"learning_rate":
|
| 802 |
-
"loss": 0.
|
| 803 |
"step": 1070
|
| 804 |
},
|
| 805 |
{
|
| 806 |
-
"epoch":
|
| 807 |
-
"grad_norm": 0.
|
| 808 |
-
"learning_rate":
|
| 809 |
-
"loss": 0.
|
| 810 |
"step": 1080
|
| 811 |
},
|
| 812 |
{
|
| 813 |
-
"epoch":
|
| 814 |
-
"grad_norm": 0.
|
| 815 |
-
"learning_rate":
|
| 816 |
-
"loss": 0.
|
| 817 |
"step": 1090
|
| 818 |
},
|
| 819 |
{
|
| 820 |
-
"epoch":
|
| 821 |
-
"grad_norm":
|
| 822 |
-
"learning_rate":
|
| 823 |
-
"loss": 0.
|
| 824 |
"step": 1100
|
| 825 |
},
|
| 826 |
{
|
| 827 |
-
"epoch":
|
| 828 |
-
"grad_norm":
|
| 829 |
-
"learning_rate":
|
| 830 |
-
"loss": 0.
|
| 831 |
"step": 1110
|
| 832 |
},
|
| 833 |
{
|
| 834 |
-
"epoch":
|
| 835 |
-
"grad_norm":
|
| 836 |
-
"learning_rate":
|
| 837 |
-
"loss": 0.
|
| 838 |
"step": 1120
|
| 839 |
},
|
| 840 |
{
|
| 841 |
-
"epoch":
|
| 842 |
-
"grad_norm":
|
| 843 |
-
"learning_rate":
|
| 844 |
-
"loss": 0.
|
| 845 |
"step": 1130
|
| 846 |
},
|
| 847 |
{
|
| 848 |
-
"epoch":
|
| 849 |
-
"grad_norm":
|
| 850 |
-
"learning_rate":
|
| 851 |
-
"loss": 0.
|
| 852 |
"step": 1140
|
| 853 |
},
|
| 854 |
{
|
| 855 |
-
"epoch":
|
| 856 |
-
"grad_norm": 0.
|
| 857 |
-
"learning_rate":
|
| 858 |
-
"loss": 0.
|
| 859 |
"step": 1150
|
| 860 |
},
|
| 861 |
{
|
| 862 |
-
"epoch":
|
| 863 |
-
"grad_norm":
|
| 864 |
-
"learning_rate": 1.
|
| 865 |
-
"loss": 0.
|
| 866 |
"step": 1160
|
| 867 |
},
|
| 868 |
{
|
| 869 |
-
"epoch":
|
| 870 |
-
"grad_norm":
|
| 871 |
-
"learning_rate": 1.
|
| 872 |
-
"loss": 0.
|
| 873 |
"step": 1170
|
| 874 |
},
|
| 875 |
{
|
| 876 |
-
"epoch":
|
| 877 |
-
"grad_norm":
|
| 878 |
-
"learning_rate":
|
| 879 |
-
"loss": 0.
|
| 880 |
"step": 1180
|
| 881 |
},
|
| 882 |
{
|
| 883 |
-
"epoch":
|
| 884 |
-
"grad_norm": 0.
|
| 885 |
-
"learning_rate":
|
| 886 |
-
"loss": 0.
|
| 887 |
"step": 1190
|
| 888 |
},
|
| 889 |
{
|
| 890 |
-
"epoch":
|
| 891 |
-
"grad_norm": 0.
|
| 892 |
-
"learning_rate":
|
| 893 |
-
"loss": 0.
|
| 894 |
"step": 1200
|
| 895 |
},
|
| 896 |
{
|
| 897 |
-
"epoch":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 898 |
"eval_accuracy": 0.8529411764705882,
|
| 899 |
-
"eval_loss": 0.
|
| 900 |
-
"eval_runtime": 6.
|
| 901 |
-
"eval_samples_per_second": 5.
|
| 902 |
-
"eval_steps_per_second":
|
| 903 |
-
"step":
|
| 904 |
},
|
| 905 |
{
|
| 906 |
-
"epoch":
|
| 907 |
-
"
|
| 908 |
-
"
|
| 909 |
-
"
|
| 910 |
-
"
|
| 911 |
-
"train_samples_per_second": 3.874,
|
| 912 |
-
"train_steps_per_second": 1.937
|
| 913 |
},
|
| 914 |
{
|
| 915 |
-
"epoch":
|
| 916 |
-
"
|
| 917 |
-
"
|
| 918 |
-
"
|
| 919 |
-
"
|
| 920 |
-
|
| 921 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 922 |
}
|
| 923 |
],
|
| 924 |
"logging_steps": 10,
|
| 925 |
-
"max_steps":
|
| 926 |
"num_input_tokens_seen": 0,
|
| 927 |
"num_train_epochs": 9223372036854775807,
|
| 928 |
-
"save_steps":
|
| 929 |
"stateful_callbacks": {
|
| 930 |
"TrainerControl": {
|
| 931 |
"args": {
|
|
@@ -938,8 +1191,8 @@
|
|
| 938 |
"attributes": {}
|
| 939 |
}
|
| 940 |
},
|
| 941 |
-
"total_flos":
|
| 942 |
-
"train_batch_size":
|
| 943 |
"trial_name": null,
|
| 944 |
"trial_params": null
|
| 945 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": 0.8529411764705882,
|
| 3 |
+
"best_model_checkpoint": "videomae-base-finetuned-ucf101-subset-hand\\checkpoint-1500",
|
| 4 |
+
"epoch": 177.004375,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 1600,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
| 10 |
"log_history": [
|
| 11 |
{
|
| 12 |
+
"epoch": 1.000625,
|
| 13 |
+
"grad_norm": 15.27369499206543,
|
| 14 |
+
"learning_rate": 3.125e-06,
|
| 15 |
+
"loss": 1.689,
|
| 16 |
"step": 10
|
| 17 |
},
|
| 18 |
{
|
| 19 |
+
"epoch": 2.00125,
|
| 20 |
+
"grad_norm": 12.487483024597168,
|
| 21 |
+
"learning_rate": 6.25e-06,
|
| 22 |
+
"loss": 1.6397,
|
| 23 |
"step": 20
|
| 24 |
},
|
| 25 |
{
|
| 26 |
+
"epoch": 3.001875,
|
| 27 |
+
"grad_norm": 21.91889762878418,
|
| 28 |
+
"learning_rate": 9.375000000000001e-06,
|
| 29 |
+
"loss": 1.6462,
|
| 30 |
"step": 30
|
| 31 |
},
|
| 32 |
{
|
| 33 |
+
"epoch": 4.0025,
|
| 34 |
+
"grad_norm": 10.232108116149902,
|
| 35 |
+
"learning_rate": 1.25e-05,
|
| 36 |
+
"loss": 1.6326,
|
| 37 |
"step": 40
|
| 38 |
},
|
| 39 |
{
|
| 40 |
+
"epoch": 5.003125,
|
| 41 |
+
"grad_norm": 15.612605094909668,
|
| 42 |
+
"learning_rate": 1.5625e-05,
|
| 43 |
+
"loss": 1.5915,
|
| 44 |
"step": 50
|
| 45 |
},
|
| 46 |
{
|
| 47 |
+
"epoch": 6.00375,
|
| 48 |
+
"grad_norm": 16.83026123046875,
|
| 49 |
+
"learning_rate": 1.8750000000000002e-05,
|
| 50 |
+
"loss": 1.6447,
|
| 51 |
"step": 60
|
| 52 |
},
|
| 53 |
{
|
| 54 |
+
"epoch": 7.004375,
|
| 55 |
+
"grad_norm": 11.247737884521484,
|
| 56 |
+
"learning_rate": 2.1875e-05,
|
| 57 |
+
"loss": 1.6781,
|
| 58 |
"step": 70
|
| 59 |
},
|
| 60 |
{
|
| 61 |
+
"epoch": 8.005,
|
| 62 |
+
"grad_norm": 10.558433532714844,
|
| 63 |
+
"learning_rate": 2.5e-05,
|
| 64 |
+
"loss": 1.6657,
|
| 65 |
"step": 80
|
| 66 |
},
|
| 67 |
{
|
| 68 |
+
"epoch": 9.005625,
|
| 69 |
+
"grad_norm": 20.691017150878906,
|
| 70 |
+
"learning_rate": 2.8125000000000003e-05,
|
| 71 |
+
"loss": 1.709,
|
| 72 |
"step": 90
|
| 73 |
},
|
| 74 |
{
|
| 75 |
+
"epoch": 11.000625,
|
| 76 |
+
"grad_norm": 11.228839874267578,
|
| 77 |
+
"learning_rate": 3.125e-05,
|
| 78 |
+
"loss": 1.6518,
|
| 79 |
"step": 100
|
| 80 |
},
|
| 81 |
{
|
| 82 |
+
"epoch": 12.00125,
|
| 83 |
+
"grad_norm": 12.333444595336914,
|
| 84 |
+
"learning_rate": 3.4375e-05,
|
| 85 |
+
"loss": 1.6348,
|
| 86 |
"step": 110
|
| 87 |
},
|
| 88 |
{
|
| 89 |
+
"epoch": 13.001875,
|
| 90 |
+
"grad_norm": 8.79792308807373,
|
| 91 |
+
"learning_rate": 3.7500000000000003e-05,
|
| 92 |
+
"loss": 1.7793,
|
| 93 |
"step": 120
|
| 94 |
},
|
| 95 |
{
|
| 96 |
+
"epoch": 14.0025,
|
| 97 |
+
"grad_norm": 11.819356918334961,
|
| 98 |
+
"learning_rate": 4.0625000000000005e-05,
|
| 99 |
+
"loss": 1.7018,
|
| 100 |
"step": 130
|
| 101 |
},
|
| 102 |
{
|
| 103 |
+
"epoch": 15.003125,
|
| 104 |
+
"grad_norm": 10.883867263793945,
|
| 105 |
+
"learning_rate": 4.375e-05,
|
| 106 |
+
"loss": 1.6195,
|
| 107 |
"step": 140
|
| 108 |
},
|
| 109 |
{
|
| 110 |
+
"epoch": 16.00375,
|
| 111 |
+
"grad_norm": 11.140106201171875,
|
| 112 |
+
"learning_rate": 4.6875e-05,
|
| 113 |
+
"loss": 1.6123,
|
| 114 |
"step": 150
|
| 115 |
},
|
| 116 |
{
|
| 117 |
+
"epoch": 17.004375,
|
| 118 |
+
"grad_norm": 9.668269157409668,
|
| 119 |
+
"learning_rate": 5e-05,
|
| 120 |
+
"loss": 1.7115,
|
| 121 |
"step": 160
|
| 122 |
},
|
| 123 |
{
|
| 124 |
+
"epoch": 18.005,
|
| 125 |
+
"grad_norm": 7.844611167907715,
|
| 126 |
+
"learning_rate": 4.965277777777778e-05,
|
| 127 |
+
"loss": 1.6477,
|
| 128 |
"step": 170
|
| 129 |
},
|
| 130 |
{
|
| 131 |
+
"epoch": 19.005625,
|
| 132 |
+
"grad_norm": 15.249470710754395,
|
| 133 |
+
"learning_rate": 4.930555555555556e-05,
|
| 134 |
+
"loss": 1.6387,
|
| 135 |
"step": 180
|
| 136 |
},
|
| 137 |
{
|
| 138 |
+
"epoch": 21.000625,
|
| 139 |
+
"grad_norm": 7.852767467498779,
|
| 140 |
+
"learning_rate": 4.8958333333333335e-05,
|
| 141 |
+
"loss": 1.6663,
|
| 142 |
"step": 190
|
| 143 |
},
|
| 144 |
{
|
| 145 |
+
"epoch": 22.00125,
|
| 146 |
+
"grad_norm": 10.084273338317871,
|
| 147 |
+
"learning_rate": 4.8611111111111115e-05,
|
| 148 |
+
"loss": 1.6515,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 149 |
"step": 200
|
| 150 |
},
|
| 151 |
{
|
| 152 |
+
"epoch": 23.001875,
|
| 153 |
+
"grad_norm": 6.365440368652344,
|
| 154 |
+
"learning_rate": 4.8263888888888895e-05,
|
| 155 |
+
"loss": 1.6323,
|
| 156 |
"step": 210
|
| 157 |
},
|
| 158 |
{
|
| 159 |
+
"epoch": 24.0025,
|
| 160 |
+
"grad_norm": 7.19619607925415,
|
| 161 |
+
"learning_rate": 4.791666666666667e-05,
|
| 162 |
+
"loss": 1.6334,
|
| 163 |
"step": 220
|
| 164 |
},
|
| 165 |
{
|
| 166 |
+
"epoch": 25.003125,
|
| 167 |
+
"grad_norm": 9.672527313232422,
|
| 168 |
+
"learning_rate": 4.756944444444444e-05,
|
| 169 |
+
"loss": 1.6093,
|
| 170 |
"step": 230
|
| 171 |
},
|
| 172 |
{
|
| 173 |
+
"epoch": 26.00375,
|
| 174 |
+
"grad_norm": 6.754970073699951,
|
| 175 |
+
"learning_rate": 4.722222222222222e-05,
|
| 176 |
+
"loss": 1.6237,
|
| 177 |
"step": 240
|
| 178 |
},
|
| 179 |
{
|
| 180 |
+
"epoch": 27.004375,
|
| 181 |
+
"grad_norm": 8.783395767211914,
|
| 182 |
+
"learning_rate": 4.6875e-05,
|
| 183 |
+
"loss": 1.5612,
|
| 184 |
"step": 250
|
| 185 |
},
|
| 186 |
{
|
| 187 |
+
"epoch": 28.005,
|
| 188 |
+
"grad_norm": 9.81071949005127,
|
| 189 |
+
"learning_rate": 4.652777777777778e-05,
|
| 190 |
+
"loss": 1.6906,
|
| 191 |
"step": 260
|
| 192 |
},
|
| 193 |
{
|
| 194 |
+
"epoch": 29.005625,
|
| 195 |
+
"grad_norm": 14.682830810546875,
|
| 196 |
+
"learning_rate": 4.618055555555556e-05,
|
| 197 |
+
"loss": 1.5813,
|
| 198 |
"step": 270
|
| 199 |
},
|
| 200 |
{
|
| 201 |
+
"epoch": 31.000625,
|
| 202 |
+
"grad_norm": 6.165945529937744,
|
| 203 |
+
"learning_rate": 4.5833333333333334e-05,
|
| 204 |
+
"loss": 1.6764,
|
| 205 |
"step": 280
|
| 206 |
},
|
| 207 |
{
|
| 208 |
+
"epoch": 32.00125,
|
| 209 |
+
"grad_norm": 10.008849143981934,
|
| 210 |
+
"learning_rate": 4.5486111111111114e-05,
|
| 211 |
+
"loss": 1.657,
|
| 212 |
"step": 290
|
| 213 |
},
|
| 214 |
{
|
| 215 |
+
"epoch": 33.001875,
|
| 216 |
+
"grad_norm": 8.975122451782227,
|
| 217 |
+
"learning_rate": 4.5138888888888894e-05,
|
| 218 |
+
"loss": 1.4942,
|
| 219 |
"step": 300
|
| 220 |
},
|
| 221 |
{
|
| 222 |
+
"epoch": 34.0025,
|
| 223 |
+
"grad_norm": 9.013062477111816,
|
| 224 |
+
"learning_rate": 4.4791666666666673e-05,
|
| 225 |
+
"loss": 1.5295,
|
| 226 |
"step": 310
|
| 227 |
},
|
| 228 |
{
|
| 229 |
+
"epoch": 35.003125,
|
| 230 |
+
"grad_norm": 5.317986965179443,
|
| 231 |
+
"learning_rate": 4.4444444444444447e-05,
|
| 232 |
+
"loss": 1.4473,
|
| 233 |
"step": 320
|
| 234 |
},
|
| 235 |
{
|
| 236 |
+
"epoch": 36.00375,
|
| 237 |
+
"grad_norm": 6.352290153503418,
|
| 238 |
+
"learning_rate": 4.4097222222222226e-05,
|
| 239 |
+
"loss": 1.378,
|
| 240 |
"step": 330
|
| 241 |
},
|
| 242 |
{
|
| 243 |
+
"epoch": 37.004375,
|
| 244 |
+
"grad_norm": 6.145012378692627,
|
| 245 |
+
"learning_rate": 4.375e-05,
|
| 246 |
+
"loss": 1.2513,
|
| 247 |
"step": 340
|
| 248 |
},
|
| 249 |
{
|
| 250 |
+
"epoch": 38.005,
|
| 251 |
+
"grad_norm": 9.207514762878418,
|
| 252 |
+
"learning_rate": 4.340277777777778e-05,
|
| 253 |
+
"loss": 1.7593,
|
| 254 |
"step": 350
|
| 255 |
},
|
| 256 |
{
|
| 257 |
+
"epoch": 39.005625,
|
| 258 |
+
"grad_norm": 33.3981819152832,
|
| 259 |
+
"learning_rate": 4.305555555555556e-05,
|
| 260 |
+
"loss": 1.5538,
|
| 261 |
"step": 360
|
| 262 |
},
|
| 263 |
{
|
| 264 |
+
"epoch": 41.000625,
|
| 265 |
+
"grad_norm": 8.981507301330566,
|
| 266 |
+
"learning_rate": 4.270833333333333e-05,
|
| 267 |
+
"loss": 1.3785,
|
| 268 |
"step": 370
|
| 269 |
},
|
| 270 |
{
|
| 271 |
+
"epoch": 42.00125,
|
| 272 |
+
"grad_norm": 9.220932006835938,
|
| 273 |
+
"learning_rate": 4.236111111111111e-05,
|
| 274 |
+
"loss": 1.4187,
|
| 275 |
"step": 380
|
| 276 |
},
|
| 277 |
{
|
| 278 |
+
"epoch": 43.001875,
|
| 279 |
+
"grad_norm": 6.985467433929443,
|
| 280 |
+
"learning_rate": 4.201388888888889e-05,
|
| 281 |
+
"loss": 1.3837,
|
| 282 |
"step": 390
|
| 283 |
},
|
| 284 |
{
|
| 285 |
+
"epoch": 44.0025,
|
| 286 |
+
"grad_norm": 14.327313423156738,
|
| 287 |
+
"learning_rate": 4.166666666666667e-05,
|
| 288 |
+
"loss": 1.5855,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 289 |
"step": 400
|
| 290 |
},
|
| 291 |
{
|
| 292 |
+
"epoch": 45.003125,
|
| 293 |
+
"grad_norm": 5.2364821434021,
|
| 294 |
+
"learning_rate": 4.1319444444444445e-05,
|
| 295 |
+
"loss": 1.5322,
|
| 296 |
"step": 410
|
| 297 |
},
|
| 298 |
{
|
| 299 |
+
"epoch": 46.00375,
|
| 300 |
+
"grad_norm": 11.791383743286133,
|
| 301 |
+
"learning_rate": 4.0972222222222225e-05,
|
| 302 |
+
"loss": 1.2774,
|
| 303 |
"step": 420
|
| 304 |
},
|
| 305 |
{
|
| 306 |
+
"epoch": 47.004375,
|
| 307 |
+
"grad_norm": 14.766942977905273,
|
| 308 |
+
"learning_rate": 4.0625000000000005e-05,
|
| 309 |
+
"loss": 1.5675,
|
| 310 |
"step": 430
|
| 311 |
},
|
| 312 |
{
|
| 313 |
+
"epoch": 48.005,
|
| 314 |
+
"grad_norm": 15.439301490783691,
|
| 315 |
+
"learning_rate": 4.027777777777778e-05,
|
| 316 |
+
"loss": 1.2914,
|
| 317 |
"step": 440
|
| 318 |
},
|
| 319 |
{
|
| 320 |
+
"epoch": 49.005625,
|
| 321 |
+
"grad_norm": 18.859539031982422,
|
| 322 |
+
"learning_rate": 3.993055555555556e-05,
|
| 323 |
+
"loss": 1.4614,
|
| 324 |
"step": 450
|
| 325 |
},
|
| 326 |
{
|
| 327 |
+
"epoch": 51.000625,
|
| 328 |
+
"grad_norm": 6.83065128326416,
|
| 329 |
+
"learning_rate": 3.958333333333333e-05,
|
| 330 |
+
"loss": 1.4297,
|
| 331 |
"step": 460
|
| 332 |
},
|
| 333 |
{
|
| 334 |
+
"epoch": 52.00125,
|
| 335 |
+
"grad_norm": 7.136291027069092,
|
| 336 |
+
"learning_rate": 3.923611111111111e-05,
|
| 337 |
+
"loss": 1.4145,
|
| 338 |
"step": 470
|
| 339 |
},
|
| 340 |
{
|
| 341 |
+
"epoch": 53.001875,
|
| 342 |
+
"grad_norm": 6.295926094055176,
|
| 343 |
+
"learning_rate": 3.888888888888889e-05,
|
| 344 |
+
"loss": 1.2577,
|
| 345 |
"step": 480
|
| 346 |
},
|
| 347 |
{
|
| 348 |
+
"epoch": 54.0025,
|
| 349 |
+
"grad_norm": 5.792518615722656,
|
| 350 |
+
"learning_rate": 3.854166666666667e-05,
|
| 351 |
+
"loss": 1.0809,
|
| 352 |
"step": 490
|
| 353 |
},
|
| 354 |
{
|
| 355 |
+
"epoch": 55.003125,
|
| 356 |
+
"grad_norm": 14.563287734985352,
|
| 357 |
+
"learning_rate": 3.8194444444444444e-05,
|
| 358 |
+
"loss": 0.9168,
|
| 359 |
"step": 500
|
| 360 |
},
|
| 361 |
{
|
| 362 |
+
"epoch": 55.003125,
|
| 363 |
+
"eval_accuracy": 0.5294117647058824,
|
| 364 |
+
"eval_loss": 1.1491186618804932,
|
| 365 |
+
"eval_runtime": 6.3087,
|
| 366 |
+
"eval_samples_per_second": 5.389,
|
| 367 |
+
"eval_steps_per_second": 1.902,
|
| 368 |
+
"step": 500
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 56.00375,
|
| 372 |
+
"grad_norm": 13.170215606689453,
|
| 373 |
+
"learning_rate": 3.7847222222222224e-05,
|
| 374 |
+
"loss": 1.3475,
|
| 375 |
"step": 510
|
| 376 |
},
|
| 377 |
{
|
| 378 |
+
"epoch": 57.004375,
|
| 379 |
+
"grad_norm": 14.003524780273438,
|
| 380 |
+
"learning_rate": 3.7500000000000003e-05,
|
| 381 |
+
"loss": 1.1146,
|
| 382 |
"step": 520
|
| 383 |
},
|
| 384 |
{
|
| 385 |
+
"epoch": 58.005,
|
| 386 |
+
"grad_norm": 57.568973541259766,
|
| 387 |
+
"learning_rate": 3.715277777777778e-05,
|
| 388 |
+
"loss": 1.0694,
|
| 389 |
"step": 530
|
| 390 |
},
|
| 391 |
{
|
| 392 |
+
"epoch": 59.005625,
|
| 393 |
+
"grad_norm": 18.669395446777344,
|
| 394 |
+
"learning_rate": 3.6805555555555556e-05,
|
| 395 |
+
"loss": 1.1377,
|
| 396 |
"step": 540
|
| 397 |
},
|
| 398 |
{
|
| 399 |
+
"epoch": 61.000625,
|
| 400 |
+
"grad_norm": 8.16624927520752,
|
| 401 |
+
"learning_rate": 3.6458333333333336e-05,
|
| 402 |
+
"loss": 1.0386,
|
| 403 |
"step": 550
|
| 404 |
},
|
| 405 |
{
|
| 406 |
+
"epoch": 62.00125,
|
| 407 |
+
"grad_norm": 2.5847628116607666,
|
| 408 |
+
"learning_rate": 3.611111111111111e-05,
|
| 409 |
+
"loss": 0.9063,
|
| 410 |
"step": 560
|
| 411 |
},
|
| 412 |
{
|
| 413 |
+
"epoch": 63.001875,
|
| 414 |
+
"grad_norm": 18.86676788330078,
|
| 415 |
+
"learning_rate": 3.576388888888889e-05,
|
| 416 |
+
"loss": 0.7986,
|
| 417 |
"step": 570
|
| 418 |
},
|
| 419 |
{
|
| 420 |
+
"epoch": 64.0025,
|
| 421 |
+
"grad_norm": 3.9623453617095947,
|
| 422 |
+
"learning_rate": 3.541666666666667e-05,
|
| 423 |
+
"loss": 0.6622,
|
| 424 |
"step": 580
|
| 425 |
},
|
| 426 |
{
|
| 427 |
+
"epoch": 65.003125,
|
| 428 |
+
"grad_norm": 2.4036800861358643,
|
| 429 |
+
"learning_rate": 3.506944444444444e-05,
|
| 430 |
+
"loss": 0.6984,
|
| 431 |
"step": 590
|
| 432 |
},
|
| 433 |
{
|
| 434 |
+
"epoch": 66.00375,
|
| 435 |
+
"grad_norm": 10.653820037841797,
|
| 436 |
+
"learning_rate": 3.472222222222222e-05,
|
| 437 |
+
"loss": 0.9318,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 438 |
"step": 600
|
| 439 |
},
|
| 440 |
{
|
| 441 |
+
"epoch": 67.004375,
|
| 442 |
+
"grad_norm": 10.026162147521973,
|
| 443 |
+
"learning_rate": 3.4375e-05,
|
| 444 |
+
"loss": 1.4336,
|
| 445 |
"step": 610
|
| 446 |
},
|
| 447 |
{
|
| 448 |
+
"epoch": 68.005,
|
| 449 |
+
"grad_norm": 17.375564575195312,
|
| 450 |
+
"learning_rate": 3.402777777777778e-05,
|
| 451 |
+
"loss": 0.7662,
|
| 452 |
"step": 620
|
| 453 |
},
|
| 454 |
{
|
| 455 |
+
"epoch": 69.005625,
|
| 456 |
+
"grad_norm": 1.6536837816238403,
|
| 457 |
+
"learning_rate": 3.368055555555556e-05,
|
| 458 |
+
"loss": 0.6028,
|
| 459 |
"step": 630
|
| 460 |
},
|
| 461 |
{
|
| 462 |
+
"epoch": 71.000625,
|
| 463 |
+
"grad_norm": 3.789870262145996,
|
| 464 |
+
"learning_rate": 3.3333333333333335e-05,
|
| 465 |
+
"loss": 0.4272,
|
| 466 |
"step": 640
|
| 467 |
},
|
| 468 |
{
|
| 469 |
+
"epoch": 72.00125,
|
| 470 |
+
"grad_norm": 22.276479721069336,
|
| 471 |
+
"learning_rate": 3.2986111111111115e-05,
|
| 472 |
+
"loss": 1.0059,
|
| 473 |
"step": 650
|
| 474 |
},
|
| 475 |
{
|
| 476 |
+
"epoch": 73.001875,
|
| 477 |
+
"grad_norm": 10.886910438537598,
|
| 478 |
+
"learning_rate": 3.263888888888889e-05,
|
| 479 |
+
"loss": 0.4979,
|
| 480 |
"step": 660
|
| 481 |
},
|
| 482 |
{
|
| 483 |
+
"epoch": 74.0025,
|
| 484 |
+
"grad_norm": 14.757866859436035,
|
| 485 |
+
"learning_rate": 3.229166666666667e-05,
|
| 486 |
+
"loss": 1.3088,
|
| 487 |
"step": 670
|
| 488 |
},
|
| 489 |
{
|
| 490 |
+
"epoch": 75.003125,
|
| 491 |
+
"grad_norm": 6.947025775909424,
|
| 492 |
+
"learning_rate": 3.194444444444444e-05,
|
| 493 |
+
"loss": 0.6417,
|
| 494 |
"step": 680
|
| 495 |
},
|
| 496 |
{
|
| 497 |
+
"epoch": 76.00375,
|
| 498 |
+
"grad_norm": 10.190616607666016,
|
| 499 |
+
"learning_rate": 3.159722222222222e-05,
|
| 500 |
+
"loss": 0.5463,
|
| 501 |
"step": 690
|
| 502 |
},
|
| 503 |
{
|
| 504 |
+
"epoch": 77.004375,
|
| 505 |
+
"grad_norm": 12.709113121032715,
|
| 506 |
+
"learning_rate": 3.125e-05,
|
| 507 |
+
"loss": 0.5217,
|
| 508 |
"step": 700
|
| 509 |
},
|
| 510 |
{
|
| 511 |
+
"epoch": 78.005,
|
| 512 |
+
"grad_norm": 0.8149201273918152,
|
| 513 |
+
"learning_rate": 3.090277777777778e-05,
|
| 514 |
+
"loss": 0.6461,
|
| 515 |
"step": 710
|
| 516 |
},
|
| 517 |
{
|
| 518 |
+
"epoch": 79.005625,
|
| 519 |
+
"grad_norm": 0.7477602958679199,
|
| 520 |
+
"learning_rate": 3.055555555555556e-05,
|
| 521 |
+
"loss": 0.3186,
|
| 522 |
"step": 720
|
| 523 |
},
|
| 524 |
{
|
| 525 |
+
"epoch": 81.000625,
|
| 526 |
+
"grad_norm": 20.247188568115234,
|
| 527 |
+
"learning_rate": 3.0208333333333334e-05,
|
| 528 |
+
"loss": 0.4566,
|
| 529 |
"step": 730
|
| 530 |
},
|
| 531 |
{
|
| 532 |
+
"epoch": 82.00125,
|
| 533 |
+
"grad_norm": 2.7397711277008057,
|
| 534 |
+
"learning_rate": 2.9861111111111113e-05,
|
| 535 |
+
"loss": 0.673,
|
| 536 |
"step": 740
|
| 537 |
},
|
| 538 |
{
|
| 539 |
+
"epoch": 83.001875,
|
| 540 |
+
"grad_norm": 0.6794953346252441,
|
| 541 |
+
"learning_rate": 2.951388888888889e-05,
|
| 542 |
+
"loss": 0.1364,
|
| 543 |
"step": 750
|
| 544 |
},
|
| 545 |
{
|
| 546 |
+
"epoch": 84.0025,
|
| 547 |
+
"grad_norm": 30.472972869873047,
|
| 548 |
+
"learning_rate": 2.916666666666667e-05,
|
| 549 |
+
"loss": 0.4537,
|
| 550 |
"step": 760
|
| 551 |
},
|
| 552 |
{
|
| 553 |
+
"epoch": 85.003125,
|
| 554 |
+
"grad_norm": 25.230920791625977,
|
| 555 |
+
"learning_rate": 2.8819444444444443e-05,
|
| 556 |
+
"loss": 0.5033,
|
| 557 |
"step": 770
|
| 558 |
},
|
| 559 |
{
|
| 560 |
+
"epoch": 86.00375,
|
| 561 |
+
"grad_norm": 12.282685279846191,
|
| 562 |
+
"learning_rate": 2.8472222222222223e-05,
|
| 563 |
+
"loss": 0.4521,
|
| 564 |
"step": 780
|
| 565 |
},
|
| 566 |
{
|
| 567 |
+
"epoch": 87.004375,
|
| 568 |
+
"grad_norm": 21.0340633392334,
|
| 569 |
+
"learning_rate": 2.8125000000000003e-05,
|
| 570 |
+
"loss": 0.4044,
|
| 571 |
"step": 790
|
| 572 |
},
|
| 573 |
{
|
| 574 |
+
"epoch": 88.005,
|
| 575 |
+
"grad_norm": 10.24821949005127,
|
| 576 |
+
"learning_rate": 2.777777777777778e-05,
|
| 577 |
+
"loss": 0.1864,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 578 |
"step": 800
|
| 579 |
},
|
| 580 |
{
|
| 581 |
+
"epoch": 89.005625,
|
| 582 |
+
"grad_norm": 0.11952634900808334,
|
| 583 |
+
"learning_rate": 2.743055555555556e-05,
|
| 584 |
+
"loss": 0.0229,
|
| 585 |
"step": 810
|
| 586 |
},
|
| 587 |
{
|
| 588 |
+
"epoch": 91.000625,
|
| 589 |
+
"grad_norm": 116.37321472167969,
|
| 590 |
+
"learning_rate": 2.7083333333333332e-05,
|
| 591 |
+
"loss": 0.4937,
|
| 592 |
"step": 820
|
| 593 |
},
|
| 594 |
{
|
| 595 |
+
"epoch": 92.00125,
|
| 596 |
+
"grad_norm": 137.15133666992188,
|
| 597 |
+
"learning_rate": 2.6736111111111112e-05,
|
| 598 |
+
"loss": 0.4143,
|
| 599 |
"step": 830
|
| 600 |
},
|
| 601 |
{
|
| 602 |
+
"epoch": 93.001875,
|
| 603 |
+
"grad_norm": 0.07484932988882065,
|
| 604 |
+
"learning_rate": 2.6388888888888892e-05,
|
| 605 |
+
"loss": 0.1536,
|
| 606 |
"step": 840
|
| 607 |
},
|
| 608 |
{
|
| 609 |
+
"epoch": 94.0025,
|
| 610 |
+
"grad_norm": 6.392551422119141,
|
| 611 |
+
"learning_rate": 2.604166666666667e-05,
|
| 612 |
+
"loss": 0.1022,
|
| 613 |
"step": 850
|
| 614 |
},
|
| 615 |
{
|
| 616 |
+
"epoch": 95.003125,
|
| 617 |
+
"grad_norm": 0.13532483577728271,
|
| 618 |
+
"learning_rate": 2.5694444444444445e-05,
|
| 619 |
+
"loss": 0.508,
|
| 620 |
"step": 860
|
| 621 |
},
|
| 622 |
{
|
| 623 |
+
"epoch": 96.00375,
|
| 624 |
+
"grad_norm": 29.035808563232422,
|
| 625 |
+
"learning_rate": 2.534722222222222e-05,
|
| 626 |
+
"loss": 0.5411,
|
| 627 |
"step": 870
|
| 628 |
},
|
| 629 |
{
|
| 630 |
+
"epoch": 97.004375,
|
| 631 |
+
"grad_norm": 0.5214412808418274,
|
| 632 |
+
"learning_rate": 2.5e-05,
|
| 633 |
+
"loss": 0.2972,
|
| 634 |
"step": 880
|
| 635 |
},
|
| 636 |
{
|
| 637 |
+
"epoch": 98.005,
|
| 638 |
+
"grad_norm": 3.654513120651245,
|
| 639 |
+
"learning_rate": 2.465277777777778e-05,
|
| 640 |
+
"loss": 0.7897,
|
| 641 |
"step": 890
|
| 642 |
},
|
| 643 |
{
|
| 644 |
+
"epoch": 99.005625,
|
| 645 |
+
"grad_norm": 11.014652252197266,
|
| 646 |
+
"learning_rate": 2.4305555555555558e-05,
|
| 647 |
+
"loss": 0.3799,
|
| 648 |
"step": 900
|
| 649 |
},
|
| 650 |
{
|
| 651 |
+
"epoch": 101.000625,
|
| 652 |
+
"grad_norm": 0.1979626715183258,
|
| 653 |
+
"learning_rate": 2.3958333333333334e-05,
|
| 654 |
+
"loss": 0.0879,
|
| 655 |
"step": 910
|
| 656 |
},
|
| 657 |
{
|
| 658 |
+
"epoch": 102.00125,
|
| 659 |
+
"grad_norm": 56.55485153198242,
|
| 660 |
+
"learning_rate": 2.361111111111111e-05,
|
| 661 |
+
"loss": 0.1458,
|
| 662 |
"step": 920
|
| 663 |
},
|
| 664 |
{
|
| 665 |
+
"epoch": 103.001875,
|
| 666 |
+
"grad_norm": 0.8609234690666199,
|
| 667 |
+
"learning_rate": 2.326388888888889e-05,
|
| 668 |
+
"loss": 0.2159,
|
| 669 |
"step": 930
|
| 670 |
},
|
| 671 |
{
|
| 672 |
+
"epoch": 104.0025,
|
| 673 |
+
"grad_norm": 53.709442138671875,
|
| 674 |
+
"learning_rate": 2.2916666666666667e-05,
|
| 675 |
+
"loss": 0.6802,
|
| 676 |
"step": 940
|
| 677 |
},
|
| 678 |
{
|
| 679 |
+
"epoch": 105.003125,
|
| 680 |
+
"grad_norm": 0.06115364283323288,
|
| 681 |
+
"learning_rate": 2.2569444444444447e-05,
|
| 682 |
+
"loss": 0.0572,
|
| 683 |
"step": 950
|
| 684 |
},
|
| 685 |
{
|
| 686 |
+
"epoch": 106.00375,
|
| 687 |
+
"grad_norm": 0.18275491893291473,
|
| 688 |
+
"learning_rate": 2.2222222222222223e-05,
|
| 689 |
+
"loss": 0.0073,
|
| 690 |
"step": 960
|
| 691 |
},
|
| 692 |
{
|
| 693 |
+
"epoch": 107.004375,
|
| 694 |
+
"grad_norm": 0.8914717435836792,
|
| 695 |
+
"learning_rate": 2.1875e-05,
|
| 696 |
+
"loss": 0.043,
|
| 697 |
"step": 970
|
| 698 |
},
|
| 699 |
{
|
| 700 |
+
"epoch": 108.005,
|
| 701 |
+
"grad_norm": 0.08716151118278503,
|
| 702 |
+
"learning_rate": 2.152777777777778e-05,
|
| 703 |
+
"loss": 0.0249,
|
| 704 |
"step": 980
|
| 705 |
},
|
| 706 |
{
|
| 707 |
+
"epoch": 109.005625,
|
| 708 |
+
"grad_norm": 0.025017954409122467,
|
| 709 |
+
"learning_rate": 2.1180555555555556e-05,
|
| 710 |
+
"loss": 0.2096,
|
| 711 |
"step": 990
|
| 712 |
},
|
| 713 |
{
|
| 714 |
+
"epoch": 111.000625,
|
| 715 |
+
"grad_norm": 2.6935932636260986,
|
| 716 |
+
"learning_rate": 2.0833333333333336e-05,
|
| 717 |
+
"loss": 0.3827,
|
| 718 |
"step": 1000
|
| 719 |
},
|
| 720 |
{
|
| 721 |
+
"epoch": 111.000625,
|
| 722 |
"eval_accuracy": 0.8235294117647058,
|
| 723 |
+
"eval_loss": 0.7478592395782471,
|
| 724 |
+
"eval_runtime": 6.5367,
|
| 725 |
+
"eval_samples_per_second": 5.201,
|
| 726 |
+
"eval_steps_per_second": 1.836,
|
| 727 |
"step": 1000
|
| 728 |
},
|
| 729 |
{
|
| 730 |
+
"epoch": 112.00125,
|
| 731 |
+
"grad_norm": 54.73932647705078,
|
| 732 |
+
"learning_rate": 2.0486111111111113e-05,
|
| 733 |
+
"loss": 0.198,
|
| 734 |
"step": 1010
|
| 735 |
},
|
| 736 |
{
|
| 737 |
+
"epoch": 113.001875,
|
| 738 |
+
"grad_norm": 0.056511297821998596,
|
| 739 |
+
"learning_rate": 2.013888888888889e-05,
|
| 740 |
+
"loss": 0.3577,
|
| 741 |
"step": 1020
|
| 742 |
},
|
| 743 |
{
|
| 744 |
+
"epoch": 114.0025,
|
| 745 |
+
"grad_norm": 5.711917877197266,
|
| 746 |
+
"learning_rate": 1.9791666666666665e-05,
|
| 747 |
+
"loss": 0.0365,
|
| 748 |
"step": 1030
|
| 749 |
},
|
| 750 |
{
|
| 751 |
+
"epoch": 115.003125,
|
| 752 |
+
"grad_norm": 0.0672488734126091,
|
| 753 |
+
"learning_rate": 1.9444444444444445e-05,
|
| 754 |
+
"loss": 0.3771,
|
| 755 |
"step": 1040
|
| 756 |
},
|
| 757 |
{
|
| 758 |
+
"epoch": 116.00375,
|
| 759 |
+
"grad_norm": 137.31533813476562,
|
| 760 |
+
"learning_rate": 1.9097222222222222e-05,
|
| 761 |
+
"loss": 0.2123,
|
| 762 |
"step": 1050
|
| 763 |
},
|
| 764 |
{
|
| 765 |
+
"epoch": 117.004375,
|
| 766 |
+
"grad_norm": 31.655075073242188,
|
| 767 |
+
"learning_rate": 1.8750000000000002e-05,
|
| 768 |
+
"loss": 0.19,
|
| 769 |
"step": 1060
|
| 770 |
},
|
| 771 |
{
|
| 772 |
+
"epoch": 118.005,
|
| 773 |
+
"grad_norm": 0.03197183832526207,
|
| 774 |
+
"learning_rate": 1.8402777777777778e-05,
|
| 775 |
+
"loss": 0.5969,
|
| 776 |
"step": 1070
|
| 777 |
},
|
| 778 |
{
|
| 779 |
+
"epoch": 119.005625,
|
| 780 |
+
"grad_norm": 0.11024000495672226,
|
| 781 |
+
"learning_rate": 1.8055555555555555e-05,
|
| 782 |
+
"loss": 0.0148,
|
| 783 |
"step": 1080
|
| 784 |
},
|
| 785 |
{
|
| 786 |
+
"epoch": 121.000625,
|
| 787 |
+
"grad_norm": 0.566511332988739,
|
| 788 |
+
"learning_rate": 1.7708333333333335e-05,
|
| 789 |
+
"loss": 0.2468,
|
| 790 |
"step": 1090
|
| 791 |
},
|
| 792 |
{
|
| 793 |
+
"epoch": 122.00125,
|
| 794 |
+
"grad_norm": 0.12642063200473785,
|
| 795 |
+
"learning_rate": 1.736111111111111e-05,
|
| 796 |
+
"loss": 0.016,
|
| 797 |
"step": 1100
|
| 798 |
},
|
| 799 |
{
|
| 800 |
+
"epoch": 123.001875,
|
| 801 |
+
"grad_norm": 69.47509002685547,
|
| 802 |
+
"learning_rate": 1.701388888888889e-05,
|
| 803 |
+
"loss": 0.3782,
|
| 804 |
"step": 1110
|
| 805 |
},
|
| 806 |
{
|
| 807 |
+
"epoch": 124.0025,
|
| 808 |
+
"grad_norm": 8.828600883483887,
|
| 809 |
+
"learning_rate": 1.6666666666666667e-05,
|
| 810 |
+
"loss": 0.5294,
|
| 811 |
"step": 1120
|
| 812 |
},
|
| 813 |
{
|
| 814 |
+
"epoch": 125.003125,
|
| 815 |
+
"grad_norm": 0.020838657394051552,
|
| 816 |
+
"learning_rate": 1.6319444444444444e-05,
|
| 817 |
+
"loss": 0.0505,
|
| 818 |
"step": 1130
|
| 819 |
},
|
| 820 |
{
|
| 821 |
+
"epoch": 126.00375,
|
| 822 |
+
"grad_norm": 0.021230794489383698,
|
| 823 |
+
"learning_rate": 1.597222222222222e-05,
|
| 824 |
+
"loss": 0.075,
|
| 825 |
"step": 1140
|
| 826 |
},
|
| 827 |
{
|
| 828 |
+
"epoch": 127.004375,
|
| 829 |
+
"grad_norm": 0.053505346179008484,
|
| 830 |
+
"learning_rate": 1.5625e-05,
|
| 831 |
+
"loss": 0.0074,
|
| 832 |
"step": 1150
|
| 833 |
},
|
| 834 |
{
|
| 835 |
+
"epoch": 128.005,
|
| 836 |
+
"grad_norm": 0.09297660738229752,
|
| 837 |
+
"learning_rate": 1.527777777777778e-05,
|
| 838 |
+
"loss": 0.0192,
|
| 839 |
"step": 1160
|
| 840 |
},
|
| 841 |
{
|
| 842 |
+
"epoch": 129.005625,
|
| 843 |
+
"grad_norm": 144.81719970703125,
|
| 844 |
+
"learning_rate": 1.4930555555555557e-05,
|
| 845 |
+
"loss": 0.1825,
|
| 846 |
"step": 1170
|
| 847 |
},
|
| 848 |
{
|
| 849 |
+
"epoch": 131.000625,
|
| 850 |
+
"grad_norm": 0.07185923308134079,
|
| 851 |
+
"learning_rate": 1.4583333333333335e-05,
|
| 852 |
+
"loss": 0.0079,
|
| 853 |
"step": 1180
|
| 854 |
},
|
| 855 |
{
|
| 856 |
+
"epoch": 132.00125,
|
| 857 |
+
"grad_norm": 0.029854485765099525,
|
| 858 |
+
"learning_rate": 1.4236111111111111e-05,
|
| 859 |
+
"loss": 0.002,
|
| 860 |
"step": 1190
|
| 861 |
},
|
| 862 |
{
|
| 863 |
+
"epoch": 133.001875,
|
| 864 |
+
"grad_norm": 0.027790095657110214,
|
| 865 |
+
"learning_rate": 1.388888888888889e-05,
|
| 866 |
+
"loss": 0.0227,
|
| 867 |
"step": 1200
|
| 868 |
},
|
| 869 |
{
|
| 870 |
+
"epoch": 134.0025,
|
| 871 |
+
"grad_norm": 0.0227198526263237,
|
| 872 |
+
"learning_rate": 1.3541666666666666e-05,
|
| 873 |
+
"loss": 0.1834,
|
| 874 |
+
"step": 1210
|
| 875 |
+
},
|
| 876 |
+
{
|
| 877 |
+
"epoch": 135.003125,
|
| 878 |
+
"grad_norm": 0.016053209081292152,
|
| 879 |
+
"learning_rate": 1.3194444444444446e-05,
|
| 880 |
+
"loss": 0.0285,
|
| 881 |
+
"step": 1220
|
| 882 |
+
},
|
| 883 |
+
{
|
| 884 |
+
"epoch": 136.00375,
|
| 885 |
+
"grad_norm": 0.01930040493607521,
|
| 886 |
+
"learning_rate": 1.2847222222222222e-05,
|
| 887 |
+
"loss": 0.1674,
|
| 888 |
+
"step": 1230
|
| 889 |
+
},
|
| 890 |
+
{
|
| 891 |
+
"epoch": 137.004375,
|
| 892 |
+
"grad_norm": 0.019892092794179916,
|
| 893 |
+
"learning_rate": 1.25e-05,
|
| 894 |
+
"loss": 0.1771,
|
| 895 |
+
"step": 1240
|
| 896 |
+
},
|
| 897 |
+
{
|
| 898 |
+
"epoch": 138.005,
|
| 899 |
+
"grad_norm": 0.09665486216545105,
|
| 900 |
+
"learning_rate": 1.2152777777777779e-05,
|
| 901 |
+
"loss": 0.0505,
|
| 902 |
+
"step": 1250
|
| 903 |
+
},
|
| 904 |
+
{
|
| 905 |
+
"epoch": 139.005625,
|
| 906 |
+
"grad_norm": 0.0465998500585556,
|
| 907 |
+
"learning_rate": 1.1805555555555555e-05,
|
| 908 |
+
"loss": 0.0802,
|
| 909 |
+
"step": 1260
|
| 910 |
+
},
|
| 911 |
+
{
|
| 912 |
+
"epoch": 141.000625,
|
| 913 |
+
"grad_norm": 0.03291332721710205,
|
| 914 |
+
"learning_rate": 1.1458333333333333e-05,
|
| 915 |
+
"loss": 0.002,
|
| 916 |
+
"step": 1270
|
| 917 |
+
},
|
| 918 |
+
{
|
| 919 |
+
"epoch": 142.00125,
|
| 920 |
+
"grad_norm": 67.8191146850586,
|
| 921 |
+
"learning_rate": 1.1111111111111112e-05,
|
| 922 |
+
"loss": 0.1477,
|
| 923 |
+
"step": 1280
|
| 924 |
+
},
|
| 925 |
+
{
|
| 926 |
+
"epoch": 143.001875,
|
| 927 |
+
"grad_norm": 0.019737206399440765,
|
| 928 |
+
"learning_rate": 1.076388888888889e-05,
|
| 929 |
+
"loss": 0.1562,
|
| 930 |
+
"step": 1290
|
| 931 |
+
},
|
| 932 |
+
{
|
| 933 |
+
"epoch": 144.0025,
|
| 934 |
+
"grad_norm": 0.018836895003914833,
|
| 935 |
+
"learning_rate": 1.0416666666666668e-05,
|
| 936 |
+
"loss": 0.2501,
|
| 937 |
+
"step": 1300
|
| 938 |
+
},
|
| 939 |
+
{
|
| 940 |
+
"epoch": 145.003125,
|
| 941 |
+
"grad_norm": 0.024592338129878044,
|
| 942 |
+
"learning_rate": 1.0069444444444445e-05,
|
| 943 |
+
"loss": 0.1302,
|
| 944 |
+
"step": 1310
|
| 945 |
+
},
|
| 946 |
+
{
|
| 947 |
+
"epoch": 146.00375,
|
| 948 |
+
"grad_norm": 0.02780103124678135,
|
| 949 |
+
"learning_rate": 9.722222222222223e-06,
|
| 950 |
+
"loss": 0.0105,
|
| 951 |
+
"step": 1320
|
| 952 |
+
},
|
| 953 |
+
{
|
| 954 |
+
"epoch": 147.004375,
|
| 955 |
+
"grad_norm": 0.03460738807916641,
|
| 956 |
+
"learning_rate": 9.375000000000001e-06,
|
| 957 |
+
"loss": 0.0255,
|
| 958 |
+
"step": 1330
|
| 959 |
+
},
|
| 960 |
+
{
|
| 961 |
+
"epoch": 148.005,
|
| 962 |
+
"grad_norm": 0.025903472676873207,
|
| 963 |
+
"learning_rate": 9.027777777777777e-06,
|
| 964 |
+
"loss": 0.1031,
|
| 965 |
+
"step": 1340
|
| 966 |
+
},
|
| 967 |
+
{
|
| 968 |
+
"epoch": 149.005625,
|
| 969 |
+
"grad_norm": 0.03954395651817322,
|
| 970 |
+
"learning_rate": 8.680555555555556e-06,
|
| 971 |
+
"loss": 0.0041,
|
| 972 |
+
"step": 1350
|
| 973 |
+
},
|
| 974 |
+
{
|
| 975 |
+
"epoch": 151.000625,
|
| 976 |
+
"grad_norm": 0.04806604981422424,
|
| 977 |
+
"learning_rate": 8.333333333333334e-06,
|
| 978 |
+
"loss": 0.0054,
|
| 979 |
+
"step": 1360
|
| 980 |
+
},
|
| 981 |
+
{
|
| 982 |
+
"epoch": 152.00125,
|
| 983 |
+
"grad_norm": 51.491668701171875,
|
| 984 |
+
"learning_rate": 7.98611111111111e-06,
|
| 985 |
+
"loss": 0.1716,
|
| 986 |
+
"step": 1370
|
| 987 |
+
},
|
| 988 |
+
{
|
| 989 |
+
"epoch": 153.001875,
|
| 990 |
+
"grad_norm": 28.071439743041992,
|
| 991 |
+
"learning_rate": 7.63888888888889e-06,
|
| 992 |
+
"loss": 0.1991,
|
| 993 |
+
"step": 1380
|
| 994 |
+
},
|
| 995 |
+
{
|
| 996 |
+
"epoch": 154.0025,
|
| 997 |
+
"grad_norm": 5.893643379211426,
|
| 998 |
+
"learning_rate": 7.2916666666666674e-06,
|
| 999 |
+
"loss": 0.0053,
|
| 1000 |
+
"step": 1390
|
| 1001 |
+
},
|
| 1002 |
+
{
|
| 1003 |
+
"epoch": 155.003125,
|
| 1004 |
+
"grad_norm": 44.624385833740234,
|
| 1005 |
+
"learning_rate": 6.944444444444445e-06,
|
| 1006 |
+
"loss": 0.0189,
|
| 1007 |
+
"step": 1400
|
| 1008 |
+
},
|
| 1009 |
+
{
|
| 1010 |
+
"epoch": 156.00375,
|
| 1011 |
+
"grad_norm": 0.021997345611453056,
|
| 1012 |
+
"learning_rate": 6.597222222222223e-06,
|
| 1013 |
+
"loss": 0.0069,
|
| 1014 |
+
"step": 1410
|
| 1015 |
+
},
|
| 1016 |
+
{
|
| 1017 |
+
"epoch": 157.004375,
|
| 1018 |
+
"grad_norm": 0.01736549474298954,
|
| 1019 |
+
"learning_rate": 6.25e-06,
|
| 1020 |
+
"loss": 0.0037,
|
| 1021 |
+
"step": 1420
|
| 1022 |
+
},
|
| 1023 |
+
{
|
| 1024 |
+
"epoch": 158.005,
|
| 1025 |
+
"grad_norm": 49.95341491699219,
|
| 1026 |
+
"learning_rate": 5.902777777777778e-06,
|
| 1027 |
+
"loss": 0.2011,
|
| 1028 |
+
"step": 1430
|
| 1029 |
+
},
|
| 1030 |
+
{
|
| 1031 |
+
"epoch": 159.005625,
|
| 1032 |
+
"grad_norm": 0.027713486924767494,
|
| 1033 |
+
"learning_rate": 5.555555555555556e-06,
|
| 1034 |
+
"loss": 0.0452,
|
| 1035 |
+
"step": 1440
|
| 1036 |
+
},
|
| 1037 |
+
{
|
| 1038 |
+
"epoch": 161.000625,
|
| 1039 |
+
"grad_norm": 0.01983625814318657,
|
| 1040 |
+
"learning_rate": 5.208333333333334e-06,
|
| 1041 |
+
"loss": 0.147,
|
| 1042 |
+
"step": 1450
|
| 1043 |
+
},
|
| 1044 |
+
{
|
| 1045 |
+
"epoch": 162.00125,
|
| 1046 |
+
"grad_norm": 0.021740276366472244,
|
| 1047 |
+
"learning_rate": 4.861111111111111e-06,
|
| 1048 |
+
"loss": 0.0015,
|
| 1049 |
+
"step": 1460
|
| 1050 |
+
},
|
| 1051 |
+
{
|
| 1052 |
+
"epoch": 163.001875,
|
| 1053 |
+
"grad_norm": 0.012222817167639732,
|
| 1054 |
+
"learning_rate": 4.513888888888889e-06,
|
| 1055 |
+
"loss": 0.0274,
|
| 1056 |
+
"step": 1470
|
| 1057 |
+
},
|
| 1058 |
+
{
|
| 1059 |
+
"epoch": 164.0025,
|
| 1060 |
+
"grad_norm": 0.017902400344610214,
|
| 1061 |
+
"learning_rate": 4.166666666666667e-06,
|
| 1062 |
+
"loss": 0.017,
|
| 1063 |
+
"step": 1480
|
| 1064 |
+
},
|
| 1065 |
+
{
|
| 1066 |
+
"epoch": 165.003125,
|
| 1067 |
+
"grad_norm": 7.585049629211426,
|
| 1068 |
+
"learning_rate": 3.819444444444445e-06,
|
| 1069 |
+
"loss": 0.1415,
|
| 1070 |
+
"step": 1490
|
| 1071 |
+
},
|
| 1072 |
+
{
|
| 1073 |
+
"epoch": 166.00375,
|
| 1074 |
+
"grad_norm": 2.5589301586151123,
|
| 1075 |
+
"learning_rate": 3.4722222222222224e-06,
|
| 1076 |
+
"loss": 0.0043,
|
| 1077 |
+
"step": 1500
|
| 1078 |
+
},
|
| 1079 |
+
{
|
| 1080 |
+
"epoch": 166.00375,
|
| 1081 |
"eval_accuracy": 0.8529411764705882,
|
| 1082 |
+
"eval_loss": 0.5559400916099548,
|
| 1083 |
+
"eval_runtime": 6.6757,
|
| 1084 |
+
"eval_samples_per_second": 5.093,
|
| 1085 |
+
"eval_steps_per_second": 1.798,
|
| 1086 |
+
"step": 1500
|
| 1087 |
},
|
| 1088 |
{
|
| 1089 |
+
"epoch": 167.004375,
|
| 1090 |
+
"grad_norm": 0.014812835492193699,
|
| 1091 |
+
"learning_rate": 3.125e-06,
|
| 1092 |
+
"loss": 0.0019,
|
| 1093 |
+
"step": 1510
|
|
|
|
|
|
|
| 1094 |
},
|
| 1095 |
{
|
| 1096 |
+
"epoch": 168.005,
|
| 1097 |
+
"grad_norm": 0.4938020706176758,
|
| 1098 |
+
"learning_rate": 2.777777777777778e-06,
|
| 1099 |
+
"loss": 0.0317,
|
| 1100 |
+
"step": 1520
|
| 1101 |
+
},
|
| 1102 |
+
{
|
| 1103 |
+
"epoch": 169.005625,
|
| 1104 |
+
"grad_norm": 0.0188869908452034,
|
| 1105 |
+
"learning_rate": 2.4305555555555557e-06,
|
| 1106 |
+
"loss": 0.0038,
|
| 1107 |
+
"step": 1530
|
| 1108 |
+
},
|
| 1109 |
+
{
|
| 1110 |
+
"epoch": 171.000625,
|
| 1111 |
+
"grad_norm": 0.01809411682188511,
|
| 1112 |
+
"learning_rate": 2.0833333333333334e-06,
|
| 1113 |
+
"loss": 0.0016,
|
| 1114 |
+
"step": 1540
|
| 1115 |
+
},
|
| 1116 |
+
{
|
| 1117 |
+
"epoch": 172.00125,
|
| 1118 |
+
"grad_norm": 35.047969818115234,
|
| 1119 |
+
"learning_rate": 1.7361111111111112e-06,
|
| 1120 |
+
"loss": 0.0828,
|
| 1121 |
+
"step": 1550
|
| 1122 |
+
},
|
| 1123 |
+
{
|
| 1124 |
+
"epoch": 173.001875,
|
| 1125 |
+
"grad_norm": 0.020055020228028297,
|
| 1126 |
+
"learning_rate": 1.388888888888889e-06,
|
| 1127 |
+
"loss": 0.4192,
|
| 1128 |
+
"step": 1560
|
| 1129 |
+
},
|
| 1130 |
+
{
|
| 1131 |
+
"epoch": 174.0025,
|
| 1132 |
+
"grad_norm": 0.026819027960300446,
|
| 1133 |
+
"learning_rate": 1.0416666666666667e-06,
|
| 1134 |
+
"loss": 0.0251,
|
| 1135 |
+
"step": 1570
|
| 1136 |
+
},
|
| 1137 |
+
{
|
| 1138 |
+
"epoch": 175.003125,
|
| 1139 |
+
"grad_norm": 21.63642692565918,
|
| 1140 |
+
"learning_rate": 6.944444444444445e-07,
|
| 1141 |
+
"loss": 0.2385,
|
| 1142 |
+
"step": 1580
|
| 1143 |
+
},
|
| 1144 |
+
{
|
| 1145 |
+
"epoch": 176.00375,
|
| 1146 |
+
"grad_norm": 0.058039579540491104,
|
| 1147 |
+
"learning_rate": 3.4722222222222224e-07,
|
| 1148 |
+
"loss": 0.161,
|
| 1149 |
+
"step": 1590
|
| 1150 |
+
},
|
| 1151 |
+
{
|
| 1152 |
+
"epoch": 177.004375,
|
| 1153 |
+
"grad_norm": 0.018306700512766838,
|
| 1154 |
+
"learning_rate": 0.0,
|
| 1155 |
+
"loss": 0.1146,
|
| 1156 |
+
"step": 1600
|
| 1157 |
+
},
|
| 1158 |
+
{
|
| 1159 |
+
"epoch": 177.004375,
|
| 1160 |
+
"step": 1600,
|
| 1161 |
+
"total_flos": 5.540151384371036e+18,
|
| 1162 |
+
"train_loss": 0.6977802323055221,
|
| 1163 |
+
"train_runtime": 1181.4515,
|
| 1164 |
+
"train_samples_per_second": 4.063,
|
| 1165 |
+
"train_steps_per_second": 1.354
|
| 1166 |
+
},
|
| 1167 |
+
{
|
| 1168 |
+
"epoch": 177.004375,
|
| 1169 |
+
"eval_accuracy": 0.7916666666666666,
|
| 1170 |
+
"eval_loss": 0.8197056651115417,
|
| 1171 |
+
"eval_runtime": 9.9982,
|
| 1172 |
+
"eval_samples_per_second": 4.801,
|
| 1173 |
+
"eval_steps_per_second": 1.6,
|
| 1174 |
+
"step": 1600
|
| 1175 |
}
|
| 1176 |
],
|
| 1177 |
"logging_steps": 10,
|
| 1178 |
+
"max_steps": 1600,
|
| 1179 |
"num_input_tokens_seen": 0,
|
| 1180 |
"num_train_epochs": 9223372036854775807,
|
| 1181 |
+
"save_steps": 500,
|
| 1182 |
"stateful_callbacks": {
|
| 1183 |
"TrainerControl": {
|
| 1184 |
"args": {
|
|
|
|
| 1191 |
"attributes": {}
|
| 1192 |
}
|
| 1193 |
},
|
| 1194 |
+
"total_flos": 5.540151384371036e+18,
|
| 1195 |
+
"train_batch_size": 3,
|
| 1196 |
"trial_name": null,
|
| 1197 |
"trial_params": null
|
| 1198 |
}
|
training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 5240
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6ca6db7927270d16689303527bdbc2b0983313bc2bcb0d36e7c6e9fc30fd97f2
|
| 3 |
size 5240
|